Transcendental Dialectic

Transcendental Logic Second Division Transcendental Dialectic

Introduction

I. On Transcendental Illusion

Above we called dialectic in general a logic of illusion. This does not mean that it is a doctrine of probability. Probability is truth cognized through insufficient grounds; cognition of it is therefore deficient, but not on that account deceptive, and hence need not be separated from the analytic part of logic. Still less may appearance and illusion be regarded as identical.

B 350 For truth or illusion lies not in the object insofar as it is intuited, but in the judgment about it insofar as it is thought. Thus one can indeed correctly say that the senses do not err, but not because they always judge correctly: rather, because they do not judge at all. Hence truth as well as error, and therefore illusion as the inducement to error, are to be found only in judgment, only in the relation of the object to our understanding. In a cognition agreeing throughout with the laws of understanding there is no error. Nor is there error in a representation of the senses, because it contains no judgment at all. No force of nature, however, can deviate of itself from its own laws. Thus neither understanding alone, without the influence of another cause, nor the senses by themselves would err. Understanding would not, because when it acts merely according to its laws, the effect, the judgment, must necessarily agree with those laws. Agreement with the laws of understanding constitutes the formal element of all truth. In the senses there is no judgment at all, either true or false. Since we have no sources of cognition other than these two, it follows that error is produced only through the unnoticed influence of sensibility on understanding.

B 351 Through this influence the subjective grounds of judgment merge with the objective grounds and cause them to deviate from their determination*, just as a moving body would by itself always maintain a straight line in the same direction, but moves in a curve when another force simultaneously influences it in another direction. To distinguish the understanding’s own action from the force interfering with it, we must therefore regard an erroneous judgment as the diagonal between two forces determining judgment in two different directions, enclosing, as it were, an angle. We must resolve that composite effect into the simple effects of understanding and sensibility. In pure a priori judgments this must be done through transcendental reflection, which, as already indicated, assigns each representation its place in the cognitive power appropriate to it and thus also distinguishes the influence of that power on the representation.

* Sensibility, placed beneath understanding as the object to which it applies its function, is the source of real cognitions. But this same sensibility, insofar as it influences the very action of understanding and determines it to judge, is the ground of error.

Our business here is not to discuss empirical illusion, optical illusion for example.

B 352 This occurs in the empirical use of otherwise correct rules of understanding, and through it the power of judgment is misled by the influence of imagination. We are concerned solely with transcendental illusion. This influences principles whose use is not even directed toward experience, where we would at least have a touchstone of their correctness. Instead, despite all the warnings of critique, it leads us entirely beyond the empirical use of the categories and holds out to us the deceptive prospect of extending pure understanding. We will call principles whose application remains wholly within the limits of possible experience immanent, and those that are to fly beyond these boundaries transcendent principles. By the latter, however, I do not mean the transcendental use or misuse of categories, which is merely an error of the power of judgment insufficiently restrained by critique, failing to attend adequately to the boundary of the ground on which alone pure understanding is permitted to operate. I mean actual principles urging us to tear down all those boundary markers and lay claim to an entirely new territory that recognizes no demarcation anywhere. Thus transcendental and transcendent are not the same.

B 353 The principles of pure understanding set out above are to have merely empirical and not transcendental use, that is, use reaching beyond the boundary of experience. But a principle removing these limits, indeed commanding us to overstep them, is called transcendent. If our critique can succeed in exposing the illusion of these pretended principles, the principles of merely empirical use can, in contrast to them, be called immanent principles of pure understanding.

Logical illusion, which consists in the mere imitation of the form of reason, the illusion of fallacies, arises solely from a lack of attention to the logical rule. As soon as attention is sharpened upon the case at hand, it therefore disappears entirely. Transcendental illusion, by contrast, does not cease even after it has been exposed and its nullity clearly understood through transcendental critique: for example, the illusion in the proposition that the world must have a beginning in time. The cause is that in our reason, considered subjectively as a human faculty of cognition, there lie fundamental rules and maxims of its use that have entirely the appearance of objective principles. Through these the subjective necessity of a certain connection of our concepts for the benefit of understanding is taken for an objective necessity in the determination of things in themselves.

B 354 This is an illusion that cannot be avoided at all, any more than we can avoid the sea’s appearing higher in the middle than at the shore because we see the former through higher rays of light than the latter; or, still more, any more than even the astronomer can prevent the moon from appearing larger as it rises, although he is not deceived by this illusion.

Transcendental dialectic will therefore content itself with exposing the illusion in transcendent judgments and at the same time preventing it from deceiving us. But that it should also disappear like logical illusion, and cease to be an illusion, is something dialectic can never bring about. For we are dealing with a natural and unavoidable illusion, itself resting on subjective principles that it passes off as objective, whereas logical dialectic, in resolving fallacies, deals only with an error in following principles, or an artificial illusion in imitating them. Thus there is a natural and unavoidable dialectic of pure reason: not one in which a bungler becomes entangled through lack of knowledge, or one artificially devised by some sophist to confuse reasonable people, but one inseparably attached to human reason. Even after we have exposed its deception, it will still not cease to beguile reason.

B 355 It will continually thrust reason into momentary aberrations that always need to be removed.

II. On Pure Reason as the Seat of Transcendental Illusion

A. On Reason in General

All our cognition begins with the senses, proceeds from there to understanding, and ends with reason. Nothing higher is found in us for working on the material of intuition and bringing it under the highest unity of thinking. Now that I am to give an explanation of this supreme cognitive power, I find myself in some difficulty. As with understanding, it has a merely formal, logical use, in which reason abstracts from all content of cognition, but also a real use, in which reason itself contains the origin of certain concepts and principles that it borrows neither from the senses nor from understanding. Logicians have of course long since explained the first faculty as the faculty of making mediate inferences, in distinction from immediate inferences (consequentiis immediatis); but the second, which itself generates concepts, is not yet understood through this.

B 356 Since reason is here divided into a logical and a transcendental faculty, a higher concept of this source of cognition must be sought that includes both concepts under it. Meanwhile, by analogy with the concepts of the understanding, we may expect the logical concept to supply the key to the transcendental concept, and the table of functions of the former to supply the genealogy of concepts of reason.

In the first part of our transcendental logic we explained understanding as the faculty of rules. Here we distinguish reason from it by calling reason the faculty of principles.

The expression “principle” is ambiguous, and commonly signifies only a cognition that can be used as a principle, although in itself and in its own origin it is not a principle. Every universal proposition, even one drawn from experience through induction, can serve as the major premise in a syllogism, but is not on that account itself a principle. Mathematical axioms, for example that between two points there can be only one straight line, are even universal a priori cognitions, and are therefore rightly called principles relative to the cases that can be subsumed under them.

B 357 Yet I cannot on that account say that I cognize this property of straight lines in general and in itself from principles, but only in pure intuition.

I would therefore call cognition from principles that cognition in which I cognize the particular in the universal through concepts. Every syllogism is thus a form of deriving a cognition from a principle. For the major premise always supplies a concept such that everything subsumed under its condition is cognized from it according to a principle. Now since every universal cognition can serve as the major premise in a syllogism, and understanding offers such universal propositions a priori, these too can be called principles in respect of their possible use.

But if we consider these principles of pure understanding in themselves as to their origin, they are anything but cognitions from concepts. For they would not even be possible a priori unless we brought in pure intuition, in mathematics, or the conditions of a possible experience in general. That everything happening has a cause cannot at all be inferred from the concept of what happens in general. Rather, the principle shows how we can first obtain a determinate experiential concept of what happens.

Understanding therefore cannot supply synthetic cognitions from concepts at all.

B 358 These are properly what I call principles without qualification, whereas all universal propositions in general can be called principles comparatively.

It is an old wish, which may one day be fulfilled, who knows how late, that instead of the endless variety of civil laws their principles might at last be sought. In this alone can lie the secret of simplifying legislation, as the saying goes. But laws here are also only restrictions of our freedom to conditions under which it agrees thoroughly with itself. They thus concern something entirely our own work, of which we ourselves can be the cause through these concepts. But how objects in themselves, how the nature of things, should stand under principles and be determined according to mere concepts is, if not impossible, at least a very unreasonable demand. However matters may stand here, since the investigation still lies before us, this at least is clear: cognition from principles, in itself, is something entirely different from mere cognition through understanding. The latter can indeed also precede other cognitions in the form of a principle, but in itself, insofar as it is synthetic, neither rests on mere thinking nor contains a universal according to concepts.

B 359 If understanding is a faculty for the unity of appearances by means of rules, reason is the faculty for the unity of the rules of understanding under principles. Thus it never relates immediately to experience or to any object, but to understanding, in order to give its manifold cognitions unity a priori through concepts. This may be called unity of reason and is of a wholly different kind from the unity understanding can provide.

This is the general concept of the faculty of reason, insofar as it could be made comprehensible in the complete absence of examples, which are to be supplied only later.

B. On the Logical Use of Reason

A distinction is made between what is cognized immediately and what is only inferred. That a figure bounded by three straight lines has three angles is cognized immediately; that these angles together equal two right angles is only inferred. Because we constantly need inference and thereby finally become entirely accustomed to it, we eventually cease to notice this distinction and often take something as immediately perceived that we have only inferred, as in the so-called deception of the senses.

B 360 In every inference there is a proposition laid down as ground, another proposition, the conclusion drawn from it, and finally the consequence according to which the truth of the latter is inevitably connected with the truth of the former. If the inferred judgment already lies in the first in such a way that it can be derived without the mediation of a third representation, the inference is called immediate (consequentia immediata); I would rather call it an inference of understanding. But if, besides the cognition laid down as ground, another judgment is needed to bring about the consequence, the inference is called a syllogism. In the proposition “All human beings are mortal” there already lie the propositions “Some human beings are mortal,” “Some mortals are human beings,” and “Nothing immortal is a human being”; these are therefore immediate conclusions from the first. By contrast, the proposition “All scholars are mortal” does not lie in the underlying judgment, since the concept of scholars does not occur in it at all; it can be inferred from it only through an intermediate judgment.

In every syllogism I first think a rule (major) through understanding. Second, through the power of judgment I subsume a cognition under the condition of the rule (minor). Finally, I determine my cognition through the predicate of the rule (conclusio), and hence a priori through reason.

B 361 The relation that the major premise, as rule, represents between a cognition and its condition therefore constitutes the different kinds of syllogism. There are accordingly exactly three kinds, just as with judgments in general insofar as they differ in how they express the relation of cognition in understanding: categorical, hypothetical, or disjunctive syllogisms.

When, as usually happens, the conclusion is given as a judgment in order to see whether it follows from judgments already given, through which an entirely different object is thought, I search understanding for the assertion of this concluding proposition, to see whether it is found there under certain conditions according to a universal rule. If I now find such a condition, and the object of the concluding proposition can be subsumed under the given condition, the proposition is inferred from the rule, which also holds for other objects of cognition. From this we see that in inference reason seeks to bring the great manifold of understanding’s cognition under the smallest number of principles, universal conditions, and thereby bring about its highest unity.

B 362 C. On the Pure Use of Reason

Can reason be isolated, and is it then still a source of its own of concepts and judgments arising solely from it, through which it relates to objects? Or is it merely a subordinate faculty for giving given cognitions a certain form called logical, through which the cognitions of understanding are merely subordinated to one another, and lower rules to higher ones whose condition includes the condition of the former within its sphere, insofar as comparing them can accomplish this? This is the question with which we are now concerned only provisionally. In fact, a manifold of rules and unity of principles is a demand of reason for bringing understanding into thoroughgoing connection with itself, just as understanding brings the manifold of intuition under concepts and thereby brings that manifold into connection. But such a principle prescribes no law to objects and contains no ground of the possibility of cognizing and determining them as such in general. It is merely a subjective law for managing the stock of our understanding: by comparing its concepts, to reduce their universal use to the smallest possible number of them.

B 363 This does not entitle us to demand from the objects themselves such an accord as would promote the convenience and extension of our understanding, and thereby give that maxim objective validity as well. In a word, the question is whether reason in itself, pure reason, contains synthetic a priori principles and rules, and what these principles may consist in.

Its formal and logical procedure in syllogisms already gives us sufficient guidance as to the ground on which its transcendental principle in synthetic cognition through pure reason will rest.

First, a syllogism relates not to intuitions in order to bring them under rules, as understanding does with its categories, but to concepts and judgments. Thus even if pure reason relates to objects, it has no immediate relation to them and their intuition, but only to understanding and its judgments, which turn immediately to the senses and their intuition in order to determine their object for them. Unity of reason is therefore not unity of a possible experience, but is essentially different from the latter as unity of understanding. That everything happening has a cause is no principle cognized and prescribed by reason. It makes the unity of experience possible and borrows nothing from reason.

B 364 Without this relation to possible experience, reason could not have commanded such synthetic unity from mere concepts.

Second, in its logical use reason seeks the universal condition of its judgment, the concluding proposition. A syllogism itself is nothing other than a judgment made by subsuming its condition under a universal rule, the major premise. Since this rule is in turn exposed to the same attempt by reason, and the condition of the condition must therefore be sought through a prosyllogism as far as possible, it is clear that reason’s distinctive principle in general, in its logical use, is to find for the conditioned cognition of understanding the unconditioned through which its unity is completed.

But this logical maxim can become a principle of pure reason only by assuming that, if the conditioned is given, the entire series of conditions subordinated to one another, which is therefore itself unconditioned, is also given (that is, contained in the object and its connection).

Such a principle of pure reason is manifestly synthetic. For the conditioned relates analytically to some condition, but not to the unconditioned. Various synthetic propositions must also arise from this principle, of which pure understanding knows nothing.

B 365 It is concerned only with objects of a possible experience, whose cognition and synthesis are always conditioned. But the unconditioned, if it actually occurs, can be considered separately according to all the determinations distinguishing it from everything conditioned, and must thereby provide material for many synthetic a priori propositions.

The principles arising from this supreme principle of pure reason will, however, be transcendent with respect to all appearances: no empirical use adequate to it can ever be made of it. It will therefore differ entirely from all principles of understanding, whose use is wholly immanent since their sole theme is the possibility of experience. Is that principle—that the series of conditions, in the synthesis of appearances or even in thinking things in general, extends to the unconditioned—objectively correct or not, and what consequences flow from it for the empirical use of understanding? Or is there, rather, no such objectively valid proposition of reason anywhere, but merely a logical prescription to approach completeness by ascending to ever higher conditions, and thereby to bring into our cognition the highest unity of reason possible for us?

B 366 Has this need of reason, I say, been taken through a misunderstanding for a transcendental principle of pure reason, one that hastily postulates such unlimited completeness of the series of conditions in the objects themselves? And what misinterpretations and delusions may creep, even in this case, into syllogisms whose major premise is taken from pure reason (and is perhaps more a petition than a postulate), and which ascend from experience to its conditions? This will be our business in transcendental dialectic. We will now develop it from its sources, deeply hidden in human reason. We will divide it into two chapters, the first treating the transcendent concepts of pure reason, the second its transcendent and dialectical syllogisms.

Transcendental Dialectic First Book On the Concepts of Pure Reason

Whatever the possibility of concepts from pure reason may involve, they are not merely reflected but inferred concepts.

B 367 Concepts of the understanding are also thought a priori before experience and for its sake, but contain nothing beyond the unity of reflection on appearances insofar as these must necessarily belong to a possible empirical consciousness. Through these concepts alone do cognition and determination of an object become possible. They therefore first provide material for inference, and no a priori concepts of objects precede them from which they could be inferred. Their objective reality, however, rests solely on this: because they constitute the intellectual form of all experience, their application must always be capable of being shown in experience.

The designation “concept of reason,” however, already provisionally indicates that it will not allow itself to be confined within experience, because it concerns a cognition of which every empirical cognition is only a part, perhaps the whole of possible experience or of its empirical synthesis. No actual experience ever fully reaches this whole, though it always belongs to it. Concepts of reason serve for comprehension, as concepts of the understanding serve for understanding perceptions. If they contain the unconditioned, they concern something under which all experience belongs, but which is never itself an object of experience: something to which reason leads in its inferences from experience, and according to which it estimates and measures the degree of its empirical use, but which never constitutes a member of empirical synthesis.

B 368 If such concepts nevertheless have objective validity, they can be called conceptus ratiocinati (correctly inferred concepts); if not, they have at least been surreptitiously obtained through an illusion of inference and may be called conceptus ratiocinantes (speciously reasoning concepts). Since this can be settled only in the chapter on the dialectical inferences of pure reason, we cannot yet take it into account. Provisionally, just as we called pure concepts of the understanding categories, we will give the concepts of pure reason a new name, calling them transcendental ideas, and will now explain and justify this designation.

First Book of Transcendental Dialectic First Section On Ideas in General

Despite the great wealth of our languages, a thinking mind often finds itself at a loss for an expression exactly fitting its concept, without which it cannot make itself properly intelligible to others or even to itself.

B 369 To coin new words is a presumption to legislate in language that rarely succeeds. Before resorting to this desperate remedy, it is advisable to look in a dead and learned language to see whether the concept, together with its appropriate expression, is already to be found there. Even if the old use has become somewhat uncertain through the carelessness of its originators, it is still better to fix the meaning that was especially proper to it, even should it remain doubtful whether precisely the same meaning was intended at the time, than to spoil one’s undertaking merely by making oneself unintelligible.

Thus if only a single word should be found for a certain concept, fitting it exactly in an already established meaning, and if distinguishing this concept from related concepts is of great importance, it is advisable not to use that word extravagantly or merely for variety as a synonym in place of others. Rather, one should carefully preserve its distinctive meaning. Otherwise it easily happens that, once the expression no longer particularly engages attention but is lost among a mass of others of very different significance, the thought it alone could have preserved is also lost.

B 370 Plato used the expression “idea” in a way that makes it clear he understood by it something not only never borrowed from the senses, but even far surpassing the concepts of the understanding with which Aristotle concerned himself, since nothing congruent with it is ever found in experience. For him ideas are archetypes of things themselves, not merely keys to possible experiences, like the categories. In his opinion they flowed from supreme reason and were thence imparted to human reason. The latter, however, is no longer in its original state, but must laboriously recall the old, now much obscured ideas through recollection, which is called philosophy. I will not enter here into a literary investigation to establish the meaning this sublime philosopher attached to his expression. I note only that it is not at all unusual, in ordinary conversation as well as in writings, to understand an author even better than he understood himself by comparing the thoughts he expresses about his subject. For he did not determine his concept sufficiently, and thereby sometimes spoke, or even thought, contrary to his own intention.

Plato very clearly observed that our cognitive power feels a far higher need than merely spelling out appearances according to synthetic unity in order to read them as experience.

B 371 He observed too that our reason naturally soars to cognitions that go far beyond anything with which any object experience can give could ever be congruent, yet which nevertheless have their reality and are by no means mere fantasies.

Plato found his ideas especially in everything practical*, that is, everything resting on freedom, which in turn stands under cognitions that are a distinctive product of reason. Anyone who wished to draw the concepts of virtue from experience, who wished to make what can at most serve as an example for imperfect elucidation into a model and source of cognition, as many have actually done, would turn virtue into an ambiguous nonentity, changing with time and circumstances and useless for any rule.

B 372 By contrast, everyone becomes aware that, when someone is presented to him as a model of virtue, he still has the true original only in his own mind, with which he compares this supposed model and by which alone he estimates it. This is the idea of virtue. In relation to it all possible objects of experience serve indeed as examples, proofs of the practicability to a certain degree of what the concept of reason demands, but not as archetypes. That no human being will ever act adequately to what the pure idea of virtue contains proves nothing chimerical in this thought. For every judgment of moral worth or unworthiness is nevertheless possible only through this idea. Hence it necessarily underlies every approach to moral perfection, however far the obstacles in human nature, indeterminable in degree, may keep us from it.

* He admittedly extended his concept also to speculative cognitions, provided only that they were pure and given entirely a priori, even to mathematics, although its object lies nowhere but in possible experience. Here I cannot follow him, any more than in the mystical deduction of these ideas or the exaggerations through which he, as it were, hypostatized them; although the lofty language he used in this field is quite capable of a milder interpretation appropriate to the nature of things.

Plato’s Republic has become proverbial as a supposedly striking example of dreamed-up perfection that can have its seat only in an idle thinker’s brain, and Brucker finds it ridiculous that the philosopher asserted that a prince would never govern well unless he participated in the ideas. Yet one would do better to pursue this thought further and illuminate it through fresh effort where that excellent man leaves us unaided, rather than set it aside as useless under the very wretched and harmful pretext of impracticability.

B 373 A constitution providing the greatest human freedom according to laws that allow each person’s freedom to coexist with everyone else’s—not the greatest happiness, for this will follow of itself—is at least a necessary idea. It must underlie not merely the first design of a political constitution but all laws as well. At the outset one must abstract from present obstacles, which may arise not so much inevitably from human nature as from neglect of genuine ideas in legislation. For nothing can be more harmful or more unworthy of a philosopher than the vulgar appeal to experience contradicting in vain, which would not exist at all if those institutions had been established at the proper time according to ideas, instead of crude concepts, precisely because they were drawn from experience, frustrating every good intention. The more legislation and government were arranged in agreement with this idea, the rarer punishments would certainly become; and it is then quite reasonable, as Plato asserts, that with a perfect arrangement of them none would be needed at all. Although this last state may never come about, the idea is nevertheless entirely correct.

B 374 It sets up this maximum as an archetype, in order to bring the legal constitution of human beings ever closer to the greatest possible perfection. For what the highest degree may be at which humanity must stop, and hence how great the gulf necessarily remaining between the idea and its execution may be, no one can or should determine, precisely because freedom can overstep every specified boundary.

But not only in what displays the true causality of human reason, where ideas become efficient causes of actions and their objects, namely in morality, but also with respect to nature itself, Plato rightly sees clear proofs of its origin in ideas. A plant, an animal, the orderly arrangement of the universe, and presumably therefore the whole order of nature, clearly show that they are possible only according to ideas. No individual creature under the individual conditions of its existence is congruent with the idea of what is most perfect of its kind, any more than a human being is with the idea of humanity that he even carries in his own soul as the archetype of his actions. Yet those ideas in the highest understanding are individual, unalterable, thoroughly determined, and the original causes of things; only the whole of their combination in the universe is fully adequate to that idea.

B 375 If we set aside the exaggeration of expression, the philosopher’s soaring from a consideration of the physical world order as a copy to its architectonic connection according to ends, that is, ideas, is an endeavor deserving respect and emulation. With regard to the principles of morality, legislation, and religion, however, where ideas first make experience itself (of the good) possible, though they can never be fully expressed in it, it is a quite distinctive merit. This merit goes unrecognized only because it is judged by those very empirical rules whose validity as principles was precisely what the ideas were intended to abolish. For with respect to nature experience supplies the rule and is the source of truth; with respect to moral laws, however, experience is (sadly!) the mother of illusion. It is utterly reprehensible to derive the laws of what I ought to do from what is done, or to wish to restrict them by it.

Instead of all these considerations, whose proper elaboration in fact constitutes the distinctive dignity of philosophy, we are now engaged in a less brilliant but not unmeritorious task: making the ground level and firm for those majestic moral edifices.

B 376 In that ground lie all kinds of mole tunnels dug by a reason searching for treasure in vain but with great confidence, which make the building insecure. Thus it is the transcendental use of pure reason, its principles and ideas, that we must now know precisely, in order properly to determine and estimate the influence and worth of pure reason. Yet before laying aside this preliminary introduction, I ask those to whom philosophy matters—which is saying more than one commonly finds—if they should find themselves convinced by this and what follows, to protect the expression “idea” in its original meaning. It should no longer fall among the other expressions commonly used to designate all kinds of representation in careless disorder, to the detriment of science. We do not lack designations properly suited to every kind of representation, without needing to encroach on the property of another. Here is their ascending scale. The genus is representation in general (repraesentatio). Beneath it stands representation with consciousness (perceptio). A perception relating solely to the subject as the modification of its state is sensation (sensatio); an objective perception is cognition (cognitio). Cognition is either intuition or concept (intuitus vel conceptus).

B 377 The former relates immediately to the object and is singular; the latter mediately, through a mark that can be common to several things. A concept is either empirical or pure; and a pure concept, insofar as it originates solely in understanding, not in the pure image of sensibility, is called notio. A concept made up of notions that surpasses the possibility of experience is an idea, or concept of reason. Anyone once accustomed to this distinction must find it intolerable to hear the representation of the color red called an idea. It cannot even be called a notion, a concept of the understanding.

First Book of Transcendental Dialectic Second Section On Transcendental Ideas

The transcendental analytic gave us an example of how the mere logical form of our cognition can contain the origin of pure a priori concepts that represent objects before all experience, or rather indicate the synthetic unity that alone makes empirical cognition of objects possible.

B 378 The form of judgments, transformed into a concept of the synthesis of intuitions, produced categories guiding every use of understanding in experience. Likewise we may expect the form of syllogisms, when applied to the synthetic unity of intuitions according to the categories, to contain the origin of special a priori concepts. These we can call pure concepts of reason, or transcendental ideas, and they will determine the use of understanding in the whole of all experience according to principles.

The function of reason in its inferences consisted in the universality of cognition according to concepts; a syllogism itself is a judgment determined a priori in the entire extent of its condition. I could also draw the proposition “Caius is mortal” from experience merely through understanding. But I seek a concept containing the condition under which the predicate, the assertion in general, of this judgment is given, here the concept of a human being. After subsuming under this condition taken in its entire extent, “All human beings are mortal,” I determine accordingly the cognition of my object: “Caius is mortal.”

Thus in the conclusion of a syllogism we restrict a predicate to a certain object.

B 379 We have previously thought it in the major premise in its entire extent under a certain condition. This completed magnitude of extent in relation to such a condition is called universality (Universalitas). Corresponding to it in the synthesis of intuitions is allness (Universitas), or the totality of conditions. Thus the transcendental concept of reason is none other than the concept of the totality of conditions for a given conditioned. Since the unconditioned alone makes the totality of conditions possible, and conversely the totality of conditions is always itself unconditioned, a pure concept of reason in general can be explained through the concept of the unconditioned insofar as it contains a ground of the synthesis of the conditioned.

There will be as many kinds of pure concepts of reason as there are kinds of relation that understanding represents through the categories. Thus we must seek, first, an unconditioned of categorical synthesis in a subject; second, of hypothetical synthesis of the members of a series; and third, of disjunctive synthesis of the parts in a system.

There are, namely, just as many kinds of syllogism, each advancing through prosyllogisms to the unconditioned: one to the subject that is no longer itself a predicate, another to the presupposition that presupposes nothing further.

B 380 The third advances to an aggregate of the members of a division, to which nothing further is required in order to complete the division of a concept. Thus pure concepts of reason concerning totality in the synthesis of conditions are necessary, at least as tasks for extending the unity of understanding, if possible, to the unconditioned, and are grounded in the nature of human reason. These transcendental concepts may otherwise lack a suitable use in concreto and consequently have no utility other than to guide understanding in a direction in which its use, while extended to the utmost, is at the same time brought into thoroughgoing agreement with itself.

But in speaking here of the totality of conditions and the unconditioned as the common heading of all concepts of reason, we again encounter an expression we cannot dispense with, yet cannot safely use owing to an ambiguity attached to it through long misuse. The word “absolute” is one of the few words originally fitted to a concept for which afterward no other word in the same language fits exactly. Its loss, or what amounts to the same thing, its uncertain use, must therefore entail the loss of the concept itself.

B 381 It is a concept that, because it greatly occupies reason, cannot be dispensed with without great harm to all transcendental judging. The word “absolute” is now often used merely to indicate that something holds of a thing considered in itself, and thus internally. In this sense “absolutely possible” would mean what is possible in itself (interne), which is in fact the least one can say of an object. By contrast, it is also sometimes used to indicate that something holds in every relation (unrestrictedly), as with absolute dominion. “Absolutely possible” in this sense would mean what is possible in every respect and every relation, which is in turn the most I can say about the possibility of a thing. Now these meanings sometimes coincide. For example, what is internally impossible is also impossible in every relation, and hence absolutely impossible. But in most cases they are infinitely far apart, and I can in no way infer that, because something is possible in itself, it is therefore possible in every relation and hence absolutely possible. Indeed, I will show later of absolute necessity that it by no means depends in every case on inner necessity, and therefore must not be regarded as synonymous with it.

B 382 If a thing’s opposite is internally impossible, its opposite is certainly also impossible in every respect, and hence the thing itself is absolutely necessary. But I cannot infer conversely that the opposite of what is absolutely necessary is internally impossible, that the absolute necessity of things is an inner necessity. For this inner necessity is in certain cases an entirely empty expression with which we can associate not the least concept, whereas the concept of the necessity of a thing in every relation, to everything possible, carries quite special determinations with it. Since the loss of a concept widely used in speculative philosophy can never be indifferent to a philosopher, I hope that determining and carefully preserving the expression on which the concept depends will not be indifferent to him either.

I will therefore use the word “absolute” in this extended meaning and oppose it to what holds merely comparatively or in a particular respect. For the latter is restricted to conditions, while the former holds without restriction.

Now the transcendental concept of reason always concerns only absolute totality in the synthesis of conditions, and ends only with what is unconditioned absolutely, in every relation.

B 383 For pure reason leaves everything to understanding that relates immediately to objects of intuition, or rather to their synthesis in imagination. Reason reserves for itself only absolute totality in the use of concepts of the understanding, and seeks to carry the synthetic unity thought in the category as far as the absolutely unconditioned. We can therefore call this the unity of reason in appearances, just as the unity expressed by the category is called unity of understanding. Thus reason relates only to the use of understanding, not insofar as this contains the ground of possible experience—for absolute totality of conditions is no concept usable in an experience, since no experience is unconditioned—but to prescribe its direction toward a certain unity of which understanding has no concept. This unity aims to gather all actions of understanding with respect to each object into an absolute whole. Hence the objective use of pure concepts of reason is always transcendent, whereas that of pure concepts of the understanding must by its nature always be immanent, since it is restricted solely to possible experience.

By an idea I understand a necessary concept of reason to which no congruent object can be given in the senses. Thus the pure concepts of reason we are now considering are transcendental ideas.

B 384 They are concepts of pure reason, for they regard all experiential cognition as determined through an absolute totality of conditions. They are not arbitrarily invented, but set as tasks by the nature of reason itself, and therefore necessarily relate to the entire use of understanding. Finally, they are transcendent and overstep the boundary of all experience, in which no object adequate to the transcendental idea can therefore ever occur. In naming an idea, one says a great deal with respect to the object, as an object of pure understanding, but for that very reason very little with respect to the subject, that is, its actuality under empirical conditions, because as the concept of a maximum it can never be given congruently in concreto. Since in the merely speculative use of reason this latter is properly the entire aim, and approaching a concept never attained in practice amounts to missing the concept altogether, we say of such a concept that it is only an idea. Thus one could say that the absolute whole of all appearances is only an idea, for since we can never portray such a whole in an image, it remains a problem without any solution. By contrast, in the practical use of understanding everything is concerned solely with carrying out rules.

B 385 Hence the idea of practical reason can always actually be given in concreto, though only in part; indeed, it is the indispensable condition of every practical use of reason. Its execution is always bounded and deficient, but under indeterminable boundaries, and thus always under the influence of the concept of absolute completeness. Accordingly the practical idea is always supremely fruitful and indispensable with respect to actual actions. In it pure reason even has causality for actually producing what its concept contains. Hence one cannot say of wisdom with a sort of contempt that it is only an idea. Rather, precisely because it is the idea of the necessary unity of all possible ends, it must serve as a rule for everything practical, as an original and at least limiting condition.

Although we must therefore say of transcendental concepts of reason that they are only ideas, we must by no means regard them as superfluous and null. For although no object can be determined through them, they can nevertheless, fundamentally and unnoticed, serve understanding as a canon of its extended and harmonious use. Through this it indeed cognizes no more objects than it would cognize according to its concepts, but is guided better and further in that cognition.

B 386 Not to mention that they may perhaps make possible a transition from concepts of nature to practical concepts, and in this way give the moral ideas themselves support and connection with the speculative cognitions of reason. We must await clarification of all this in what follows.

In keeping with our purpose, however, we set practical ideas aside here and therefore consider reason only in its speculative use, and within this more narrowly still, only in its transcendental use. Here we must follow the same path we took above in the deduction of the categories: consider the logical form of cognition through reason, and see whether reason thereby also becomes a source of concepts for regarding objects in themselves as synthetically determined a priori with respect to one or another function of reason.

Reason, considered as the faculty of a certain logical form of cognition, is the faculty of inference, of judging mediately by subsuming the condition of a possible judgment under the condition of a given judgment. The given judgment is the universal rule, the major premise (Major). Subsuming the condition of another possible judgment under the condition of the rule is the minor premise (Minor). The actual judgment asserting the rule’s assertion in the subsumed case is the conclusion (Conclusio).

B 387 The rule, namely, asserts something universally under a certain condition. Now in a case at hand the condition of the rule obtains. Thus what held universally under that condition is also regarded as valid in the case at hand, which carries that condition with it. It is easy to see that reason arrives at a cognition through actions of understanding that constitute a series of conditions. Suppose I arrive at the proposition “All bodies are alterable” only by beginning with the more remote cognition, which does not yet contain the concept of body but nevertheless contains its condition, “Everything composite is alterable”; proceeding from this to a nearer cognition standing under the condition of the former, “Bodies are composite”; and only from this to a third, which now connects the remote cognition, alterable, with the present one: “Therefore bodies are alterable.” I have then arrived at a cognition, a conclusion, through a series of conditions, premises. Now every series whose exponent, of the categorical or hypothetical judgment, is given can be continued. Hence this same action of reason leads to the ratiocinatio polysyllogistica: a series of inferences that can be continued to indefinite lengths on the side of conditions (per prosyllogismos).

B 388 Or it can be continued on the side of the conditioned (per episyllogismos).

But we soon become aware that the chain or series of prosyllogisms, inferred cognitions on the side of grounds or conditions for a given cognition—in other words, the ascending series of syllogisms—must stand in a different relation to the faculty of reason from the descending series, reason’s advance on the side of the conditioned through episyllogisms. For in the first case the cognition, the conclusion (conclusio), is given only as conditioned. We can therefore arrive at it through reason only under at least the presupposition that all members of the series on the side of conditions are given, totality in the series of premises, because only under that presupposition is the judgment before us possible a priori. On the side of the conditioned or the consequences, by contrast, we think only a developing series, not one already entirely presupposed or given, and hence only a potential advance. Thus when a cognition is regarded as conditioned, reason is compelled to regard the ascending series of conditions as completed and given in its totality.

B 389 But when the very same cognition is regarded at the same time as the condition of other cognitions that together constitute a descending series of consequences, reason can be entirely indifferent to how far this advance extends a parte posteriori and whether totality of this series is even possible anywhere. For it does not need such a series for the conclusion before it, since that conclusion is already sufficiently determined and secured by its grounds a parte priori. Now whether the series of premises on the side of conditions has a first member as supreme condition or not, and thus is without bounds a parte priori, it must nevertheless contain totality of condition, even supposing we could never succeed in grasping it. The whole series must be unconditionally true if the conditioned, regarded as a consequence arising from it, is to count as true. This is a demand of reason, which declares its cognition determined a priori and necessary: either in itself, in which case it needs no grounds, or, if derived, as a member of a series of grounds that is itself unconditionally true.

B 390 First Book of Transcendental Dialectic Third Section System of Transcendental Ideas

We are concerned here not with a logical dialectic that abstracts from all content of cognition and merely exposes false illusion in the form of syllogisms, but with a transcendental dialectic. This is to contain, entirely a priori, the origin of certain cognitions from pure reason and of inferred concepts whose object cannot be given empirically at all, and which therefore lie wholly outside the power of pure understanding. From the natural relation that the transcendental use of our cognition, in inferences as well as judgments, must have to its logical use, we have gathered that there will be only three kinds of dialectical inference. These relate to the three kinds of inference through which reason can arrive at cognitions from principles, and in all of them its business is to ascend from the conditioned synthesis to which understanding always remains bound to the unconditioned synthesis that it can never attain.

Now the universal forms of all relation our representations can have are 1.

B 391 relation to the subject, 2. relation to objects, either as appearances or as objects of thinking in general. Combining this subdivision with the division above, every relation of representations of which we can form either a concept or an idea is threefold: 1. the relation to the subject, 2. to the manifold of the object in appearance, 3. to all things in general.

Now all pure concepts in general concern the synthetic unity of representations, but concepts of pure reason, transcendental ideas, concern the unconditioned synthetic unity of all conditions in general. Consequently all transcendental ideas can be brought under three classes: the first containing the absolute, unconditioned unity of the thinking subject; the second the absolute unity of the series of conditions of appearance; the third the absolute unity of the condition of all objects of thinking in general.

The thinking subject is the object of psychology; the sum total of all appearances (the world) is the object of cosmology; and the thing containing the supreme condition of the possibility of everything that can be thought, the being of all beings, is the object of theology.

B 392 Thus pure reason supplies the idea for a transcendental doctrine of the soul (psychologia rationalis), a transcendental science of the world (cosmologia rationalis), and finally a transcendental cognition of God (Theologia transzendentalis). Even the mere design of any one of these sciences does not derive from understanding, even if understanding were combined with the highest logical use of reason, all conceivable inferences, in order to advance from one of its objects, an appearance, to all others as far as the most remote members of empirical synthesis. It is solely a pure and genuine product, or problem, of pure reason.

What modi of pure concepts of reason stand under these three headings of all transcendental ideas will be fully set out in the following chapter. They proceed along the thread of the categories. For pure reason never relates directly to objects, but to the concepts of the understanding concerning them. Likewise only in the full development will it become clear how reason, solely through the synthetic use of the very same function it employs in categorical syllogism, must necessarily arrive at the concept of the absolute unity of the thinking subject; how the logical procedure in hypothetical ideas must necessarily entail the idea of the absolutely unconditioned in a series of given conditions.

B 393 Finally, it will become clear how the mere form of disjunctive syllogism must necessarily entail the highest concept of reason, that of a being of all beings. At first sight this thought seems exceedingly paradoxical.

No objective deduction of these transcendental ideas is properly possible, as we were able to provide for the categories. For in fact they have no relation to any object that could be given congruently with them, precisely because they are only ideas. But we could undertake a subjective introduction to them from the nature of our reason, and this has also been accomplished in the present chapter.

It is easy to see that pure reason aims at nothing but absolute totality of synthesis on the side of conditions, whether of inherence, dependence, or concurrence, and has nothing to do with absolute completeness on the side of the conditioned. For only the former is needed to presuppose the entire series of conditions and thereby give it to understanding a priori. Once a completely and unconditionally given condition is present, no further concept of reason is needed for continuing the series, since understanding takes every downward step from condition to conditioned by itself.

B 394 Thus transcendental ideas serve only for ascending in the series of conditions to the unconditioned, to principles. With respect to descending to the conditioned, however, there is indeed a far-reaching logical use that our reason makes of the laws of understanding, but no transcendental use at all. If we form an idea of the absolute totality of such a synthesis (progressus), for example of the entire series of all future alterations of the world, this is a thought-entity (ens rationis), thought merely arbitrarily and not necessarily presupposed by reason. For the possibility of the conditioned presupposes the totality of its conditions, but not of its consequences. Consequently such a concept is no transcendental idea, with which alone we are concerned here.

Finally, we also become aware that a certain connection and unity shines forth among transcendental ideas themselves, and that through them pure reason brings all its cognitions into a system. To proceed from cognition of oneself (the soul) to cognition of the world, and through this to the original being, is so natural an advance. It seems similar to reason’s logical progress from the premises to the conclusion*.

B 395 Whether a kinship of the kind found between logical and transcendental procedure really underlies this in secret is also one of the questions whose answer must await the continuation of these investigations. Provisionally we have already achieved our purpose.

B 396 We have been able to extract the transcendental concepts of reason from the ambiguous position in which they are ordinarily mingled with others in philosophers’ theories, without those philosophers even properly distinguishing them from concepts of the understanding. We have been able to indicate their origin and thereby their determinate number, beyond which there can be no more, and present them in a systematic connection through which a special field is marked out and restricted for pure reason.

* Metaphysics has only three ideas as the proper aim of its inquiry: God, freedom, and immortality. The second concept, combined with the first, is to lead to the third as a necessary conclusion. Everything else this science concerns itself with serves merely as a means of arriving at these ideas and their reality. It needs them not for natural science, but to go beyond nature. Insight into them would make theology, morality, and through their combination religion, and hence the highest ends of our existence, depend solely on the speculative faculty of reason and on nothing else. In a systematic presentation of these ideas, the order given above, as the synthetic order, would be the most suitable. But in the treatment that must necessarily precede it, the analytic order, reversing this order, will be more appropriate to our purpose: to carry out our grand design by proceeding from what experience immediately supplies, the doctrine of the soul, to the doctrine of the world, and from there to cognition of God.

Transcendental Dialectic Second Book On the Dialectical Inferences of Pure Reason

One can say that the object of a mere transcendental idea is something of which we have no concept, although this idea has been generated quite necessarily in reason according to its original laws. For in fact no concept of the understanding is possible of an object that is to be adequate to reason’s demand: no concept that can be exhibited and made intuitive in a possible experience.

B 397 Yet one would express oneself better, and with less danger of misunderstanding, by saying that we can have no acquaintance with the object corresponding to an idea, though we can have a problematic concept of it.

Now at least the transcendental, subjective reality of pure concepts of reason rests on our being led to such ideas through a necessary syllogism. Thus there will be syllogisms containing no empirical premises, through which we infer from something we know to something else of which we nevertheless have no concept, yet to which we give objective reality through an unavoidable illusion. With respect to their result such inferences should therefore be called specious rather than rational inferences. With respect to their occasion, however, they may well bear the latter name, since they are neither invented nor accidentally arisen, but have sprung from the nature of reason. They are sophistries not of human beings but of pure reason itself. Even the wisest of all human beings cannot free himself from them, and although after much effort he may perhaps prevent the error, he can never entirely rid himself of the illusion that incessantly torments and mocks him.

There are therefore only three kinds of these dialectical syllogisms, as many as the ideas in which their conclusions terminate.

B 398 In the syllogism of the first class I infer from the transcendental concept of the subject, which contains no manifold, to the absolute unity of this subject itself, of which in this way I have no concept whatever. I will call this dialectical inference the transcendental paralogism. The second class of specious inferences is directed toward the transcendental concept of the absolute totality of the series of conditions for a given appearance in general. From always having a self-contradictory concept of the unconditioned synthetic unity of the series on one side, I infer the correctness of the opposite unity, of which I nevertheless also have no concept. I will call the state of reason in these dialectical inferences the antinomy of pure reason. Finally, according to the third kind of specious inference, I infer from the totality of conditions for thinking objects in general insofar as they can be given to me, to the absolute synthetic unity of all conditions of the possibility of things in general. That is, from things I do not know according to their mere transcendental concept, I infer a being of all beings that I know still less through a transcendent concept, and of whose unconditioned necessity I can form no concept. I will call this dialectical syllogism the ideal of pure reason.

B 399 Second Book of Transcendental Dialectic First Chapter On the Paralogisms of Pure Reason

A logical paralogism consists in the falsity of a syllogism as to its form, whatever its content may otherwise be. A transcendental paralogism, however, has a transcendental ground for inferring falsely as to form. In this way such a false inference will have its ground in the nature of human reason and carry with it an unavoidable, though not insoluble, illusion.

We now come to a concept not recorded above in the general list of transcendental concepts, yet which must be counted among them without thereby altering that table in the least or declaring it deficient. This is the concept, or, if one prefers, the judgment “I think.” But it is easy to see that it is the vehicle of all concepts in general, and hence also of transcendental concepts; it is therefore always included among them and is accordingly just as transcendental. Yet it can have no special heading, because it serves only to introduce all thinking as belonging to consciousness.

B 400 Meanwhile, however pure it is of the empirical, the impression of the senses, it nevertheless serves to distinguish two kinds of object from the nature of our power of representation. I, as thinking, am an object of inner sense and am called soul. What is an object of outer senses is called body. Thus the expression “I, as a thinking being” already signifies the object of psychology, which can be called the rational doctrine of the soul if I wish to know nothing further about the soul than what can be inferred, independently of all experience that determines me more closely and in concreto, from this concept “I” insofar as it occurs in all thinking.

The rational doctrine of the soul is indeed an undertaking of this kind. For if the slightest empirical element of my thinking, any particular perception of my inner state, were mixed among the grounds of cognition of this science, it would no longer be a rational but an empirical doctrine of the soul. We therefore already have before us a supposed science built on the single proposition “I think,” whose soundness or unsoundness we can very appropriately investigate here in accordance with the nature of a transcendental philosophy. One must not take offense that in this proposition, which expresses the perception of oneself, I nevertheless have an inner experience.

B 401 Nor must one take offense that rational psychology built on it is therefore never pure, but partly grounded on an empirical principle. For this inner perception is nothing further than mere apperception, “I think,” which even makes possible all transcendental concepts, in which we say “I think substance, cause,” and so on. For inner experience in general and its possibility, or perception in general and its relation to another perception, without any particular difference or determination being empirically given, cannot be regarded as empirical cognition. It must be regarded as cognition of the empirical in general, and belongs to the investigation of the possibility of every experience, which is certainly transcendental. The least object of perception, for example mere pleasure or displeasure, added to the universal representation of self-consciousness would immediately transform rational psychology into empirical psychology.

“I think” is therefore the sole text of rational psychology, from which it is to unfold all its wisdom. It is easy to see that this thought, if it is to be related to an object, myself, can contain nothing but transcendental predicates of it, because the slightest empirical predicate would spoil the science’s rational purity and independence from all experience.

B 402 Here, however, we need only follow the guiding thread of the categories. Since a thing, I as a thinking being, has first been given here, we will not alter the order of the categories among themselves as presented in their table, but will nevertheless begin here with the category of substance, through which a thing in itself is represented, and thus follow their series backward. The topics of the rational doctrine of the soul, from which everything else it may contain must be derived, are accordingly as follows:

1 The soul is substance.

2 In its quality, simple.

3 In the different times in which it exists, numerically identical, that is, unity, not plurality.

4 In relation to possible objects in space*.

* Readers who cannot readily discern from these expressions, in their transcendental abstraction, their psychological sense, or why the soul’s last attribute belongs to the category of existence, will find them sufficiently explained and justified in what follows. As for the Latin expressions that have slipped in instead of their German equivalents, contrary to good literary taste, both in this section and throughout the work, I offer this excuse: I preferred to sacrifice something of elegance in language rather than make academic use more difficult through the slightest unintelligibility.

B 403 From these elements all concepts of the pure doctrine of the soul arise merely through composition, without recognizing any other principle whatever. This substance, merely as an object of inner sense, gives the concept of immateriality; as simple substance, that of incorruptibility; its identity as an intellectual substance gives personality; all three together give spirituality. Relation to objects in space gives commerce with bodies; hence it represents thinking substance as the principle of life in matter, that is, as soul (anima) and as the ground of animality; this, restricted by spirituality, gives immortality.

To these there now relate four paralogisms of a transcendental doctrine of the soul, falsely regarded as a science of pure reason concerning the nature of our thinking being.

B 404 But we can ground it on nothing other than the simple representation, entirely empty of content in itself, “I.” Of this we cannot even say that it is a concept, but only a mere consciousness accompanying all concepts. Through this I, or he, or it, the thing that thinks, nothing further is represented than a transcendental subject of thoughts = x. This is cognized only through the thoughts that are its predicates, and apart from these we can never have the slightest concept of it. We therefore turn in a perpetual circle around it, since we must always already employ its representation to judge anything about it. This inconvenience is inseparable from it, because consciousness in itself is not so much a representation distinguishing a particular object as a form of representation in general insofar as it is to be called cognition. For of this alone can I say that through it I think something.

Yet at the very outset it must seem strange that the condition under which I think in general, which is therefore merely a constitution of my subject, should also hold for everything that thinks; and that we can presume to ground on a seemingly empirical proposition an apodictic and universal judgment, namely that everything thinking is constituted as the pronouncement of self-consciousness declares me to be.

B 405 The cause of this lies in our having necessarily to ascribe to things a priori all the properties constituting the conditions under which alone we think them. Now I can have the slightest representation of a thinking being through no outer experience, but solely through self-consciousness. Thus such objects are nothing further than the transfer of this consciousness of mine to other things, which are represented as thinking beings only in this way. But the proposition “I think” is here taken only problematically, not insofar as it may contain a perception of an existence (the Cartesian cogito, ergo sum), but according to its mere possibility, to see what properties may flow from this so simple proposition to its subject, whether such a subject exists or not.

If our pure rational cognition of thinking beings in general were grounded on more than the cogito, if we also enlisted observations of the play of our thoughts and the natural laws of the thinking self to be drawn from them, an empirical psychology would arise. This would be a kind of physiology of inner sense and might serve to explain its appearances, but could never disclose properties that do not belong to possible experience at all, such as that of the simple.

B 406 Nor could it teach apodictically anything concerning the nature of thinking beings in general. It would therefore be no rational psychology.

Now since the proposition “I think,” taken problematically, contains the form of every judgment of understanding in general and accompanies all categories as their vehicle, it is clear that inferences from it can contain a merely transcendental use of understanding, rejecting every admixture of experience. After what we have shown above, we can form no favorable expectation in advance of its progress. We will therefore follow it with a critical eye through all the predicaments of the pure doctrine of the soul, but, for brevity’s sake, we will let their examination proceed in an uninterrupted connection.

First, the following general remark may sharpen our attention to this kind of inference. I do not cognize any object merely by thinking. I can cognize an object only by determining a given intuition with regard to the unity of consciousness in which all thinking consists. Thus I do not cognize myself by being conscious of myself as thinking, but by being conscious of the intuition of myself as determined with respect to the function of thinking.

B 407 All modi of self-consciousness in thinking as such are therefore not yet concepts of the understanding concerning objects, categories, but mere functions that give thinking no object to cognize, and hence do not give myself as an object either. The object is not consciousness of the determining self, but only that of the determinable self, my inner intuition, insofar as its manifold can be combined in accordance with the universal condition of the unity of apperception in thinking.

1) In all judgments I am always the determining subject of the relation constituting the judgment. But that I, who think, must always count in thinking as subject, and as something that cannot be regarded as merely attaching to thought as a predicate, is an apodictic and indeed identical proposition. It does not, however, mean that I as object am for myself a self-subsistent being, or substance. The latter goes very far, and therefore requires data not found in thinking at all, perhaps, insofar as I consider merely the thinking being as such, more than I will ever find in it anywhere.

2) That the I of apperception, and consequently in every act of thinking, is a singular that cannot be resolved into a plurality of subjects, and thus designates a logically simple subject, already lies in the concept of thinking and is therefore an analytic proposition.

B 408 But this does not mean that the thinking I is a simple substance, which would be a synthetic proposition. The concept of substance always relates to intuitions that in my case can only be sensible, and thus lie entirely outside the field of understanding and its thinking, which is properly all that is being discussed here when we say that the I in thinking is simple. It would indeed be astonishing if what elsewhere requires so much preparation—to distinguish in what intuition presents what in it is substance, and still more whether this can also be simple, as with the parts of matter—were here given so directly in the poorest of all representations, as though through a revelation.

3) The proposition of my own identity through all the manifold of which I am conscious likewise lies in the concepts themselves and is therefore analytic. But this identity of the subject, of which I can become conscious in all its representations, does not concern the intuition of the subject through which it is given as an object. Hence it cannot signify personal identity, by which is understood consciousness of the identity of one’s own substance as a thinking being throughout all change of states.

B 409 To prove this, mere analysis of the proposition “I think” would not suffice; various synthetic judgments grounded on the given intuition would be required.

4) “I distinguish my own existence as a thinking being from other things outside me, including my body” is likewise an analytic proposition, for other things are those I think as distinct from myself. But whether this consciousness of myself is even possible without things outside me through which representations are given to me, and whether I can therefore exist merely as a thinking being without being human, I do not thereby know at all.

Thus through analysis of the consciousness of myself in thinking in general, not the slightest gain is made toward cognition of myself as an object. The logical exposition of thinking in general is falsely taken for a metaphysical determination of the object.

It would be a great, indeed the only, stumbling block against our entire critique if there were a possibility of proving a priori that all thinking beings are in themselves simple substances, and as such therefore, as a consequence of the same ground of proof, inseparably possess personality and are conscious of their existence apart from all matter. For in this way we would after all have taken a step beyond the world of sense and entered the field of noumena.

B 410 No one could then deny us the right to spread further within it, settle it, and take possession there as each person’s lucky star favored him. For the proposition “Every thinking being as such is a simple substance” is a synthetic a priori proposition, because, first, it goes beyond the concept laid at its foundation and adds a way of existing to thinking in general; and second, it adds to that concept a predicate, simplicity, that can be given in no experience whatever. Thus synthetic a priori propositions would be feasible and admissible not merely, as we have maintained, in relation to objects of possible experience, as principles of the possibility of this experience itself, but could also relate to things in general and in themselves. This conclusion would put an end to this entire critique and command us to leave matters as they were. But on closer examination the danger is not so great here.

In the procedure of rational psychology there prevails a paralogism represented by the following syllogism:

What cannot be thought otherwise than as subject also cannot exist otherwise than as subject, and is therefore substance.

B 411 Now a thinking being, considered merely as such, cannot be thought otherwise than as subject.

Therefore it also exists only as such, that is, as substance.

The major premise speaks of a being that can be thought in general in every respect, consequently also as it may be given in intuition. But the minor premise speaks of it only insofar as it considers itself as subject solely in relation to thinking and the unity of consciousness, not at the same time in relation to the intuition through which it is given to thinking as an object. Thus the conclusion is inferred per Sophisma figurae dictionis, and hence through a fallacy*.

* Thinking is taken in entirely different senses in the two premises: in the major premise as relating to an object in general, and hence as that object may be given in intuition; in the minor premise only as it consists in relation to self-consciousness. Here no object at all is therefore thought, but only the relation to oneself as subject, as the form of thinking, is represented. The first speaks of things that cannot be thought otherwise than as subjects; the second speaks not of things but of thinking, abstracting from every object, in which the I always serves as the subject of consciousness. Hence the conclusion cannot be “I cannot exist otherwise than as subject,” but only “In thinking my existence I can employ myself only as the subject of judgment.” This is an identical proposition disclosing absolutely nothing about the manner of my existence.

B 412 That this resolution of the celebrated argument into a paralogism is entirely correct becomes clear if one consults the general remark on the systematic presentation of principles and the section on noumena. There it was proved that the concept of a thing that can exist by itself as subject, but not as mere predicate, carries no objective reality at all: we cannot know whether any object whatever could correspond to it, since we do not understand the possibility of such a way of existing. Consequently it supplies absolutely no cognition. If, therefore, under the name of substance it is to indicate an object that can be given, if it is to become a cognition, a persistent intuition must be laid at its foundation as the indispensable condition of the objective reality of a concept, that through which alone the object is given.

B 413 But in inner intuition we have nothing persistent at all, for the I is only the consciousness of my thinking. Thus if we remain with thinking alone, we also lack the necessary condition for applying the concept of substance, of a self-subsistent subject, to ourselves as thinking beings. The simplicity of substance bound up with it entirely falls away together with the objective reality of this concept, and is transformed into a merely logical qualitative unity of self-consciousness in thinking in general, whether the subject is composite or not.

Refutation of Mendelssohn’s Proof of the Persistence of the Soul

This acute philosopher soon noticed an insufficiency in the ordinary argument intended to prove that the soul, if conceded to be a simple being, cannot cease to be through division. The argument was inadequate to secure its necessary continued existence, since one could still assume that its existence ceases through disappearance. In his Phaedo he now sought to ward off this perishability, which would be a genuine annihilation, by venturing to prove that a simple being cannot cease to be at all. For it cannot be diminished, and so cannot gradually lose something of its existence.

B 414 Nor can it be transformed gradually into nothing, since it has no parts and hence no plurality within itself. Thus no time would be found between an instant in which it is and another in which it no longer is, which is impossible.—But he did not consider that even if we concede this simple nature to the soul, its containing no manifold of parts outside one another and hence no extensive magnitude, we can no more deny it intensive magnitude than we can deny this to anything existing. That is, it has a degree of reality with respect to all its faculties, indeed everything constituting its existence. This degree can decrease through all the infinitely many smaller degrees, and thus the supposed substance, the thing whose persistence is not already otherwise established, can be transformed into nothing, though not through division, yet through the gradual relaxation (remissio) of its forces, and hence through elanguescence, if I may use this expression. For even consciousness always has a degree that can still be further diminished*, and consequently so does the faculty of being conscious of oneself.

B 415 The same holds of all the other faculties.—Thus the persistence of the soul, as a mere object of inner sense, remains unproved and even unprovable. Its persistence in life, when the thinking being as a human being is at the same time an object of outer senses to itself, is indeed clear in itself. But this by no means satisfies the rational psychologist, who undertakes to prove from mere concepts the soul’s absolute persistence even beyond life∗∗.

* Clarity is not, as logicians say, consciousness of a representation. For a certain degree of consciousness, insufficient for recollection, must be found even in many obscure representations, since without any consciousness at all we would make no distinction in combining obscure representations. Yet we can do this with the marks of many concepts, such as those of right and equity, and as a musician does when striking many notes together in improvisation. Rather, a representation is clear when consciousness in it is sufficient for consciousness of its difference from others. If consciousness is sufficient for distinction but not for consciousness of the difference, the representation would still have to be called obscure. Thus there are infinitely many degrees of consciousness down to its disappearance.

∗∗ Those who believe they have done enough to introduce a new possibility by defiantly claiming that no one can show a contradiction in their presuppositions can be greatly embarrassed by other possibilities no bolder in the least. This includes all those who believe they understand the possibility of thinking even after human life ceases, although they have an example of it only in empirical intuitions during human life. One such possibility is the division of a simple substance into several substances, and conversely the flowing together, coalition, of several into one simple substance. For although divisibility presupposes something composite, it does not necessarily require a composite of substances, but merely of degrees of the various faculties of one and the same substance. Just as we can think all the soul’s forces and faculties, even consciousness itself, diminished by half, yet so that substance still remains, we can also without contradiction represent this extinguished half as preserved, but not in the soul: outside it. Since everything in the soul that is real, and hence has a degree, and thus its entire existence without anything omitted, has here been halved, a separate substance would then arise outside it. For the plurality that was divided was already present, not as a plurality of substances, but of every reality as a quantum of existence in the soul. The unity of substance was only a way of existing that through this division alone could be transformed into a plurality of subsistence. But several simple substances could likewise flow together again into one, with nothing lost except the plurality of subsistence, since the one would contain the combined degree of reality of all the previous ones. Perhaps the simple substances that give us the appearance of matter might, through such dynamical division of the parents’ souls as intensive magnitudes, produce children’s souls—not, admittedly, through a mechanical or chemical influence on one another, but through an influence unknown to us, of which the former would be only the appearance—while the parents in turn replenished their loss through coalition with new material of the same kind. I am far from granting such fantasies the slightest worth or validity; moreover, the principles of the analytic above have sufficiently insisted that no use of the categories, such as that of substance, be made except in experience. But if the rationalist is bold enough to make a self-subsistent being out of the mere faculty of thinking, without any persistent intuition through which an object would be given, merely because the unity of apperception in thinking allows him no explanation from the composite—when he would do better to confess that he does not know how to explain the possibility of a thinking nature—why should the materialist not be entitled to equal boldness in using his principle for the opposite purpose while retaining the formal unity of the former, although he can no more cite experience in support of his possibilities?

B 416 Let us now take our propositions above in synthetic connection, as they must also be taken in rational psychology as a system valid for all thinking beings, and proceed from the category of relation.

B 417 We begin with the proposition that all thinking beings, as such, are substances. Going backward through their series until the circle closes, we finally encounter their existence. In this system they are not only conscious of their existence independently of outer things.

B 418 They can also determine it from themselves, with respect to the persistence that necessarily belongs to the character of substance. It follows, however, that idealism is unavoidable in that same rationalistic system, at least problematic idealism. If the existence of outer things is not at all required to determine one’s own existence in time, the former is also assumed quite gratuitously, without any proof of it ever being possible.

If, by contrast, we follow the analytic procedure, where “I think,” as a proposition already including an existence, is laid down as given, and hence modality is the foundation, and analyze it to cognize its content—whether and how this I determines its existence in space or time merely through that proposition—the propositions of rational psychology would begin not with the concept of a thinking being in general but with an actuality.

B 419 From the way this actuality is thought, after everything empirical in it has been separated off, what belongs to a thinking being in general would be inferred, as the following table shows.

1 I think,

2 as subject,

3 as simple subject,

4 as identical subject, in every state of my thinking.

Since the second proposition here does not determine whether I can exist and be thought only as subject and not also as predicate of another, the concept of a subject is taken merely logically, and it remains undetermined whether substance is to be understood by it or not. But in the third proposition the absolute unity of apperception, the simple I in the representation to which all combination or separation constituting thinking relates, becomes important in itself, even though I have not yet established anything about the subject’s constitution or subsistence. Apperception is something real, and its simplicity already lies in its possibility. Now in space there is nothing real that is simple, for points, which constitute the only simple things in space, are mere boundaries, not themselves anything serving to constitute space as a part.

B 420 Thus it follows that explaining my constitution as a merely thinking subject on materialist grounds is impossible. But since my existence is regarded in the first proposition as given—it does not say “Every thinking being exists,” which would assert absolute necessity of them and hence too much, but only “I exist thinking”—the proposition is empirical and contains the determinability of my existence solely with respect to my representations in time. But for this I again first need something persistent, of a kind not given to me at all in inner intuition insofar as I think myself. Hence the way I exist, whether as substance or accident, cannot be determined through this simple self-consciousness at all. Thus if materialism is unfit as a way of explaining my existence, spiritualism is equally insufficient for it. The conclusion is that in no way whatever can we cognize anything about the constitution of our soul concerning the possibility of its separate existence in general.

And how could it be possible to go beyond experience, our existence in life, through the unity of consciousness, which we ourselves know only because we indispensably need it for the possibility of experience?

B 421 How could it even be possible to extend our cognition to the nature of all thinking beings in general through the empirical proposition “I think,” which is undetermined with respect to every kind of intuition?

Thus there is no rational psychology as a doctrine supplying an addition to our self-cognition, but only as a discipline setting impassable boundaries to speculative reason in this field. On the one hand, it keeps us from throwing ourselves into the arms of soulless materialism; on the other, from losing ourselves in fanciful wanderings through a spiritualism groundless for us in life. Rather, it reminds us to regard reason’s refusal to give satisfying answers to inquisitive questions reaching beyond this life as its indication that we should turn our self-cognition from fruitless, extravagant speculation to fruitful practical use. Though this use too is always directed only toward objects of experience, it draws its principles from higher sources and determines conduct as though our vocation reached infinitely far beyond experience, and hence beyond this life.

From all this we see that a mere misunderstanding gives rise to rational psychology. The unity of consciousness underlying the categories is here taken for an intuition of the subject as object, and the category of substance is applied to it.

B 422 But it is only unity in thinking, through which alone no object is given. The category of substance, which always presupposes a given intuition, therefore cannot be applied to it, and this subject cannot be cognized at all. Thus the subject of the categories cannot, by thinking them, obtain a concept of itself as an object of the categories. For in order to think them it must lay at their foundation its pure self-consciousness, which was precisely what was to be explained. Likewise the subject in which the representation of time originally has its ground cannot thereby determine its own existence in time; and if the latter cannot occur, neither can the former occur through categories as a determination of itself as a thinking being in general*.

* As already stated, “I think” is an empirical proposition and contains within itself the proposition “I exist.” But I cannot say “Everything that thinks exists,” for then the property of thinking would make all beings possessing it necessary beings. Hence my existence also cannot be regarded as inferred from the proposition “I think,” as Descartes held, since otherwise the major premise “Everything that thinks exists” would have to precede it; rather, my existence is identical with it. The proposition expresses an indeterminate empirical intuition, perception, and thus does prove that sensation, which consequently belongs to sensibility, already underlies this existential proposition. But it precedes experience, which is to determine the object of perception through the category with respect to time. Existence here is not yet a category, since a category relates not to an indeterminately given object but only to one of which we have a concept and wish to know whether it is also posited outside this concept or not. An indeterminate perception here signifies only something real that has been given, and indeed only for thinking in general, thus neither as appearance nor as a thing in itself (noumenon), but as something that in fact exists and is designated as such in the proposition “I think.” For it should be noted that, when I called “I think” an empirical proposition, I did not mean that the I in this proposition is an empirical representation. Rather, it is purely intellectual because it belongs to thinking in general. Yet without some empirical representation supplying the material for thinking, the act “I think” would not take place; and the empirical is only the condition of the application or use of the pure intellectual faculty.

* * *

B 423 Thus a cognition attempted beyond the boundaries of possible experience, yet belonging to humanity’s highest interest, dissolves into disappointed expectation insofar as it is to be owed to speculative philosophy.

B 424 Nevertheless, the rigor of critique, by proving at the same time the impossibility of dogmatically establishing anything about an object of experience beyond the boundary of experience, renders reason a service of no small importance in this interest: it also secures reason against every possible assertion of the contrary. This can be accomplished only by either proving one’s proposition apodictically or, if that fails, seeking the sources of this inability. If these lie in the necessary limits of our reason, they must then subject every opponent to precisely the same law of renouncing all claims to dogmatic assertion.

Nevertheless, not the slightest thing is thereby lost for the right, indeed the necessity, of assuming a future life according to principles of the practical use of reason combined with its speculative use. For the merely speculative proof could never in any case have had any influence on common human reason. It is so poised on a hair’s breadth that even the schools can keep it there only as long as they make it spin incessantly around itself like a top; thus even in their own eyes it supplies no persistent foundation on which anything could be built.

B 425 All proofs usable for the world retain their undiminished worth and indeed gain clarity and unaffected conviction through the removal of those dogmatic pretensions. They place reason in its proper domain, the order of ends, which is nevertheless also an order of nature. But reason at the same time, as a practical faculty in itself, without being restricted to the conditions of nature’s order, is entitled to extend the order of ends, and with it our own existence, beyond the boundaries of experience and life. Judging by analogy with the nature of living beings in this world—where reason must necessarily adopt as a principle that no organ, no faculty, no impulse, hence nothing dispensable or disproportionate to its use, and thus purposeless, is to be found, but everything is precisely suited to its vocation in life—the human being, who alone can contain within himself the ultimate final end of all this, would have to be the only creature exempted from it. For his natural predispositions, not merely his talents and impulses to use them, but above all the moral law within him, go so far beyond every use and advantage he could draw from them in this life that the latter even teaches him to value above everything the mere consciousness of integrity of disposition, in the absence of all advantages and even of the mere shadow of posthumous fame.

B 426 He feels inwardly called to make himself fit, through his conduct in this world and the renunciation of many advantages, to be a citizen of a better world that he has in idea. This powerful ground of proof, never to be refuted, remains, accompanied by an incessantly increasing cognition of purposiveness in everything we see before us and a prospect into the immensity of creation, hence also by consciousness of a certain boundlessness in the possible extension of our cognitions together with a corresponding impulse, even though we must abandon the attempt to understand the necessary continuance of our existence from merely theoretical cognition of ourselves.

Conclusion of the Resolution of the Psychological Paralogism

The dialectical illusion in rational psychology rests on confusing an idea of reason (a pure intelligence) with the concept, indeterminate in every respect, of a thinking being in general. I think myself for the purpose of a possible experience while still abstracting from all actual experience, and infer from this that I can become conscious of my existence even outside experience and its empirical conditions.

B 427 Thus I confuse the possible abstraction from my empirically determined existence with the supposed consciousness of a separately possible existence of my thinking self, and believe I cognize what is substantial in me as the transcendental subject, when what I have in mind is merely the unity of consciousness that underlies all determination as the mere form of cognition.

The task of explaining the communion of the soul with the body does not properly belong to the psychology under discussion here, because its intention is to prove the personality of the soul even outside this communion (after death), and it is therefore transcendent in the strict sense, although it concerns an object of experience—but only insofar as that object ceases to be an object of experience. Yet our doctrine can also provide an adequate answer to this question. As is well known, the difficulty that prompted the task lies in the presumed heterogeneity of the object of inner sense (the soul) and the objects of the outer senses: time alone attaches to the former as the formal condition of its intuition, while space attaches to the latter as well. But consider that these two kinds of objects differ in this respect not intrinsically, but only insofar as one appears externally to the other.

B 428 Then what underlies the appearance of matter, as a thing in itself, may perhaps not be so heterogeneous, and this difficulty disappears. No other remains than the question of how a communion of substances is possible at all. Its solution lies entirely outside the field of psychology and, as the reader will readily judge from what the Analytic said about fundamental powers and faculties, undoubtedly outside the field of all human cognition as well.

General Remark Concerning the Transition from Rational Psychology to Cosmology

The proposition “I think,” or “I exist thinking,” is an empirical proposition. But such a proposition rests on empirical intuition, and consequently on the object thought as appearance. It therefore seems that, according to our theory, the soul would be transformed entirely into appearance, even in thinking, and that in this way our consciousness itself, as mere illusion, would in fact refer to nothing.

Thinking, taken by itself, is merely the logical function, and thus sheer spontaneity of the combination of the manifold of a merely possible intuition, and by no means presents the subject of consciousness as appearance.

B 429 This is simply because it takes no account at all of the kind of intuition, whether sensible or intellectual. Through it I represent myself neither as I am nor as I appear to myself; I merely think myself as any object in general, abstracting from its manner of intuition. If I represent myself here as the subject of thoughts, or as the ground of thinking, these modes of representation do not signify the categories of substance or cause, for these are those functions of thinking (judging) already applied to our sensible intuition, which would certainly be required if I wished to cognize myself. But now I wish only to become conscious of myself as thinking; I set aside how my own self is given in intuition. In that intuition it might be mere appearance to me, who think, though not insofar as I think. In consciousness of myself in mere thinking, I am the being itself, though admittedly nothing about it is thereby yet given to me for thought.

But the proposition “I think,” insofar as it means “I exist thinking,” is not a mere logical function. It determines the subject (which is then also an object) with respect to existence, and cannot occur without inner sense, whose intuition always provides the object not as a thing in itself, but merely as appearance.

B 430 Thus it no longer contains mere spontaneity of thinking, but also receptivity of intuition: that is, the thinking of myself applied to the empirical intuition of that same subject. In this latter intuition the thinking self would now have to seek the conditions for using its logical functions as categories of substance, cause, and so on, in order not merely to designate itself as an object in itself through the “I,” but also to determine the manner of its existence—that is, to cognize itself as a noumenon. This, however, is impossible, since inner empirical intuition is sensible and provides nothing but data of appearance. These furnish the object of pure consciousness with nothing toward knowledge of its separate existence, but can serve only the purposes of experience.

Suppose, however, that in what follows we found occasion—not in experience, but in certain (not merely logical rules, but) laws of the pure use of reason, established a priori and concerning our existence—to presuppose ourselves as legislating wholly a priori with regard to our own existence, and as ourselves determining that existence. This would reveal a spontaneity through which our actuality could be determined without requiring the conditions of empirical intuition. Here we would become aware that consciousness of our existence contains something a priori.

B 431 This can serve to determine our existence, which is thoroughly determinable only sensibly, nevertheless in respect of a certain inner faculty, in relation to an intelligible world (admittedly one that is only thought).

Nonetheless, this would not advance any of the attempts of rational psychology in the least. For through that admirable faculty, first disclosed to me by consciousness of the moral law, I would indeed have a purely intellectual principle for determining my existence—but through what predicates? Through none other than those that must be given to me in sensible intuition. Thus I would again arrive where I was in rational psychology: needing sensible intuitions to give meaning to my concepts of the understanding—substance, cause, and so on—through which alone I can have cognition of myself. But those intuitions can never carry me beyond the field of experience. Yet with regard to practical use, which is always directed toward objects of experience, I would be entitled to apply these concepts to freedom and its subject in a meaning analogous to their theoretical use. By them I understand merely the logical functions of subject and predicate, ground and consequence.

B 432 In accordance with these, actions or effects are determined under those laws so that, together with the laws of nature, they can always be explained in conformity with the categories of substance and cause, although they arise from an entirely different principle. This has been said only to prevent the misunderstanding to which the doctrine of our self-intuition as appearance is easily exposed. We will have occasion to make use of it in what follows.

Transcendental Dialectic Book Two

Chapter Two The Antinomy of Pure Reason

In the introduction to this part of our work, we showed that all transcendental illusion of pure reason rests on dialectical inferences, whose schema logic supplies in the three formal kinds of syllogism in general, just as the categories find their logical schema in the four functions of all judgments. The first kind of these sophistical inferences concerned the unconditioned unity of the subjective conditions of all representations in general (the subject or the soul), corresponding to categorical syllogisms, whose major premise, as principle, asserts the relation of a predicate to a subject.

B 433 The second kind of dialectical argument will therefore, by analogy with hypothetical syllogisms, take as its content the unconditioned unity of objective conditions in appearance, just as the third kind, which will occur in the following chapter, has as its theme the unconditioned unity of the objective conditions of the possibility of objects in general.

It is noteworthy, however, that the transcendental paralogism produced a merely one-sided illusion with regard to the idea of the subject of our thinking, and that not the slightest illusion arising from concepts of reason can be found in support of the contrary assertion. The advantage lies entirely on the side of pneumatism, although it cannot disown its inherited defect: despite all the illusion in its favor, it dissolves into mere vapor in the trial by fire of critique.

The outcome is entirely different when we apply reason to the objective synthesis of appearances. Here reason thinks it can establish its principle of unconditioned unity with much illusion, but soon becomes entangled in contradictions that force it, so far as its cosmological aim is concerned, to abandon its demand.

Here a new phenomenon of human reason appears: an entirely natural antithetic, which no one needs to contrive through subtle speculation or by artfully laying snares.

B 434 Reason falls into it of itself, and indeed unavoidably. It is thereby protected from the slumber of an imagined conviction produced by a merely one-sided illusion, but at the same time tempted either to surrender to skeptical hopelessness or to assume a dogmatic defiance, obstinately insisting on certain assertions without giving the grounds for the contrary a fair hearing. Either is the death of sound philosophy, although the former might, if necessary, still be called the euthanasia of pure reason.

Before presenting the scenes of division and disruption brought about by this conflict of laws (antinomy) of pure reason, we will offer certain discussions that may explain and justify the method we employ in treating our subject. I call all transcendental ideas, insofar as they concern absolute totality in the synthesis of appearances, world-concepts. This is partly because of that very unconditioned totality, on which the concept of the world-whole also rests—a concept that is itself only an idea—and partly because they concern solely the synthesis of appearances, hence empirical synthesis. By contrast, there is the absolute totality in the synthesis of the conditions of all possible things in general.

B 435 This will give rise to an ideal of pure reason, entirely distinct from the world-concept, though related to it. Accordingly, just as the paralogisms of pure reason laid the foundation for a dialectical psychology, the antinomy of pure reason will display the transcendental principles of a supposed pure (rational) cosmology. It does so not to establish their validity and adopt them, but, as the very name of a conflict of reason already indicates, to present them in their dazzling but false illusion as an idea that cannot be reconciled with appearances.

The Antinomy of Pure Reason Section One System of Cosmological Ideas

To enumerate these ideas according to a principle with systematic precision, we must first note that pure and transcendental concepts can arise only from the understanding. Reason does not, strictly speaking, generate any concept at all; at most it frees the concept of the understanding from the unavoidable limitations of a possible experience, and thus seeks to extend it beyond the boundaries of the empirical, yet in connection with it.

B 436 It does so by demanding absolute totality on the side of the conditions for a given conditioned (under which the understanding subjects all appearances to synthetic unity), thereby making the category into a transcendental idea, in order to give empirical synthesis absolute completeness by continuing it to the unconditioned (which is never found in experience, but only in the idea). Reason demands this according to the principle: if the conditioned is given, then the whole sum of conditions, and thus the absolutely unconditioned through which alone the conditioned was possible, is also given. First, then, transcendental ideas will properly be nothing but categories extended to the unconditioned, and can be arranged in a table ordered under the headings of those categories. Second, however, not all categories will be suited to this, but only those in which the synthesis constitutes a series, specifically of conditions subordinated to one another (not coordinated) for a conditioned. Reason demands absolute totality only insofar as it concerns the ascending series of conditions for a given conditioned, not when the question concerns the descending line of consequences, or the aggregate of coordinated conditions for those consequences.

B 437 For conditions are already presupposed in relation to the given conditioned and must be regarded as given along with it, whereas consequences do not make their conditions possible but rather presuppose them. Hence in proceeding to the consequences (or descending from the given condition to the conditioned), we need not concern ourselves with whether the series comes to an end or not, and the question of its totality is in general no presupposition of reason at all.

Thus we necessarily think a time that has wholly elapsed up to the given moment as also given (although not determinable by us). But as for future time, since it is not a condition of arriving at the present, it is entirely immaterial to our comprehension of the present how we choose to regard the future: whether we let it end somewhere or run on to infinity. Let there be a series m, n, o, in which n is given as conditioned in relation to m, but also as the condition of o. Let the series ascend from the conditioned n to m (l, k, i, etc.), and likewise descend from the condition n to the conditioned o (p, q, r, etc.). Then I must presuppose the former series in order to regard n as given, and according to reason (the totality of conditions), n is possible only through that series. Its possibility does not, however, rest on the following series o, p, q, r.

B 438 This series can therefore be regarded not as given, but only as dabilis.

I will call the synthesis of a series on the side of the conditions, beginning with the one nearest the given appearance and proceeding to more remote conditions, regressive synthesis; the synthesis that proceeds on the side of the conditioned, from the nearest consequence to more remote ones, I will call progressive synthesis. The former proceeds to antecedentia, the latter to consequentia. Cosmological ideas therefore concern the totality of regressive synthesis and proceed to antecedentia, not to consequentia. When the latter occurs, it is an arbitrary and not a necessary problem of pure reason, because to make what is given in appearance completely comprehensible we need its grounds, but not its consequences.

To arrange the table of ideas according to the table of categories, we first take the two original quanta of all our intuition, time and space. Time is in itself a series (and the formal condition of all series), and therefore, in relation to a given present, the antecedentia as conditions (the past) can be distinguished a priori from the consequentibus (the future). Consequently, the transcendental idea of the absolute totality of the series of conditions for a given conditioned concerns only all past time.

B 439 According to the idea of reason, the whole elapsed time, as the condition of the given moment, is necessarily thought as given. As for space, however, in itself it contains no distinction between progression and regression, because it constitutes an aggregate, not a series, since its parts are all simultaneous. I could regard the present moment, in relation to past time, only as conditioned, never as its condition, because this moment arises only through elapsed time (or rather through the elapsing of the preceding time). But since the parts of space are not subordinated to one another but coordinated, one part is not the condition of the possibility of another, and space does not in itself constitute a series as time does. Yet the synthesis of the manifold parts of space, through which we apprehend it, is successive, and thus occurs in time and contains a series. And since, in this series of aggregated spaces (for example, the feet in a rod), the spaces thought as added beyond a given one are always the condition of the boundary of the preceding ones, measuring a space must also be regarded as a synthesis of a series of conditions for a given conditioned. The side of the conditions, however, is not in itself distinguished from the side toward which the conditioned lies.

B 440 Consequently regression and progression in space seem to be one and the same. Yet because one part of space is not given through another, but only bounded by it, we must also regard every bounded space as conditioned insofar as it presupposes another space as the condition of its boundary, and so onward. With regard to bounding, then, proceeding in space is also a regression, and the transcendental idea of the absolute totality of synthesis in the series of conditions applies to space as well. I can ask about the absolute totality of appearance in space just as well as about its totality in elapsed time. Whether an answer to either question is possible at all will be determined later.

Second, reality in space—that is, matter—is a conditioned whose inner conditions are its parts, while the parts of those parts are its more remote conditions. Here, therefore, a regressive synthesis takes place whose absolute totality reason demands. This can occur only through a completed division in which the reality of matter vanishes either into nothing or into what is no longer matter, namely the simple. Here, too, then, there is a series of conditions and an advance toward the unconditioned.

B 441 Third, as regards the categories of real relation among appearances, the category of substance with its accidents is unsuited to a transcendental idea; that is, reason has no ground for proceeding regressively to conditions in its case. For accidents, insofar as they inhere in a single substance, are coordinated with one another and do not constitute a series. In relation to substance, however, they are not properly subordinated to it, but are the way the substance itself exists. What might still seem to be an idea of transcendental reason here would be the concept of the substantial. But this means nothing other than the concept of an object in general that subsists, insofar as we think in it merely the transcendental subject without any predicates. Since we are concerned here only with the unconditioned in the series of appearances, it is clear that the substantial cannot constitute a member of that series. The same holds for substances in communion: they are mere aggregates and have no exponent of a series, since they are not subordinated to one another as conditions of their possibility—which one could indeed say of spaces, whose boundary was never determined in itself, but always through another space. Thus only the category of causality remains. It presents a series of causes for a given effect.

B 442 In this series we can ascend from the latter, as the conditioned, to the former, as conditions, and answer reason’s question.

Fourth, the concepts of the possible, the actual, and the necessary lead to no series except insofar as what is contingent in existence must always be regarded as conditioned and, according to the rule of the understanding, points to a condition under which it is necessary; this condition points to a higher one, until reason finds unconditioned necessity only in the totality of this series.

Accordingly, there are only four cosmological ideas, corresponding to the four headings of the categories, if we select those that necessarily involve a series in the synthesis of the manifold.

B 443 1. Absolute completeness of the composition of the given whole of all appearances

2 Absolute completeness of the division of a given whole in appearance

3 Absolute completeness of the origination of an appearance

4 Absolute completeness of the dependence of the existence of what is alterable in appearance

First, we must note here that the idea of absolute totality concerns nothing but the exposition of appearances, not the pure concept of the understanding of a whole of things in general. Here, then, appearances are regarded as given, and reason demands the absolute completeness of the conditions of their possibility insofar as these constitute a series: hence an absolutely (that is, in every respect) complete synthesis through which the appearance can be expounded according to laws of the understanding.

Second, it is properly only the unconditioned that reason seeks.

B 444 It seeks it in this synthesis of conditions, continued serially and indeed regressively: completeness, as it were, in the series of premises that together presuppose no others. Now this unconditioned is always contained in the absolute totality of the series when we represent it in imagination. But this absolutely completed synthesis is itself only an idea; for we cannot know, at least in advance, whether such a synthesis is possible in the case of appearances. If we represent everything through mere pure concepts of the understanding, without conditions of sensible intuition, we can say outright that with a given conditioned the whole series of conditions subordinated to one another is also given, for the former is given only through the latter. But with appearances there is a special restriction on the manner in which conditions are given: namely, through the successive synthesis of the manifold of intuition, which is supposed to be complete in the regression. Whether this completeness is sensibly possible remains a problem. Yet the idea of this completeness lies in reason, regardless of the possibility or impossibility of connecting empirical concepts in a way adequate to it.

B 445 Thus, since the unconditioned is necessarily contained in the absolute totality of the regressive synthesis of the manifold in appearance (under the guidance of the categories, which represent it as a series of conditions for a given conditioned)—even if one leaves unsettled whether and how this totality can be brought about—reason here takes the path of beginning with the idea of totality, although its ultimate aim is properly the unconditioned, whether of the entire series or of a part of it.

We can now think this unconditioned either as consisting solely in the whole series, in which all members without exception would be conditioned and only their whole would be absolutely unconditioned—in which case the regression is called infinite—or as the absolutely unconditioned being only a part of the series, to which its remaining members are subordinated but which itself stands under no other condition.* In the first case, the series a parte priori is without boundaries (without a beginning), that is, infinite, and yet given in its entirety; regression within it, however, is never completed and can be called infinite only potentialiter.

B 446 In the second case, there is a first member of the series, called, with respect to elapsed time, the beginning of the world; with respect to space, the boundary of the world; with respect to the parts of a whole given within its boundaries, the simple; with respect to causes, absolute self-activity (freedom); and with respect to the existence of alterable things, absolute natural necessity.

* The absolute whole of the series of conditions for a given conditioned is always unconditioned, because outside it there are no further conditions in relation to which it could be conditioned. But this absolute whole of such a series is only an idea, or rather a problematic concept whose possibility must be investigated, specifically in relation to the way the unconditioned, as the transcendental idea properly at issue, may be contained in it.

We have two expressions, “world” and “nature,” which sometimes overlap. The first signifies the mathematical whole of all appearances and the totality of their synthesis, both in the large and in the small—that is, both in the advance of this synthesis through composition and through division. But this same world is called nature* insofar as it is considered as a dynamical whole, and attention is directed not to aggregation in space or time.

B 447 Aggregation serves to produce it as a magnitude; attention goes instead to unity in the existence of appearances. Here the condition of what happens is called the cause; the unconditioned causality of the cause in appearance is called freedom, while conditioned causality is called a natural cause in the narrower sense. What is conditioned in existence generally is called contingent, and what is unconditioned, necessary. The unconditioned necessity of appearances can be called natural necessity.

* Nature, taken adjectivally (formaliter), signifies the connection of the determinations of a thing according to an inner principle of causality. By nature taken substantivally (materialiter), by contrast, we mean the sum total of appearances insofar as they are thoroughly connected through an inner principle of causality. In the first sense we speak of the nature of fluid matter, of fire, etc., and use this word adjectivally; by contrast, when we speak of the things of nature, we have an existing whole in mind.

I called the ideas with which we are now concerned cosmological ideas above, partly because “world” means the sum total of all appearances, and our ideas, too, are directed only toward the unconditioned among appearances; partly also because the word “world,” in the transcendental sense, signifies the absolute totality of the sum of existing things, and we direct our attention solely to completeness of synthesis (though properly only in the regression to the conditions). Moreover, all these ideas are transcendent. Although they do not go beyond their object—appearances—in kind, and concern only the sensible world, not noumena, they nevertheless carry synthesis to a degree that exceeds all possible experience. In view of this, I think they can all quite appropriately be called world-concepts.

B 448 With regard to the distinction between the mathematically and the dynamically unconditioned, at which the regression aims, I would nevertheless call the first two world-concepts in a narrower sense (of the world in the large and in the small), and the other two transcendent concepts of nature. This distinction is not yet particularly significant, but may become more important as we proceed.

The Antinomy of Pure Reason Section Two The Antithetic of Pure Reason

If a thetic is any body of dogmatic doctrines, then by an antithetic I mean not dogmatic assertions of the opposite, but the conflict of seemingly dogmatic cognitions (thesin cum antithesi), without attributing to either a superior claim to acceptance. An antithetic, then, is not concerned at all with one-sided assertions, but considers universal cognitions of reason solely in terms of their conflict with one another and its causes. The transcendental antithetic is an investigation of the antinomy of pure reason, its causes, and its result.

B 449 If we employ our reason not merely on objects of experience, for the use of principles of the understanding, but venture to extend those principles beyond the boundary of experience, sophistical propositions arise that may neither hope for confirmation in experience nor fear refutation by it. Each is not only free of contradiction in itself but even finds conditions of its necessity in the nature of reason. Unfortunately, however, its opposite has equally valid and necessary grounds for assertion on its side.

The questions that naturally arise in such a dialectic of pure reason are therefore: 1. In which propositions, then, pure reason is in fact inevitably subject to an antinomy. 2. On what causes this antinomy rests. 3. Whether, and in what way, a path to certainty nevertheless remains open to reason under this contradiction.

A dialectical proposition of pure reason must therefore have this characteristic, distinguishing it from all sophistical propositions: it concerns not an arbitrary question raised merely for some chosen purpose, but one that every human reason must necessarily encounter in its advance. Second, together with its opposite, it must carry not merely an artificial illusion that disappears as soon as it is understood, but a natural and unavoidable illusion.

B 450 Even when one is no longer taken in by it, this illusion still deludes, though it does not deceive, and thus can be rendered harmless but never eradicated.

Such a dialectical doctrine will refer not to the unity of the understanding in concepts of experience, but to the unity of reason in mere ideas. Since the conditions of these ideas are to agree, first, as a synthesis according to rules, with the understanding, and yet at the same time, as the absolute unity of that synthesis, with reason, they will be too great for the understanding if they are adequate to the unity of reason, and too small for reason if they are suited to the understanding. From this a conflict must arise that cannot be avoided, whatever approach we take.

These sophistical assertions thus open a dialectical arena where whichever side is permitted to attack gains the upper hand, while the side forced to remain on the defensive is certain to succumb. Hence even valiant knights, whether championing a good or a bad cause, are sure of carrying off the victor’s wreath if they only secure the privilege of making the last attack and are not obliged to withstand a renewed assault from their opponent. We can readily imagine that this arena has been entered often enough from time immemorial, and that many victories have been won on both sides.

B 451 We can imagine, too, that for the final victory, which decided the matter, care was always taken that the champion of the good cause alone held the field, by forbidding his opponent to take up arms again. As impartial judges, we must set entirely aside whether the cause for which the contestants fight is good or bad, and first let them settle their dispute between themselves. Perhaps, after exhausting rather than injuring one another, they will recognize the futility of their quarrel for themselves and part as good friends.

This method of watching a conflict of assertions—or rather of provoking it ourselves—not to decide finally in favor of one side or the other, but to investigate whether its object may not be a mere mirage, which each pursues in vain and from which neither could gain anything even if no resistance were offered: this procedure, I say, may be called the skeptical method. It is entirely distinct from skepticism, a principle of artful and scientific ignorance that undermines the foundations of all cognition in order, if possible, to leave it nowhere reliable or secure. For the skeptical method aims at certainty.

B 452 It does so, in a dispute honestly intended on both sides and intelligently conducted, by seeking to discover the point of misunderstanding, in order, as wise legislators do, to draw instruction for itself from the perplexity of judges in lawsuits, concerning what is defective and insufficiently determined in their laws. Given our limited wisdom, the antinomy disclosed in applying laws is the best test of legislation, making reason, which does not easily notice its missteps in abstract speculation, attentive to the factors involved in determining its principles.

But this skeptical method is essentially peculiar to transcendental philosophy alone; it may be dispensed with in every other field of investigation, but not in this one. Its use in mathematics would be absurd, since false assertions cannot hide there and become invisible: proofs must always proceed along the thread of pure intuition, through a synthesis that is evident at every step. In experimental philosophy a doubt that postpones judgment may indeed be useful; at least, however, no misunderstanding is possible that could not readily be removed, and experience must ultimately contain the final means of deciding the dispute, whether these are discovered sooner or later.

B 453 Morality, too, can present all its principles in concreto, together with their practical consequences, at least in possible experiences, and thereby avoid the misunderstanding of abstraction. Transcendental assertions, by contrast, presume insights extending even beyond the field of all possible experiences. Their abstract synthesis can neither be given in any intuition a priori, nor can misunderstanding be discovered through any experience. Transcendental reason therefore admits no other touchstone than the attempt to reconcile its assertions with one another, and consequently, beforehand, their free and unhindered contest with one another. This is what we will now undertake.*

* The antinomies follow one another in the order of the transcendental ideas listed above.

The Antinomy of Pure Reason

B 454 First Conflict of the Transcendental Ideas

Thesis

The world has a beginning in time and is also enclosed within boundaries in space.

Proof

For suppose the world has no beginning in time. Then an eternity has elapsed up to every given point in time, and consequently an infinite series of successive states of things in the world has passed. But the infinity of a series consists precisely in its never being capable of completion through successive synthesis. An infinite elapsed world-series is therefore impossible; hence a beginning of the world is a necessary condition of its existence. This was the first point to be proved.

As for the second point, again assume the opposite. Then the world will be an infinite given whole of simultaneously existing things. Now consider the magnitude of a quantum that is not given to any intuition within certain boundaries*.

B 456 We can think it in no other way than through the synthesis of its parts, and the totality of such a quantum only through completed synthesis, or through the repeated addition of the unit to itself.∗∗ Accordingly, to think the world that fills all spaces as a whole, the successive synthesis of the parts of an infinite world would have to be regarded as completed; that is, an infinite time would have to be regarded as having elapsed in enumerating all coexisting things. This is impossible. Consequently, an infinite aggregate of actual things cannot be regarded as a given whole, and thus not as given simultaneously. A world is therefore not infinite in spatial extension, but enclosed within its boundaries. This was the second point.

* We can intuit an indeterminate quantum as a whole if it is enclosed within boundaries, without having to construct its totality through measurement, that is, the successive synthesis of its parts. For the boundaries already determine completeness by cutting off everything more.

∗∗ The concept of totality in this case is nothing other than the representation of the completed synthesis of its parts, because, since we cannot derive the concept from intuition of the whole (which in this case is impossible), we can grasp it only through the synthesis of the parts up to completion of the infinite, at least in the idea.

B 455 Antithesis

The world has no beginning and no boundaries in space, but is infinite with regard to both time and space.

Proof

For suppose it has a beginning. Since a beginning is an existence preceded by a time in which the thing does not exist, there must have preceded a time in which the world did not exist, that is, an empty time. But in an empty time no thing can arise, because no part of such a time, compared with another, contains any distinguishing condition of existence rather than nonexistence (whether we assume that the thing arises of itself or through another cause). Thus although many a series of things can begin in the world, the world itself can have no beginning, and is therefore infinite with regard to past time.

As for the second point, first assume the opposite: that the world is spatially finite and bounded. It then exists in an empty space that is not bounded. There would therefore be not only a relation of things in space, but also of things to space. Now the world is an absolute whole.

B 457 Outside it no object of intuition, and thus no correlate of the world to which it could stand in relation, is found, so the relation of the world to empty space would be its relation to no object. But such a relation, and therefore the bounding of the world by empty space, is nothing. Thus the world is not spatially bounded at all; that is, it is infinite in extension.*

* Space is merely the form of outer intuition (formal intuition), not an actual object that can be intuited externally. Space prior to all things that determine it (fill or bound it), or rather that give an empirical intuition conforming to its form, is, under the name of absolute space, nothing other than the mere possibility of outer appearances insofar as they either exist in themselves or can be added to given appearances. Empirical intuition is therefore not composed of appearances and space (perception and empty intuition). One is not the other’s correlate in synthesis; rather, they are combined only in one and the same empirical intuition as its matter and form. If we try to place one of these two components outside the other (space outside all appearances), all sorts of empty determinations of outer intuition result, which are nevertheless not possible perceptions: for example, the motion or rest of the world in infinite empty space, a determination of their mutual relation that can never be perceived and is therefore also the predicate of a mere thought-entity.

B 458 Remark on the First Antinomy I. On the Thesis

In these conflicting arguments I have not sought deceptive tricks in order perhaps (as one says) to offer an advocate’s proof, exploiting an opponent’s carelessness and willingly allowing his appeal to a misunderstood law so as to build one’s own unlawful claims on its refutation. Each of these proofs has been drawn from the nature of the matter, setting aside the advantage that the fallacies of dogmatists on either side might afford us.

I could also have seemingly proved the thesis by first introducing a faulty concept of the infinity of a given magnitude, as dogmatists habitually do. An infinite magnitude would be one than which no greater is possible (that is, no greater than the multitude of a given unit contained in it). Now no multitude is the greatest, because one or more units can always be added. Therefore an infinite given magnitude, and consequently a world infinite either in its elapsed series or in its extension, is impossible: it is therefore bounded in both respects. I could have conducted my proof in this way. But this concept does not agree with what is understood by an infinite whole. It does not represent how great that whole is, so its concept is not the concept of a maximum.

B 460 It thinks only the whole’s relation to a unit that may be assumed at will, in respect of which the whole is greater than any number. Depending on whether the unit is assumed greater or smaller, the infinite would be greater or smaller; but its infinity, consisting merely in its relation to this given unit, would always remain the same. Admittedly, the absolute magnitude of the whole would not thereby be cognized at all, but that is not at issue here.

The true (transcendental) concept of infinity is this: the successive synthesis of the unit in measuring a quantum can never be completed.* From this it follows with certainty that an eternity of actual successive states cannot have elapsed up to a given (the present) point in time, and therefore that the world must have a beginning.

* It thereby contains a multitude (of the given unit) greater than any number, which is the mathematical concept of the infinite.

With regard to the second part of the thesis, the difficulty of an infinite yet elapsed series does indeed disappear, for the manifold of a world infinite in extension is given simultaneously. But to think the totality of such a multitude, since we cannot appeal to boundaries that themselves constitute this totality in intuition, we must account for our concept. In such a case it cannot proceed from the whole to the determinate multitude of parts, but must demonstrate the possibility of a whole through the successive synthesis of the parts. Since this synthesis would have to constitute a series that could never be completed, we cannot think a totality prior to it, and thus not through it either. For in this case the concept of totality itself is the representation of a completed synthesis of the parts, and this completion, hence also its concept, is impossible.

B 459 II. Remark on the Antithesis

The proof of the infinity of the given world-series and of the sum total of the world rests on this: in the opposite case an empty time, and likewise an empty space, would have to constitute the boundary of the world. I am well aware that attempts are made to evade this consequence by claiming that a boundary of the world in time and space is quite possible without having to assume an absolute time before the world’s beginning, or an absolute space extending outside the actual world—which is impossible. I am quite content with the latter part of this opinion of the philosophers of the Leibnizian school. Space is merely the form of outer intuition, not an actual object that can be intuited externally, and not a correlate of appearances but the form of appearances themselves. Space, therefore, cannot occur absolutely (on its own) as something determining the existence of things, because it is no object at all, but only the form of possible objects. Things as appearances do indeed determine space: that is, among all its possible predicates (magnitude and relation), they bring it about that these or those belong to actuality. But conversely, space, as something existing on its own, cannot determine the actuality of things with regard to magnitude or shape, because it is nothing actual in itself. A space (whether full or empty)* can therefore indeed be bounded by appearances.

B 461 Appearances, however, cannot be bounded by an empty space outside them. The same holds for time. Granting all this, it nevertheless remains indisputable that we must absolutely assume these two nonentities—empty space outside the world and empty time before it—if we assume a boundary of the world, whether in space or in time.

* It is readily apparent that this means: empty space insofar as it is bounded by appearances, and thus empty space within the world, at least does not contradict transcendental principles and can therefore be admitted with respect to them (although its possibility is not thereby immediately asserted).

For the escape by which people seek to evade the consequence we state—that if the world has boundaries in time and space, the infinite void must determine the existence of actual things with regard to their magnitude—secretly consists only in this: instead of a sensible world, they think some intelligible world, who knows what; instead of a first beginning (an existence preceded by a time of nonexistence), they think an existence in general that presupposes no other condition in the world; instead of a boundary of extension, they think limits of the world-whole, and thereby evade time and space. But here we are concerned only with the mundus phaenomenon and its magnitude. We can by no means abstract from the stated conditions of sensibility without abolishing its essence. If the sensible world is bounded, it necessarily lies in the infinite void. If we wish to leave this out, and hence space in general as an a priori condition of the possibility of appearances, the entire sensible world disappears. In our problem this world alone is given to us. The mundus intelligibilis is nothing but the general concept of a world in general, in which we abstract from all conditions of its intuition, and with respect to which, consequently, no synthetic proposition, affirmative or negative, is possible at all.

B 462 The Antinomy of Pure Reason Second Conflict of the Transcendental Ideas

Thesis

Every composite substance in the world consists of simple parts, and nothing exists anywhere but the simple or what is composed of it.

Proof

For suppose composite substances did not consist of simple parts. Then, if all composition were abolished in thought, no composite part would remain, and (since there are no simple parts) no simple part either; thus nothing at all would remain, and consequently no substance would have been given. Either, then, it is impossible to abolish all composition in thought, or after its abolition something existing without any composition—that is, the simple—must remain. In the first case, however, the composite would in turn not consist of substances (because in these, composition is only a contingent relation of substances, without which they must exist as beings enduring on their own).

B 464 Since this case contradicts the supposition, only the second remains: namely, that the substantial composite in the world consists of simple parts.

From this it follows immediately that all things in the world are simple beings, that composition is only their external state, and that although we can never entirely remove the elementary substances from this state of combination and isolate them, reason must nevertheless think them as the first subjects of all composition, and thus as simple beings prior to it.

B 463 Antithesis

No composite thing in the world consists of simple parts, and nothing simple exists anywhere in it.

Proof

Suppose a composite thing (as substance) consists of simple parts. Since all external relation, and therefore all composition of substances, is possible only in space, the space occupied by the composite must consist of as many parts as the composite itself. Now space does not consist of simple parts, but of spaces. Thus every part of the composite must occupy a space. But the absolutely first parts of every composite are simple. Therefore the simple occupies a space. Now everything real that occupies a space contains a manifold whose elements are external to one another, and is thus composite. As a real composite, it consists not of accidents (for these cannot be external to one another without substance), but of substances. Thus the simple would be a substantial composite, which contradicts itself.

The second proposition of the antithesis is that nothing simple exists in the world at all.

B 465 Here it is meant to say only this: the existence of the absolutely simple cannot be demonstrated from any experience or perception, either outer or inner. The absolutely simple is therefore a mere idea whose objective reality can never be demonstrated in any possible experience, and hence is without any application or object in the exposition of appearances. For suppose an object of experience could be found for this transcendental idea. Then the empirical intuition of some object would have to be cognized as containing absolutely no manifold whose elements are external to one another and combined into unity. But from our lack of consciousness of such a manifold no inference is valid to its complete impossibility in any intuition of an object; yet this latter is absolutely required for absolute simplicity. It follows that simplicity cannot be inferred from any perception whatever. Thus, since something can never be given as an absolutely simple object in any possible experience, while the sensible world must be regarded as the sum total of all possible experiences, nothing simple is given anywhere in it.

This second proposition of the antithesis goes much further than the first, which excludes the simple only from the intuition of the composite; this one, by contrast, removes it from the whole of nature. Hence it could be proved not from the concept of a given object of outer intuition (the composite), but from its relation to a possible experience in general.

B 466 Remark on the Second Antinomy I. On the Thesis

When I speak of a whole that necessarily consists of simple parts, I mean only a substantial whole as the proper compositum: that is, the contingent unity of a manifold that is given separately (at least in thought), placed in reciprocal combination, and thereby constitutes one thing. Space should properly be called not compositum but totum, because its parts are possible only in the whole, not the whole through its parts. At most it could be called a compositum ideale, but not reale. Yet this is only a subtlety. Since space is not a composite of substances (not even of real accidents), if I abolish all composition in it, nothing must remain, not even a point; for a point is possible only as the boundary of a space (and thus of a composite).

B 468 Space and time therefore do not consist of simple parts, and neither does what belongs only to the state of a substance, even if it has a magnitude (for example, alteration); that is, a certain degree of alteration does not arise through an accumulation of many simple alterations. Our inference from the composite to the simple holds only for things existing on their own. Accidents of a state, however, do not exist on their own. Thus we can easily ruin the proof of the necessity of the simple as the constituent parts of every substantial composite, and thereby our case in general, if we extend it too far and try to make it hold for everything composite without distinction, as has in fact often happened.

Moreover, I speak here only of the simple insofar as it is necessarily given in the composite, since the latter can be resolved into it as into its constituent parts.

B 470 The proper meaning of the word “Monas” (according to Leibniz’s usage) should probably concern only the simple given immediately as simple substance (for example, in self-consciousness), not as an element of the composite, which might better be called the Atomus. And since I wish to prove simple substances only with regard to the composite, as its elements, I could call the antithesis of the second antinomy transcendental atomism. But since this word has long been used to designate a particular way of explaining corporeal appearances (molecularum), and thus presupposes empirical concepts, let it be called the dialectical principle of monadology.

B 467 II. Remark on the Antithesis

Against this proposition of an infinite division of matter, whose ground of proof is merely mathematical, the monadists raise objections that already arouse suspicion by refusing to accept the clearest mathematical proofs as insights into the constitution of space, insofar as it is in fact the formal condition of the possibility of all matter. Instead, they regard them only as inferences from abstract but arbitrary concepts that could not be related to actual things. As if it were even possible to invent a kind of intuition other than the one given in the original intuition of space, and as if its a priori determinations did not also concern everything possible only by filling that space. If we listen to them, we would have to think, besides the mathematical point, which is simple but not a part, merely the boundary of a space, physical points as well. These, too, would be simple, but would have the advantage of filling space as its parts through their mere aggregation. Without repeating here the many common and clear refutations of this absurdity—it is wholly futile to try to reason away mathematical evidence through merely discursive concepts—I note only the following.

B 469 When philosophy quibbles with mathematics here, it does so because it forgets that this question concerns only appearances and their condition. Here it is not enough to find the concept of the simple for the pure concept of the understanding of the composite; we must find the intuition of the simple for the intuition of the composite (matter). This is wholly impossible according to the laws of sensibility, and thus also in objects of the senses. It may therefore always hold of a whole made of substances, thought merely through the pure understanding, that we must have the simple prior to all its composition. Yet this does not hold of the totum substantiale phaenomenon, which, as empirical intuition in space, necessarily has the property that no part of it is simple, because no part of space is simple. The monadists, however, have been subtle enough to try to evade this difficulty by presupposing not space as a condition of the possibility of objects of outer intuition (bodies), but these objects and the dynamical relation of substances in general as the condition of the possibility of space. Now we have a concept of bodies only as appearances; as such, however, they necessarily presuppose space as the condition of the possibility of all outer appearance. The evasion is therefore futile, as it was also sufficiently blocked above in the Transcendental Aesthetic. Were they things in themselves, the monadists’ proof would certainly hold.

B 471 The second dialectical assertion has this peculiarity: opposed to it is a dogmatic assertion that, alone among all sophistical assertions, undertakes to prove visibly in an object of experience the actuality of what we counted above merely among transcendental ideas, namely the absolute simplicity of substance. It asserts that the object of inner sense, the “I” that thinks, is an absolutely simple substance. Without entering into this now (since it was considered more fully above), I note only that when something is thought merely as an object, without adding any synthetic determination of its intuition (as occurs through the entirely bare representation “I”), then certainly no manifold and no composition can be perceived in such a representation. Moreover, since the predicates through which I think this object are merely intuitions of inner sense, nothing can occur in them that would prove a manifold whose elements are external to one another, and hence real composition. Self-consciousness therefore entails only that, because the subject that thinks is also its own object, it cannot divide itself (although it can divide the determinations inhering in it); for in relation to itself every object is absolute unity. Nevertheless, if this subject is considered externally, as an object of intuition, it would indeed display composition in appearance. But it must always be considered in this way if we wish to know whether or not it contains a manifold whose elements are external to one another.

B 472 The Antinomy of Pure Reason Third Conflict of the Transcendental Ideas

Thesis

Causality according to laws of nature is not the only causality from which all the appearances of the world can be derived. To explain them, it is necessary to assume a causality through freedom as well.

Proof

Suppose there is no causality other than that according to laws of nature. Then everything that happens presupposes a previous state, on which it inevitably follows according to a rule. But the previous state must itself be something that happened (came to be in time, since it did not previously exist), because if it had always existed, its consequence, too, would not first have arisen, but would always have existed. Thus the causality of the cause through which something happens is itself something that has happened, which according to the law of nature again presupposes a previous state and its causality, and this in turn a still earlier one, and so on. If, therefore, everything happens according to mere laws of nature, there is always only a subordinate beginning.

B 474 There is never, however, a first beginning, and thus no completeness at all of the series on the side of causes deriving from one another. But the law of nature consists precisely in this: nothing happens without a cause sufficiently determined a priori. Thus the proposition that all causality is possible only according to laws of nature contradicts itself in its unrestricted universality, and this causality cannot therefore be assumed to be the only kind.

Accordingly, we must assume a causality through which something happens without its cause being further determined according to necessary laws by another preceding cause: that is, an absolute spontaneity of causes to begin of themselves a series of appearances that proceeds according to laws of nature, and hence transcendental freedom, without which even in the course of nature the succession of appearances is never complete on the side of causes.

B 473 Antithesis

There is no freedom; everything in the world happens solely according to laws of nature.

Proof

Suppose there is freedom in the transcendental sense, as a special kind of causality according to which events in the world could occur: namely, a power to begin a state, and thus a series of its consequences, absolutely. Then not only will a series begin absolutely through this spontaneity, but the determination of this spontaneity itself to produce the series—that is, its causality—will also begin absolutely, so that nothing precedes by which this occurring action is determined according to constant laws. But every beginning of action presupposes a state of the cause in which it is not yet acting, and a dynamically first beginning of action presupposes a state having no causal connection at all with the preceding state of that same cause: that is, it in no way follows from it. Transcendental freedom is therefore contrary to the causal law.

B 475 It is a combination of the successive states of efficient causes of a kind according to which no unity of experience is possible; it is thus also found in no experience, and is consequently an empty thought-entity.

We therefore have nothing but nature in which to seek the connection and order of events in the world. Freedom (independence) from laws of nature is indeed a liberation from constraint, but also from the guiding thread of all rules. For we cannot say that laws of freedom enter into the causality of the world’s course in place of laws of nature, because if this causality were determined according to laws, it would not be freedom but itself nothing other than nature. Nature and transcendental freedom therefore differ as lawfulness and lawlessness. The former admittedly burdens the understanding with the difficulty of seeking the derivation of events ever higher in the series of causes, because their causality is always conditioned; but in compensation it promises thoroughgoing and lawful unity of experience. The mirage of freedom, by contrast, promises the inquiring understanding rest in the chain of causes by leading it to an unconditioned causality that begins to act of itself. But since it is itself blind, it breaks the guiding thread of rules through which alone a thoroughly connected experience is possible.

B 476 Remark on the Third Antinomy I. On the Thesis

The transcendental idea of freedom is far from constituting the entire content of the psychological concept bearing this name, which is largely empirical. It constitutes only the absolute spontaneity of action, as the proper ground of its imputability. Yet it is the real stumbling block for philosophy, which finds insuperable difficulties in admitting such a kind of unconditioned causality. Thus what has always placed speculative reason in such great perplexity in the question of freedom of the will is properly only transcendental. It concerns solely whether we must assume a power to begin a series of successive things or states of itself. It is not equally necessary to be able to answer how such a power is possible, since even with causality according to laws of nature we must be content to cognize a priori that such causality must be presupposed, although we cannot in any way comprehend the possibility of one existence positing the existence of another, and in this respect must hold solely to experience. Now we have properly demonstrated this necessity of a first beginning of a series of appearances through freedom only insofar as it is required to make an origin of the world comprehensible, while all subsequent states can be taken as a succession according to mere laws of nature.

B 478 But since the power to begin a series in time entirely of itself has thereby once been proved (though not understood), we are now also permitted to let various series begin of themselves, as regards causality, in the midst of the world’s course, and to attribute to its substances a power to act from freedom. We must not, however, let a misunderstanding obstruct us here: namely, that because a successive series in the world can have only a comparatively first beginning, since a state of things in the world always precedes it, perhaps no absolutely first beginning of series is possible during the world’s course. For here we are speaking not of an absolutely first beginning in time, but in causality. If I now (for example) rise from my chair entirely freely, without the necessarily determining influence of natural causes, then in this event, together with its natural consequences to infinity, a new series begins absolutely, although in time this event is only the continuation of a preceding series. For this decision and deed do not lie in the succession of mere natural effects and are not their mere continuation. Rather, the determining natural causes cease entirely above it with regard to this event, which follows them but does not follow from them, and must therefore be called an absolutely first beginning of a series of appearances, not in time, but with respect to causality.

The confirmation of reason’s need to appeal to a first beginning through freedom in the series of natural causes is especially clear in this: all ancient philosophers (except the Epicurean school) felt compelled to assume a first mover to explain the motions of the world—that is, a freely acting cause that first began this series of states of itself. For they did not venture to make a first beginning comprehensible from mere nature.

B 477 II. Remark on the Antithesis

The defender of nature’s omnipotence (transcendental physiocracy), opposed to the doctrine of freedom, would maintain his proposition against the sophistical inferences of the latter as follows. If you assume no mathematically first term in time in the world, you also have no need to seek a dynamically first term in causality. Who told you to invent an absolutely first state of the world, and thus an absolute beginning of the series of appearances that unfolds successively, and to set boundaries to unrestricted nature in order to provide your imagination a resting place? Since substances in the world have always existed—at least the unity of experience makes such a presupposition necessary—there is no difficulty in also assuming that the succession of their states, that is, a series of their alterations, has always existed, and consequently that no first beginning, either mathematical or dynamical, need be sought. The possibility of such an infinite derivation, without a first member in relation to which everything else is merely subsequent, cannot be made comprehensible. But if you wish to discard these riddles of nature on that account, you will find yourselves compelled to reject many synthetic fundamental properties (fundamental powers) that you can comprehend just as little.

B 479 Even the possibility of alteration in general must offend you. For if you did not find through experience that it is actual, you could never devise a priori how such an unceasing succession of being and nonbeing is possible.

Even if a transcendental power of freedom were conceded, if necessary, to begin the alterations of the world, this power would at least have to be only outside the world (although it always remains a bold presumption to assume, outside the sum total of all possible intuitions, an object that can be given in no possible perception). But attributing such a power to substances within the world itself can never be permitted, because then the connection according to universal laws of appearances necessarily determining one another, which we call nature, would largely disappear, and with it the criterion of empirical truth that distinguishes experience from dreams. For alongside such a lawless power of freedom, nature can scarcely be thought any longer: its laws would be incessantly altered by the influences of freedom, and the play of appearances, which according to mere nature would be regular and uniform, would thereby be made confused and disconnected.

B 480 The Antinomy of Pure Reason Fourth Conflict of the Transcendental Ideas

Thesis

There belongs to the world something that, either as its part or as its cause, is an absolutely necessary being.

Proof

The sensible world, as the whole of all appearances, also contains a series of alterations. For without these, even the representation of the temporal series, as a condition of the possibility of the sensible world, would not be given to us.* But every alteration stands under its condition, which precedes it in time and under which it is necessary. Now every conditioned that is given presupposes, with respect to its existence, a complete series of conditions up to the absolutely unconditioned, which alone is absolutely necessary. Therefore something absolutely necessary must exist if an alteration exists as its consequence. But this necessary being itself belongs to the sensible world. For suppose it were outside that world: then the series of alterations in the world would derive its beginning from it.

B 482 Yet this necessary cause itself would not belong to the sensible world. Now this is impossible. For since the beginning of a temporal series can be determined only by what precedes in time, the highest condition of the beginning of a series of alterations must exist in the time when that series did not yet exist (for a beginning is an existence preceded by a time in which the thing that begins did not yet exist). Thus the causality of the necessary cause of alterations, and hence the cause itself, belongs to time, and consequently to appearance (in which alone time is possible as its form). It therefore cannot be thought as separate from the sensible world as the sum total of all appearances. Thus something absolutely necessary is contained in the world itself (whether this is the entire world-series itself or a part of it).

* Time, as the formal condition of the possibility of alterations, does indeed objectively precede them. Subjectively, however, and in the actuality of consciousness, this representation, like every other, is given only on the occasion of perceptions.

B 481 Antithesis

No absolutely necessary being exists anywhere, either in the world or outside the world as its cause.

Proof

Suppose the world itself is a necessary being, or a necessary being exists within it. Then in the series of its alterations there would either be a beginning that was unconditionally necessary, and thus without a cause, which conflicts with the dynamical law determining all appearances in time; or the series itself would be without any beginning and, although contingent and conditioned in all its parts, nevertheless absolutely necessary and unconditioned as a whole. This contradicts itself, because the existence of a multitude cannot be necessary if not a single part of it possesses an existence necessary in itself.

Suppose, on the other hand, there is an absolutely necessary cause of the world outside the world.

B 483 Then, as the highest member in the series of causes of alterations in the world, it would first begin their existence and their series.* But it would then also have to begin to act, and its causality would belong to time, and for that very reason to the sum total of appearances, that is, to the world. Consequently the cause itself would not be outside the world, contrary to the supposition. Thus neither in the world nor outside it (but in causal combination with it) is there any absolutely necessary being.

* The word “begin” is taken in two senses. The first is active, when the cause begins a series of states as its effect (infit.). The second is passive, when causality arises in the cause itself (fit.). Here I infer from the former to the latter.

B 484 Remark on the Fourth Antinomy I. On the Thesis

To prove the existence of a necessary being, I am obliged here to use no argument other than a cosmological one: one that ascends from the conditioned in appearance to the unconditioned in the concept, regarding this as the necessary condition of the absolute totality of the series. Attempting a proof from the mere idea of a highest being of all beings in general belongs to another principle of reason, and such a proof will therefore have to occur separately.

Now the pure cosmological proof can demonstrate the existence of a necessary being only by leaving unsettled whether that being is the world itself or a thing distinct from it. For to establish the latter requires principles that are no longer cosmological and do not proceed in the series of appearances, but rather concepts of contingent beings in general (insofar as they are considered merely as objects of the understanding), and a principle for connecting these with a necessary being through mere concepts. All this belongs to a transcendent philosophy for which this is not yet the place.

But once we begin the proof cosmologically, taking as our basis the series of appearances and the regression within it according to empirical laws of causality, we cannot afterward leap away from it and pass to something that does not belong to the series as a member at all.

B 486 For something must be regarded as a condition in the very same sense in which the relation of the conditioned to its condition was taken within the series that was to lead to this highest condition through continuous advance. Now if this relation is sensible and belongs to the possible empirical use of the understanding, the highest condition or cause can conclude the regression only according to laws of sensibility, and thus only as belonging to the temporal series; the necessary being must be regarded as the highest member of the world-series.

Nevertheless, people have taken the liberty of making such a leap (metabasis eis allo genos). From alterations in the world they inferred empirical contingency—that is, their dependence on empirically determining causes—and obtained an ascending series of empirical conditions, which was entirely correct. But since they could find no first beginning and no highest member in this series, they suddenly abandoned the empirical concept of contingency and took up the pure category. This then occasioned a merely intelligible series whose completeness rested on the existence of an absolutely necessary cause. Since this cause was now bound to no sensible conditions, it was also freed from the temporal condition of itself beginning its causality. But this procedure is wholly illegitimate, as we can conclude from the following.

Contingent, in the pure sense of the category, is that whose contradictory opposite is possible. Now we cannot infer that intelligible contingency from empirical contingency at all.

B 488 The opposite of what is altered (of its state) is actual at another time, and thus also possible. But this is not the contradictory opposite of the previous state; that would require that at the same time when the previous state existed, its opposite could have existed in its place, which cannot be inferred from alteration at all. A body that was in motion = A comes to rest = non A. From the fact that a state opposed to A follows it, we cannot infer that the contradictory opposite of A is possible, and hence that A is contingent; for that would require that at the same time when the motion existed, rest could have existed instead. Now all we know is that rest was actual, and thus possible, at the subsequent time. But motion at one time and rest at another are not contradictorily opposed to one another. Thus the succession of opposed determinations, that is, alteration, by no means proves contingency according to concepts of the pure understanding, and therefore cannot lead to the existence of a necessary being according to pure concepts of the understanding. Alteration proves only empirical contingency: that is, according to the law of causality, the new state could not have occurred on its own without a cause belonging to the preceding time. Even if this cause is assumed to be absolutely necessary, it must in this way still be found in time and belong to the series of appearances.

B 485 II. Remark on the Antithesis

If, in ascending the series of appearances, we think we encounter difficulties against the existence of an absolutely necessary highest cause, these must not rest on mere concepts of the necessary existence of a thing in general, and thus must not be ontological. They must emerge instead from the causal combination with a series of appearances when we assume for that series a condition that is itself unconditioned; consequently, they must be cosmological and inferred according to empirical laws. Specifically, we must show that ascent in the series of causes (in the sensible world) can never end at an empirically unconditioned condition, and that the cosmological argument from the contingency of states of the world, as evidenced by their alterations, goes against the assumption of a first cause that begins the series absolutely.

B 487 But a strange contrast appears in this antinomy: from the very same ground of proof from which the thesis inferred the existence of an original being, the antithesis infers its nonexistence, and with equal rigor. First it was said: there is a necessary being because the whole past time contains the series of all conditions, and thus also the unconditioned (necessary). Now it is said: there is no necessary being precisely because the whole elapsed time contains the series of all conditions (which are consequently all in turn conditioned). The cause is this. The first argument considers only the absolute totality of the series of conditions, one of which determines another in time, and thereby obtains something unconditioned and necessary. The second, by contrast, considers the contingency of everything determined in the temporal series (because each is preceded by a time in which the condition itself must in turn be determined as conditioned), whereby everything unconditioned and all absolute necessity fall away entirely.

B 489 Yet the mode of inference in both is quite in keeping even with ordinary human reason, which often comes to be divided against itself when considering its object from two different standpoints. Mr. von Mairan considered the dispute between two famous astronomers, arising from a similar difficulty about the choice of standpoint, a phenomenon sufficiently remarkable to warrant a separate treatise. One reasoned as follows: the moon rotates on its axis because it constantly turns the same side toward the earth. The other: the moon does not rotate on its axis, precisely because it constantly turns the same side toward the earth. Both inferences were correct, depending on the standpoint from which one chose to observe the moon’s motion.

B 490 The Antinomy of Pure Reason Section Three The Interest of Reason in This Conflict of Its Own

We now have the entire dialectical play of cosmological ideas. These permit neither an object congruent with them to be given in any possible experience, nor even reason to think them in harmony with universal laws of experience. Yet they are not arbitrarily invented: reason is necessarily led to them in the continuous advance of empirical synthesis when it wishes to free from every condition what can always be determined only conditionally according to rules of experience, and to grasp it in its unconditioned totality. These sophistical assertions are so many attempts to solve four natural and unavoidable problems of reason. There can therefore be precisely this many, neither more nor fewer, because there are no more series of synthetic presuppositions that bound empirical synthesis a priori.

We have presented the brilliant pretensions of reason, which extends its domain beyond all boundaries of experience, only in dry formulas containing merely the ground of its legal claims.

B 491 As befits a transcendental philosophy, we have stripped these of everything empirical, although the full splendor of reason’s assertions can shine forth only in combination with it. But in this application, and in the progressive extension of the use of reason, beginning in the field of experiences and gradually soaring to these sublime ideas, philosophy displays a dignity that, if it could only vindicate its pretensions, would leave the value of all other human science far beneath it. For it promises the foundation for our greatest expectations and prospects concerning the ultimate ends in which all endeavors of reason must finally unite. Whether the world has a beginning and any boundary to its extension in space; whether anywhere, and perhaps in my thinking self, there is an indivisible and indestructible unity, or nothing but the divisible and perishable; whether in my actions I am free or, like other beings, led along the thread of nature and fate; whether, finally, there is a highest cause of the world, or whether natural things and their order constitute the final object at which we must stop in all our reflections: these are questions for whose solution the mathematician would gladly surrender his entire science; for that science can give him no satisfaction with regard to humanity’s highest and most urgent ends.

B 492 Even the proper dignity of mathematics (this pride of human reason) rests on its guiding reason to insight into nature, both in the large and in the small, in its order and regularity and in the admirable unity of the forces that move it, far beyond every expectation of philosophy founded on ordinary experience. It thereby itself provides occasion and encouragement for the use of reason extended beyond all experience, and supplies the philosophy concerned with that use with the finest materials for supporting its investigation, as far as its character permits, through appropriate intuitions.

Unfortunately for speculation (though perhaps fortunately for the practical vocation of the human being), reason finds itself, amid its greatest expectations, so entangled in a press of grounds and countergrounds that—since neither its honor nor even its security permits it to withdraw and watch this dispute indifferently as a mere mock battle, still less to command peace outright, because the object of the dispute is of great interest—nothing remains for it but to reflect on the origin of this disagreement of reason with itself, and to ask whether a mere misunderstanding may not be to blame.

B 493 Once that is clarified, the proud claims on both sides might indeed fall away, but in their place a permanently peaceful rule of reason over understanding and senses would begin.

For now we will postpone this thorough clarification a little longer and first consider which side we would most readily join if compelled to take a party. Since in this case we consult not the logical touchstone of truth but merely our interest, such an investigation, though it decides nothing about the disputed rights of either side, will nevertheless help us understand why participants in this dispute have joined one side rather than the other without any superior insight into the object being the cause. It will likewise explain other incidental matters: for example, the zealous heat of one side and the cold assertions of the other, why people gladly acclaim one party with joyful applause and are irreconcilably prejudiced against the other in advance.

There is, however, something that determines the standpoint from which alone this preliminary assessment can be undertaken with due thoroughness: comparison of the principles from which the two sides proceed. Among the assertions of the antithesis we notice a complete uniformity of thought and an entire unity of maxim.

B 494 The maxim is, namely, a principle of pure empiricism, not only in explaining appearances in the world but also in resolving the transcendental ideas of the universe itself. The assertions of the thesis, by contrast, take intellectual beginnings as their basis in addition to the empirical mode of explanation within the series of appearances, and to that extent their maxim is not simple. But from its essential distinguishing characteristic I will call it the dogmatism of pure reason.

On the side of dogmatism, then, in determining the cosmological ideas of reason, or on the side of the thesis, there appears:

First, a certain practical interest in which every well-disposed person, if he understands his true advantage, takes a heartfelt part. That the world has a beginning; that my thinking self is simple and therefore imperishable in nature; that it is also free in its voluntary actions and raised above natural compulsion; and that, finally, the entire order of things constituting the world derives from an original being from which everything borrows its unity and purposive connection—these are so many foundation stones of morality and religion. The antithesis robs us of all these supports, or at least seems to do so.

Second, a speculative interest of reason also manifests itself on this side.

B 495 For if we accept and employ transcendental ideas in this way, we can grasp the whole chain of conditions entirely a priori and comprehend the derivation of the conditioned by beginning with the unconditioned. The antithesis cannot achieve this, and recommends itself very poorly by being unable to answer the question about the conditions of its synthesis without leaving us to ask further questions endlessly. According to it, we must ascend from a given beginning to a still higher one; every part leads to a still smaller part; every event always has another event above it as cause; and the conditions of existence in general always rest in turn on others, without ever obtaining unconditioned footing and support in a self-subsisting thing as original being.

Third, this side also has the advantage of popularity, which certainly contributes no small part to its appeal. Ordinary understanding finds not the slightest difficulty in the ideas of the unconditioned beginning of all synthesis, since it is in any case more accustomed to descend to consequences than to ascend to grounds. In the concepts of the absolutely first (whose possibility it does not ponder), it finds ease and also a firm point to which it can attach the guiding cord of its steps. By contrast, it can find no satisfaction in the restless ascent from the conditioned to its condition, always with one foot in the air.

B 496 On the side of empiricism in determining cosmological ideas, or of the antithesis, there is, first, no such practical interest arising from pure principles of reason as morality and religion carry with them. Rather, mere empiricism seems to deprive both of all force and influence. If there is no original being distinct from the world; if the world is without a beginning and thus also without an author; if our will is not free and the soul is as divisible and perishable as matter, then moral ideas and principles also lose all validity and fall together with the transcendental ideas that constituted their theoretical support.

In compensation, however, empiricism offers reason’s speculative interest advantages that are very enticing and far surpass those the dogmatic teacher of ideas of reason can promise. According to empiricism, the understanding is always on its own proper ground, namely the field of nothing but possible experiences, whose laws it can trace and through which it can extend its secure and comprehensible cognition without end. Here it can and should present the object to intuition, both in itself and in its relations, or at least in concepts whose image can be exhibited clearly and distinctly in given similar intuitions. Not only does it have no need to leave this chain of the natural order.

B 497 It need not do so in order to cling to ideas whose objects it does not know, because as thought-entities they can never be given. It is not even permitted to abandon its work and, on the pretext that it is now complete, pass into the domain of idealizing reason and to transcendent concepts. There it would no longer need to observe and investigate according to laws of nature, but only to think and invent, certain that it could not be refuted by facts of nature, since it is not bound to their testimony but may pass them by, or even subordinate them to a higher authority, that of pure reason.

The empiricist will therefore never permit any epoch of nature to be assumed as absolutely the first, or any boundary to his view of its extent to be regarded as the outermost; nor will he permit a transition from objects of nature that he can analyze through observation and mathematics and determine synthetically in intuition (the extended) to those that neither sense nor imagination can ever present in concreto (the simple). Nor will he concede that a power to act independently of laws of nature (freedom) should be taken as a basis within nature itself, thereby curtailing the understanding’s work of tracing the origin of appearances along the guiding thread of necessary rules.

B 498 Nor, finally, will he admit that we should seek a cause for anything outside nature (an original being), because we know nothing beyond nature, since it alone offers us objects and can instruct us in its laws.

To be sure, if the empirical philosopher intends his antithesis only to suppress the prying curiosity and presumption of reason that mistakes its true vocation, boasting of insight and knowledge precisely where insight and knowledge cease, and trying to pass off what is accepted with regard to the practical interest as advancing the speculative interest, so as, whenever convenience dictates, to break the thread of physical investigations and, under the pretense of extending cognition, attach it to transcendental ideas through which we properly cognize only that we know nothing—if, I say, the empiricist were content with this, his principle would be a maxim of moderation in claims, modesty in assertions, and at the same time the greatest possible extension of our understanding through the teacher properly appointed for us, namely experience. For in such a case intellectual presuppositions and faith would not be taken from us for the purposes of our practical affairs. Only, they could not be allowed to appear under the title and pomp of science and rational insight.

B 499 For properly speculative knowledge can concern no object other than that of experience; and when we cross its boundary, the synthesis that attempts new cognitions independent of experience has no substratum of intuition on which it could be exercised.

But when empiricism itself becomes dogmatic concerning ideas (as usually happens) and boldly denies what lies beyond the sphere of its intuitive cognitions, it itself falls into the fault of immodesty, which is all the more blameworthy here because it causes irreparable harm to reason’s practical interest.

This is the opposition of Epicureanism* to Platonism.

* It remains a question, however, whether Epicurus ever presented these principles as objective assertions. If they were perhaps nothing more than maxims of the speculative use of reason, he showed in them a more genuine philosophical spirit than any of the ancient philosophers. That in explaining appearances we must proceed as if the field of investigation were cut off by no boundary or beginning of the world; that we must assume the material of the world as it must be if we are to learn about it through experience; that we must employ no production of events other than as they are determined through unalterable laws of nature, and finally no cause distinct from the world—these are still very correct, though little observed, principles for extending speculative philosophy and likewise for discovering the principles of morality independently of outside resources. Someone who demands that we ignore those dogmatic propositions as long as we are occupied with mere speculation must not on that account be accused of wishing to deny them.

B 500 Each says more than he knows, though in such a way that the former encourages and advances knowledge, albeit to the detriment of the practical, while the latter indeed supplies excellent principles for the practical, but thereby allows reason, in everything in which speculative knowledge alone is granted us, to pursue ideal explanations of natural appearances and neglect physical investigation for their sake.

As for the third consideration relevant to a preliminary choice between the two contending parties, it is exceedingly strange that empiricism is entirely contrary to all popularity. One might have thought ordinary understanding would eagerly embrace a plan promising to satisfy it through nothing but experiential cognitions and their rational connection, whereas transcendental dogmatics forces it to ascend to concepts far exceeding the insight and rational powers of even minds most practiced in thinking. But this is precisely its motive.

B 501 For it then finds itself in a condition in which even the most learned can claim no superiority over it. If it understands little or nothing of the matter, no one else can boast of understanding much more. And although it cannot speak about it as scholastically as others, it can nevertheless reason sophistically about it infinitely more, because it wanders among mere ideas about which we are most eloquent precisely because we know nothing of them. In the investigation of nature, by contrast, it would have to fall completely silent and confess its ignorance. Convenience and vanity thus already strongly recommend these principles. Moreover, although a philosopher finds it very difficult to accept something as a principle without being able to account for it to himself, or even to introduce concepts whose objective reality cannot be understood, nothing is more usual for ordinary understanding. It wants something with which it can confidently begin. The difficulty of comprehending such a presupposition itself does not trouble it, because this never occurs to it (since it does not know what comprehension means), and it takes as known what frequent use has made familiar. Finally, all speculative interest vanishes for it before the practical, and it imagines that it understands and knows what its fears or hopes drive it to assume or believe.

B 502 Thus the empiricism of transcendentally idealizing reason is wholly deprived of popularity. However much it may contain that is detrimental to the highest practical principles, there is no danger that it will ever cross the boundaries of the school and acquire any considerable standing in the community or any favor among the multitude.

Human reason is architectonic by nature: that is, it regards all cognitions as belonging to a possible system, and therefore admits only principles that at least do not make an intended cognition incapable of standing together with others in some system. But the propositions of the antithesis are such as to make the completion of an edifice of cognitions wholly impossible. According to them, beyond a state of the world there is always a still earlier one; in every part there are always further parts that are themselves divisible; before every event there is another that was itself likewise produced by something else; and in existence generally everything is always only conditioned, without acknowledging any unconditioned and first existence. Since the antithesis therefore admits no first term anywhere, and no beginning that could serve absolutely as the foundation of the structure, a complete edifice of cognition is wholly impossible under such presuppositions.

B 503 Hence reason’s architectonic interest (which demands not empirical unity, but pure rational unity a priori) carries with it a natural recommendation for the assertions of the thesis.

But if a person could renounce every interest and consider the assertions of reason, indifferent to all consequences, solely according to the content of their grounds, then, supposing he knew no way out of the predicament except to profess one or the other of the contending doctrines, he would be in a state of ceaseless vacillation. Today it would seem convincing to him that the human will is free; tomorrow, considering the indissoluble chain of nature, he would hold freedom to be nothing but self-deception and everything to be mere nature. But when it came to doing and acting, this play of merely speculative reason would vanish like the shadowy images of a dream, and he would choose his principles solely according to practical interest. Yet it befits a reflective and inquiring being to devote certain periods solely to examining his own reason, laying aside all partiality completely and publicly communicating his observations for others to judge.

B 504 No one can therefore be blamed, much less forbidden, for letting theses and antitheses appear as they can defend themselves, undeterred by any threat, before jurors of his own estate (namely, that of weak human beings).

The Antinomy of Pure Reason Section Four The Transcendental Problems of Pure Reason Insofar as They Must Absolutely Be Capable of Solution

To want to solve all problems and answer all questions would be shameless boasting and such extravagant conceit that one would thereby immediately have to forfeit all trust. Yet there are sciences whose nature requires that every question arising within them must be absolutely answerable from what is known, because the answer must arise from the very sources from which the question arises. Here it is by no means permissible to plead unavoidable ignorance; a solution can be demanded. What is right or wrong in all possible cases must be knowable according to the rule, because it concerns our obligation, and we have no obligation to what we cannot know.

B 505 In explaining appearances of nature, however, much must remain uncertain and many a question insoluble, because what we know of nature is far from sufficient in every case for what we are supposed to explain. The question now is whether, in transcendental philosophy, any question concerning an object presented to reason is unanswerable by this same pure reason, and whether we may rightly evade a decisive answer by counting it as absolutely uncertain (from everything we can cognize), among matters of which we have enough of a concept to raise a question, but entirely lack the means or power ever to answer it.

I now maintain that transcendental philosophy has this peculiarity among all speculative cognition: no question concerning an object given to pure reason is insoluble for this same human reason, and no plea of unavoidable ignorance and unfathomable depth of the problem can absolve us from the obligation to answer it thoroughly and completely. For the very concept that enables us to ask must also make us fully capable of answering this question, since the object is not found at all outside the concept.

B 506 (As in the case of right and wrong.)

But in transcendental philosophy it is only cosmological questions for which we can rightly demand a satisfactory answer concerning the constitution of the object, without allowing the philosopher to evade it by pleading impenetrable obscurity; and these questions can concern only cosmological ideas. For the object must be given empirically, and the question concerns only its adequacy to an idea. If the object is transcendental and thus itself unknown—for example, whether the something whose appearance (in ourselves) is thinking (the soul) is a being simple in itself, whether there is an absolutely necessary cause of all things together, and so on—then we are to seek an object for our idea of which we can confess that it is unknown to us, but not therefore impossible*.

B 507 Cosmological ideas alone have the peculiarity that they can presuppose their object and the empirical synthesis required for its concept as given, and the question arising from them concerns only the advance of this synthesis insofar as it is supposed to contain absolute totality, which is no longer anything empirical, since it cannot be given in any experience. Since here we are speaking solely of a thing as an object of possible experience, not as a thing in itself, the answer to the transcendent cosmological question can lie nowhere outside the idea, for it concerns no object in itself. And with regard to possible experience, the question is not about what can be given in concreto in any experience, but about what lies in the idea, which the empirical synthesis is merely supposed to approach. Thus it must be capable of solution from the idea alone; for the idea is a mere creature of reason, which therefore cannot cast off its responsibility and shift it onto the unknown object.

* We can give no answer to the question of the constitution of a transcendental object, namely what it is, but we can indeed answer that the question itself is nothing, because no object for it has been given. Hence all questions of the transcendental doctrine of the soul are also answerable, and have actually been answered; for they concern the transcendental subject of all inner appearances, which is not itself appearance and thus is not given as an object, and in which none of the categories (to which the question is properly directed) finds conditions of its application. Here, then, the common expression applies that no answer is also an answer: namely, that a question about the constitution of that something which cannot be thought through any determinate predicate, because it is placed entirely outside the sphere of objects that can be given to us, is wholly null and empty.

B 508 It is not as extraordinary as it first seems that a science can demand and expect certain solutions to all questions belonging within its domain (quaestiones domesticae), even if these solutions have perhaps not yet been found. Besides transcendental philosophy, there are two other sciences of pure reason, one of merely speculative, the other of practical content: pure mathematics and pure morality. Has anyone ever heard it declared uncertain, as though because of a necessary ignorance of the conditions, exactly what ratio the diameter bears to the circle in rational or irrational numbers? Since it cannot be given congruently through the former, but has not yet been found through the latter, it was judged that at least the impossibility of such a solution could be cognized with certainty, and Lambert gave a proof of it. In the universal principles of morals nothing can be uncertain, because the propositions are either entirely null and meaningless or must flow solely from our concepts of reason. In natural science, by contrast, there is an infinity of conjectures about which certainty can never be expected, because natural appearances are objects given to us independently of our concepts. Their key thus lies not in us and our pure thinking, but outside us, and for that very reason cannot be found in many cases; hence no certain explanation can be expected.

B 509 I do not include here the questions of the Transcendental Analytic concerning the deduction of our pure cognition, because we are now treating only the certainty of judgments regarding objects, not regarding the origin of our concepts themselves.

We therefore cannot evade the obligation to give at least a critical solution to the questions of reason before us by lamenting the narrow limits of our reason and confessing, with the illusion of humble self-cognition, that it is beyond our reason to determine whether the world has existed from eternity or has a beginning; whether the space of the world is filled with beings to infinity or enclosed within certain boundaries; whether anything in the world is simple or everything must be divided to infinity; whether there is generation and production from freedom or everything hangs on the chain of the natural order; finally, whether there is any entirely unconditioned and intrinsically necessary being, or everything is conditioned in its existence, and thus externally dependent and intrinsically contingent. For all these questions concern an object that can be given nowhere but in our thoughts: the absolutely unconditioned totality of the synthesis of appearances.

B 510 If we can say and establish nothing certain about it from our own concepts, we must not lay the blame on the thing that conceals itself from us. For such a thing cannot be given to us at all (since it is found nowhere outside our idea). Instead, we must seek the cause in our idea itself, which is a problem admitting no solution, yet which we stubbornly assume has an actual object corresponding to it. A clear exposition of the dialectic lying in our concept itself would soon bring us to complete certainty about what we must judge concerning such a question.

Against your pretext of uncertainty about these problems, we can first pose this question, which at least you must answer clearly: where do you get the ideas whose solution entangles you in such difficulty here? Are they perhaps appearances that you need to explain, for which, according to these ideas, you have only to seek the principles or the rule of their exposition? Assume that nature is wholly revealed before you, that nothing is hidden from your senses and your consciousness of everything presented to your intuition. Even then, through no single experience will you be able to cognize the object of your ideas in concreto.

B 511 (For besides this complete intuition, a completed synthesis and consciousness of its absolute totality are also required, which no empirical cognition can provide.) Consequently, your question can by no means be necessary for explaining any appearance that occurs, and thus be assigned, as it were, by the object itself. For the object can never occur to you, since it cannot be given through any possible experience. With all possible perceptions you always remain caught within conditions, whether in space or time, and reach nothing unconditioned from which to determine whether this unconditioned is to be placed in an absolute beginning of synthesis or in an absolute totality of the series without any beginning. The “all” in the empirical sense, however, is always only comparative. The absolute all of magnitude (the universe), of division, of derivation, of the condition of existence in general, together with all questions whether it can be achieved through a finite synthesis or one continued to infinity, has nothing to do with any possible experience. You could not, for example, explain the appearances of a body in the least better, or even differently, whether you assumed that it consisted of simple parts or always and throughout of composite ones; for neither a simple appearance nor an infinite composition can ever occur to you. Appearances require explanation only as far as the conditions of their explanation are given in perception.

B 512 But everything that may ever be given in them, taken together in an absolute whole, is itself a perception. Yet it is precisely this all whose explanation is demanded in the transcendental problems of reason.

Since even the solution of these problems can never occur in experience, you cannot say it is uncertain what should be attributed to the object in this regard. For your object is merely in your brain and cannot be given outside it at all. You therefore have only to ensure agreement with yourself and prevent the amphiboly that turns your idea into a supposed representation of an empirically given object, thus one also cognizable according to laws of experience. The dogmatic solution is therefore not uncertain, but impossible. The critical solution, however, which can be wholly certain, considers the question not objectively at all, but according to the foundation of cognition on which it rests.

B 513 The Antinomy of Pure Reason Section Five Skeptical Representation of the Cosmological Questions through All Four Transcendental Ideas

We would gladly relinquish the demand to have our questions answered dogmatically if we already understood in advance that, whatever the answer, it would only increase our ignorance, plunging us from one incomprehensibility into another, from one obscurity into a still greater one, and perhaps even into contradictions. If our question calls merely for affirmation or denial, it is wise to leave the presumable grounds of the answer aside for the moment and first consider what we would gain if the answer fell on one side, and what if it fell on the opposite side. If it turns out that in both cases the result is sheer meaninglessness (nonsense), we have good grounds for critically examining our question itself and seeing whether it rests on an unfounded presupposition and plays with an idea that reveals its falsity better in application and through its consequences than in its isolated representation.

B 514 This is the great benefit of the skeptical way of treating the questions pure reason addresses to pure reason: with little effort we can rid ourselves of a great mass of dogmatic clutter and put in its place a sober critique that, as a true Katarktikon, will successfully expel delusion together with its retinue, polymathy.

If, accordingly, I could see in advance of a cosmological idea that, whichever side of the unconditioned in the regressive synthesis of appearances it took, it would be either too great or too small for every concept of the understanding, then I would comprehend that, since this idea concerns only an object of experience that ought to be commensurate with a possible concept of the understanding, it must be entirely empty and meaningless, because the object does not fit it however I try to accommodate it to the idea. This is indeed the case with all world-concepts, which for that very reason entangle reason in an unavoidable antinomy as long as it clings to them. For suppose:

First, that the world has no beginning. Then it is too great for your concept, since this concept, consisting in a successive regression, can never reach the whole elapsed eternity. Suppose it has a beginning: then, in the necessary empirical regression, it is again too small for your concept of the understanding.

B 515 For since the beginning still presupposes a preceding time, it is not yet unconditioned, and the law of the empirical use of the understanding requires you to ask for a still higher temporal condition. The world is thus plainly too small for this law.

The same holds for the two answers to the question of the world’s spatial magnitude. For if it is infinite and unbounded, it is too great for every possible empirical concept. If it is finite and bounded, you rightly ask further: what determines this boundary? Empty space is no self-subsisting correlate of things and can be no condition at which you can stop, still less an empirical condition constituting part of a possible experience. (For who can have an experience of the absolutely empty?) But the absolute totality of empirical synthesis always requires the unconditioned to be a concept of experience. Thus a bounded world is too small for your concept.

Second, if every appearance in space (matter) consists of infinitely many parts, the regression of division is always too great for your concept; and if the division of space is to stop anywhere at some member (the simple), then it is too small for the idea of the unconditioned.

B 516 For this member still leaves a regression to further parts contained in it.

Third, suppose you assume that in everything that happens in the world there is nothing but consequence according to laws of nature. Then the causality of the cause is always in turn something that happens, making your regression to a still higher cause, and hence the extension of the series of conditions a parte priori, necessary without end. Mere efficient nature is therefore too great for any of your concepts in the synthesis of events in the world.

If you choose, here and there, events brought about of themselves, hence production from freedom, then the “why” pursues you according to an unavoidable law of nature and forces you to go beyond this point according to the causal law of experience. You find that such a totality of connection is too small for your necessary empirical concept.

Fourth, if you assume an absolutely necessary being (whether the world itself, something in the world, or the cause of the world), you place it in a time infinitely distant from the given point in time, because otherwise it would depend on another and earlier existence. But then this existence is inaccessible to your empirical concept and too great for you ever to reach it through any continued regression.

B 517 But if, in your view, everything belonging to the world (whether as conditioned or as condition) is contingent, then every existence given to you is too small for your concept. For it forces you to keep looking for another existence on which it depends.

In all these cases we have said that the world-idea is either too great or too small for empirical regression, and hence for every possible concept of the understanding. Why did we not express ourselves conversely and say that in the first case the empirical concept is always too small for the idea, and in the second too great, so that the fault would attach, as it were, to empirical regression, instead of accusing the cosmological idea of departing by excess or deficiency from its aim, namely possible experience? The reason was this. Possible experience alone can give reality to our concepts; without it every concept is only an idea, without truth or relation to an object. Hence the possible empirical concept was the standard by which the idea had to be judged as to whether it was a mere idea and thought-entity, or found its object in the world. For we say something is too great or too small in relation to something else only of what is assumed solely for the sake of that other thing and must be adapted to it.

B 518 Among the playthings of the ancient dialectical schools was this question: if a ball does not pass through a hole, what should we say—is the ball too large or the hole too small? In this case it is immaterial how you choose to express yourself, for you do not know which exists for the sake of the other. By contrast, you will not say that a man is too tall for his coat, but that the coat is too short for the man.

We have therefore at least been led to a well-founded suspicion that cosmological ideas, and with them all the mutually conflicting sophistical assertions, may be based on an empty and merely imagined concept of the way the object of these ideas is given to us. This suspicion can already put us on the right track toward discovering the mirage that has so long led us astray.

The Antinomy of Pure Reason Section Six Transcendental Idealism as the Key to Resolving the Cosmological Dialectic

In the Transcendental Aesthetic we have sufficiently proved that everything intuited in space or time, and thus all objects of an experience possible for us, are nothing but appearances, that is, mere representations.

B 519 As they are represented—as extended beings or series of alterations—they have no existence grounded in themselves outside our thoughts. I call this doctrine transcendental idealism.* The realist in the transcendental sense turns these modifications of our sensibility into things subsisting in themselves, and hence mere representations into things in themselves.

* I have also sometimes called it formal idealism, to distinguish it from material idealism, that is, the ordinary kind, which doubts or denies the existence of external things themselves. In some cases it seems advisable to use this expression rather than the one above, to prevent all misinterpretation.

It would be unjust to attribute to us the long-discredited empirical idealism that assumes space has a reality of its own while denying, or at least doubting, the existence of extended beings in it, and admits no sufficiently demonstrable distinction between dream and truth in this respect. As for the appearances of inner sense in time, it finds no difficulty in accepting them as actual things; indeed, it even maintains that this inner experience alone sufficiently proves the actual existence of its object (in itself), (with all this temporal determination).

B 520 Our transcendental idealism, by contrast, allows that objects of outer intuition are actual just as they are intuited in space, and all alterations in time just as inner sense represents them. For since space is already a form of the intuition we call outer, and without objects in it there would be no empirical representation at all, we can and must assume extended beings in it as actual; the same holds for time. But space itself, together with this time and, along with both, all appearances, are nevertheless not things in themselves, but nothing but representations, and cannot exist outside our mind at all. Even the inner sensible intuition of our mind (as an object of consciousness), whose determination is represented through the succession of different states in time, is not the proper self as it exists in itself, or the transcendental subject, but only an appearance given to the sensibility of this being unknown to us. The existence of this inner appearance as a thing existing in itself in this way cannot be admitted, because its condition is time, which cannot be a determination of any thing in itself. In space and time, however, the empirical truth of appearances is sufficiently secured.

B 521 It is adequately distinguished from affinity with dreams if both are correctly and thoroughly connected in one experience according to empirical laws.

Accordingly, objects of experience are never given in themselves, but only in experience, and do not exist outside it at all. That there could be inhabitants on the moon, although no human being has ever perceived them, must certainly be admitted; but this means only that we could encounter them in the possible advance of experience. For everything is actual that stands in a context with a perception according to laws of empirical progression. Thus they are actual if they stand in an empirical connection with my actual consciousness, although they are not therefore actual in themselves, that is, outside this advance of experience.

Nothing is actually given to us but perception and the empirical advance from it to other possible perceptions. For appearances in themselves, as mere representations, are actual only in perception, which is in fact nothing other than the actuality of an empirical representation, that is, appearance. To call an appearance an actual thing prior to perception means either that in the advance of experience we must encounter such a perception, or it has no meaning at all. For we could certainly say that it exists in itself, without relation to our senses and possible experience, if we were speaking of a thing in itself.

B 522 But we are speaking only of an appearance in space and time, neither of which is a determination of things in themselves, but only of our sensibility. Hence what is in them (appearances) is not something in itself, but mere representations, which, unless given in us (in perception), are found nowhere at all.

The faculty of sensible intuition is properly only a receptivity for being affected in a certain way by representations. Their relation to one another is a pure intuition of space and time (mere forms of our sensibility), and insofar as they are connected and determinable in this relation (space and time) according to laws of the unity of experience, they are called objects. The nonsensible cause of these representations is entirely unknown to us, and we therefore cannot intuit it as an object; for such an object would have to be represented neither in space nor in time (as mere conditions of sensible representation), without which conditions we cannot think any intuition at all. Yet we can call the merely intelligible cause of appearances in general the transcendental object, simply so that we have something corresponding to sensibility as a receptivity. To this transcendental object we can ascribe the whole extent and connection of our possible perceptions.

B 523 We can say that it is given in itself prior to all experience. But in conformity with it, appearances are given not in themselves but only in this experience, because they are mere representations that signify an actual object only as perceptions, namely when this perception is connected with all others according to the rules of the unity of experience. Thus we can say that actual things of past time are given in the transcendental object of experience; but for me they are objects and actual in past time only insofar as I represent that a regressive series of possible perceptions (whether along the guiding thread of history or in the traces of causes and effects), according to empirical laws—in a word, the course of the world—leads to an elapsed temporal series as the condition of the present time. That series is then represented as actual only in the connection of a possible experience, not in itself. Thus all events that have elapsed from time immemorial before my existence signify nothing but the possibility of extending the chain of experience upward from present perception to the conditions determining it in time.

If, accordingly, I represent to myself all existing objects of the senses collectively in all time and all spaces, I do not place them in these prior to experience.

B 524 Rather, this representation is nothing other than the thought of a possible experience in its absolute completeness. In it alone are those objects (which are nothing but mere representations) given. To say that they exist prior to all my experience means only that they are to be found in the part of experience to which I must first advance, beginning from perception. The cause of the empirical conditions of this advance, and hence which members, or how far, I can encounter in such a regression, is transcendental and therefore necessarily unknown to me. But this is not at issue; only the rule of the advance of experience is, in which objects, namely appearances, are given to me. The outcome is exactly the same whether I say that in an empirical advance through space I could encounter stars a hundred times farther away than the most distant ones I see, or say that perhaps such stars are to be found in cosmic space, although no human being has ever perceived or ever will perceive them. For even if they were given as things in themselves, without relation to possible experience generally, they are nothing for me, and hence no objects, except insofar as they are contained in the series of empirical regression. Only in another relation is the distinction significant.

B 525 When these very appearances are to be used for the cosmological idea of an absolute whole, and thus when the question goes beyond the boundaries of possible experience, it matters how we take the actuality of the objects of the senses in question; the distinction then prevents a deceptive delusion that must inevitably arise from misinterpreting our own concepts of experience.

The Antinomy of Pure Reason Section Seven Critical Decision of Reason’s Cosmological Dispute with Itself

The entire antinomy of pure reason rests on the dialectical argument: if the conditioned is given, then the whole series of all its conditions is also given; now objects of the senses are given to us as conditioned; consequently, etc. Through this syllogism, whose major premise seems so natural and evident, as many cosmological ideas are introduced as there are different conditions in the synthesis of appearances insofar as these constitute a series. These ideas postulate the absolute totality of those series and thereby inevitably place reason in conflict with itself. But before exposing the deception in this sophistical argument, we must prepare ourselves.

B 526 We must do so by correcting and determining certain concepts occurring in it.

First, the following proposition is clear and indubitably certain: if the conditioned is given, then by that very fact a regression in the series of all its conditions is set as a task for us. For the concept of the conditioned already entails that something is thereby referred to a condition and, if this is in turn conditioned, to a more remote condition, and so through all members of the series. This proposition is therefore analytic and beyond all fear of a transcendental critique. It is a logical postulate of reason to pursue through the understanding, and continue as far as possible, the connection of a concept with its conditions that already attaches to the concept itself.

Further, if both the conditioned and its condition are things in themselves, then, when the former is given, not merely is regression to the latter assigned as a task; the latter is actually already given along with it. Since this holds for all members of the series, the complete series of conditions, and thus also the unconditioned, is thereby simultaneously given—or rather presupposed, since the conditioned, which was possible only through that series, is given. Here the synthesis of the conditioned with its condition is a synthesis of the mere understanding, which represents things as they are without considering whether and how we can come to know them.

B 527 If, by contrast, I am dealing with appearances, which as mere representations are not given at all unless I come to know them (that is, come to them themselves, for they are nothing but empirical cognitions), then I cannot say in the same sense that if the conditioned is given, all its conditions (as appearances) are also given. Consequently, I can by no means infer the absolute totality of their series. For in apprehension appearances themselves are nothing other than an empirical synthesis (in space and time), and are therefore given only in that synthesis. Now it by no means follows that if the conditioned (in appearance) is given, the synthesis constituting its empirical condition is thereby also given and presupposed. Rather, this synthesis takes place only in the regression, and never without it. But in such a case we can indeed say that a regression to the conditions, that is, a continued empirical synthesis on this side, is commanded or assigned as a task, and that conditions to be given through this regression cannot be lacking.

This shows that the major premise of the cosmological syllogism takes the conditioned in the transcendental sense of a pure category, while the minor premise takes it in the empirical sense of a concept of the understanding applied to mere appearances.

B 528 Consequently, it contains the dialectical deception called Sophisma figurae dictionis. But this deception is not contrived; it is an entirely natural delusion of ordinary reason. For through it we presuppose (in the major premise) the conditions and their series, as it were without looking, whenever something is given as conditioned, because this is nothing other than the logical requirement to assume complete premises for a given conclusion. Here no temporal order is found in the connection of the conditioned with its condition; they are presupposed as given simultaneously in themselves. Further, it is just as natural (in the minor premise) to regard appearances as things in themselves and as objects given to the mere understanding, just as occurred in the major premise, where I abstracted from all conditions of intuition under which alone objects can be given. But here we overlooked a remarkable difference between the concepts. The synthesis of the conditioned with its condition and the whole series of the latter (in the major premise) carried no limitation through time and no concept of succession. Empirical synthesis and the series of conditions in appearance (subsumed in the minor premise), by contrast, are necessarily successive and given only one after another in time. Consequently, I could not presuppose here, as I could there, the absolute totality of synthesis and of the series represented through it.

B 529 For there all members of the series are given in themselves (without a temporal condition), whereas here they are possible only through successive regression, which is given only by actually carrying it out.

Once such a misstep has been demonstrated in the argument that serves as their common basis (for cosmological assertions), both contending parties can rightly be dismissed as grounding their demand on no sound title. But their dispute is not thereby ended in the sense that they, or one of them, have been proved wrong about the matter itself they assert (in the conclusion), even if they did not know how to build it on sound grounds of proof. Nothing seems clearer than that of two people, one asserting that the world has a beginning, the other that it has no beginning but has existed from eternity, one must be right. But if this is so, then, since the clarity is equal on both sides, it is impossible ever to determine which side is right, and the dispute continues as before, even though the parties have been ordered to keep the peace before the tribunal of reason. Thus no means remains of ending the dispute thoroughly and to the satisfaction of both sides except that, since they can refute one another so well, they should finally be convinced that they are disputing about nothing.

B 530 They should be convinced, too, that a certain transcendental illusion has painted an actuality for them where none is to be found.

We will now take this path toward settling a dispute that cannot be adjudicated.

* * *

Zeno of Elea, a subtle dialectician, was already severely censured by Plato as a willful sophist for trying, to display his skill, to prove one and the same proposition through plausible arguments and soon afterward overturn it through others equally strong. He asserted that God (by which he presumably meant nothing but the world) is neither finite nor infinite, neither in motion nor at rest, neither similar nor dissimilar to any other thing. Those who judged him on this thought he meant to deny two mutually contradictory propositions entirely, which is absurd. But I do not find that this charge can rightly be brought against him. I will shortly examine the first of these propositions more closely. As for the others, if by “God” he understood the universe, then he certainly had to say that it neither remains persistently present in its place (at rest) nor changes its place (moves), because all places are only in the universe, which itself is therefore in no place. If the universe contains everything that exists, then in that respect it is also neither similar nor dissimilar to any other thing, because outside it there is no other thing with which it could be compared.

B 531 If two opposed judgments presuppose an inadmissible condition, then despite their conflict (which is nevertheless no genuine contradiction), both fall away, because the condition under which alone either proposition was supposed to hold falls away.

If someone said that every body either smells good or smells not good, a third case is possible: that it has no smell at all (gives off no scent), and thus both conflicting propositions can be false. If I say that it is either fragrant or not fragrant (vel suaveolens vel non suaveolens), the two judgments are contradictorily opposed, and only the first is false; its contradictory opposite, namely that some bodies are not fragrant, also includes bodies that have no smell at all. In the former opposition (per disparata), the contingent condition of the concept of bodies (smell) remained in the conflicting judgment and was not abolished along with it; hence the latter was not the contradictory opposite of the former.

Accordingly, if I say that the world is either infinite in space or is not infinite (non est infinitus), then if the former proposition is false, its contradictory opposite, that the world is not infinite, must be true. I would thereby merely abolish an infinite world without positing another, namely a finite one.

B 532 But if the statement were that the world is either infinite or finite (noninfinite), both could be false. For I then regard the world as determined in its magnitude in itself, since in the opposite I do not merely abolish infinity, and with it perhaps the world’s entire separate existence, but add a determination to the world as a thing actual in itself. This may equally be false if the world should not be given as a thing in itself at all, and thus not in its magnitude either, whether as infinite or finite. Allow me to call such opposition dialectical, while calling the opposition of contradiction analytical. Thus two dialectically opposed judgments can both be false, because one not only contradicts the other but says more than contradiction requires.

If we regard the two propositions “The world is infinite in magnitude” and “The world is finite in magnitude” as contradictorily opposed, we assume that the world (the entire series of appearances) is a thing in itself. For it remains whether I abolish the infinite or the finite regression in the series of its appearances. But if I remove this presupposition, or this transcendental illusion, and deny that the world is a thing in itself.

B 533 The contradictory conflict of the two assertions then becomes merely dialectical. Since the world does not exist in itself at all (independently of the regressive series of my representations), it exists neither as a whole infinite in itself nor as one finite in itself. It is found only in the empirical regression of the series of appearances, and not at all on its own. Hence, if that series is always conditioned, it is never given in its entirety, and the world is therefore no unconditioned whole, nor does it exist as such, whether with infinite or finite magnitude.

What has been said here of the first cosmological idea, namely the absolute totality of magnitude in appearance, holds also for all the others. The series of conditions is found only in the regressive synthesis itself, not in itself in appearance as an independent thing given prior to all regression. Hence I must also say that the multitude of parts in a given appearance is in itself neither finite nor infinite, because appearance is not something existing in itself, and the parts are given only through and within the regression of decomposing synthesis. This regression is never given absolutely in its entirety, either as finite or as infinite. The same holds for the series of causes ordered one above another, or of conditioned existence up to unconditionally necessary existence.

B 534 This series can never be regarded in its totality as finite or infinite in itself, because as a series of subordinated representations it consists only in the dynamical regression; prior to this, and as a self-subsisting series of things, it cannot exist in itself at all.

Thus the antinomy of pure reason in its cosmological ideas is removed by showing that it is merely dialectical: a conflict of an illusion arising from applying the idea of absolute totality, which holds only as a condition of things in themselves, to appearances. These exist only in representation and, when they constitute a series, in successive regression; otherwise they do not exist at all. Conversely, however, we can also draw a genuine benefit from this antinomy, not dogmatic but nevertheless critical and doctrinal: indirectly proving the transcendental ideality of appearances, should anyone find the direct proof in the Transcendental Aesthetic insufficient. The proof would consist in this dilemma. If the world is a whole existing in itself, it is either finite or infinite. Now both the former and the latter are false (according to the proofs of the antithesis on the one hand and the thesis on the other, given above).

B 535 Thus it is also false that the world (the sum total of all appearances) is a whole existing in itself. From this it follows that appearances in general are nothing outside our representations, which is precisely what we meant by their transcendental ideality.

This remark is important. It shows that the above proofs of the fourfold antinomy were not deceptions but sound, under the presupposition that appearances, or a sensible world embracing them all, were things in themselves. But the conflict of the propositions inferred from this reveals a falsity in the presupposition, and thereby leads us to discover the true constitution of things as objects of the senses. Transcendental dialectic therefore gives no support whatever to skepticism, but does support the skeptical method, which can display in it an example of its great usefulness. For if we allow reason’s arguments to confront one another with the greatest freedom, then although in the end they may not deliver what we sought, they will always deliver something useful and conducive to correcting our judgments.

B 536 The Antinomy of Pure Reason Section Eight The Regulative Principle of Pure Reason with Regard to Cosmological Ideas

Since the cosmological principle of totality gives no maximum of the series of conditions in a sensible world as a thing in itself, but can only assign it as a task in the regression through that series, the stated principle of pure reason, in its meaning thus corrected, still retains its full validity. It does so not as an axiom for thinking totality as actual in the object, but as a problem for the understanding, hence for the subject, to undertake and continue the regression in the series of conditions for a given conditioned in conformity with completeness in the idea. For in sensibility, that is, in space and time, every condition we can reach in expounding given appearances is itself conditioned, because these are not objects in themselves, in which the absolutely unconditioned might perhaps occur, but merely empirical representations that must always find their condition in intuition, determining them in space or time. The principle of reason is therefore properly only a rule that commands a regression in the series of conditions of given appearances.

B 537 This regression is never permitted to stop at something absolutely unconditioned. Thus it is no principle of the possibility of experience and of empirical cognition of objects of the senses, and hence no principle of the understanding; for every experience is enclosed within its boundaries (according to the given intuition). Nor is it a constitutive principle of reason for extending the concept of the sensible world beyond all possible experience. Rather, it is a principle of the greatest possible continuation and extension of experience, according to which no empirical boundary must count as an absolute boundary. It is therefore a principle of reason that, as a rule, postulates what we are to do in the regression, and does not anticipate what is given in itself in the object prior to all regression. Hence I call it a regulative principle of reason. By contrast, the principle of the absolute totality of the series of conditions as given in itself in the object (appearances) would be a constitutive cosmological principle. Through this very distinction I have sought to indicate its nullity, and thereby to prevent what otherwise inevitably happens (through transcendental subreption): attributing objective reality to an idea that serves merely as a rule.

To determine the meaning of this rule of pure reason properly, we must first note the following.

B 538 It cannot say what the object is, but how the empirical regression is to be undertaken in order to arrive at the complete concept of the object. For if the former occurred, it would be a constitutive principle, such as can never be possible from pure reason. Thus we can by no means intend to say through it that the series of conditions for a given conditioned is in itself finite or infinite. For then a mere idea of absolute totality, created solely within itself, would think an object that can be given in no experience, by conferring on a series of appearances an objective reality independent of empirical synthesis. The idea of reason will therefore prescribe only a rule to regressive synthesis in the series of conditions, according to which it proceeds from the conditioned through all subordinated conditions to the unconditioned, although the latter is never reached. For the absolutely unconditioned is not found in experience at all.

For this purpose we must first determine precisely the synthesis of a series insofar as it is never complete. Two expressions are commonly used here, supposedly to distinguish something in it, though no one quite knows how to specify the ground of the distinction. Mathematicians speak solely of a progressus in infinitum.

B 539 Those who investigate concepts (philosophers) would allow instead only the expression of a progressus in indefinitum. Without pausing to examine the scruple that recommended such a distinction to them, or its useful or fruitless employment, I will try to determine these concepts precisely in relation to my purpose.

We can rightly say of a straight line that it can be extended to infinity; here the distinction between infinite and indefinitely far progression (progressus in indefinitum) would be an empty subtlety. Admittedly, if the instruction is “Extend a line,” it is more correct to add in indefinitum than to say in infinitum: the former means no more than “Extend it as far as you wish,” the latter “You must never stop extending it” (which is not the intention here). But when the question concerns only what can be done, the former expression is entirely correct; for you can keep making it greater to infinity. The same holds in all cases where we speak only of progression, that is, of proceeding from the condition to the conditioned: this possible advance in the series of appearances goes to infinity. From a pair of parents you can proceed without end in the descending line of generation, and can quite well think that it actually proceeds in this way in the world.

B 540 For here reason never needs absolute totality of the series, since it does not presuppose this as a condition and as given (datum), but only as something conditioned that can be given (dabile) and is added without end.

The matter is quite different with the problem of how far the regression ascending from the given conditioned to the conditions in a series extends: whether I can say it is a regression to infinity, or only one extending indefinitely far (in indefinitum); and thus whether I can ascend to infinity from the people now living in the series of their ancestors, or whether we can say only that, however far I have gone back, no empirical ground is ever found for regarding the series as bounded anywhere, so that I am entitled and at the same time obliged to seek a further forebear for each ancestor, although not actually to presuppose one.

Accordingly, I say: if the whole has been given in empirical intuition, regression in the series of its inner conditions proceeds to infinity. But if only one member of the series is given, from which regression is first to proceed toward absolute totality, then only a regression in an indeterminate manner (in indefinitum) takes place.

B 541 Thus of the division of a matter (a body) given within its boundaries, we must say that it proceeds to infinity. For this matter is given as a whole, and consequently with all its possible parts, in empirical intuition. Now since the condition of this whole is its part, and the condition of this part the part of that part, and so on, and since in this regression of decomposition no unconditioned (indivisible) member of this series of conditions is ever encountered, not only is there nowhere any empirical ground for stopping the division, but the further members of the division to be continued are themselves empirically given even before that further division: that is, the division proceeds to infinity. By contrast, the series of ancestors of a given person is given in its absolute totality in no possible experience; yet regression proceeds from each member of this generation to a higher one, so that no empirical boundary is found presenting a member as absolutely unconditioned. But since the members that could supply the condition are not already present in the empirical intuition of the whole before the regression, this regression proceeds not to infinity (in dividing what is given), but indefinitely far in seeking further members for the given ones, which are themselves always given only as conditioned.

B 542 In neither of the two cases, the regressus in infinitum as well as that in indefinitum, is the series of conditions regarded as given as infinite in the object. They are not things given in themselves, but only appearances given, as conditions of one another, solely in the regression itself. Thus the question is no longer how great this series of conditions is in itself, whether finite or infinite—for it is nothing in itself—but how we should undertake empirical regression and how far we should continue it. And here there is a significant difference in the rule for this advance. If the whole has been empirically given, it is possible to regress to infinity in the series of its inner conditions. But if the whole is not given and is first to be given through empirical regression, I can say only that it is infinitely possible to proceed to still higher conditions of the series. In the first case I could say that there are always more members present and empirically given than I reach through the regression (of decomposition); in the second, that I can always go further in regression, because no member is empirically given as absolutely unconditioned, and thus it always admits a still higher member as possible and hence the search for it as necessary. There it was necessary to encounter more members of the series; here it is always necessary to inquire after more, because no experience sets an absolute boundary.

B 543 For either you have no perception that absolutely bounds your empirical regression, in which case you must not regard your regression as completed; or you have a perception bounding your series, in which case it cannot be part of the series you have traversed (because what bounds must be distinct from what it bounds), and you must therefore continue your regression to this condition as well, and so onward.

The following section will put these remarks in their proper light through their application.

The Antinomy of Pure Reason Section Nine The Empirical Use of the Regulative Principle of Reason with Regard to All Cosmological Ideas

As we have repeatedly shown, there is no transcendental use either of pure concepts of the understanding or of concepts of reason. The absolute totality of the series of conditions in the sensible world rests solely on a transcendental use of reason, which demands this unconditioned completeness of what it presupposes as a thing in itself; but the sensible world contains no such thing.

B 544 Hence the issue can no longer be the absolute magnitude of the series in it, whether bounded or unbounded in themselves, but only how far we should go back in empirical regression, tracing experience to its conditions, so that, according to reason’s rule, we stop only at an answer to its questions commensurate with the object.

Thus all that remains to us is the validity of reason’s principle as a rule for the continuation and magnitude of a possible experience, after its invalidity as a constitutive principle of appearances in themselves has been sufficiently demonstrated. If we can exhibit that validity beyond doubt, reason’s dispute with itself will also be wholly ended. For not only will critical resolution have removed the illusion that divided reason against itself; in its place, the meaning in which reason agrees with itself, whose misinterpretation alone occasioned the dispute, will be disclosed, and an otherwise dialectical principle transformed into a doctrinal one. Indeed, suppose this principle can be vindicated in its subjective meaning—determining the greatest possible use of the understanding in experience in conformity with its objects.

B 545 Then it is precisely as much as if it determined objects in themselves a priori like an axiom (which is impossible from pure reason). For even this could have no greater influence on extending and correcting our cognition of objects of experience than proving itself active in the most extensive experiential use of our understanding.

I. Resolution of the Cosmological Idea of the Totality of the Composition of Appearances of a World-Whole

Here, as with the other cosmological questions, the ground of reason’s regulative principle is the proposition that in empirical regression no experience of an absolute boundary can be found, and hence of no condition as empirically absolutely unconditioned. The reason is that such an experience would have to contain a bounding of appearances by nothing, or by the void, which continued regression could encounter through a perception. This is impossible.

Now this proposition amounts to saying that in empirical regression I always reach only a condition.

B 546 That condition must itself in turn be regarded as empirically conditioned. The proposition thus contains the rule in terminis that, however far I may have come in the ascending series, I must always inquire after a higher member of the series, whether or not it becomes known to me through experience.

Nothing further is now needed to resolve the first cosmological problem except to determine whether, in regression toward the unconditioned magnitude of the world-whole (in time and space), this never-bounded ascent can be called a regression to infinity, or only an indefinitely continued regression (in indefinitum).

The mere general representation of the series of all past states of the world, and likewise of the things simultaneously present in cosmic space, is itself nothing other than a possible empirical regression that I think, though still indeterminately, and through which alone the concept of such a series of conditions for the given perception can arise*.

B 547 Now I always have the world-whole only in a concept, never in intuition (as a whole). Thus I cannot infer from its magnitude the magnitude of the regression and determine the latter according to the former; rather, I must first form a concept of the world’s magnitude through the magnitude of empirical regression. But of this regression I never know more than that from every given member of the series of conditions I must always proceed empirically to a higher (more remote) member. Thus the magnitude of the whole of appearances is not thereby determined absolutely at all. Consequently, we also cannot say that this regression proceeds to infinity, because this would anticipate the members the regression has not yet reached and represent their multitude as so great that no empirical synthesis can reach it. It would therefore determine the world’s magnitude before the regression (even if only negatively), which is impossible. For this is given to me through no intuition (in its totality), and hence its magnitude too is not given at all before the regression. Accordingly, we can say nothing whatever about the world’s magnitude in itself, not even that a regressus in infinitum takes place within it; we must seek the concept of its magnitude only according to the rule determining empirical regression in it. But this rule says no more than that, however far we may have come in the series of empirical conditions, we should nowhere assume an absolute boundary.

B 548 Rather, we must subordinate every appearance, as conditioned, to another as its condition, and thus proceed further to it. This is the regressus in indefinitum, which, because it determines no magnitude in the object, can be distinguished clearly enough from that in infinitum.

* This world-series can therefore be neither greater nor smaller than the possible empirical regression on which alone its concept rests. And since this can yield neither a determinate infinite nor a determinate finite (something absolutely bounded), it is clear that we can assume the world’s magnitude neither as finite nor as infinite, because the regression (through which it is represented) admits neither.

Accordingly, I cannot say that the world is infinite in past time or in space. For such a concept of magnitude as a given infinity is absolutely impossible empirically, and hence also with regard to the world as an object of the senses. Nor will I say that regression from a given perception to everything that bounds it in a series, both in space and in past time, proceeds to infinity; for this presupposes the world’s infinite magnitude. Nor will I say that it is finite, for an absolute boundary is likewise empirically impossible. Thus I will be able to say nothing about the entire object of experience (the sensible world), but only about the rule according to which experience should be undertaken and continued in conformity with its object.

To the cosmological question concerning the world’s magnitude, then, the first and negative answer is: the world has no first beginning in time and no outermost boundary in space.

For in the opposite case it would be bounded by empty time on the one hand and empty space on the other.

B 549 Since, as appearance, it can be neither of these in itself—for appearance is no thing in itself—a perception of its bounding by absolutely empty time or empty space would have to be possible, through which these ends of the world would be given in a possible experience. But such an experience, as wholly empty of content, is impossible. Thus an absolute boundary of the world is empirically, and hence absolutely, impossible*.

* It will be noticed that the proof here has been conducted in an entirely different way from the dogmatic proof above in the antithesis of the first antinomy. There we allowed the sensible world, according to the ordinary and dogmatic manner of representation, to count as a thing given in itself in its totality prior to all regression; and if it did not occupy all time and all spaces, we denied it any determinate position in either at all. Hence the inference was also different from the present one: we inferred its actual infinity.

From this follows at the same time the affirmative answer: regression in the series of appearances of the world, as a determination of the world’s magnitude, proceeds in indefinitum. This amounts to saying that the sensible world has no absolute magnitude, but empirical regression (through which alone it can be given on the side of its conditions) has its rule. From every member of the series, as conditioned, it must always proceed to a still more remote one.

B 550 It must do so whether through one’s own experience, or the guiding thread of history, or the chain of effects and their causes, and must nowhere exempt itself from extending the possible empirical use of one’s understanding. This is, indeed, reason’s proper and sole business with its principles.

This prescribes no determinate empirical regression proceeding incessantly within a particular kind of appearances—for example, that from a living person we must always ascend in a series of ancestors without expecting a first pair, or in the series of celestial bodies without admitting an outermost sun. Rather, it commands only progression from appearances to appearances, even if these yield no actual perception (when they are too weak in degree for our consciousness to become experience), because they nevertheless belong to possible experience.

Every beginning is in time, and every boundary of the extended is in space. But space and time are only in the sensible world. Consequently, only appearances in the world are bounded conditionally; the world itself is bounded neither conditionally nor unconditionally.

For this very reason, and because the world can never be given in its entirety, nor even the series of conditions for a given conditioned as a world-series, the concept of the world’s magnitude is given only through regression, not prior to it in a collective intuition.

B 551 But regression always consists only in determining magnitude, and therefore yields no determinate concept, nor any concept of a magnitude infinite in relation to a certain measure. It thus proceeds not to infinity (as though already given), but indefinitely far, in order to yield a magnitude (of experience) that first becomes actual through this regression.

II. Resolution of the Cosmological Idea of the Totality of the Division of a Given Whole in Intuition

When I divide a whole given in intuition, I proceed from a conditioned to the conditions of its possibility. The division of the parts (subdivisio or decompositio) is a regression in the series of these conditions. The absolute totality of this series would be given only if the regression could reach simple parts. But if, in a continuously proceeding decomposition, all parts are always divisible in turn, then division—that is, regression from the conditioned to its conditions—proceeds in infinitum, because the conditions (the parts) are contained in the conditioned itself and, since the latter is given in its entirety in an intuition enclosed within its boundaries, are all given along with it.

B 552 Regression must therefore not be called merely a regression in indefinitum, as the preceding cosmological idea alone allowed, where I was to proceed from the conditioned to its conditions outside it, which were consequently not thereby given along with it but were first added in empirical regression. Despite this, we are by no means permitted to say of such an infinitely divisible whole that it consists of infinitely many parts. For although all parts are contained in the intuition of the whole, the entire division is not contained in it: this consists only in the continuing decomposition, or regression itself, which first makes the series actual. Since this regression is infinite, all members (parts) it reaches are indeed contained in the given whole as aggregates, but not the entire series of division, which is successively infinite and never entire, and therefore can represent neither an infinite multitude nor its collection in a whole.

This general reminder is first very easily applied to space. Every space intuited within its boundaries is such a whole, whose parts, throughout all decomposition, are always spaces in turn, and it is therefore divisible to infinity.

B 553 From this follows quite naturally the second application, to an outer appearance enclosed within its boundaries (a body). Its divisibility rests on the divisibility of space, which constitutes the possibility of the body as an extended whole. The body is therefore divisible to infinity, without thereby consisting of infinitely many parts.

It admittedly seems that, since a body must be represented as substance in space, it will differ from space with regard to the law of divisibility. For we can perhaps concede that decomposition in space can never remove all composition, since then even all space, which otherwise has nothing self-subsisting, would cease (which is impossible). But that, if all composition of matter were abolished in thought, nothing at all should remain seems irreconcilable with the concept of a substance, which should properly be the subject of all composition and would have to remain in its elements even if their connection in space, through which they constitute a body, were abolished. But what is called substance in appearance is not constituted as we might think a thing in itself through a pure concept of the understanding. It is not an absolute subject.

B 554 It is an enduring image of sensibility and nothing but intuition, in which nothing unconditioned is found anywhere.

Although this rule of progression to infinity undoubtedly holds in subdividing an appearance as a mere filling of space, it cannot hold if we also wish to extend it to the multitude of parts already separated in a certain way within the given whole, thereby constituting a quantum discretum. We cannot think that in every articulated (organized) whole each part is itself articulated, and that in this way, in analyzing the parts to infinity, we always encounter new artfully formed parts—in a word, that the whole is articulated to infinity—although we can indeed think that the parts of matter could be articulated in its decomposition to infinity. For the infinity of the division of a given appearance in space rests solely on the fact that this appearance gives only divisibility, that is, a multitude of parts absolutely indeterminate in itself, while the parts themselves are given and determined only through subdivision: in short, the whole is not already divided in itself. Hence division can determine a multitude in it extending as far as we wish to proceed in the regression of division.

B 555 By contrast, in an organic body articulated to infinity, the whole is represented through this very concept as already divided, and a multitude of parts determinate in itself but infinite is found in it prior to all regression of division, whereby one contradicts oneself, since this infinite unfolding is regarded as a series that can never be completed (infinite), yet also as completed in a collection. Infinite division designates only appearance as quantum continuum and is inseparable from the filling of space, because in that filling lies the ground of infinite divisibility. But as soon as something is assumed as quantum discretum, the multitude of units in it is determined, and hence always equal to a number. Thus how far organization may extend in an articulated body can be determined only by experience; and even if experience never reached an inorganic part with certainty, such parts must nevertheless lie at least in possible experience. But how far the transcendental division of an appearance in general extends is no matter of experience at all. It is a principle of reason never to regard empirical regression in decomposing the extended as absolutely completed, in conformity with the nature of this appearance.

B 556 Concluding Remark on the Resolution of the Mathematical-Transcendental Ideas, and Preliminary Reminder Concerning the Resolution of the Dynamical-Transcendental Ideas

When we presented the antinomy of pure reason through all transcendental ideas in a table, indicating the ground of this conflict and the only means of removing it—declaring both opposed assertions false—we everywhere represented conditions as belonging to their conditioned according to relations of space and time. This is the usual presupposition of ordinary human understanding, on which that conflict entirely rested. In this respect, all dialectical representations of totality in the series of conditions for a given conditioned were also thoroughly of the same kind. There was always a series in which condition and conditioned were connected as its members and were thereby homogeneous; hence regression could never be thought as completed, or, if it were to be so thought, an intrinsically conditioned member would have to be falsely assumed as first, and thus as unconditioned. Thus, although not everywhere the object, that is, the conditioned, was considered merely in its magnitude, the series of its conditions was.

B 557 The difficulty, which could be removed not by any compromise but only by cutting the knot entirely, was that reason made the series either too long or too short for the understanding, so that the latter could never match its idea.

But here we overlooked an essential distinction among the objects, that is, the concepts of the understanding, which reason seeks to elevate to ideas. According to our table of categories above, two of them signify a mathematical synthesis of appearances, while the other two signify a dynamical synthesis. Until now this could quite well be done, since, just as in the general representation of all transcendental ideas we always remained only among conditions in appearance, so too in the two mathematical-transcendental ideas we had no object other than one in appearance. But now, as we proceed to dynamical concepts of the understanding insofar as they are to conform to the idea of reason, that distinction becomes important and opens an entirely new prospect concerning the dispute in which reason is entangled. Previously dismissed as built on false presuppositions on both sides, it may now admit settlement to the satisfaction of both parties from this standpoint.

B 558 Perhaps in the dynamical antinomy such a presupposition occurs that can coexist with reason’s claim, while the judge supplies the deficiency in legal grounds that both sides had misunderstood. This could not be done in the dispute of the mathematical antinomy.

The series of conditions are indeed all homogeneous insofar as we consider only their extent: whether they conform to the idea, or whether the idea is too great or too small for them. But the concept of the understanding underlying these ideas contains either solely a synthesis of the homogeneous (which is presupposed in every magnitude, both in its composition and in its division), or also a synthesis of the heterogeneous, which can at least be admitted in dynamical synthesis, both in causal combination and in the combination of the necessary with the contingent.

Thus, in the mathematical connection of series of appearances, no condition can enter except a sensible one, that is, one that is itself part of the series. The dynamical series of sensible conditions, by contrast, still admits a heterogeneous condition that is not part of the series but, as merely intelligible, lies outside it.

B 559 This satisfies reason and places the unconditioned before appearances without thereby confusing their series, which is always conditioned, or breaking it off contrary to principles of the understanding.

Now because dynamical ideas admit a condition of appearances outside their series, that is, one that is not itself appearance, something occurs wholly different from the outcome of the antinomy. The antinomy required both dialectical counterassertions to be declared false. By contrast, the thoroughly conditioned character of dynamical series, inseparable from them as appearances, connected with a condition that is empirically unconditioned but also nonsensible, can satisfy the understanding on the one hand and reason on the other.* The dialectical arguments that sought unconditioned totality in mere appearances in one way or another fall away.

B 560 The propositions of reason, by contrast, can both be true in their meaning corrected in this way. This can never occur with cosmological ideas concerning merely mathematically unconditioned unity, because in them no condition of the series of appearances is found except one that is itself appearance and, as such, also constitutes a member of the series.

* For the understanding allows no condition among appearances that would itself be empirically unconditioned. But if an intelligible condition, thus one not belonging to the series of appearances as a member, could be thought for a conditioned (in appearance) without in the least interrupting the series of empirical conditions, then such a condition could be admitted as empirically unconditioned, so that continuous empirical regression would nowhere be impaired.

III. Resolution of the Cosmological Ideas of the Totality of the Derivation of Events in the World from Their Causes

With regard to what happens, we can think only two kinds of causality: according to nature or from freedom. The first is the connection of a state with a preceding one in the sensible world, on which it follows according to a rule. Now since the causality of appearances rests on temporal conditions, and the preceding state, if it had always existed, would also have produced no effect first arising in time, the causality of the cause of what happens or arises has itself also arisen and, according to the principle of the understanding, itself requires a cause in turn.

B 561 By freedom in the cosmological sense, by contrast, I understand the power to begin a state of itself, whose causality therefore does not in turn stand under another cause determining it in time according to the law of nature. Freedom in this sense is a pure transcendental idea, which, first, contains nothing borrowed from experience and, second, has an object that also cannot be given determinately in any experience. For it is a universal law, even of the possibility of all experience, that everything that happens must have a cause; hence the causality of the cause, which itself has happened or arisen, must in turn have a cause. Thereby the whole field of experience, however far it may extend, is transformed into a sum total of mere nature. But since in this way no absolute totality of conditions in the causal relation can be obtained, reason creates for itself the idea of a spontaneity that can begin to act of itself, without another cause being permitted to be placed before it to determine it in turn to action according to the law of causal connection.

It is exceedingly noteworthy that the practical concept of freedom is grounded on this transcendental idea, and that the latter constitutes in it the real source of the difficulties that have always surrounded the question of its possibility.

B 562 Freedom in the practical sense is the independence of choice from compulsion by impulses of sensibility. For a power of choice is sensible insofar as it is pathologically affected (through motives of sensibility); it is called animal (arbitrium brutum) if it can be pathologically necessitated. Human choice is indeed an arbitrium sensitivum, but not brutum, rather liberum, because sensibility does not necessitate its action, and the human being possesses a power to determine himself of himself, independently of compulsion by sensible impulses.

It is easy to see that if all causality in the sensible world were mere nature, every event would be determined by another in time according to necessary laws. Consequently, since appearances, insofar as they determine choice, would have to necessitate every action as their natural consequence, abolishing transcendental freedom would also annihilate all practical freedom. For the latter presupposes that although something did not happen, it nevertheless ought to have happened, and hence its cause in appearance was not so determining as to exclude a causality in our choice independent of those natural causes, capable even against their power and influence of producing something determined in the temporal order according to empirical laws, and thus of beginning a series of events entirely of itself.

B 563 What occurs here, then, is what we find generally in the conflict of a reason venturing beyond the boundaries of possible experience: the problem is properly not physiological, but transcendental. Hence the question of the possibility of freedom does indeed trouble psychology, but because it rests on dialectical arguments of merely pure reason, it must, together with its solution, occupy transcendental philosophy alone. To enable that philosophy—which cannot decline to give a satisfactory answer—to address it, I must first try to determine its procedure in this problem more closely through a remark.

If appearances were things in themselves, and hence space and time forms of the existence of things in themselves, the conditions would always belong with the conditioned as members of one and the same series. From this there would arise in the present case, too, the antinomy common to all transcendental ideas: this series would inevitably turn out too great or too small for the understanding. But the dynamical concepts of reason with which we are concerned in this and the following subsection have this peculiarity: since they concern not an object considered as magnitude but only its existence, we can also abstract from the magnitude of the series of conditions.

B 564 What matters in them is solely the dynamical relation of condition to conditioned. Thus in the question of nature and freedom we already encounter the difficulty whether freedom is possible at all and, if it is, whether it can coexist with the universality of the natural law of causality; hence whether it is a correctly disjunctive proposition that every effect in the world must arise either from nature or from freedom, or whether both might instead occur simultaneously in different relations in one and the same event. The correctness of the principle of the thoroughgoing connection of all events in the sensible world according to immutable laws of nature is already established as a principle of the Transcendental Analytic and admits no impairment. The only question, then, is whether, despite this, freedom may also occur with regard to the very same effect determined according to nature, or whether that inviolable rule wholly excludes it. Here the common but deceptive presupposition of the absolute reality of appearances at once shows its harmful influence in confusing reason. For if appearances are things in themselves, freedom cannot be saved. Nature is then the complete and intrinsically sufficient determining cause of every event, and its condition is always contained only in the series of appearances, which, together with their effect, are necessary under every law of nature.

B 565 If, by contrast, appearances count as no more than they actually are—not things in themselves, but mere representations connected according to empirical laws—then they must themselves still have grounds that are not appearances. But such an intelligible cause is not determined in its causality by appearances, although its effects appear and can thus be determined by other appearances. It therefore stands, together with its causality, outside the series, while its effects are found in the series of empirical conditions. The effect can thus be regarded as free in relation to its intelligible cause, and yet at the same time, in relation to appearances, as a consequence of them according to the necessity of nature. This distinction must seem extremely subtle and obscure when presented generally and quite abstractly, but it will become clear in application. Here I wished only to note that, since the thoroughgoing connection of all appearances in a context of nature is an indispensable law, it would necessarily overthrow all freedom if we stubbornly clung to the reality of appearances. Hence those who follow the common opinion here have never been able to reconcile nature and freedom.

B 566 The Possibility of Causality through Freedom in Union with the Universal Law of Natural Necessity

I call that in an object of the senses which is not itself appearance intelligible. Accordingly, if what must be regarded as appearance in the sensible world also has in itself a power that is no object of sensible intuition, but through which it can nevertheless be the cause of appearances, then the causality of this being can be considered from two sides: as intelligible in its action as a thing in itself, and as sensible in its effects as an appearance in the sensible world. We would accordingly form both an empirical and an intellectual concept of the causality of such a subject’s power, both occurring together in one and the same effect. Thinking the power of an object of the senses from such a twofold aspect contradicts none of the concepts we must form of appearances and of a possible experience. For appearances, being no things in themselves, must have underlying them a transcendental object that determines them as mere representations.

B 567 Nothing therefore prevents us from attributing to this transcendental object, besides the property through which it appears, a causality that is not appearance, although its effect is nevertheless found in appearance. But every efficient cause must have a character, that is, a law of its causality, without which it would not be a cause at all. Thus in a subject of the sensible world we would have, first, an empirical character through which its actions, as appearances, would be thoroughly connected with other appearances according to constant laws of nature and could be derived from them as their conditions. Combined with these, they would therefore constitute members of a single series in the natural order. Second, we would also have to grant it an intelligible character, through which it is indeed the cause of those actions as appearances, but which itself stands under no conditions of sensibility and is not itself appearance. We could also call the former the character of such a thing in appearance, the latter the character of the thing in itself.

Now this acting subject, according to its intelligible character, would stand under no temporal conditions, for time is only the condition of appearances, not of things in themselves. In it no action would arise or pass away.

B 568 Hence it would also not be subject to the law of all temporal determination, of everything alterable: that everything that happens finds its cause in appearances (of the preceding state). In a word, its causality, insofar as it is intellectual, would not stand at all in the series of empirical conditions that necessitate the event in the sensible world. This intelligible character could indeed never be known immediately, because we can perceive nothing except insofar as it appears; but it would nevertheless have to be thought in conformity with the empirical character, just as we must generally think a transcendental object as underlying appearances, although we know nothing of what it is in itself.

According to its empirical character, then, this subject as appearance would be subject to all laws of determination through causal combination, and to that extent would be nothing but a part of the sensible world whose effects, like every other appearance, flowed inevitably from nature. Once we knew through experience how outer appearances influenced it and what its empirical character—that is, the law of its causality—was, all its actions would have to be explicable according to laws of nature, and all requirements for their complete and necessary determination would have to be found in a possible experience.

B 569 But according to its intelligible character (of which we can have nothing but its general concept), that same subject would nevertheless have to be declared free from all influence of sensibility and determination through appearances. And since in it, insofar as it is a noumenon, nothing happens—no alteration requiring dynamical temporal determination, and hence no connection with appearances as causes is found—this active being would to that extent be independent and free in its actions from all natural necessity, which is found solely in the sensible world. We would say of it quite correctly that it begins its effects in the sensible world of itself, without the action beginning in itself. This would hold without the effects in the sensible world therefore being allowed to begin of themselves, because in that world they are always predetermined through empirical conditions in the preceding time, though only by means of the empirical character (which is merely the appearance of the intelligible character), and are possible only as a continuation of the series of natural causes. Thus freedom and nature, each in its full meaning, would be found simultaneously and without any conflict in the very same actions, according as we compare them with their intelligible or sensible cause.

B 570 Elucidation of the Cosmological Idea of Freedom in Combination with Universal Natural Necessity

I thought it advisable first to sketch the outline of the solution to our transcendental problem, so that the course of reason in solving it might be more easily surveyed. We will now set out the factors on which its decision properly depends and consider each separately.

The law of nature states that everything that happens has a cause; that the causality of this cause, that is, the action, since it precedes in time and, in view of an effect that has arisen, cannot itself always have existed but must have happened, also has its cause among appearances through which it is determined; and that consequently all events are empirically determined in a natural order. This law, through which alone appearances can constitute a nature and provide objects of an experience, is a law of the understanding that we are permitted under no pretext to depart from, or to exempt any appearance from. Otherwise we would place that appearance outside all possible experience, and would thereby distinguish it from all objects of possible experience.

B 571 It would be made into a mere thought-entity and a fantasy.

Although this presents nothing but a chain of causes that admits no absolute totality in regression to its conditions, this difficulty does not detain us at all; for it has already been removed in the general assessment of reason’s antinomy when it seeks the unconditioned in the series of appearances. If we choose to yield to the delusion of transcendental realism, neither nature nor freedom remains. Here the only question is whether, if we acknowledge sheer natural necessity throughout the entire series of all events, it is nevertheless possible to regard the very same event, which on one side is merely a natural effect, on the other as an effect from freedom, or whether there is a direct contradiction between these two kinds of causality.

Among causes in appearance there can certainly be nothing capable of beginning a series absolutely and of itself. Every action as appearance, insofar as it produces an event, is itself an event or occurrence that presupposes another state in which its cause is found. Thus everything that happens is only a continuation of the series, and no beginning occurring of itself is possible within it.

B 572 All actions of natural causes in temporal succession are therefore themselves effects in turn, which likewise presuppose their causes in the temporal series. An original action through which something happens that did not exist before is not to be expected from the causal connection of appearances.

But is it also necessary that, if effects are appearances, the causality of their cause—which is itself also appearance—must be merely empirical? Is it not rather possible that, although every effect in appearance certainly requires a connection with its cause according to laws of empirical causality, this empirical causality itself could nevertheless, without in the least interrupting its connection with natural causes, be an effect of a causality that is not empirical but intelligible? That is, an action original with regard to appearances, of a cause that is therefore to that extent not appearance but intelligible in this power, although in other respects it must be counted wholly as a member of the chain of nature belonging to the sensible world.

We need the proposition of the causality of appearances among one another in order to seek and specify natural conditions—that is, causes in appearance—for natural events. If this is conceded and weakened by no exception, the understanding has everything it can demand.

B 573 In its empirical use it sees nothing but nature in all occurrences, and is entitled to do so. Physical explanations continue their unhindered course. Now it does not impair the understanding in the least—even supposing the assumption were otherwise a mere invention—if we assume that among natural causes there are some possessing a power that is only intelligible, since its determination to action never rests on empirical conditions but on mere grounds of the understanding, provided that the action in appearance arising from this cause conforms to all laws of empirical causality. For in this way the acting subject, as causa phaenomenon, would be linked with nature in the unbroken dependence of all its actions, and only the phaenomenon of this subject (with all its causality in appearance) would contain certain conditions that, if we wish to ascend from the empirical object to the transcendental one, would have to be regarded as merely intelligible. For if we only follow the rule of nature in whatever may be the cause among appearances, we can remain unconcerned about what ground of these appearances and their connection is thought in the transcendental subject, which is empirically unknown to us. This intelligible ground does not affect empirical questions at all, but perhaps concerns only thinking in the pure understanding.

B 574 Although the effects of this thinking and acting of the pure understanding are found in appearances, they must nevertheless be capable of complete explanation from their cause in appearance according to laws of nature, by following the merely empirical character of that cause as the highest ground of explanation and passing over its intelligible character, the transcendental cause of the former, as entirely unknown, except insofar as it is indicated through the empirical character as its sensible sign. Let us apply this to experience. The human being is one of the appearances of the sensible world, and thus also one of the natural causes whose causality must stand under empirical laws. As such, he must therefore also have an empirical character, like all other natural things. We observe it through the forces and faculties it manifests in its effects. In inanimate nature, or nature animated only in an animal way, we find no ground for thinking any faculty otherwise than as merely sensibly conditioned. But the human being, who otherwise knows the whole of nature solely through the senses, also cognizes himself through mere apperception, and indeed in actions and inner determinations that he cannot count among impressions of the senses at all. He is to himself certainly in one respect a phenomenon, but in another, namely with regard to certain faculties, a merely intelligible object, because their action cannot be counted as belonging to the receptivity of sensibility at all.

B 575 We call these faculties understanding and reason; especially the latter is properly and preeminently distinguished from all empirically conditioned forces, since it considers its objects solely according to ideas and determines the understanding accordingly, which then makes an empirical use of its concepts (though these, too, are pure).

That this reason has causality, or at least that we represent such causality in it, is clear from the imperatives we prescribe as rules to the executive forces in all practical matters. The “ought” expresses a kind of necessity and connection with grounds that occurs nowhere else in the whole of nature. Of nature, the understanding can cognize only what is, has been, or will be. It is impossible that something in nature ought to be other than it actually is in all these temporal relations; indeed, if we have only the course of nature before us, the “ought” has no meaning at all. We cannot ask what ought to happen in nature, any more than what properties a circle ought to have, but only what happens in nature or what properties the circle has.

Now this “ought” expresses a possible action whose ground is nothing other than a mere concept; by contrast, the ground of a merely natural action must always be an appearance.

B 576 The action must certainly be possible under natural conditions if the “ought” is directed toward it; but these natural conditions concern not the determination of choice itself, but only its effect and outcome in appearance. However many natural grounds may impel me to will, however many sensible incentives there may be, they cannot produce the “ought,” but only a willing that is far from necessary and is always conditioned. Against this, the “ought” pronounced by reason sets measure and goal, indeed prohibition and authority. Whether the object is one of mere sensibility (the agreeable) or of pure reason (the good), reason does not yield to the ground empirically given and does not follow the order of things as they present themselves in appearance. Rather, with complete spontaneity it makes for itself an order of its own according to ideas, into which it fits empirical conditions, and according to which it even declares actions necessary that have not occurred and perhaps will not occur. Yet of all these actions it presupposes that reason could have causality in relation to them; otherwise it would not expect effects in experience from its ideas.

Let us now stop here and assume as at least possible that reason actually has causality with regard to appearances.

B 577 Then, however much it is reason, it must nevertheless exhibit an empirical character, because every cause presupposes a rule according to which certain appearances follow as effects, and every rule requires a uniformity of effects grounding the concept of the cause (as a power). Insofar as this must become evident from mere appearances, we can call it the cause’s empirical character, which is constant while its effects appear in changing forms according to differences in accompanying and partly limiting conditions.

Thus every human being has an empirical character of choice, which is nothing other than a certain causality of his reason insofar as it exhibits in its effects in appearance a rule through which we can discern the grounds of reason and its actions in their kinds and degrees, and judge the subjective principles of his choice. Because this empirical character must itself be drawn from appearances as effect and from their rule supplied by experience, all human actions in appearance are determined according to the order of nature by his empirical character and the other cooperating causes.

B 578 If we could investigate all appearances of his choice to their foundation, there would be no human action we could not predict with certainty and cognize as necessary from its preceding conditions. With regard to this empirical character, then, there is no freedom; and yet this is the only character according to which we can consider the human being when we merely observe and, as happens in anthropology, seek to investigate the moving causes of his actions physiologically.

But if we consider these same actions in relation to reason—not speculative reason, to explain them in their origin, but solely insofar as reason is the cause of producing them itself; in a word, if we compare them with reason for a practical purpose—we find an entirely different rule and order from that of nature. For perhaps everything that did happen according to nature’s course, and inevitably had to happen according to its empirical grounds, ought not to have happened. Sometimes, however, we find, or at least believe we find, that the ideas of reason have actually demonstrated causality with regard to human actions as appearances, and that these occurred not because they were determined by empirical causes—no, but because they were determined by grounds of reason.

B 579 Suppose, then, that we could say reason has causality with regard to appearance. Could its action be called free when it is precisely determined and necessary in reason’s empirical character (its sensible disposition)? This character is in turn determined in the intelligible character (the way of thinking). But we do not know the latter; we designate it through appearances, which properly make known immediately only the sensible disposition (empirical character).* Now the action, insofar as it is attributable to the way of thinking as its cause, nevertheless does not follow from it according to empirical laws: that is, not so that the conditions of pure reason precede, but only so that their effects precede in the appearance of inner sense. Pure reason, as a merely intelligible faculty, is not subject to the form of time, and hence not to the conditions of temporal succession either. The causality of reason in its intelligible character does not arise, or begin at some particular time, in order to produce an effect.

B 580 Otherwise it would itself be subject to the natural law of appearances insofar as that law determines causal series in time, and its causality would then be nature, not freedom. Thus we will be able to say: if reason can have causality with regard to appearances, it is a faculty through which the sensible condition of an empirical series of effects first begins. For the condition lying in reason is not sensible and therefore does not itself begin. Accordingly, what we missed in all empirical series then occurs: the condition of a successive series of events could itself be empirically unconditioned. For here the condition is outside the series of appearances (in the intelligible), and hence subject to no sensible condition and no temporal determination through a preceding cause.

* The proper morality of actions (merit and guilt), even that of our own conduct, therefore remains entirely hidden from us. Our imputations can refer only to the empirical character. But how much of this is to be ascribed to the pure effect of freedom, how much to mere nature and the blameless fault of temperament or its fortunate constitution (merito fortunae), no one can fathom, and hence no one can judge with complete justice.

Nevertheless, this same cause also belongs in another relation to the series of appearances. The human being is himself appearance. His choice has an empirical character that is the (empirical) cause of all his actions. None of the conditions determining the human being according to this character fails to be contained in the series of natural effects and to obey its law, according to which no empirically unconditioned causality of what happens in time is found at all. Hence no given action (since it can be perceived only as appearance) can begin absolutely of itself.

B 581 But of reason we cannot say that before the state in which it determines choice another precedes in which this state itself is determined. For since reason itself is no appearance and is subject to no conditions of sensibility at all, no temporal succession occurs in it even with regard to its causality. Hence the dynamical law of nature, which determines temporal succession according to rules, cannot be applied to it.

Reason is therefore the enduring condition of all voluntary actions through which the human being appears. Each is predetermined in the human being’s empirical character before it occurs. With regard to the intelligible character, of which the former is only the sensible schema, no “before” or “after” holds. Every action, regardless of the temporal relation in which it stands to other appearances, is the immediate effect of the intelligible character of pure reason, which therefore acts freely without being dynamically determined in the chain of natural causes by grounds that are external or internal but precede in time. This freedom can be regarded not merely negatively, as independence from empirical conditions (for thereby the faculty of reason would cease to be a cause of appearances).

B 582 It can also be designated positively as a power to begin a series of events of itself, so that nothing begins in reason itself, but, as the unconditioned condition of every voluntary action, it admits above itself no conditions preceding in time. Its effect nevertheless begins in the series of appearances, but can never constitute an absolutely first beginning within it.

To elucidate the regulative principle of reason through an example of its empirical use—not to confirm it, since proofs of this kind are unsuited to transcendental assertions—take a voluntary action, for example a malicious lie through which a person has caused a certain confusion in society. We first investigate the motives from which it arose, then judge how the action, together with its consequences, can be imputed to him. For the first purpose we trace his empirical character to its sources, seeking these in a bad upbringing, bad company, and partly also in the malice of a natural disposition insensitive to shame, partly attributing it to carelessness and thoughtlessness, while not overlooking the occasional causes that prompted it. In all this we proceed as we generally do in investigating the series of determining causes of a given natural effect.

B 583 Although we believe the action to be determined in this way, we nevertheless blame its author—not for his unfortunate natural disposition, not for the circumstances influencing him, not even for his previous conduct. For we presuppose that we can set aside entirely what that conduct was, regard the elapsed series of conditions as not having happened, and regard this deed as wholly unconditioned with respect to the previous state, as if through it the agent began a series of consequences entirely of himself. This blame rests on a law of reason, under which reason is regarded as a cause that could and ought to have determined the person’s conduct otherwise, regardless of all the empirical conditions mentioned. Indeed, we regard reason’s causality not merely as a contributing factor, but as complete in itself, even if sensible incentives were not in its favor but actually against it. The action is attributed to his intelligible character; now, at the moment he lies, he is entirely at fault. Thus despite all empirical conditions of the deed, reason was wholly free, and the deed is entirely attributable to reason’s omission.

It is easy to see in this imputative judgment that we have the following in mind. Reason is not affected at all by all that sensibility.

B 584 It does not alter (although its appearances, namely the way it shows itself in its effects, do alter). In it no state precedes to determine the next, and hence it does not belong at all to the series of sensible conditions that necessitate appearances according to laws of nature. Reason is present and the same in all human actions under all temporal circumstances, but is not itself in time or entering some new state in which it was not before. With respect to that state, it determines but is not determinable. Hence we cannot ask why reason did not determine itself differently, but only why through its causality it did not determine appearances differently. Yet no answer to this is possible. For a different intelligible character would have given a different empirical one. When we say that despite his entire previous conduct the agent could nevertheless have refrained from the lie, this means only that the lie stands immediately under reason’s power, and that reason in its causality is subject to no conditions of appearance or of the course of time. Difference in time does indeed make a principal difference among appearances in relation to one another; but since these are not things, and thus not causes in themselves either, it can make no difference to the action in relation to reason.

B 585 Thus in judging free actions with respect to their causality, we can reach only as far as the intelligible cause, but not beyond it. We can cognize that this cause can be free, that is, determined independently of sensibility, and in this way can be the sensibly unconditioned condition of appearances. But why the intelligible character yields precisely these appearances and this empirical character under the circumstances at hand exceeds all our reason’s power to answer, indeed all its authority even to ask, just as much as asking why the transcendental object of our outer sensible intuition yields precisely intuition in space and not some other kind. But the problem we had to solve does not oblige us to this at all. It was only whether freedom conflicts with natural necessity in one and the same action. We have sufficiently answered this by showing that, since freedom can have a relation to an entirely different kind of conditions from those of natural necessity, the law of the latter does not affect the former; thus both can occur independently of and undisturbed by one another.

* * *

It must be carefully noted that we have not thereby sought to demonstrate the actuality of freedom.

B 586 We did not mean to show it as one of the faculties containing the cause of the appearances of our sensible world. Apart from the fact that this would not have been a transcendental consideration, which concerns only concepts, it could not succeed, since from experience we can never infer something that need not be thought according to laws of experience at all. Further, we have not even sought to prove the possibility of freedom; this, too, would not have succeeded, because from mere concepts a priori we cannot cognize the possibility of any real ground or any causality at all. Freedom is treated here only as a transcendental idea through which reason thinks to begin the series of conditions in appearance absolutely through the sensibly unconditioned, but thereby becomes entangled in an antinomy with its own laws prescribed for the empirical use of the understanding. That this antinomy rests on a mere illusion, and that nature at least does not conflict with causality from freedom, was the only thing we could accomplish, and the only thing that concerned us.

B 587 IV. Resolution of the Cosmological Idea of the Totality of the Dependence of Appearances with Regard to Their Existence in General

In the preceding subsection we considered alterations of the sensible world in their dynamical series, where each stands under another as its cause. Now this series of states serves only as a guide toward an existence that could be the highest condition of everything alterable: namely, the necessary being. Here the issue is not unconditioned causality, but the unconditioned existence of substance itself. Thus the series before us is properly only one of concepts, not of intuitions insofar as one is the condition of another.

But it is easy to see that, since everything in the sum total of appearances is alterable, and hence conditioned in existence, nowhere in the series of dependent existence can there be an unconditioned member whose existence would be absolutely necessary. Thus, suppose appearances were things in themselves, and for that very reason their condition always belonged with the conditioned to one and the same series of intuitions.

B 588 Then a necessary being as the condition of the existence of appearances in the sensible world could never occur.

But dynamical regression has this peculiarity distinguishing it from mathematical regression: since the latter properly concerns only the composition of parts into a whole, or the decomposition of a whole into its parts, the conditions of this series must always be regarded as parts of it, hence as homogeneous, and consequently as appearances. In the former regression, by contrast, the question concerns not the possibility of an unconditioned whole from given parts, or of an unconditioned part for a given whole, but the derivation of a state from its cause, or of the contingent existence of substance itself from necessary existence. Here the condition need not necessarily constitute an empirical series with the conditioned.

Thus in the apparent antinomy before us, one escape remains open: both conflicting propositions can be true simultaneously in different relations. All things in the sensible world are thoroughly contingent, and therefore always have only empirically conditioned existence; yet for the whole series there can also be a nonempirical condition, that is, an unconditionally necessary being. For as an intelligible condition, this would not belong to the series as a member at all (not even as its highest member).

B 589 Nor would it make any member of the series empirically unconditioned, but would leave the entire sensible world in its empirically conditioned existence running through all its members. This way of grounding appearances in an unconditioned existence would therefore differ from the empirically unconditioned causality (freedom) in the preceding article in this: with freedom, the thing itself as cause (Substantia phaenomenon) nevertheless belonged to the series of conditions, and only its causality was thought as intelligible. Here, however, the necessary being would have to be thought wholly outside the series of the sensible world (as ens extramundanum) and as merely intelligible. Only in this way can we prevent its being itself subjected to the law of the contingency and dependence of all appearances.

Reason’s regulative principle with regard to this problem is therefore that everything in the sensible world has empirically conditioned existence, and that nowhere within it is there unconditioned necessity with regard to any property; that there is no member of the series of conditions for which we must not always expect, and seek as far as we can, the empirical condition in a possible experience; and that nothing entitles us to derive any existence from a condition outside the empirical series, or even to regard it within the series itself as absolutely independent and self-subsisting.

B 590 Nevertheless, we must not thereby deny that the entire series could be grounded in some intelligible being (which is therefore free from every empirical condition and instead contains the ground of the possibility of all these appearances).

But the intention here is by no means to prove the unconditionally necessary existence of a being, or even to ground on this the possibility of a merely intelligible condition of the existence of appearances in the sensible world. Rather, just as we restrict reason so that it does not abandon the thread of empirical conditions and wander into transcendent grounds of explanation incapable of any presentation in concreto, so, on the other hand, we restrict the law of the merely empirical use of the understanding so that it does not decide about the possibility of things in general or declare the intelligible impossible merely because we cannot use it to explain appearances. Thus we show only that the thoroughgoing contingency of all natural things and all their (empirical) conditions can quite well coexist with the arbitrary presupposition of a necessary, though merely intelligible, condition. There is therefore no true contradiction between these assertions, and both can be true. Such an absolutely necessary being of the understanding may indeed be impossible in itself.

B 591 But this can by no means be inferred from the universal contingency and dependence of everything belonging to the sensible world, or from the principle that we must stop at no single member of it insofar as it is contingent and appeal to a cause outside the world. Reason follows its course in empirical use and its separate course in transcendental use.

The sensible world contains nothing but appearances, and these are mere representations that are always sensibly conditioned in turn. Since here our objects are never things in themselves, it is no wonder that we are never entitled to leap from any member whatever of the empirical series outside the connection of sensibility, as if these were things in themselves existing outside their transcendental ground, which we could leave in order to seek the cause of their existence outside them. With contingent things this would indeed finally have to occur, but not with mere representations of things, whose contingency is itself only a phenomenon and can lead to no regression other than one determining phenomena, that is, an empirical one. But thinking an intelligible ground of appearances, that is, of the sensible world, and thinking it free from the contingency of that world, conflicts neither with unrestricted empirical regression in the series of appearances nor with their thoroughgoing contingency.

B 592 Yet this is also the only thing we had to accomplish to remove the apparent antinomy, and it could be done only in this way. For if the respective condition of every conditioned (with regard to existence) is sensible, and for that very reason belongs to the series, it is itself conditioned in turn (as the antithesis of the fourth antinomy shows). Thus either a conflict with reason, which demands the unconditioned, had to remain, or the unconditioned had to be placed outside the series in the intelligible, whose necessity requires and admits no empirical condition and is therefore unconditionally necessary in relation to appearances.

The empirical use of reason (with regard to the conditions of existence in the sensible world) is not affected by granting a merely intelligible being, but proceeds according to the principle of thoroughgoing contingency from empirical conditions to higher ones that are always equally empirical. Yet this regulative principle no more excludes the assumption of an intelligible cause outside the series when the issue is the pure use of reason (with regard to ends). For there that cause signifies only the ground of the possibility of the sensible series in general, a ground that for us is merely transcendental and unknown.

B 593 Its existence, independent of all conditions of that series and unconditionally necessary in relation to it, in no way conflicts with the unlimited contingency of the series, or therefore with regression that nowhere ends in the series of empirical conditions.

Concluding Remark on the Entire Antinomy of Pure Reason

As long as our concepts of reason have as their object only the totality of conditions in the sensible world, and what can be done in their regard to serve reason, our ideas are indeed transcendental, but still cosmological. But as soon as we place the unconditioned (which is properly at issue) in what lies wholly outside the sensible world, and hence outside all possible experience, the ideas become transcendent. They serve not merely to complete the empirical use of reason (which always remains an idea never realizable, yet still to be followed), but detach themselves entirely from it and make objects for themselves whose material is not taken from experience and whose objective reality rests not on the completion of the empirical series but on pure concepts a priori. Such transcendent ideas have a merely intelligible object. We are certainly permitted to admit it as a transcendental object of which we otherwise know nothing. But this permission does not extend to thinking it as a thing determinable through its distinguishing and inner predicates.

B 594 For that we have on our side neither grounds of possibility (as independent of all concepts of experience) nor the slightest justification for assuming such an object. It is therefore a mere thought-entity. Yet of all cosmological ideas, the one that occasioned the fourth antinomy impels us to venture this step. For the existence of appearances, not grounded in itself at all but always conditioned, calls on us to look for something distinct from all appearances, hence an intelligible object in which this contingency would cease. But once we have taken the liberty of assuming a self-subsisting actuality outside the field of all sensibility, and of regarding appearances merely as contingent ways in which intelligible objects are represented by beings that are themselves intelligences, nothing remains to us but analogy, by which we use experiential concepts to form some concept of intelligible things of which in themselves we have not the slightest knowledge. Since we learn of the contingent only through experience, whereas here we are speaking of things that are not to be objects of experience at all, we will have to derive our knowledge of them from what is necessary in itself, from pure concepts of things in general.

B 595 Hence the first step we take outside the sensible world compels us to begin our new knowledge with the investigation of the absolutely necessary being, and to derive from its concepts the concepts of all things insofar as they are merely intelligible. We will undertake this attempt in the following chapter.

Book Two of the Transcendental Dialectic Chapter Three The Ideal of Pure Reason

Section One The Ideal in General

We have seen above that no objects can be represented through pure concepts of the understanding without any conditions of sensibility, because the conditions of their objective reality are lacking, and nothing but the mere form of thinking is found in them. Yet they can be presented in concreto when applied to appearances; for in appearances they properly have the material for the concept of experience, which is nothing but a concept of the understanding in concreto. Ideas, however, are still further removed from objective reality than categories, for no appearance can be found in which they could be represented in concreto.

B 596 They contain a certain completeness that no possible empirical cognition reaches, and in them reason has in mind only a systematic unity to which it seeks to bring empirically possible unity nearer, without ever attaining it completely.

But what I call the ideal seems still further removed from objective reality than the idea. By it I understand the idea not merely in concreto, but in individuo, that is, as a single thing determinable or even determined through the idea alone.

Humanity in its entire perfection contains not only the extension of all essential properties belonging to this nature, which constitute our concept of it, to complete congruence with its ends—which would be our idea of perfect humanity—but also everything beyond this concept that belongs to the idea’s thoroughgoing determination. For of all opposed predicates, only one can accord with the idea of the most perfect human being. What to us is an ideal was to Plato an idea of the divine understanding, a single object in its pure intuition, the most perfect of each kind of possible beings and the original ground of all copies in appearance.

B 597 Without ascending so far, however, we must confess that human reason contains not only ideas but ideals as well. Though these have no creative force like the Platonic ones, they nevertheless have practical force (as regulative principles) and underlie the possibility of perfection in certain actions. Moral concepts are not entirely pure concepts of reason, because something empirical (pleasure or displeasure) underlies them. Yet in respect of the principle through which reason sets limits to freedom, which in itself is lawless (thus if we attend only to their form), they can quite well serve as examples of pure concepts of reason. Virtue, and with it human wisdom in its entire purity, are ideas. But the wise person (of the Stoic) is an ideal: a human being existing merely in thought, yet wholly congruent with the idea of wisdom. Just as the idea gives the rule, so in such a case the ideal serves as the archetype for the thoroughgoing determination of the copy. We have no other standard for our actions than the conduct of this divine human being within us, with which we compare and judge ourselves, thereby improving ourselves although we can never attain it. Even if we do not grant these ideals objective reality (existence), they must not on that account be regarded as fantasies. They supply an indispensable standard for reason.

B 598 Reason needs the concept of what is entirely complete in its kind in order to estimate and measure by it the degree and deficiencies of the incomplete. But to try to realize the ideal in an example, that is, in appearance, as one might portray the wise person in a novel, is impracticable and moreover has something absurd and hardly edifying about it. For the natural limits that continually impair completeness in the idea make all illusion in such an attempt impossible, thereby making even the good lying in the idea suspect and similar to a mere fiction.

Such is the ideal of reason, which must always rest on determinate concepts and serve as rule and archetype, whether for following or for judging. Matters are quite different with those creatures of imagination that no one can explain or give an intelligible concept of: monograms, as it were, consisting only of individual strokes determined according to no alleged rule, which constitute less a determinate image than a drawing hovering, as it were, amid various experiences. Such are what painters and physiognomists claim to have in their heads, supposedly an incommunicable shadow-image of their products or judgments. They can be called ideals of sensibility, though only improperly, because they are supposed to be the unattainable model of possible empirical intuitions.

B 599 Yet they supply no rule capable of explanation and examination.

Reason’s intention with its ideal, by contrast, is thoroughgoing determination according to a priori rules. Hence it thinks an object supposed to be thoroughly determinable according to principles, although sufficient conditions for this are lacking in experience, and the concept itself is therefore transcendent.

Chapter Three Section Two The Transcendental Ideal (Prototypon transzendentale)

Every concept is indeterminate with regard to what is not contained in it and stands under the principle of determinability: only one of any two contradictorily opposed predicates can belong to it. This rests on the principle of contradiction and is therefore a merely logical principle, abstracting from all content of cognition and having only its logical form in view.

But every thing, with regard to its possibility, also stands under the principle of thoroughgoing determination, according to which one of all possible predicates of things, insofar as they are compared with their opposites, must belong to it.

B 600 This rests not merely on the principle of contradiction; for besides the relation of two conflicting predicates, it also considers every thing in relation to the entirety of possibility as the sum total of all predicates of things in general. Presupposing this as an a priori condition, it represents every thing as deriving its own possibility from its share in that entirety of possibility.* The principle of thoroughgoing determination thus concerns content, not merely logical form. It is the principle of the synthesis of all predicates that are to constitute the complete concept of a thing, not merely of analytical representation through one of two opposed predicates.

B 601 It contains a transcendental presupposition: that of the material for all possibility, which is supposed to contain a priori the data for the particular possibility of each thing.

* Through this principle every thing is thus referred to a common correlate, namely the entirety of possibility. If this (that is, the material for all possible predicates) were found in the idea of a single thing, it would prove an affinity of everything possible through the identity of the ground of its thoroughgoing determination. The determinability of every concept is subordinated to the universality (Universalitas) of the principle excluding a middle between two opposed predicates, whereas the determination of a thing is subordinated to the allness (Universitas), or sum total, of all possible predicates.

The proposition “Everything existing is thoroughly determined” means not only that one predicate from every pair of given opposed predicates always belongs to it, but also one from all possible predicates. Through this proposition, predicates are not merely compared logically with one another; the thing itself is compared transcendentally with the sum total of all possible predicates. It means this: to cognize a thing completely, we must cognize everything possible and thereby determine the thing, whether affirmatively or negatively. Thoroughgoing determination is consequently a concept we can never present in concreto in its totality, and thus rests on an idea whose seat lies solely in reason, which prescribes to the understanding the rule of its complete use.

Now although this idea of the sum total of all possibility, insofar as it underlies the thoroughgoing determination of each thing as its condition, is itself still indeterminate regarding the predicates that might constitute it, and through it we think nothing more than a sum total of all possible predicates in general, closer investigation nevertheless shows that this idea, as an original concept, excludes a multitude of predicates that, as derivative, are already given through others or cannot coexist.

B 602 It is purified into a concept thoroughly determined a priori, thereby becoming the concept of a single object thoroughly determined through the mere idea, and thus must be called an ideal of pure reason.

If we consider all possible predicates not merely logically, but transcendentally, that is, according to the content that can be thought in them a priori, we find that through some of them a being is represented, through others a mere nonbeing. Logical negation, indicated solely by the little word “not,” never properly attaches to a concept, but only to its relation to another in a judgment, and thus falls far short of being sufficient to designate a concept with regard to its content. The expression “not mortal” cannot at all indicate that mere nonbeing in the object is thereby represented; it leaves all content untouched. Transcendental negation, by contrast, signifies nonbeing in itself, opposed to transcendental affirmation, which is a something whose concept itself already expresses a being. Hence it is called reality (thinghood), because through it alone, and as far as it extends, objects are something (things).

B 603 The opposed negation, by contrast, signifies a mere lack, and where this alone is thought, the abolition of every thing is represented.

Now no one can think a negation determinately without having the opposed affirmation as its basis. A person born blind cannot form the slightest representation of darkness, because he has none of light; a person living outside civilization cannot represent poverty, because he does not know prosperity.* The ignorant person has no concept of his ignorance because he has none of science, and so on. Thus all concepts of negations are derivative, and realities contain the data and, so to speak, the material, or transcendental content, for the possibility and thoroughgoing determination of all things.

* The observations and calculations of astronomers have taught us much that is admirable, but perhaps the most important thing is that they have disclosed to us the abyss of ignorance that human reason, without this knowledge, could never have imagined so great. Reflection on it must bring about a great change in determining the ultimate aims of our use of reason.

If, therefore, thoroughgoing determination in our reason is grounded in a transcendental substratum containing, as it were, the whole stock of material from which all possible predicates of things can be taken, this substratum is nothing other than the idea of an all of reality (omnitudo realitatis).

B 604 All true negations are then nothing but limits, which they could not be called unless the unlimited (the all) lay at their basis.

But through this possession of all reality, the concept of a thing in itself is also represented as thoroughly determined, and the concept of an ens realissimum is the concept of a single being, because of all possible opposed predicates, one—namely, what belongs absolutely to being—is found in its determination. Thus a transcendental ideal underlies the thoroughgoing determination necessarily found in everything that exists, and constitutes the highest and complete material condition of its possibility, to which all thinking of objects in general must be traced back with regard to its content. Yet it is also the only proper ideal of which human reason is capable, because only in this single case is a concept of a thing that is in itself universal thoroughly determined through itself and cognized as the representation of an individual.

The logical determination of a concept through reason rests on a disjunctive syllogism, whose major premise contains a logical division (the division of the sphere of a universal concept), whose minor premise restricts this sphere to one part, and whose conclusion determines the concept through that part.

B 605 The universal concept of a reality in general cannot be divided a priori, because without experience we know no determinate kinds of reality contained under that genus. Thus the transcendental major premise of the thoroughgoing determination of all things is nothing other than the representation of the sum total of all reality: not merely a concept comprising all predicates under itself according to their transcendental content, but one comprising them within itself. The thoroughgoing determination of every thing rests on the limitation of this all of reality, some of it being attributed to the thing while the rest is excluded. This agrees with the “either–or” of the disjunctive major premise and the determination of the object through one member of this division in the minor premise. Accordingly, reason’s use of the transcendental ideal as the basis for its determination of all possible things is analogous to its procedure in disjunctive syllogisms. This was the proposition I laid above at the foundation of the systematic division of all transcendental ideas, according to which they are generated parallel to and corresponding with the three kinds of syllogism.

Reason’s purpose here is merely to represent the necessary thoroughgoing determination of things.

B 606 It goes without saying that for this purpose it presupposes not the existence of a being conforming to the ideal, but only its idea, in order to derive conditioned determination, that is, that of the limited, from an unconditioned totality of thoroughgoing determination. Thus for reason the ideal is the archetype (Prototypon) of all things, which collectively, as deficient copies (ectypa), take from it the material of their possibility and, while approaching it more or less closely, nevertheless always fall infinitely short of attaining it.

Thus all possibility of things (of the synthesis of the manifold with regard to its content) is regarded as derivative, and only the possibility of what includes all reality within itself as original. For all negations (which are the only predicates through which everything else can be distinguished from the most real being) are mere limitations of a greater and ultimately of the highest reality; hence they presuppose this reality and are merely derived from it in content. All manifoldness of things is only a correspondingly manifold way of limiting the concept of the highest reality, which is their common substratum, just as all figures are possible only as different ways of limiting infinite space. Hence the object of reason’s ideal, which is found merely in reason, is also called the original being (ens originarium); insofar as it has none above it, the highest being (ens summum); and insofar as everything, as conditioned, stands under it, the being of all beings (ens entium).

B 607 But all this signifies not the objective relation of an actual object to other things, but that of the idea to concepts, and leaves us in complete ignorance about the existence of a being of such exceptional preeminence.

Since we also cannot say that an original being consists of many derivative beings, because each of them presupposes it and therefore cannot constitute it, the ideal of the original being must also be thought as simple.

Strictly speaking, the derivation of all other possibility from this original being therefore cannot be regarded as a limitation of its highest reality, and as though a division of it. For then the original being would be regarded as a mere aggregate of derivative beings, which according to what precedes is impossible, although we initially represented it this way in our first rough outline. Rather, the highest reality would underlie the possibility of all things as a ground, not as a sum total, and their manifoldness would rest not on limiting the original being itself, but on its complete consequence. To this would also belong our entire sensibility, together with all reality in appearance, which cannot belong as an ingredient to the idea of the highest being.

B 608 If we now pursue this idea further by hypostatizing it, we will be able to determine the original being through the mere concept of the highest reality as one, simple, all-sufficient, eternal, and so on—in a word, in its unconditioned completeness through all predicaments. The concept of such a being is that of God, thought in the transcendental sense, and thus the ideal of pure reason is the object of a transcendental theology, as I also stated above.

Yet this use of the transcendental idea would already exceed the boundaries of its purpose and admissibility. For reason laid it, merely as the concept of all reality, at the foundation of the thoroughgoing determination of things in general, without demanding that all this reality be objectively given and itself constitute a thing. The latter is a mere invention through which we collect and realize the manifold of our idea in an ideal as a particular being. We have no authority for this, not even to assume outright the possibility of such a hypothesis. Moreover, all consequences flowing from such an ideal have nothing to do with the thoroughgoing determination of things in general, for whose sake alone the idea was needed, and have not the slightest influence on it.

B 609 It is not enough to describe our reason’s procedure and its dialectic; we must also seek to discover their sources in order to explain this illusion itself as a phenomenon of the understanding. For the ideal of which we speak is grounded in a natural, not merely arbitrary, idea. I therefore ask: how does reason come to regard all possibility of things as derived from a single underlying possibility, namely that of the highest reality, and then to presuppose this as contained in a particular original being?

The answer presents itself from the discussions of the Transcendental Analytic. The possibility of objects of the senses is their relation to our thinking, in which something (namely empirical form) can be thought a priori, while what constitutes the matter, the reality in appearance (what corresponds to sensation), must be given; otherwise it could not even be thought, and hence its possibility could not be represented. Now an object of the senses can be thoroughly determined only if it is compared with all predicates of appearance and represented affirmatively or negatively through them. But what constitutes the thing itself (in appearance), namely the real, must be given in it, without which it could not even be thought.

B 610 Moreover, that in which the real of all appearances is given is the one all-embracing experience. The material for the possibility of all objects of the senses must therefore be presupposed as given in one sum total. On its limitation alone can rest all possibility of empirical objects, their difference from one another, and their thoroughgoing determination. Now in fact no objects can be given to us except those of the senses, and nowhere except in the context of a possible experience; consequently, nothing is an object for us unless it presupposes the sum total of all empirical reality as the condition of its possibility. Through a natural illusion, we now take what properly holds only of things given as objects of our senses for a principle that must hold of all things in general. Consequently, by omitting this restriction, we will regard the empirical principle of our concepts of the possibility of things as appearances as a transcendental principle of the possibility of things in general.

That we afterward hypostatize this idea of the sum total of all reality comes from dialectically transforming the distributive unity of the experiential use of the understanding into the collective unity of a whole of experience. In this whole of appearance we think a single thing containing all empirical reality within itself.

B 611 This is then, through the transcendental subreption already mentioned, confused with the concept of a thing standing at the summit of the possibility of all things and supplying the real conditions of their thoroughgoing determination.*

* Thus this ideal of the most real being, although a mere representation, is first realized, that is, made into an object, then hypostatized, and finally, through a natural advance of reason toward completing unity, even personified, as we will shortly explain. For the regulative unity of experience rests not on appearances themselves (sensibility alone), but on the connection of their manifold through the understanding (in one apperception). Hence the unity of the highest reality and the thoroughgoing determinability (possibility) of all things seem to lie in a highest understanding, and thus in an intelligence.

Chapter Three Section Three The Grounds of Proof of Speculative Reason for Inferring the Existence of a Highest Being

Despite reason’s urgent need to presuppose something that could serve the understanding as a complete foundation for the thoroughgoing determination of its concepts, it notices the ideal and merely invented character of such a presupposition much too readily.

B 612 It cannot be persuaded by this alone to accept a mere creature of its own thinking immediately as an actual being, unless it were driven by something else to seek a resting place somewhere in the regression from the given conditioned to the unconditioned. The latter is indeed not given as actual in itself and according to its mere concept, yet it alone can complete the series of conditions carried out to their grounds. This is the natural course taken by every human reason, even the most ordinary, though not every reason persists in it. It begins not from concepts but from ordinary experience, thus taking something existing as its foundation. But this ground gives way unless it rests on the immovable rock of the absolutely necessary. Yet this rock itself floats unsupported if there is still empty space outside and beneath it, and it does not itself fill everything, thereby leaving no room for any further “why”—that is, unless it is infinite in reality.

If anything, whatever it may be, exists, we must also admit that something exists necessarily. For the contingent exists only under the condition of another as its cause, and the inference continues to hold of this cause until we reach one that is not contingent and therefore exists necessarily without condition. This is the argument on which reason grounds its advance to the original being.

B 613 Reason now looks for the concept of a being suited to such a preeminence of existence as unconditioned necessity. It does so not so much in order then to infer its existence a priori from its concept (for if it trusted itself to do this, it would need to investigate only among mere concepts and would have no need to take a given existence as its basis), but only to find among all concepts of possible things the one containing nothing contrary to absolute necessity. For after the first inference it already regards it as settled that something must exist absolutely necessarily. If it can now eliminate everything incompatible with this necessity except one thing, that is the absolutely necessary being, whether or not we can comprehend its necessity, that is, derive it from its concept alone.

Now that whose concept contains the “because” for every “why,” which is defective in no part and no respect, which suffices everywhere as a condition, seems for that very reason to be the being suited to absolute necessity. For in possessing within itself all conditions of everything possible, it needs no condition itself, indeed is not even capable of one, and consequently satisfies, at least in one respect, the concept of unconditioned necessity.

B 614 No other concept can equal it in this, since, being deficient and needing completion, it displays no such mark of independence from all further conditions. It is true that from this we still cannot securely infer that what does not contain within itself the highest condition, complete in every respect, must therefore itself be conditioned in its existence. But it does not possess the sole distinguishing mark of unconditioned existence available to reason for cognizing any being as unconditioned through an a priori concept.

Thus among all concepts of possible things, the concept of a being of the highest reality would best accord with the concept of an unconditionally necessary being. Even if it does not fully satisfy that concept, we have no choice, but see ourselves compelled to hold to it, because we cannot dismiss the existence of a necessary being. Yet if we admit this existence, we can find nothing throughout the field of possibility that could make a better-founded claim to such preeminence in existence.

Such, then, is the natural course of human reason. First it convinces itself of the existence of some necessary being. In this it cognizes an unconditioned existence.

B 615 It now seeks the concept of what is independent of every condition, and finds it in what is itself the sufficient condition of everything else, that is, what contains all reality. But the all without limits is absolute unity and carries with it the concept of one being, namely the highest. Thus reason concludes that the highest being, as the original ground of all things, exists absolutely necessarily.

This concept cannot be denied a certain soundness when a decision is at issue: namely, once the existence of some necessary being is conceded and we agree that we must choose what to take it to be. We can then make no more appropriate choice—or rather, we have no choice, but are compelled to cast our vote for the absolute unity of complete reality as the original source of possibility. But if nothing drives us to decide, and we would rather leave the whole matter unsettled until compelled to assent by the full weight of grounds of proof—that is, if the issue is merely to judge how much we know of this problem and what we merely flatter ourselves that we know—then the above inference appears in a far less favorable light and needs favor to make up for its lack of rightful claims.

For let us allow everything to stand as it is before us.

B 616 First, from some given existence (even merely my own) there is a valid inference to the existence of an unconditionally necessary being. Second, I must regard a being containing all reality, and hence every condition, as absolutely unconditioned, so that the concept of the thing suited to absolute necessity has thereby been found. Even so, we cannot infer that the concept of a limited being, which does not possess the highest reality, therefore contradicts absolute necessity. For although I do not find in its concept the unconditioned that already carries with it the all of conditions, we cannot infer from this that its existence must for that very reason be conditioned; just as in a hypothetical syllogism I cannot say that where a certain condition (here, completeness according to concepts) is absent, the conditioned is also absent. Rather, we remain free to regard all other limited beings as likewise unconditionally necessary, although we cannot infer their necessity from the general concept we have of them. In this way, however, the argument would have supplied us not the slightest concept of the properties of a necessary being and would have accomplished nothing at all.

Nevertheless, this argument retains a certain importance and authority that its objective inadequacy cannot immediately take away.

B 617 For suppose there were obligations that would be entirely correct in the idea of reason, but without any reality in application to ourselves—that is, without incentives—unless a highest being were presupposed that could give practical laws effect and force. Then we would also have an obligation to follow concepts that, though perhaps not objectively sufficient, nevertheless preponderate by the measure of our reason, and in comparison with which we cognize nothing better or more convincing. Here the duty to choose would tip speculation’s indecision out of equilibrium through a practical addition. Indeed, reason would find no justification even before itself, as the most indulgent judge, if under pressing motives, though with only deficient insight, it had failed to follow these grounds of judgment, than which we at least know none better.

Although this argument is in fact transcendental, since it rests on the inner inadequacy of the contingent, it is so simple and natural that it accords with the most ordinary human sense as soon as this sense is once directed to it. We see things alter, arise, and pass away; they, or at least their state, must therefore have a cause. But the same question can be asked in turn of every cause that may ever be given in experience.

B 618 Where, then, should we more reasonably place the supreme causality than where the highest causality also is: in that being which originally contains within itself sufficiency for the possible effect, whose concept is also very easily formed through the single feature of all-embracing perfection? We then regard this highest cause as absolutely necessary because we find it absolutely necessary to ascend to it and find no ground to go beyond it. Hence among all peoples, even through their blindest polytheism, we see some sparks of monotheism shining through. They were led to it not by reflection and deep speculation, but only by a natural course of ordinary understanding that gradually became intelligible.

Only Three Kinds of Proof of God’s Existence from Speculative Reason Are Possible

All paths we may take for this purpose begin either from determinate experience and the particular constitution of our sensible world cognized through it, ascending from this according to laws of causality to the highest cause outside the world; or they take only indeterminate experience, that is, some existence, as their empirical basis; or, finally, they abstract from all experience and infer the existence of a highest cause entirely a priori from mere concepts.

B 619 The first proof is the physicotheological, the second the cosmological, the third the ontological. There are no more, nor can there be any more.

I will show that reason accomplishes as little on the one path (the empirical) as on the other (the transcendental), and spreads its wings in vain to pass beyond the sensible world through the mere power of speculation. As for the order in which these kinds of proof must be presented for examination, it will be precisely the reverse of the order taken by reason as it gradually extends itself, in which we also first presented them. For it will become evident that, although experience gives the first occasion, the transcendental concept alone guides reason in this endeavor and in all such attempts marks out the goal it has set itself. I will therefore begin by examining the transcendental proof and then see what the addition of the empirical can do to increase its force.

B 620 Chapter Three Section Four The Impossibility of an Ontological Proof of God’s Existence

It is easy to see from what precedes that the concept of an absolutely necessary being is a pure concept of reason, that is, a mere idea whose objective reality is far from proved by reason’s need for it. It points only to a certain, though unattainable, completeness, and properly serves more to bound the understanding than to extend it to new objects. Here we now find the strange and absurd situation that the inference from a given existence in general to some absolutely necessary existence seems compelling and correct, yet all the conditions of the understanding for forming a concept of such necessity are entirely against us.

People have always spoken of an absolutely necessary being, taking pains not so much to understand whether and how one can even think a thing of this kind as to prove its existence. A nominal definition of this concept is indeed quite easy: it is something whose nonbeing is impossible.

B 621 But this makes us no wiser about the conditions that make it impossible to regard a thing’s nonbeing as absolutely unthinkable, which are properly what we want to know: whether through this concept we think anything at all or not. For throwing away, by means of the word “unconditioned,” all conditions the understanding always needs to regard something as necessary does not remotely make it intelligible to me whether through a concept of the unconditionally necessary I still think something, or perhaps nothing at all.

Still more: people believed they had explained this concept, ventured at mere random and finally become quite familiar, through a multitude of examples, so that any further inquiry into its intelligibility seemed quite unnecessary. Every proposition of geometry—for example, that a triangle has three angles—is absolutely necessary, and so they spoke of an object lying wholly outside the sphere of our understanding as if they understood perfectly well what they meant by its concept.

All the alleged examples, without exception, are taken only from judgments, not from things and their existence. But the unconditioned necessity of judgments is not an absolute necessity of things. For the absolute necessity of the judgment is only a conditioned necessity of the thing, or of the predicate in the judgment.

B 622 The preceding proposition did not say that three angles are absolutely necessary, but that under the condition that a triangle exists (is given), three angles also necessarily exist (in it). Yet this logical necessity has demonstrated such a powerful illusion that, after forming an a priori concept of a thing, a concept so framed that, in their opinion, existence was included in its scope, people believed they could securely infer that, because existence necessarily belongs to the object of this concept, that is, under the condition that I posit this thing as given (existing), its existence is also necessarily posited (according to the rule of identity), and this being is therefore itself absolutely necessary, because its existence is thought along with it in a concept arbitrarily assumed and under the condition that I posit its object.

If I abolish the predicate in an identical judgment while retaining the subject, a contradiction arises; hence I say the former necessarily belongs to the latter. But if I abolish the subject together with the predicate, no contradiction arises, for nothing remains that could be contradicted. To posit a triangle while abolishing its three angles is contradictory; but to abolish the triangle together with its three angles is no contradiction. Precisely the same holds for the concept of an absolutely necessary being.

B 623 If you abolish its existence, you abolish the thing itself with all its predicates; where, then, is the contradiction to come from? Externally there is nothing to contradict, since the thing is not supposed to be externally necessary; internally there is also nothing, because by abolishing the thing itself you have abolished everything internal along with it. “God is omnipotent” is a necessary judgment. Omnipotence cannot be abolished if you posit a deity, that is, an infinite being with whose concept that predicate is identical. But if you say “God does not exist,” neither omnipotence nor any other predicate of his is given; for all are abolished together with the subject, and not the slightest contradiction appears in this thought.

You have therefore seen that if I abolish a judgment’s predicate together with its subject, no inner contradiction can ever arise, whatever the predicate may be. You now have no escape but to say that there are subjects that cannot be abolished at all and must therefore remain. But this would amount to saying that there are absolutely necessary subjects: the very presupposition whose correctness I have doubted and whose possibility you intended to show me. For I cannot form the slightest concept of a thing that, if abolished together with all its predicates, would leave a contradiction behind.

B 624 And without contradiction I have, through mere pure concepts a priori, no mark of impossibility.

Against all these general inferences (which no one can refuse), you challenge me with a case you present as proof by the fact itself: that there is nevertheless one concept, and only this one, in which the nonbeing or abolition of its object is self-contradictory, namely the concept of the most real being. It has, you say, all reality, and you are entitled to assume such a being as possible (which I concede for now, although a concept free of contradiction is far from proving the possibility of its object)*. Now existence is also included among all reality. Thus existence lies in the concept of something possible.

B 625 If this thing is now abolished, the inner possibility of the thing is abolished, which is contradictory.

* A concept is always possible if it does not contradict itself. This is the logical mark of possibility, distinguishing its object from the nihil negativum. Yet it can still be an empty concept unless the objective reality of the synthesis through which it is generated is separately demonstrated. As shown above, this always rests on principles of possible experience, not on the principle of analysis (the principle of contradiction). This is a warning not to infer immediately from the (logical) possibility of concepts to the (real) possibility of things.

I answer: you have already committed a contradiction when, into the concept of a thing you intended to think solely with regard to its possibility, you have already introduced, under whatever concealed name, the concept of its existence. If we grant you this, you have seemingly won the game, but in fact said nothing, for you have committed a mere tautology. I ask you: is the proposition “This or that thing (which I grant you as possible, be it what it may) exists”—is this proposition, I ask, analytic or synthetic? If the former, then through the thing’s existence you add nothing to your thought of the thing; but then either the thought in you would have to be the thing itself, or you have presupposed an existence as belonging to possibility and then purported to infer existence from inner possibility, which is nothing but a miserable tautology. The word “reality,” which sounds different in the concept of the thing from “existence” in the concept of the predicate, makes no difference. For even if you call all positing reality (leaving indeterminate what you posit), you have already posited the thing with all its predicates in the concept of the subject and assumed it as actual; in the predicate you merely repeat it.

B 626 If, by contrast, you admit, as every rational person rightly must, that every existential proposition is synthetic, how can you then maintain that the predicate of existence cannot be abolished without contradiction, when this privilege belongs peculiarly only to analytic propositions, whose character rests precisely on it?

I would hope to dispose of this subtle sophistry without further ado through a precise determination of the concept of existence, had I not found that the illusion arising from confusing a logical predicate with a real one (that is, the determination of a thing) resists nearly all instruction. Anything we please can serve as a logical predicate; even the subject can be predicated of itself, for logic abstracts from all content. But a determination is a predicate added to the concept of the subject and enlarging it. It must therefore not already be contained in it.

Being is plainly no real predicate, that is, no concept of something that could be added to the concept of a thing. It is merely the positing of a thing, or of certain determinations in themselves. In logical use it is merely the copula of a judgment. The proposition “God is omnipotent” contains two concepts having their objects: God and omnipotence.

B 627 The little word “is” is not an additional predicate, but only what posits the predicate in relation to the subject. If I now take the subject (God) together with all its predicates (including omnipotence) and say “God is,” or “There is a God,” I posit no new predicate in addition to the concept of God, but only the subject in itself with all its predicates, and indeed the object in relation to my concept. Both must contain exactly the same thing, and hence nothing further can be added to the concept, which expresses only possibility, by my thinking its object as absolutely given (through the expression “it is”). Thus the actual contains no more than the merely possible. A hundred actual thalers contain not the slightest bit more than a hundred possible ones. For since the latter signify the concept, while the former signify the object and its positing in itself, if the object contained more than the concept, my concept would not express the whole object and thus would not be its adequate concept. But my financial condition contains more with a hundred actual thalers than with their mere concept (that is, their possibility). For in actuality the object is not merely analytically contained in my concept, but is synthetically added to my concept (which is a determination of my state), without this being outside my concept in the least increasing those hundred thalers themselves as thought.

B 628 Thus if I think a thing through whatever predicates and however many I please (even in thoroughgoing determination), adding that this thing exists adds not the slightest thing to it. Otherwise, not exactly the same thing, but more than I had thought in the concept, would exist, and I could not say that precisely the object of my concept exists. Even if I think in a thing all reality except one, saying that such a deficient thing exists does not add the missing reality; it exists burdened with precisely the same deficiency with which I thought it. Otherwise something other than what I thought would exist. If I now think a being as the highest reality (without deficiency), the question still remains whether it exists or not. For although nothing is lacking in my concept of the possible real content of a thing in general, something is still lacking in its relation to my entire state of thinking: namely, that cognition of that object is also possible a posteriori. Here the cause of the difficulty becomes evident. If we were speaking of an object of the senses, I could not confuse the existence of the thing with its mere concept. For through the concept the object is thought only as agreeing with the universal conditions of a possible empirical cognition in general, whereas through existence it is thought as contained in the context of all experience.

B 629 Through connection with the content of all experience, the concept of the object is not in the least enlarged, but our thinking gains one more possible perception through it. If, by contrast, we wish to think existence through the pure category alone, it is no wonder that we can specify no mark distinguishing it from mere possibility.

Thus whatever and however much our concept of an object may contain, we must go outside it to confer existence on the object. With objects of the senses this occurs through connection with some perception of mine according to empirical laws. But for objects of pure thinking there is no means whatever of cognizing their existence, because it would have to be cognized entirely a priori, whereas our consciousness of all existence (whether immediately through perception or through inferences connecting something with perception) belongs wholly to the unity of experience. An existence outside this field cannot indeed be declared absolutely impossible, but it is a presupposition we can justify through nothing.

The concept of a highest being is an idea very useful in many respects; but precisely because it is a mere idea, it is wholly incapable, by itself alone, of extending our cognition regarding what exists.

B 630 It cannot even instruct us about the possibility of anything further. The analytical mark of possibility—that mere positings (realities) produce no contradiction—cannot indeed be denied it. But the connection of all real properties in one thing is a synthesis whose possibility we cannot judge a priori, because realities are not given to us specifically; and even if they were, no judgment could be made in this matter at all, because the mark of the possibility of synthetic cognitions must always be sought only in experience, to which the object of an idea cannot belong. Thus the famous Leibniz has by far not accomplished what he flattered himself with, namely wanting to have a priori insight into the possibility of so sublime an ideal being.

All effort and labor expended on the famous ontological (Cartesian) proof of the existence of a highest being from concepts is therefore lost. A person would be no more likely to become richer in insight from mere ideas than a merchant would in wealth by adding a few zeros to his cash balance to improve his position.

B 631 Chapter Three Section Five The Impossibility of a Cosmological Proof of God’s Existence

Trying to extract the existence of the corresponding object itself from an entirely arbitrarily devised idea was something wholly unnatural, a mere innovation of scholastic ingenuity. In fact, no one would ever have attempted this path had there not first been our reason’s need to assume, for existence in general, something necessary (at which we could stop in ascending); and had reason not been compelled, since this necessity must be unconditioned and certain a priori, to seek a concept that could, if possible, satisfy such a demand and make an existence known entirely a priori. People now believed they had found it in the idea of a most real being, and thus used this idea only for a more determinate knowledge of what they were already otherwise convinced or persuaded must exist, namely the necessary being. But they concealed this natural course of reason and, instead of ending with this concept, tried to begin with it in order to derive from it the necessity of existence, which it was intended only to complete.

B 632 From this arose the unsuccessful ontological proof, which offers satisfaction neither to natural and sound understanding nor to scholastic examination.

The cosmological proof we will now investigate retains the connection of absolute necessity with the highest reality. But instead of inferring, like the former proof, from highest reality to necessity in existence, it infers rather from the previously given unconditioned necessity of some being to its unbounded reality. To that extent, it at least brings everything into the track of an inference that, whether rational or sophistical I do not know, is at least natural and carries the greatest persuasive force for both ordinary and speculative understanding. Indeed, it visibly draws the first outlines for all proofs of natural theology, which people have always followed and will continue to follow, however much foliage and ornament they may use to embellish and conceal them. We will now present and examine this proof, which Leibniz also called the proof a contingentia mundi.

It runs as follows: if anything exists, an absolutely necessary being must also exist. Now at least I myself exist. Therefore an absolutely necessary being exists.

B 633 The minor premise contains an experience; the major contains the inference from an experience in general to the existence of the necessary.* Thus the proof properly begins from experience and is consequently not conducted wholly a priori, or ontologically. Since the object of all possible experience is called the world, it is therefore called the cosmological proof. Since it also abstracts from every particular property of objects of experience through which this world might differ from any possible world, its very name distinguishes it from the physicotheological proof, which uses observations of the particular constitution of this sensible world of ours as grounds of proof.

* This inference is too familiar to need an extensive presentation here. It rests on the supposedly transcendental natural law of causality: everything contingent has its cause, which, if itself contingent, must likewise have a cause, until the series of causes subordinated to one another must end at an absolutely necessary cause, without which it would have no completeness.

The proof now continues: the necessary being can be determined in only one way, that is, with regard to all possible opposed predicates, only through one of them; consequently, it must be thoroughly determined through its concept. Now only one concept of a thing is possible that determines it thoroughly a priori, namely that of the ens realissimum.

B 634 Thus the concept of the most real being is the only one through which a necessary being can be thought; that is, a highest being necessarily exists.

So many sophistical principles converge in this cosmological argument that speculative reason seems here to have summoned all its dialectical art to produce the greatest possible transcendental illusion. We will set their examination aside for a while, however, to expose just one of its tricks. It presents an old argument in disguise as a new one and appeals to the agreement of two witnesses, one a witness of pure reason, the other empirically accredited, when in fact there is only the former, who merely changes his dress and voice to be taken for a second. To secure its foundation, this proof rests on experience and thereby gives itself the appearance of differing from the ontological proof, which puts all its trust in nothing but pure concepts a priori. But the cosmological proof uses this experience only to take a single step: to the existence of a necessary being in general. The empirical ground of proof cannot teach what properties this being has; here reason takes complete leave of it and investigates among mere concepts what properties an absolutely necessary being in general must have.

B 635 (That is, it asks which among all possible things contains the required conditions (requisita) for absolute necessity.) Now it believes these requirements are found solely in the concept of a most real being, and then concludes that this is the absolutely necessary being. But it is clear that this presupposes that the concept of a being of the highest reality fully satisfies the concept of absolute necessity in existence, that is, that the latter can be inferred from the former. This is a proposition asserted by the ontological argument, which is therefore accepted and taken as the basis in the cosmological proof, though it was precisely what we intended to avoid. For absolute necessity is an existence from mere concepts. If I now say that the concept of the ens realissimum is such a concept, indeed the only one suited and adequate to necessary existence, I must also concede that the latter can be inferred from it. Thus it is properly only the ontological proof from mere concepts that contains all the force of proof in the so-called cosmological argument. The alleged experience is quite idle, useful perhaps only to lead us to the concept of absolute necessity, but not to demonstrate this in any determinate thing. For as soon as we intend this, we must immediately abandon all experience and seek among pure concepts which of them contains the conditions of the possibility of an absolutely necessary being.

B 636 But if in this way only the possibility of such a being is understood, its existence is also demonstrated; for this amounts to saying that among everything possible there is one thing carrying absolute necessity with it—that is, this being exists absolutely necessarily.

All deceptions in inference are most easily discovered when presented in scholastic form. Here is such a presentation.

If the proposition “Every absolutely necessary being is also the most real being” is correct (which is the nervus probandi of the cosmological proof), then, like all affirmative judgments, it must allow conversion at least per accidens: some most real beings are also absolutely necessary beings. Now an ens realissimum does not differ from another in any respect, and what holds of some contained under this concept therefore holds of all. Hence I will also be able to convert it (in this case) simply, that is, every most real being is a necessary being. Now since this proposition is determined solely from its concepts a priori, the mere concept of the most real being must also carry its absolute necessity with it. This is precisely what the ontological proof asserted.

B 637 The cosmological proof refused to acknowledge it, yet covertly made it the basis of its inferences.

Thus the second path speculative reason takes to prove the existence of the highest being is not only as deceptive as the first, but has this further fault: it commits an ignoratio elenchi, promising to lead us along a new path but, after a short detour, bringing us back to the old one we had abandoned for its sake.

I said shortly before that a whole nest of dialectical pretensions lies hidden in this cosmological argument, which transcendental critique can readily discover and destroy. I will now merely list them, leaving the already practiced reader to investigate and remove these deceptive principles further.

We find here, for example: 1. The transcendental principle of inferring a cause from the contingent, which has significance only in the sensible world, but outside it has not even a meaning. For the merely intellectual concept of the contingent can produce no synthetic proposition such as that of causality, and the principle of causality has no significance or mark of its use except in the sensible world; yet here it was intended precisely to carry us beyond that world. 2.

B 638 The inference from the impossibility of an infinite series of causes given one above another in the sensible world to a first cause. The principles of reason’s use do not entitle us to this even in experience, much less to be able to extend this principle beyond experience (to which this chain cannot be prolonged at all). 3. Reason’s false self-satisfaction regarding the completion of this series: finally removing every condition, without which no concept of necessity can occur, and, since we can then comprehend nothing further, taking this for a completion of the concept. 4. The confusion of the logical possibility of a concept of all reality united (without inner contradiction) with its transcendental possibility, which requires a principle of the feasibility of such a synthesis—a principle that in turn can concern only the field of possible experiences, and so on.

The artifice of the cosmological proof aims solely to evade proving the existence of a necessary being a priori through mere concepts, which would have to proceed ontologically, something we feel wholly incapable of doing. For this purpose we infer, as best we can, from an actual existence taken as our basis (an experience in general) to some absolutely necessary condition of it. We then have no need to explain that condition’s possibility.

B 639 For if its existence is proved, the question of its possibility is quite unnecessary. If we now wish to determine this necessary being more closely in its constitution, we seek not what is sufficient to comprehend the necessity of existence from its concept—for if we could do that, we would need no empirical presupposition—but only the negative condition (conditio sine qua non), without which a being would not be absolutely necessary. This would work well enough in every other kind of inference from a given consequence to its ground. Unfortunately, however, here the condition required for absolute necessity can be found in only one being, which would therefore have to contain in its concept everything required for absolute necessity and thus make an a priori inference to it possible. That is, I would also have to be able to infer conversely: whatever thing this concept (of the highest reality) belongs to is absolutely necessary. If I cannot infer in this way (as I must confess if I wish to avoid the ontological proof), I have also failed on my new path and find myself back where I began. The concept of the highest being does indeed satisfy a priori all questions that can be raised about the inner determinations of a thing.

B 640 It is therefore an incomparable ideal, because the universal concept also singles it out as an individual among all possible things. But it does not at all satisfy the question of its own existence, which was the sole point properly at issue. To someone who assumed the existence of a necessary being and wished only to know which among all things must be regarded as such, we could not answer: this here is the necessary being.

It may well be permissible to assume the existence of a being of the highest sufficiency as cause of all possible effects, to facilitate for reason the unity of explanatory grounds it seeks. But to presume so far as to say that such a being necessarily exists is no longer the modest expression of a permissible hypothesis, but the bold pretension to apodictic certainty. For the cognition of what we profess to cognize as absolutely necessary must itself carry absolute necessity with it.

The entire problem of the transcendental ideal comes down to finding either a concept for absolute necessity or, for the concept of some thing, its absolute necessity. If we can do one, we must also be able to do the other; for reason cognizes as absolutely necessary only what is necessary from its concept.

B 641 But both wholly exceed every utmost effort to satisfy our understanding on this point, and also every attempt to reconcile it to this incapacity.

The unconditioned necessity that we so indispensably need as the ultimate support of all things is the true abyss for human reason. Even eternity, however awesomely sublime a Haller may depict it, falls far short of producing the same dizzying impression on the mind; for it merely measures the duration of things, but does not support them. We cannot resist the thought, yet cannot bear it either, that a being we represent as the highest among all possible beings should say to itself, as it were: “I am from eternity to eternity; outside me there is nothing except what is something merely through my will. But where, then, do I come from?” Here everything beneath us gives way, and the greatest perfection, like the smallest, floats unsupported before speculative reason, which has no difficulty letting either vanish without the slightest hindrance.

Many forces of nature that manifest their existence through certain effects remain unfathomable to us, for we cannot trace them far enough through observation. The transcendental object underlying appearances is and remains unfathomable to us.

B 642 And so is the ground why our sensibility has these rather than other highest conditions, although the thing itself is otherwise given, only not understood. But an ideal of pure reason cannot be called unfathomable, because it can display no further warrant for its reality than reason’s need to complete all synthetic unity through it. Since it is thus not even given as a thinkable object, it is also not unfathomable as such. Rather, as a mere idea, it must find its seat and its solution in the nature of reason, and thus be capable of investigation. For reason consists precisely in our being able to account for all our concepts, opinions, and assertions, whether from objective grounds or, if they are mere illusion, from subjective ones.

Discovery and Explanation of the Dialectical Illusion in All Transcendental Proofs of the Existence of a Necessary Being

Both proofs conducted so far were transcendental, that is, attempted independently of empirical principles. For although the cosmological proof takes an experience in general as its basis, it proceeds not from any particular constitution of that experience but from pure principles of reason in relation to an existence given through empirical consciousness in general, and even abandons this guidance to rest on nothing but pure concepts.

B 643 What, then, in these transcendental proofs causes the dialectical yet natural illusion that connects the concepts of necessity and highest reality, realizing and hypostatizing what can only be an idea? What causes the inevitability of assuming something necessary in itself among existing things, yet at the same time recoiling before the existence of such a being as before an abyss? And how can reason be brought to understand itself on this point and pass from the wavering state of a timid assent, repeatedly withdrawn, to tranquil insight?

It is exceedingly noteworthy that if we presuppose that something exists, we cannot avoid the inference that something also exists necessarily. The cosmological argument rested on this entirely natural (though not therefore secure) inference. By contrast, whatever concept of a thing I may assume, I find that I can never represent its existence as absolutely necessary, and that nothing prevents me, whatever may exist, from thinking its nonbeing. Thus I must indeed assume something necessary for the existing in general, but can think no single thing itself as necessary in itself.

B 644 That is, I can never complete the regression to the conditions of existing without assuming a necessary being, but I can never begin from it.

If I must think something necessary for existing things in general, but am not entitled to think any thing as necessary in itself, it follows inevitably that necessity and contingency must not concern and apply to the things themselves, since otherwise a contradiction would arise. Consequently, neither of these principles is objective; at most they can be only subjective principles of reason: on the one hand, to seek for everything given as existing something necessary, that is, never to stop except at an explanation completed a priori; on the other, never to hope for this completion, that is, to assume nothing empirical as unconditioned and thereby exempt ourselves from further derivation. In this meaning, both principles can quite well coexist as merely heuristic and regulative, concerned only with reason’s formal interest. For one says: you should philosophize about nature as if everything belonging to existence had a necessary first ground, solely to bring systematic unity into your cognition by following such an idea, namely an imagined highest ground.

B 645 The other, however, warns you not to take any single determination concerning the existence of things as such a highest ground, that is, as absolutely necessary, but always to keep open the path to further derivation, and therefore always to treat it as still conditioned. But if we must regard everything perceived in things as conditionally necessary, then no thing (that may be empirically given) can be regarded as absolutely necessary either.

From this it follows, however, that you must assume the absolutely necessary outside the world, because it is to serve only as a principle of the greatest possible unity of appearances, as their highest ground, and you can never reach it within the world, since the second rule commands you always to regard all empirical causes of unity as derivative.

The ancient philosophers regarded all form in nature as contingent, but matter, according to the judgment of ordinary reason, as original and necessary. Had they considered matter in its existence not relatively, as the substratum of appearances, but in itself, the idea of absolute necessity would immediately have disappeared. For nothing absolutely binds reason to this existence; it can always abolish it in thought without conflict, yet absolute necessity lay only in thought as well. Some regulative principle must therefore have underlain this persuasion.

B 646 Indeed, extension and impenetrability (which together constitute the concept of matter) are the highest empirical principle of the unity of appearances, and insofar as it is empirically unconditioned, it possesses a property of the regulative principle. Nevertheless, since every determination of matter constituting its reality, and hence impenetrability too, is an effect (action) that must have its cause and is therefore always still derivative, matter is unsuited to the idea of a necessary being as a principle of all derivative unity. For each of its real properties, as derivative, is only conditionally necessary and can therefore be abolished in itself; but thereby matter’s entire existence would be abolished. If this did not happen, we would have reached empirically the highest ground of unity, which the second regulative principle forbids. It follows that matter, and generally whatever belongs to the world, is unsuited to the idea of a necessary original being as a mere principle of the greatest empirical unity, and that this being must instead be placed outside the world. We can then always confidently derive the appearances of the world and their existence from others, as if there were no necessary being.

B 647 And yet we can strive unceasingly for completeness of derivation, as if such a being were presupposed as a highest ground.

According to these considerations, the ideal of the highest being is nothing other than a regulative principle of reason: to regard every combination in the world as if it arose from an all-sufficient necessary cause, in order to ground on this the rule of a systematic unity, necessary according to universal laws, in explaining that combination. It is not an assertion of an existence necessary in itself. Yet it is at the same time inevitable, through a transcendental subreption, to represent this formal principle as constitutive and think this unity hypostatically. For just as space, because it originally makes possible all shapes, which are merely different limitations of it, is nevertheless taken for that very reason as an absolutely necessary, self-subsisting something and an object given a priori in itself, although it is only a principle of sensibility, so too it happens quite naturally that, since the systematic unity of nature can be established as a principle of the empirical use of our reason only insofar as we take the idea of a most real being as highest cause as our foundation, this idea is thereby represented as an actual object, and this object in turn, because it is the highest condition, as necessary.

B 648 Thus a regulative principle is transformed into a constitutive one. This substitution reveals itself in the fact that, if I now consider this highest being—which was absolutely (unconditionally) necessary in relation to the world—as a thing on its own, this necessity admits of no concept. Hence it must have been found in my reason only as a formal condition of thinking, not as a material and hypostatic condition of existence.

Chapter Three Section Six The Impossibility of the Physicotheological Proof

If neither the concept of things in general nor the experience of any existence in general can accomplish what is required, one means remains to be tried: whether a determinate experience, namely that of the things of the present world, their constitution and arrangement, provides a ground of proof that can securely convince us of the existence of a highest being. Such a proof we would call physicotheological. If this too should be impossible, then no satisfactory proof from merely speculative reason is possible at all for the existence of a being corresponding to our transcendental idea.

B 649 After all the remarks above, it will readily be seen that we can expect a quite easy and conclusive answer to this inquiry. For how could experience ever be given that was commensurate with an idea? The peculiarity of the latter consists precisely in the fact that no experience can ever be congruent with it. The transcendental idea of a necessary, all-sufficient original being is so extravagantly great, so elevated above everything empirical, which is always conditioned, that on the one hand we can never find enough material in experience to fill such a concept, and on the other we will always grope among the conditioned, searching forever in vain for the unconditioned, of which no law of any empirical synthesis gives us an example or the slightest guidance.

If the highest being stood in this chain of conditions, it would itself be a member of their series and, like the lower members placed beneath it, would require further investigation concerning its still higher ground. If, by contrast, we wish to separate it from this chain and not include it, as a merely intelligible being, in the series of natural causes, what bridge can reason build to reach it? For all laws of transition from effects to causes, indeed all synthesis and extension of our cognition in general, are founded on nothing other than possible experience.

B 650 They are thus founded merely on objects of the sensible world and can have meaning only in relation to them.

The present world opens before us so immeasurable a theater of manifoldness, order, purposiveness, and beauty, whether we pursue these into the infinity of space or into its unbounded division, that even after the knowledge our feeble understanding has been able to acquire of it, all language lacks its force before so many unsurveyably great wonders, all numbers their power to measure, and even our thoughts all boundary. Thus our judgment of the whole must dissolve into a speechless yet all the more eloquent astonishment. Everywhere we see a chain of effects and causes, of ends and means, regularity in arising and passing away; and since nothing has entered of itself into the state in which it is found, that state always points further to another thing as its cause, which makes precisely the same further inquiry necessary. In this way the entire universe would have to sink into the abyss of nothingness unless we assumed something that, outside this infinite contingent whole, originally and independently subsisting on its own, sustained it and, as the cause of its origin, also secured its continuance. This highest cause (in regard to all things of the world)—how great are we to think it?

B 651 We do not know the world in its entire content, still less do we know how to estimate its magnitude by comparison with everything possible. But since we once need an ultimate and highest being with regard to causality, what prevents us from placing it above everything else possible in degree of perfection as well? We can easily accomplish this, though admittedly only through the delicate outline of an abstract concept, by representing in it, as one single substance, all possible perfection united. This concept favors our reason’s demand for economy of principles, is subject to no contradictions in itself, and even helps extend reason’s use within experience through the guidance such an idea gives toward order and purposiveness, while nowhere decisively conflicting with any experience.

This proof always deserves to be mentioned with respect. It is the oldest, clearest, and most suited to ordinary human reason. It enlivens the study of nature, just as it owes its own existence to that study and continually gains new strength from it. It introduces ends and intentions where our observation would not have discovered them on its own, and extends our knowledge of nature along the guiding thread of a special unity whose principle lies outside nature.

B 652 But this knowledge in turn acts back on its cause, namely the idea that prompted it, increasing faith in a highest author to an irresistible conviction.

It would therefore be not only disheartening but quite futile to try to diminish the authority of this proof. Reason, ceaselessly elevated by grounds of proof so powerful and continually growing in its hands, though only empirical, cannot be so depressed by the doubts of subtle abstract speculation that a glance at the wonders of nature and the majesty of the world’s structure would not tear it from every brooding indecision as from a dream, to rise from greatness to greatness up to the very highest, from the conditioned to the condition up to the supreme and unconditioned author.

Yet although we have nothing to object to in the rationality and usefulness of this procedure, and must instead recommend and encourage it, we cannot therefore approve the claims this kind of proof might make to apodictic certainty and to an assent needing no favor or outside support. And it can by no means harm the good cause to lower the dogmatic language of a scornful reasoner to the tone of moderation and modesty, of a faith sufficient to bring peace, though not commanding unconditional submission.

B 653 I therefore maintain that the physicotheological proof can never by itself demonstrate the existence of a highest being, but must always leave it to the ontological proof (to which it serves only as an introduction) to remedy this deficiency. The latter thus still contains the only possible ground of proof (if a speculative proof takes place at all) which no human reason can bypass.

The main elements of this physicotheological proof are these: 1. Everywhere in the world are clear signs of an arrangement according to a determinate intention, carried out with great wisdom, in a whole of indescribable manifoldness of content and unbounded magnitude of extent. 2. This purposive arrangement is quite foreign to the things of the world and attaches to them only contingently. That is, the nature of different things could not of itself harmonize toward determinate ultimate ends through so many different means converging together unless they had expressly been chosen and designed for this by an ordering rational principle according to underlying ideas. 3. Thus there exists a sublime and wise cause (or several), which must be the cause of the world not merely through fecundity, as blindly acting omnipotent nature, but through freedom, as intelligence. 4. Its unity can be inferred from the unity of the reciprocal relation of the world’s parts.

B 654 These are members of an artful structure; the inference is certain in what our observation reaches, and beyond that probable according to all principles of analogy.

Natural reason infers, from the analogy of certain natural products with what human art produces when it does violence to nature, compelling it to proceed not according to its own ends but to accommodate ours (from their resemblance to houses, ships, and clocks), that just such a causality, namely understanding and will, also underlies nature. It derives the inner possibility of freely operating nature (which first makes all art, and perhaps even reason itself, possible) from yet another, though superhuman, art. Without quibbling here with this inference, which might not withstand the strictest transcendental critique, we must nevertheless confess that if we are once to name a cause, we can proceed no more securely than by analogy with such purposive productions, the only ones whose causes and mode of operation are fully known to us. Reason could not justify to itself passing from a causality it knows to obscure and unprovable grounds of explanation it does not know.

B 655 According to this inference, the purposiveness and harmony of so many natural arrangements would have to prove merely the contingency of form, not of matter, that is, of the substance in the world. For the latter would additionally require proof that the things of the world were in themselves unsuited to such order and harmony according to universal laws unless, even in their substance, they were products of a highest wisdom. This would require quite different grounds of proof from those of analogy with human art. Thus the proof could at most demonstrate an architect of the world, always very much limited by the suitability of the material he works with, but not a creator of the world, to whose idea everything is subject. This falls far short of sufficing for the great aim before us: proving an all-sufficient original being. If we wished to prove the contingency of matter itself, we would have to take refuge in a transcendental argument, which was precisely what was to be avoided here.

The inference therefore proceeds from the order and purposiveness so thoroughly observable in the world, as a thoroughly contingent arrangement, to the existence of a cause proportionate to it. But the concept of this cause must make something quite determinate about it known to us; thus it can be none other than the concept of a being possessing all power, wisdom, and so on—in a word, all perfection—as an all-sufficient being.

B 656 For predicates of very great, astonishing, or immeasurable power and excellence give no determinate concept and do not properly say what the thing is in itself. They are only relative representations of the magnitude of the object that the observer (of the world) compares with himself and his power of comprehension, and turn out equally laudatory whether we enlarge the object or diminish the observing subject in relation to it. Where the magnitude (of perfection) of a thing in general is at issue, no concept is determinate except one comprehending all possible perfection, and only the all (omnitudo) of reality is thoroughly determined in the concept.

Now I would not expect anyone to presume to understand the relation of the world’s magnitude as he observes it (both in extent and content) to omnipotence, of the world’s order to the highest wisdom, of the world’s unity to the author’s absolute unity, and so on. Thus physicotheology can give no determinate concept of the supreme cause of the world, and therefore cannot suffice as a principle of theology, which in turn is to constitute the foundation of religion.

The step to absolute totality is wholly impossible by the empirical path. Yet this step is nevertheless taken in the physicotheological proof.

B 657 What means, then, is used to cross so vast a gulf?

After reaching admiration for the greatness of the wisdom, power, and so on of the author of the world, and being unable to go further, one suddenly abandons this argument conducted through empirical grounds of proof and turns to the world’s contingency, inferred at the very beginning from its order and purposiveness. From this contingency alone one now proceeds, solely through transcendental concepts, to the existence of something absolutely necessary; and from the concept of the absolute necessity of the first cause, to its thoroughly determined or determining concept, namely an all-embracing reality. Thus the physicotheological proof became stuck in its undertaking and, in this predicament, suddenly leaped over to the cosmological proof. Since that is only a concealed ontological proof, it actually accomplished its aim solely through pure reason, although at first it had denied all kinship with pure reason and staked everything on evident proofs from experience.

Physicotheologians therefore have no reason to be so disdainful toward the transcendental kind of proof and, with the conceit of clear-sighted students of nature, look down on it as the cobweb of obscure speculators. For if they would only examine themselves, they would find the following.

B 658 After advancing a good distance on the ground of nature and experience and nevertheless seeing themselves still just as far from the object that appears before their reason, they suddenly abandon this ground and pass into the realm of mere possibilities. There, on the wings of ideas, they hope to approach what has eluded all their empirical searching. After finally imagining they have gained firm footing through so mighty a leap, they extend the now determinate concept (which they have come to possess without knowing how) over the whole field of creation, and elucidate through experience—though poorly enough and far beneath the dignity of its object—the ideal that was solely a product of pure reason, without wishing to confess that they reached this knowledge or presupposition by a path other than experience.

Thus the physicotheological proof rests on the cosmological, and that in turn on the ontological proof of the existence of one original being as the highest being. Since no path beyond these three is open to speculative reason, the ontological proof from nothing but pure concepts of reason is the only possible one, if any proof at all is possible of a proposition so far elevated above every empirical use of the understanding.

B 659 Chapter Three Section Seven Critique of All Theology from Speculative Principles of Reason

If by theology I understand cognition of the original being, it is either from mere reason (theologia rationalis) or from revelation (revelata). The former thinks its object either solely through pure reason, by means of nothing but transcendental concepts (ens originarium, realissimum, ens entium), and is called transcendental theology; or through a concept it borrows from nature (our soul), as the highest intelligence, and would have to be called natural theology. Someone who admits only a transcendental theology is called a deist; someone who also accepts a natural theology, a theist. The former concedes that we can at most cognize the existence of an original being through mere reason, but that our concept of it is merely transcendental: only that of a being possessing all reality, which we cannot determine more closely. The latter maintains that reason is capable of determining the object more closely by analogy with nature: as a being that through understanding and freedom contains within itself the original ground of all other things.

B 660 Thus the former represents by it merely a cause of the world (whether through the necessity of its nature or through freedom remains undecided), the latter an author of the world.

Transcendental theology either intends to derive the existence of the original being from an experience in general (without determining anything more closely about the world to which it belongs), and is called cosmotheology; or believes it can cognize its existence through mere concepts without the aid of the slightest experience, and is called ontotheology.

Natural theology infers the properties and existence of an author of the world from the constitution, order, and unity found in this world, in which two kinds of causality and their rule must be assumed, namely nature and freedom. It therefore ascends from this world to the highest intelligence, either as the principle of all natural order and perfection or of all moral order and perfection. In the former case it is called physicotheology, in the latter moral theology.*

* Not theological morality; for that contains moral laws presupposing the existence of a highest ruler of the world, whereas moral theology is a conviction of the existence of a highest being grounded in moral laws.

By the concept of God we are accustomed to understand not merely a blindly acting eternal nature as the root of things, but a highest being.

B 661 It is to be the author of things through understanding and freedom. And since this concept alone interests us, strictly speaking we could deny the deist all faith in God, leaving him only the assertion of an original being or supreme cause. Yet because no one may be accused of wishing to deny something merely because he does not venture to assert it, it is gentler and fairer to say that the deist believes in a God, the theist in a living God (summam intelligentiam). We will now seek the possible sources of all these attempts of reason.

I am content here to explain theoretical cognition as cognition through which I cognize what is, practical cognition as that through which I represent what ought to be. Accordingly, the theoretical use of reason is that through which I cognize a priori (as necessary) that something is; the practical use is that through which what ought to happen is cognized a priori. Now if either that something is, or that it ought to happen, is indubitably certain but still only conditioned, then either a certain determinate condition of it can be absolutely necessary, or it can be presupposed only as optional and contingent. In the first case the condition is postulated (per thesin), in the second supposed (per hypothesin).

B 662 Since there are practical laws that are absolutely necessary (the moral laws), if these necessarily presuppose some existence as the condition of the possibility of their binding force, this existence must be postulated, because the conditioned from which we infer this determinate condition is itself cognized a priori as absolutely necessary. Later we will show of moral laws that they not only presuppose the existence of a highest being, but also, since in another respect they are absolutely necessary, rightly postulate it—though admittedly only practically. For now we set this mode of inference aside.

Since, when we speak merely of what is (not what ought to be), the conditioned given to us in experience is always also thought as contingent, its corresponding condition cannot be cognized from it as absolutely necessary. Rather, it serves for rational cognition of the conditioned only as a relatively necessary—or, more properly, requisite—presupposition, but one arbitrary in itself and a priori. Thus if the absolute necessity of a thing is to be cognized in theoretical cognition, this could occur only from concepts a priori, never as the necessity of a cause in relation to an existence given through experience.

A theoretical cognition is speculative if it concerns an object, or concepts of an object, that we can reach in no experience.

B 663 It is opposed to cognition of nature, which concerns no objects or predicates of them except those that can be given in a possible experience.

The principle of inferring a cause from what happens (the empirically contingent) as effect is a principle of cognition of nature, but not of speculative cognition. For if we abstract from its being a principle containing the condition of possible experience in general and, omitting everything empirical, intend to assert it of the contingent in general, not the slightest justification of such a synthetic proposition remains by which to see how I can pass from something that is to something entirely different from it (called a cause). Indeed, in such merely speculative use, the concept of a cause, like that of the contingent, loses all meaning whose objective reality could be made comprehensible in concreto.

Now if we infer from the existence of things in the world to their cause, this belongs not to the natural but to the speculative use of reason. For the former refers not things themselves (substances), but only what happens, hence their states as empirically contingent, to some cause. That substance itself (matter) is contingent in existence would have to be a merely speculative rational cognition.

B 664 But even if we spoke only of the world’s form, the manner of its combination and its changes, yet wished to infer from this a cause wholly distinct from the world, this would again be a judgment of merely speculative reason, because the object here is no object of a possible experience at all. The principle of causality, however, which holds only within the field of experiences and outside it has no use, indeed no meaning, would then be wholly diverted from its purpose.

I now maintain that all attempts at a merely speculative use of reason regarding theology are wholly fruitless and, by their inner constitution, null and void; but that the principles of reason’s natural use lead to no theology at all. Consequently, unless moral laws are taken as a basis or used as a guiding thread, there can be no theology of reason whatever. For all synthetic principles of the understanding have an immanent use, whereas cognition of a highest being requires a transcendent use of them, for which our understanding is not equipped at all. If the empirically valid law of causality is to lead to the original being, this being would have to belong to the chain of objects of experience; but then, like all appearances, it would itself be conditioned in turn.

B 665 Even if we permitted the leap beyond the boundary of experience by means of the dynamical law relating effects to their causes, what concept could this procedure supply? By no means a concept of a highest being, because experience never presents us with the greatest of all possible effects (which is supposed to bear witness to its cause). If, merely to leave nothing empty in our reason, we are to be allowed to fill this lack of complete determination through a mere idea of the highest perfection and original necessity, this can indeed be granted as a favor, but not demanded by right of an irresistible proof. Thus the physicotheological proof might perhaps lend force to other proofs, if such are available, by connecting speculation with intuition; but by itself it does more to prepare the understanding for theological cognition and give it a straight and natural direction toward it than it could complete the work alone.

We can thus clearly see that transcendental questions admit only transcendental answers, that is, answers from nothing but concepts a priori without the slightest empirical admixture. But here the question is plainly synthetic and demands an extension of our cognition beyond all boundaries of experience: to the existence of a being supposed to correspond to our mere idea, which no experience can ever equal.

B 666 Now according to our proofs above, all synthetic cognition a priori is possible only by expressing the formal conditions of a possible experience, and all principles thus have only immanent validity: they refer solely to objects of empirical cognition, or appearances. Thus nothing is accomplished by a transcendental procedure either, regarding the theology of a merely speculative reason.

But if someone would rather cast doubt on all the above proofs of the Analytic than let himself be deprived of his persuasion about the weight of the long-used grounds of proof, he cannot refuse to meet the challenge when I demand that he at least justify how, and through what illumination, he ventures to soar above all possible experience by the power of mere ideas. I would ask to be spared new proofs or revisions of old ones. Admittedly, there is not much choice here, since all merely speculative proofs ultimately come down to a single one, the ontological, so I need not fear being particularly burdened by the fertility of the dogmatic champions of that reason free of the senses.

B 667 Moreover, without imagining myself especially combative, I will not decline the challenge to expose the fallacy in every attempt of this kind and thereby frustrate its pretension. Yet this never entirely extinguishes the hope of better fortune in those once accustomed to dogmatic persuasions. I therefore hold to the single reasonable demand that they justify, in general and from the nature of human understanding together with all other sources of cognition, how they intend to extend their cognition wholly a priori, carrying it where no possible experience, and hence no means, reaches to assure any concept we have devised ourselves of its objective reality. However the understanding may have arrived at this concept, the existence of its object cannot be found analytically in it, because cognition of the object’s existence consists precisely in its being posited in itself outside the thought. But it is wholly impossible to go outside a concept of itself and, without following empirical connection (through which only appearances are ever given), arrive at the discovery of new objects and extravagant beings.

Although reason in its merely speculative use falls far short of sufficing for this great aim, namely reaching the existence of a supreme being, it is nevertheless of very great use.

B 668 It corrects cognition of that being, if it could be drawn from elsewhere; makes it agree with itself and with every intelligible purpose; and purifies it of everything that might conflict with the concept of an original being and of every admixture of empirical limitations.

Despite all its inadequacy, transcendental theology thus retains an important negative use and is a constant censor of our reason when reason deals only with pure ideas, which for that very reason admit no standard other than a transcendental one. For if, in another, perhaps practical relation, the presupposition of a highest and all-sufficient being as supreme intelligence once maintained its validity without dispute, it would be of the greatest importance to determine this concept precisely on its transcendental side as the concept of a necessary and most real being; to remove what conflicts with the highest reality and belongs to mere appearance (anthropomorphism in the wider sense); and at the same time to clear away all opposed assertions, whether atheistic, deistic, or anthropomorphic. This is very easy in such a critical treatment, since the same grounds by which human reason’s inability to assert the existence of such a being is displayed must also suffice to prove the unfitness of every counterassertion.

B 669 For where could anyone derive through pure speculation of reason the insight that there is no highest being as the original ground of everything; or that none of the properties belongs to it that, according to their consequences, we represent as analogous to the dynamical realities of a thinking being; or that, in the latter case, these properties must also be subject to all the limitations that sensibility inevitably imposes on the intelligences we know through experience?

Thus for the merely speculative use of reason, the highest being remains a mere but faultless ideal, a concept that completes and crowns the whole of human cognition, whose objective reality can neither be proved nor refuted on this path. And if there should be a moral theology capable of remedying this deficiency, the previously merely problematic transcendental theology then proves its indispensability by determining its concept and ceaselessly censuring a reason often enough deceived by sensibility and not always in agreement with its own ideas. Necessity, infinity, unity, existence outside the world (not as a world-soul), eternity without conditions of time, omnipresence without conditions of space, omnipotence, and so on, are all transcendental predicates.

B 670 Hence their purified concept, which every theology so greatly needs, can be drawn solely from transcendental theology.

Appendix to the Transcendental Dialectic The Regulative Use of the Ideas of Pure Reason

The outcome of all dialectical attempts of pure reason confirms not only what we have already proved in the Transcendental Analytic: that all inferences meant to carry us beyond the field of possible experience are deceptive and unfounded. It also teaches us something more specific: human reason has a natural tendency to cross this boundary, and transcendental ideas are as natural to it as the categories are to the understanding. The difference is that, while the latter lead to truth—that is, to the agreement of our concepts with the object—the former produce a mere but irresistible illusion whose deception even the most searching critique can scarcely prevent.

Everything grounded in the nature of our powers must be purposive and consistent with their proper use.

B 671 That holds, if only we can prevent a certain misunderstanding and discover their true direction. We may therefore presume that transcendental ideas have a sound and consequently immanent use, although, when their meaning is misunderstood and they are taken for concepts of actual things, their application can become transcendent and thus deceptive. For it is not the idea itself but only its use that can either soar beyond all possible experience (be transcendent) or remain at home within it (be immanent), depending on whether we direct it immediately to an object supposedly corresponding to it or merely to the use of the understanding in general with regard to the objects with which it deals. All errors of subreption must always be attributed to a deficiency in the power of judgment, never to the understanding or reason.

Reason never relates directly to an object, but only to the understanding and, through it, to its own empirical use. It therefore creates no concepts (of objects), but merely orders them and gives them the unity they can possess in their greatest possible extension: that is, in relation to the totality of series. The understanding pays no attention to this totality, but only to the connection through which series of conditions arise everywhere in accordance with concepts.

B 672 Thus reason has as its proper object only the understanding and its purposive employment. Just as the understanding unites the manifold in the object through concepts, reason in turn unites the manifold of concepts through ideas, setting a certain collective unity as the goal of the understanding’s actions, which are otherwise concerned only with distributive unity.

I therefore maintain that transcendental ideas never have a constitutive use, as though they supplied concepts of certain objects; when understood in this way, they are merely sophistical (dialectical) concepts. They do, however, have an excellent and indispensably necessary regulative use: they direct the understanding toward a certain goal, with a view to which the lines of direction of all its rules converge at a point. Although this point is only an idea (focus imaginarius), that is, a point from which the concepts of the understanding do not actually proceed, since it lies entirely outside the boundaries of possible experience, it nevertheless serves to give them the greatest unity together with the greatest extension. This does indeed give rise to the deception that these lines of direction issue from an object itself lying outside the field of empirically possible cognition (just as objects are seen behind the surface of a mirror). Yet this illusion (which can be prevented from deceiving us) is nonetheless indispensably necessary.

B 673 That is so if, besides the objects before our eyes, we also want to see those lying far off behind our backs—that is, in our case, if we want to train the understanding to go beyond every given experience (as a part of the whole of possible experience), and thus to its greatest and furthest possible extension.

If we survey the cognitions of our understanding in their full extent, we find that what reason peculiarly prescribes for them and seeks to bring about is the systematic character of cognition: its coherence through a principle. This unity of reason always presupposes an idea, namely that of the form of a whole of cognition that precedes determinate cognition of the parts and contains the conditions for determining a priori each part’s place and relation to the others. This idea accordingly postulates complete unity in the cognition of the understanding, making it not a merely accidental aggregate but a system connected according to necessary laws. Strictly speaking, we cannot say that this idea is a concept of the object; it is a concept of the thoroughgoing unity of these concepts, insofar as that unity serves the understanding as a rule. Such concepts of reason are not drawn from nature. Rather, we question nature in accordance with these ideas and regard our cognition as deficient so long as it is not adequate to them.

B 674 It is admitted that pure earth, pure water, pure air, and so on can scarcely be found. Nevertheless, we need concepts of them (which, as regards complete purity, therefore originate only in reason) to determine properly the share each of these natural causes has in the appearance. We thus reduce all materials to earths (as it were the mere load), salts and combustible substances (as the force), and finally water and air as vehicles (as it were machines through which the former act), in order to explain the chemical interactions of materials according to the idea of a mechanism. For although people do not actually express themselves this way, such an influence of reason on natural scientists’ classifications is very easy to discover.

If reason is a faculty for deriving the particular from the universal, then either the universal is already certain in itself and given, in which case only the power of judgment is needed for subsumption, and the particular is thereby necessarily determined. I shall call this the apodictic use of reason. Or the universal is assumed only problematically and is a mere idea; the particular is certain, but the universality of the rule for this consequence remains a problem. Several particular cases, all of them certain, are then tested against the rule to see whether they follow from it.

B 675 In that case, if it appears that all the particular cases that can be cited follow from it, we infer the universality of the rule, and subsequently infer from this universality all cases, including those not given in themselves. I shall call this the hypothetical use of reason.

The hypothetical use of reason, proceeding from ideas laid down as problematic concepts, is not properly constitutive. That is, judged with complete rigor, it is not such that the truth of the universal rule assumed as a hypothesis follows from it. For how could we know all the possible consequences which, by following from the same assumed principle, would prove its universality? Rather, it is only regulative, serving to bring unity into particular cognitions as far as possible and thereby to bring the rule closer to universality.

The hypothetical use of reason thus aims at the systematic unity of the understanding’s cognitions, and this unity is the touchstone of the truth of rules. Conversely, systematic unity (as a mere idea) is only a projected unity, which must be regarded not as given in itself but only as a problem. Yet it serves to find a principle for the manifold and particular use of the understanding, and thereby to guide that use beyond the cases that are given and make it coherent.

B 676 From this we see only that the systematic unity, or unity of reason, of the understanding’s manifold cognition is a logical principle. Where the understanding alone cannot arrive at rules, this principle assists it through ideas; at the same time it brings its diverse rules into agreement under one principle (systematic) and thus gives them coherence as far as possible. But whether the constitution of objects, or the nature of the understanding that cognizes them as such, is in itself destined for systematic unity; whether this unity can to some degree be postulated a priori, even without regard to such an interest of reason; and whether we can therefore say that all possible cognitions of the understanding (including empirical ones) possess unity of reason and fall under common principles from which, despite their diversity, they can be derived—this would be a transcendental principle of reason. It would make systematic unity necessary not merely subjectively and logically, as a method, but objectively.

Let us illustrate this with a case of reason’s use. Among the different kinds of unity according to concepts of the understanding is the unity of a substance’s causality, which is called force. The various appearances of one and the same substance at first seem so heterogeneous that we must initially assume almost as many kinds of force in it as there are effects that manifest themselves.

B 677 Examples in the human mind are sensation, consciousness, imagination, memory, wit, the power of discrimination, pleasure, desire, and so on. A logical maxim initially commands us to reduce this apparent diversity as far as possible by discovering hidden identity through comparison and examining whether imagination combined with consciousness might be memory, wit, the power of discrimination, perhaps even understanding and reason. The idea of a fundamental power—though logic does not at all establish whether such a thing exists—is at least the problem of a systematic representation of the manifold of forces. The logical principle of reason requires us to bring about this unity as far as possible. The more the appearances of one force and another are found to be identical, the more probable it becomes that they are merely different manifestations of one and the same force, which can be called (comparatively) their fundamental power. We proceed in the same way with the rest.

These comparatively fundamental powers must in turn be compared with one another, so that, by discovering their agreement, we may bring them closer to a single radical, that is, absolute fundamental power. But this unity of reason is merely hypothetical. We do not assert that such a fundamental power must actually be found, but that we must seek it for reason’s sake, namely to establish certain principles for the various rules that experience may supply.

B 678 In this way, wherever it can be done, we must bring systematic unity into cognition.

Yet if we attend to the transcendental use of the understanding, we find that this idea of a fundamental power in general is not intended merely as a problem for hypothetical use. It lays claim to objective reality, thereby postulating the systematic unity of the various forces of a substance and establishing an apodictic principle of reason. For even before we have tested the agreement of the various forces—and even if all our attempts to discover it fail—we nevertheless presuppose that such agreement will be found. We do so not only, as in the case cited, on account of the unity of the substance. Where very many forces, though to some degree homogeneous, are found, as in matter in general, reason likewise presupposes the systematic unity of manifold forces, since particular laws of nature fall under more general ones, and economy of principles becomes not merely an economical principle of reason but an inner law of nature.

Indeed, it is impossible to see how a logical principle of the unity of reason in rules could hold unless a transcendental principle were presupposed, through which such systematic unity, as belonging to the objects themselves, is assumed a priori to be necessary.

B 679 For by what right can reason, in its logical use, require us to treat the manifold forces that nature makes known to us as a merely hidden unity, and to derive them as far as possible from some fundamental power, if reason were free to concede that all forces might equally well be heterogeneous and that systematic unity in their derivation might not accord with nature? In that case, reason would act directly against its own vocation by making an idea its goal that wholly contradicted nature’s arrangement. Nor can we say that reason has first gathered this unity, according to its own principles, from nature’s contingent constitution. For reason’s law requiring us to seek this unity is necessary: without it we would have no reason at all; without reason, no coherent use of the understanding; and without this, no sufficient mark of empirical truth. With regard to this last, therefore, we must absolutely presuppose nature’s systematic unity as objectively valid and necessary.

We also find this transcendental presupposition concealed in a remarkable way in philosophers’ principles, although they have not always recognized it there or admitted it to themselves.

B 680 That all the manifold differences of individual things do not exclude identity of species; that the various species must be treated merely as different determinations of a few genera, and these in turn of still higher genera, and so on; and that a certain systematic unity of all possible empirical concepts must therefore be sought, insofar as they can be derived from higher and more general ones—this is a scholastic rule or logical principle without which reason could not be used at all. For we can infer from the universal to the particular only insofar as universal properties of things are laid down under which the particular properties fall.

But philosophers also presuppose that such agreement is found in nature in the familiar scholastic rule that beginnings (principles) must not be multiplied unnecessarily (entia praeter necessitatem non esse multiplicanda). This says that the nature of things itself provides material for the unity of reason, and that its apparently infinite diversity must not prevent us from suspecting behind it a unity of fundamental properties, from which the manifold can be derived merely through further determination. Although this unity is a mere idea, it has always been pursued so eagerly that there has been more reason to moderate the desire for it than to encourage it. It was already a considerable achievement for chemists to reduce all salts to two principal genera, acids and alkalis; they even attempt to regard this distinction itself as merely a variation or different manifestation of one and the same basic material.

B 681 People have gradually tried to reduce the various kinds of earths (the material of stones and even of metals) to three, and finally to two. Yet, still dissatisfied, they cannot abandon the thought of a single genus behind these varieties, or even of a principle common to them and to salts. One might perhaps believe this to be merely an economical expedient of reason, intended to save as much effort as possible, and a hypothetical attempt which, if successful, lends probability to the presupposed ground of explanation through this very unity. But such a self-serving intention is very easily distinguished from the idea according to which everyone presupposes that this unity of reason accords with nature itself, and that reason here does not beg but commands, although it cannot determine the boundaries of this unity.

Suppose the appearances presented to us displayed such great diversity—not, I mean, of form, for they might resemble one another in that respect, but of content, that is, in the manifold of existing beings—that even the sharpest human understanding could not, by comparing one with another, discover the slightest similarity. Such a case can certainly be conceived. The logical law of genera would then have no place at all.

B 682 There would be no concept of a genus or any universal concept, indeed no understanding, since the understanding deals only with such concepts. Thus the logical principle of genera presupposes a transcendental principle if it is to be applied to nature (by which I here understand only objects given to us). According to this principle, homogeneity is necessarily presupposed in the manifold of a possible experience (although we cannot determine its degree a priori); for without it no empirical concepts, and therefore no experience, would be possible.

The logical principle of genera, which postulates identity, is opposed by another, that of species. This requires a manifold and differences among things despite their agreement under the same genus, and prescribes that the understanding attend no less to the former than to the latter. This principle (of acuteness, or of the faculty of discrimination) greatly restricts the recklessness of the first principle (that of wit). Here reason displays a twofold, conflicting interest: on the one hand, an interest in extension (universality) with regard to genera; on the other, an interest in content (determinacy) with regard to the manifold of species, because in the first case the understanding thinks much under its concepts, in the second all the more within them.

B 683 This is also expressed in the very different ways natural scientists think. Some (who are chiefly speculative), as it were hostile to heterogeneity, always look toward the unity of the genus; others (chiefly empirical minds) constantly seek to divide nature into so great a manifold that we might almost have to abandon hope of judging its appearances according to universal principles.

This latter way of thinking clearly also rests on a logical principle aimed at the systematic completeness of all cognitions when, beginning with the genus, I descend to the manifold that may be contained under it, seeking in this way to give the system extension, just as in the first case, by ascending to the genus, I seek to give it simplicity. For the sphere of the concept that designates a genus no more reveals how far its division can proceed than does the space that matter can occupy. Every genus therefore requires different species, and these in turn different subspecies. Since none of the latter can exist without again having a sphere (extension as conceptus communis), reason, in its full scope, requires that no species be regarded as in itself the lowest. For it always remains a concept containing only what different things have in common; it therefore cannot be thoroughly determined, and so cannot relate immediately to an individual.

B 684 It must consequently always contain other concepts, that is, subspecies, under itself. This law of specification could be expressed as follows: entium varietates non temere esse minuendas.

It is easy to see, however, that this logical law too would lack meaning and application unless it rested on a transcendental law of specification. This law certainly does not demand an actual infinity of differences among the things that can become our objects; the logical principle gives no grounds for that, since it asserts only the indeterminacy of the logical sphere with regard to possible division. Nevertheless, it obliges the understanding to seek subspecies under every species we encounter, and smaller differences within every difference. For if there were no lower concepts, there would be no higher ones either. Now the understanding cognizes everything only through concepts; hence, however far it proceeds in division, it never cognizes through mere intuition but always through lower concepts in turn. Cognition of appearances in their thoroughgoing determination (which is possible only through the understanding) requires an endlessly continued specification of its concepts and a progression toward differences that still remain, from which abstraction was made in the concept of the species and still more in that of the genus.

B 685 Nor can this law of specification be borrowed from experience, for experience cannot yield disclosures of such reach. Empirical specification soon comes to a halt in distinguishing the manifold unless it is guided by the antecedent transcendental law of specification, as a principle of reason, to seek differences and always to suspect them even when they do not reveal themselves to the senses. The discovery that absorbent earths are of different kinds (calcareous and muriatic earths) required a prior rule of reason. This set the understanding the task of seeking the difference, presupposing nature to be rich enough to warrant suspecting it. For we possess understanding only on the presupposition of differences in nature, just as we possess it only on the condition that nature’s objects are homogeneous in themselves, because the manifold of what can be gathered under a concept is precisely what constitutes the use of that concept and the activity of the understanding.

Reason thus prepares the understanding’s field for it: 1. through a principle of the homogeneity of the manifold under higher genera; 2. through a principle of the variety of the homogeneous under lower species; and, to complete systematic unity, 3. it adds a law of the affinity of all concepts.

B 686 This law commands a continuous transition from every species to every other through a graduated increase in difference. We can call these the principles of homogeneity, specification, and continuity of forms. The last arises from uniting the first two after we have completed systematic coherence in the idea, both in ascending to higher genera and in descending to lower species. For all the manifold is then related, because all of it descends through every degree of expanded determination from a single highest genus.

We can give sensible form to systematic unity under these three logical principles in the following way. Every concept can be regarded as a point which, as the standpoint of a spectator, has its horizon: a multitude of things that can be represented and, as it were, surveyed from it. Within this horizon an infinite multitude of points must be assignable, each of which in turn has its narrower field of view. That is, every species contains subspecies according to the principle of specification, and the logical horizon consists only of smaller horizons (subspecies), not of points without extension (individuals). But for different horizons—that is, genera determined by as many concepts—we can conceive a common horizon drawn, from which they can all be surveyed as from a center.

B 687 This is the higher genus, until finally the highest genus is the universal and true horizon, determined from the standpoint of the highest concept and encompassing all the manifold as genera, species, and subspecies.

The law of homogeneity leads me to this highest standpoint; the law of specification leads me to all lower ones and their greatest variety. But since in this way there is nothing empty throughout the entire extension of all possible concepts, and nothing can be found outside it, the presupposition of that universal field of view and its thoroughgoing division gives rise to the principle: non datur vacuum formarum. That is, there are no different original and first genera that are, as it were, isolated and separated from one another (by an empty intervening space). Rather, all the manifold genera are merely divisions of a single highest and universal genus. From this principle follows its immediate consequence: datur continuum formarum. That is, all differences of species border on one another and permit no transition from one to another by a leap, but only through all the smaller degrees of difference by which one can get from one to the other. In a word, no species or subspecies are nearest to one another (in reason’s concept); intermediate species are always still possible.

B 688 Their difference from the first and the second is smaller than the difference of those two from one another.

The first law thus prevents us from straying into a manifold of different original genera and recommends homogeneity. The second in turn restricts this inclination toward agreement and commands us to distinguish subspecies before turning with our universal concept to individuals. The third unites the two: even amid the greatest manifold, it prescribes homogeneity through a graduated transition from one species to another. This indicates a kind of kinship among the different branches, insofar as they have all sprung from one stem.

But this logical law of the continui specierum (formarum logicarum) presupposes a transcendental one (lex continui in natura). Without it, the understanding’s use would only be misled by this precept, since it might take a path directly opposed to nature. This law must therefore rest on purely transcendental, not empirical, grounds. In the latter case it would come later than the systems, whereas it has in fact first produced the systematic character of our cognition of nature. Nor do these laws conceal intentions to put them to the test as mere experiments.

B 689 Certainly this coherence, wherever it holds, provides a powerful ground for regarding the hypothetically devised unity as well founded, and the laws are thus useful in this respect as well. But they clearly show that they judge economy of fundamental causes, multiplicity of effects, and the resulting kinship among nature’s members to be in themselves rational and appropriate to nature. These principles thus carry their recommendation directly, not merely as expedients of method.

It is easy to see, however, that this continuity of forms is a mere idea for which no congruent object can be exhibited in experience. This is not only because species in nature are actually divided and must therefore in themselves constitute a quantum discretum, and, if the graduated progression in their kinship were continuous, would also have to contain a true infinity of intermediate members lying between two given species, which is impossible. It is also because we can make no determinate empirical use of this law: it indicates not the slightest mark of affinity by which, or how far, we should seek the graduated sequence of their differences. It gives only a general indication that we are to seek it.

B 690 If we rearrange the principles just cited to bring them into line with their use in experience, the principles of systematic unity would stand roughly as follows: manifoldness, kinship, and unity, each taken as an idea at the highest degree of its completeness. Reason presupposes the understanding’s cognitions, which are applied in the first instance to experience, and seeks their unity according to ideas that extend far beyond experience’s reach. The kinship of the manifold under a principle of unity, without prejudice to its differences, concerns not only things but still more their mere properties and forces. Thus, for example, if an experience (not yet fully corrected) presents the planets’ courses as circular, and we find differences, we suspect them in that which can, according to a constant law and through all infinitely many intermediate degrees, transform a circle into one of these deviating orbits. That is, the planets’ motions that are not circular will perhaps approximate the circle’s properties more or less closely, and fall upon the ellipse. Comets show still greater differences in their paths, since (as far as observation reaches) they do not even return in a circuit. Yet we conjecture a parabolic course, which is nevertheless related to the ellipse and cannot in any of our observations be distinguished from it when the ellipse’s major axis is very greatly extended.

B 691 Guided by these principles, we thus arrive at unity of genus among these orbits in respect of their shape, and thereby proceed to the unity of the cause of all the laws of their motion (gravitation). From there we extend our conquests, seeking to explain all varieties and apparent deviations from those rules by the same principle. Finally, we even add more than experience can ever confirm: following the rules of kinship, we conceive even hyperbolic cometary orbits in which these bodies leave our solar system altogether and, passing from sun to sun, unite in their course the remoter parts of a cosmic system unbounded to us and connected by one and the same moving force.

What is remarkable about these principles, and alone concerns us here, is this: they appear to be transcendental. Although they contain merely ideas for the empirical use of reason to follow—ideas which that use can follow only asymptotically, as it were, by approaching without ever reaching them—they nevertheless have objective though indeterminate validity as synthetic propositions a priori. They serve as rules of possible experience and are indeed successfully employed as heuristic principles in working on it.

B 692 Yet no transcendental deduction of them can be accomplished, which, as proved above, is always impossible with regard to ideas.

In the Transcendental Analytic we distinguished, among the principles of the understanding, the dynamical principles, as merely regulative principles of intuition, from the mathematical principles, which are constitutive with regard to intuition. Despite this, those dynamical laws are certainly constitutive with regard to experience, since they make possible a priori the concepts without which no experience occurs. Principles of pure reason, by contrast, cannot be constitutive even with regard to empirical concepts, because no corresponding schema of sensibility can be given for them, and they can therefore have no object in konkreto. If I now abandon such an empirical use of them as constitutive principles, how can I nevertheless secure for them a regulative use and, with it, some objective validity? And what meaning can that use have?

The understanding is an object for reason just as sensibility is for the understanding. Making the unity of all possible empirical acts of the understanding systematic is reason’s task, just as the understanding connects the manifold of appearances through concepts and brings it under empirical laws. But without schemata of sensibility, acts of the understanding are indeterminate.

B 693 Likewise, the unity of reason is in itself indeterminate both as to the conditions under which the understanding should combine its concepts systematically and as to how far it should do so. Yet although no schema can be found in intuition for the thoroughgoing systematic unity of all concepts of the understanding, an analogue of such a schema can and must be given. This is the idea of the maximum of division and unification of the understanding’s cognition in one principle. For the greatest and absolutely complete can be thought determinately, since all restrictive conditions, which yield an indeterminate manifold, are omitted. The idea of reason is thus an analogue of a schema of sensibility, but with this difference: applying concepts of the understanding to reason’s schema does not yield cognition of the object itself (as applying the categories to their sensible schemata does). It yields only a rule or principle of the systematic unity of every use of the understanding. Now every principle that establishes thoroughgoing unity a priori for the understanding’s use also holds, although only indirectly, of the object of experience. The principles of pure reason will therefore also have objective reality with regard to this object, not to determine anything about the objects, but only to indicate the procedure.

B 694 By this procedure the understanding’s empirical and determinate use in experience can become thoroughly consistent with itself, through being brought into connection with the principle of thoroughgoing unity as far as possible, and derived from it.

I call all subjective principles drawn not from the constitution of the object but from reason’s interest in a certain possible perfection of cognition of that object maxims of reason. Thus there are maxims of speculative reason resting solely on its speculative interest, although they may seem to be objective principles.

If merely regulative principles are regarded as constitutive, they can conflict as objective principles. But if they are regarded merely as maxims, there is no true conflict, only a difference in reason’s interests that causes a divergence in ways of thinking. In fact, reason has only one single interest, and the conflict of its maxims is merely a difference and reciprocal limitation among methods of satisfying that interest.

In this way, one reasoner is more strongly moved by the interest in manifoldness (according to the principle of specification), and another by the interest in unity (according to the principle of aggregation).

B 695 Each believes his judgment arises from insight into the object, yet bases it solely on a greater or lesser attachment to one of two principles, neither of which rests on objective grounds but only on reason’s interest; they might therefore be better called maxims than principles. When I see discerning men disputing about the characteristics of human beings, animals, plants, or even bodies in the mineral kingdom—for example, when some assume particular national characters grounded in descent, or definite and hereditary differences among families, races, and so on, while others insist that nature has provided entirely identical predispositions in this respect and that every difference rests only on external contingencies—I need only consider the nature of the object to understand that it lies far too deeply concealed from both parties for them to speak from insight into its nature. It is nothing other than reason’s twofold interest, one side taking one interest to heart, the other the other, or even merely affecting to do so. Thus it is the difference between the maxims of nature’s manifoldness and nature’s unity. These can very well be united; but so long as they are taken for objective insights, they occasion not only conflict but also obstacles that long hold up truth.

B 696 So it remains until a means is found to reconcile the conflicting interest and to satisfy reason on this point.

The same holds of the assertion or denial of the much-discussed law of a continuous scale of creatures, put into circulation by Leibniz and splendidly embellished by Bonnet. This is nothing but obedience to the principle of affinity resting on reason’s interest; for observation and insight into nature’s arrangement could never have supplied it as an objective assertion. The rungs of such a ladder, as experience can disclose them to us, stand much too far apart, and our supposedly small differences are commonly such wide chasms in nature itself that nothing at all can be built on such observations (especially given a great manifold of things, where it must always be easy to find certain similarities and approximations) as evidence of nature’s intentions. By contrast, the method of seeking order in nature according to such a principle, and the maxim of regarding such order as grounded in nature in general, although leaving indeterminate where or how far, is certainly a legitimate and excellent regulative principle of reason. As such, however, it goes far beyond anything experience or observation could match, yet determines nothing, but merely traces out for experience the path toward systematic unity.

B 697 On the Final Aim of the Natural Dialectic of Human Reason

The ideas of pure reason can never be dialectical in themselves. Their misuse alone must produce the deceptive illusion that arises for us from them. For they are assigned to us by the nature of our reason, and this highest court of all the rights and claims of our speculation cannot possibly itself contain original deceptions and delusions. Presumably, then, they have a sound and purposive vocation in the natural predisposition of our reason. But the rabble of reasoners cries out, as usual, about absurdity and contradictions and reviles the government whose innermost plans it cannot penetrate, although it ought to owe to that government’s beneficent influence its own preservation and even the cultivation that enables it to criticize and condemn it.

We cannot use a concept a priori with any security without having accomplished its transcendental deduction. The ideas of pure reason admittedly permit no deduction of the kind possible for the categories. Suppose, however, that they are to have any objective validity at all, even if only indeterminate, and are not to represent merely empty thought-entities (entia rationis ratiocinantis).

B 698 Then a deduction of them must certainly be possible, even if it differs greatly from the one that can be undertaken with the categories. This is the completion of pure reason’s critical task, and we shall now undertake it.

There is a great difference between something being given to my reason as an object without qualification and being given merely as an object in the idea. In the first case, my concepts aim to determine the object. In the second, it is really only a schema to which no object is directly assigned, not even hypothetically; it serves only to represent other objects to us indirectly in their systematic unity, through their relation to this idea. Thus I say that the concept of a highest intelligence is a mere idea. That is, its objective reality is not to consist in its relating directly to an object (for in that sense we could not justify its objective validity). Rather, it is only a schema of the concept of a thing in general, ordered according to the conditions of the greatest unity of reason. It serves only to maintain the greatest systematic unity in our reason’s empirical use, by deriving the object of experience, as it were, from the imagined object of this idea as its ground or cause. We then say, for example, that things in the world must be regarded as if they owed their existence to a highest intelligence.

B 699 In this way the idea is properly only a heuristic, not an ostensive, concept. It indicates not how an object is constituted but how, under its guidance, we should seek the constitution and connection of objects of experience in general. Now suppose we can show that, although the three kinds of transcendental ideas (psychological, cosmological, and theological) are not directly related to any corresponding object and its determination, all rules of reason’s empirical use nevertheless lead to systematic unity on the presupposition of such an object in the idea, always extending experiential cognition and never being able to contradict it. It is then a necessary maxim of reason to proceed according to such ideas. This is the transcendental deduction of all ideas of speculative reason: not as constitutive principles for extending our cognition to more objects than experience can give, but as regulative principles of the systematic unity of the manifold of empirical cognition in general. Through them this cognition is cultivated and corrected within its own boundaries more than could be accomplished without such ideas through the mere use of the understanding’s principles.

B 700 Let me make this clearer. Following the ideas just mentioned as principles, we shall first (in psychology) connect all the appearances, actions, and receptivity of our mind along the guiding thread of inner experience as if the mind were a simple substance that, with personal identity, exists enduringly (at least during life), while its states, to which those of the body belong only as external conditions, change continuously. Secondly (in cosmology), we must pursue the conditions of both inner and outer natural appearances in an inquiry that can nowhere be completed, as if this inquiry were in itself infinite and without a first or highest member. We do not thereby deny their merely intelligible first grounds outside all appearances, but we must never bring these into the context of explanations of nature, since we do not know them at all. Finally and thirdly (with regard to theology), we must consider everything that can possibly belong to the context of possible experience as if this experience constituted an absolute unity, yet one that is thoroughly dependent and always still conditioned within the sensible world. At the same time, we must regard the sum of all appearances (the sensible world itself) as if it had outside its compass one single highest and all-sufficient ground: a reason that is, as it were, self-subsistent, original, and creative.

B 701 In relation to it, we direct every empirical use of our reason in its greatest extension as if the objects themselves had sprung from that archetype of all reason. This means: not to derive the soul’s inner appearances from a simple thinking substance, but to derive them from one another according to the idea of a simple being; not to derive the world’s order and its systematic unity from a highest intelligence, but to take from the idea of a supremely wise cause the rule according to which reason may best be used, to its own satisfaction, in connecting causes and effects in the world.

Now there is not the slightest thing to prevent us from also assuming these ideas objectively and hypostatically, except in the case of the cosmological idea, where reason encounters an antinomy if it tries to accomplish this (the psychological and theological ideas contain nothing of the kind). For there is no contradiction in them. How, then, could anyone dispute their objective reality, since he knows as little of their possibility in denying it as we do in affirming it? Yet the absence of a positive obstacle is not enough to warrant an assumption. We cannot be permitted to introduce thought-entities that surpass all our concepts, though they contradict none of them, as actual and determinate objects merely on the credit of speculative reason, which is eager to complete its task.

B 702 They are therefore not to be assumed in themselves. Their reality is to count only as that of a schema of the regulative principle of the systematic unity of all cognition of nature. They are thus to be laid down only as analogues of actual things, not as such things in themselves. We remove from the object of the idea the conditions that restrict our concept of the understanding, but which alone also make it possible for us to have a determinate concept of anything. We now think of a something of which we have no concept at all as to what it is in itself, but whose relation to the sum of appearances we nevertheless conceive as analogous to the relation appearances have to one another.

When we assume such ideal beings, then, we do not really extend our cognition beyond the objects of possible experience. We extend only the empirical unity of these objects through the systematic unity for which the idea supplies the schema; the idea thus holds not as a constitutive but merely as a regulative principle. For when we posit a thing corresponding to the idea—a something, or an actual being—we do not thereby mean to extend our cognition of things through transcendent concepts. For this being is laid down only in the idea, not in itself.

B 703 It is thus laid down only to express the systematic unity that is to guide reason’s empirical use. Nothing is thereby settled about the ground of this unity or about the inner character of such a being on which it rests as its cause.

Thus the transcendental and only determinate concept of God that merely speculative reason gives us is deistic in the strictest sense. That is, reason does not even supply the objective validity of such a concept, but only the idea of a something on which all empirical reality grounds its highest and necessary unity. We can think of this something only by analogy with an actual substance that is the cause of all things according to laws of reason—if indeed we undertake to think of it at all as a particular object, rather than remain content with the mere idea of reason’s regulative principle and set aside the completion of all conditions of thinking as extravagant for the human understanding. Yet this latter course cannot be reconciled with the aim of a perfect systematic unity in our cognition, to which reason at least sets no limits.

Hence it happens that, when I assume a divine being, I have not the slightest concept either of the inner possibility of its supreme perfection or of the necessity of its existence.

B 704 Nevertheless, I can then satisfy all other questions concerning the contingent and provide reason with the most perfect satisfaction regarding the greatest unity to be sought in its empirical use, though not regarding this presupposition itself. This proves that it is reason’s speculative interest, not its insight, that entitles it to proceed from a point so far above its sphere, in order to view its objects from there within a complete whole.

Here, given one and the same presupposition, a difference in ways of thinking emerges that is rather subtle, but nevertheless very important in transcendental philosophy. I can have sufficient ground for assuming something relatively (suppositio relativa), without being entitled to assume it absolutely (suppositio absoluta). This distinction applies when we are concerned only with a regulative principle whose necessity in itself we cognize, but not its source, and for which we assume a highest ground merely in order to think the principle’s universality more determinately: for example, when I think as existing a being corresponding to a mere, specifically transcendental, idea. For in that case I can never assume the existence of this thing in itself.

B 705 That is because no concepts through which I can think any object determinately reach it, and because the conditions of my concepts’ objective validity are excluded by the idea itself. The concepts of reality, substance, causality, and even necessity in existence have no object-determining meaning apart from the use in which they make empirical cognition of an object possible. They can therefore be used to explain the possibility of things in the sensible world, but not the possibility of a world-whole itself, since this ground of explanation would have to lie outside the world and thus be no object of a possible experience. Nevertheless, I can assume such an incomprehensible being, the object of a mere idea, in relation to the sensible world, though not in itself. For suppose the greatest possible empirical use of my reason rests on an idea (of systematically complete unity, about which I shall shortly speak more determinately) that can never itself be adequately exhibited in experience, although it is unavoidably necessary for bringing empirical unity as close as possible to the highest degree. I shall then be not only entitled but compelled to realize this idea, that is, to posit an actual object for it, but only as a something in general that I do not know at all in itself. Only as a ground of that systematic unity, and in relation to it, do I give it properties analogous to concepts of the understanding in their empirical use.

B 706 Thus, by analogy with the realities in the world of substances, of causality, and of necessity, I shall think of a being that possesses all these in the highest perfection. Since this idea rests solely on my reason, I shall be able to think of this being as self-subsistent reason that, through ideas of the greatest harmony and unity, is the cause of the world-whole. I thereby omit all conditions restricting the idea, solely to make possible, under the protection of such an original ground, the systematic unity of the manifold in the world-whole and, through it, the greatest possible empirical use of reason. I regard all combinations as if they were arrangements of a highest reason, of which ours is a faint copy. I then think of this highest being entirely through concepts that properly have their application only in the sensible world. But since I have this transcendental presupposition only for a relative use, namely to provide the substratum of the greatest possible unity of experience, I may quite legitimately think of a being that I distinguish from the world through properties belonging solely to the sensible world. For I by no means demand, nor am I entitled to demand, to cognize this object of my idea according to what it may be in itself.

B 707 For I have no concepts for that, and even the concepts of reality, substance, causality, and indeed necessity in existence lose all meaning and become empty titles for concepts, without any content, if I venture with them beyond the field of the senses. I think only the relation of a being wholly unknown to me in itself to the greatest systematic unity of the world-whole, solely to make it the schema of the regulative principle of the greatest possible empirical use of my reason.

If we now turn our gaze to the transcendental object of our idea, we see that we cannot presuppose its actuality in itself according to the concepts of reality, substance, causality, and so on, because these concepts have not the slightest application to something wholly distinct from the sensible world. Thus reason’s supposition of a highest being as supreme cause is conceived merely relatively, for the sake of the systematic unity of the sensible world, and a mere something in the idea, of which we have no concept as to what it is in itself. This also explains why, in relation to what is given to the senses as existing, we need the idea of an original being necessary in itself, yet can never have the slightest concept of this being and its absolute necessity.

We can now clearly present the result of the entire Transcendental Dialectic.

B 708 We can also determine precisely the final aim of the ideas of pure reason, which become dialectical only through misunderstanding and carelessness. Pure reason is in fact occupied with nothing but itself, nor can it have any other task, because it is given not objects to bring to the unity of the concept of experience, but cognitions of the understanding to bring to the unity of reason’s concept, that is, to coherence in a principle. The unity of reason is the unity of a system. This systematic unity serves reason not objectively as a principle for extending it over objects, but subjectively as a maxim for extending it over all possible empirical cognition of objects. Nevertheless, the systematic coherence that reason can give the empirical use of the understanding not only furthers its extension but at the same time attests its correctness. The principle of such systematic unity is also objective, but in an indeterminate way (principium vagum), not as a constitutive principle for determining something concerning its direct object, but as a merely regulative principle and maxim. By opening new paths unknown to the understanding, it furthers and secures the empirical use of reason to infinity (the indeterminate), without ever contradicting the laws of empirical use in the slightest.

B 709 But reason cannot think this systematic unity without at the same time giving its idea an object, which, however, cannot be given through any experience; for experience never supplies an example of perfect systematic unity. This being of reason (ens rationis ratiocinatae) is admittedly a mere idea. It is therefore not assumed absolutely and in itself as something actual, but is laid down only problematically (since we cannot reach it through any concepts of the understanding). Its purpose is to let us regard every connection of things in the sensible world as if they had their ground in this being of reason, solely in order to ground on it the systematic unity that is indispensable to reason and can further the understanding’s empirical cognition in every way, yet can never hinder it.

We immediately misunderstand this idea if we take it for the assertion, or even merely the presupposition, of an actual thing to which we intend to ascribe the ground of the world’s systematic constitution. Rather, we leave entirely unsettled what character this ground, which eludes our concepts, has in itself, and set up only an idea as a standpoint from which alone we can extend that unity so essential to reason and so beneficial to the understanding.

B 710 In a word, this transcendental thing is merely the schema of the regulative principle through which reason, so far as it can, extends systematic unity over all experience. The first object of such an idea is myself, considered merely as a thinking nature (soul). If I want to investigate the properties with which a thinking being exists in itself, I must consult experience; even among all the categories, I can apply none to this object except insofar as its schema is given in sensible intuition. But in this way I never arrive at a systematic unity of all appearances of inner sense. Instead of the experiential concept (of what the soul actually is), which cannot carry us far, reason therefore takes the concept of the empirical unity of all thinking. By thinking this unity as unconditioned and original, it makes that concept into a concept of reason (idea) of a simple substance that is in itself unchangeable (personally identical) and in community with other actual things outside it: in a word, of a simple self-subsistent intelligence. Here, however, reason has in view nothing but principles of systematic unity in explaining the soul’s appearances: to regard all determinations as in one single subject, all forces as far as possible as derived from one fundamental power, and all change as belonging to the states of one and the same enduring being.

B 711 Its principles likewise require us to represent all appearances in space as entirely distinct from acts of thinking. That simplicity of substance and so on was meant to be only the schema for this regulative principle; it is not presupposed as the actual ground of the soul’s properties. For these may rest on entirely different grounds of which we know nothing, just as we could not properly cognize the soul in itself through these assumed predicates even if we wanted to grant them to it without qualification, since they constitute a mere idea that cannot be represented in concreto at all. Nothing but benefit can arise from such a psychological idea, provided we take care to allow it to count as no more than a mere idea, that is, merely in relation to reason’s systematic use concerning the appearances of our soul. For then no empirical laws of bodily appearances, which are of an entirely different kind, intrude into explanations of what belongs solely to inner sense. No flimsy hypotheses about the generation, destruction, and palingenesis of souls, and so on, are admitted. The consideration of this object of inner sense is thus kept entirely pure and unmixed with heterogeneous properties. Moreover, reason’s inquiry is directed toward reducing the grounds of explanation in this subject, as far as possible, to a single principle.

B 712 All this is best accomplished—indeed, can be accomplished only—through such a schema, as if it were an actual being. The psychological idea can also mean nothing but the schema of a regulative concept. For if I were even to ask whether the soul is not of spiritual nature in itself, this question would have no meaning at all. For through such a concept I remove not merely bodily nature but all nature whatsoever: all predicates of any possible experience, and thus all conditions for thinking an object for such a concept. Yet this alone makes it possible to say that the concept has a meaning.

The second regulative idea of merely speculative reason is the concept of the world in general. For nature is properly the only given object with regard to which reason needs regulative principles. This nature is twofold: either thinking or corporeal nature. But to think the latter according to its inner possibility, that is, to determine the application of the categories to it, we need no idea, that is, no representation exceeding experience. Nor is any possible with regard to it, because here we are guided solely by sensible intuition, unlike in the psychological fundamental concept (the I), which contains a priori a certain form of thinking, namely its unity.

B 713 Thus nothing remains for pure reason but nature in general and the completeness of its conditions according to some principle. The absolute totality of the series of these conditions in the derivation of their members is an idea that can never be fully realized in reason’s empirical use, but nevertheless serves as a rule for how we should proceed with regard to them. In explaining given appearances (by going backward or ascending) we proceed as if the series were infinite in itself, that is, in indefinitum. But where reason itself is regarded as a determining cause (in freedom), and thus in the case of practical principles, we proceed as if we had before us not an object of the senses but one of the pure understanding. Here conditions can be posited no longer within the series of appearances but outside it, and the series of states can be regarded as if it began absolutely (through an intelligible cause). All this proves that cosmological ideas are nothing but regulative principles, far from positing, as though constitutively, an actual totality of such series. The rest may be sought in its proper place under the Antinomy of Pure Reason.

The third idea of pure reason, which contains a merely relative supposition of a being as the single and all-sufficient cause of all cosmological series, is reason’s concept of God.

B 714 We have not the slightest ground for assuming the object of this idea absolutely (supposing it in itself). For what could move us, or even entitle us, to believe in or assert a being of the highest perfection, absolutely necessary by its nature, merely from its concept in itself, were it not for the world, in relation to which alone this supposition can be necessary? Here it becomes clear that the idea of this being, like all speculative ideas, means nothing more than that reason commands us to consider every connection in the world according to principles of systematic unity, hence as if they had all sprung from one single all-encompassing being as their highest and all-sufficient cause. It is clear from this that reason can intend nothing here but its own formal rule in extending its empirical use, never an extension beyond all boundaries of empirical use. Consequently, this idea conceals no constitutive principle for reason’s use directed toward possible experience.

This highest formal unity, which rests solely on concepts of reason, is the purposive unity of things. Reason’s speculative interest makes it necessary to regard every arrangement in the world as if it had sprung from the intention of a supreme reason.

B 715 Such a principle opens entirely new prospects for our reason as applied to the field of experience: to connect things in the world according to teleological laws and thereby arrive at their greatest systematic unity. The presupposition of a highest intelligence as the sole cause of the world-whole, though admittedly only in the idea, can therefore always benefit reason and never harm it. For if, with regard to the earth’s shape (round, though somewhat flattened)*, or mountains and seas, and so on, we presuppose wise intentions of an author throughout, we can make a host of discoveries along this path. Provided we hold to this presupposition as a merely regulative principle, even error cannot harm us. For at most, it can only follow that, where we expected a teleological connection (nexus finalis), we encounter a merely mechanical or physical one (nexus effectivus).

B 716 In such a case we merely miss one additional unity, but do not spoil the unity of reason in its empirical use. Yet even this setback cannot touch the law itself in its general teleological intent. For although an anatomist can be convicted of error if he relates some member of an animal body to an end that can clearly be shown not to result from it, it is nevertheless entirely impossible to prove in any case that a natural arrangement, whatever it may be, has no end at all. Thus physiology (of physicians) also extends its very limited empirical knowledge of the ends of the structure of an organic body through a principle supplied by pure reason alone, to the point where we quite boldly assume, with the agreement of all discerning people, that everything in an animal has its use and good purpose. If this presupposition were meant to be constitutive, it would go much further than observation to date can justify. From this we can see that it is nothing but a regulative principle of reason for arriving at the highest systematic unity through the idea of the purposive causality of the highest cause of the world, as if this cause, as a highest intelligence, were the cause of everything according to the wisest intention.

* The advantage afforded by a spherical earth is well enough known. But few know that its flattening into a spheroid alone prevents protrusions of the mainland, or even smaller mountains perhaps thrown up by earthquakes, from continuously and, within no very long time, considerably shifting the earth’s axis, were the earth’s bulge at the equator not so mighty a mountain that the impetus of any other mountain can never appreciably displace it from its position in relation to the axis. Yet we have no hesitation in explaining this wise arrangement from the equilibrium of the earth’s formerly fluid mass.

B 717 But if we abandon this restriction of the idea to merely regulative use, reason is misled in many ways. It then leaves the ground of experience, which must nevertheless contain the landmarks of its course, and ventures above it into the incomprehensible and inscrutable. At that height it necessarily grows dizzy, because from this standpoint it sees itself entirely cut off from every use consistent with experience.

The first error arising from using the idea of a highest being not merely regulatively but (which is contrary to the nature of an idea) constitutively, is lazy reason (ignava ratio)*. We can give this name to every principle that causes us to regard our investigation of nature, at any point whatsoever, as absolutely complete.

B 718 Reason then retires to rest, as if it had fully accomplished its task. Thus even the psychological idea, when used as a constitutive principle for explaining the appearances of our soul, and afterward even for extending our cognition of this subject beyond all experience (its state after death), admittedly makes things very comfortable for reason, but also completely spoils and destroys every natural use of it guided by experience. Thus the dogmatic spiritualist explains the unity of the person, which remains unchanged through every change of states, by the unity of the thinking substance that he believes he immediately perceives in the I. He explains our interest in things that are to occur only after our death by our consciousness of the immaterial nature of our thinking subject, and so on. He thereby dispenses with all natural inquiry into the cause of these inner appearances through physical grounds of explanation. As though by the decree of a transcendent reason, he bypasses the immanent sources of cognition in experience for the sake of his own comfort, but at the cost of all insight. This harmful consequence is still more evident in the dogmatism of our idea of a highest intelligence and the theological system of nature (physicotheology) falsely grounded on it.

B 719 For here all the ends manifest in nature—often only made into ends by us—serve to make the investigation of causes very comfortable for us. Instead of seeking these causes in the universal laws of the mechanism of matter, we appeal directly to the inscrutable decree of supreme wisdom and regard reason’s labor as complete once we dispense with its use. Yet reason finds a guiding thread nowhere except where nature’s order and the series of changes, according to their inner and more general laws, supply one. This error can be avoided if we consider from the standpoint of ends not merely some parts of nature, such as the distribution of the mainland, its structure, and the constitution and location of mountains, or even merely organization in the plant and animal kingdoms, but make this systematic unity of nature, in relation to the idea of a highest intelligence, entirely universal. For then we take as our basis a purposiveness according to universal laws of nature, from which no particular arrangement is excepted, but is only marked out more or less recognizably for us. We have a regulative principle of the systematic unity of a teleological connection, which we must not determine in advance.

B 720 Rather, in expectation of it, we may only pursue the physical-mechanical connection according to universal laws. Only in this way can the principle of purposive unity always extend reason’s use with regard to experience without impairing it in any case.

* The ancient dialecticians gave this name to a fallacy that ran as follows: if it is your fate to recover from this illness, it will happen whether you consult a physician or not. Cicero says this kind of inference received its name because, if we followed it, no use of reason in life would remain. This is why I give the sophistical argument of pure reason the same name.

The second error arising from a misinterpretation of this principle of systematic unity is that of inverted reason (perversa ratio, ysteron proteron rationis). The idea of systematic unity was meant to serve only as a regulative principle for seeking that unity in the combination of things according to universal laws of nature; insofar as something of it can be found empirically, we should believe that we have come that much closer to completeness in its use, although we shall admittedly never attain it. Instead, people reverse the matter. They begin by laying down as hypostatic the actuality of a principle of purposive unity, determine the concept of such a highest intelligence anthropomorphically because it is in itself entirely inscrutable, and then impose ends on nature forcibly and dictatorially instead of seeking them, as they rightly should, along the path of physical inquiry. Thus teleology, which should serve merely to supplement nature’s unity according to universal laws, now acts rather to abolish it.

B 721 Reason also defeats its own purpose of subsequently proving from nature the existence of such an intelligent highest cause. For unless we can presuppose the highest purposiveness in nature a priori, that is, as belonging to its essence, how could we be directed to seek it and, ascending its scale, to approach an author’s highest perfection as an absolutely necessary and thus a priori cognizable perfection? The regulative principle requires us to presuppose systematic unity as a unity of nature that is not merely cognized empirically but presupposed a priori, though still indeterminately—to presuppose it absolutely, therefore, as following from the essence of things. But if I first lay down a highest ordering being, the unity of nature is in fact abolished. For it is then entirely foreign and contingent to the nature of things and cannot be cognized from its universal laws. A vicious circle in proof thus arises, since we presuppose what was actually to be proved.

Taking the regulative principle of nature’s systematic unity for a constitutive one, and hypostatically presupposing as a cause what is laid down only in the idea as the ground of reason’s harmonious use, merely confuses reason.

B 722 The investigation of nature follows its course solely along the chain of natural causes according to their universal laws. It does so according to the idea of an author, but not in order to derive from him the purposiveness it pursues everywhere. Rather, it seeks to cognize his existence from this purposiveness, which is sought in the essences of natural things and, if possible, in the essences of all things whatsoever, and thus as absolutely necessary. Whether this last aim succeeds or not, the idea always remains correct, as does its use when restricted to the conditions of a merely regulative principle.

Complete purposive unity is perfection (considered absolutely). If we do not find it in the essence of the things that constitute the whole object of experience, that is, of all our objectively valid cognition, and thus in universal and necessary laws of nature, how can we infer directly from it the idea of the highest and absolutely necessary perfection of an original being that is the source of all causality? The greatest systematic and consequently purposive unity is the school, and indeed the very foundation of the possibility, of the greatest use of human reason.

B 723 Its idea is therefore inseparably bound up with the essence of our reason. This same idea is thus legislative for us, and so it is very natural to assume a corresponding legislative reason (intellectus archetypus) from which all systematic unity of nature, as the object of our reason, is to be derived.

On the occasion of the Antinomy of Pure Reason, we said that all questions raised by pure reason must be absolutely answerable, and that an excuse based on the limits of our cognition, as unavoidable as it is legitimate in many questions about nature, cannot be permitted here, because these questions are posed to us not by the nature of things but only by the nature of reason, and solely concerning its inner constitution. We can now confirm this seemingly bold assertion with regard to the two questions in which pure reason takes its greatest interest, and thereby bring our consideration of its dialectic to full completion.

Suppose, then, that we ask first (with a view to a transcendental theology)*.

B 724 Is there something distinct from the world that contains the ground of the world’s order and its coherence according to universal laws? The answer is: undoubtedly. For the world is a sum of appearances, and there must therefore be some transcendental ground of them, that is, a ground thinkable only by the pure understanding. If the second question is whether this being is substance, possesses the greatest reality, is necessary, and so on, I answer that this question has no meaning at all. For all the categories through which I try to form a concept of such an object have only an empirical use, and have no meaning unless applied to objects of possible experience, that is, to the sensible world. Outside this field they are merely titles for concepts, which one can concede but through which one can understand nothing. Finally, if the third question is whether we may at least think of this being distinct from the world by analogy with objects of experience, the answer is: certainly.

B 725 But only as an object in the idea, not in reality; namely, only insofar as it is a substratum, unknown to us, of the systematic unity, order, and purposiveness of the world’s arrangement, which reason must make a regulative principle of its investigation of nature. More than this, we can freely and blamelessly permit in this idea certain anthropomorphisms that further the regulative principle in question. For it always remains only an idea, related not directly to a being distinct from the world but to the regulative principle of the world’s systematic unity, and only through a schema of that unity: a highest intelligence that is its author according to wise intentions. We were not thereby meant to think what this original ground of the world’s unity is in itself, but how we should use it—or rather its idea—in relation to reason’s systematic use concerning things in the world.

* What I have already said about the psychological idea and its true vocation as a principle for reason’s merely regulative use saves me the lengthy task of separately discussing the transcendental illusion through which that systematic unity of the whole manifold of inner sense is represented hypostatically. The procedure here is very similar to that followed by the critique with regard to the theological ideal.

But in this way can we nevertheless (someone will continue to ask) assume a single wise and almighty author of the world? Without any doubt; and not only that, we must presuppose such an author. But do we not then extend our cognition beyond the field of possible experience? Not at all.

B 726 For we have presupposed only a something of which we have no concept whatever as to what it is in itself (a merely transcendental object). But in relation to the systematic and purposive order of the world’s structure, which we must presuppose when we study nature, we have thought of this unknown being only by analogy with an intelligence (an empirical concept). That is, with regard to the ends and the perfection grounded on them, we have endowed it with precisely those properties that, according to the conditions of our reason, can contain the ground of such systematic unity. This idea is thus entirely well founded in relation to our reason’s use in the world. But if we wanted to give it objective validity without qualification, we would forget that what we think is merely a being in the idea. By then starting from a ground that cannot be determined at all through consideration of the world, we would make ourselves unable to apply this principle appropriately to reason’s empirical use.

But (someone will ask further) in this way I can nevertheless make use of the concept and presupposition of a highest being in rational consideration of the world? Yes; this was indeed the very purpose for which reason laid down this idea.

B 727 But may I now regard arrangements resembling ends as intentions, by deriving them from the divine will, though through particular predispositions established in the world for that purpose? Yes, you can do this too, but in such a way that it must be all the same to you whether someone says that divine wisdom has ordered everything thus for its highest ends, or that the idea of the highest wisdom is a regulative guide in investigating nature and a principle of its systematic and purposive unity according to universal laws of nature, even where we do not perceive that unity. That is, where you do perceive it, it must be entirely the same to you to say, “God has wisely willed it so,” or, “Nature has wisely ordered it so.” For the greatest systematic and purposive unity, which your reason required you to lay down as a regulative principle for every investigation of nature, was precisely what entitled you to lay down the idea of a highest intelligence as a schema of that regulative principle. As much purposiveness as you now find in the world according to it, so much confirmation do you have of your idea’s legitimacy. But since the principle in question intended nothing but the search for nature’s necessary and greatest possible unity, we shall admittedly owe this unity, insofar as we attain it, to the idea of a highest being.

B 728 Yet we cannot, without contradicting ourselves, bypass the universal laws of nature, for whose sake alone the idea was laid down, in order to regard nature’s purposiveness as contingent and hyperphysical in origin. For we were not entitled to assume a being above nature with the properties in question, but only to lay down its idea, in order to regard appearances as systematically connected with one another by analogy with a causal determination.

For this very reason, we are also entitled to think the cause of the world in the idea not only according to a subtler anthropomorphism (without which nothing at all could be thought of it), namely as a being that has understanding, pleasure and displeasure, and likewise a desire and will appropriate to these, and so on, but also to ascribe to it infinite perfection. This far exceeds the perfection we could be entitled to assume through empirical knowledge of the world’s order. For the regulative law of systematic unity requires us to study nature as if systematic and purposive unity were to be found everywhere to infinity, amid the greatest possible manifold. For although we shall discern or attain only a little of this perfection of the world, it nevertheless belongs to our reason’s legislation to seek and suspect it everywhere. Considering nature according to this principle must always be advantageous to us and can never be harmful.

B 729 But under this representation of the underlying idea of a highest author, it is also clear that I lay down not the existence and knowledge of such a being, but only its idea. Thus I actually derive nothing from this being, but only from its idea, that is, from the nature of things in the world according to such an idea. A certain, though undeveloped, consciousness of the genuine use of this concept of our reason also seems to have prompted the modest and fair language of philosophers in every age. They speak of nature’s wisdom and providence and of divine wisdom as equivalent expressions; indeed, so long as merely speculative reason is at issue, they prefer the former expression, because it restrains the presumption of making a greater assertion than we are entitled to make and at the same time directs reason back to its own proper field, nature.

Thus pure reason, which at first seemed to promise us nothing less than an extension of knowledge beyond all boundaries of experience, contains, when rightly understood, nothing but regulative principles. These admittedly command a greater unity than the empirical use of the understanding can attain.

B 730 Yet precisely by moving the goal of its approximation so far away, they bring its agreement with itself through systematic unity to the highest degree. But if these principles are misunderstood and taken for constitutive principles of transcendent cognitions, they produce, through a brilliant but deceptive illusion, persuasion and imagined knowledge, and with these, endless contradictions and disputes.

* * *

Thus all human cognition begins with intuitions, proceeds from them to concepts, and ends with ideas. Although it has sources of cognition a priori for all three elements, which at first sight seem to disdain the boundaries of all experience, a completed critique nevertheless convinces us that reason, in its speculative use of these elements, can never get beyond the field of possible experience, and that the true vocation of this highest faculty of cognition is to use all methods and their principles only to pursue nature into its innermost depths according to every possible principle of unity, foremost among them the unity of ends, but never to fly beyond nature’s boundary, outside which there is nothing for us but empty space.

B 731 The critical investigation in the Transcendental Analytic of all propositions capable of extending our cognition beyond actual experience has admittedly convinced us sufficiently that they can never lead to anything more than a possible experience. If we were not distrustful even of the clearest abstract and universal doctrines, and if enticing and apparent prospects did not tempt us to throw off the constraint of the former, we could certainly have been spared the laborious examination of all the dialectical witnesses whom a transcendent reason calls to support its pretensions. For we already knew in advance, with complete certainty, that all their allegation, though perhaps honestly intended, must be utterly null, because it concerned information no human being can ever obtain. Yet talk will never end unless we discover the true cause of the illusion through which even the most rational person can be deceived. And the analysis of all our transcendent cognition into its elements (as a study of our inner nature) has no small value in itself and is even a duty for the philosopher. It was therefore necessary not only to trace this entire, though futile, labor of speculative reason in detail back to its first sources.

B 732 Since dialectical illusion is here not only deceptive in its judgment but also alluring in the interest we take in that judgment, and is always natural and will remain so for all time to come, it was also advisable to draw up, as it were, a detailed record of these proceedings and deposit it in the archive of human reason, to prevent future errors of a similar kind.

Transcendental Logic Second Division Transcendental Dialectic

Introduction

I. Transcendental Illusion

Earlier we described dialectic as a whole as a logic of illusion. That doesn’t mean it is a theory of probability. For probability is truth known from insufficient grounds. Such knowledge is deficient, but it isn’t deceptive. So probability need not be cut off from the analytic part of logic. Even less may appearance and illusion be taken as the same.

B 350 For truth and illusion are not in the object as it is intuited. They are in the judgment about the object as it is thought. It is therefore right to say that the senses never err. That’s not because they always judge correctly. It is because they don’t judge at all. So truth and error, and with them illusion as the lure toward error, exist only in judgment, that is, only in how the object stands to our understanding. A cognition that fully agrees with the laws of the understanding contains no error. A representation of the senses contains no error either, since it holds no judgment. No force of nature, though, can swerve from its own laws on its own. So neither the understanding alone (with no other cause acting on it) nor the senses alone would err. The understanding wouldn’t, because if it acts purely by its own laws, its effect (the judgment) has to agree with those laws. And agreement with the laws of the understanding is the formal side of all truth. In the senses there is no judgment, either true or false. Since we have no sources of cognition besides these two, it follows that error comes about only when sensibility influences the understanding without being noticed. That influence mixes the subjective grounds of judgment into the objective ones.

B 351 It pulls the objective ones away from their purpose*. Compare a moving body. Left to itself it would always keep to a straight line in one direction. But if another force acts on it at the same time in a different direction, its motion bends into a curve. To separate the understanding’s own action from the force that gets mixed in, we need to treat the erring judgment as a diagonal between two forces. The two forces pull the judgment in different directions, as if enclosing an angle. That compound effect then has to be split into its two simple parts: the effect of understanding and the effect of sensibility. For pure judgments a priori, transcendental reflection does this. As already noted, it assigns every representation its place in the power of cognition that suits it, and so it also separates the influence of the one power on the other.

* Sensibility, set beneath the understanding as the object that the understanding’s activity works on, is the source of real cognitions. The same sensibility is the ground of error, however, when it flows into the understanding’s own activity and moves it to judge.

We are not here to deal with empirical illusion (optical illusion, for example).

B 352 It appears when rules of the understanding that are otherwise correct are used empirically, and it misleads the power of judgment through the influence of imagination. Our concern is transcendental illusion alone. It affects principles that aren’t even meant to be used on experience; if they were, we would at least have a touchstone for checking them. Instead, in spite of every warning from critique, it carries us entirely past the empirical use of the categories. It keeps stringing us along with the deceptive prospect of widening pure understanding. Principles that are applied wholly inside the limits of possible experience we will call immanent. Principles that are meant to fly over those boundaries we will call transcendent. By transcendent principles I don’t mean the transcendental use or misuse of the categories. That is a plain fault of a power of judgment that critique hasn’t properly reined in, and that doesn’t take enough care about the boundary of the ground where pure understanding alone is allowed to play. I mean real principles that call on us to pull up all those boundary posts and claim a whole new ground, one that acknowledges no boundary line anywhere. So transcendental and transcendent are different things. The principles of pure understanding that we presented above are meant for empirical use only, and not for transcendental use.

B 353 Transcendental use means a use that goes beyond the boundary of experience. A principle that removes these limits, or even commands us to cross them, is called transcendent. If our critique can uncover the illusion in such presumptuous principles, then the principles of purely empirical use can be called immanent principles of pure understanding, as opposed to the transcendent ones.

Logical illusion is the illusion of fallacies. It consists in simply copying the form of reason, and it comes only from not paying attention to the logical rule. As soon as the rule is brought to bear on the case in front of us, the illusion vanishes completely. Transcendental illusion is different. It doesn’t stop even once it has been uncovered, and transcendental critique has shown plainly that it is empty. (Example: the illusion in the statement that, in time, the world must have a beginning.) The reason is this. Our reason (viewed subjectively, as a human power of cognition) contains basic rules and maxims for its own use, and they have the complete look of objective principles. So the subjective necessity of connecting our concepts in a certain way, which suits the understanding, gets taken for an objective necessity: a determination of things in themselves. There’s no escaping this illusion.

B 354 We can’t help seeing the sea as higher out in the middle than it is at the shore (we see the middle by higher rays of light). Or take something stronger still: even an astronomer can’t stop the moon from looking larger when it rises, although this illusion doesn’t deceive the astronomer.

So transcendental dialectic will settle for uncovering the illusion in transcendent judgments and preventing it from deceiving us. It can never make that illusion disappear, as logical illusion does, and stop it from being an illusion. For here we face a natural and unavoidable illusion, which rests on subjective principles itself and slips them in as objective ones. Logical dialectic, when it resolves fallacies, faces only a mistake in following principles, or an artificial illusion produced by imitating them. So there is a dialectic of pure reason that is natural and unavoidable. It’s not a dialectic that some bungler stumbles into for lack of knowledge, or that some sophist has devised to confuse sensible people. It is stuck to human reason beyond any hope of removal. Even after we have exposed its deceptive show, it won’t stop playing tricks on reason.

B 355 It will keep pushing reason again and again into momentary errors, which always have to be corrected.

II. Pure Reason as the Seat of Transcendental Illusion

A. Reason in General

All our cognition starts with the senses, moves on to the understanding, and finishes in reason. Nothing higher in us can work over the material of intuition and bring it under the highest unity of thinking. Now I must explain this highest power of cognition, and I’m somewhat at a loss. Reason, like the understanding, has a merely formal use, a logical one, in which it ignores all content of cognition. It also has a real use, because it holds within itself the source of certain concepts and principles, which it takes neither from the senses nor from the understanding. Logicians explained the first use long ago: it is the power of inferring mediately (as distinct from immediate inferences, consequentiis immediatis). That doesn’t yet make the second use clear, the one that produces concepts itself. Here reason is split into a logical power and a transcendental one.

B 356 So we must find a higher concept for this source of cognition, one that takes in both. We can expect, following the model of the understanding’s concepts, that the logical concept will also give us the key to the transcendental one, and that the table of the logical functions will give us the family tree of the concepts of reason.

The first part of our transcendental logic defined the understanding as the power of rules. Here we set reason apart from it by calling reason the power of principles.

The word “principle” is ambiguous. Usually it means just a cognition that can be put to work as a principle, even if in itself and by origin it isn’t a principium. Any general proposition can be the major premise of an inference of reason, even one drawn from experience by induction. That doesn’t make the proposition a principium. The mathematical axioms are even general cognitions a priori (for example, “only one straight line can lie between two points”). So we rightly call them principles, relative to the cases that can fall under them. Yet that doesn’t let me say I know this property of straight lines from principles, in general and in itself.

B 357 I cognize it only in pure intuition.

I would therefore call cognition from principles the cognition in which I know the particular in the general by means of concepts. Every inference of reason is thus a way of deriving a cognition from a principle. The major premise, you see, always supplies a concept, and that concept makes everything that is subsumed under its condition known from the concept, according to a principle. Any general cognition can serve as the major premise of an inference of reason, and the understanding supplies general propositions of this kind a priori. So these, too, can be called principles, in view of how they might be used.

Look at these principles of pure understanding in themselves, though, and at where they come from. They are nothing like cognitions from concepts. For they wouldn’t even be possible a priori if we didn’t call on pure intuition (in mathematics) or on the conditions of possible experience in general. You can’t infer, from the concept of anything at all that happens, that whatever happens has a cause. The principle instead shows how we can first form a determinate empirical concept of what happens.

The understanding therefore can’t provide synthetic cognitions from concepts.

B 358 These are what I properly call principles, pure and simple, while all general propositions can be called comparative principles.

People have long wished (and who knows, the wish may come true, however late) that someone would look for the principles behind civil laws, instead of their endless variety. For simplifying legislation, as the saying goes, may lie only there. But laws here are just limits on our freedom, which keep it within conditions where it agrees with itself throughout. They therefore concern something that is entirely our own work, and that we can cause through those very concepts. A different matter is how objects in themselves, the very nature of things, could stand under principles and be determined by concepts alone. To demand that is, if not impossible, at least very absurd. Whatever the truth of that (we still have to investigate it), this much is clear: cognition from principles (in itself) is quite different from mere cognition by the understanding. Cognition of the understanding can also go ahead of other cognitions, in the shape of a principle. Taken in itself, though, as far as it is synthetic, it doesn’t rest on mere thinking and holds nothing general according to concepts.

B 359 The understanding may be a power of unifying appearances by rules. Reason is then the power of unifying the rules of the understanding under principles. It therefore never goes first to experience or to some object. It goes to the understanding, to give a priori unity, through concepts, to the understanding’s varied cognitions. Call this the unity of reason. It is a completely different sort of unity from anything the understanding can achieve.

That is the general concept of the power of reason, as far as it could be made clear with no examples at all (they can only come later).

B. The Logical Use of Reason

We distinguish between what is known directly and what is only inferred. A figure enclosed by three straight lines has three angles: that is known directly. That these angles add up to two right angles is only inferred. Because we need to infer all the time, we end up used to it. We stop noticing the difference. Often we treat as directly perceived what we have only inferred, as with the so-called deception of the senses.

B 360 Every inference has a proposition at its base, another that is drawn from it (the conclusion), and the consequence, by which the truth of the second is unfailingly connected with the truth of the first. If the inferred judgment is already so contained in the first that it can be derived without a third representation in between, the inference is called immediate (consequentia immediata). I’d prefer to call it an inference of the understanding. But if another judgment is needed besides the cognition used as the base, to produce the conclusion, then the inference is called an inference of reason. The proposition “All humans are mortal” already holds the propositions “Some humans are mortal,” “Some mortals are human,” and “Nothing immortal is human.” These are immediate conclusions from the first. “All scholars are mortal” is different. It isn’t in the underlying judgment, which never mentions the concept of scholars. It can be inferred from that judgment only through a mediating judgment.

In every inference of reason, I start by thinking a rule (major) by means of the understanding. Next, I place a cognition under the condition of that rule (minor), using the power of judgment. Last, I fix my cognition by the rule’s predicate (conclusio).

B 361 I do this a priori, then, by means of reason. The major premise is the rule, and it represents a relation between a cognition and its condition. That relation is what sets the different kinds of inference of reason apart. There are exactly three, just like the kinds of judgment in general. Judgments are divided by the way they express a cognition’s relation in the understanding. The three are categorical, hypothetical and disjunctive inferences of reason.

In most cases the conclusion is put forward as a judgment to be tested. The question is whether it follows from judgments we already have, through which a wholly different object is thought. So I search the understanding for the assertion made in this conclusion. I want to know whether it turns up there, under certain conditions, according to a general rule. Suppose I do find such a condition, and the conclusion’s object fits under it. Then the conclusion is derived from a rule that also holds for other objects of cognition. This shows that reason, when it infers, tries to reduce the great variety of the understanding’s cognitions to the fewest possible principles (general conditions), and so to bring about the highest unity among them.

B 362 C. The Pure Use of Reason

Can reason be isolated? If so, is it still a source of its own, producing concepts and judgments that come from it alone, and through which it relates to objects? Or is it only a subordinate power that gives given cognitions a certain form, called logical? Through it the understanding’s cognitions get ranked beneath one another, and lower rules beneath higher ones, as far as comparison can carry this out. (A higher rule’s condition takes in the condition of the lower ones within its range.) We are now dealing with this question only for a first look. It is true that variety of rules and unity of principles is something reason demands. Reason wants the understanding to hang together with itself all the way through. The understanding does the same for the manifold of intuition: it brings it under concepts and so connects it. But a principle like that gives objects no law. Nor does it hold the ground that would let us cognize and determine them as such. It is merely a subjective law of thrift, for handling what our understanding has in store. By comparing the concepts, it cuts their general use down to the fewest possible. It doesn’t give us the right to demand a certain harmony from the objects themselves.

B 363 Such a harmony would favor the convenience and growth of our understanding. Nor does it let us give that maxim objective validity at the same time. In short, the question is whether reason in itself, that is, pure reason, contains synthetic principles and rules a priori, and what those principles may be.

The formal and logical way reason proceeds in its inferences already gives us enough to go on here. It shows what ground the transcendental principium of reason will have in synthetic cognition from pure reason.

First, an inference of reason isn’t aimed at intuitions, to put them under rules the way the understanding does with its categories. It goes to concepts and judgments. Even when pure reason does go to objects, it has no direct relation to them or to their intuition. It relates only to the understanding and its judgments, which turn first to the senses and their intuition to fix an object for them. Reason’s unity is therefore not the unity of a possible experience. It is essentially different, since the unity of experience is the understanding’s unity. That everything that happens has a cause is not at all a principle that reason knows and lays down. It is what makes the unity of experience possible. It borrows nothing from reason.

B 364 Reason, lacking this relation to possible experience, could never have commanded such a synthetic unity from mere concepts.

Second, in its logical use, reason looks for the general condition of its judgment (the conclusion). The inference of reason is itself just a judgment, reached by subsuming its condition under a general rule (the major premise). That rule faces the same attempt by reason in turn, and so the condition of the condition has to be looked for (by a prosyllogism) as long as that can be done. So we can see what reason’s distinctive principle is, in its logical use generally: for the understanding’s conditioned cognition, find the unconditioned, which completes its unity.

A logical maxim like this can become a principium of pure reason only if we assume something. If the conditioned is given, then the whole series of conditions that are subordinate to one another is given too, and that series is itself unconditioned (that is, it lies in the object itself and in its connection).

A principle of pure reason of this kind is plainly synthetic. From the conditioned we can analytically reach some condition, but not the unconditioned. Various synthetic propositions must also come from it, and pure understanding knows nothing of them.

B 365 Pure understanding deals only with objects of a possible experience, and the cognition and synthesis of such objects are always conditioned. The unconditioned, though, if it really exists, can be considered on its own, with all the determinations that set it apart from anything conditioned. That must supply the material for many synthetic propositions a priori.

Still, the principles that follow from this highest principle of pure reason will be transcendent toward all appearances. No empirical use that fits it will ever be possible. It will therefore be entirely unlike all the principles of the understanding, which are used wholly immanently, since their only topic is the possibility of experience. Take the principle that the series of conditions extends to the unconditioned, whether in the synthesis of appearances or in thinking of things in general. Is it objectively correct, or not? What follows from it for the empirical use of the understanding? Or is there perhaps no objectively valid proposition of reason like it at all? Perhaps there is only a logical rule: climbing toward ever higher conditions, we should come closer to completing them, and so give our cognition the greatest unity of reason that we can achieve. Has this need of reason been mistaken, through a misunderstanding?

B 366 Has it been counted as a transcendental principle of pure reason? Such a principle would rashly postulate that the objects themselves contain an unlimited, complete series of conditions. And what misreadings and delusions may creep in, even then, to inferences of reason that take their major premise from pure reason (a premise that may be closer to a petition than to a postulate), and that climb from experience up to its conditions? All of this will be our task in transcendental dialectic. We now want to develop that dialectic out of its sources, which lie deep in human reason. We will divide it into two chapters. The first will treat the transcendent concepts of pure reason. The second will treat its transcendent and dialectical inferences of reason.

Transcendental Dialectic First Book The Concepts of Pure Reason

However concepts from pure reason may be possible, they are not merely reflected concepts. They are inferred. The understanding’s concepts are likewise thought a priori, prior to experience and serving it.

B 367 But they hold only the unity of reflection on appearances, as far as appearances are supposed to belong necessarily to a possible empirical consciousness. They alone make cognition and determination of an object possible. They therefore supply the first material for inferring, and no earlier concepts a priori of objects exist from which they could be inferred. Their objective reality rests on one thing only. They make up the intellectual form of all experience, and so it must always be possible to show them at work in experience.

The very name “concept of reason” gives a first hint that such a concept won’t be confined within experience. It is about a cognition, and every empirical cognition is only one part of it. That cognition might be the whole of possible experience, or its empirical synthesis. No actual experience ever fully reaches that whole, but every one belongs to it. Concepts of reason serve to comprehend, as concepts of the understanding serve to understand (perceptions). Where they contain the unconditioned, they are about something that all experience falls under, though it is itself never an object of experience. Reason is led to it by its inferences from experience. It uses this thing to estimate and measure how far its empirical use extends.

B 368 But it never forms a member of the empirical synthesis. If such concepts have objective validity all the same, we may call them conceptus ratiocinati (concepts correctly inferred). If they have none, they have at least been got by trickery, through an illusion of inference, and can be called conceptus ratiocinantes (sophistical concepts). That can be settled only in the chapter about pure reason’s dialectical inferences, so for now we can’t take it up. For now, we’ll do what we did when we named the pure concepts of the understanding categories. We’ll give a new name of its own to the concepts of pure reason, transcendental ideas. We will now explain and justify that name.

First Book of the Transcendental Dialectic First Section Ideas in General

Languages are very rich, yet a thinking person is often stuck for the expression that exactly fits their concept. Without it, they can’t make themselves properly understood to others, or even to themselves.

B 369 To forge new words is to presume to make laws for a language, and that rarely works. Before resorting to such a desperate means, it is wise to look around in some dead, learned language. The concept and its fitting expression may already be there. Perhaps the old use of the word has become shaky through its creators’ carelessness. Even so, it is better to pin down the meaning that was especially its own (even if we stay unsure that people then had exactly this meaning in mind) than to ruin one’s purpose by being unintelligible.

So suppose a certain concept has just one word that fits it exactly, in an established meaning, and it is very important to distinguish that concept from related ones. Then it is wise not to be wasteful with the word. Don’t use it for variety’s sake as a synonym for others. Keep its own meaning for it, with care. Otherwise it easily happens that the word doesn’t draw special attention, and gets lost in the crowd of other words with very different meanings. Then the thought is lost as well, the one thought that the word alone could have kept.

B 370 Plato used the word “idea” in a way that makes clear he meant something that is never borrowed from the senses. It even goes far beyond the concepts of the understanding that Aristotle dealt with, since nothing in experience is ever found to match it. Ideas, for Plato, are archetypes of the things themselves, and not merely keys to possible experiences, as the categories are. In his view they flowed from the highest reason, and human reason received them from there. Human reason is no longer in its original state. It has to call the old ideas back, now much dimmed, by effort and recollection (which is called philosophy). I don’t mean to start a literary investigation into the sense this exalted philosopher gave his word. I only point out that it is nothing unusual, in ordinary conversation as in writing, to understand an author better than the author understood themselves. You do this by comparing the thoughts the author gives on the subject. Perhaps the author didn’t define the concept well enough, and so now and then spoke, or even thought, against their own intention.

Plato saw clearly that our power of cognition feels a much higher need than just spelling out appearances by synthetic unity so that they can be read as experience.

B 371 He saw that our reason naturally climbs to cognitions that go much further than any object experience can give could ever match. Yet these cognitions are real, and are in no way mere figments of the brain.

Plato found his ideas mainly in everything practical*, that is, everything that rests on freedom, which in turn is among the cognitions that are a special product of reason. Imagine someone trying to get the concepts of virtue out of experience. Suppose they took what can serve at best as an example for imperfect illustration, and made it a model that is a source of cognition (as many have really done). They would turn virtue into an ambiguous non-thing that shifts with time and circumstances and can’t be used as a rule.

B 372 Everyone, on the other hand, becomes aware of this. When someone is held up to them as a model of virtue, they still carry the true original only in their own head. They compare the supposed model against it, and judge it only by that. That original is the idea of virtue. For this idea, all possible objects of experience are useful as examples (proof that what the concept of reason demands can be done to some degree), but not as archetypes. That no human will ever act fully as the pure idea of virtue requires does not show that the thought is somehow chimerical. For all judgment of moral worth or worthlessness is possible only by means of this idea. So the idea has to lie beneath every approach to moral perfection, even if obstacles in human nature, which can’t be measured by degree, hold us far from it.

* Of course he stretched his concept to speculative cognitions as well, as long as they were given purely and wholly a priori. He even stretched it to mathematics, though mathematics finds its object only in possible experience. I can’t follow him there. I can’t follow his mystical deduction of these ideas either, or the exaggerations that made them, so to speak, into things in their own right. Still, the elevated way he spoke on this subject can easily be read more gently, in a way that suits how things really are.

Plato’s Republic has become a proverb. People take it for a striking example of dreamed-up perfection, the kind that could live only in the head of an idle thinker. Brucker finds it ridiculous that the philosopher said that no prince would ever rule well without a share in the ideas. But we would do better to follow this thought further. Where the excellent man leaves us without help, we should bring it to light by fresh effort, and not push it aside as useless on the excuse that it can’t be done.

B 373 That excuse is very wretched and harmful. Consider a constitution that gives people the greatest freedom possible, under laws that let each person’s freedom stand together with everyone else’s. (I don’t mean the greatest happiness, because happiness will follow of itself.) At the very least that is a necessary idea. We have to put it at the base of every law, and not only of the first draft of a state constitution. And at the start we have to look past the obstacles in our way today. Those obstacles may not spring unavoidably from human nature so much as from neglect of the true ideas when laws were made. For nothing is more harmful, or less worthy of a philosopher, than the crude appeal to an experience that objects to the idea to no purpose. That experience wouldn’t even exist if the right arrangements had been made in good time, following the ideas. It exists because crude concepts took their place, and these concepts, just because they were drawn from experience, have wrecked every good intention. Suppose lawmaking and government were set up to fit this idea ever more closely. Punishments would certainly become rarer. So Plato is quite reasonable when he says that a perfect arrangement of lawmaking and government would need no punishments of that kind at all. This last point may never come about.

B 374 Even so, the idea is entirely right when it sets up this maximum as the archetype, so that we can bring the lawful constitution of human beings closer and closer to the greatest perfection possible. For no one can say, and no one should try to say, what the highest degree is at which humanity has to stop. No one can say how wide the gap must be that always remains between the idea and its carrying out. That is just because it is freedom, and freedom can go beyond any boundary anyone names.

Plato is right to see clear proofs that things originate in ideas. He sees them in morals, where human reason shows real causality and ideas become effective causes (of actions and of their objects). But he also sees them in nature itself. A plant, an animal, the orderly layout of the cosmos, and presumably the whole order of nature along with it, make it plain that they could exist only by following ideas. No single creature, under the particular conditions of its existence, matches the idea of the most perfect of its kind. In the same way a human being doesn’t match the idea of humanity, which each of us carries in our own soul as the archetype of our actions. Yet in the highest understanding those ideas are individual, unchangeable and thoroughly determined, and they are the original causes of things.

B 375 Only the whole that they form together in the universe fits that idea completely. Strip away the exaggeration in how he puts it. Then the philosopher’s soaring leap deserves respect and imitation. He rises from a copy-like way of looking at the physical side of the world’s order up to its architectonic connection by purposes, that is, by ideas. In the area of the principles of morality, lawmaking and religion, though, it is a quite special merit. Here the ideas are what make experience (of the good) possible at all, even though experience can never express them completely. People fail to see this merit only because they judge it by the same empirical rules whose validity as principles the ideas were meant to abolish. For where nature is concerned, experience gives us the rule and is where truth comes from. Where moral laws are concerned, experience is (sadly!) the mother of illusion. It is utterly objectionable to take the laws for my conduct from what people actually do, or to hem them in by it.

Working out all these reflections properly would, in fact, give philosophy its own special dignity. But instead of them we are busy now with a job that is less splendid, though not without merit.

B 376 We must level the ground for those majestic moral buildings and make it firm enough to build on. That ground is full of mole tunnels, dug by a reason that goes on digging for treasure, in vain but with good confidence, and these tunnels make the building unsafe. So what we now have to know exactly is the transcendental use of pure reason, with its principles and ideas. Only then can we properly determine and assess the influence and value of pure reason. Before I put this preliminary introduction aside, though, I have a request. It is for those who really care about philosophy (which is more than you commonly find). If this and what follows convinces them, I ask them to take the word “idea” under their protection in its original meaning. It shouldn’t slip any further in among the other words that people use for all sorts of representations, in careless disorder, with science the loser. We aren’t short of names that fit each kind of representation properly, and we don’t need to trespass on what belongs to another. Here is a ladder of them. The genus is representation in general (repraesentatio). Below it comes representation with consciousness (perceptio). A perception that refers only to the subject, as a change in its state, is sensation (sensatio). An objective perception is cognition (cognitio).

B 377 Cognition comes in two kinds, intuition and concept (intuitus vel conceptus); an intuition refers directly to the object and is singular, while a concept refers to it indirectly, by means of a mark that several things can share. A concept is either empirical or pure. If a pure concept comes only from the understanding (and not from the pure image of sensibility), it is named a notio. A concept built from notions that goes beyond the possibility of experience is the idea, or concept of reason. Anyone who has got used to this distinction will find it unbearable that the representation of the color red should be called an idea. It doesn’t even deserve the name notion (a concept of the understanding).

First Book of the Transcendental Dialectic Second Section The Transcendental Ideas

The analytic showed us by example that the bare logical form of our cognition can be where pure concepts a priori come from. These concepts represent objects before all experience. Or rather, they indicate a synthetic unity.

B 378 That unity alone makes an empirical cognition of objects possible. The form of judgments, once changed into a concept of how intuitions are synthesized, gave rise to the categories, which guide every use of the understanding in experience. In the same way we can expect the form of inferences of reason to hold the source of special concepts a priori. This will happen when we apply that form to the synthetic unity of intuitions, as the categories direct. We can call these concepts pure concepts of reason, or transcendental ideas. They will determine, by principles, how the understanding is used in all of experience.

In its inferences, the function of reason was to give cognition universality through concepts. An inference of reason is itself a judgment that is settled a priori across the whole range of its condition. I could take the proposition “Caius is mortal” from experience, through the understanding alone. But I look for a concept that contains the condition under which the predicate of this judgment (assertion in general) is given. Here that is the concept of a human being. Once I have subsumed my object under that condition, taken in its full range (all human beings are mortal), I determine my cognition of it accordingly (Caius is mortal).

In the conclusion of an inference of reason, then, we narrow a predicate down.

B 379 We narrow it to one particular object, after first thinking the predicate in the major premise, in its whole extent, under some condition. When the extent is complete with respect to such a condition, we call its size universality (Universalitas). In the synthesis of intuitions, what matches it is allness (Universitas), the totality of conditions. So the transcendental concept of reason is just the concept of a totality of conditions for any conditioned thing that is given. Only the unconditioned can make a totality of conditions possible. And the other way around, a totality of conditions is itself always unconditioned. That lets us explain a pure concept of reason, in general, through the concept of the unconditioned, to the extent that it holds a ground for the synthesis of what is conditioned.

The categories give the understanding a certain number of kinds of relation. So there will be that many different pure concepts of reason. So we shall have to look for three things. First, an unconditioned for the categorical synthesis in a subject. Second, one for the hypothetical synthesis that links the members of a series. Third, one for the disjunctive synthesis that joins the parts in a system.

For there are just as many kinds of inferences of reason. Each kind climbs through prosyllogisms toward the unconditioned. One climbs to a subject that is itself no longer a predicate. Another climbs to a presupposition.

B 380 That presupposition presupposes nothing further. The third climbs to a collection made up of the members of a division, and nothing more is needed to complete the division of a concept. Pure concepts of reason about totality in the synthesis of conditions are therefore necessary, and they are rooted in the nature of human reason. They are so at least as tasks: the task of carrying the unity of the understanding on, if possible, all the way to the unconditioned. This stays true even if these transcendental concepts have no fitting use in concrete cases. They would then serve no purpose beyond steering the understanding in a direction where its use is stretched as far as it will go and, at the same time, brought into complete agreement with itself.

Now we are speaking of the totality of conditions and of the unconditioned, as the common title of all concepts of reason. Here again we meet an expression that we can’t do without, yet can’t use safely either, because long misuse has left it ambiguous. “Absolute” is among the few words that began life fitted to one concept, a concept for which no other word of the same language fits exactly afterwards. Suppose the word is lost. Or suppose, and it comes to the same thing, that its use keeps wavering.

B 381 Then the concept itself must be lost with it, and this concept occupies reason a great deal, so that all transcendental judging would suffer badly without it. These days “absolute” is often used merely to say that something is true of a thing as it is in itself, and so inwardly. Then “absolutely possible” would mean possible in itself (interne). That is really the least we can say about an object. At other times the word is used to say that something is valid in every relation, without restriction (absolute rule, for example). Then “absolutely possible” would mean possible in every respect and in every relation. That, in turn, is as much as I can say about whether a thing is possible. These two meanings do sometimes coincide. Whatever is inwardly impossible, for example, is also impossible in every relation, and so absolutely impossible. But in most cases they lie infinitely far apart. Because something is possible in itself, I can’t conclude that it is possible in every relation, and so absolutely possible. Indeed, I’ll show later that absolute necessity by no means depends on inner necessity in every case. The two must not be treated as having the same meaning. Take something whose opposite is inwardly impossible.

B 382 Its opposite is of course also impossible in every respect, and so the thing itself is absolutely necessary. But I can’t turn this around. I can’t say that if something is absolutely necessary, its opposite is inwardly impossible. That would mean that, for things, absolute necessity and inner necessity are the same. For in certain cases inner necessity is a completely empty expression, and we can’t connect the slightest concept with it. By contrast, the concept of a thing’s necessity in every relation (to everything possible) brings quite particular determinations with it. A philosopher can never be indifferent to losing a concept that serves speculative philosophy so widely. So I hope they won’t be indifferent to fixing the expression the concept hangs on, and to keeping it with care.

I will use the word “absolute” in this wider meaning, then. I will set it against what is valid only comparatively, or in some particular respect. The latter is restricted to conditions. The former holds without restriction.

The transcendental concept of reason always aims at one thing only: absolute totality in the synthesis of conditions. It never comes to rest until it reaches what is unconditioned without qualification, that is, unconditioned in every respect. For pure reason leaves everything to the understanding.

B 383 The understanding is the faculty that first deals with objects of intuition, or rather with their synthesis in the imagination. Reason reserves for itself alone the absolute totality that comes up when concepts of the understanding are used. It tries to push the synthetic unity that a category thinks all the way to what is unconditioned without qualification. So we may call this kind of unity reason’s unity of appearances, and call the unity that a category expresses the understanding’s unity. Reason, then, is concerned only with how the understanding is used. It is not concerned with the understanding as the ground of possible experience. (The absolute totality of conditions can’t be used as a concept in any experience, because no experience is unconditioned.) What reason does is prescribe a direction for the understanding, toward a certain unity of which the understanding itself has no concept. This unity aims to gather all the understanding’s acts, for each object, into one absolute whole. That is why pure concepts of reason, when used on objects, are always transcendent. Pure concepts of the understanding are different. By their nature they must always be used immanently, because they are confined to possible experience.

By an idea I mean a necessary concept of reason for which no matching object can ever be given to the senses. The pure concepts of reason we have been weighing are therefore transcendental ideas. They are concepts of pure reason, because they treat all cognition from experience as settled by an absolute totality of conditions.

B 384 Nobody invented them at will. The nature of reason itself sets them as a task, so they necessarily bear on the whole use of the understanding. Finally, they are transcendent. They go beyond the boundary of all experience, and so no object can ever turn up in experience that would be adequate to a transcendental idea. To call something an idea says a great deal about the object, if we think of it as an object of the pure understanding. It says very little about the subject, meaning its actuality under empirical conditions. The reason is this: as the concept of a maximum, an idea can never be given in concreto as a perfect match. In the merely speculative use of reason, though, that match is really the whole point. And getting close to a concept that is never reached in practice is no better than missing it altogether. So people say of such a concept: it is only an idea. They could say, for instance, of the absolute whole made up of all appearances that it is only an idea, because we can never draw up anything of the kind as an image, so it stays a problem with no solution. The practical use of the understanding is different. There all that matters is acting by rules.

B 385 So the idea of practical reason can always be actually given in concreto, though only in part. It is even indispensable: no practical use of reason works without it. Acting on it is always bounded and falls short, but within boundaries that can’t be fixed. So the concept of absolute completeness always has some influence on it. For that reason the practical idea is always highly fruitful, and for real actions it can’t be done without. In it pure reason actually has causality: it brings about in fact what its concept contains. So we can’t say of wisdom, as if brushing it off: it is only an idea. Wisdom is the idea of one necessary unity among all possible purposes, and for exactly that reason it has to serve as the rule for everything practical, as the original condition, or at least a restricting one.

We have to say that the transcendental concepts of reason are only ideas. That doesn’t make them superfluous or empty. For even though they can’t determine any object, deep down and without our noticing, they can serve the understanding as a canon for using itself widely and consistently. They don’t let the understanding cognize more objects than it would by its own concepts. But they do guide it better and further in that cognition.

B 386 Not to mention that they may perhaps open a way over from concepts of nature to practical ones. In that way they could give the moral ideas themselves a hold, and a connection with the speculative cognitions of reason. We will have to wait for light on all this in what follows.

Our aim lets us set the practical ideas aside here. We consider reason in its speculative use alone, and within that, more narrowly still, in its transcendental use alone. The path to take is the one we took earlier in deducing the categories. We weigh the logical form of the cognition of reason, and we ask whether that form might also make reason a source of concepts. Such concepts would let us regard objects in themselves as determined synthetically a priori, in connection with one or another function of reason.

Reason, taken as the power behind one particular logical form of cognition, is the power of inferring, which means judging indirectly: the condition of one possible judgment is brought under the condition of a given one. The given judgment is the general rule (major premise, Major). Next comes a second possible judgment. Its condition is brought under the rule’s condition, and that gives the minor premise (Minor). The actual judgment states the rule’s assertion in the case brought under it.

B 387 That judgment is the conclusion (Conclusio). The rule, after all, says something in general under a certain condition. In the case at hand that condition is met. So what held in general under that condition is also taken to hold in this case, which carries the condition with it. It is easy to see that reason gets to a cognition by acts of the understanding that make up a series of conditions. Say I reach the proposition “All bodies are changeable” only like this. I begin with a more remote cognition, in which the concept of body does not yet occur but which holds the condition for it: “Everything composite is changeable.” From there I move to a nearer one that stands under the first one’s condition: “Bodies are composite.” Only then do I move to a third cognition. This one now connects the remote cognition (changeable) with the one in front of me: “So bodies are changeable.” Through a series of conditions (premises) I have reached a cognition (conclusion). Now take any series whose exponent is given (whether of a categorical or of a hypothetical judgment). Such a series can always be continued. The very same act of reason therefore leads to the ratiocinatio polysyllogistica, a chain of inferences that can be run on to any length. It can run toward the conditions (per prosyllogismos).

B 388 Or it can go the other way, toward what is conditioned (per episyllogismos).

Soon, though, we notice something. The chain, or series, of prosyllogisms has to stand differently toward the faculty of reason than the downward series does. By prosyllogisms I mean the inferred cognitions on the side of the grounds, or conditions, of a given cognition, so, in other words, the upward series of inferences of reason. The downward series is how reason moves forward toward what is conditioned, through episyllogisms. For in the first case the cognition (conclusio) is given only as something conditioned. Reason can reach it only by assuming, at the very least, that every member of the series on the side of the conditions is given (a totality in the series of premises). Only if that is assumed can the judgment in front of us be made a priori. On the side of the conditioned, the consequences, we think only of a series that is still becoming. It isn’t already fully presupposed or given, so what we think is merely a potential advance. So when a cognition is viewed as conditioned, reason is forced to treat the upward series of conditions as complete and as given in its totality. Now suppose the same cognition is also viewed as the condition of other cognitions.

B 389 These form a series of consequences running downward. Then reason doesn’t care how far this progression reaches a parte posteriori, or whether this series can ever be a totality at all. It has no need of such a series for the conclusion lying before it, since its grounds a parte priori already determine and secure that conclusion well enough. On the side of the conditions, the series of premises may or may not have a first member, a highest condition, and so may or may not be unbounded a parte priori. Either way it has to hold a totality of the condition, even if we could never succeed in grasping it. And the series as a whole must be true without condition, if what is conditioned, which counts as a consequence flowing from it, is to be accepted as true. This is a demand of reason. Reason announces its cognition as determined a priori and as necessary. That cognition is necessary either in itself, and then it needs no grounds, or, if it is derived, as one link in a chain of grounds that is itself true without condition.

B 390 First Book of the Transcendental Dialectic Third Section System of the Transcendental Ideas

This is not a logical dialectic that sets aside all the content of cognition and does nothing but expose the false illusion in the form of inferences of reason. We are dealing with a transcendental dialectic. It has to hold, wholly a priori, the source of certain cognitions out of pure reason, and of inferred concepts. Their object can’t be given empirically at all, so they lie entirely beyond the power of the pure understanding. Our use of cognition in the transcendental way must relate naturally to the logical way, in inferences and in judgments alike. From that relation we concluded two things. First, there can be only three kinds of dialectical inference, matching the three kinds of inference by which reason moves from principles to cognitions. Second, all their work is to climb from the conditioned synthesis, to which the understanding stays bound, up to the unconditioned one, which the understanding can never reach.

Now what is common to every kind of relation that our representations can have is: 1.

B 391 the relation to the subject, 2. the relation to objects, taken as appearances or else as objects of thought generally. Combine this subdivision with the one above. Every relation among representations that we can turn into a concept or an idea is then threefold: 1. the relation to the subject, 2. the relation to what is manifold in the object as it appears, 3. the relation to all things in general.

All pure concepts, in general, deal with the synthetic unity of representations. Concepts of pure reason (transcendental ideas) deal with the unconditioned synthetic unity of all conditions, whatever they may be. So we can sort all transcendental ideas into three classes. The first class holds one thing: the thinking subject’s absolute (unconditioned) unity. The second holds the series of conditions of appearance, in its absolute unity. The third holds the condition of all objects of thought, in its absolute unity.

Psychology has the thinking subject as its object. Cosmology has the sum of all appearances, the world. Theology has the thing that holds the highest condition for the possibility of everything thinkable, the being of all beings. From this, pure reason hands us the idea of a transcendental teaching about the soul (psychologia rationalis).

B 392 It hands us another idea for a transcendental world-science (cosmologia rationalis). And it hands us a last one for a transcendental knowledge of God (Theologia transzendentalis). The understanding is not where even a bare outline of one of these sciences comes from. That holds even if the understanding were joined with reason’s highest logical use, meaning every inference one can think of, to move on from one of its objects (appearance) to all the others, out to the farthest links of the empirical synthesis. The outline is simply a pure, authentic product of pure reason, or a problem it poses.

Which modi of the pure concepts of reason fall under these three headings of transcendental ideas will be set out in full in the next chapter. They follow the thread of the categories. For pure reason never relates straight to objects. It relates to the concepts of the understanding about them. Likewise only a full working-out can make clear how reason, simply by using in a synthetic way the same function it uses in the categorical inference of reason, has to reach the concept of the thinking subject’s absolute unity. It will also show how the logical procedure in the hypothetical kind has to lead to the idea of the absolutely unconditioned within a series of given conditions.

B 393 And it will show how the bare form of the disjunctive inference of reason has to bring with it the highest concept of reason: the concept of a being of all beings. At first glance this thought seems very paradoxical.

An objective deduction of these transcendental ideas, like the one we could give for the categories, isn’t really possible. For in fact there is no object for them to relate to that could be given as a match, simply because they are only ideas. A subjective account of where they come from in the nature of our reason was something we could attempt, and the present chapter has done that.

It’s plain that pure reason has one aim: the absolute totality of the synthesis, on the side of the conditions (whether of inherence, of dependence, or of concurrence). Reason has no business with absolute completeness on the conditioned side. It needs only the former, in order to presuppose the whole series of conditions and so to give it to the understanding a priori. Once a condition has been given completely (and unconditionally), no concept of reason is needed any more for carrying on the series. The understanding takes every step downward by itself. It goes from the condition to the conditioned.

B 394 In this way the transcendental ideas serve only for climbing up the series of conditions, as far as the unconditioned, that is, as far as the principles. Going down to the conditioned is another matter. Our reason does make wide logical use of the laws of the understanding there, but no transcendental use at all. Say we try to form an idea of such a synthesis (of the progressus) in its absolute totality, for example the whole series of every future change in the world. That is a thing of thought (ens rationis), which we think up at will and which reason does not necessarily presuppose. The possibility of something conditioned does presuppose the totality of its conditions, but not the totality of its consequences. So such a concept is not a transcendental idea, and transcendental ideas are all we are dealing with here.

Last of all, we also notice that a certain connection and unity shows among the transcendental ideas themselves. Through them pure reason puts all its cognitions into a system. Going from the cognition of oneself (the soul) to the cognition of the world, and by way of that to the original being, is such a natural step that it seems to resemble the logical progress of reason.

B 395 That progress runs from the premises to the conclusion*. Whether a hidden kinship really lies underneath here, like the one between the logical and the transcendental procedure, is another question. We can expect its answer only later, as these investigations go on. For now we have already reached our aim.

B 396 That is because we were able to pull the transcendental concepts of reason out of an ambiguous position, in which they usually get mixed in with other concepts in philosophers’ theories, and philosophers don’t even properly tell them apart from concepts of the understanding. We have been able to give their origin and, with it, their definite number, beyond which there can be none. We have also been able to present them in a systematic connection. This marks off and limits a special field for pure reason.

* Metaphysics has only three ideas as the real purpose of its inquiry: God, freedom and immortality. The second, together with the first, is meant to bring us to the third, as a conclusion that has to follow. Everything else this science deals with serves it only as a means of reaching these ideas and their reality. It doesn’t need them for the sake of natural science. It needs them to get beyond nature. Insight into them would make theology, morals and, through the two together, religion, and so the highest purposes of our existence, depend only on the speculative power of reason and on nothing else. In a systematic presentation of those ideas, the order just given, as the synthetic one, would be the most fitting. But in the work that has to come before it, the analytic order, which turns this order around, will suit the purpose better. We start with what experience hands us directly, the doctrine of the soul. We go on to the doctrine of the world, and then to the cognition of God, and so carry out our great plan.

Transcendental Dialectic Second Book The Dialectical Inferences of Pure Reason

One might put it this way: for a mere transcendental idea there is an object of which we have no concept at all, even though reason produces the idea quite necessarily by its original laws. Indeed, for an object that is to meet the demand of reason, no concept of the understanding can exist. By that I mean a concept that can be shown, and made intuitive, in a possible experience. It would still be better, and less risky for misunderstanding, to put it another way.

B 397 We should say that we can know nothing of the object that answers to an idea, though we can form a problematic concept of it.

Now the transcendental (subjective) reality of the pure concepts of reason rests, at least, on this: that a necessary inference of reason leads us to such ideas. So there will be inferences of reason that contain no empirical premises. Using them, we reason from something known to something else, of which we have no concept, and to which we nonetheless give objective reality through an unavoidable illusion. Judged by their result, such inferences should be called sophistical rather than inferences of reason. Because of what prompts them, though, they can still carry the latter name, since they aren’t made up and didn’t arise by chance. They sprang from the nature of reason. These are tricks of argument that come not from people but from pure reason itself. Not even the wisest person can break free of them. After much effort, they may perhaps avoid the error. But they can never get rid of the illusion that keeps nagging and mocking them.

These dialectical inferences of reason come in only three kinds, one for each of the ideas in which their conclusions end. Take the first class.

B 398 I start with the subject’s transcendental concept, which holds nothing manifold, and conclude to the absolute unity of that very subject. In that way I have no concept at all of the subject. This dialectical inference I will name the transcendental paralogism. The second class of sophistical inferences starts from the transcendental concept of an absolute totality: the series of conditions for any given appearance. From the fact that my concept of the series’ unconditioned synthetic unity is always self-contradictory on one side, I conclude that the opposite unity is the correct one. Yet of this one too I have no concept. The condition reason is in during these dialectical inferences I will call the antinomy of pure reason. Finally, in the third kind of sophistical inference I start from the totality of conditions needed to think objects in general, as far as they can be given to me. From that I conclude to an absolute synthetic unity that covers every condition of the possibility of things in general. That is, from things I don’t know through their bare transcendental concept, I infer to a being of all beings. I know this being through a transcendent concept even less, and I can form no concept of its unconditioned necessity. This dialectical inference of reason I will name the Ideal of pure reason.

B 399 Second Book of the Transcendental Dialectic First Chapter The Paralogisms of Pure Reason

A logical paralogism is an inference of reason that is false in form, whatever its content may be. A transcendental paralogism, though, has a transcendental ground for reasoning falsely in form. Such a fallacy, then, is rooted in the nature of human reason. It carries with it an illusion that can’t be avoided, though it can be resolved.

Now we come to a concept that didn’t appear in the general list of transcendental concepts earlier. It still has to be counted among them, yet that doesn’t change the table in the slightest, nor does it show the table to be defective. This is the concept, or, if you prefer, the judgment: I think. But one can see right away that it serves as the vehicle for every concept, the transcendental ones included. It is always counted in among them, and so it is transcendental as well. But it can have no special heading of its own.

B 400 That is because it serves only to bring in all thinking as belonging to consciousness. Even so, though it is free of anything empirical (the impression of the senses), it serves to tell apart, by how our power of representation works, two kinds of objects. As thinking, I am an object of inner sense, and I am called soul. What is an object of the outer senses is called body. So the word “I,” said of a thinking being, already points to the object of psychology. We may call that psychology the rational doctrine of the soul, as long as all I ask to know about the soul is what can be concluded from the concept “I” as it occurs in all thinking, apart from any experience (which would determine me more closely and in concreto).

The rational doctrine of the soul really is an undertaking of that sort. For if even the slightest empirical element of my thinking, some particular perception of my inner state, got mixed into what this science rests its knowledge on, it would no longer be a rational doctrine of the soul. It would be an empirical one. So we have a science already, or so it is claimed, resting on the single proposition “I think.” We can quite fittingly examine whether it has a ground or not, as the nature of a transcendental philosophy requires. We needn’t be bothered that I do have an inner experience in this proposition, the one that states my perception of myself.

B 401 Nor should it bother us that, as a result, the rational doctrine of the soul built on this proposition is never pure but rests partly on an empirical principium. For this inner perception is nothing more than the bare apperception “I think.” It is this that makes all transcendental concepts possible, the ones in which we say: I think substance, cause, and so on. Inner experience in general and how it is possible, or perception in general and how it relates to other perception, can’t be counted as empirical cognition so long as no particular difference or determination of it is given empirically. Instead it has to be taken as cognition of the empirical in general. It belongs in the study of how any experience is possible, and that study is certainly transcendental. The slightest object of perception (pleasure or displeasure, say) added to the general representation of self-consciousness would at once turn rational psychology into an empirical one.

So the only text rational psychology has is “I think.” From it, this science is supposed to unfold all its wisdom. One can see at once that this thought, once it is applied to an object (myself), can hold only transcendental predicates of that object. The slightest empirical predicate would spoil the rational purity of the science and its independence from all experience.

B 402 Here we will only have to follow the guiding thread of the categories. But a thing, I, as a thinking being, is given here first. So we will not alter the order the table gives the categories among themselves. We will, however, begin here with substance, since that category represents a thing in itself, and then work back through the others in order. The topics of rational psychology, from which all its other content has to be derived, are therefore these:

1 The soul is substance.

2 Simple in its quality.

3 Numerically identical across the different times in which it exists, that is, unity (not plurality).

4 In relation to possible objects in space*.

* A reader may not easily guess from these expressions, in their transcendental abstractness, what they mean psychologically, or why the last attribute of the soul belongs to the category of existence. That reader will find them explained and justified well enough in what follows. As for the Latin expressions that have crept in where equivalent German ones should stand, against the taste of good style, I have this excuse to offer, for this section and for the whole work: I preferred to give up some elegance of language rather than make school use harder by the slightest obscurity.

B 403 The pure doctrine of the soul gets all its concepts from these elements, simply by putting them together, without recognizing any other principium at all. Substance considered only as an object of inner sense yields the concept of immateriality. As simple substance, it yields incorruptibility. Its identity, as intellectual substance, gives personality. All three of these together give spirituality. Being related to objects in space gives rise to commerce with bodies. The result is thinking substance presented as the principium of life in matter: soul (anima), and the ground of animality. Animality, restricted by spirituality, gives immortality.

Four paralogisms connect with this. They belong to a transcendental doctrine of the soul, which people wrongly take for a science of pure reason about the nature of our thinking being.

B 404 The only thing we can lay at its base is the simple representation “I,” which is completely empty of content in itself. One can’t even say it is a concept. It is a bare consciousness that accompanies all concepts. All that is represented through this I, or this he or it (the thing that thinks), is a transcendental subject of thoughts = x and nothing more. We know this subject solely by the thoughts that serve as its predicates. Taken apart from them, we can have no concept of it, not even the slightest. So we keep turning around it in a constant circle, because we always have to use its representation already in order to judge anything about it. This inconvenience can’t be separated from it, because consciousness by itself isn’t so much a representation that picks out some particular object as the form that any representation has, so far as it deserves to be called cognition. For only of cognition can I say that by it I think something.

It has to seem strange right from the start that the condition under which I think at all, which is therefore only a feature of my own subject, should hold at the same time for everything that thinks. It also seems strange that we can presume to build an apodictic and universal judgment on a proposition that looks empirical. The judgment is this: that everything that thinks is constituted as the declaration of self-consciousness says it is in me.

B 405 The reason for this lies in the fact that we necessarily have to attribute to things a priori all the properties that make up the conditions under which we can think them at all. Now no outer experience can give me even the faintest representation of a thinking being. Only self-consciousness can. So such objects are simply this consciousness of mine, carried over to other things. Only by that transfer are they represented as thinking beings at all. The proposition “I think” is taken here only problematically. I do not take it as holding the perception of any existence (the Cartesian cogito, ergo sum). I take it in its bare possibility. My only aim is to see which properties might follow, for the subject of this very simple proposition, whether or not such a subject exists.

Suppose our pure cognition by reason of thinking beings in general had more than the cogito underneath it. Suppose we also drew on observations of how our thoughts play out, and on the laws of nature governing the thinking self that could be gathered from them. The result would be an empirical psychology, a sort of physiology of inner sense. It could perhaps explain the appearances of inner sense. But it could never serve to uncover properties that lie outside possible experience altogether (such as simplicity).

B 406 It also couldn’t teach anything apodictically about thinking beings in general, about their nature. So it would not be a rational psychology.

Taken problematically, the proposition “I think” holds the form of every judgment of the understanding. It also accompanies all the categories as their vehicle. So it is clear that the inferences drawn from it can involve only a transcendental use of the understanding. This use turns away every admixture of experience. From what we showed above, we can already tell in advance that we can form no favorable idea of how it will turn out. Let us trace it with a critical eye through every predicament of the pure doctrine of the soul. To keep it short, though, we will let the examination run on in one continuous line.

First, a general remark can make us more watchful about this kind of inference. Merely thinking never gives me cognition of an object. I cognize an object only when I determine a given intuition in light of the unity of consciousness, which is what all thinking consists of. So I don’t cognize myself just by being aware that I think. I cognize myself when I am aware of my own intuition as determined in light of the function of thinking.

B 407 None of the modi of self-consciousness in thinking, taken by itself, is yet a concept of the understanding for objects (a category); they are mere functions, and from them thinking gets no object to cognize, so it doesn’t get me myself as an object either. What counts as the object is not the consciousness of what determines. It is only the consciousness of the self that can be determined, meaning my inner intuition, so far as its manifold can be combined in keeping with the general requirement of apperception’s unity in thinking.

1) In every judgment, it is always I who am the subject that determines the relation making up the judgment. That I, who think, can always be taken in thinking as a subject, as something that doesn’t merely attach to thinking like a predicate, and must count as one, is an apodictic and even identical proposition. But it doesn’t mean that I, as an object, am a being that subsists on its own, or a substance. That claim goes very far. It therefore requires data that are not found in thinking at all, and perhaps, if I regard the thinker purely as a thinker, more than I will ever meet in it.

2) In every act of thinking, the I of apperception is a singular, and it can’t be dissolved into a plurality of subjects. So it stands for a logically simple subject. That much is already contained in the concept of thinking, so this is an analytic proposition.

B 408 But that doesn’t mean the thinking I is a simple substance. That would be a synthetic proposition. The concept of substance always refers to intuitions, and in my case these can only be sensible. These lie entirely outside the field of the understanding and its thinking. Yet when we call the I simple in thinking, the only thing we are really talking about is thinking. It would also be a wonder if something that otherwise takes so much effort were handed to me here so directly, in the poorest representation of all, as if by a revelation. I mean the effort of working out which part of what an intuition lays out is substance, and, still harder, whether that substance might be simple too (the way the parts of matter are).

3) The proposition of my identity amid whatever manifold I am conscious of is likewise one that lies in the concepts themselves, and so it is an analytic proposition. But this identity of the subject, which I can become aware of in each of its representations, has nothing to do with the intuition by which it is given as an object. So it also can’t mean personal identity, which is the consciousness that one’s own substance, as a thinking being, stays the same through every change of state. Merely analyzing the proposition “I think” could not prove that.

B 409 We would need various synthetic judgments, which rest on the given intuition.

4) I set apart the existence that is mine, that of a thinking being, from the things outside me (my body is one of them). That too is an analytic proposition. After all, other things are simply what I think of as different from me. But I learn nothing from this about whether I could be conscious of myself at all without things outside me, the things that supply my representations. So I learn nothing about whether I could exist as a mere thinking being (without being a human being).

So analyzing my consciousness of myself in thinking in general gains not the slightest thing toward cognizing myself as an object. A logical account of thinking in general is wrongly taken to be a metaphysical determination of the object.

Our whole Critique would face a great stumbling block, in fact its only one, if some way existed to prove a priori that thinking beings, one and all, are simple substances in themselves. As such they would inseparably carry personality with them (which follows from the same ground of proof), and they would be conscious of their existence as separate from all matter. For in that way we would have taken a step beyond the sensible world. We would have entered the field of noumena.

B 410 Now let no one dispute our right to spread out further in that field, to cultivate it and, if each person’s lucky star favors them, to take possession of it. The reason is that the claim “A thinking being is, just as such, a simple substance” is a synthetic proposition a priori. For one thing, it goes beyond the concept at its base and adds a kind of existence to thinking in general. For another, it attaches to that concept a predicate (simplicity) that no experience can supply. So synthetic propositions a priori would be workable and admissible not merely, as we have claimed, in relation to objects of possible experience, and as principles of the possibility of that experience itself. They could also apply to things in general and in themselves. That consequence would put an end to this whole Critique and would order us to leave everything as it was. But the danger is not so great here, once one looks at the matter more closely.

In the procedure of rational psychology a paralogism is at work. It is set out in the following inference of reason:

What can’t be thought other than as a subject also doesn’t exist other than as a subject, and so is substance.

B 411 Now, if we regard a thinking being just as such, it cannot be thought except as a subject.

Therefore it exists only as a subject too, that is, as a substance.

The major premise talks about a being that can be thought of in any respect at all, and so also in whatever way intuition may give it. The minor premise, though, talks about that being only as it considers itself, as subject, purely in connection with thinking and the unity of consciousness, and not also in connection with the intuition by which it is given as an object for thinking. So the conclusion is drawn per sophisma figurae dictionis, that is, by a fallacy*.

* Thinking is taken in two quite different meanings in the two premises. In the major premise it means thinking as it goes to an object in general (and so as the object may be given in intuition). In the minor premise it means thinking only as it consists in the relation to self-consciousness. There no object at all is thought. What gets represented is only one’s relation to oneself as subject, which is the form of thinking. The first premise speaks of things that can only be thought as subjects. The second does not speak of things at all. It speaks of thinking (leaving every object out of account), in which the I is always what serves as consciousness’s subject. So the conclusion cannot be: I can’t exist other than as a subject. It can only be: in thinking my existence I can use myself only as the subject of the judgment. That is an identical proposition, and it reveals absolutely nothing about the kind of existence I have.

B 412 That this resolution of the famous argument into a paralogism is quite correct is plainly shown if one turns to the general remark attached to the systematic presentation of the principles, and to the section on noumena. That discussion proved that the concept of a thing able to exist for itself as subject, and not as a mere predicate, has no objective reality of its own. That is, one can’t know whether any object can correspond to it, because we can’t see how such a way of existing could be possible. So it yields no cognition whatsoever. Suppose the concept is to point, under the name of substance, to an object that can be given, and to become a cognition. Then a persistent intuition has to be laid at its base. Such an intuition is the one condition a concept cannot do without for objective reality: it is what alone gives the object.

B 413 Now inner intuition gives us nothing persistent whatsoever, because the I is nothing but the awareness of my thinking. So long as we stay with thinking alone, the necessary condition is missing for applying the concept of substance (a subject subsisting for itself) to ourselves as thinking beings. The simplicity of substance connected with that concept then drops away entirely along with the concept’s objective reality. What remains is only a logical unity of self-consciousness in thinking generally, a unity of quality, whether or not the subject is composite.

Refutation of Mendelssohn’s Proof of the Soul’s Persistence

This sharp-minded philosopher soon noticed a shortcoming in the usual argument meant to prove that the soul (if one grants that it is a simple being) cannot cease to be by being divided. The argument isn’t sufficient for the aim of securing its necessary continuation, because one could still assume that its existence ends by vanishing. In his Phaedo he then tried to keep this perishing, which would be a true annihilation, away from the soul. He dared to prove that a simple being cannot stop existing at all. His reason was this. A simple being can’t be diminished. So it can’t lose something of its existence little by little.

B 414 And so it can’t be turned gradually into nothing (it has no parts, and so no plurality in itself). Then no time at all would lie between the moment when it is and the next moment when it is no longer, and that is impossible.—But he didn’t consider this. Grant the soul this simple nature, in that it contains no manifold outside one another and so no extensive magnitude. Even so, nobody can deny it intensive magnitude, any more than they can deny it to anything else that exists. Intensive magnitude is a degree of reality with regard to all its faculties, indeed to everything that makes up its existence. That degree can sink through all the infinitely many lesser degrees. And so the supposed substance (the thing whose persistence isn’t already established otherwise) can be turned into nothing. It does not happen by division, but by a gradual slackening (remissio) of its forces (that is, by elanguescence, if I may use this expression). For consciousness too always has some degree, and that degree can always be lowered further*.

B 415 So can the faculty of self-awareness, and so can all the other faculties.—The soul’s persistence, then, as the mere object of inner sense, stays unproved and even unprovable. That it persists during life is obvious by itself, since the thinker, as a human being, is also an object of the outer senses. That doesn’t satisfy a rational psychologist, who sets out to prove, from bare concepts alone, that the soul persists absolutely, even beyond life∗∗.

* Clarity is not, as the logicians say, the consciousness of a representation. For a certain degree of consciousness, one that is not enough for recollection, must be found even in some obscure representations, because with no consciousness whatever we could tell nothing apart when obscure representations are combined. Yet we can do just that with the marks of some concepts (for instance those of right and fairness, or of a musician who plays many notes at once while improvising). Rather, a representation is clear when the consciousness in it is enough for consciousness of its difference from others. If it is enough for telling them apart, but not enough for being conscious of the difference, we would still have to call the representation obscure. Consciousness therefore has infinitely many degrees, all the way down to vanishing.

∗∗ People who think they’ve done enough to bring a new possibility forward, simply by insisting that no one can show them a contradiction in their assumptions, can be thrown into great embarrassment by other possibilities that are no bolder at all. (All who believe they can see the possibility of thinking continuing after life has ended, though they have an example of thinking only in empirical intuitions within human life, are in this position.) One of these is that a simple substance might be divided into several substances. The reverse is the possibility of several substances flowing together (coalition) into one simple one. For divisibility presupposes something composite. But it doesn’t necessarily require something composite of substances. It requires only a composite of degrees (of the soul’s various faculties) belonging to one and the same substance. We can think of all the forces and faculties of the soul, even that of consciousness, as dwindled by half, so that substance would still remain. In the same way we can picture this extinguished half, without contradiction, as preserved, not in the soul but outside it. Everything in the soul that is real has a degree, and so the soul’s whole existence, with nothing lacking, would have been halved here. Outside it, a substance of its own would then come into being. For the plurality that was divided was there beforehand. It was not a plurality of substances, though. It was a plurality of every reality, each being a quantum of existence in the soul. The unity of the substance was only one way of existing, and this division alone would turn it into a plurality of subsistence. In the same way, several simple substances could flow together again into one. The only loss would be the plurality of subsistence, because the single substance would hold within itself the degree of reality that all the earlier ones had between them. Perhaps the simple substances that appear to us as matter could produce child-souls by a dynamic division of parent-souls, understood as intensive magnitudes. (They would do so not by a mechanical or chemical action on one another, but by some action unknown to us, of which those would be only the appearance.) The parent-souls would make good their loss in turn, by coalition with fresh matter of the same sort. I am far from granting such figments of the brain the slightest value or validity. The principles of the Analytic above have also impressed on us enough that we should use the categories (the category of substance, for instance) only in experience. But suppose the rationalist is bold enough to turn the mere power of thinking into a self-subsisting being, with no persistent intuition by which an object would be given. The only reason is that the unity of apperception in thinking allows no explanation from something composite. It would be wiser for the rationalist to confess that they can’t explain how a thinking nature is possible. If they do it anyway, why shouldn’t the materialist be entitled to the same boldness, even though the materialist can no more cite experience for their own possibilities? Why shouldn’t the materialist use their own principle for the opposite purpose, while keeping the rationalist’s formal unity?

B 416 Let us now put our propositions above into synthetic connection, as rational psychology has to when it treats them as one system that holds good for every thinking being. Start from the category of relation.

B 417 The proposition is: all thinking beings, as such, are substances; then work backward through the series of categories until the circle closes. In the end we meet the existence of these beings. In this system they are conscious of this existence independently of outer things.

B 418 And they can determine it from themselves, as regards the persistence that necessarily goes with being a substance. It follows from this that idealism is unavoidable in this same rationalist system, at least problematic idealism. And suppose determining one’s own existence in time doesn’t require outer things to exist at all. Then their existence is assumed for nothing, and no proof of it can ever be given.

Suppose instead we follow the analytic procedure. Then “I think” lies at the base as given. It is a proposition that already includes an existence, and so modality. We take it apart to learn what it contains: whether this I settles its existence in space or in time, and how, simply by thinking. In that case the propositions of the rational doctrine of the soul would begin with an actuality. They would not begin with the concept of a thinking being at large. They would work from the way this actuality is thought, once everything empirical in it has been set aside.

B 419 From that they would infer whatever belongs to any thinking being, as the following table shows.

1 I think,

2 as subject,

3 as simple subject,

4 as identical subject, whatever state my thinking is in.

The second proposition doesn’t settle whether I can exist and be thought of solely as a subject, or also as the predicate of something else. So the concept of a subject is taken here as merely logical. Nothing is settled about whether substance is meant by it. In the third proposition, though, the simple I, apperception’s absolute unity, becomes important in its own right. It is the I in the representation to which every combination or separation that makes up thinking relates. This holds even though I haven’t yet settled anything about the constitution or subsistence of the subject. Apperception is something real, and its simplicity lies already in its possibility. Now nothing real in space is simple, since points (the only simple things in space) are merely boundaries. They are not themselves something that serves to make up space as a part.

B 420 So it follows from this that it is impossible to explain my constitution, as a merely thinking subject, from the grounds of materialism. In the first proposition, however, my existence is regarded as given. It doesn’t say that every thinking being exists (which would say absolute necessity of them, and so too much). It says only: I exist thinking. So that proposition is empirical. It contains the determinability of my existence only with regard to my representations in time. But for this I again need something persistent first, and as far as I think myself, nothing of the kind is given to me in inner intuition. So the way in which I exist, whether as substance or as accident, can’t possibly be determined by this simple self-consciousness. If materialism is unfit to explain my existence, then spiritualism is just as inadequate for it. The conclusion is that we cannot cognize anything at all, in any way whatever, about the make-up of our soul as it bears on whether it could exist separately at all.

How could we possibly get beyond experience (our existence in life) by means of the unity of consciousness? The only reason we know that unity is that we cannot do without it for experience to be possible. And how could we widen our cognition to cover the nature of every thinking being whatever?

B 421 We would have to do it by the proposition “I think,” which is empirical but undetermined with regard to every kind of intuition.

So rational psychology doesn’t exist as a doctrine that would add anything to our self-knowledge. It exists only as a discipline. It sets uncrossable boundaries for speculative reason in this field. One purpose is to keep reason from throwing itself into the lap of soulless materialism. The other is to keep it from losing itself, wandering aimlessly in a spiritualism that, for us in this life, has no ground. Instead the discipline reminds us to see our reason’s refusal to give a satisfying answer to curious questions that reach beyond this life as a hint. The hint is to turn our self-knowledge away from fruitless, excessive speculation and toward fruitful practical use. This practical use is always directed at objects of experience. Even so, it takes its principles from higher up, and it determines conduct as if what we are meant for reached infinitely far beyond experience, and so beyond this life.

All of this shows that rational psychology originates in a mere misunderstanding. The unity of consciousness that lies at the base of the categories is taken here for an intuition of the subject as an object.

B 422 Then substance, the category, is applied to it; but that unity is only the unity in thinking, and by it alone no object is given. So substance, a category that always presupposes a given intuition, can’t be applied to it. And so this subject can’t be cognized at all. The subject of the categories, then, cannot gain a concept of itself as something the categories apply to just by thinking them. For to think them it has to lay its pure self-consciousness at the base, and that is exactly what was supposed to be explained. In the same way, the subject that is the original ground of the representation of time can’t determine its own existence in time by this means. And if that can’t happen, then the other can’t happen either: the subject’s determining itself, as a thinking being at large, through categories*.

* “I think” is, as already said, an empirical proposition, and it contains the proposition “I exist.” But I can’t say: everything that thinks exists. For if I could, the property of thinking would turn every being that has it into a necessary being. So my existence can’t be treated as something inferred from “I think,” which is what Descartes believed (otherwise the major premise would have to come first: everything that thinks exists). Instead, my existence is identical with it. It expresses an undetermined empirical intuition, that is, a perception. (So it does prove that sensation, which therefore belongs to sensibility, lies at the base of this existential proposition.) But it comes before the experience that is meant to determine the object of perception, by means of the category, in respect of time. And existence here is not yet a category. A category refers not to an object given undetermined, but only to one of which one has a concept and about which one wants to learn whether or not it is also posited outside that concept. Here an undetermined perception means just something real that was given, given only to thinking in general. So it is not given as an appearance, and not as a thing in itself (noumenon) either. It is given as something that actually exists, and the proposition “I think” marks it as such. For note this: I called “I think” an empirical proposition. By that I didn’t mean that the I in it is an empirical representation. It is rather purely intellectual, since it belongs to thinking as such. But without some empirical representation that supplies thinking with its matter, the act “I think” could not occur. The empirical is merely what has to be present for the pure intellectual faculty to be applied or used.

* * *

B 423 So a cognition that was attempted beyond the boundaries of possible experience, and that yet belongs to the highest interest of humanity, disappears, as far as it is supposed to be owed to speculative philosophy.

B 424 It ends in disappointed expectation. Even so, the strictness of the Critique does reason a service in this interest of its own, and not an unimportant one. It proves at the same time the impossibility of settling anything dogmatically, beyond the boundary of experience, about an object of experience. That also secures reason against all possible assertions of the opposite. This can happen in no other way than this: either one proves one’s proposition apodictically or, if that fails, one looks for the sources of this inability. If they are rooted in our reason’s own necessary limits, they must hold every opponent to the identical rule: give up every claim to dogmatic assertion.

Even so, nothing at all is lost here as regards the right, indeed even the necessity, of assuming a future life, on the principles of the practical use of reason that is joined with the speculative. For the merely speculative proof could never have affected ordinary human reason anyway. It is balanced on the point of a hair. Even the schools can keep it balanced there only by making it spin like a top, without pause. By their own lights, then, it offers no lasting footing to build anything on. The proofs that are usable for the world all keep their undiminished worth here.

B 425 Rather, they gain in clarity and unforced conviction when those dogmatic presumptions are removed. They move reason into its own proper territory, namely the order of purposes, which is at the same time an order of nature. There reason is entitled, as a practical faculty in itself, to stretch the former order, and our own existence along with it, past the boundaries of experience and of life. It is not limited to the conditions of the latter. Consider by analogy the living creatures of this world. In them reason must assume, as a principle, that you will find no organ, no faculty, no drive that could be spared, or that is out of proportion to its use, and so unsuited to its purpose. Everything is exactly fitted to what it is meant for in life. Judging by that analogy, the human being would have to be the one creature exempted from it. Yet the human being alone can hold within themselves the final end of all this. For their natural predispositions, not merely the talents and impulses as far as making use of them goes, but above all the moral law within them, go so far beyond any benefit or advantage that this life could bring them that the moral law even teaches them to esteem above everything the bare consciousness of an upright disposition, when all advantages are lacking.

B 426 That holds even when the mere shadow of fame after death is lacking too. They also feel an inner call to qualify themselves, by how they behave in this world and at the cost of many advantages, as citizens of a better world, one they hold in the idea. This powerful ground of proof, which can never be refuted, remains in place. It is accompanied by a steadily growing cognition of purposiveness in all we see around us, and by a view into the immensity of creation. It is accompanied too by the sense that our knowledge could grow without any fixed end, along with a drive to match. All of this remains even though we must give up seeing, from merely theoretical cognition of ourselves, that our existence necessarily continues.

Conclusion of the Resolution of the Psychological Paralogism

The dialectical illusion in rational psychology comes from a mix-up. An idea of reason (a pure intelligence) gets confused with the concept of a thinking being in general, a concept that is undetermined in every respect. I think of myself for the sake of a possible experience, abstracting from all actual experience while I do. From this I conclude that I could become aware of my existence even outside experience, apart from its empirical conditions.

B 427 In doing so I mistake two things for each other. The first is the possibility of abstracting from my existence as experience determines it. The second is an alleged awareness that my thinking self could exist on its own, set apart from all that. And I take myself to cognize the substantial in me, the transcendental subject, when all I have in thought is the unity of consciousness, which underlies all determining as the mere form of cognition.

Explaining how the soul and the body are in community isn’t really a job for the psychology we’re discussing here. That psychology means to prove that the soul stays a person even without this community (after death). In the strict sense, then, it is transcendent. It does deal with something from experience, but only as that thing ceases to be something we can experience. Still, our own doctrine can answer this question adequately too. The difficulty, as is well known, comes from an assumed difference in kind. The soul is what inner sense takes as its object, and the things the outer senses take as theirs are supposed to be of another sort. For the soul, only time is the formal condition of its intuition. For the others, space is one too. But the two sorts of things don’t differ from one another inwardly in this; they differ only in that one appears outside the other.

B 428 So the thing in itself that lies at the basis of matter’s appearance may well not be so different in kind. Then the difficulty vanishes. What is left is only the question how a community of substances is possible at all. Solving that lies completely outside the field of psychology. And, as the reader will readily see from the Analytic’s remarks on fundamental powers and faculties, it surely lies outside the field of all human cognition, too.

General remark on the move from rational psychology to cosmology

Take the sentence “I think,” or “I exist thinking.” It is an empirical sentence. Such a sentence rests on empirical intuition, and so the object thought in it counts as an appearance. That makes it seem as if our theory turned the soul completely into appearance, even in its thinking. In that case our own consciousness would be mere illusion, and it would really be of nothing.

Thinking, taken by itself, is only the logical function. It is pure spontaneity in combining the manifold of a merely possible intuition. It does not present the subject of consciousness as an appearance.

B 429 The reason is simply that thinking pays no attention to what kind of intuition there is, whether sensible or intellectual. Through thinking, then, what I represent is not myself as I am, and not myself as I appear either. I only think of myself as I would think of any object whatever, with its kind of intuition left aside. When I take myself here to be a subject of thoughts, or even the ground of thinking, these ways of representing do not stand for the categories of substance or of cause. Those categories are the functions of thinking (of judging) once they are applied to our sensible intuition, and they would be needed if I wanted to cognize myself. But all I want is to become aware of myself as thinking. In what way my own self is given to intuition, I leave aside. On that question I, the one who thinks, could be mere appearance to myself, though not insofar as I think. In my consciousness of myself while merely thinking, I am the being itself. But of course nothing about this being is thereby given to me to think.

But the statement “I think,” insofar as it says the same as “I exist thinking,” is not just a logical function. It determines the subject (which is an object as well) in respect of existence. And it can’t occur without inner sense, whose intuition always supplies its object as appearance only, never as a thing in itself.

B 430 So this statement already holds more than the bare spontaneity of thinking; it also holds the receptivity of intuition, that is, my thinking about myself brought to bear on the empirical intuition of this very same subject. In that empirical intuition the thinking self would then have to look for the conditions under which its logical functions can be put to work as the categories substance, cause, and the rest. Only then could it do more than mark itself with the word “I” as an object in itself. Only then could it also determine how it exists, which means cognizing itself as a noumenon. That can’t be done, since inner empirical intuition is sensible and supplies only data of appearance. These data give the object of pure consciousness nothing toward knowing its separate existence. They are good only for serving experience.

Suppose, though, that we later found grounds, not in experience but in certain laws of the pure use of reason that concern our existence and stand fixed a priori (laws, and not merely logical rules), for assuming, completely a priori, that we legislate for our own being and also determine that existence ourselves. Then we would discover a spontaneity by which our reality could be determined without any need for the conditions of empirical intuition. And we would notice that something a priori is already contained in our consciousness of being here. This something would serve to fix our existence, which is determinable all the way through only by way of the senses.

B 431 It would do so with regard to a certain inner faculty, with a view to an intelligible world, though to be sure one we can only think.

Even so, this would not move any attempt of rational psychology forward in the least. That admirable faculty, which first discloses to me the consciousness of the moral law, would indeed give me a principle for determining my existence, and a purely intellectual one. But through which predicates? Only those that have to be given to me in sensible intuition. So I would end up back where I stood in rational psychology. Once more I would need sensible intuitions, in order to give meaning to the concepts of the understanding I use (substance, cause and so on), the only concepts through which I can have cognition of myself. And those intuitions can never help me out beyond the field of experience. In practical use, though, which always aims at objects of experience, I would still have the right to bring these concepts to bear on freedom and on the subject that has it. The meaning would be analogous to the one they have in theoretical use. For all I mean by them is the logical functions of subject and predicate, of ground and consequence.

B 432 In line with these functions, actions or effects are determined, according to those laws, in such a way that they can at the same time always be explained in agreement with the laws of nature, that is, with the categories of substance and cause, even though they come from a wholly different principle. All this has been said only to prevent a misunderstanding, to which the teaching about our intuition of ourselves as appearance is easily exposed. We will have a chance to put it to use later.

The Transcendental Dialectic Book Two

Chapter Two The Antinomy of Pure Reason

We showed, in the introduction to this division of the book, that the transcendental illusion of pure reason, all of it, is founded on dialectical inferences. Logic provides their pattern in the three formal kinds of inference by reason, just as the categories find their logical pattern in the four functions of all judgments. The first kind of these sophistical inferences went after unconditioned unity among subjective conditions, those of every representation whatever (that is, those of the subject or soul). It corresponds to categorical inferences, whose major premise, as a principle, asserts that a predicate relates to a subject.

B 433 By analogy with hypothetical inferences, then, the second kind of dialectical argument is going to deal with an unconditioned unity again, but among objective conditions, in appearance. The third kind, which comes in the next chapter, deals with unconditioned unity among the objective conditions that make objects as such possible.

It is remarkable, though, that the transcendental paralogism produced only a one-sided illusion, when it comes to the idea we have of our thinking subject. For the opposite claim, not the slightest illusion from concepts of reason can be found. The advantage lies wholly with pneumatism. Yet pneumatism can’t deny its inherited flaw. For all the illusion that favors it, it dissolves into pure vapor in the fire test of critique.

Things turn out very differently once reason is applied to the objective synthesis of appearances. There reason thinks it can make its principle of unconditioned unity count, and with a great deal of illusion. But it soon gets tangled in such contradictions that, for cosmological purposes, it has to give up its demand.

Here a new phenomenon of human reason shows itself: a completely natural antithetic. No one has to rack their brains over it or lay artful traps to find it.

B 434 Reason runs into it on its own, and unavoidably. That keeps reason safe from the slumber of an imagined conviction, which a merely one-sided illusion produces. But it also tempts reason either to hand itself over to skeptical hopelessness, or to take up a dogmatic defiance and stubbornly dig in on certain claims, without giving the grounds of the opposite side a hearing or a fair judgment. Either one is the death of a healthy philosophy, though the former might at most be called pure reason’s euthanasia.

Before we display the scenes of discord and confusion that this conflict of laws (antinomy) brings on pure reason, we want to give some explanations. They are meant to clarify and justify the method we use in handling our subject. I call a transcendental idea a world concept when it has to do with absolute totality in how appearances are synthesized. There are two reasons for the name. One is this very unconditioned totality. The concept of the world as a whole also rests on it, and that concept is itself only an idea. The other is that these ideas go only to the synthesis of appearances, and so only to the empirical. Absolute totality is a different matter when it is found in synthesizing the conditions of all possible things whatever.

B 435 That will lead to an ideal of pure reason, which is wholly distinct from the world concept, although it refers back to it. The paralogisms of pure reason laid the groundwork for a dialectical psychology. In the same way, the antinomy will put before our eyes the transcendental principles of a so-called pure (rational) cosmology. It will not do this to find them valid and appropriate them. As the very name for a conflict of reason already indicates, it will do it to show them for what they are: an idea that can’t be reconciled with appearances, shining with a dazzling but false illusion.

Pure Reason’s Antinomy First Section The Cosmological Ideas as a System

To list these ideas systematically, by one principle and with precision, we first have to note a few things. Pure, transcendental concepts can only spring from the understanding. Reason doesn’t really produce any concept. At most it takes a concept of the understanding and frees it from the unavoidable restrictions of possible experience, so that it reaches past the boundaries of the empirical while still staying connected with the empirical.

B 436 Reason does this as follows. For any conditioned thing that is given, it asks that the conditions be gathered up in an absolute totality. (The understanding brings all appearances under synthetic unity by way of these conditions.) That turns the category into a transcendental idea. Reason wants to make the empirical synthesis absolutely complete by continuing it until it reaches the unconditioned, and the unconditioned is met with only in the idea, never in experience. Its principle is this: if a conditioned thing is given, the entire sum of its conditions is given as well, and so is the absolutely unconditioned, which alone made that thing possible. First, then, the transcendental ideas will turn out to be nothing other than categories stretched out to the unconditioned, and they can be put in a table arranged under the headings of the categories. Second, not every category will suit this. Only those will do in which the synthesis forms a series, a series of conditions for a conditioned thing that are subordinate to one another (not coordinate). Reason asks for absolute totality only on the way up: the rising series of conditions that leads to a conditioned thing already given. It does not ask for it where the talk is of the line of consequences running down, or of the aggregate of coordinate conditions for those consequences.

B 437 For conditions, in relation to the conditioned thing that is given, are already presupposed, and so are to be regarded as given along with it. Consequences do not make their conditions possible; they presuppose them. So as we go forward to consequences (or down from a given condition to what it conditions), we can be unconcerned whether the series ends. Whether that series is a totality is nothing reason has to assume.

In the same way we necessarily think of the time that has fully run its course up to the present moment as given, too (though we can’t determine it). As for future time, it is not the condition of our reaching the present. So, for grasping the present, it makes no difference how we choose to handle the future, whether to end it somewhere or to let it run on without end. Suppose a series m, n, o. In it n is given as conditioned by m and at the same time as the condition for o. The series goes upward, from n as the conditioned one back to m (l, k, i, and so on), and also downward, from n as condition to o as the conditioned (p, q, r, and so on). To regard n as given, I have to presuppose the first series. Reason’s standard, the totality of conditions, lets n be possible only by means of that first series. But its possibility does not depend on the later series, o, p, q, r.

B 438 That series therefore can’t count as given either, only as dabilis (something that can be given).

I’ll use the name regressive synthesis for the synthesis of a series on the conditions’ side. It begins with the condition closest to the appearance we’re given and goes on to conditions farther away. I’ll use progressive synthesis for the one on the side of the conditioned: it runs from the nearest consequence out to those farther along. The regressive one heads in antecedentia (toward what comes before), the progressive one in consequentia (toward what follows). The cosmological ideas, then, are about the regressive synthesis taken as a whole, and they head in antecedentia, never in consequentia. Heading the other way would be a problem reason might take up if it liked, but doesn’t have to. For a complete grasp of what appearance gives us, we need grounds. We have no need of consequences.

To build the table of ideas on the pattern of the table of categories, we start with the two original quanta in all of our intuition: time and space. Time is a series by its own nature, and it is the formal condition for every series. So once a present moment is given, we can tell a priori the antecedentia, which serve as conditions (the past), from the consequentibus (the future). The transcendental idea asks for a series of conditions to be complete.

B 439 Where a conditioned thing is given, that idea reaches only as far as the whole of past time. The idea of reason requires us to think of all the time that has gone by as given, because it is what conditions the moment we are in. Space is different. Space has no built-in difference between progressus and regressus, since it is an aggregate and not a series: all its parts exist together. I could treat the present instant, relative to the past, only as conditioned and never as a condition of it, because the instant arises only out of the time that has passed (or rather out of the passing of earlier time). The parts of space, though, are not ranked under one another. They stand side by side. So no part conditions the possibility of another, and space is not a series the way time is. Even so, we take in the manifold parts of space by a synthesis that goes step by step. That happens in time, and it contains a series. Look at a series of spaces added together (the feet in a rod, say). From any given one on, each further space we think of adding is always the condition for the boundary of the one before it. Measuring a space is therefore also a synthesis of a series of conditions, leading to something conditioned that is given. The only difference is that nothing in space itself separates the conditions’ side from the side where the conditioned thing lies.

B 440 For that reason, regressus and progressus look like one and the same thing in space. But one part of space is not handed to us by another part; the other part only bounds it. So we have to regard any bounded space as conditioned in this way too, since it assumes a second space to condition its boundary, and that one a third, and so on. Where bounding is concerned, then, moving on in space is also a regressus. The transcendental idea of absolute totality in the synthesis of a series of conditions therefore reaches space too. I may ask for a complete totality of appearance in space, just as I may for the time that has passed. Whether there can be an answer at all will be decided later.

Second, matter, which is reality in space, has conditions too. Its parts are its inner conditions, and the parts of those parts are conditions farther out. So we find a regressive synthesis here, and reason wants it carried to absolute totality. The only way to do that is complete division, which makes the reality of matter disappear: either into nothing, or into something that is no longer matter, the simple. So this too gives us a series of conditions and a way forward to the unconditioned.

B 441 Third, consider the categories of real relation among appearances. The category of substance, along with its accidents, doesn’t fit a transcendental idea. In other words, reason has no ground for going back to conditions where substance is concerned. Accidents that belong to one and the same substance are on the same level as each other, and they don’t form a series. With respect to the substance, they aren’t strictly subordinated to it either; they are simply its way of existing. One thing that might still seem an idea of transcendental reason is the concept of the substantial. But this concept means just the concept of an object that subsists, when all that is thought in it is the transcendental subject with no predicates. Here the topic is only the unconditioned in the series of appearances. So the substantial clearly can’t be one of the links in that series. The same goes for substances in community. They are mere aggregates, and there is no exponent of a series in them. They aren’t ranked under one another as conditions that make each other possible. One could well say that of spaces, since the boundary of a space was never fixed by the space itself, only by another space. So only the category of causality remains. It supplies a series of causes behind a given effect.

B 442 In that series one can climb from the effect, which is the conditioned thing, to the causes, which are its conditions, and so answer the question reason puts.

As for the fourth heading, the possible, the actual and the necessary give us a series in only one way: contingent existence always has to be seen as conditioned. The rule of the understanding sends it back to a condition, and under that condition it is necessary to refer this condition in turn to a still higher one. That goes on until reason meets unconditioned necessity, which it meets only in the totality of the series.

That gives us only four cosmological ideas, matching the four headings of the categories, once we select the categories that bring a series along with them in the synthesis of the manifold.

B 443 1. Absolute completeness in putting together the given whole made up of all appearances

2 Absolute completeness in dividing up a given whole within appearance

3 Absolute completeness in how an appearance comes to be

4 Absolute completeness in the dependence of existence, for whatever changes in appearance

A first point to note: the idea of absolute totality has to do with the setting out of appearances and nothing else. It is not about the pure concept of the understanding for a whole of things in general. So appearances are being treated as given, and what reason demands is that the conditions for their being possible be complete, as far as they form a series. That means a synthesis complete without qualification (that is, in every respect), through which an appearance could be set out according to laws of the understanding.

Second, what reason is really after is just the unconditioned.

B 444 It looks for it in this synthesis of conditions, which it carries on in a series, step by step, backward. The unconditioned is, so to speak, completeness in the line of premises, where all the premises together call for no further one. Now the unconditioned is always included in the absolute totality of such a series, if we form a picture of that totality in the imagination. Yet a synthesis finished in this complete sense is itself only an idea. At least ahead of time, we can’t know whether a synthesis like that can exist for appearances at all. Suppose we picture everything by means of bare pure concepts of the understanding, without the conditions of sensible intuition. Then we could simply say that when something conditioned is given, the full series of conditions ranked under one another is given with it, since only through the conditions is the conditioned thing given. With appearances, however, there is a particular restriction on how conditions are given. They are given by a synthesis of the manifold of intuition that proceeds in succession, and that synthesis is meant to be complete as it goes backward. Whether such completeness is possible for the senses remains a problem. All the same, the idea of that completeness is there in reason, whether or not empirical concepts that match it can be connected with it. Take the synthesis of the manifold in appearance, carried out as a regress, and view it as an absolute totality, following the lead of the categories, which present it as a series of conditions behind a given conditioned thing.

B 445 That totality necessarily contains the unconditioned. We can leave open whether, and how, that totality could ever be achieved. So reason goes about it by setting out from the idea of totality, even though what it ultimately aims at is the unconditioned, either of the entire series or of some part of it.

There are two ways to think of this unconditioned. In one, it exists only in the series as a whole: every member of the series is conditioned, without exception, and only the series taken together is unconditioned. In that case the regressus is called infinite. In the other, the absolutely unconditioned is a part of the series, which the remaining members are subordinate to, while it is not itself under any other condition.* On the first view, the series has no boundary a parte priori (looking backward), and so no beginning. That makes it infinite, and yet the whole of it is given. The regressus through it never gets finished, though, and can be called infinite only potentialiter (potentially).

B 446 On the second view, the series has a first member. Reckoned in elapsed time, this first member is called the beginning of the world. Reckoned in space, it is the world’s boundary. Reckoned among the parts of a whole that is given inside its boundaries, it is the simple. Among causes it is absolute self-activity (freedom), and in the existence of things that change, it is absolute natural necessity.

* A series of conditions for a given conditioned thing, taken as an absolute whole, is unconditioned every time, since there are no conditions outside it that could condition it. This absolute whole of such a series, however, is only an idea, or really a problematic concept. We need to investigate whether it is possible, specifically in regard to how the unconditioned, the true transcendental idea we are concerned with, might be contained in it.

Two terms, “world” and “nature,” sometimes get run together. “World” means the mathematical whole formed by all appearances, with the totality of their synthesis, whether large or small, that is, whether the synthesis advances by composing or by dividing. The very same world is called “nature”* once it is viewed as a dynamical whole. Then the thing we look at is not an aggregate in space or time, put together to make a quantity.

B 447 It is the unity in the existence of appearances. In this view, whatever conditions an event is called its cause. The cause’s unconditioned causality in appearance is called freedom; its conditioned causality, in the narrower sense, is called natural cause. In existence generally, the conditioned is called contingent and the unconditioned necessary. The unconditioned necessity found in appearances may be named natural necessity.

* “Nature” taken adjectivally (formaliter) means how the determinations of a thing hang together by an inner principle of causality. Taken as a noun (substantively, materialiter), it means the sum of appearances insofar as they are linked all the way through by an inner principle of causality. In the first meaning we talk about the nature of liquid matter, of fire, and so on, and use the word as an adjective. But when we talk about things of nature, we have in mind an existing whole.

The ideas we’re dealing with now are the ones I called cosmological ideas earlier. I had two reasons. First, “world” is understood as the sum of all appearances, and our ideas aim only at what is unconditioned among appearances. Second, in the transcendental sense, “world” means the absolute totality of all existing things taken together, and we are directing our attention only to the completeness of the synthesis (though strictly only in going back to the conditions). Besides, all these ideas are transcendent. They don’t overstep their object, appearances, as to its kind, since they concern the world of sense alone (not noumena). They still drive the synthesis to a degree beyond all possible experience. For these reasons I think they can all quite fittingly be called world concepts.

B 448 In view of the difference between the mathematically unconditioned and the dynamically unconditioned, at which the regressus aims, I would use “world concepts” in the narrower meaning for the first two (they deal with the world on the large scale and the small), and I would call the other two transcendent concepts of nature. For now this distinction is not especially important, but it may become more so as we go further.

Pure Reason’s Antinomy Second Section The Antithetic of Pure Reason

If a thetic is any collection of dogmatic teachings, then by antithetic I mean something else: not assertions of the opposite that are themselves dogmatic, but the clash among cognitions that seem to be dogmatic (thesin cum antithesi, a thesis against an antithesis), with neither side granted a better claim to our assent. The antithetic is never about one-sided assertions. It looks at general cognitions of reason solely for how they clash with each other and why. The transcendental antithetic is an inquiry into the antinomy of pure reason, covering its causes and its result. Suppose we do more than use reason, along with the principles of the understanding, on objects of experience.

B 449 Suppose we venture to stretch it beyond the boundary of experience. Out come sophistical doctrines. These can’t expect experience to confirm them, and they needn’t fear that it will refute them. Each is free of contradiction within itself, and it even meets the conditions of its necessity in the very nature of reason. The only trouble, and an unlucky one, is that its opposite has just as valid and necessary grounds for its own claim.

The questions that present themselves naturally with a dialectic like this one are the following. 1. Which propositions, exactly, is pure reason bound to fall into an antinomy over? 2. What causes lie behind the antinomy? 3. Does a way to certainty remain open to reason, despite this contradiction? If so, in what manner?

So a dialectical proposition of pure reason has to carry the following mark, which separates it from every sophistical proposition. First, it does not concern an arbitrary question, one raised only for some purpose or other. It concerns a question that any human reason, as it advances, is bound to meet. Second, it comes with its opposite, and together they carry not a contrived illusion that vanishes the moment someone sees through it, but a natural and unavoidable illusion.

B 450 Even when you are no longer led astray by it, this illusion still deceives, though it does not trick you. So it can be made harmless, but never wiped out.

A dialectical doctrine like this will not deal with the understanding’s unity in empirical concepts. It will deal with reason’s unity in mere ideas. The conditions of that unity have to fit two things at once. As a synthesis that follows rules, they have to fit the understanding, and yet as the absolute unity of that synthesis, they have to fit reason. If they match the unity of reason, they will be too big for the understanding. If they suit the understanding, they will be too small for reason. A conflict has to come out of this, and no one can avoid it, however they go about it.

These sophistical claims therefore open up a dialectical battlefield. There the side that is allowed to attack keeps the upper hand, and the side that is forced to merely defend is bound to lose. So sturdy knights, whether they stand surety for the good cause or the bad one, are sure to carry off the victor’s wreath, provided they make sure they have the privilege of the last attack and needn’t withstand a fresh assault from the other side. You can easily picture how often this field has been trodden since long ago, and how many victories each side has won.

B 451 But as for the last victory, the one that settled the question, it was always arranged that the defender of the good cause held the field alone. That was done by forbidding the opponent to take up arms any longer. As impartial referees, we have to set aside entirely whether the cause they are fighting over is the good one or the bad one. We should let them settle it first between themselves. Perhaps, after they have tired each other out more than they have hurt each other, they will see for themselves how worthless their quarrel is and part as good friends.

This method of watching a quarrel between assertions, or rather of starting one ourselves, is not meant to decide in the end for one side or the other. Its aim is to investigate whether what they are fighting over is perhaps a mere mirage, one that each side grabs at in vain and could gain nothing from even if no one stood in its way. This procedure, I say, can be called the skeptical method. It is completely different from skepticism, which is a principle of ignorance practiced as an art and a science. Skepticism undermines the foundations of all cognition, so that, if possible, no reliability or certainty is left anywhere. The skeptical method aims at certainty. The dispute it looks at is one that both sides mean honestly and carry on with intelligence.

B 452 In it, the method tries to find the point of misunderstanding. Wise lawgivers do the same: they draw a lesson for themselves, about what is defective or imprecise in their laws, from the judges’ perplexity in lawsuits. The antinomy that shows itself when laws are applied is, given our limited wisdom, the best way to test the lawmaking. It makes reason, which doesn’t easily notice its own missteps in abstract speculation, aware of the moments that matter in fixing its principles.

This skeptical method belongs essentially to transcendental philosophy alone. In any other field of inquiry it could be done without, though not in this one. In mathematics, using it would be absurd. No false claims can hide there or make themselves invisible, because proofs there always have to follow the guiding thread of pure intuition, one evident synthesis after another. In experimental philosophy, a doubt that holds back judgment can be useful. But at least no misunderstanding is possible that couldn’t easily be removed, and the last means for settling the dispute must lie in experience, whether they are found sooner or later.

B 453 Morality can show every one of its principles in concreto too, together with what follows from them in practice, at least within possible experiences, and so it can avoid the misunderstanding that comes from abstraction. Transcendental claims are different. They presume to insights that reach beyond the field of all possible experience. Their abstract synthesis can’t be given in any intuition a priori, and they aren’t the kind of claim whose misunderstanding could be found out through some experience. Transcendental reason therefore allows no other touchstone than the attempt to unite its claims with one another. For that, it needs first of all a free and unhindered contest among them. This is what we now want to stage.*

* The antinomies come one after another in the sequence of the transcendental ideas given earlier.

Pure Reason’s Antinomy

B 454 First Conflict among the Transcendental Ideas

Thesis

In time, the world has a beginning. In space, it is also enclosed within boundaries.

Proof

Take it that the world has no start in time. Then by any moment you pick, an eternity has run out, and an unending series of states of the world’s things, each following the last, has gone by. But what makes a series unending is exactly this: successive synthesis can never finish it. An unending series of world-states that has already elapsed is therefore impossible. So a beginning of the world is a condition its existence necessarily requires. That is what had to be proved first.

For the second claim, again assume the opposite. The world would then be an infinite, given whole of things that exist at the same time. Now, the magnitude of a quantum that isn’t given within certain boundaries of any intuition* can be thought in only one way.

B 456 That way is the synthesis of its parts. And the totality of such a quantum can be thought only through the completed synthesis, or through adding the unit to itself over and over.∗∗ Now think of the world, which fills every space, as a single whole. For that, we would have to treat the step-by-step synthesis of an infinite world’s parts as finished. That means an infinite time would have to count as run out while we go through all the things that exist side by side. This can’t be. So an infinite aggregate of actual things can’t be regarded as a given whole, and so not as given all at once either. Therefore a world is not infinite in its extent in space. It is enclosed within boundaries. That was the second claim.

* An indeterminate quantum that is enclosed within boundaries can be intuited as a whole. We don’t have to build up its totality by measuring, that is, by the step-by-step synthesis of its parts. The boundaries already fix its completeness, because they cut off whatever more there might be.

∗∗ Here the concept of totality simply stands for the finished synthesis of the parts. We can’t get the concept out of an intuition of the whole, since that is impossible in this case. The only way to reach it is by synthesizing the parts, all the way to the end of the infinite, in the idea at least.

B 455 Antithesis

There is no beginning to the world and no boundary to it in space. It is infinite in respect of time, and equally so in respect of space.

Proof

Suppose, for the sake of argument, that it does have a beginning. A beginning is a kind of existence. It is preceded by a time when the thing was not there. So a time must have gone before in which the world was not, that is, an empty time. But nothing can come to be in an empty time. No part of such a time has, over another part, any distinguishing condition for existing rather than not existing (whether we assume the thing arises of itself or through some other cause). So within the world many a series of things can begin, but the world itself can’t have a beginning. In regard to past time it is therefore infinite.

As for the second claim, first assume the opposite: the world is finite in space and has boundaries. It would then lie in an empty space with no boundary. So we would find not only a relation of things within space, but also a relation of things to space. Now the world is an absolute whole.

B 457 Outside it there is no object of intuition at all, and so no correlate of the world for it to be related to. If the world stood in a relation to empty space, it would stand in a relation to no object whatever. But a relation like that is nothing, and so is any bounding of the world by empty space. So the world is not bounded in space at all. That is, it is infinite in extent.*

* Space is nothing but the form of outer intuition (formal intuition); it is not a real object that outer intuition could take in. Before any things that determine it (by filling or bounding it), or rather that supply an empirical intuition fitting its form, space is what people call absolute space. Under that name it is merely the possibility of outer appearances, whether they exist in themselves or could still be added to appearances already given. Empirical intuition is not, then, put together out of appearances and space (that is, out of perception and empty intuition). Neither one is the other’s correlate in a synthesis. They are united only inside a single empirical intuition, as its matter and its form. Someone who wants to set one of the two apart from the other (space apart from all appearances) ends up with all kinds of empty determinations of outer intuition, none of which could be perceived. For example, take the world’s moving or resting in infinite empty space. That determines how the two stand to each other, and it can never be perceived. It is therefore also just the predicate of a thing of thought.

B 458 Remark on the First Antinomy I. On the Thesis

In these conflicting arguments I haven’t looked for deceptive tricks, to put on what’s called a lawyer’s proof. A lawyer’s proof takes advantage of the opponent’s carelessness, and gladly lets the opponent lean on a law that has been misunderstood, so that it can then base its own unjust claims on refuting that law. All of these proofs came out of the nature of the case itself. I have set aside the advantage that the fallacies of the dogmatists on both sides could have given us.

I could also have “proved” the thesis, so it would seem, by first laying down, as dogmatists tend to, a mistaken notion of what it is for a given magnitude to be infinite. A magnitude is infinite if none larger than it is possible, larger, that is, than the amount of a given unit it holds. But no amount is the largest, since a unit or more can always be added to any. So no given magnitude can be infinite, and neither can a world that is infinite, whether in its elapsed series or in its extent. It follows that the world is bounded on both sides. That is how I could have conducted my proof. But the concept doesn’t match what is meant by an infinite whole. It doesn’t tell us how big the whole is, so its concept isn’t the concept of a maximum either.

B 460 What is thought through it is only the whole’s relation to a unit we may take as we please, and relative to that unit the whole exceeds every number. If the unit is taken bigger or smaller, the infinite would be bigger or smaller too. Yet the infinity, because it consists only in the relation to this given unit, would remain the same. Of course the absolute size of the whole would not be known from this at all, and that isn’t the point here.

The true (transcendental) concept of the infinite is as follows: when a quantum is measured through, the successive synthesis of the unit can never reach its end.* It follows with complete certainty that no eternity of actual successive states can have run out by a given moment (the present one), so the world must have a beginning.

* In this way it holds a multitude (of a given unit) larger than any number. That is the mathematical concept of the infinite.

The second part of the thesis is free of the difficulty about a series that is infinite and yet already run through. The manifold of a world infinite in extent is given all at once. Even so, there is a problem in thinking the totality of such a multitude. We can’t point to boundaries that would form this totality by themselves in intuition. So we have to account for our concept. In a case like this the concept can’t start from the whole and move to a definite number of parts. It has to establish the possibility of a whole by synthesizing the parts one after another. That synthesis would have to form a series that can never be finished. So one cannot think a totality ahead of the synthesis, and therefore cannot think one by means of it. The concept of totality is itself, here, the thought of a synthesis of the parts brought to completion. Such completion, and with it the concept of it, is impossible.

B 459 II. Remark on the Antithesis

The proof that the world’s given series, and the world as a sum, are infinite rests on this: if they were not, an empty time and an empty space would have to be what bounds the world. I know that people look for ways out of this consequence. They claim that the world can very well have a boundary in time and in space without our needing to assume an absolute time before the world began, or an absolute space spread out beyond the real world, which is impossible. I am quite satisfied with the second half of this view, held by philosophers of the Leibnizian school. Space is only the form of outer intuition. It is no actual object to be intuited outwardly, and it is not a correlate of appearances. It is the very form of appearances. So space cannot, taken absolutely and by itself, figure as something that determines the existence of things. It is no object, just the form of possible objects. Things, as appearances, do determine space. Among all of its possible predicates (size and relation), they decide which belong to what is real. But space, as something that exists by itself, cannot determine the reality of things as to their size or shape, since in itself it is nothing real. So it is quite possible for a space (full or empty)* to be bounded by appearances.

B 461 But no empty space outside them can bound appearances. The same is true of time. Once all this is granted, it remains beyond dispute that anyone who assumes a boundary of the world, in space or in time, has to assume these two non-things as well: the empty space beyond the world and the empty time that precedes it.

* It’s easy to see what this is meant to say. Empty space that appearances bound, that is, empty space inside the world, at least doesn’t contradict the transcendental principles. As far as those principles go, then, it can be granted (though that doesn’t mean its possibility is straightaway asserted).

People try to get around the consequence we draw. We say that a world with boundaries in time and space would have the infinite void fixing the size of the real things that exist. The way they do it, though they don’t say so, is this. They put some intelligible world, who knows which, where a world of sense should be. Where the first beginning should be (an existence with a time of non-being before it), they put an existence of some sort that needs no other condition in the world. Where a boundary of extent should be, they put limits of the whole world. This lets them step around time and space. Our topic here, though, is only the mundus phaenomenon (the world as appearance), and how big it is. You can’t set aside the conditions of sensibility that we think into it without doing away with what it is. A world of sense that is bounded lies, necessarily, within the infinite void. Drop that void and, with it, space in general as the a priori condition that makes appearances possible, and the whole world of sense is gone. In our problem it is the only world we are given. The mundus intelligibilis (the intelligible world) is just the general concept of a world, in which we abstract from all the conditions of its intuition. So no synthetic proposition about it is possible, either to affirm or to deny.

B 462 Pure Reason’s Antinomy Second Conflict among the Transcendental Ideas

Thesis

Each composite substance in the world is made of simple parts. Nothing exists anywhere except the simple and whatever is put together out of it.

Proof

Let’s suppose that composite substances are not built from simple parts. Imagine all composition being removed in thought. No composite part would be left. There would be no simple part left either, since by assumption there are none, so nothing at all would be left, and that means no substance would have been given. So either it can’t be done at all, to remove all composition in thought, or something that exists without any composition has to remain after it is removed. That something is the simple. If the first is the case, then the composite again would not be made up of substances. (In substances, composition is just a contingent relation among them, and without that relation they have to continue to exist as beings that persist on their own.)

B 464 But that contradicts the assumption. So only the second option is left: whatever is composite in the world, and substantial, is built from simple parts.

It follows directly that the things of the world are all simple beings, that composition is only an external state of them, and that reason has to think of the elementary substances as the subjects that come first in every composition, and so as simple beings before any composing. This holds even if we can never take those substances entirely out of this state of combination and isolate them.

B 463 Antithesis

Nothing composite in the world is made of simple parts. And nothing simple exists anywhere in the world.

Proof

Suppose a composite thing (as a substance) is made up of simple parts. All outer relation, and so all composition out of substances, is possible only in space. That means the space a composite occupies must have as many parts as the composite does. Now space is not made of simple parts, but of spaces. So every portion of what is composite takes up some space. But the very first parts of anything composite are simple. So the simple occupies a space. Whatever is real and occupies a space includes a manifold whose members are outside one another. So it is composite, and since it is a real composite, it is composed not of accidents (they can’t be outside one another without a substance) but of substances. That would make the simple a substantial composite, and that is a contradiction.

In the antithesis, the second proposition is that nothing simple exists at all in the world.

B 465 It means only the following. No experience or perception, outer or inner, can establish that the absolutely simple exists. That makes the absolutely simple nothing but an idea. Its objective reality can’t be established in any possible experience, so in setting out appearances it has no use and no object. Consider this. Say that experience did offer an object matching this transcendental idea. We would then have to recognize, in the empirical intuition of some object, one that contains no manifold at all in which the members are outside one another and joined into a unity. But we can’t infer from not being aware of such a manifold that it is wholly impossible in any intuition of an object. And that impossibility is exactly what absolute simplicity requires. So absolute simplicity can’t be concluded from any perception whatsoever. Something absolutely simple can never be given as an object in any possible experience. The world of sense, though, has to be taken as the sum of all possible experiences. So nothing simple is given anywhere in it.

The second proposition of the antithesis reaches far beyond the first. The first only keeps the simple out of the intuition of the composite. The second wipes it out of nature as a whole. For that reason its proof couldn’t start from the concept of a composite, an object given in outer intuition. It had to come from how such an object relates to possible experience in general.

B 466 Note on the Second Antinomy I. Concerning the Thesis

By a whole that must consist of simple parts, I mean a substantial whole only: the true compositum. A compositum is a contingent unity of a manifold. The manifold is given as separate (in thought, at least), is put in mutual combination, and so makes one thing. Space, strictly speaking, shouldn’t be called a compositum at all. It should be called a totum, because its parts are possible only inside the whole, and the whole isn’t possible through the parts. You might call it a compositum ideale, but never a reale. That, however, is mere hairsplitting. Space is not composed of substances (nor even of real accidents). So when I remove all composition from it, nothing remains, not even a point. A point is only possible as a space’s boundary, and a space is something composite.

B 468 Space and time, then, are not made of simple parts. Something that only belongs to the state of a substance, though it has a magnitude (change, for instance), is not made of the simple either. A given degree of change does not arise by many simple changes piling up. Our inference from composite to simple is valid only for things that exist on their own. Accidents of a state don’t exist on their own. It is easy, then, to spoil the proof that the simple is necessary as the component of everything substantially composite, and with it one’s whole position. One does this by overextending the proof and trying to make it count for every composite without distinction, which has in fact been done more than once.

Note also that I am speaking of the simple here only insofar as it is necessarily given within the composite, which can be broken down into it as its components.

B 470 The word monas (Leibniz’s usage) should properly apply only to what is simple and given directly as a simple substance (in self-consciousness, for example), not as an element of a composite. For that second kind, atomus would be the better name. I only want to prove simple substances as elements of the composite. So I might give the name transcendental atomism to the second antinomy’s antithesis. Yet this word was long ago taken up to name a special way of explaining bodily appearances (molecularum), which presupposes empirical concepts. So it may instead be called the dialectical principle of monadology.

B 467 II. Remark on the Antithesis

The monadists object to this proposition of infinite divisibility of matter, whose ground of proof is purely mathematical. Their objections are suspect right from the start. The clearest mathematical proofs they won’t let count as insights into how space is made, although space truly is the formal condition for the possibility of any matter. They see them merely as conclusions drawn from concepts that are abstract but arbitrary, and so can’t be applied to real things. As though anyone could even dream up an intuition of a different kind from the one that space gives us originally. As though the determinations of space a priori did not also cover everything that is possible only by filling this space. If we listened to them, we would have to think of physical points besides the mathematical point. The mathematical point is simple, but it is no part of space, only a boundary of a space. Physical points are simple too, and they have the privilege of being parts of space and filling it up just by being aggregated. There are many common, clear refutations of this absurdity. I won’t repeat them here, since it is entirely pointless to try to argue away the evidence of mathematics with nothing but discursive concepts. I’ll only remark this.

B 469 When philosophy quibbles with mathematics here, the reason is that it forgets the question concerns only appearances and their condition. And here it isn’t enough to find a concept of the simple to match the pure concept of the composite in the understanding. We need to find an intuition of the simple to match the intuition of the composite (matter). By the laws of sensibility, and so for objects of the senses too, that is completely impossible. Now take a whole of substances that is thought through the pure understanding alone. For it, it may always be valid that we must have the simple before any composition of it. But that is not valid for the totum substantiale phaenomenon. As an empirical intuition in space, it necessarily carries this property: none of its parts is simple, since no part of space is simple. The monadists were still sharp enough to want to evade this difficulty. They don’t take space to be what makes outer objects of intuition (bodies) possible. They take the reverse to be true: bodies, and substances in their dynamical relation in general, make space possible. But the only concept we have of bodies is as appearances. As such, they necessarily presuppose space as the condition that makes every outer appearance possible. So the evasion is in vain, and it was cut off sufficiently in the Transcendental Aesthetic above. If bodies were things in themselves, the monadists’ proof would hold.

B 471 The second dialectical claim has one special feature. The dogmatic claim set against it is the only one, among all the sophistical claims, that sets out to prove clearly, on an object of experience, the reality of something we earlier ranked as a mere transcendental idea: the absolute simplicity of substance. The claim is that the object of inner sense, the thinking I, is a substance that is simple without qualification. I won’t get into this now, since it was weighed at greater length earlier. I only note this. Suppose something is thought only as an object, and no synthetic determination of how it is intuited is added to it (as happens with the completely bare representation “I”). Then of course no manifold and no composition can be noticed in the representation. Besides, the predicates by which I think this object are only intuitions of inner sense. Nothing in them can show a manifold with members outside one another, and so real composition. It is only self-consciousness, then, that makes it so: because the thinking subject is at the same time its own object, it cannot divide itself (though it can divide the determinations that inhere in it). With respect to itself, after all, every object is absolute unity. Even so, if this subject is considered from outside, as an object of intuition, it would surely show composition in appearance. And it has to be considered that way whenever the question is whether it contains a manifold with members outside one another, or not.

B 472 Pure Reason’s Antinomy Third Conflict among the Transcendental Ideas

Thesis

Causality that follows natural laws is not the only causality that can account for all the appearances of the world. We also have to assume a causality through freedom in order to explain them.

Proof

Suppose there is no causality other than the kind that follows laws of nature. Then every event presupposes an earlier state, and the event follows that state inevitably, by a rule. But the earlier state must itself be something that happened, something that came to be in time since it wasn’t there before. For if it had always been there, its consequence too would have always been there, and not just now come about. So the causality of the cause by which something happens is itself something that happened. By the law of nature it, in turn, presupposes an earlier state with its causality, and that presupposes a still older one, and so on. If everything happens by laws of nature alone, then, there is only ever a subordinate beginning.

B 474 There is never a first beginning, and so the series on the side of causes that stem from one another is never complete. Yet the law of nature is just this: nothing happens without a cause that is adequately determined a priori. So the claim that all causality is possible only by laws of nature contradicts itself when it is stated with no restriction on its generality. That claim can’t, then, be accepted as the only kind of causality.

We therefore have to assume a causality through which something happens although its cause is not determined, in turn, by an earlier cause according to necessary laws. In other words: an absolute spontaneity on the part of causes, which starts a series of appearances running by natural laws entirely on its own. That is transcendental freedom. Without it, even in nature’s own course, the causal side of the sequence of appearances can never be completed.

B 473 Antithesis

Freedom does not exist. All that happens in the world takes place by the laws of nature and by nothing else.

Proof

Suppose there is such a thing as freedom in the transcendental sense. It would be a special kind of causality by which the events of the world could come about, namely a power to begin a state absolutely, and with it a series of its consequences. Then not only would a series begin through this spontaneity. The very determination of this spontaneity to produce the series, that is, the causality itself, would begin absolutely, so that nothing goes before it by which this action that takes place would be determined according to permanent laws. But every beginning to act presupposes a state of the cause that is not yet acting. And a dynamically first beginning of the action presupposes a state that has no causal link at all with the earlier state of that same cause, that is, a state that does not follow from it in any way. So transcendental freedom runs against the law of causality.

B 475 It is also a kind of combination of the successive states of acting causes under which no unity of experience is possible. Nothing like it is ever found in any experience, and so it is an empty thing of thought.

All we have, then, is nature, and in it we have to look for the coherence and order of the events of the world. Freedom (independence) from the laws of nature frees us from compulsion, true. It also cuts us loose from every rule that could serve us as a guide. For we can’t say that laws of freedom step in to take the place of the laws of nature in the causality of the world’s course. If that causality were determined by laws, it would not be freedom. It would be nothing other than nature. So nature and transcendental freedom differ as lawfulness and lawlessness. Nature burdens the understanding with a difficulty: it has to keep looking higher and higher up the series of causes for where events come from, because the causality in them is always conditioned. But to make up for that, nature promises a thoroughgoing, law-governed unity of experience. The mirage of freedom does the opposite. To the understanding as it investigates the chain of causes, it promises rest, because it leads the understanding to an unconditioned causality that starts to act on its own. But that causality is itself blind, and so it tears away the guiding thread of rules, which is the only thing that makes a thoroughly connected experience possible.

B 476 Note on the Third Antinomy I. Concerning the Thesis

The transcendental idea of freedom comes nowhere near covering all that the psychological concept of the same name contains, and that concept is mostly empirical. The transcendental idea is just the absolute spontaneity of an action, taken as what really grounds our holding the agent responsible for it. Still, this idea is the true stumbling block for philosophy, because philosophy finds it impossibly hard to allow an unconditioned causality of this kind. So whatever it was in the question of free will that has always left speculative reason so embarrassed is, strictly, a transcendental matter. The only question is whether we have to assume a power of starting a series of successive things or states from itself. We are not equally obliged to explain how such a power could be possible. After all, in the case of causality by natural laws we also have to settle for knowing a priori that such causality must be assumed. We don’t grasp at all how the existence of one thing sets up the existence of another, and we simply have to go by experience. Now, what we have shown is this necessity of a first beginning, from freedom, for a series of appearances. We have shown it only to the extent needed to make an origin of the world understandable; every later state can be treated as following by natural laws alone.

B 478 But once the power to start a series in time wholly from itself has been proved (though not seen into), we may also let several series begin on their own, as far as causality goes, at points along the world’s course, and credit the substances in them with a power to act out of freedom. A misunderstanding shouldn’t hold us up here. A successive series in the world, it is said, can have only a relatively first beginning, because one state or another of the world’s things always lies before it. So perhaps no absolutely first beginning of series can occur while the world runs its course. But we aren’t talking about an absolutely first beginning in time. We mean one in causality. Say I get up from my chair right now, entirely freely, with no necessarily determining pressure from natural causes. Then a new series begins outright, and this event is its start. Its natural consequences stretch on without end. In time, the event is just the continuation of an earlier series. But this decision and deed isn’t part of the sequence of mere natural effects, and doesn’t merely continue it. The determining natural causes cease altogether above it, so far as this event is concerned. It follows them, but it doesn’t result from them. So in terms of causality, though not in terms of time, this must be called the absolutely first beginning of a run of appearances.

The need of reason, when it deals with the series of natural causes, to fall back on a first beginning from freedom, shows up very plainly in the following. Apart from the Epicurean school, all the thinkers of antiquity saw themselves forced to posit a first mover to account for the motions of the world. In other words, they posited a freely acting cause that began this series of states first, and by itself. They did not dare to rely on nature alone to make a first beginning intelligible.

B 477 II. Remark on the Antithesis

The defender of nature’s omnipotence (transcendental physiocracy), the opponent of the doctrine of freedom, would defend the position against the latter’s sophistical inferences like this. “If you don’t posit anything first in the world in the mathematical sense, as to time, you needn’t hunt for something first in the dynamical sense, as to causality. Who ordered you to make up a first state of the world, absolute and unqualified, and so an absolute starting point for the series of appearances unrolling bit by bit, and then, to give your imagination a resting place, to put boundaries on nature, which has none? Substances in the world have always existed. At the very least, the unity of experience makes that assumption necessary. So there is no problem in assuming that the alternation of their states, that is, the run of their changes, has also always been going on. Then no first beginning, mathematical or dynamical, should be sought at all. That such an infinite descent, with no first member to which all the rest merely succeed, is possible can’t be made comprehensible. But if you want to throw these riddles of nature away because of that, you will be compelled to discard many basic synthetic properties (fundamental powers) that you can no more comprehend.

B 479 Even the possibility of change in general will have to look objectionable to you. Unless experience showed you it is real, you could never think up a priori how such a constant succession of being and not-being could be possible.”

Now suppose a transcendental power of freedom is granted, at a pinch, for beginning the world’s changes. Even then, it would have to lie, at the least, outside the world. (Even so, it would remain a bold presumption to assume, beyond the sum of all possible intuitions, yet another object that can’t be given in any possible perception.) But it can never be permitted to credit substances within the world itself with such a power. If it were, the coherence of appearances that necessarily determine one another by universal laws, which is what we call nature, would largely vanish. So would the mark of empirical truth that separates experience from a dream. Next to a lawless power of freedom like that, hardly any nature can be thought at all. The influences of freedom would constantly change nature’s laws, and the play of appearances, which under nature alone would be regular and uniform, would turn confused and disconnected.

B 480 Pure Reason’s Antinomy Fourth Conflict among the Transcendental Ideas

Thesis

Something that belongs to the world, whether as a part of it or as its cause, is a being that is necessary without any qualification.

Proof

The world of sense, being the whole of all appearances, includes a series of changes. Without them we wouldn’t even have the representation of the time series, which is a condition that makes the world of sense possible.* Each change falls under a condition that comes earlier in time, and given that condition the change is necessary. Now whenever a conditioned thing is given, then as far as its existence goes, a complete series of conditions is presupposed, extending to what is unconditioned without qualification, and that alone is absolutely necessary. Something absolutely necessary must therefore exist if there is a change that exists as its consequence. This necessary thing, however, belongs to the world of sense itself, for suppose it were outside that world, so that the series of world changes drew its start from it.

B 482 Yet this necessary cause would not itself belong to the world of sense. But that is impossible. A time series can have its beginning fixed only by what precedes it in time. So the supreme condition for the beginning of a series of changes has to exist in time, at a point when that series was not yet there. (A beginning is an existence with a time before it in which the thing that begins was not yet there.) The causality of the necessary cause of the changes, and so the cause itself, therefore belongs to time and so to appearance (time is possible only as appearance’s form). It follows that it can’t be thought of as separated from the world of sense, the sum of all appearances. So the world itself contains something absolutely necessary (whether it be the entire world series or only a part of it).

* As the formal condition for changes to be possible, time does come before them objectively. But subjectively, and in the reality of consciousness, the representation of time is given only on occasion of perceptions, just as any other representation is.

B 481 Antithesis

An absolutely necessary being does not exist anywhere, neither within the world nor outside it as its cause.

Proof

Suppose the world itself, or something within it, is a necessary being. In the series of its changes there would then be one of two things. One is a beginning that is unconditionally necessary and therefore has no cause. That clashes with the dynamical law governing how all appearances in time are determined. The other is that the series has no beginning whatsoever. Every part of it would be contingent and conditioned, and yet the whole would be absolutely necessary and unconditioned. This is self-contradictory, since a multitude can’t exist necessarily if no single part of it has existence that is necessary in itself.

Suppose, on the other hand, that an absolutely necessary cause of the world existed outside the world.

B 483 As the supreme member in the series of causes of the world’s changes, it would be the one that first started those changes and their series.* But then it would also have to start acting, and its causality would fall in time, and for that very reason within the sum of appearances, that is, in the world. That makes the cause itself no longer external to the world, which goes against the assumption. So there is no absolutely necessary being, either in the world or outside it (though causally combined with it).

* “Begin” is used in two ways. In the active sense, the cause sets going a series of states that are its effect (infit.). In the passive sense, the causality begins in the cause itself (fit.). I am inferring here from the former to the latter.

B 484 Note on the Fourth Antinomy I. Concerning the Thesis

My task here, in proving that a necessary being exists, is to use nothing but a cosmological argument. Such an argument climbs from what is conditioned in appearance to what is unconditioned in concept, by seeing the unconditioned as the condition that the series’ absolute totality necessarily needs. An attempt at a proof from the bare idea of a supreme being among all beings falls under a different principle of reason, so that kind of proof will have to be taken up on its own.

The pure cosmological proof can establish that a necessary being exists only if it also leaves open whether the necessary being is the very world, or a thing that differs from it. To find out which, we would need principles that are no longer cosmological and don’t go along the series of appearances. They would be concepts of contingent beings as such (considered only as objects of the understanding), along with a principle that lets us connect them with a necessary being by concepts alone. All this belongs to transcendent philosophy, which has no place here yet.

But once the proof begins cosmologically, with the series of appearances, and the regressus that runs through it by empirical laws of causality, as its basis, it can’t later leap away from that to something that isn’t a member of the series at all.

B 486 After all, whatever counts as a condition has to count in the same sense in which the conditioned stood to its condition when the series was set up, a series that was to lead us by continuous steps to the highest condition. Now suppose this relation is sensible and falls within the possible empirical use of the understanding. Then the supreme condition or cause can only end the regressus by laws of sensibility, that is, only as part of the time series, and the necessary being has to be seen as the highest link of the world series.

All the same, people have allowed themselves the freedom to make a jump of this kind (metabasis eis allo genos, a shift into another kind). They reasoned from the changes in the world to their empirical contingency, that is, to the fact that they depend on empirically determining causes. This gave them a series of empirical conditions rising upward, and that was perfectly correct. But they could find no first beginning or highest member in that series. So all at once they dropped the empirical concept of contingency and adopted the pure category. This gave rise to a series that is merely intelligible. That series was complete only if an absolutely necessary cause existed. And because this cause was tied to no sensible conditions, it was released from the condition of time as well, the condition that its causality has to begin itself. This procedure is entirely unjustified, as the following shows.

A thing is contingent, taking the category in its pure sense, if the opposite that contradicts it is possible. But empirical contingency gives us no basis for inferring that intelligible kind.

B 488 If something changes, its opposite state is actual at some other time, and so also possible. But that doesn’t make it the contradictory opposite of the earlier state. For that we would need to know that, at the very time the earlier state was there, its opposite could have taken its place. Change can’t tell us this. Take a body at first in motion (= A) that then comes to rest (= non A). That a state opposed to A comes after A doesn’t allow us to infer that A’s contradictory opposite is possible, and therefore that A is contingent. We would need rest to have been possible during the time of the motion, as its replacement. But what we know is only that rest was actual later, and therefore possible. Being in motion at one time and being at rest at another are not set against each other as contradictories. So the succession of opposed determinations, that is, change, doesn’t show contingency in the sense of the pure concepts of the understanding. Nor can it take us, through pure concepts of the understanding, to the existence of a necessary being. What change shows is only empirical contingency. By the law of causality, the new state couldn’t have arisen by itself, without a cause that belongs to the earlier time. This cause, even if it is assumed to be absolutely necessary, has to turn up in time on this account, and belong to the series of appearances.

B 485 II. Remark on the Antithesis

Suppose someone thinks they find, in climbing up the series of appearances, difficulties that tell against a supreme cause which is necessary without qualification. Such difficulties can’t be grounded on bare concepts of a thing’s necessary existence as such. They would not be ontological then. They have to emerge from the causal combination of a cause with a series of appearances, that is, from positing, for that series, a condition that is unconditioned in itself. They have to be cosmological and drawn according to empirical laws. What has to be shown is that the climb along the series of causes (in the world of sense) can never stop at a condition that is empirically unconditioned. And it has to be shown that the cosmological argument, which starts from how contingent the world’s states are as their changes show, comes out against assuming a first cause that starts the series absolutely.

B 487 This antinomy shows a strange contrast, though. The thesis and the antithesis both use one and the same ground of proof, and they use it with equal sharpness. The thesis concludes from it that an original being exists. The antithesis concludes from it that no such being exists. First the argument was: a necessary being exists, because all of past time contains the whole series of conditions, and with it the unconditioned (the necessary). Now it is: no necessary being exists, for the very reason that all elapsed time contains the whole series of conditions (and these are all conditioned in turn). The cause of this is the following. The first argument looks only at the series of conditions as an absolute totality, where one condition determines another in time, and so it ends up with something unconditioned and necessary. The second argument takes into account the contingency of everything determined in the time series. (Each of them is preceded by a time, and in that time the condition itself has to be determined, in its turn, as conditioned.)

B 489 That makes everything unconditioned, and all absolute necessity, vanish altogether. Still, the way of inferring in both suits even common human reason, which often ends up divided against itself when it weighs its object from two different standpoints. Mr. von Mairan took the dispute between two famous astronomers, which came from a similar difficulty over the choice of standpoint, to be a phenomenon remarkable enough to deserve a special treatise. One of them argued: the moon turns on its axis, because it always turns the same side toward the earth. The other: the moon doesn’t turn on its axis, precisely because it always turns the same side toward the earth. Each inference was right, according to which standpoint a person took for observing how the moon moves.

B 490 Pure Reason’s Antinomy Third Section On Reason’s Interest in This Conflict with Itself

There we have it: the whole dialectical play of the cosmological ideas. No object matching them can be given in any possible experience. Reason can’t even think them in harmony with the general laws of experience. Yet these ideas aren’t dreamed up at will. When reason, advancing steadily through empirical synthesis, tries to free from every condition what the rules of experience can only ever determine conditionally, and to take it in as an unconditioned totality, it is led to them of necessity. These sophistical claims are each an attempt to solve one of four natural, unavoidable problems of reason. There can be just that many, neither more nor fewer, since there are only so many series of synthetic presuppositions bounding empirical synthesis a priori.

The dazzling pretensions of a reason that widens its territory beyond all boundaries of experience have been shown here only in dry formulas.

B 491 These contain just the basis for its legitimate claims. And, as befits a transcendental philosophy, we stripped them of everything empirical, though the whole glory of reason’s assertions can shine out only together with the empirical. When philosophy is applied this way, though, and reason’s use keeps widening, starting in the field of experience and gradually rising to these exalted ideas, philosophy shows a dignity. If it could only make good on its pretensions, this dignity would leave the worth of every other human science far behind. For philosophy promises a foundation for our greatest hopes and prospects, concerning the final ends in which all the efforts of reason must finally come together. Consider these questions. Does the world have a beginning, and some boundary to its extent in space? Is there anywhere, maybe in my own thinking self, a unity that can’t be divided or destroyed, or is there only what can be divided and what passes away? Am I free in what I do, or, like other beings, led along by the thread of nature and fate? Is there, in the end, a supreme cause of the world, or do natural things and their order form the last object, where we must stop in all our thinking? A mathematician would gladly give up the whole of mathematics to get these questions resolved. For mathematics can bring no satisfaction when it comes to the highest and most urgent purposes of humanity.

B 492 Even the true dignity of mathematics (the pride of human reason) rests on this. Mathematics guides reason in seeing how orderly and regular nature is, from the largest things to the smallest, and in seeing the wonderful unity of the forces that move it. This goes far beyond anything a philosophy built on ordinary experience could expect. In this way mathematics gives reason itself an occasion and encouragement to use itself beyond all experience. It also supplies the philosophy dealing with these matters with excellent materials, so that its inquiry can be backed up, as far as its nature permits, with fitting intuitions.

Unfortunately for speculation (though perhaps happily for humanity’s practical vocation), reason finds itself, in the very midst of its greatest hopes, so caught in a throng of arguments and counterarguments that it is stuck. Retreating isn’t possible, either for its honor or for its safety, and neither is looking on indifferently at the quarrel as if it were a mere skirmish for show. Ordering outright peace is even less possible, since the object of the dispute holds great interest. Nothing is left but to ponder where this division of reason against itself comes from, and whether a mere misunderstanding might be at fault.

B 493 Clearing that up may well make the proud claims on both sides drop away, but in return a lasting, quiet government of reason over understanding and senses would get its start.

For now we will put this thorough inquiry off a little. First let us consider which side we would most like to take if we were somehow forced to take sides. In this case we consult only our interest, not the logical touchstone of truth. So an inquiry like this settles nothing about the disputed rights of the two parties. Still, it will be useful. It will make clear why the participants in this dispute have taken one side rather than the other, though no superior insight into the matter was the reason. It will also explain some side issues, for example the zealous heat of one party and the cold assertion of the other, and why people gladly cheer one party with joyful applause and are implacably prejudiced against the other beforehand.

There is one thing that settles the viewpoint for this preliminary assessment, the only viewpoint from which it can be done with real thoroughness. That is the comparison of the principles the two parties start out from. In the claims of the antithesis one notices perfect uniformity in the way of thinking, and complete unity of maxim.

B 494 That maxim is the principle of pure empiricism, applied not just in explaining appearances in the world but in solving the transcendental ideas about the cosmos itself. The claims of the thesis, on the other hand, base themselves on intellectual beginnings in addition to the empirical way of explaining things within the series of appearances, and for this reason their maxim isn’t simple. From what chiefly marks it out, I would name it pure reason’s dogmatism.

On the side of dogmatism, then, in settling the cosmological ideas of reason, that is, on the side of the thesis, we find:

First of all there is a practical interest, one that every well-disposed person, if they understand their own true advantage, shares warmly. That the world has a beginning, that my thinking self has a simple, and therefore indestructible, nature, that this self is at the same time free in its deliberate actions and lifted above natural compulsion, and finally that the entire order of the things that make up the world has an original being as its source, and from that being everything borrows its unity and its purposeful connection: these are so many cornerstones of morality and religion. The antithesis takes all these supports away from us, or seems to.

Second, reason has a speculative interest on this side too. Accept the transcendental ideas and use them in this way.

B 495 Then one can take in the entire chain of conditions entirely a priori, and see how the conditioned is derived, by beginning with the unconditioned. The antithesis can’t do that, and this makes it a poor recommendation. Ask it about the conditions for its synthesis, and any answer it gives leaves us asking further, endlessly. It says we must rise from a given beginning to a higher one, that each part leads on to a smaller part, that each event has yet another event above it as its cause, and that the conditions of existence as such lean on further conditions again and again, never finding unconditioned hold and support in a self-standing thing, an original being.

Third, this side has the advantage of being popular, and that is certainly a major part of its appeal. Ordinary understanding finds no difficulty at all in the ideas of an unconditioned beginning for all synthesis. It is in any case more used to moving down toward consequences than up toward grounds. Concepts of the absolutely first give it ease, and a fixed point where it can fasten the line that guides its steps. It doesn’t brood over whether that first is even possible. Rising ceaselessly from conditioned to condition, always with one foot in midair, can’t please it at all.

B 496 On the side of empiricism, in settling the cosmological ideas (the antithesis), there is, first, no practical interest of the kind morality and religion bring with them out of pure principles of reason. If anything, plain empiricism seems to strip both of all their force and influence. Suppose no original being exists apart from the world, and the world has no beginning and so no creator. Suppose we have no free will, and the soul can be split up and can decay just as matter can. Then the moral ideas and principles lose all validity too, and they collapse along with the transcendental ideas that were their theoretical support.

Empiricism, though, does hold out advantages for reason’s speculative interest. They are enticing, and they exceed by far those that the dogmatic teacher of the ideas of reason might promise. In empiricism the understanding is always standing on ground of its own, the field of nothing but possible experiences. It can trace their laws, and through them it can enlarge its cognition without limit, a cognition that is secure and easy to grasp. The understanding can, and should, show the object to intuition, both as it is and in how it stands to other things, or else in concepts whose picture can be shown clearly and distinctly in similar given intuitions. The understanding has no need to step out of this chain of the natural order.

B 497 If it did, it would be attaching itself to ideas whose objects it doesn’t know, since as things of thought they can never be given. Beyond that, the understanding is not even allowed to quit its work. It can’t say the job is now done and move on into the territory of idealizing reason and transcendent concepts. In that territory it would no longer have to observe and to investigate by the laws of nature. It would only have to think and make things up. It would be sure that facts about nature could never refute it, since it isn’t bound by what they testify. It may simply ignore them, or even place them beneath a higher authority, that of pure reason.

So the empiricist will never allow anyone to take some epoch of nature as the absolutely first one. The empiricist will never treat any boundary of the view into nature’s extent as the outermost one. Nor will the empiricist move on from the objects of nature, which can be broken down by observation and mathematics and determined synthetically in intuition (the extended), to objects that neither the senses nor the imagination can ever present in a concrete case (the simple). The empiricist also won’t agree to lay down, even inside nature itself, a power to act independently of the laws of nature (freedom). That would cut into the job of the understanding, which is to trace how appearances arise with necessary rules as its guide.

B 498 And in the end the empiricist will not grant that anyone should look for the cause of anything outside nature (an original being). We know nothing except nature, because it alone presents objects to us and can teach us its laws.

Granted, the empirical philosopher may mean the antithesis to do just one job: to put down the nosy presumption of a reason that misjudges its true vocation. That kind of reason puts on a big show of insight and knowledge at the very point where insight and knowledge stop. It wants to sell what we allow for practical purposes as a boost to the speculative interest. And whenever it suits its comfort, it snaps the thread of physical inquiry, then ties it to transcendental ideas while claiming to widen our cognition. All those ideas really teach us is that we know nothing. Now suppose the empiricist stopped there. Then the empiricist’s principle would be a rule of restraint in what we claim and modesty in what we assert. It would also be a rule for stretching our understanding as far as it can go, under the teacher we were actually given, which is experience. We would not lose our intellectual assumptions or our faith, which serve our practical affairs. We just couldn’t put them forward under the grand title and show of science.

B 499 Nor could we call them rational insight. Real speculative knowledge can only ever reach objects of experience. Step across the boundary of experience, and the synthesis that tries for new cognitions, independent of experience, has no intuition underneath it to work on.

But empiricism often turns dogmatic itself when it comes to ideas. When it does, and flatly denies whatever goes beyond the cognitions it gets through intuition, it falls into the very same fault of immodesty. Here the fault is worse, because it does irreplaceable damage to reason’s practical interest.

That is the conflict between Epicureanism* and Platonism.

* It is still unsettled whether Epicurus ever offered these principles as claims about objects. If they were only rules for how reason should work in speculation, he showed a truer philosophical spirit than any thinker of the ancient world. The rules are these. Explain appearances as if no boundary or beginning of the world shut off the field of inquiry. Take the world’s matter to be whatever it must be if experience is to teach us about it. Allow events to come about only through unchanging laws of nature. And bring in no cause that is different from the world. These rules are still very sound, though few follow them, for widening speculative philosophy. They are just as sound for finding the principles of morals without outside help. And someone who asks us to leave those dogmatic claims aside while we are busy with pure speculation can’t fairly be accused, for that, of wanting to deny them.

B 500 Each side says more than it knows. But they differ. The first encourages and advances knowledge, even though it harms the practical side. The second does offer excellent principles for practical matters. Yet by that very fact, in everything where only speculative knowledge is open to us, it lets reason chase idealizing explanations of natural appearances and neglect physical research.

The third point to weigh in choosing provisionally between the two sides is popular appeal. Here something is very strange. Empiricism has no popular appeal at all. You would expect common understanding to jump at a scheme that promises to satisfy it with nothing but knowledge from experience and the reasoned links between such items. Transcendental dogmatism, on the other hand, makes common understanding climb up to concepts that are far beyond the insight and reasoning power of the best-trained thinkers. But that is just what attracts it.

B 501 For then common understanding finds itself in a position where even the most learned person can claim no advantage over it. If it understands little or nothing here, nobody else can boast of understanding much more. It may not be able to talk about these matters in proper school style, but it can reason about them endlessly. It is wandering among ideas, and people are most eloquent about ideas precisely because they know nothing about them. When it comes to studying nature, though, it would have to keep quiet and admit that it doesn’t know. So comfort and vanity are already a strong recommendation for these principles. There is more. A philosopher has a hard time taking anything as a principle without being able to give an account of it, or bringing in concepts whose objective reality can’t be seen. Common understanding finds nothing more ordinary. It wants something to start from with confidence. It isn’t bothered by the difficulty of grasping such an assumption, since it doesn’t know what grasping means and the difficulty never crosses its mind. What frequent use has made familiar, it counts as known. In the end, though, speculative interest gives way entirely to the practical, and common understanding fancies that it sees and knows whatever it feels pushed to assume or believe.

B 502 What pushes it is worry or hope. So the empiricism of transcendentally idealizing reason has no popular appeal whatever. It may hold a good deal that is harmful to the highest practical principles. Still, there is nothing to fear: it will never get out of the schools, or win any real standing or favor with the crowd.

Human reason is by nature architectonic. That is, it looks at all cognitions as parts of a possible system. So it allows only principles that at the least don’t stop a planned piece of cognition from fitting alongside others in some system. The propositions of the antithesis, though, make it entirely impossible to complete a building of cognitions. On them, every state of the world has an older one behind it. Every part has other parts in it, and those can be divided again. Every event has an earlier one before it, which was itself produced by something else. Everything in existence is conditioned only, and no unconditioned, first existence is ever recognized. So the antithesis nowhere grants anything first, no beginning that could be the base of the structure. A complete building of cognition is therefore wholly impossible on these assumptions.

B 503 That is why reason’s architectonic interest (which asks for pure unity of reason, a priori, and not for empirical unity) carries a natural recommendation for the claims of the thesis.

Suppose someone could drop every interest and judge the assertions of reason only by the weight of their grounds, whatever the consequences. Suppose too that the only way out of the squeeze was to side with one of the two disputed doctrines. That person would waver without end. Today it would seem convincing that the human will is free. Tomorrow, looking at the unbreakable chain of nature, they would decide that freedom is only self-deception and that everything is nature. Yet once they had to act, this game played by purely speculative reason would vanish like the figures in a dream, and they would choose their principles by practical interest alone. Still, a being that thinks and inquires should set aside certain times for testing its own reason, drop all partiality while doing it, and share its observations in public for others to judge. So nobody can be blamed for this, let alone stopped, if they let the claims and counterclaims have their say.

B 504 Let them defend themselves as well as they can, frightened by no threat, before a jury of their own kind: weak human beings.

Pure Reason’s Antinomy Section Four Problems of Transcendental Philosophy That Reason Absolutely Must Be Able to Solve

Trying to solve every problem and answer every question would be shameless bragging. It would be such wild conceit that you would lose everyone’s trust at once. Yet some sciences are built so that any question arising inside them must have an answer drawn from what is already known. That is because the answer has to spring from the same sources as the question. In those sciences you can’t plead unavoidable ignorance. A solution can be demanded. We must be able to know, as a rule, what is right and what is wrong in every possible case, because that is a matter of our duty, and we have no duty regarding what we can’t know.

B 505 When it comes to explaining the appearances of nature, by contrast, a lot has to stay uncertain and many questions have to stay unsolved, because what we know of nature falls far short, in many cases, of everything we are asked to explain. So here is the question. Does transcendental philosophy contain any question, about an object put before reason, that this same pure reason can’t answer? And may we rightly get out of giving a decisive answer by filing the matter as absolutely uncertain, given everything we can cognize? That is, may we put it among the things for which we have just enough of a concept to ask a question, but have no means or power to answer it?

I claim that transcendental philosophy has this special feature among all speculative knowledge: no question about an object given to pure reason is unsolvable for that same human reason. Nobody can plead unavoidable ignorance, or a problem too deep to fathom, to get free of the duty to answer it fully and thoroughly. The very concept that lets us ask the question must also make us able to answer it, since the object is nowhere to be found outside the concept.

B 506 Right and wrong are like that.

In transcendental philosophy the cosmological questions are the only ones where we may rightly demand a satisfying answer about the nature of the object. The philosopher isn’t allowed to dodge them by pleading impenetrable obscurity. These questions can only be about cosmological ideas. The reason is that the object has to be given empirically, and the only issue is whether it matches an idea. Things differ when the object is transcendental, and so itself unknown. Examples are these: whether the something that appears (in us) as thinking, the soul, is a simple being in itself; whether there is a cause of all things together that is absolutely necessary; and so on. Here we have to look for an object to go with our idea, one that we can admit is unknown to us, though not for that reason impossible*.

B 507 Only the cosmological ideas have this distinctive feature: they can take their object, and the empirical synthesis its concept requires, as already given. The question they raise concerns just how that synthesis goes on, as far as it is supposed to reach absolute totality. Such totality is no longer anything empirical, because no experience can give it. Now, here we are dealing with a thing only as an object that experience could give us, never as something that exists on its own. So the transcendent cosmological question can be answered only from within the idea, since it isn’t about any object in itself. And as for possible experience, we aren’t asking what might be given in some concrete experience. We are asking what the idea contains, the idea that empirical synthesis is only meant to approach. So the question has to be solvable from the idea alone. The idea is just a creation of reason. Reason therefore can’t disown responsibility for it, or push that responsibility onto the unknown object.

* Asked what a transcendental object is like, that is, what it is, we can give no answer. What we can say is that the question itself amounts to nothing, since no object for it has been given. For that reason all the questions of transcendental psychology can be answered, and in fact have been. They concern the transcendental subject that underlies all inner appearances. That subject is not itself an appearance, so it is not given as an object, and none of the categories (which are what the question is really aimed at) finds any conditions for being applied to it. This is a case where the common saying holds that no answer is also an answer. Suppose we ask about the nature of a something that no determinate predicate can express, since it is set entirely outside the range of objects that could ever be given to us. Such a question is completely void and empty.

B 508 It is not as odd as it first seems that a science could demand certain answers to all the questions inside its own territory (quaestiones domesticae, “home questions”), and expect to get them, even if some haven’t been found yet. Transcendental philosophy isn’t the only science of pure reason. There are two more: pure mathematics, purely speculative in content, and pure morals, practical in content. Who has ever been told that the exact ratio of a circle’s diameter to the circle is uncertain, in rational or in irrational numbers, as if we were bound to be ignorant of the conditions? Rational numbers can’t give that ratio exactly, and irrational ones haven’t yielded it yet. So people concluded that we could at least know for certain that no such solution is possible, and Lambert gave a proof. In the general principles of morals nothing can be uncertain. Either the propositions are wholly empty and meaningless, or they must follow simply from our concepts of reason. In the study of nature it is different. There are endless conjectures there that can never be certain. The appearances of nature are objects that reach us apart from our concepts. So the key to them lies outside us, not within us or our pure thinking. For that reason it often can’t be found.

B 509 So no certain answer can be expected. I’m not counting here the questions of the transcendental analytic that concern the deduction of our pure cognition. We are concerned now only with how certain our judgments about objects are, not with where our concepts themselves come from.

So we can’t wriggle out of the duty to give at least a critical answer to the questions put to reason. We can’t do it by complaining about the narrow limits of our reason and, with the illusion of humble self-knowledge, admitting that it is beyond us to settle these matters: whether the world has always existed or had a beginning; whether space is filled with beings forever or is closed in by certain boundaries; whether anything in the world is a simple unit, or all of it divides forever; whether things are generated and produced by freedom, or all of them hang from the chain of natural order; and finally whether some being is wholly unconditioned and necessary in itself, or everything exists conditionally, depends on something outside, and is contingent in itself. All of these questions are about an object found only in our thinking: the wholly unconditioned totality of the synthesis of appearances.

B 510 Suppose that from our own concepts we can’t state or settle anything sure about it. Then we mustn’t put the blame on the thing for hiding from us. No such thing can be given to us at all, because it is found nowhere outside our idea. The cause has to be sought in our idea itself. That idea is a problem with no solution, and yet we stubbornly assume that a real object matches it. If we laid out clearly the dialectic that lies in our concept itself, we would soon be completely certain what we ought to judge about such a question.

Your excuse is that these problems are uncertain. To that we can first reply with a question that you at least must answer clearly. Where do the ideas come from, whose solution tangles you in such difficulty? Are they perhaps appearances that you need explained, so that these ideas only send you looking for principles, or for the rule for setting those appearances out? Imagine nature laid wide open to you. Nothing is hidden from your senses or from your awareness of all that is set before your intuition. Still, no experience at all would let you know the object of your ideas in a concrete case. (Besides the complete intuition, a finished synthesis would be needed.

B 511 And you would need to be conscious of its absolute totality, which no empirical cognition can provide.) So your question can’t be needed to explain any appearance that comes up. The object itself doesn’t set it for you, so to speak. The object can never come up for you, because no possible experience can give it. All your possible perceptions leave you stuck inside conditions, in space or in time. You never reach anything unconditioned that would let you decide whether the unconditioned lies in an absolute first beginning of the synthesis, or in an absolute totality of a series with no beginning. But “all” in the empirical sense is only ever relative. The absolute all of size (the universe), of division, of descent, of the condition of existence in general, with every question whether it could be reached by a finite synthesis or by one that goes on forever, is none of experience’s business. Take a body. You couldn’t explain its appearances any better, or even differently, whether you assume it consists of simple parts or of parts that are made up of further parts all the way down. No simple appearance, and no infinite composition either, can ever come up for you. Appearances need explaining only as far as the conditions for explaining them are given in perception.

B 512 Everything that might ever be given in them, gathered into one absolute whole, would itself have to be a perception. And it is exactly this all that the transcendental problems of reason ask us to explain.

Experience can never contain even the solution of these problems. So you can’t claim it is uncertain what to ascribe to the object here. Your object is just in your head. It can’t be given anywhere outside it. All you have to do is make sure you are in agreement with yourselves, and avoid the amphiboly that makes your idea look like the representation of something given empirically, an object to be known by the laws of experience. So a dogmatic solution isn’t uncertain; it is impossible. A critical solution, which can be perfectly certain, doesn’t treat the question objectively at all. It looks to the ground of cognition on which the question is built.

B 513 Pure Reason’s Antinomy Section Five A Skeptical Presentation of the Cosmological Questions Across All Four Transcendental Ideas

We would gladly drop our demand for dogmatic answers if we could see in advance that any answer would only increase our ignorance. It would throw us from one thing we can’t grasp to another, from darkness into deeper darkness, and perhaps even into contradictions. If our question asks only for a yes or a no, it is smart to put the likely grounds for an answer aside for the moment. First weigh what we would gain if the answer went one way, and what we would gain if it went the other. Suppose both turn out to be pure nonsense (Nonsens). Then we have good reason to examine the question itself critically, and to see whether it rests on a groundless assumption and toys with an idea whose falseness shows up better when the idea is applied, and in its consequences, than when it is taken by itself. That is the great advantage of the skeptical way.

B 514 It lies in how it treats the questions that pure reason puts to pure reason, and at little cost it can spare us a huge pile of dogmatic clutter and put a sober critique in its place. Like a true purge (Katarktikon), that critique will happily clear out the delusion, together with its companion, the habit of knowing many things.

Suppose I could see ahead of time that a cosmological idea would come out either too big or too small for any concept of the understanding. It would do so in the regressive synthesis of appearances, on whichever side of the unconditioned it came down. Then I would understand the following. The idea is only about an object of experience, and experience has to fit some possible concept of the understanding. So the idea must be wholly empty and meaningless, because no object fits it however I bend the object to suit. And that is exactly the case with every concept of the world. This is why these concepts tangle reason in an unavoidable antinomy for as long as it holds on to them. For consider the following.

First, suppose the world has no beginning. Then it is too big for your concept. Your concept is a successive regress, and that can never get back through all the eternity that has passed. Now suppose the world has a beginning. Then it is too small again, this time for your concept of the understanding in the necessary empirical regress.

B 515 A beginning still presupposes an earlier time, so it is not yet unconditioned. And the law that governs the understanding’s empirical use makes you ask for a time-condition still further back. So plainly the world is too small for this law.

The same goes for the two possible answers about the size of the world in space. If the world is infinite and unbounded, it is too big for any possible empirical concept. If it is finite and bounded, you can fairly go on to ask what fixes the boundary. Empty space isn’t something that stands on its own as a correlate of things. It can’t be a condition where you could stop, and it is even less an empirical condition that could be part of a possible experience. (Who can have an experience of the completely empty?) But for the absolute totality of the empirical synthesis, the unconditioned must always be a concept taken from experience. A bounded world, then, turns out too small for your concept.

Second, suppose every appearance in space (matter) is made of infinitely many parts. Then the regress of division is always too big for your concept. Suppose instead the division of space is to end at some member (the simple). Then the regress turns out too small to match the idea of the unconditioned.

B 516 That member still leaves room for a regress to further parts inside it.

Third, suppose you assume that everything that happens in the world is nothing but a result following laws of nature. Then the causality of the cause is itself again something that happens. That forces you to regress to a still higher cause, and so to extend the series of conditions a parte priori (backward) without end. Mere working nature is therefore too big for any concept you have, in the synthesis of world events.

Now suppose you pick, here and there, events that bring themselves about, and so production from freedom. Even then the question “why” chases you back to a law of nature you can’t avoid. It forces you to go past this point, following the causal law of experience. And you find that a totality of connection like this proves too small for the empirical concept you cannot do without.

Fourth. You might assume a being that is absolutely necessary. It may be the world itself, or a part of the world, or the world’s cause. Then you have to place it in a time infinitely far from any given moment. Otherwise it would depend on an older existence. But then this existence is out of reach for your empirical concept. It is too big for you ever to get to it by any continuing regress.

B 517 Suppose instead you hold that everything belonging to the world, as conditioned or as condition, is contingent. Then any existence you are given is too small for your concept, because it forces you always to look around for another existence on which it depends.

In all these cases we said that the idea of the world is too big for the empirical regress, or else too small, and so for every possible concept of the understanding. Why didn’t we say it the other way around? We might have said that in the first case our empirical concept always turns out too small beside the idea, and in the second too large, so that the fault lies, in a way, with the empirical regress. Instead we blamed the cosmological idea for missing its goal, which is possible experience, by going too far or not far enough. Here is why. Possible experience is the one thing that can give our concepts reality. Without it, any concept is only an idea, with no truth and no relation to an object. So the possible empirical concept had to be the measuring stick for judging the idea: is it a mere idea, a thing of thought, or does it find its object in the world? We only call something too big or too small relative to something else when it was taken on because of that other thing, and has to be shaped to suit it.

B 518 The old schools of dialectic had a puzzle like this to play with. A ball won’t fit through a hole. Should we say the ball is too big, or the hole too small? In that case it doesn’t matter how you put it, since you don’t know which of the two exists for the other’s sake. But no one will say a man is too long for his coat. They’ll say that it is the coat that is too short.

So at least we have a well-founded suspicion. The cosmological ideas, and with them all the sophistical assertions that are set against one another, may rest on a concept that is empty and only imagined, a notion of the way their object is given to us. This suspicion alone can set us on the right track to uncovering the mirage that has led us astray for so long.

Pure Reason’s Antinomy Section Six Transcendental Idealism: The Key to Resolving the Cosmological Dialectic

In the transcendental aesthetic we proved well enough that whatever is intuited in space or time, and so every object of any experience we can have, is only appearance. That is, it is mere representation.

B 519 As such, whether it is shown as extended beings or as series of changes, it has no existence of its own outside our thoughts. I call this theory transcendental idealism.* The transcendental realist makes these modifications of our sensibility into things that stand on their own. In that way the realist turns mere representations into things in themselves.

* I have at times also called it formal idealism, to set it apart from the material kind, the ordinary idealism that questions or denies the very existence of outer things. In some cases it seems wiser to use this name rather than the one above, so as to prevent any misunderstanding.

It would be unjust to pin on us the empirical idealism that has been shouted down for so long. This idealism grants that space is real in its own right. But it denies that extended beings exist in space, or at least calls their existence doubtful. On this point it allows no difference between dream and truth that could be proved well enough. Inner sense is another matter. This idealism has no trouble with the appearances of inner sense in time, which it treats as actual things. It even claims that inner experience, by itself, fully proves that its object actually exists (in itself), along with all this determination in time.

B 520 Our transcendental idealism, on the other hand, lets the objects of outer intuition be real as well, exactly as we intuit them in space. The same goes for every change in time, as inner sense presents it. Space is already a form of what we call outer intuition. With no objects in space, we would have no empirical representation whatever. So we can and must treat extended beings in space as actual. Time works the same way. But space itself, together with time, and with both of them all appearances, are not things in themselves. They are nothing but representations and can’t exist anywhere outside our mind. Now take the inner, sensible intuition we have of our own mind (the mind as an object of consciousness), in which its determination is represented as a succession of different states in time. This is not the true self as it exists in itself, and not the transcendental subject. It is only an appearance, handed to the sensibility of a being we know nothing about. We can’t concede that this inner appearance exists as a thing existing in itself in that way, because its condition is time, and time can’t describe any thing in itself. Inside space and time, though, appearances have all the empirical truth they need to count as secure, and they are kept clearly apart from the kinship with a dream.

B 521 That holds whenever both hang together correctly and without a gap, by empirical laws, in a single experience.

So objects of experience never come to us as they are in themselves. They come only within experience, and outside it they don’t exist at all. We do have to allow that the moon might have inhabitants, even though nobody has ever perceived them. But all that means is that we might meet them as experience moves forward. Anything is actual if it stands in one context with some perception, by the laws of the empirical advance. So those inhabitants are actual if they are linked, empirically, to my actual consciousness. That doesn’t make them actual in themselves, apart from this forward movement of experience.

Nothing is actually given to us except perception, and the empirical move from it to other possible perceptions. Appearances in themselves are mere representations, and they are actual only in perception. Perception just is the reality of an empirical representation, which means an appearance. To call an appearance a real thing before any perception can mean only that, as experience goes on, we will have to run into such a perception. Otherwise it means nothing at all. Of course one could say that an appearance exists in itself, with no relation to our senses or possible experience.

B 522 But that would be right only if we were talking about a thing in itself. Here we are talking about an appearance in space and time. Neither is a determination of things in themselves; both belong only to our sensibility. So whatever is in space and time, namely appearances, is nothing in itself. It is representation only, and if it isn’t given in us (in perception), it is found nowhere at all.

Our sensible power of intuition is really just a receptivity: it can be affected by representations in a certain way. Their relation to one another amounts to a pure intuition of space and time (nothing but forms of our sensibility). Insofar as laws that unify experience connect them and make them determinable in this relation (space and time), they are called objects. The cause of these representations isn’t sensible, and we know nothing of it. So we can’t intuit it as an object. An object like that would have to be represented in neither space nor time, which are only conditions of sensible representation, and without those conditions we can’t think of any intuition at all. Still, we may call the cause of appearances in general, a cause that exists only for the intellect, the “transcendental object,” purely so that something answers to sensibility as a receptivity. To this transcendental object we can assign the entire reach and connection of our possible perceptions.

B 523 We can then say that it is given in itself before all experience. Appearances, according to it, are given not in themselves but only within this experience. They are mere representations. As perceptions they mean an actual object only if that perception connects with all the others by the rules of the unity of experience. So the real things of past time can be called given in the transcendental object of experience. For me, though, they are objects, and real in past time, only to the extent that I picture a backward series of possible perceptions. The series follows empirical laws, whether the thread of history or the traces of causes and effects; in short, the way the world runs. It leads back to an earlier stretch of time as the condition of the present time. And even that earlier time is represented as actual only in the setting of a possible experience, and never as it is in itself. So all the events that happened in times beyond remembering, before I existed, mean nothing other than this: it is possible to extend the chain of experience upward, starting from my present perception, to the conditions that fix it in time.

So when I picture all the existing objects of the senses together, in all time and all spaces, I am not placing them into time and space before experience.

B 524 What I am picturing is just the thought of a possible experience, complete in every respect. Only in that experience are those objects given. (They are nothing but representations.) To say they exist before all my experience means only that they will be found in the part of experience that I have yet to reach, starting from perception. What causes the empirical conditions of that advance, which members I will meet in the regress, or how far I will run into them, is transcendental, and so necessarily unknown to me. But that isn’t the concern anyway. The concern is the rule for how experience advances, the experience in which objects (appearances) are given to me. It also comes to the same thing in the end whether I say that, advancing empirically through space, I might meet stars a hundred times farther off than the most distant ones I can see; or whether I say that there may be such stars out in space, even though no human has ever perceived them or ever will. For suppose they were given as things in themselves, with no relation to possible experience. They would still be nothing to me, and so not objects, except to the degree that they are contained in the series of the empirical regress. Only in a different connection does it matter how one understands the reality of objects of the senses that are merely thought: namely, when these same appearances are put to work for the cosmological idea of a whole that is absolute.

B 525 Then the question reaches past the boundaries of any possible experience, and the distinction matters, in order to head off a deceptive delusion. That delusion is bound to arise, unavoidably, from our misreading of our own concepts of experience.

Pure Reason’s Antinomy Section Seven Settling Reason’s Cosmological Quarrel with Itself, Critically

The whole antinomy of pure reason rests on one dialectical argument: if the conditioned is given, then the entire series of its conditions is given too. What our senses give us as objects is conditioned. Therefore, and so on. The major premise of this syllogism looks so natural and obvious. Through it, cosmological ideas come in, one for each kind of condition (in the synthesis of appearances), to the extent that the conditions form a series. These ideas demand the absolute totality of such series. By that very demand they throw reason, unavoidably, into conflict with itself. Before we can expose what is misleading in this sophistical argument, we have to prepare ourselves.

B 526 We do it by correcting and pinning down certain concepts that appear in the argument.

First, here is a proposition that is clear and certain beyond doubt. Whenever the conditioned is given, a regress through the series of all its conditions is thereby set before us as a task. The concept of the conditioned already carries this with it: something is tied to a condition, then, if that condition is itself conditioned, to a more distant condition, and so on through every member of the series. This proposition is therefore analytic, and it has nothing to fear from a transcendental critique. It is a logical demand of reason: to follow, by means of the understanding, the connection that a concept already has with its conditions, and to carry it as far as possible.

Next, suppose the conditioned and its condition are both things in themselves. Then once the first is given, the regress to the second isn’t only set as a task. The second is actually already given along with the first. And since that holds for every member of the series, the whole series of conditions, and the unconditioned with it, is given at the same time. Or better, it is presupposed in the very fact that the conditioned, which was possible only by way of that series, is given. In this case the joining of the conditioned to its condition is the work of the bare understanding. The understanding represents things as they are.

B 527 It doesn’t ask whether, or how, we could come to know them. But take the case where I am dealing with appearances. As mere representations they aren’t given at all unless I arrive at some knowledge of them (that is, at the appearances themselves, for they are nothing but empirical knowledge). Then it isn’t open to me to say, in that same sense: the conditioned is given, so its conditions are given too, as appearances. So I can’t conclude to the absolute totality of their series. In apprehension, appearances are themselves nothing but an empirical synthesis carried out in space and time, and they are given only in that synthesis. It doesn’t follow that giving the conditioned (in appearance) also gives and presupposes the synthesis that makes up its empirical condition. That synthesis happens first in the regress, and never without it. In a case like this, though, we can say something: a regress to the conditions, meaning an empirical synthesis that keeps going on this side, is ordered or set as a task. And there can be no shortage of conditions, since the regress supplies them.

From this it is clear what happens. The major premise of the cosmological syllogism uses “conditioned” in the transcendental meaning of a pure category. The minor premise uses it in the empirical meaning, as an understanding-concept applied to nothing more than appearances.

B 528 So the dialectical trick in it is the one called a sophisma figurae dictionis (a fallacy resting on a figure of speech). The trick isn’t contrived. It is a quite natural deception of common reason. We presuppose the conditions and their series in the major premise almost unseen, any time something is given as conditioned. That’s because this is only the logical demand to take complete premises for a given conclusion. In the connection of a conditioned thing with its condition there is no time order at all. They are presupposed in themselves as given together. And in the minor premise it is just as natural to treat appearances as things in themselves, and also as objects that the bare understanding is handed, as we did in the major premise when we set aside all the conditions of intuition, though only under those conditions can objects be given at all. But here we overlooked a striking difference between the concepts. In the major premise, joining a conditioned thing to its condition, and the whole series of conditions, carried no restriction by time and no idea of succession. In the minor premise, empirical synthesis, and the line of conditions inside appearance (which is what gets placed under that premise), are necessarily successive. They are given one after another, in time only. So here I couldn’t presuppose the absolute totality of the synthesis, and of the series it stands for, the way I could there.

B 529 There, each link in the series is given in itself, with no time condition. Here they are possible only through the successive regress, and the regress is given only by really carrying it out.

Once the common argument behind the cosmological assertions has been shown to contain such a slip, both parties to the dispute can rightly be thrown out of court. Neither has a sound title for its claim. Yet that doesn’t settle their quarrel, if settling it means proving that they, or one of them, is wrong about the matter itself, the conclusion, even if it couldn’t base its claim on solid proofs. Nothing seems clearer than this. If one says the world has a beginning and the other says it has none but has existed from eternity, one of them has to be right. If so, then since both sides are equally clear, nobody can ever work out which side is right. The quarrel carries on just as it did, though the court of reason has told both parties to keep quiet. So the one remedy left, for ending the quarrel for good and to both sides’ satisfaction, is this. They refute each other so well that they must finally be shown they are fighting over nothing.

B 530 They must also see that a certain transcendental illusion has painted before them something actual, where nothing of the kind is to be found.

This is the way we will now take to settle a dispute that cannot be decided.

* * *

Zeno of Elea, a clever dialectician, was already scolded by Plato as a reckless sophist. To show off his art, he would try to prove one statement with arguments that looked sound, and soon afterward overthrow it with others just as strong. He claimed that God (for him, probably nothing but the world) was neither finite nor infinite, neither moving nor at rest, and neither like nor unlike anything else. To the people who passed judgment on him here, it seemed he wanted to deny flat out two statements that contradict each other, which is absurd. I don’t think that charge is fair. I will look at the first of those statements closely in a moment. Take the others. Suppose by “God” he meant the universe. Then he had to say it neither stays put in its place (rest) nor changes that place (motion), since all places exist only inside the universe, so the universe itself is in no place. And if the universe takes in everything that exists, then it also resembles no other thing, and is unlike none.

B 531 That is because there is no other thing outside it to compare it with. When two opposed judgments presuppose a condition that can’t stand, both drop out, despite their conflict (which is not a true contradiction), since the condition on which alone each of the statements was meant to hold falls away.

Suppose someone says: every body either smells good or doesn’t smell good. Then a third case exists, namely that it doesn’t smell at all (it gives off no scent whatever), and so both of the conflicting statements can be false. Now say instead that it is sweet-scented or it is not sweet-scented (vel suaveolens vel non suaveolens). These two judgments are contradictory opposites, and only the first is false. Its contradictory opposite, “some bodies are not sweet-scented,” also covers the bodies that have no smell at all. In the earlier opposition (per disparata, between unlike things) the accidental condition of the concept of body (smell) stayed in the opposing judgment too, so that judgment didn’t cancel it. That is why the second judgment wasn’t the contradictory opposite of the first.

Now take the statement: in space the world is either infinite or not infinite (non est infinitus). If the first is false, its contradictory opposite, “the world is not infinite,” has to be true. All I would be doing is canceling an infinite world, without putting any other world, the finite one, in its place.

B 532 But suppose the statement were: the world is either infinite or finite (non-infinite). Then both could be false. For now I would be treating the world as something whose size is fixed in itself. In the opposition I would not merely be canceling infinity, and with it perhaps the world’s whole separate existence. I would be adding to the world a determination, as if it were a thing actual in itself. That addition can just as easily be false, namely if the world is not a thing in itself in any way, and so has no size of its own, neither infinite nor finite. Let me call this kind of opposition dialectical, and the opposition of contradiction analytic. Then two judgments opposed dialectically can both be false. The reason is that one doesn’t merely contradict the other. It says something more than a contradiction needs.

Now treat these two claims as contradictory opposites: the world is infinite in size, and the world is finite in size. Then we are taking the world (the entire run of appearances) to be a thing in itself. The world stays put whether I cancel the infinite regress or the finite one in the series of its appearances. But take that assumption away, the transcendental illusion, and say the world is not a thing in itself.

B 533 Then the contradictory conflict between the two claims turns into a merely dialectical one. The world doesn’t exist in itself at all, independent of the backward series of my representations. So it exists neither as a whole that is infinite in itself nor as a whole that is finite in itself. It can be found only in the empirical regress through the series of appearances, and by itself not at all. A series that is conditioned at every step is never given complete. The world is therefore not an unconditioned whole, and it doesn’t exist as one either, neither with infinite size nor with finite size.

What I have just said about the first cosmological idea, the absolute totality of size in appearance, is true of all the rest. The series of conditions can be met with only in the regressive synthesis itself. It isn’t found in the appearance on its own account, as a separate thing given ahead of any regress. So I will also have to say that the multitude of parts a given appearance has is neither finite nor infinite in itself. An appearance isn’t anything that exists in itself, and its parts are given first of all through the regress of dividing it up (the decomposing synthesis), and only within that regress. That regress is never given as absolutely complete, as finite or as infinite. The same goes for a series of causes stacked one above the other, or of conditioned things leading up to unconditionally necessary existence.

B 534 Such a series can never be regarded, in itself and in its totality, as either finite or infinite. As a series of subordinate representations it exists only in the dynamic regress. Before that regress, as a series of things standing on its own, it can’t exist in itself at all.

In this way the antinomy of pure reason is lifted for its cosmological ideas. It is shown to be merely dialectical: a conflict produced by an illusion. The illusion arises from using the idea of absolute totality, which is valid solely as a condition of things in themselves, on appearances. Appearances exist only in representation and, when they form a series, only in the successive regress, and otherwise not at all. But we can also turn this around and draw a real benefit from the antinomy. It isn’t a dogmatic benefit; it is critical and doctrinal. We can use it to prove, indirectly, the transcendental ideality of appearances, if someone isn’t convinced by the direct proof given in the transcendental aesthetic. The proof would be this dilemma. Suppose the world is a whole that exists in itself. Then it is either finite or infinite. But the first is false, and so is the second, according to the proofs given earlier for the antithesis and the thesis respectively.

B 535 So the world, the sum of all appearances, can’t be a whole that exists in itself either. It follows that appearances, taken generally, are nothing apart from our representations. That is exactly what we meant by their transcendental ideality.

This remark is important. It shows that the proofs of the fourfold antinomy given above were not deceptive tricks, but sound, on the assumption that appearances, or a world of sense that contains them all, are things in themselves. But the conflict of the propositions drawn from that assumption reveals that the assumption holds a falsehood, and so brings us to a discovery of the true nature of things as objects of the senses. The transcendental dialectic therefore does nothing to help skepticism. It does help the skeptical method, which can point to it as an example of its great usefulness. That happens when the arguments of reason are let loose against each other in their fullest freedom. Even if they don’t in the end deliver what was sought, they will always deliver something useful for correcting our judgments.

B 536 Pure Reason’s Antinomy Section Eight Pure Reason’s Regulative Principle, as It Bears on the Cosmological Ideas

The cosmological principle of totality gives us no maximum for the series of conditions in a world of sense that is treated as a thing in itself. That maximum can only be set as a task within the regress of the series. So this principle of pure reason, once corrected in this way, still keeps its good validity. It is not an axiom telling us to think of the totality as actually there in the object. It is a problem for the understanding, and so for the subject: to start and continue the regress through the conditions of a given conditioned, as the idea’s completeness requires. In sensibility, meaning in space and time, any condition we can reach when we set out given appearances is conditioned in turn. These appearances aren’t objects in themselves, where the absolutely unconditioned could perhaps turn up. They are just empirical representations, and each must find in intuition its condition, which fixes it in space or in time. So the principle of reason is really only a rule, a rule that commands a regress in the series of the conditions of given appearances.

B 537 That regress is never allowed to stop at anything absolutely unconditioned. The principle is therefore not a principle of the possibility of experience or of empirical cognition of the objects of the senses, and so it is not a principle of the understanding. Every experience stays within the boundaries set by the intuition it is given. It is also not a constitutive principle, one that would stretch the concept of the world of sense beyond all possible experience. It is a principle for carrying experience on and widening it as far as possible. By it, no empirical boundary may count as an absolute boundary. So reason’s principle here works as a rule: it lays down what we are to do in the regress. It doesn’t anticipate what is given in the object itself before any regress. That is why I name it a regulative principle for reason. A principle that treated the series of conditions as an absolute totality given in the object (the appearances) itself would, on the other hand, be a constitutive cosmological principle. I wanted to show by this distinction that such a principle is void. I also wanted to prevent what would otherwise happen unavoidably (through transcendental subreption): giving objective reality to an idea that serves only as a rule.

To pin down the sense of this rule of pure reason correctly, we must first note one thing.

B 538 The rule cannot say what the object is. It says how the empirical regress is to proceed, so as to reach the object’s complete concept. If it did say what the object is, it would be a constitutive principle, and pure reason can never yield one of those. So the rule can’t possibly be saying that a given conditioned’s series of conditions is finite in itself, or infinite. If it did, a mere idea of absolute totality, which is created only in itself, would be thinking an object that no experience can give. It would lend a series of appearances an objective reality that doesn’t depend on empirical synthesis. What the idea of reason will do is prescribe a rule to the backward synthesis in the series of conditions. By that rule the synthesis goes from the conditioned through all the conditions subordinate to one another toward the unconditioned, though it never gets there. The absolutely unconditioned is simply not met with in experience.

For this purpose we first have to define precisely what the synthesis of a series is when it can never be completed. People usually use two expressions here that are meant to differ somehow, though no one really knows how to state the ground of the difference. Mathematicians talk only of a progressus in infinitum (a progress to infinity).

B 539 The investigators of concepts (philosophers) will accept only the expression progressus in indefinitum (a progress with no fixed end). I won’t stop to examine the scruple that led them to recommend this distinction, or whether it has been put to good use or none. I will try to fix these concepts exactly for my own purpose.

About a straight line one may rightly say that it can be extended to infinity. Here the distinction between the infinite and an advance of unspecified length (progressus in indefinitum) would be a hollow quibble. Granted, if you say “draw a line further,” it is more accurate to add in indefinitum than in infinitum. The first means only “make it as long as you like”; the second means “never stop making it longer,” which is not what is meant here. But when the question is only what is possible, the first phrase is quite correct, because you can always make the line longer without limit. It is the same in every case where we speak only of the progress, that is, the step from condition to conditioned. In the series of appearances this possible progress goes on to infinity. Starting from a pair of parents, you can go down the line of generation without end, and you can quite well imagine that the line really does go on like this in the world.

B 540 For here reason never needs the absolute totality of the series. It doesn’t presuppose that totality as a condition, as something given (datum). It presupposes only something conditioned, something merely able to be given (dabile), and added on without end.

The task is quite different when the question is how far the regress reaches, the regress that rises from the given conditioned up to the conditions in a series. Can I say that it goes back without end, or only that it reaches back indefinitely far (in indefinitum)? Can I climb to infinity from people now living up through their ancestors? Or can I only say that no matter how far back I have gone, I never meet an empirical reason to think the series stops anywhere? If so, I am entitled, and also bound, to look for a forebear of each ancestor, though I am not really to presuppose one.

Here is my claim. Suppose the whole is given to us in empirical intuition. Then the regress through the series of its inner conditions goes on to infinity. But suppose only a single member of the series is given, and the regress to absolute totality is only now to start from it; then the backward move is merely indefinite (in indefinitum).

B 541 Take the division of a piece of matter (a body) that is given within its boundaries. We have to say it goes on to infinity. That piece of matter is given whole in empirical intuition, and so with all its possible parts. Its whole has the part as its condition, that part has its own part as a condition, and so on. In this regress of decomposition we never meet an unconditioned (indivisible) member of the series. So we have no empirical reason anywhere to stop dividing. What is more, the further members of the division yet to be carried out are themselves empirically given before that further division. In other words, the division goes on to infinity. Compare the line of ancestors leading back from a given person. No possible experience gives it in its absolute totality. The regress still goes from each member of the line to a higher one, and no empirical boundary turns up that would show some member to be absolutely unconditioned. Yet before the regress begins, the members that could supply that condition are not yet part of what we empirically intuit of the whole. So the regress does not go to infinity, as the division of what is given does. It goes on indefinitely far, as we look for further members beyond those given, and those are again always given only as conditioned.

B 542 In neither case, the regressus in infinitum or the one in indefinitum, is the series of conditions treated as given in the object as infinite. They aren’t things in themselves. They are appearances, which as conditions of one another are given only within the regress. So the question stops being how big this series of conditions is in itself, finite or infinite, since in itself it is nothing. The question becomes how we should carry out the empirical regress and how far we should push it. And here the rule for moving forward differs in a way that matters. If the whole has been given empirically, we can go back without end through the series of its inner conditions. If instead the whole isn’t given, and is only to be given by empirical regress, all I can say is this: it is possible to go on without end to ever higher conditions of the series. In the former case, more members always exist, and are empirically given, than my regress (of decomposition) reaches. In the latter I can say that I can always go further in the regress. No member is given empirically as absolutely unconditioned. So a higher member is still possible, and asking after it is still necessary. In the one case it was necessary to find more members of the series. In the other it is always necessary to ask for more.

B 543 The reason is that no experience is absolutely bounded. Either no perception of yours sets an outright boundary to your empirical regress, and then you mustn’t regard your regress as finished. Or you do have a perception that bounds your series. Then that perception can’t belong to the stretch of the series you have already covered (what bounds must differ from what it bounds). So you have to carry the regress on to this condition too, and so on.

The next section will show these remarks in their proper light, by applying them.

Pure Reason’s Antinomy Section Nine The Empirical Use of Reason’s Regulative Principle, as It Bears on All the Cosmological Ideas

We have shown more than once that pure concepts, whether of the understanding or of reason, have no transcendental use. A series of conditions in the world of sense can be absolutely total only on the basis of a transcendental use of reason. Reason demands this unconditioned completeness from whatever it takes to be a thing in itself.

B 544 The world of sense contains nothing like that. So we can no longer talk about the absolute size of the series in the world of sense, about whether they are bounded or unbounded in themselves. The only matter left is how far back we should go in the empirical regress when we trace experience to its conditions. According to the rule of reason, we are to stop at no answer to its questions except one that fits the object.

All that remains is the validity of the principle of reason as a rule for continuing experience and enlarging a possible experience, now that we have shown clearly enough that it is invalid as a constitutive principle about appearances in themselves. If we can put that validity beyond doubt, reason’s quarrel with itself comes to a complete end. Critical resolution doesn’t just remove the illusion that split reason against itself. It also opens up the sense in which reason agrees with itself, a sense whose misreading was the only cause of the quarrel. A principle that would otherwise be dialectical becomes doctrinal. Suppose this principle can be confirmed in its subjective meaning, as a way of fixing how far the understanding may be used in experience at most, so that this use fits the objects of experience. Then the result is exactly the same.

B 545 It would be as if this principle were an axiom (impossible from pure reason) fixing the objects in themselves a priori. Even an axiom, with respect to objects of experience, could have no greater effect on enlarging and correcting our cognition than showing itself at work in the widest empirical use of our understanding.

I. Solving the Cosmological Idea of Totality in the Composition of Appearances Making Up a World-Whole

In this question, as in the other cosmological ones, reason’s regulative principle rests on one proposition: as we regress empirically, we can never experience an absolute boundary, and so never a condition that is empirically absolutely unconditioned. That in turn is because such an experience would have to include a boundary of appearances set by nothing, by the void. The regress, driven forward by a perception, would have to run into that void. And this is impossible.

This proposition amounts to the following. In the empirical regress I never arrive at more than a condition.

B 546 That condition itself has to be seen, once more, as empirically conditioned. So in terminis (in plain words) it contains the rule: no matter how far up the series I have got, I must always ask about a higher member, whether or not experience ever tells me about it.

Now, to settle the first cosmological problem, only one thing is left to decide. When the regress climbs to the unconditioned size of the world-whole (in time and in space), is this climb that never meets a boundary a regress to infinity, or just a regress continued indefinitely (in indefinitum)?

The bare general idea of all the world’s past states in a series, together with the things that exist side by side in space, is just a possible empirical regress. I think of that regress, though so far undetermined. Through it alone does the concept arise of a series of conditions reaching up from the perception we are given*.

B 547 Now, the world-whole is always only in my concept, and never (as a whole) in intuition. So I can’t infer from the world’s size to the size of the regress, or fix the regress by it. I first have to make myself a concept of the size of the world, by way of the size of the empirical regress. About that regress I never know more than this: from whatever member of the series of conditions happens to be given, I have to go on empirically to a higher (more remote) member. So the size of the whole of appearances isn’t fixed absolutely by it. That is also why it can’t be said that this regress goes to infinity. To say so would anticipate the members that the regress hasn’t reached yet, and would picture their number as so large that no empirical synthesis could get to them. It would then determine the world’s size before the regress (though only negatively), and that is impossible. No intuition gives me the world in its totality, so no intuition gives me its size before the regress. We therefore can say nothing at all about the size of the world in itself. We can’t even say that a regressus in infinitum happens in it. All we can do is look for the concept of its size by following the rule that governs the empirical regress in it. That rule says only this: however far we have got in the series of empirical conditions, we are not to assume an absolute boundary anywhere.

B 548 Instead, we have to put every appearance, as conditioned, under another as its condition, and go on further to that one. That is the regressus in indefinitum. It fixes no size in the object, and so it is clearly different from the one in infinitum.

* So this series of the world can be neither bigger nor smaller than the possible empirical regress, the only thing its concept rests on. This regress can give neither a determinate infinite nor, just as little, a determinately finite thing (one that is absolutely bounded). It is clear from this that we can’t take the world’s size as finite, or as infinite, since the regress, by which it is pictured, allows neither.

So I can’t say that the world is infinite, whether in past time or in space. A concept of size like that, as a given infinity, is empirical. It is therefore completely impossible for the world, which is an object of the senses. I also won’t say that the regress from a given perception, through everything that bounds it in a series, in space as well as in past time, goes on to infinity. That would presuppose an infinite size for the world. I won’t claim it is finite either, because an absolute boundary is empirically impossible too. So about the whole object of experience (the world of sense) I will be able to say nothing. I can speak only of the rule by which experience, fitted to its object, is to be carried out and continued.

The first answer to the cosmological question about the size of the world is a negative one. In time the world has no first beginning, and in space it has no outermost boundary.

If it were otherwise, the world would be bounded by empty time on one side and by empty space on the other.

B 549 As an appearance, though, it can’t be either of those in itself, because an appearance is not a thing in itself. So it would have to be possible to perceive such a limit, set by wholly empty time or empty space, and by that perception the world’s ends would be there for a possible experience to meet. But an experience like that, with no content whatever, is impossible. An absolute boundary of the world is thus impossible empirically, and so impossible outright*.

* You will notice that the proof here is built in a quite different way from the dogmatic proof given earlier, in the antithesis of the first antinomy. There we accepted the common, dogmatic way of picturing the world of sense: as a thing already given, in itself and as a complete whole, before any regress. And we denied it any definite place in time and space at all, if it didn’t fill all of time and every space. So the inference was different from the one here. There we inferred that the world is actually infinite.

This also gives the affirmative answer. The regress through the series of world appearances, as a way of fixing the world’s size, goes on in indefinitum. That is to say, the world of sense has no absolute size. The empirical regress does have its rule, and only through that regress can the world be given as far as its conditions go. The rule says: from each member of the series, as something conditioned, always go on to one still more remote.

B 550 That can happen through one’s own experience, or by the thread of history, or along the chain of effects and their causes. And never excuse oneself from widening the possible empirical use of the understanding. That is also the real and only business of reason in its principles.

This doesn’t prescribe a particular empirical regress that goes on forever within one kind of appearance. For example, it doesn’t say that from a living person you must always climb a line of ancestors without expecting a first pair. Nor does it say that in the line of heavenly bodies you must never allow an outermost sun. All it commands is the move from appearance to appearance, even if these gave no actual perception (when a perception is too weak in degree for our awareness to make it experience), because they belong to possible experience all the same.

A beginning is always in time, and the boundary of anything extended is always in space. But time and space exist only in the world of sense. So appearances are conditioned within the world, but the world itself is neither conditioned nor bounded in an unconditioned way.

For that reason, and since the world can never come to us whole, nor can even the conditions of a given conditioned be given whole as a world-series, we get the concept of how big the world is through the regress only.

B 551 It isn’t given in advance, in an all-at-once intuition. The regress, though, consists only in fixing a size. So it gives no determinate concept, and no concept of a size that would be infinite relative to a given measure. It doesn’t go on to the infinite, as if that were given. It goes on indefinitely far, to supply a size (of experience) that becomes actual only through this regress.

II. Solving the Cosmological Idea of Totality in the Division of a Whole That Is Given in Intuition

Say I divide a whole that is given in intuition. I am then going from something conditioned back to the conditions that make it possible. Dividing the parts further (subdivisio or decompositio) is a step back through the series of those conditions. That series would be given as an absolute totality only if the regress could end at simple parts. But if decomposing never stops, and every part can always be divided again, then the division goes on in infinitum. This division is the regress from the conditioned to its conditions. It has no end, because the conditions (the parts) lie inside the conditioned itself.

B 552 The conditioned is given whole in an intuition. That intuition is shut in between its boundaries, and so all the conditions are given with it too. This means we can’t just call the regress a going back in indefinitum. That was the most the earlier cosmological idea allowed. There the conditions lay outside the thing I started from, so they were not given along with it, and I had to move on to them. They were added only as the empirical regress went forward. Still, we may not say that a whole divisible to infinity is made up of infinitely many parts. Each part is included in the intuition of the whole, but the complete division is not. The division consists only in the continuing decomposition, the regress itself, and the regress is what first turns the series into something actual. Because the regress is endless, each member (each part) that it reaches is held in the given whole as an aggregate. The entire series of division is not. It is endless by succession and is never complete. So it can’t present an infinite multitude, and it can’t present a gathering of that multitude into one whole.

This general reminder is easiest to apply first to space. Any space that is intuited within its boundaries is a whole of this kind. Its parts, however far it is decomposed, are always spaces again. So it can be divided to infinity.

B 553 The second application follows from this quite naturally. It concerns an outer appearance that is shut in by its boundaries: a body. Whether a body can be divided depends on whether space can, since space is what makes a body possible as an extended whole. So a body can be divided to infinity, and yet it does not consist of infinitely many parts.

It does seem that a body must differ from space in the law of divisibility, because a body has to be represented in space as a substance. We might grant that however far decomposition goes in space, it can never take away all composition. For then even space as a whole would cease to exist, space which otherwise has nothing self-standing about it (and that cannot happen). But it seems that the concept of substance can’t be reconciled with the claim that nothing at all would remain once all composition of matter is removed in thought. Substance is supposed to be the subject of all composition. Its elements would have to stay behind, even if the connection that makes them a body in space were lifted. But the substance we speak of in appearance is not like this. It isn’t what we would think a thing in itself to be, through a pure concept of the understanding. It is no absolute subject.

B 554 It is a lasting image in sensibility, nothing but intuition, and no unconditioned is met with anywhere in intuition.

The rule of going on to infinity certainly holds when an appearance is subdivided, as long as the appearance is a mere filling of space. But it fails once we try to carry it over to the multitude of parts that are already marked off in some way inside the given whole. Parts marked off like this make a quantum discretum. Suppose every articulated (organized) whole had parts that are themselves articulated, so that when we break the parts down to infinity we keep finding new, artfully made parts. In short, suppose the whole were articulated to infinity. That can’t be thought at all. What can be thought is that the parts of matter might become articulated as they are decomposed to infinity. Here is why. A given appearance in space can be divided without end for one reason only. The appearance supplies divisibility and nothing more, that is, a multitude of parts that is, taken by itself, completely undetermined. The parts themselves get given and determined only by the act of subdividing. The whole, in other words, isn’t already divided in itself. So division can fix a multitude in it, and that multitude reaches exactly as far as we choose to carry the regress of division. An organic body articulated to infinity is another matter.

B 555 Its concept already presents the whole as divided. Before any regress of division, it would hold a multitude of parts that is fixed in itself and yet infinite. That is a contradiction. The endless unfolding would have to count as a series with no possible end (infinite), and in the same breath as a series that is finished once its members are taken together. Infinite division marks an appearance only as a quantum continuum. It can’t be pulled apart from the filling of space, since the ground of endless divisibility lies right there. As soon as something is taken to be a quantum discretum, however, the multitude of units in it is fixed, and it therefore always equals a number. Only experience can decide how far organization runs in an articulated body. Even if experience certainly never reached a part that isn’t organized, such parts must still lie at least within possible experience. But how far an appearance in general can be divided transcendentally is not a question for experience at all. What decides it is a principle of reason: given what this appearance is like, never treat the empirical regress, when the extended is decomposed, as absolutely finished.

B 556 Closing Note on Solving the Mathematical-Transcendental Ideas, and an Introductory Note on Solving the Dynamical-Transcendental Ideas

Earlier we set out the antinomy of pure reason in one table, covering all the transcendental ideas. We named the source of the conflict and the one way of ending it, which was to declare both of the opposed assertions false. In every case we pictured the conditions as belonging to what they condition by relations of space and time. Common human understanding usually assumes this, and the whole conflict depended on it. So, looked at this way, every dialectical picture of totality was alike. Each concerned the series of conditions behind some given conditioned thing. In each, the condition and the conditioned were connected as members of one series, and so were alike in kind. Then the regress can never be thought of as complete. If it is to be complete, some member that is itself conditioned has to be wrongly taken as the first, and so as unconditioned. The object, the conditioned, may not have been weighed everywhere.

B 557 But the series of conditions behind it was, and only by its magnitude. No compromise could remove the difficulty there. Only cutting the knot altogether could. It was this: reason stretched the series out too long for the understanding, or cut it too short, and either way the understanding could never come up to reason’s idea.

But in all this we overlooked an important difference among the objects. By objects I mean the concepts of the understanding that reason tries to lift up into ideas. Look at the table of categories given earlier. Two of them stand for a mathematical synthesis of appearances, and the other two for a dynamical one. Up to this point the oversight could easily occur. In the general presentation of all the transcendental ideas we stayed among conditions within appearance. In just the same way the two mathematical-transcendental ideas had for us no object but one within appearance. Now we turn to the dynamical concepts of the understanding, as far as they are meant to fit reason’s idea. At this point the distinction matters. It opens a wholly new view of the dispute in which reason is tangled. We threw that dispute out earlier, since each side rested on a false assumption.

B 558 The dynamical antinomy, though, may contain an assumption that can live alongside reason’s claim. Then the dispute could be looked at from this angle. The judge would fill in the missing grounds in law that both sides had failed to see, and the parties could be reconciled to the satisfaction of both. In the mathematical antinomy this could not be done.

It is true that the series of conditions are all alike, if we look only at how far each extends, that is, whether it matches the idea or the idea turns out too large or too small for it. The concept of the understanding that lies under these ideas, however, can contain two different things. It can contain only a synthesis of things of the same kind. Every magnitude is assumed to have this, both when it is put together and when it is divided. Or it can contain a synthesis of things of different kinds too. At least dynamical synthesis can allow this, when a cause is combined with its effect, and also when the necessary is combined with the contingent.

So when series of appearances are connected mathematically, the only condition that can enter is a sensible one, which means one that is itself a part of the series. A dynamical series of sensible conditions can take a further condition of a different kind. That condition is no part of the series. It is merely intelligible, and so it lies outside the series.

B 559 This satisfies reason. The unconditioned is placed before the appearances, and the series of appearances, which is always conditioned, is neither muddled by it nor, against the principles of the understanding, cut off.

The dynamical ideas allow a condition of the appearances outside their series. Such a condition is not an appearance itself. This leads to a result quite unlike that of the antinomy. The antinomy made us declare both dialectical counter-assertions false. Now the thoroughly conditioned character of the dynamical series, inseparable from them as appearances, can be connected with a condition that is non-sensible and, empirically, unconditioned. It then does justice to the understanding on one side and to reason on the other*. The dialectical arguments, which in some way or other sought unconditioned totality in mere appearances, fall away.

B 560 The propositions of reason, with their sense set right in this manner, can then both be true. This can never happen with the cosmological ideas that deal only with mathematically unconditioned unity. With these, the only condition to be found for the series of appearances is one that is an appearance as well, and so a member of the series itself.

* For the understanding permits no condition among appearances that could itself be empirically unconditioned. Now imagine that we could conceive, for something conditioned (in appearance), an intelligible condition that doesn’t belong to the series of appearances as one of its members, and that doesn’t break into the series of empirical conditions in the least. Such a condition could then be allowed as empirically unconditioned, and the continuous empirical regress would suffer no interruption anywhere.

III. Solving the Cosmological Ideas of Totality in Deriving World Events from Their Causes

We can think of only two kinds of causality for what happens: one working by nature, the other from freedom. Causality by nature connects a state in the world of sense with an earlier state, and the state follows that earlier one by a rule. But the causality of appearances depends on conditions of time. If the earlier state had always been there, it could not have brought about an effect that begins only in time. So the cause of whatever happens or comes to be has a causality that itself came to be, and under the principle of the understanding it needs a further cause.

B 561 By freedom, in the cosmological sense, I mean the power to start a state on its own. The causality of such a power is not itself placed, under natural law, beneath another cause that fixes it in time. Taken this way, freedom is a pure transcendental idea. For one thing, it holds nothing borrowed from experience. For another, no experience can give its object in a determinate form. The reason is a universal law, which holds even for the possibility of any experience at all: whatever happens must have a cause, and that includes the causality of a cause that has itself happened or come to be. So the entire field of experience, however far it extends, gets changed into a mere sum of nature. But by this route we never get an absolute totality of conditions for causal relations. Reason therefore produces its own idea of a spontaneity, one that could start acting on its own. No other cause need be sent ahead to set it acting under the law of causal connection.

It is highly striking that the practical concept of freedom rests on this transcendental idea. In the practical concept, the transcendental idea is the real core of the difficulties that have always hung over the question of how freedom is possible.

B 562 Freedom in the practical sense means that the power of choice is independent of compulsion by the impulses of sensibility. For a power of choice counts as sensible when it is affected pathologically, that is, by motives that come from sensibility. It is called animal choice (arbitrium brutum) when it can be compelled pathologically. The human power of choice is an arbitrium sensitivum, true, but it is not brutum. It is liberum, since sensibility does not make its action necessary. Human beings have within them a capacity to determine themselves, independent of compulsion by sensible impulses.

One can see easily that if all causality in the world of sense were mere nature, each event would be fixed by another event in time, by necessary laws. Appearances, in so far as they fix the power of choice, would have to make each action necessary, as its natural outcome. Doing away with transcendental freedom would therefore destroy all practical freedom as well. For practical freedom assumes that something that did not happen still ought to have happened. Its cause in appearance was therefore not so decisive as to rule out a causality in our power of choice. That causality works independently of the natural causes, and even against their power and influence it can bring about something that the empirical laws fix in the order of time. It can thus start a series of events entirely by itself.

B 563 This is what happens in any conflict of a reason that dares to go beyond the boundaries of possible experience: the task turns out to be transcendental, not physiological. The question how freedom is possible does bother psychology. Because it depends on dialectical arguments from pure reason alone, though, both the question and its answer fall to transcendental philosophy, and to nothing else. That philosophy can’t turn down a satisfying answer here. To equip it for giving one, I first have to explain more closely, in a remark, how it should proceed on this task.

Suppose appearances were things in themselves, and space and time were the forms in which things in themselves exist. Then conditions and conditioned would always be links in one and the same series. Here, as with every transcendental idea, an antinomy would follow: the series would inevitably come out too long for the understanding, or too short. The dynamical concepts of reason, which occupy us in this number and the next, are special in one way. They concern an object’s existence, not the object as a magnitude. So we may leave aside how long the series of conditions is. In them, only one thing matters.

B 564 It is the dynamical relation between condition and conditioned. The difficulty then shows up already in the question of nature and freedom. Is freedom possible in the first place? If so, can it live alongside the universal natural law of causality? Put differently, is it a sound either/or to say that every effect in the world has to come from nature or else from freedom? Or might both hold at once for a single event, taken in different relations? The principle that all events of the world of sense are thoroughly connected under unchangeable laws of nature is already settled as a principle of the Transcendental Analytic. It stays unharmed. What remains to ask, then, is this. Could freedom still apply to the very same effect that nature determines? Or does that unbreakable rule shut freedom out completely? Here the common but deceptive assumption that appearances are absolutely real quickly does its damage, by confusing reason. For if appearances are things in themselves, there is no rescuing freedom. Nature would then be the complete cause of every event, fully sufficient by itself to determine it. The condition of the event would lie only in the series of appearances, and these are necessary under every law of nature, along with their effect.

B 565 If instead appearances count for only what they truly are, namely representations tied together by empirical laws and not things in themselves, then they must have grounds of their own that are not appearances. Such an intelligible cause is not determined by appearances as to its causality. Its effects, however, do appear, and so other appearances can determine them. The cause and its causality therefore lie outside the series, whereas the effects are met with in the series of empirical conditions. So the effect may be viewed as free with respect to its intelligible cause, while with respect to the appearances it may at the same moment be viewed as following from them by the necessity of nature. When stated in general and in the abstract, this distinction is bound to seem extremely subtle and obscure. Applying it will make it clear. All I wanted to point out here is the following. The thorough connection of all appearances in one context of nature is a law that allows no exception. If someone held on stubbornly to the reality of appearances, that law would have to wipe out all freedom. That is why people who hold the common view on this point have never managed to unite nature with freedom.

B 566 The Possibility of Causality through Freedom, Joined With the General Law of Natural Necessity

I call intelligible whatever in an object of the senses is not itself an appearance. Now suppose that something we must regard as an appearance in the world of sense also has a power in itself. This power is no object of sensible intuition, yet through it the thing can be a cause of appearances. Then the causality of such a being can be looked at in two ways. Its action, as the action of a thing in itself, is intelligible. Its effects, as an appearance in the world of sense, are sensible. So for what such a subject can do, we would form two concepts of its causality, an empirical one and an intellectual one, and both apply to one and the same effect. To think of an object of the senses as having a power with two sides like this conflicts with no concept that we need to have of appearances or of possible experience. For appearances are not things in themselves, so a transcendental object has to underlie them, one that determines them as mere representations. Nothing stops us, then, from attributing a causality to this transcendental object.

B 567 That causality would not be appearance. It would come in addition to the property by which the object appears. And its effect is nevertheless met with in appearance. Each cause that acts has to have a character. By that I mean a law of its causality, and without it the cause wouldn’t be a cause at all. In a subject of the world of sense we would therefore have, first, an empirical character. Through it, the subject’s actions, being appearances, would hang together with other appearances by steady laws of nature. They could be derived from those appearances as their conditions, and in combination with them they would be members of one single series of the natural order. Second, we would have to grant the subject an intelligible character too. Through it the subject causes those actions in their form as appearances, but this character is bound by no condition of sensibility, and is no appearance either. The first could be called the character of the thing in appearance, the second its character as a thing in itself.

According to its intelligible character, this acting subject would be under no conditions of time. Time is a condition only of appearances, not of things in themselves. No action would arise or pass away in it.

B 568 It would therefore not be subject to the law governing every determination in time and everything that changes, the law that whatever happens has its cause in the appearances (in the previous state). To put it briefly, its causality, to the extent that it is intellectual, would have no place in the series of empirical conditions that make events in the world of sense necessary. We could never know this intelligible character directly, since we can perceive only what appears. Yet we would have to think of it in agreement with the empirical character. We do the same when we place a transcendental object beneath appearances in our thinking, even though we have no knowledge of what that object is in itself.

As for its empirical character, then, this subject would be an appearance, subject to every law that fixes things by causal combination. In that respect it would be only a part of the world of sense. Its effects would pour out of nature with no exception, as with any other appearance. To the extent that outer appearances act upon it, and its empirical character (meaning the rule by which it acts as a cause) is known from experience, all its actions would have to be explainable by laws of nature. Whatever is required to determine them completely and necessarily would be found in a possible experience.

B 569 By its intelligible character, though, the same subject would have to be released from all influence of sensibility and from all determination by appearances. (All we can have of this character is its general concept.) In the subject, as noumenon, nothing happens. No change takes place there that would demand a dynamical determination in time. Hence no connection with appearances as causes turns up in it. So this acting being would be independent and free, in its actions, of all natural necessity, which is found only in the world of sense. It would be entirely right to say that it starts its effects in the world of sense by itself, although the action does not start in the being itself. This remains true even though the effects in the world of sense need not for that reason start by themselves. In the world of sense, those effects are always fixed in advance by empirical conditions of the preceding time, and only by way of the empirical character, which is just the appearance of the intelligible one. They are possible only as the continuation of the series of natural causes. So freedom and nature, each in its full sense, would exist together in the same actions, without any conflict, according to whether we compare the actions with their intelligible cause or their sensible one.

B 570 An Explanation of the Cosmological Idea of Freedom as Joined to Universal Natural Necessity

I decided to draw a rough sketch of how our transcendental problem is solved before anything else. It lets us survey more easily the path reason takes in solving it. Next we’ll set out the decisive points, the ones that really matter, and take up each of them by itself.

There is a law of nature that everything that happens has a cause. The causality of this cause, its action, comes earlier in time. Since an effect has come into being, the action can’t have gone on forever. It must itself have happened, and so it has its own cause among the appearances, which determines it. All events are therefore empirically determined within an order of nature. This law alone lets appearances form a nature at all and become objects of experience. It is a law of the understanding, and on no pretext may we depart from it or exempt any appearance from it. To do so would put that appearance outside possible experience altogether.

B 571 It would be set apart from every object of possible experience, and become a mere thought-thing, a creature of the brain.

True, all this looks like nothing but a chain of causes that allows no absolute totality in the regress toward its conditions. But this worry doesn’t hold us up at all, because it was already removed in the general assessment of the antinomy of reason, when reason aims at the unconditioned in the series of appearances. If we give in to the deception of transcendental realism, neither nature nor freedom is left. Here the only question is this. Grant that we recognize nothing but natural necessity in the whole series of all events. Is it still possible to regard one and the same event, which on one side is a mere effect of nature, as on the other side an effect from freedom? Or is a flat contradiction found between these two kinds of causality?

Nothing among the causes in appearance can start a series absolutely, from itself. Any action, as appearance, that brings about an event is itself an event, a happening. Such an event presupposes a prior state where its cause is found. In this way everything that happens only continues the series. No beginning that comes about by itself is possible within it.

B 572 All the actions of natural causes, in their order in time, are in turn effects, and they equally presuppose their causes in the series of time. From the causal connection of appearances, we cannot expect an original action, by which something happens that wasn’t there before.

Is it truly necessary, though, that the causality of a cause must be merely empirical, if the effects are appearances and the cause is itself an appearance? Isn’t it possible after all that each effect in appearance does require a connection with its cause under laws of empirical causality, and yet this empirical causality is itself the effect of a causality that isn’t empirical but intelligible? It could be so without in the least interrupting its tie to the natural causes. That would be an action of a cause that is original in regard to appearances. To that extent the cause is not appearance, but intelligible by its power, though in all other respects it must be counted as a link in the chain of nature, and so as belonging to the world of sense.

To look for natural conditions of natural events, and to name them, we need the principle that appearances cause one another. These conditions are causes within appearance. If this principle is granted and no exception weakens it, the understanding gets all that it can demand.

B 573 In its empirical use it sees only nature in every event, and it is entitled to do so. Physical explanations then proceed without any obstacle. Now take the following supposition: some of the causes in nature have a power that is only intelligible. What moves such a power to act never depends on empirical conditions. It depends only on grounds of the understanding. (Grant, for the rest, that this may be a mere invention.) Even so, the action this cause produces among the appearances must conform to every law of empirical causality. The supposition does the understanding no harm at all. In that case the acting subject, called causa phaenomenon, would be tied into nature through all its actions, with no break in its dependence. Only this subject’s phaenomenon would hold certain conditions (together with everything the subject causes in appearance). These conditions would have to count as merely intelligible for anyone who wants to climb up from the empirical object toward the transcendental one. For as long as we follow nature’s rule in whichever thing among the appearances serves as the cause, we can stay unconcerned about whatever ground of these appearances, and of how they hang together, may be thought to lie in the transcendental subject. That subject is unknown to us empirically. This intelligible ground leaves the empirical questions untouched. It may concern only thinking in the pure understanding.

B 574 Pure understanding thinks and acts, and the effects of that do turn up among the appearances. Still, they must be completely explainable from their cause in appearance, through laws of nature. The way to do that is to keep to the merely empirical character as the highest ground of explanation, and to pass over the intelligible character entirely as unknown. The intelligible character is the transcendental cause of the empirical one. We mention it only to the extent that the empirical character points to it as its sensible sign. Let us apply this to experience. Human beings are among the appearances of the world of sense. To that extent they are among the natural causes, and their causality has to follow empirical laws. As natural causes they must also have an empirical character, as every other natural thing does. We notice the character through the forces and powers a thing shows in its effects. With lifeless nature, or with nature that is alive only in the way animals are, we have no ground to think of any power as anything but conditioned by the senses. A human being is different. Humans know the whole rest of nature only through their senses, but they know themselves too, through mere apperception. They do this in actions and inner determinations that they can’t possibly count as impressions of the senses. To themselves, humans are phenomena on one side. On the other side, namely for certain powers, they are purely intelligible objects.

B 575 That is because the action of those powers can’t be counted as part of sensibility’s receptivity. We call these powers understanding and reason. Reason especially is properly and sharply set apart from every force that depends on empirical conditions. It considers its objects solely in terms of ideas, and it determines the understanding by them. The understanding then uses its concepts empirically, though they too are pure.

The imperatives we set up as rules for the active powers, throughout the practical realm, make it clear that reason has some causality, or that we at least picture it so. “Ought” expresses a kind of necessity, and a connection with grounds, found nowhere else in nature. Where nature is concerned, the understanding can know only what is, what has been, or what will be. Nothing in nature can possibly be other than it is in all these time-relations. When our eyes are only on the course of nature, “ought” simply means nothing. It makes no sense to ask what should happen in nature, just as it makes none to ask what properties a circle should have. The question is what does happen there, or what properties the circle does have.

“Ought” points to a possible action, and the ground of that action is nothing more than a concept. Take a mere action of nature instead.

B 576 Its ground has to be an appearance every time. It’s true that an action which the “ought” is aimed at must be possible under natural conditions. But those conditions do not bear on how the power of choice is determined. They bear only on what it brings about, and how, in appearance. Natural grounds may push me to will something, and there may be any number of sensible attractions. They still cannot produce the “ought.” They can produce only a willing that is anything but necessary and is always conditioned. Against that willing the “ought” spoken by reason sets a measure and an aim, and even a prohibition and an authority. It makes no difference whether the object belongs to mere sensibility (the pleasant) or to pure reason (the good). Either way, reason won’t give way to a ground that is given empirically. It doesn’t follow the order of things as they show themselves in appearance. In complete spontaneity, reason sets up its own order for itself, shaped by ideas. It makes the empirical conditions fit into that order. By that order it even declares actions necessary that haven’t happened, and perhaps never will. Of all of them, though, it presupposes that reason could have causality over them. Otherwise it wouldn’t expect its ideas to have any effects in experience.

Let’s stay with this point. Suppose, at least as a possibility, that reason really has causality over appearances.

B 577 Then reason, for all that it is reason, has to display an empirical character. Every cause presupposes a rule, by which certain appearances follow as its effects. Every rule needs the effects to be uniform. That uniformity is what grounds the concept of a cause (as a power). Because this concept has to be made clear from appearances alone, we can call it the empirical character. It stays constant, whereas the effects show up in shifting forms, depending on the conditions that accompany them and partly restrict them.

This means every person has an empirical character of choice. That character is nothing but a certain causality of their reason, to the extent that reason shows a rule in its effects among the appearances. By this rule we can infer the grounds of reason and its actions, their kind and degree, and form a judgment on the subjective principles of the person’s power of choice. That empirical character has to be taken from the appearances themselves, as their effect, and from their rule, which experience hands us. So everything a person does in appearance is determined, in the order of nature, by their empirical character together with the other causes that help produce it. Suppose we could investigate all the appearances of their power of choice.

B 578 Suppose we could follow them right down to the ground. Then there would be not one human action that we couldn’t predict for certain and see as necessary from its earlier conditions. Regarding this empirical character, then, there’s no freedom. It is the only character through which we can consider a human being, when we simply observe and try to investigate the driving causes of their actions physiologically, as anthropology does.

Now consider the very same actions in relation to reason. I don’t mean speculative reason, in order to explain where they come from. I mean reason alone, as the cause that produces them. In a word, we compare the actions with reason for a practical purpose. Then we find a rule and an order that are entirely different from the order of nature. For in that order, perhaps, none of what has happened by the course of nature should have happened, though by its empirical grounds it could not fail to happen. At times, however, we find (or think we find, at least) that the ideas of reason have actually shown causality in human actions as appearances. Those actions happened, we find, not because empirical causes determined them but because grounds of reason did.

B 579 Suppose we can say that reason has causality over appearance. Could its action be called free, then, when it is fixed precisely and necessarily in its empirical character, its way of sensing? That character is in turn fixed in the intelligible character, its way of thinking. We don’t know this second one. We mark it by appearances, and these, strictly speaking, reveal directly only the way of sensing, the empirical character*. Now take the action insofar as it is to be put down to the way of thinking, which is its cause. Even then it doesn’t follow from that cause by empirical laws. That is, it is not that the conditions of pure reason come first. Only the effects of pure reason, in the appearance of inner sense, come first. As a purely intelligible power, pure reason isn’t subject to the form of time, and so not to the conditions of temporal succession. The causality of reason in the intelligible character doesn’t come into being, and doesn’t start at some moment to produce an effect.

B 580 For if it did, it would itself be bound by the natural law of appearances, to the extent that this law sets causal series in time. Its causality would then be nature and not freedom. So we may say: if reason can have causality over appearances, it is a power through which the sensible condition for an empirical series of effects begins for the first time. The condition in reason isn’t sensible, so it does not itself start. Then we get what was missing in every empirical series. The condition for a series of events following one another can itself be unconditioned empirically. For here the condition stands apart from the series of appearances, in the realm of the intelligible, and so it is subject to no sensible condition and to no time-determination by a cause that comes before it.

* So the real morality of actions (merit and guilt) stays entirely hidden from us, even in the case of our own conduct. What we hold ourselves or others accountable for can be referred only to the empirical character. How much of it is the pure effect of freedom, though, and how much belongs to mere nature and to faults of temperament that aren’t anyone’s fault, or to a fortunate makeup of temperament (merito fortunae), nobody can work out. For that reason nobody can pass judgment with full justice either.

Still, in another respect this same cause also belongs to the series of appearances. People are themselves appearances. Their power of choice has an empirical character, and this is the (empirical) cause of everything they do. Every condition that determines a person according to this character belongs to the series of natural effects and obeys the law of that series. By that law, no empirically unconditioned causality is found in anything that happens in time. So no given action can begin absolutely by itself.

B 581 After all, it can be perceived only as appearance. But of reason we can’t say that another state precedes the one in which it determines the power of choice, a state in which the first state is itself determined. For reason is no appearance and is not subject to any conditions of sensibility. So no succession in time occurs in it, even regarding its causality. The dynamical law of nature, which sets succession in time by rules, can’t therefore be applied to reason.

Reason is therefore the lasting condition of all the voluntary actions in which a human being appears. Each of them is fixed beforehand in the person’s empirical character, before it happens. In the intelligible character, of which the empirical one is only the sensible schema, there is no earlier or later. Whatever the time relations that an action has to other appearances, it flows directly from the intelligible character of pure reason. Pure reason, then, acts freely. Within the chain of natural causes, no ground that comes earlier in time, outer or inner, determines it dynamically. This freedom shouldn’t be viewed only negatively, as independence from empirical conditions (that would stop reason from being a cause of appearances).

B 582 It can also be described positively, as a power to start a series of events by itself. Nothing starts in reason itself. Being the unconditioned condition of each voluntary action, it allows no earlier condition above it in time. Its effect, though, does start in the series of appearances. It can never make an absolutely first beginning there.

An example from the empirical use of reason can illustrate its regulative principle. It is meant only to illustrate, not to confirm, since this kind of proof is no good for transcendental claims. Take a voluntary action, say a malicious lie by which someone has thrown society into some confusion. First we look into the motives out of which the lie came about. Then we judge how the lie, with its consequences, can be blamed on that person. For the first purpose, finding out the motives, we trace the person’s empirical character back to its sources. We look for them in a poor upbringing, in bad company, and partly in the wickedness of a temperament that feels no shame. Part of it we put down to frivolity and thoughtlessness. In doing so, we don’t lose sight of the chance occasions that set the lie off. In all of this we go about it as we do in any investigation of the chain of determining causes behind a given effect in nature.

B 583 We may believe that all this determines the action. Even so, we blame the person who did it. We don’t blame them for an unlucky temperament, or for the circumstances that acted on them, or even for the way they lived before. For we assume that we can set aside entirely how that life went. We assume we can treat the earlier series of conditions as if it had never happened, and treat this deed as wholly unconditioned by the earlier state, as though the person began a series of consequences entirely on their own. The ground of this blame is a law of reason. In it we look on reason as a cause that could have determined the person’s conduct differently, and should have, whatever the empirical conditions named above. And we see the causality of reason not merely as one contributor among others. We see it as complete in itself, even if the sensible drives were not for the action at all and were, in fact, against it. The action is charged to the liar’s intelligible character. At the moment of the lie, the guilt is wholly theirs. So reason was fully free, whatever the empirical conditions of the deed, and the deed is to be put wholly down to reason’s failure to act.

It is easy to see what this judgment of blame has in mind. It has in mind that sensibility doesn’t affect reason at all. Reason doesn’t change.

B 584 Its appearances do change, in other words, how it appears in what it does, but reason itself does not. No state in it comes before the next and determines it. So it doesn’t belong in the series of sensible conditions that make appearances necessary by laws of nature. Reason is present in all the actions of a human being, in all circumstances of time, and is the same in all of them. But it is not itself in time, and it doesn’t fall into some new state it wasn’t in before. In relation to that state it is determining but not determinable. So the question can’t be why reason failed to determine itself otherwise. The only question can be why its causality did not determine the appearances in some other way. And to that no answer is possible. For with another intelligible character we would have been given another empirical one. When we say that the liar could still have refrained from the lie, in spite of their whole earlier way of life, we mean only this. The act stands directly under the power of reason. Reason’s causality is not tied to any condition that belongs to appearance or to the passing of time. Time does mark a main difference when appearances are compared with each other. Yet it cannot change anything about how the action stands to reason, because appearances are not things, and so are not causes in themselves.

B 585 In judging free actions, then, as to their causality, we can get as far as the intelligible cause but no further. We can recognize that it may be free, meaning that sensibility does not decide it, and that in this way it may be the condition of appearances which has no sensible condition behind it. But why the intelligible character gives just these appearances and this empirical character in the circumstances at hand goes beyond anything our reason can answer. It even goes beyond reason’s right to ask. It is like asking why the transcendental object behind our outer, sensible intuition yields intuition in space only, and none of any other kind. But the task we had to solve doesn’t oblige us to do this at all, for it was only this: can freedom and natural necessity clash within a single action? We have answered that well enough. We showed that freedom can refer to a quite different kind of conditions than natural necessity does, so the law of natural necessity doesn’t touch freedom. The two can therefore both be present, independent of each other and undisturbed by each other.

* * *

We must note that we have not tried to show that freedom is real.

B 586 We did not try to show it as one of the powers that contain the cause of the appearances of our world of sense. For one thing, that would not have been a transcendental inquiry, which handles concepts alone. For another, it couldn’t have succeeded in any case, since from experience we can never infer to something that need not be thought by the laws of experience at all. Nor have we tried even to prove that freedom is possible. That too would have failed, because from concepts a priori alone we can’t know how any real ground or causality is possible. Here freedom is treated only as a transcendental idea. By means of it, reason thinks it can open the series of conditions among appearances with an absolute beginning, using something that has no sensible conditions. In doing so, reason gets itself tangled in an antinomy with its own laws, which it prescribes for the empirical use of the understanding. We could do just one thing, and it was the only thing we cared about: to show that this antinomy rests on mere illusion, and that nature at any rate does not clash with causality from freedom.

B 587 IV. Solving the Cosmological Idea of Totality in the Dependence of Appearances as to Their Existence in General

In the previous number we looked at the changes of the world of sense in their dynamical series, where each change stands under another as its cause. Now this series of states serves us only as a guide, to arrive at an existence that might be the supreme condition of all that changes, namely the necessary being. Here the issue is not an unconditioned causality. It is the unconditioned existence of substance itself. So the series before us is really only a series of concepts and not of intuitions, to the extent that one is the condition of the other.

It is easy to see, though, the following. All that is contained in the sum of appearances can change, and so is conditioned as to its existence. That means the series of dependent existence can have no unconditioned member, none whose existence would be necessary without qualification. Suppose, then, that appearances were things in themselves. For that very reason a condition would always belong to the same series of intuitions as what it conditions.

B 588 In that case a necessary being could never exist as the condition of the existence of the appearances of the world of sense.

The dynamical regress, though, has something peculiar to it that sets it apart from the mathematical one. The mathematical regress deals only with putting parts together into a whole, or breaking a whole down into its parts. So the conditions in its series must always count as parts of the series, hence as of the same kind, and so as appearances. In the dynamical regress it is otherwise. There, the question is not whether an unconditioned whole could be made from given parts, or an unconditioned part found for a given whole. The question is how a state comes from its cause, or how the contingent existence of substance itself comes from a necessary one. There the condition need not form an empirical series with the conditioned.

So in the apparent antinomy before us, one way out is still open. Both of the conflicting propositions can be true at once, in different respects. All things of the world of sense are altogether contingent, and so always have only empirically conditioned existence. Yet there can still be a non-empirical condition for the whole series, that is, an unconditionally necessary being. For such a being, as an intelligible condition, would not belong to the series as one of its members. It would not belong even as the highest member.

B 589 Nor would it turn any member of the series into an empirically unconditioned one. The entire world of sense would stay as it is, empirically conditioned in its existence through every member. This way of placing an unconditioned existence beneath the appearances therefore differs from the empirically unconditioned causality (freedom) of the previous article. In the case of freedom, the thing itself, as a cause (Substantia phaenomenon), still had its place among the conditions of the series; only its causality was conceived as intelligible. Here, the necessary being has to be conceived as lying wholly outside the series of the world of sense (as ens extramundanum), and as merely intelligible. This is the only way to prevent it from falling itself under the law of contingency and dependence that governs all appearances.

For our present task, then, the regulative principle of reason is this. Everything in the world of sense has empirically conditioned existence. Nowhere in it is there an unconditioned necessity as to any property. For every member of the series of conditions we should always expect an empirical condition in a possible experience, and search for it as far as we can. Nothing gives us the right to trace any existence back to a condition that lies beyond the empirical series, nor to treat an existence inside the series as wholly independent and self-sufficient.

B 590 Yet this in no way denies that the whole series may be grounded in some intelligible being. That being is free of all empirical conditions, and instead holds the ground that makes all these appearances possible.

The aim here is not to prove that some being exists with unconditional necessity. Nor is it meant to base on this the mere possibility of a merely intelligible condition for the existence of the appearances of the world of sense. The point is only this. We restrict reason so that it won’t drop the thread of empirical conditions and wander off into transcendent grounds of explanation, which could never be shown in a concrete case (in concreto). So, from the other side, we must hold the law of the understanding’s purely empirical use within bounds as well. It may not rule on whether things in general are possible, and it may not declare the intelligible impossible just because we can’t use it to explain appearances. What this shows, then, is only that all natural things, along with every one of their (empirical) conditions, can be contingent through and through, and that this fits perfectly well with the optional assumption of a necessary condition, though only an intelligible one. So no true contradiction is to be found between these claims, and both can be true. It may be that a being like this, absolutely necessary and of the understanding only, is impossible in itself.

B 591 Even so, that cannot be concluded from the general contingency and dependence of everything that belongs to the world of sense, or from the principle of never stopping at any single member of the world, so far as it is contingent, to appeal to a cause outside the world. Reason goes its own way in its empirical use and its own special way in its transcendental use.

The world of sense contains nothing but appearances. These are mere representations, which are in turn always conditioned by the senses. Here we never have things in themselves as our objects. So we should not be surprised that we have no right ever to jump, from any member of the empirical series at all, out of the connection of sensibility. We can’t act as if the members were things in themselves that existed apart from their transcendental ground, and that we could leave behind in order to look for the cause of their existence outside them. With contingent things this would indeed eventually have to happen. It need not happen with mere representations of things, though, because their contingency is itself only a phenomenon. That contingency can lead to no other regress than the one that determines the phenomena, that is, the empirical one. But to think of an intelligible ground of the appearances, that is, of the world of sense, and to think of it as freed from the contingency of the world of sense, is not opposed to the unlimited empirical regress in the series of appearances.

B 592 It is not opposed to their thoroughgoing contingency either. This was all we had to do to clear up the seeming antinomy, and there was no other way to do it. For suppose the condition of each conditioned thing, in respect of existence, is sensible. Then it belongs to the series for that reason, and is itself conditioned again. The antithesis of the fourth antinomy shows this. Two outcomes were possible. Reason demands the unconditioned, so either the clash with reason had to stay, or the unconditioned had to be set outside the series, in the intelligible. The necessity of the intelligible requires no empirical condition and permits none, and so it is unconditionally necessary in relation to appearances.

To allow a merely intelligible being leaves the empirical use of reason untouched, that use being concerned with conditions of existence in the world of sense. It keeps going on the principle of complete contingency, moving from empirical conditions to higher conditions, and these are empirical too. Equally, though, this regulative principle does not rule out accepting an intelligible cause that is not in the series, when it is a matter of the pure use of reason (in regard to ends). For there that cause means only the ground that makes the sensible series possible at all, a ground that for us is only transcendental and unknown. This ground exists independently of all the conditions of that series, and is unconditionally necessary in relation to them.

B 593 So it is not at all opposed to the unlimited contingency of the series, or to the regress that never ends in the series of empirical conditions.

Closing Note on the Whole Antinomy of Pure Reason

For as long as our concepts of reason deal only with the totality of conditions inside the world of sense, and with what can be done there to serve reason, our ideas are transcendental yet still cosmological. The situation changes when we locate the unconditioned, which is what we are really after, in something wholly beyond the world of sense, and therefore beyond all possible experience. The ideas then turn transcendent. Their job is no longer only to round off reason’s use on experience. (That use remains an idea we can never carry out in full, though we must keep following it.) The ideas cut loose from that use altogether. They create objects of their own, whose stuff isn’t taken from experience, and whose objective reality doesn’t rest on completing the empirical series. It rests on concepts that are pure and a priori. Such transcendent ideas have an object that is merely intelligible. We are, to be sure, permitted to accept that object as a transcendental object, one about which nothing else is known. But to think of it as a thing that can be fixed by its own distinguishing, inner predicates, we have no grounds of possibility on our side.

B 594 Such grounds would need to be independent of every concept drawn from experience. We also lack even the faintest warrant for accepting an object like that. It is therefore just a thing of thought. Even so, one cosmological idea above the rest drives us to venture this step: the one that gave rise to the fourth antinomy. For the existence of appearances has no ground in itself, and is always conditioned. This calls on us to look about for something that differs from every appearance, an intelligible object, where this contingency comes to an end. Now, suppose we have once allowed ourselves to assume a reality standing on its own outside the whole field of sensibility. Suppose we have also allowed ourselves to regard appearances only as chance ways in which beings who are themselves intelligences represent intelligible objects. Then we have nothing left but analogy. We use it to apply concepts from experience, and so form some sort of concept of intelligible things, though we know nothing at all of them in themselves. Contingent things are known to us only through experience. But the things in question here are not supposed to be objects of experience at all. So what we know of them has to be drawn out of whatever is necessary in itself, out of the pure concepts that apply to things as such. That is why our first step outside the world of sense puts us under a constraint.

B 595 Our new knowledge has to begin with an inquiry into the absolutely necessary being, and the concept of that being must give us the concepts of every other thing, to the extent that those things are merely intelligible. We’ll make this attempt in the next chapter.

Transcendental Dialectic, Second Book Third Chapter The Ideal of Pure Reason

Section One On the Ideal in General

We saw above that pure concepts of the understanding can’t represent any objects when every condition of sensibility is taken away. What makes them objectively real is then missing, and all that can be found in them is the bare form of thinking. Yet they can be presented in concreto, in a real case, once they are applied to appearances. For appearances give them what they lack: the material for a concept of experience. And a concept of experience is nothing other than an understanding concept in concreto. Ideas reach even less far toward objective reality than the categories do. For no appearance exists in which an idea could be pictured in concreto.

B 596 Ideas carry a kind of completeness that no empirical cognition we could ever have would reach. Reason, in dealing with them, has only systematic unity in view. It tries to bring the unity that experience makes possible closer to that systematic unity, and never fully reaches it.

The ideal, as I use the word, seems to stand even further from objective reality than the idea does. I take the idea there not just in concreto but in individuo, meaning as one individual thing that the idea alone can determine, or has determined.

Take humanity in all its perfection. It includes the full extension of all the essential properties that belong to human nature and make up our concept of it, carried to complete agreement with the ends of that nature. This would be our idea of perfect humanity. But it includes something more besides. It includes everything that, apart from this concept, belongs to the complete determination of the idea. For only one out of each set of opposed predicates can fit the idea of the most perfect human. What for us is an ideal was for Plato an idea in the divine understanding, an individual object of that understanding’s pure intuition. It was the most perfect specimen of each kind of possible being, and the first ground of all the copies found in appearance.

B 597 Let’s not go so far, though. We have to grant that human reason holds ideals as well as ideas. Plato’s were creative, and these are not, but they do have practical force (as regulative principles), and they underlie the possibility that certain actions can be perfect. Moral concepts aren’t completely pure products of reason, since beneath them lies something empirical, namely pleasure or displeasure. Still, they make good examples of pure rational concepts in respect of the principle by which reason sets limits to freedom, which is lawless in itself, that is, when we attend only to their form. Virtue is an idea, and so is human wisdom in complete purity. The Stoic’s sage, though, is an ideal: a human being who exists only in thought but who fits the idea of wisdom completely. The idea gives the rule. In this case, the ideal is the original pattern for fully determining the copy. We have no yardstick for our actions other than how this godlike human being within us behaves. We compare ourselves with that figure, judge ourselves by it, and improve in that way, though we can never get there. One may be unwilling to grant these ideals objective reality (existence). Still, they are not to be dismissed as fantasies spun in the head. They provide reason with a measure it can’t do without.

B 598 Reason needs the concept of whatever is wholly complete of its kind, so that it can gauge the degree and the shortcomings of what is incomplete. To realize the ideal in some example, meaning in appearance, the way a novelist might portray the sage, is not feasible. It also has something absurd about it, and is not very uplifting. The natural limits keep cutting into the completeness in the idea, which makes any illusion impossible in such an attempt. And that makes even the good in the idea look suspicious, and like a mere invention.

That is how things stand with reason’s ideal. It always has to rest on determinate concepts, and it has to work as the rule and original pattern, whether we use it to guide our conduct or to judge. Things are very different with the creations of the imagination. No one can explain them or give an intelligible concept of them. They resemble monograms: single lines that follow no rule anyone could state. Put together, they form less a definite image than a sketch that floats, as it were, in the midst of different experiences. Painters and physiognomists say they have such things in their heads. They are supposed to be a shadowy picture, impossible to hand on, of what these people produce or judge. People could name them ideals of sensibility, though the name fits only loosely. The reason is that they are meant to serve as the model, out of reach, for possible empirical intuitions.

B 599 And yet they give no rule that could be explained and tested.

What reason wants from its ideal, in contrast, is complete determination by rules a priori. For this reason it conceives an object that is to be completely determinable by principles. Experience, though, lacks the conditions sufficient for that, so the concept itself is transcendent.

Third Chapter Second Section On the Transcendental Ideal (Prototypon transzendentale)

A concept is undetermined in regard to anything that isn’t contained in it. It falls under the principle of determinability, which says that of any two contradictory predicates, opposed to each other, only one can belong to the concept. This principle rests on the law of contradiction. That makes it a purely logical principle. It leaves out all the content of cognition and looks at nothing but its logical form.

Every thing, however, is also subject, as regards its possibility, to the principle of complete determination.

B 600 According to this principle, of all the possible predicates of things, when these are set against their opposites, one must belong to it. The law of contradiction alone doesn’t ground this. For the principle looks beyond the relation between two conflicting predicates. It also views each thing in relation to the whole of possibility, which is the whole collection of predicates for things as such. By assuming that sum as an a priori condition, the principle represents every thing as getting its own possibility from the part it has in that whole.* So the principle of complete determination is about content, not only logical form. It is the principle of synthesis for all the predicates that are supposed to make up the complete concept of a thing. It isn’t just the principle of analytic representation through one of two opposed predicates. And it carries a transcendental presupposition.

B 601 That presupposition is the matter for all possibility, and this matter is supposed to supply, a priori, the data for the special possibility of every thing.

* This principle thus relates every thing to a shared correlate, the whole of possibility. Suppose that whole (that is, the stuff for all possible predicates) turned up in the idea of a single thing. It would then demonstrate an affinity among everything possible, because the ground of its complete determination would be identical. The determinability of any concept falls under the universality (Universalitas) of the principle that excludes a middle between two opposed predicates. The determination of a thing falls under the allness (Universitas), the sum total of all possible predicates.

“Everything that exists is completely determined.” This statement does not only mean that for each pair of given opposed predicates, one belongs to the thing. It also means that one of all possible predicates always belongs to it. The statement doesn’t compare predicates only with each other, logically. It sets the thing itself beside the whole collection of possible predicates, and does so transcendentally. In effect it says that knowing a thing completely requires knowing everything possible, and determining the thing by that, one way or the other, affirmatively or negatively. Complete determination is therefore a concept that we can never display in concreto, in its totality. It is based on an idea that has its seat in reason alone. This idea prescribes to the understanding the rule for using itself completely.

Take the sum total of all possibility, as it underlies the complete determination of every thing. This idea is itself still undetermined as to the predicates that might make it up. All we think by means of it is the collected whole of possible predicates in general. A closer look shows, however, that as a primitive concept the idea excludes many predicates. Some are already given, as derived from others.

B 602 Some can’t stand together. The idea refines itself until it is a concept completely determined a priori. In this way it becomes the concept of an individual object, completely determined by the mere idea, and so it has to be called an ideal of pure reason.

Now weigh all possible predicates, not merely in a logical way but in a transcendental one, meaning by the content that each of them lets us think a priori. We then find that some represent a being and others a mere not-being. Logical negation is indicated only by the small word “not.” It never really attaches to a concept itself, but only to the relation between that concept and another within a judgment. So it falls far short of marking a concept with respect to its content. The expression “not-mortal” can’t show that a mere not-being in the object is meant. It leaves the content wholly untouched. Transcendental negation means not-being in itself. Its opposite is transcendental affirmation, a something whose concept already expresses a being. For this reason it is called reality (thinghood). Only through it, and only as far as it reaches, are objects something (things). The negation opposed to it means a mere lack.

B 603 Where negation alone is thought, the cancellation of all thinghood is represented.

Nobody can think a negation determinately unless the opposed affirmation underlies it. A person born blind can have no idea at all of darkness, since they have none of light. A savage can have none of poverty, since they know nothing of prosperity.* Someone ignorant has no concept of their ignorance, since they have none of science, and so on. So concepts of negations are also all derived. The realities supply the data and, so to speak, the matter, or the transcendental content, for the possibility and complete determination of all things.

* The observations and calculations of astronomers taught us much to admire. Probably the most important thing, though, is that they exposed to us the abyss of ignorance. Without this knowledge, human reason could never have imagined that abyss to be so great. Reflecting on it must make a large difference to how we fix the final aims of our use of reason.

Now suppose reason takes a transcendental substratum as the basis of complete determination. This substratum is, as it were, the entire supply of stuff, and every predicate that things can possibly have is taken from it. Nothing but the idea of reality as an All (omnitudo realitatis) is this substratum.

B 604 Every true negation, then, is simply a limit. That word wouldn’t fit if the unlimited (the All) did not lie underneath.

To possess all of reality in this way is also to represent the concept of a thing in itself as completely determined. An ens realissimum, a most real being, is conceived as one single being. The reason is that, of all the possible opposed predicates, one turns up in its determination: the one that belongs to being without qualification. So it is a transcendental ideal. It underlies the complete determination that is necessarily found in everything that exists, and it is the highest and complete material condition of that thing’s possibility. All thinking of objects in general, as far as its content goes, has to be traced back to this condition. It is also the only true ideal that human reason is capable of. For only in this one case is a concept that is general in itself determined completely through itself, and recognized as the representation of an individual.

Reason determines a concept logically by means of a disjunctive syllogism. The major premise contains a logical division, a splitting-up of the sphere of a general concept. The minor premise cuts that sphere down to a single part.

B 605 The conclusion then settles the concept by that part. A general concept of reality as such can’t be split up a priori. The reason is that without experience we know of no particular kinds of reality that could fall under that genus. Thus the transcendental major premise behind the complete determination of all things is simply the representation of the totality of all reality. It is not merely a concept that has all predicates under it, as to their transcendental content. It is a concept that includes them within itself. The complete determination of each thing rests on narrowing down this All of reality. Some of it is assigned to the thing, and the rest is excluded. That fits the “either … or” found in the disjunctive major premise, and fits the minor premise too, which fixes the object by one member of the division. So the use reason makes of the transcendental ideal, as the basis on which it determines all possible things, is analogous to its way of proceeding in disjunctive syllogisms. This was the proposition on which I earlier built the systematic division of all transcendental ideas. They arise in step with the three kinds of syllogisms, each matching one.

It goes without saying that reason has an aim here: merely to represent the necessary complete determination of things.

B 606 For this aim it takes for granted not that a being exists that matches the ideal, but only the idea of such a being. It wants to derive the conditioned kind of complete determination, the kind found in limited things, from a totality of complete determination that is unconditioned. So reason sees the ideal as the original model (Prototypon) for all things. Those things, as defective copies (ectypa), take the material for their possibility from it. They come more or less close to it, yet they always fall infinitely short of reaching it.

In this way the possibility of all things (the synthesis of the manifold, as to its content) counts as derived. Only the possibility of that which holds all reality inside it counts as original. The reason: every negation is a mere limitation of some larger reality, and in the end of the highest reality. (Negations are the only predicates that set everything else apart from the most real being.) So they presuppose that reality, and as to content they are only derived from it. The many different things amount to nothing but just as many different ways of restricting the concept of highest reality, which is the substratum they share. Figures work the same way: each is possible only as one of the different ways of restricting infinite space. So the object of reason’s ideal, which exists only in reason, is called the original being (ens originarium). Insofar as it has none above it, it is called the highest being (ens summum). Everything stands under it as conditioned.

B 607 In that respect it is called the being of all beings (ens entium). None of this describes how an actual object relates to other things. It describes how an idea is related to concepts. And it leaves us wholly ignorant whether a being of such outstanding rank exists.

Also, an original being cannot be said to be made up of many derived beings, because every one of those presupposes it and so can’t constitute it. This means the ideal of the original being must be thought of as simple as well.

Strictly speaking, then, deriving all other possibility from this original being cannot be seen as a limiting of its highest reality, or as a kind of division of it. For then the original being would be seen as a mere aggregate of derived beings. By what was said just now, that is impossible, although in the first rough outline we did picture it that way. Instead, all things would owe their possibility to the highest reality as a ground, and not as a sum total. On that view the variety of those possibilities depends not on any restriction of the original being itself but on everything that follows from it. All of our sensibility belongs among those consequences, together with all the reality in appearance. Such reality cannot be a component of the idea of the highest being.

B 608 Suppose we follow our idea further in this way, by hypostasizing it. We will then be in a position to determine the original being, from the bare concept of the highest reality, as single, simple, all-sufficient, everlasting, and so forth. In a word, we can determine it in its unconditioned completeness through all the predicaments. The concept of such a being is the concept of God, thought in the transcendental sense. That makes the ideal of pure reason the subject matter of a transcendental theology, just as I said earlier.

This use of the transcendental idea, however, would already go beyond the boundaries of its purpose and of what is allowed. For reason put the idea, as the concept of all reality, only at the base of the complete determination of things in general. It did not require that all this reality be given objectively and itself make up a thing. That second thing is a mere fiction. In it we gather the manifold of our idea together into an ideal, as a particular being, and make it real. We have no authority for this. We have no authority even to accept the possibility of such a hypothesis outright. Besides, whatever follows from such an ideal does not touch the complete determination of things in general at all. The idea was needed for that purpose alone, and the consequences have not the slightest effect on it.

B 609 It’s not enough to describe how our reason proceeds and its dialectic. We must also try to find its sources, so that we can account for this illusion itself, the way we would for a phenomenon of the understanding. For the ideal we’re talking about is founded on a natural idea, not a merely arbitrary one. So I ask how reason comes to treat the possibility of all things as derived from a single basic one, that of the highest reality. And how does it then go on to presuppose this reality as contained in a particular original being?

The answer comes of its own accord out of the Transcendental Analytic. Objects of the senses are possible by standing in a relation to our thinking. In that relation one thing, the empirical form, can be thought a priori. The matter is another thing: reality in appearance, which corresponds to sensation. It has to be given. Without it nothing could be thought, and the object’s possibility couldn’t be represented. An object of the senses is completely determined only when we compare it with every predicate of appearance and represent it through these, by affirming or denying them. Yet what constitutes the thing itself in appearance, the real, must be given in the comparison, or the thing could not be thought.

B 610 And the real of all appearances is given in one all-embracing experience. So for objects of the senses to be possible at all, their matter must be presupposed as given in one totality. Only by restricting this totality can empirical objects be possible, differ from each other, and be completely determined. Now only objects of the senses can actually be given to us, and only in the context of a possible experience. So nothing counts as an object for us unless, as the condition of its being possible, it presupposes all empirical reality taken together. A natural illusion leads us to count this as a principle valid for all things whatsoever, though it holds only for things given as objects of our senses. We drop the restriction, and so we take the empirical principle behind our concepts of how things are possible, as appearances, to be a transcendental principle of how things in general are possible.

We then go on to hypostasize this idea of the totality of all reality. It happens like this. When the understanding is used on experience, its unity is distributive. We dialectically change this into the collective unity of an experiential whole. In this whole of appearance we think of a single thing that holds all empirical reality in itself.

B 611 Then the transcendental subreption mentioned above does its work. By it, that thing is confused with the concept of a thing at the head of the possibility of all things, which supplies the real conditions for their complete determination.*

* This ideal of the most real being is only a representation. Even so, it is first made real, that is, turned into an object. Then it is hypostasized. Finally, as reason naturally moves on toward completing the unity, it is even personified, as we will soon show. This happens because the unity that regulates experience does not depend on the appearances themselves (on sensibility alone). It depends on the understanding connecting their manifold (in one apperception). Hence it seems that all things are completely determinable (possible) in a highest understanding, and that the highest reality has its unity there too. In other words, both seem to lie in an intelligence.

Third Chapter Third Section On the Proofs That Speculative Reason Offers for Concluding That a Highest Being Exists

Reason is under strong pressure to presuppose something that might lie fully at the base of the understanding’s complete determination of all the concepts it uses. Yet reason notices much too easily how ideal such a presupposition is, how much it is only a fiction, something merely invented.

B 612 That alone could not talk it into accepting a mere creature of its own thinking, straight away, as an actual being. It would do so only if something else drove it to look somewhere for a place to rest, in the regress from the conditioned that is given to the unconditioned. In itself, and by its bare concept, the unconditioned is not given as actual. Still, only the unconditioned can bring the series of conditions, followed back to their grounds, to completion. This is the natural course that every human reason takes, even the most ordinary, although not every one holds out in it. Reason does not start from concepts but from common experience, and so lays something existing at its base. This ground sinks, though, if it is not resting on the unshakeable rock of what is absolutely necessary. The rock itself, in turn, floats unsupported if empty space remains outside it and below it, and if the rock does not fill everything itself and so leave no further room for a “Why?”, in other words, if it is not infinite in reality.

If anything at all exists, whatever it may be, we must also grant that something exists necessarily. For the contingent exists only on the condition of another thing, as its cause. The same inference applies to that cause, and so on, until we reach a cause that is not contingent, and for that very reason is there, necessarily and without condition. This is the argument on which reason bases its advance to the original being.

B 613 Now reason looks around for the concept of a being that would suit so great a privilege of existence as unconditioned necessity. Its aim is not to infer from that concept, a priori, the being’s existence. (If reason dared to do that, it could research among mere concepts and would not need to put a given existence at its base.) Its aim is only to find, among all the concepts of possible things, the one that has nothing in it which conflicts with absolute necessity. For that something must exist with absolute necessity it regards as already settled, by the first inference. If, then, reason can remove everything that does not fit with this necessity, except for one thing, that one thing is the absolutely necessary being. That is so regardless of whether we are able to grasp its necessity, in the sense of deriving it from its concept alone.

Now take what contains in its concept the “Because” for every “Why,” what is defective in no part and in no respect, and what suffices everywhere as a condition. That is exactly why it looks like the being fit for absolute necessity. Since it possesses all the conditions for everything possible, it needs no condition itself, and is not even capable of one. So at least in one respect it satisfies the concept of unconditioned necessity.

B 614 No other concept can match it in this respect. Every other concept is lacking and needs something added, so none shows this sign of independence from further conditions. We cannot, it’s true, conclude for certain from this that something which does not contain the highest, fully complete condition must for that reason be conditioned in its own existence. But it does lack the one mark of unconditioned existence that reason has at its command for cognizing, through a concept a priori, any being as unconditioned.

Among everything we can conceive as possible, the concept of a being with the highest reality would fit best with the concept of a being that is unconditionally necessary. Even if it falls short of satisfying that concept completely, there is no alternative. We have to stay with it, since we can’t just ignore the existence of a necessary being. And if we grant that existence, we can find nothing in the whole field of possibility that could lay a better-founded claim to such a privilege in existence.

This, then, is how the natural course of human reason is made. First reason convinces itself that some necessary being exists. In that being it recognizes an unconditioned existence. Then it looks for the concept of something independent of every condition.

B 615 It finds that concept in what is itself the sufficient condition for everything else, that is, in what contains all reality. An All without any limits is absolute unity, and that brings with it the concept of a single being, the highest being. Reason therefore concludes that the highest being, being the primal ground of all things, exists absolutely and of necessity.

You can’t deny that this concept has a certain solidity when the question is one of making a decision. Suppose we’ve admitted that some necessary being exists, and we agree that we have to pick a side on what to identify it with. Then we couldn’t choose anything more fitting. In fact we’d have no choice at all. We’d be obliged to give our vote to the absolute unity of complete reality, since all possibility comes from it in the first place. But suppose nothing pushes us to decide. Suppose we’d rather leave the whole matter open until the full weight of the evidence compels our agreement. That is, suppose we only want to judge how much we know about this problem, and how much we merely flatter ourselves that we know. Then the argument above looks nothing like so favorable. It needs some indulgence to make up for having no rightful claim of its own.

Let’s allow everything to stand as it is put here.

B 616 First, from some given existence (perhaps only my own) we can validly infer the existence of an unconditionally necessary being. Second, I have to treat a being that contains all reality, and so every condition too, as absolutely unconditioned. So the concept of the thing that suits absolute necessity has been found. Even so, we can’t conclude from this that the concept of a limited being, one without the highest reality, therefore conflicts with absolute necessity. I may not find in its concept the unconditioned, which already carries the whole of the conditions with it. But I can’t infer from that that its existence must be conditioned. It’s the same in a hypothetical syllogism. I can’t say: where a certain condition is missing (here, completeness according to concepts), the conditioned thing is missing as well. We remain free to treat all other limited beings as just as unconditionally necessary, even though the general concept we have of them gives us no way to conclude that they are necessary. But then this argument would not have given us the faintest concept of what a necessary being is like. It would have achieved nothing at all.

Even so, this argument keeps some weight and standing, which its objective inadequacy can’t remove at once.

B 617 Suppose there are obligations that are perfectly sound in the idea of reason, but that would have no reality when applied to ourselves, that is, no motives to act on, unless a highest being were presupposed to give the practical laws effect and force. Then we would also be obliged to follow those concepts which may not be objectively sufficient but which, by the measure of our reason, carry more weight. We know nothing better or more convincing to compare with them. Here the duty to choose would tip the scales of speculation, which can’t settle the matter, by adding a practical weight. Reason couldn’t even justify itself before itself, though it is the most forgiving judge, if, under pressing motives, it had not followed these grounds of judgment, even though its insight was imperfect. And we know no better grounds.

This argument is transcendental in fact, because it rests on the inner inadequacy of whatever is contingent. Yet it is so simple and natural that ordinary common sense takes to it, as soon as it is pointed that way. We see things change, come to be, and pass away. So they, or at least their state, must have a cause. But the same question can be asked again of every cause that might ever be given in experience.

B 618 Where would it be fairer to place the supreme causality than at the place where the highest causality already is? That place is the being that has within itself, from the start, all it takes to bring about the possible effect. Its concept is also very easy to form, by the one feature of an all-embracing perfection. We then hold this highest cause to be necessary without qualification, because we find it absolutely necessary to climb up to it, and see no reason to go further past it. That’s why we see sparks of monotheism shining through the blindest polytheism of every people. It wasn’t reflection or deep speculation that led them there. It was only the natural course of common understanding, which gradually became clear to them.

Speculative reason can offer just three kinds of proof that God exists.

Every route that might be taken to this end starts in one of three ways. It might start from determinate experience, and from the particular character of our world of the senses that this experience reveals. From there it climbs, following the laws of cause and effect, up to a highest cause beyond the world. Or it might use only indeterminate experience, some existence or other, as its empirical basis. Or finally it might put aside all experience and argue entirely a priori, from mere concepts, that a highest cause exists.

B 619 The first is the physicotheological proof, the second the cosmological proof, the third the ontological proof. There are no others, and there can’t be any more.

I’ll show that reason gets nowhere on the one path (the empirical) any more than on the other (the transcendental). It spreads its wings in vain, trying to rise above the world of the senses by the mere power of speculation. As for the order in which these proofs have to be examined, it will be exactly the reverse of the one reason follows as it gradually widens its reach. That is also the order in which we first listed them. For it will turn out that experience gives the first occasion, but only the transcendental concept guides reason in this effort and sets the goal it has in mind in every such attempt. So I’ll begin by examining the transcendental proof. After that I’ll look at how much the empirical element can add to its force when it’s brought in.

B 620 Third Chapter Fourth Section Why There Can Be No Ontological Proof of God’s Existence

From what we’ve said so far, you can easily see that the idea of a being that is absolutely necessary is a pure concept of reason, which is to say a mere idea. The fact that reason needs it is nowhere near enough to prove that it has objective reality. It only points toward a certain completeness, which can’t be reached. And it really serves to set a boundary for the understanding more than to extend it to new objects. Here we run into something strange and absurd. Reasoning from any given existence at all to some existence that is absolutely necessary seems urgent and correct. And yet we have every condition of the understanding against us when we try to form a concept of such necessity.

People have always talked about the absolutely necessary being. They’ve worked less at understanding whether and how one could even think a thing of this kind than at proving that it exists. Giving a verbal definition of this concept is certainly easy: it is something whose nonexistence is impossible.

B 621 But that doesn’t make us any wiser about the conditions that would make it impossible to regard the nonexistence of a thing as completely unthinkable. And those conditions are exactly what we want to know. We want to know whether we are thinking anything at all through this concept, or not. If I throw out, with the word “unconditioned,” every condition the understanding always needs in order to see something as necessary, I’m still far from understanding whether I then think something through the concept of the unconditionally necessary, or perhaps nothing at all.

There’s more. People have ventured this concept at random, and in time it has become quite familiar. On top of that, they’ve believed they could explain it with a whole series of examples, so that any further question about whether it is understandable seemed pointless. Every proposition of geometry is absolutely necessary, for instance that a triangle has three angles. So people spoke about an object that lies completely beyond the reach of our understanding, as though they knew very well what they were saying when they used its concept.

Every example they give, with no exceptions, comes from judgments. None comes from things and whether they exist. Yet a judgment’s unconditional necessity is not the same as absolute necessity in the things themselves. For the absolute necessity of a judgment amounts to just a conditional necessity for the thing, or for the judgment’s predicate.

B 622 The proposition just given didn’t say that three angles are absolutely necessary. It said that if a triangle is there (is given), then three angles are necessarily there too (in it). Still, this logical necessity has shown so great a power of illusion that it has led people astray in the following way. They would form an a priori concept of a thing, and set it up so that, as they saw it, existence was included within its scope. From this they thought they could draw a safe conclusion. Existence belongs necessarily to the object of this concept. That is, if I posit the thing as given (existing), then its existence is also posited necessarily (by the rule of identity). So, they said, this being itself is absolutely necessary, because its existence gets thought along with it in a concept that people simply chose to adopt, provided I posit that concept’s object.

Suppose I remove the predicate in an identical judgment but keep the subject. Then a contradiction arises, and so I say that the predicate belongs to the subject necessarily. But suppose I remove the subject together with the predicate. Then no contradiction arises, because nothing is left that could be contradicted. To posit a triangle and yet remove its three angles is contradictory. But taking away the triangle along with its three angles is no contradiction. The concept of an absolutely necessary being is in exactly the same position.

B 623 If you remove its existence, you remove the thing itself with all its predicates. So where would the contradiction come from? Outside the thing there is nothing to be contradicted, because the thing isn’t supposed to be necessary from outside. And inside there is nothing either, since by removing the thing itself you’ve removed everything inside it at the same time. “God is almighty” is a necessary judgment. Almightiness can’t be removed once you posit a deity, an infinite being, whose concept is identical with it. But if you say “God is not,” then neither almightiness nor any other of his predicates is given. They’re all removed together with the subject, and the thought doesn’t show the slightest contradiction.

So you’ve seen it now: take away a judgment’s predicate and its subject together, and no inner contradiction can ever come up, whatever the predicate happens to be. You have no way out left except to say this: there are subjects that can’t be removed at all, so they have to remain. But that would mean there are absolutely necessary subjects. That is the very assumption whose correctness I doubted, and whose possibility you set out to show me. I have no idea at all of a thing such that removing it, predicates and all, would leave a contradiction behind.

B 624 And without a contradiction, I have no mark of impossibility from mere pure concepts a priori.

You challenge me, against all these general conclusions (which no one can refuse), with a case that you put forward as a proof by deed. You say there is one concept, and only this one, for which the nonexistence or removal of its object contradicts itself. This is the concept of the most real being. It has all reality, you say, and you’re entitled to assume that such a being is possible. (I’ll agree to that for now, although a concept that doesn’t contradict itself is far from proving that its object is possible.)* Now existence is also included in all reality. So the concept of something possible has existence in it.

B 625 If this thing is removed, the thing’s inner possibility is removed, and that is a contradiction.

* A concept is always possible if it doesn’t contradict itself. That is the logical mark of possibility, and it is what sets its object apart from the nihil negativum. But the concept can still be empty. That happens if the objective reality of the synthesis that produces the concept is not specifically shown. As we showed above, this always depends on principles of possible experience, and not on the basic rule of analysis, the law of contradiction. This is a warning not to infer straight away from the possibility of concepts (logical possibility) to the possibility of things (real possibility).

I answer: you’ve already contradicted yourself. You wanted to think of a thing only as to its possibility, and yet you slipped the concept of its existence into the concept of that thing, under some disguised name or other. If we let you have that, you seem to have won the game, but in fact you’ve said nothing. All you’ve done is state a tautology. Let me ask you something. Take any thing you like (I grant you that it is possible, whatever it may be). Is the proposition “it exists” an analytic or a synthetic proposition? If it is analytic, then saying the thing exists adds nothing to what you were already thinking of. But then one of two things must be true. Either the thought in your head is the thing itself, or you took existence to belong to possibility and then pretended to infer it from inner possibility. That is nothing but a wretched tautology. The word “reality” in the concept of the thing sounds different from the word “existence” in the predicate’s concept, but that settles nothing. Suppose you call every act of positing “reality,” whatever it is that you posit. Even then you have already set down the thing, with every one of its predicates, inside the subject’s concept, and treated it as actual. In the predicate you only repeat it.

B 626 On the other hand, you may admit, as every reasonable person fairly has to, that every existential proposition is synthetic. Then how can you maintain that the predicate of existence can’t be removed without contradiction? This privilege belongs only to analytic propositions, whose very character rests on it.

I would have hoped to wipe out this brooding, hair-splitting argument at once, just by defining the concept of existence precisely. But I’ve found that the illusion comes from mistaking a logical predicate for a real one (that is, for a determination of a thing). That illusion almost shuts out any correction. Anything you like can serve as a logical predicate. The subject itself can even be stated as its own predicate, since logic ignores all content. A determination is different. It is a predicate that goes beyond the subject’s concept and makes that concept bigger. So it must not already be contained in that concept.

Being is obviously not a real predicate. In other words, it is not the concept of any extra item that could join the concept of a thing. All it does is posit a thing, or certain determinations, as they are in themselves. In logic it is only the copula of a judgment. Take “God is almighty.” It contains two concepts, each with its own object: God and almightiness. The little word “is” isn’t a further predicate on top of those two.

B 627 It only puts the predicate into its relation to the subject. Now suppose I take the subject (God) together with all his predicates (almightiness is one of them) and say “God is,” or “A God exists.” Then I’m adding no new predicate to the concept of God. I posit only the subject itself with all its predicates, that is, the object in relation to my concept. Both must contain exactly the same thing. So the concept, which expresses mere possibility, gains nothing more when I think of its object as given outright (by saying “it is”). The actual, then, contains nothing more than the merely possible. A hundred real thalers hold nothing at all more than a hundred possible thalers do. The possible ones stand for the concept. The actual ones stand for the object and its positing in itself. Suppose the object had more in it than the concept. Then my concept would fail to express the whole object, and so wouldn’t be the right concept of it. But I am better off, in my financial position, with a hundred actual thalers than with the mere concept of them (that is, their possibility). With an actual thing, the object isn’t just analytically contained in my concept. It joins my concept synthetically, and my concept is itself a determination of my state. Yet those hundred thalers I’ve thought of aren’t increased in the least by this being outside my concept.

B 628 Suppose, then, that I think of a thing through whatever predicates I like, and as many as I like (even in its complete determination). By adding “this thing is,” I add nothing at all to the thing. If I did, then what exists would not be exactly the same thing. There would be more of it than I had in mind when I formed the concept, and I couldn’t say that the very object of my concept exists. Say I even think of a thing as having all reality except one. By saying that such a defective thing exists, I don’t supply the missing reality. It exists with exactly the same defect that I thought it with. If not, something different from what I thought of would be there. Now I think of a being as the highest reality (with no defect). We can still ask whether it exists or not. My concept may be missing none of the real content a thing could have, in general. Still, something is lacking in its relation to my whole state of thinking: namely, that cognition of that object should also be possible a posteriori. And here we can see why this is so hard. Were it an object of the senses we were talking about, I couldn’t confuse the thing’s existence with the mere concept of the thing. By way of the concept, we think of the object merely as fitting the general conditions of any possible empirical cognition. By way of existence, we think of it as lying within the whole context of experience.

B 629 Connecting it with the content of all experience doesn’t enlarge the concept of the object in the least. Our thinking, however, gains one more possible perception through it. But suppose we try to think existence by means of the pure category and nothing else. Then it’s no wonder we can’t name any mark that tells it apart from mere possibility.

So whatever our concept of an object may contain, and however much, we still have to go outside it in order to give the object existence. With objects of the senses, this happens through their link with one of my perceptions, by empirical laws. For objects of pure thought, though, there is no means at all of cognizing their existence, since it would have to be known purely a priori. But all our consciousness of any existence (whether we get it straight from perception or through inferences linking something to perception) is part of the unity of experience, without remainder. An existence outside this field can’t be declared flatly impossible. But it is an assumption that nothing can justify.

The concept of a highest being is, in some respects, a very useful idea. But it is only an idea, and for that very reason it is quite unable, by itself, to extend our cognition of what exists.

B 630 It can’t even teach us anything about the possibility of anything more. Nobody can deny it the analytic mark of possibility, which is that mere positings (realities) produce no contradiction. Yet when all real properties are connected in one thing, that is a synthesis. We can’t judge its possibility a priori, and for two reasons. The realities aren’t given to us specifically. And even if they were, no judgment would be made at all, because the mark of possibility for synthetic cognitions has to be sought in experience alone, and the object of an idea can’t belong to experience. So the famous Leibniz was far from achieving what he flattered himself he could do: to see a priori the possibility of so lofty an ideal being.

All the effort and labor spent on the famous ontological (Cartesian) proof of a highest being’s existence, from concepts, is therefore wasted. A person could grow richer in insight from mere ideas about as little as a merchant could grow richer in wealth by tacking a few zeros onto the cash on hand, hoping to improve their position.

B 631 Third Chapter Fifth Section Why There Can Be No Cosmological Proof of God’s Existence

It was thoroughly unnatural, and a mere novelty of school cleverness, to try to dig out the existence of the matching object itself from an idea sketched quite arbitrarily. In fact, nobody would ever have tried it this way, if something hadn’t come first. Our reason has a need: it wants to accept, for existence in general, something necessary (a point where the climb upward can come to rest). This necessity has to be unconditioned and certain a priori. So reason was forced to look for a concept that would, if possible, meet such a demand and give us cognition of an existence entirely a priori. People believed the idea of a most real being was where to find it. So this idea was used only to know more definitely something that people were already convinced of, or persuaded of, on other grounds: that it must exist. That something is the necessary being. But people hid this natural course of reason. Instead of ending with this concept, they tried to begin with it, hoping to get the necessity of existence out of it, when it was only meant to fill that in.

B 632 This is how the failed ontological proof arose. It offers nothing satisfying to natural, healthy understanding, and nothing satisfying to examination by the rules of the schools.

The cosmological proof, which we’re about to examine, keeps the connection between absolute necessity and highest reality. But the previous proof inferred necessity of existence from highest reality, and this one doesn’t. It infers the unlimited reality of some being from that being’s unconditioned necessity, which is given in advance. In this way it brings everything at least onto the track of a natural kind of inference. I can’t say whether it is sound reasoning or only sophistry. Either way, it persuades most, and not just ordinary understanding but speculative understanding too. It also visibly lays down the first outlines for all the proofs of natural theology. People have always followed those outlines and always will, however much foliage and flourish they use to decorate and hide them. Leibniz also called this proof the one a contingentia mundi (from the contingency of the world). We’ll now set it out and put it to the test.

It runs like this. When something exists, there must also be an absolutely necessary being. Now at least I myself exist. So an absolutely necessary being exists. The minor premise contains an experience.

B 633 The major premise holds the step from any experience at all to the existence of something necessary.* So the proof really begins from experience. It is therefore not carried out wholly a priori, or ontologically. And since the object of all possible experience goes by the name “world,” the proof is named the cosmological one. It also sets aside every particular property of the objects of experience that might distinguish this world from any other possible world. So even its name tells it apart from the physicotheological proof. That proof takes its grounds of proof from observing the particular character of this world of ours, the world of the senses.

* This inference is too well known to need a long statement here. It depends on the so-called transcendental law of nature about causality, that everything contingent has its cause. If that cause is contingent in turn, it must have a cause too. This goes on until the series of causes, each below the next, has to stop at a cause that is absolutely necessary. Without it the series would not be complete.

The proof goes on as follows. There is only one way to determine the necessary being. That is, of all the possible opposite predicates, it can be determined by only one of each pair. So it must be completely determined by its concept. Now there is just one possible concept of a thing that can determine it completely a priori. That is the concept of the entis realissimi (the most real being). So the most real being’s concept is the only one there is for this.

B 634 Through it alone can a necessary being be thought. That is, a highest being exists necessarily.

This cosmological argument brings together so many sophistical principles that speculative reason seems to have called up all its dialectical skill here to produce the greatest possible transcendental illusion. For now, though, we’ll set the examination of it aside for a while. We only want to expose one trick it plays. It dresses an old argument up as a new one, and calls on two witnesses who are supposed to agree. One is a witness of pure reason, the other a witness with empirical credentials. But there is only the first, alone. It has merely changed its clothes and its voice to pass for a second. To lay its foundation securely, this proof stands on experience. That makes it look different from the ontological proof, the one that trusts nothing but pure concepts a priori. Still, the cosmological proof needs this experience for one single step, the step to the existence of some necessary being, any at all. What properties this being has, the empirical ground of proof can’t teach. There reason takes leave of it entirely and investigates among nothing but concepts. The question it asks is which properties any absolutely necessary being would have to have.

B 635 Put differently: of all possible things, which one holds within itself the conditions (requisita) that absolute necessity requires? Reason thinks these requirements are found in one place only: the concept of a most real being. It then concludes: that is the absolutely necessary being. But it’s clear that this assumes something. It assumes that a being of the highest reality, taken as a concept, completely meets what the concept of absolute necessity in existence demands. In other words, it assumes that you can go from the first to the second by inference. That is the claim the ontological argument made. So the cosmological proof accepts it and builds on it, though it had meant to avoid it. For absolute necessity is existence from mere concepts. Suppose I say that the entis realissimi gives us just such a concept, and the only one that suits necessary existence and matches it. Then I must also grant that necessary existence follows from it by inference. So really it is only the ontological proof, from nothing but concepts, that holds all the proving force in the so-called cosmological one. The supposed experience is quite idle. At best it may bring us to the concept of absolute necessity. It can’t demonstrate that necessity in any particular thing. As soon as we have that aim, we have to leave all experience at once and search among pure concepts.

B 636 We have to ask which one of them contains the conditions needed for the possibility of an absolutely necessary being. But once the possibility of such a being has been seen in this way, its existence has been shown too. For that amounts to saying: of all possible things, one carries absolute necessity with it. That is, this being exists absolutely necessarily.

Deceptive tricks in reasoning are easiest to uncover when they’re laid out in proper school form. Here is such a layout.

Suppose this proposition is correct: every absolutely necessary being is at the same time the most real being. (That is the sinew of the cosmological proof, its nervus probandi.) If so, then it must be possible to turn it around, as with all affirmative judgments, at least per accidens. So: some most real beings are at the same time absolutely necessary beings. Now one ens realissimum doesn’t differ from another in any respect. So whatever holds of some of the beings contained under this concept holds of all of them. So here I can turn it around without limiting it: any most real being at all is a necessary being. And this proposition is determined a priori from its concepts alone. So the plain concept of the most real being has to bring its absolute necessity along with it. That is exactly what the ontological proof claimed.

B 637 The cosmological proof refused to acknowledge it, and yet it slipped it in under its conclusions, in a hidden way.

So the second route that speculative reason takes, in order to prove that the highest being exists, misleads just as the first does. It has one more fault besides: it commits an ignoratio elenchi (an argument that misses the point at issue). It promises to lead us along a new footpath. But after a small detour it brings us back to the old one, which we had left on its account.

A moment ago I said that this cosmological argument hides a whole nest of dialectical presumptions, and that transcendental critique can easily find and destroy it. Now I’ll only list them. I’ll leave it to the practiced reader to look further into the deceptive principles and take them apart.

Here we find, for example: 1. The transcendental principle that lets us conclude from the contingent to a cause. It has meaning only in the world of the senses. Outside that world it doesn’t even make sense. The merely intellectual concept of the contingent can’t produce any synthetic proposition, such as the one about causality. The principle of causality has no meaning and no mark of its use except in the world of the senses. Yet here it was supposed to do the job of getting us beyond that world. 2.

B 638 The conclusion that, because a never-ending series of causes stacked one on another is impossible in the world of the senses, there must be a first cause. The principles by which reason is used don’t entitle us to this even within experience. Much less can they let us stretch the principle beyond experience (where the chain can’t be extended at all). 3. Reason’s false self-satisfaction about completing this series. In the end it gets rid of every condition, without which no concept of a necessity can exist. Then, since nothing more can be grasped, it takes this for the completion of its concept. 4. Mixing up two kinds of possibility. One is the logical possibility of a concept that unites all reality (it has no inner contradiction). The other is transcendental possibility. That needs a principle saying such a synthesis can be done, and that principle in turn can only extend as far as possible experience, and so on.

The clever trick of the cosmological proof aims at one thing only: to dodge the a priori proof, through mere concepts, that a necessary being exists. That proof would have to be carried out ontologically, and we feel wholly unable to do that. So we take an actual existence as our basis (an experience in general) and infer from it, as well as we can manage, some absolutely necessary condition of that existence. We then don’t need to explain how that condition is possible.

B 639 Once it’s proved that the condition exists, the question of its possibility is entirely unnecessary. Now suppose we want to say more exactly what this necessary being is like. We aren’t looking for what would be enough to grasp the necessity of its existence from its concept. If we could do that, we’d need no empirical premise. No: all we look for is the condition in the negative sense (conditio sine qua non), the one a being cannot be absolutely necessary without. With any other kind of inference, one that goes from a given consequence back to its ground, this would work well enough. Here, unluckily, it turns out that the condition one demands for absolute necessity is to be found in one single being only. That being would have to hold, within its concept, everything absolute necessity calls for, and so it makes an inference a priori to that necessity possible. That is, I would have to be able to infer in reverse as well: whatever thing this concept (of highest reality) belongs to is absolutely necessary. And if I can’t infer that way (as I must admit if I want to avoid the ontological proof), then I’ve come to grief on my new road too, and I’m back where I started. The concept of the highest being does answer, a priori, every question that can be raised about a thing’s inner determinations, and so it is also an ideal without equal.

B 640 That’s because the general concept marks it out, at the same time, as an individual among all possible things. But it doesn’t answer the question about its own existence at all, and that was what the whole matter was really about. Someone accepted that a necessary being exists and only wanted to know which of all things should be taken for it. You couldn’t answer that person by saying: This here is the necessary being.

It may well be permitted to assume that a being of the highest sufficiency exists, as the cause of all possible effects, so as to make it easier for reason to find the unity of explanatory grounds that it seeks. But it takes far too much on oneself to go further and say, “Such a being exists necessarily.” That is no longer the modest statement of a permitted hypothesis. It is the bold presumption of apodictic certainty. After all, if you claim to cognize something as absolutely necessary, your cognition of it must carry absolute necessity with it.

The whole task of the transcendental ideal comes down to this: either to find a concept for absolute necessity, or to find, for the concept of some thing, that thing’s absolute necessity. If you can do the one, you must be able to do the other. Reason recognizes something as absolutely necessary only when it is necessary just from its concept.

B 641 But both lie wholly beyond our utmost efforts to satisfy our understanding on this point. They also lie beyond every attempt to calm it about its inability to do so.

We can’t do without unconditioned necessity as the last thing that bears all other things. Yet it is the real abyss for human reason. Even eternity, however shudderingly sublime a Haller may paint it, is far from making so dizzying an impression on the mind. Eternity only measures the duration of things. It doesn’t bear them up. We can’t keep the thought away, but we can’t bear it either: that a being we picture as the highest of all possible beings might say to itself, as it were: “I am from eternity to eternity. Apart from me there is nothing at all, except what exists only through my will. But then where do I come from?” Here everything sinks beneath us. The greatest perfection, like the smallest, hovers without support before speculative reason. It costs speculative reason nothing to let the one vanish as readily as the other, without the slightest obstacle.

Many forces of nature show that they exist through certain effects, and yet remain beyond our investigation, because we can’t trace them far enough by observation. The transcendental object underlying appearances is one of these, along with the reason why our sensibility has these highest conditions and not others.

B 642 They are and remain beyond our investigation, although the thing itself is given, only not understood. An ideal of pure reason, though, can’t be called beyond investigation. It has no credentials for its reality except reason’s need to complete all synthetic unity by means of it. Since it isn’t even given as a thinkable object, it isn’t beyond investigation as one either. As a mere idea, it must instead have its seat, and its resolution, in the nature of reason, and so it must be possible to investigate it. For this is exactly what reason is: the ability to give reasons for each of our concepts, opinions and assertions. If they rest on objective grounds, we can give those. If they are mere illusion, we can give the subjective grounds.

Uncovering and Explaining the Dialectical Illusion in All Transcendental Proofs That a Necessary Being Exists.

Both proofs so far were transcendental. That is, they were attempted independently of empirical principles. The cosmological proof does rest on an experience in general. But it isn’t drawn from any particular character of that experience. It comes from pure principles of reason, applied to an existence that empirical consciousness gives us in general.

B 643 It even leaves that guidance behind, to rest on pure concepts alone. Now what is the cause, in these transcendental proofs, of the dialectical but natural illusion that connects the concept of necessity with that of highest reality, and that realizes and hypostasizes what can only be an idea? What is the cause of this unavoidable need to assume something among existing things that is necessary in itself, and yet at the same time to shrink back from the existence of such a being as from an abyss? And how can one bring reason to understand itself on this matter, and to pass from a wavering state of timid approval, always taken back again, to calm insight?

It is something very remarkable that once we suppose something exists, we can’t avoid the conclusion that something also exists necessarily. The cosmological argument rested on this thoroughly natural inference (though that doesn’t make it certain). But whatever concept of a thing I may take up, I find that I can never picture its existence as absolutely necessary. Nothing stops me from thinking its nonexistence, whatever may exist. So I do have to assume something necessary for existing things in general, but I can’t think any single thing as necessary in itself.

B 644 That is, I can’t ever finish going back through the conditions of existence unless I assume a necessary being. But I can’t ever start out from one.

For existing things in general I have to think of something as necessary. But I have no right to think any one thing as necessary in itself. From this it follows unavoidably that necessity and contingency can’t concern and affect the things themselves. Otherwise there would be a contradiction. So neither of these two principles is objective. At most they can be subjective principles of reason. One tells reason to look, for everything given as existing, for something that is necessary. That is, never to stop anywhere except at an explanation completed a priori. The other tells reason never to hope for that completion. That is, to accept nothing empirical as unconditioned, and so to spare itself any further derivation. In this sense both principles can stand side by side quite well, as merely heuristic and regulative principles that look after nothing but the formal interest of reason. For the one says: you should philosophize about nature as if there were a necessary first ground for everything that belongs to existence, only so that your cognition gets systematic unity, by following up an idea of that kind, namely an imagined supreme ground.

B 645 The other, though, warns you against accepting any single determination that concerns the existence of things as such a supreme ground, that is, as absolutely necessary. It tells you to keep the way open for further derivation, and so always to treat such determinations as still conditioned. But if we must treat everything perceived in things as necessary only on a condition, then no thing (anything that may be empirically given) can count as absolutely necessary either.

What follows from this is that the absolutely necessary has to be assumed outside the world. Its only job is to be a principle for the greatest unity appearances can have, their supreme ground. Within the world you can never reach it, because the second rule orders you always to treat every empirical cause of unity as derived.

The philosophers of antiquity regarded every form in nature as contingent, but matter, by the judgment of common reason, as original and necessary. If they had looked at matter not merely in relation to appearances, as their substratum, but as it exists in itself, then the idea of absolute necessity would have disappeared immediately. For nothing binds reason to this existence outright. Reason can always, and without conflict, remove it in thought. And in thought alone lay the absolute necessity. So some regulative principle must have lain at the bottom of this conviction.

B 646 In fact, extension and impenetrability (which together make up the concept of matter) are the supreme empirical principle of the unity of appearances. Insofar as it is empirically unconditioned, this principle has a property of a regulative principle. Still, every determination of matter that makes up its reality, impenetrability included, is an effect (an action) that must have its cause, and so it is always still derived. For that reason matter can’t serve as the idea of a necessary being, a principle behind all derived unity. Each of its real properties, being derived, is only conditionally necessary, and so can be removed in itself. But removing it would remove the whole existence of matter. And if that did not happen, we would have reached the supreme ground of unity empirically, which the second regulative principle forbids. It follows that matter, and in general whatever belongs to the world, does not fit the idea of a necessary original being, when that being is meant only as a principle of the highest degree of empirical unity. Such a being must be placed outside the world. Then we can calmly keep deriving the world’s appearances and their existence from other things, as if there were no necessary being. And we can still strive without end for completeness of the derivation.

B 647 And we can do so while behaving as though we had assumed such a being, a supreme ground.

After all this, the ideal of the highest being turns out to be only a regulative principle of reason. It tells reason to look at every combination in the world as though it came from a necessary cause that is enough for everything. On that, reason can build its rule of systematic unity, necessary by general laws, for explaining that combination. It does not assert any existence that is necessary in itself. But it is at the same time unavoidable that, by a transcendental subreption, we picture this formal principle as constitutive, and think of this unity hypostatically. Take space. It makes all figures originally possible, and these figures are merely different restrictions of it. It is only a principle of sensibility, and yet for that very reason it is taken for something absolutely necessary that exists on its own, and for an object given a priori in itself. In just the same way, it happens quite naturally that the systematic unity of nature can be set up as a principle for the empirical use of our reason only if the supreme cause, the idea of a most real being, is put at the base. So this idea is pictured as an actual object, and that object, being the supreme condition, as necessary. A regulative principle is thereby turned into a constitutive one.

B 648 The substitution shows itself like this. Take this supreme being, which was unconditionally necessary relative to the world, as a thing for itself. Its necessity can’t be grasped through any concept. So the only place it can have turned up in my reason is as a condition of thinking that is merely formal. It cannot be a material and hypostatic condition of existence.

Third Chapter Sixth Section Why There Can Be No Physicotheological Proof

What is demanded can’t be delivered by the concept of things as such, and it can’t be delivered by experience of some existence or other either. One means is left to try. We can see whether a determinate experience might supply a ground of proof that could safely help us reach the conviction that a highest being exists. That would be the experience of the things in the present world, with their nature and arrangement. We would call such a proof the physicotheological one. If this proof turned out to be impossible too, then speculative reason alone has no way at all to give a satisfying proof that a being exists which would match our transcendental idea.

B 649 After all the remarks above, you’ll soon see that the answer to this inquiry can be expected to come easily and briefly. For how could an experience ever be given that would be adequate to an idea? That is just what is peculiar to an idea: no experience can ever fit it. The transcendental idea of an original being, necessary and sufficient for everything, is so excessively large, so far above anything empirical (and the empirical is always conditioned) that two things happen. On one hand, we can never scrape together enough material in experience to fill out such a concept. On the other, we keep groping around among the conditioned, and we will always search in vain for the unconditioned. Not one law of empirical synthesis shows us a case of it, or gives us the slightest guide.

Suppose the highest being had its place in this chain of conditions. Then it would itself be a member of the series. Just like the lower members it is placed above, it would call for further inquiry into a still higher ground. Suppose instead that we cut it off from the chain, and treat it, as a merely intelligible being, as not included in the series of natural causes. What bridge can reason then build to reach it? For all the laws by which we pass from effects to causes, and in fact every synthesis and every widening of our cognition, rest on nothing but possible experience.

B 650 That is, they rest only on objects of the world of the senses, and they mean something only when applied to those objects.

The present world opens before us such an immeasurable stage of variety, order, purposiveness and beauty, whether we trace them out into the endlessness of space or down through its unlimited divisibility, that even given what our weak understanding has managed to learn of it, all language loses its force in the face of so many wonders of such unforeseeable size. All numbers lose their power to measure them, and even our thoughts find no boundary. So our judgment of the whole has to dissolve into a speechless but all the more eloquent amazement. On every side we find a chain of effects and causes, of purposes and the means to them, and regularity in how things arise and pass away. Nothing has come by itself into the state it is in. Each state sends us on to some other thing as its cause, and that cause calls for exactly the same further question. On those terms the whole universe would have to fall into the abyss of nothing, if we did not assume something outside all this endless contingency. It would have to exist originally and independently by itself, hold the contingent up, and, as the cause of its origin, secure its continuing existence as well.

B 651 How great should we think this highest cause is (with respect to all the things of the world)? We don’t know everything the world contains. Still less can we estimate its size by comparing it with all that is possible. Since we do need an ultimate, supreme being for causality, what stops us from also placing it higher than anything else possible in how perfect it is? We can do that easily, although admittedly only by the delicate outline of an abstract concept, if we picture all possible perfection united in it, as one single substance. This concept suits reason’s wish to economize on principles. It is not subject to any contradictions in itself. It even helps widen how reason is used inside experience, because an idea like this points the way toward order and purposiveness. And nowhere does it clearly conflict with any experience.

This proof deserves to be spoken of with respect every time. It is the oldest, the clearest, and the one best suited to ordinary human reason. It gives life to the study of nature, just as it gets its own existence from that study and so keeps gaining new strength. It brings ends and purposes in where our observation would not have found them by itself. And it widens what we know of nature, with a special unity to guide it, a unity whose principle lies outside nature. But this knowledge has an effect on its own cause in turn. That cause is the idea that gave the occasion for it.

B 652 And so it increases belief in a highest author until the belief becomes an irresistible conviction.

So it would be not only dismal but also quite pointless to try to take anything away from the standing of this proof. Reason is lifted up without pause by grounds of proof that are so powerful and keep growing under its hands, even though they are only empirical. No doubts of subtle, abstract speculation can weigh it down so much that one look at nature’s wonders and the grandeur of the world’s structure wouldn’t pull it out of every brooding indecision, as if out of a dream. That look lifts it from one greatness to the next until the very highest, from the conditioned to its condition, until it reaches the supreme author who is unconditioned.

We have no objection to the rationality and usefulness of this way of proceeding. We would rather recommend and encourage it. Still, we can’t approve the claims this kind of proof would like to make to apodictic certainty, or to an assent that needs no favor or outside support. And it can do the good cause no harm to tone down the dogmatic language of a scoffing reasoner to a note of moderation and modesty. That is the note of a faith that is enough to bring calm, though it doesn’t demand unconditional submission.

B 653 So I maintain that the physicotheological proof can never establish the existence of a highest being on its own. It must always count on the ontological proof, of which it is only the introduction, to fill this gap. So the ontological proof is still the one that holds the only ground of proof there can be, if a speculative proof is possible at all, and no human reason can pass it by.

The main points of the physicotheological proof are these. 1. Everywhere in the world there are clear signs of an arrangement made for a definite aim and executed with great wisdom, within a whole that is beyond description in the variety of what it contains and boundless in its extent. 2. The things of the world have nothing of this purposive arrangement in themselves. It clings to them only by contingency. That is, the natures of different things could not have come together on their own, through so many different means working together, toward definite final aims. They could do so only if they had been chosen and set up for exactly that purpose by an arranging, rational principle, according to ideas lying at the base. 3. So there is a sublime and wise cause (perhaps more than one). It has to be the world’s cause, and not just as a blindly working, all-powerful nature acting through fertility, but as an intelligence acting through freedom. 4. We can conclude that this cause is one, from the unity in how the parts of the world are related to each other, as members of an artfully made structure.

B 654 As far as our observation reaches, the inference is certain. Beyond that, by all the principles of analogy, it is probable.

I won’t quarrel here with natural reason about its inference. Natural reason reasons from the analogy between some natural products and what human art makes when it forces nature, compelling it not to follow its own purposes but to bend to ours (the likeness of such products to houses, ships, clocks). It concludes that the same kind of causality, namely understanding and will, must lie at the base of nature. It does this when it derives the inner possibility of nature working freely (nature, which is what first makes all art possible, and perhaps even reason itself) from another art, though a superhuman one. Perhaps the sharpest transcendental critique could not withstand that kind of inference. Still, we have to admit that if we’re to name a cause at all, we can’t proceed more safely here than by analogy with such purposive productions. They are the only ones whose causes and manner of working are fully known to us. Reason could not answer to itself if it wanted to pass from the causality it knows to dark and unprovable grounds of explanation that it doesn’t know.

On this line of reasoning, the fit and purposiveness of so many arrangements in nature would show only that their form is contingent.

B 655 They could not show that the matter is contingent, meaning the substance in the world. For proving that, we would also need to show that the things of the world could not by themselves have fit together in such order and agreement, by general laws, unless they were, even in their substance, the product of a highest wisdom. For that we’d need grounds of proof of a very different kind from the analogy with human art. So the proof could at most establish a world-builder, always very much limited by how suitable the material is that it works on, but not a world-creator, to whose idea everything is subject. That is far from enough for the great aim we have in view, which is to prove an all-sufficient original being. Should we want to show that matter itself is contingent, we’d have to fall back on a transcendental argument. And that is exactly what we were trying to avoid here.

So the inference starts from the order and purposiveness we observe throughout the world, which it treats as an entirely contingent set-up, and concludes that a cause proportioned to it exists. The concept of this cause, however, must let us know something quite definite about it. So it can be no other than the concept of a being that has all power, wisdom and the like, in short every perfection, being sufficient for everything.

B 656 For predicates such as very great, astonishing, immeasurable power and excellence give no definite concept at all. They don’t really say what the thing is in itself. They are only representations of relation, of the size of the object, which the observer (of the world) measures against their own self and their own power to take it in. They come out equally high-sounding whether you make the object bigger or make the observing subject smaller in relation to it. Where it matters how great (in perfection) a thing is, there is no definite concept except the one that takes in all possible perfection. Only the All (omnitudo) of reality is completely determined in the concept.

Now I don’t expect anyone to presume to claim insight into how the size of the world they have observed (in both extent and content) relates to almighty power, how the order of the world relates to supreme wisdom, how the unity of the world relates to the absolute unity of its author, and so on. So physicotheology can’t hand us a definite concept of what the world’s supreme cause is. That means it falls short of being a principle for theology, and theology in turn is meant to be the foundation of religion.

The step to absolute totality is entirely impossible by the empirical path. Yet people do take it in the physicotheological proof.

B 657 What means do they use, then, to get across so wide a gap?

People get as far as marveling at how great the wisdom, the power and so forth of the world’s author are, and can’t go any further. Then they suddenly drop this argument, which was led by empirical grounds of proof, and move to the world’s contingency, which they had inferred at the very start from the order and purposiveness of the world. From that contingency alone they move on, using only transcendental concepts, to something absolutely necessary that exists. Then they go from the first cause’s absolute necessity, as a concept, to the concept that is completely determined, or determining, for it: a concept of all-embracing reality. So the physicotheological proof got stuck in its undertaking. In this predicament it jumped suddenly to the cosmological proof. And since that is only a hidden ontological proof, it really carried out its aim by pure reason alone, although at first it had denied any kinship with pure reason and had staked everything on evident proofs from experience.

So the physicotheologians have no reason at all to act so standoffish toward the transcendental kind of proof, or to look down on it, as clear-sighted students of nature full of self-regard, the way one looks down on the cobweb of gloomy brooders. A little self-examination would show them this: they have covered a good stretch of the way.

B 658 They covered it on the soil of nature and experience, and yet they find themselves still as far away as ever from the object their reason sees ahead. Then they suddenly leave this ground and cross into the realm of mere possibilities, where they hope, on the wings of ideas, to come close to what had escaped all their empirical searching. At last they think they have gained a firm footing through so mighty a leap. They then spread the concept, now definite, over all of creation. (They came to possess it without knowing how.) And they illustrate the ideal, which was only a product of pure reason, through experience, though poorly enough and far below the dignity of its object. All the while they refuse to admit that they reached this knowledge, or this assumption, by any other footpath than that of experience.

So the cosmological proof lies at the base of the physicotheological one, and under the cosmological proof lies the ontological proof, that one original being exists as the highest being. Speculative reason has no way open to it besides these three. So the only proof still possible is the ontological one, built from pure concepts of reason alone, if a proof is possible at all for a proposition raised so far above every empirical use of the understanding.

B 659 Third Chapter Seventh Section Critique of All Theology Drawn from Speculative Principles of Reason

By theology I mean cognition of the original being. It is either theology from mere reason (theologia rationalis) or theology from revelation (revelata). The first thinks of its object in one of two ways. One way is through pure reason alone, by means of purely transcendental concepts (ens originarium, realissimum, ens entium). This is called transcendental theology. The other way is by using a concept taken over from nature (our soul): the concept of the highest intelligence. This would have to be called natural theology. A person who allows only transcendental theology is called a deist. A person who accepts natural theology as well is called a theist. The deist concedes that mere reason might, at the most, let us cognize that an original being exists, but says our concept of it is merely transcendental. That is, it is the concept of a being that has all reality, which can’t be determined any more closely. The theist claims that reason, reasoning by analogy with nature, can pin the object down more closely. It can determine it as a being that contains within itself, through understanding and freedom, the original ground of all other things. So the deist pictures under it merely a cause of the world.

B 660 Whether that cause acts by the necessity of its nature or by freedom is left undecided. The theist pictures an author of the world.

Transcendental theology is of two kinds. One kind intends to get the original being’s existence out of experience taken in general, without settling anything more specific about the world to which that experience belongs. It is called cosmotheology. The other kind believes it can cognize the being’s existence through mere concepts, without the help of the slightest experience. It is called ontotheology.

Natural theology reasons to the properties of an author of the world, and to its existence, starting from the character, order and unity met with in this world. This world calls for us to assume two kinds of causality and the rule for each: nature and freedom. That is why it climbs out of this world up to the highest intelligence, which it takes either as the principle behind all natural order and perfection, or as the principle behind all moral order and perfection. In the first case it is called physicotheology, in the second moral theology.*

* Not theological morality. That contains moral laws that presuppose the existence of a supreme ruler of the world. Moral theology, by contrast, is a conviction that a highest being exists, a conviction grounded on moral laws.

By the concept of God we are used to understanding not just a blind, eternally working nature that things grow from, but a highest being who is supposed to bring things about, as their author, by means of understanding and freedom.

B 661 Only this concept interests us. So, strictly speaking, one could deny the deist any belief in God, and leave the deist only the claim of an original being, or supreme cause. Still, nobody may be accused of wanting to deny something outright just because they don’t dare to assert it. So a milder and fairer way to put it is that the deist believes in a God, while the theist believes in a living God (summam intelligentiam). Next we want to look for where all these attempts of reason could come from.

Here I’ll be content to explain theoretical cognition as the kind by which I cognize what is, and practical cognition as the kind by which I represent to myself what ought to be. By this, the theoretical use of reason is the use through which I come to know a priori, as something necessary, that a thing is. The practical use is the one by which it is cognized a priori what ought to happen. Now suppose that either that something is, or that something ought to happen, is beyond doubt certain but only conditioned. Then there are two possibilities. A certain determinate condition for it may be absolutely necessary. Or it may be presupposed merely as something optional and contingent. When the first holds, the condition gets postulated (per thesin). When the second holds, it gets supposed (per hypothesin). Some practical laws are absolutely necessary, namely the moral ones.

B 662 So suppose these laws must presuppose some existence, as the condition that makes their binding force possible. Then that existence has to be postulated. The reason is that the conditioned thing, the one the inference starts from on its way to this determinate condition, is itself known a priori to be absolutely necessary. Later we will show that moral laws do more than presuppose a highest being’s existence. Since they are absolutely necessary in another respect, they also rightly postulate it, though of course only practically. For now we leave this kind of inference aside.

When we speak only of what is (not of what ought to be), the conditioned that is given to us in experience is always thought of as contingent too. So the condition that belongs to it can’t be cognized from this as absolutely necessary. It serves only as a presupposition that is relatively necessary, or rather needed, but arbitrary in itself and a priori, for the rational cognition of the conditioned. So suppose we want to cognize a thing’s absolute necessity in theoretical cognition. That could happen only from concepts a priori. It could never happen by treating the thing as a cause, in relation to an existence that is given through experience.

A theoretical cognition counts as speculative when it deals with an object, or with concepts of one, that no experience can reach.

B 663 It is the opposite of cognition of nature, which deals only with objects, and their predicates, that a possible experience can give.

Inferring a cause from whatever happens, treating it (the empirically contingent) as an effect, is a principle for cognizing nature. It is not a principle of speculative cognition. For suppose we set aside this principle as one that contains the condition of possible experience in general. Suppose we drop everything empirical and want to say it of the contingent as such. Then not the slightest justification remains for such a synthetic proposition. Nothing would let us see how I get from something that exists to something quite different from it (called a cause). In fact, once it is used this way, in pure speculation, the concept of a cause loses all meaning, just as the concept of the contingent does, any meaning whose objective reality we could make understandable in a concrete case.

Now suppose we infer from the existence of the things in the world to their cause. This belongs to the speculative use of reason, not to the natural use. The reason is that the natural use does not relate the things themselves (substances) to a cause. It relates to a cause only what happens, which means the states of things, seen as empirically contingent. To know that substance itself (matter) is contingent in being there would have to be a purely speculative cognition of reason.

B 664 And suppose we were speaking only of the form of the world, the way its parts are combined and how they change, but wanted to infer from this a cause that is wholly different from the world. That too would be a judgment of purely speculative reason, since in this case the object is not an object of any possible experience. But then the principle of causality, which holds only inside the field of experiences and is of no use, indeed has no meaning, outside it, would be taken wholly away from its purpose.

My claim now is this: all attempts to use reason in a purely speculative way on theology are fruitless from start to finish and, by their inner nature, null and void. I also maintain that the principles of reason’s use in nature lead to no theology at all. So unless we lay moral laws as the foundation or use them as a guide, there can be no theology of reason anywhere. For all synthetic principles of the understanding are of immanent use. Cognizing a highest being, however, calls for a transcendent use of them, and our understanding is not equipped for that at all. Suppose the law of causality, which is valid for experience, is to lead us to the original being. Then that being would need to be part of the chain of objects of experience. But then it would itself be conditioned in turn, like all appearances.

B 665 Suppose, though, that we permitted a leap across the boundary of experience, using the dynamical law that relates effects to their causes. What concept can this procedure give us? Nowhere near a concept of the highest being. Experience, after all, never hands us the greatest effect that is possible (the one that is meant to bear witness to its cause). Suppose we are permitted, merely so that nothing is left empty in our reason, to fill this lack of complete determination with a mere idea of highest perfection and original necessity. That can be granted as a favor, but it can’t be demanded by the right of an irresistible proof. So the physicotheological proof might perhaps add weight to other proofs (if there are any), by connecting speculation with intuition. On its own, though, it prepares the understanding for theological cognition rather than finishing the work, and gives the understanding a straight, natural heading toward it.

You can see from this that transcendental questions permit only transcendental answers, that is, answers built wholly from concepts a priori, with not the least empirical mixed in. But the question here is plainly synthetic. It calls for us to extend our cognition past every boundary of experience, namely to the existence of a being that is supposed to correspond to our mere idea.

B 666 No experience can ever equal that idea. Now, as our earlier proofs showed, all synthetic cognition a priori is possible just because it states the formal conditions of a possible experience. So all principles have only immanent validity. That is, they relate only to objects of empirical cognition, or appearances. So a merely speculative reason accomplishes nothing for theology even by a transcendental procedure.

Suppose, though, that people would rather doubt all the proofs of the Analytic above than let themselves be robbed of their persuasion about the weight of the grounds of proof used for so long. Even so, they can’t refuse to meet this demand of mine: they should at least justify themselves on how, and thanks to what illumination, they dare to soar above all possible experience on the strength of mere ideas. I’d ask to be spared new proofs, or touched-up versions of old ones. It’s true that there isn’t much to choose from here, since all merely speculative proofs finally boil down to one only, the ontological. So I needn’t fear being troubled much by the productivity of the dogmatic champions of that reason which claims to be free of the senses. Without thinking myself very combative on that account, I am also not willing to turn down the challenge.

B 667 It is the challenge to uncover the fallacy in each attempt of this kind, and so to defeat its presumption. Even so, the hope of better luck is never wholly abandoned among those who are used to dogmatic persuasions. That is why I hold to the one fair demand: that people justify themselves in general terms, starting from the nature of human understanding and every other source of cognition, on how they mean to go about extending their cognition wholly a priori. They must say how they mean to carry it to where no possible experience, and so no means at all, is enough to assure any concept we have thought up ourselves of its objective reality. However the understanding may have arrived at such a concept, the existence of its object can’t be found analytically in it. For the cognition of an object’s existence consists precisely in this: that the object is posited outside the thought, in itself. But it is wholly impossible to go out beyond a concept by itself and, without following empirical connection (which only ever gives appearances), to reach the discovery of new objects and excessive beings.

Reason in its merely speculative use is far from adequate to this great aim, which is to arrive at the existence of a supreme being. Even so, it is very useful.

B 668 It can correct the cognition of that being, in case it could be drawn from somewhere else. It can bring that cognition into agreement with itself and with every intelligible aim. And it can clear it of whatever might clash with the concept of an original being, as well as of every trace of empirical restriction.

Transcendental theology therefore remains, for all its inadequacy, of important negative use. It acts as a permanent censor of our reason whenever reason deals with pure ideas alone. Pure ideas allow no standard of measure but a transcendental one. For suppose that, in some other relation, perhaps a practical one, the assumption of a supreme intelligence, a highest being sufficient for everything, turned out to be valid beyond dispute. Then it would matter enormously to fix this concept exactly on its transcendental side: a being that is necessary and most real. It would matter just as much to strip away whatever clashes with highest reality, whatever belongs to mere appearance (anthropomorphism, taken broadly). And it would matter, at the same time, to clear out of the way all opposing claims, whether atheistic, deistic or anthropomorphistic. That is very easy in a critical treatment like this. Grounds are at hand that bring to light the inability of human reason to assert that such a being exists.

B 669 Those very same grounds must also be enough to prove the unfitness of every counter-assertion. After all, how could anyone, by reason’s pure speculation, reach the insight that no highest being exists to serve as the original ground of all things? Or that it has none of the properties that we picture, judging by their consequences, as comparable with the dynamical realities of a thinking being? Or that, in the latter case, those properties would also have to carry every restriction that sensibility inevitably places on the intelligences we know from experience?

So where reason is used in a purely speculative way, the highest being stays a mere ideal, though a faultless one: a concept that closes and crowns all human cognition. On this path its objective reality can’t be proved, but it can’t be refuted either. And suppose there should be a moral theology that can make up this lack. Then transcendental theology, which until then was only problematic, proves itself indispensable, by determining its concept and constantly censoring a reason that is often enough deceived by sensibility and doesn’t always agree with its own ideas. Necessity, infinity, unity, existence apart from the world (not as a world soul) and eternity free of conditions of time are purely transcendental predicates.

B 670 So are omnipresence without conditions of space, almightiness and so on. That is why the purified concept of them, which every theology needs so badly, can be drawn only from transcendental theology.

Appendix to the Transcendental Dialectic Using the Ideas of Pure Reason in a Regulative Way

How all the dialectical attempts of pure reason turn out confirms what we already proved in the Transcendental Analytic: every inference that tries to lead us beyond the field of possible experience is deceptive and has no ground. But the outcome also teaches us something more particular. Here human reason has a natural tendency to overstep that boundary. Transcendental ideas are as natural to reason as the categories are to the understanding. There is one difference. The categories lead to truth, that is, to our concepts agreeing with their object. The ideas produce a mere illusion, but an irresistible one, and even the sharpest critique can hardly stop its deception from working on us.

Whatever is rooted in the nature of our forces must be suited to its purpose and agree with the proper use of those forces.

B 671 This holds as long as we can prevent a certain misunderstanding and find out which way those forces really point. So, in all likelihood, the transcendental ideas have a good use, and therefore an immanent one. If their meaning is mistaken, though, and they are taken for concepts of real things, then their application can be transcendent, and for that very reason deceptive. That is because it is not the idea in itself but only its use that can fly beyond the whole of possible experience (transcendent) or stay at home within it (immanent). Which it is depends on where we aim the idea. We can aim it straight at an object we imagine corresponds to it, or only at the use of the understanding in general, in relation to the objects the understanding deals with. And every fault of subreption must always be blamed on a defect of the power of judgment, never on the understanding or on reason.

Reason never relates straight to an object. It relates only to the understanding, and by way of the understanding to its own empirical use. So it creates no concepts (of objects). It only puts them in order, and gives them the unity they can have at their widest extent, that is, unity in relation to the totality of the series. The understanding doesn’t look at that totality at all. It looks only at the connection through which series of conditions come about everywhere according to concepts.

B 672 What reason really has as its object, then, is only the understanding and its purposeful employment. The understanding binds the manifold in an object together through concepts. Reason does the same for the manifold of concepts, using ideas. It puts a certain collective unity before the understanding’s acts as their target, whereas left to themselves those acts are busy only with distributive unity.

So here is my claim: transcendental ideas are never constitutive, in the sense of giving us concepts of certain objects. If we understand them that way, they are merely sophistical (dialectical) concepts. They do have an excellent and absolutely necessary regulative use, though. Their job is to aim the understanding at a certain goal. Looking toward that goal, the lines of direction of all the understanding’s rules meet in a single point. That point is just an idea, a focus imaginarius. No concept of the understanding really starts out from it, because it lies wholly outside the boundaries of possible experience. Even so, it serves to give those concepts the greatest unity along with the greatest breadth. Now this does give rise to a deception. It looks as if these lines of direction shot out from an object situated outside the field in which empirical cognition is possible (like the objects we see behind the face of a mirror). But this illusion (which we can keep from actually deceiving us) is still absolutely indispensable.

B 673 We need it if we want to see not only the objects right before our eyes but also those lying far off behind our backs. In our case that means we want to train the understanding to reach beyond every given experience (which is only a part of all possible experience), and so to its greatest and utmost expansion.

Look over our cognitions of the understanding across their whole range. We find that reason contributes something of its own to them and works to achieve it: the systematic side of cognition, meaning how it hangs together from one principle. Behind this unity of reason there is always an idea, namely the idea of the form of a whole body of cognition. That whole comes before the determinate cognition of the parts, and it contains the conditions for fixing a priori each part’s place and its relation to the others. So this idea postulates complete unity in the understanding’s cognition. That unity turns the cognition from an accidental collection into a system connected by necessary laws. Strictly speaking, we can’t say the idea is a concept of an object. What it is a concept of is the unity that runs through all those concepts, to the extent that this unity works as a rule for the understanding. Concepts of reason like this aren’t drawn from nature. Rather, we question nature in the light of these ideas, and we count our cognition as deficient. It stays deficient as long as it doesn’t match them.

B 674 People concede that you will hardly find pure earth, or pure water, or pure air, or anything of the kind. Still, we need the concepts of them (so, as far as complete purity goes, these concepts originate only in reason) in order to settle properly how much each of these natural causes contributes to the appearance. So all materials are reduced to the earths (the mere weight, so to speak), then salts and combustible substances (the force), and last of all water and air, which serve as vehicles (machines, so to speak, through which the others work). The aim is to explain the chemical effects of materials on one another by the idea of a mechanism. For although nobody actually puts it in these words, the influence of reason on the classifications of natural scientists is very easy to spot.

If reason is a power of deriving the particular from the universal, there are two cases. In the first, the universal is certain on its own account and already given. Then all that’s needed is the power of judgment to subsume under it, and through it the particular is determined by necessity. I’ll call this the apodictic use of reason. In the second, the universal is only assumed problematically and is a mere idea. The particular is certain, but whether the rule is universal for this consequence is still a problem. So a number of particular cases, every one of them certain, are run past the rule, to check whether they flow from it.

B 675 If it looks as if every particular case we can name follows from it, we infer that the rule is universal, and from that we infer to all cases, even those not given in themselves. I’ll call this the hypothetical use of reason.

The hypothetical use of reason works from ideas laid down as problematic concepts. Strictly speaking, it isn’t constitutive. That is, if we judge with full rigor, it doesn’t make the truth of the universal rule, which was adopted as a hypothesis, follow from it. After all, how could anyone know all the possible consequences that would prove the rule’s universality by following from the same adopted principle? The hypothetical use is only regulative. Its job is to bring unity into the particular cognitions as far as possible, and so to bring the rule closer to universality.

So the hypothetical use of reason is aimed at systematic unity among the understanding’s cognitions, and that unity is how we test whether rules are true. Turn it around, and systematic unity (taken as a mere idea) is nothing but a unity we have projected. We shouldn’t treat it as something given. It is only a problem. Still, it helps us find a principle for the varied, particular uses of the understanding, and with that principle it leads the understanding into cases that haven’t been given, and ties its work together.

B 676 But all this lets us see is that systematic unity, the unity that reason brings to the many cognitions of the understanding, is a logical principle. Where the understanding can’t get to rules by itself, ideas help it along. At the same time they bring its different rules into harmony under one principle (systematically), and so into connection, as far as that can be done. A separate question is whether the objects are by their own constitution, or the understanding that cognizes them as objects is by its nature, destined for systematic unity. A further question is whether we may, to a certain extent, postulate that unity a priori, even without regard to this interest of reason, and so say: all possible cognitions of the understanding (the empirical ones included) have unity of reason and stand under common principles from which they can be derived, however different they are. That would be a transcendental principle of reason. It would make systematic unity not merely subjective and logical, as a method, but objectively necessary.

Let’s illustrate this with a case where reason is used. One kind of unity according to the understanding’s concepts is the unity of the causality of a substance, which we call a force. The different appearances of one and the same substance look so unlike at first glance that we have to begin by assuming almost as many forces in it as there are effects showing up.

B 677 In the human mind, for example, there are sensation, consciousness, imagination, memory, wit, discernment, pleasure, desire, and so on. At the outset, a logical maxim commands us to reduce this apparent variety as far as possible. We do so by comparing, to discover the hidden identity, and by checking whether imagination combined with consciousness might not be memory, wit, discernment, or perhaps even understanding and reason. The idea of a fundamental power is at minimum a problem: how to represent the many forces systematically. Logic can’t find out at all whether there is any such power. Reason’s logical principle requires us to achieve as much of this unity as we can. The more identical the appearances of the various forces prove to be, the likelier it gets that they are only different outward expressions of a single force, which we may call (in a comparative sense) their fundamental power. We proceed the same way with the rest.

Next we have to compare the comparative fundamental powers with one another. If we find that they agree, we can draw them closer to a single root, an absolute fundamental power. This unity of reason is merely hypothetical, though. We don’t claim that such a power must actually be found. We claim only that, for the sake of reason, we must look for one, in order to set up certain principles for the many rules.

B 678 Experience may supply those rules. And wherever possible, we must work systematic unity into our cognition by that route.

But watch the transcendental use of the understanding, and it turns out that the general idea of a fundamental power is not meant merely as a problem for hypothetical use. It lays claim to objective reality, and through that the systematic unity of the many forces of one substance is postulated and an apodictic principle of reason is set up. For even before we have tried to find the agreement of the many forces—and even if every attempt to find it fails—we still presuppose that such agreement will be found. We do this not only because of the unity of the substance, as in the case just given. Even where very many forces turn up that are alike only to some degree, as in matter in general, reason presupposes systematic unity among the many forces. That is because particular laws of nature stand under more general ones, and so the saving of principles becomes not just an economizing rule for reason but an inner law of nature.

Indeed, it’s hard to see how a logical principle of unity of reason among rules could hold unless a transcendental one were presupposed. That transcendental principle assumes, a priori and as necessary, that such systematic unity belongs to the objects themselves.

B 679 What entitles reason, in its logical use, to demand that we treat the manifold of forces which nature lets us know as a merely hidden unity, and derive it from some fundamental power, to the extent that it can? Reason could not demand this if it were free to admit that all forces might just as well be unlike, and that the systematic unity of their derivation might not fit nature. In that case reason would be acting against its own vocation, since it would set itself an idea as its goal that flatly contradicts how nature is arranged. Nor can we say that reason first took this unity from the contingent constitution of nature, following principles of reason. That is because the law of reason that tells us to seek this unity is necessary: if it were removed, there would be no reason left at all. Without reason we would have no coherent use of the understanding, and without that, no sufficient mark of empirical truth. So, as far as this last is concerned, we must presuppose the systematic unity of nature as altogether objectively valid and necessary.

This transcendental presupposition turns up, hidden in a remarkable way, in the principles of philosophers too, although they haven’t always noticed it there or owned up to it. The school rule, or logical principle, is this. The variety of single things doesn’t rule out sameness of species.

B 680 The many species must be treated only as various determinations of a handful of genera, and those genera as determinations of higher ones still, and so on. So we have to look for a certain systematic unity among all possible empirical concepts, to the degree that they can be traced back to higher, more general concepts. No use of reason would take place without this rule. We can reason from the universal down to the particular only when general properties of things are laid down as a basis, with the particular ones falling under them.

Philosophers also take it for granted that this same agreement is found in nature. They do so in the well-known school rule that we shouldn’t multiply the beginnings (principles) without need (entia praeter necessitatem non esse multiplicanda). What this says is that the nature of things itself supplies material for unity of reason. The endless variety we seem to see must not stop us from suspecting a unity of basic properties behind it, from which the variety can be derived only by further determination. People have chased this unity in every age, even though it is a mere idea, and so eagerly that there has been more cause to hold back the appetite for it than to encourage it. It was already a great deal when chemists managed to trace all salts back to two main kinds, acid and alkaline. They even try to treat this difference as only a variety.

B 681 In other words, they take it as one and the same basic substance expressing itself in different ways. The many kinds of earth (the stuff of stones and even of metals) people have tried, step by step, to bring down to three kinds and finally to two. Still not satisfied, chemists can’t shake the thought that behind these varieties there is a single kind. They even suspect a common principle for these earths and the salts together. One might think this is only an economical trick of reason, a way to spare itself as much trouble as it can, and a hypothetical experiment that, if it works, lends probability to the proposed explanation just through this unity. But a self-serving aim like that is very easy to tell apart from the idea. Everyone presupposes, in line with the idea, that this unity of reason fits nature itself, and that here reason isn’t begging but commanding, even though it can’t fix the boundaries of this unity.

Suppose the appearances offered to us differed so greatly that even the sharpest human understanding, by comparing one with another, could not find the slightest similarity. I don’t mean a difference of form (in that they may resemble each other). I mean a difference in content, in the variety of beings that exist. We can certainly imagine such a case. Then the logical law of genera would have no place at all.

B 682 There would not even be a concept of a genus or any general concept, and there would be no understanding either, because the understanding works with nothing but such concepts. So when the logical principle of genera is applied to nature (and here by nature I mean only the objects that are given to us), a transcendental principle has to stand behind it. That principle requires us to presuppose sameness of kind in whatever variety a possible experience contains, though we can’t settle its degree a priori. Without such sameness there would be no empirical concepts, and so no experience.

Opposed to the logical principle of genera, which demands sameness, there is a second, the principle of species. It needs variety and differences among things, even though they agree in falling under the same genus, and it tells the understanding to give these as much attention as it gives the likeness. This principle (of acuteness, or the power to discriminate) puts a firm check on the light-mindedness of the first (the principle of wit). So reason shows a double interest here, and the two interests pull against each other. One is the interest in extent (generality), with regard to genera. The other is the interest in content (determinateness), with regard to the variety of species. In the first case a great deal is thought under the concepts. In the second, even more is thought within them.

B 683 This also shows in the sharply different mindsets of natural scientists. Some of them (the mainly speculative ones) are almost enemies of unlikeness, and always aim at unity of genus. The others (the mainly empirical minds) keep trying to split nature into so much variety that one would almost have to give up hope of judging its appearances by general principles.

The latter way of thinking plainly rests on a logical principle too, one that aims at the systematic completeness of all cognitions. I start from the genus and go down to whatever variety it may hold, and in this way I try to give the system breadth. In the first case, when I climb up to the genus, I try to give it simplicity. The sphere of a concept that marks a genus doesn’t show how far its division can go, any more than the space matter can fill shows how far matter can be divided. So every genus needs different species, and these need different subspecies. And there is no subspecies that wouldn’t again have a sphere (an extent, as conceptus communis). So reason, at its fullest reach, demands that no species be regarded as the lowest in itself. That is because a species is always a concept that contains only what is common to different things. So it isn’t thoroughly determined.

B 684 Nor can it be related directly to an individual. It must therefore always contain other concepts under itself, that is, subspecies. This law of specification could be put this way: entium varietates non temere esse minuendas.

It’s easy to see, though, that this logical law too would have no sense and no use unless a transcendental law of specification lay beneath it. That transcendental law does not, of course, require the things that can become our objects to be actually infinite in their differences. The logical principle gives no occasion for that, since all it asserts is that the logical sphere is indeterminate with regard to possible division. What the transcendental law does is require the understanding to look for subspecies under every species it meets, and for smaller differences within every difference. Without lower concepts there could be no higher ones either. The understanding cognizes by means of concepts alone. So, as far as it carries its division, it never cognizes by mere intuition, but always again through lower concepts. Cognition of appearances, if it is to be fully determined (and only the understanding can do that), requires us to keep specifying its concepts without end. It also requires us to advance to differences that always remain, differences that the concept of the species, and even more the concept of the genus, has abstracted from.

B 685 This law of specification can’t have been taken from experience either, since experience can’t disclose anything so far-reaching. Empirical specification soon stops short in telling the manifold apart, unless a transcendental law of specification is already in place, as a principle of reason, to guide it. That law leads it to look for such differences, and to keep suspecting them even when they don’t reveal themselves to the senses. It took a rule of reason running ahead of experience to discover that absorbent earths come in different kinds (lime earths and muriatic earths). That rule set the understanding the task of looking for the difference, because it presupposed nature to be rich enough to make us suspect it. We have understanding only if nature contains differences. Equally, we have it only if nature’s objects are alike in kind. After all, the very manifoldness of what can be gathered under one concept is what makes up the use of that concept and the business of the understanding.

Reason thus prepares the understanding’s field 1. through a principle of sameness of kind in the manifold under higher genera, and 2. through a principle of variety among things of the same kind under lower species. To complete the systematic unity, it also adds 3. a law of affinity among all concepts.

B 686 This law demands a continuous passage from each species to every other, by a stepwise growth of difference. They can be named the principles of homogeneity, of specification, and of the continuity of forms. The last arises when we unite the first two. First we complete the systematic connection in the idea, both by going up to higher genera and by going down to lower species. After that, all the manifold things are related to one another, because they all descend, through every degree of further determination, from a single highest genus.

We can picture the systematic unity among the three logical principles in the following way. Take every concept as a point. As the standpoint of a spectator, the point has its horizon, that is, a number of things that can be pictured from it and, as it were, looked over. Within this horizon it must be possible to name a multitude of points without end, and each of them again has its own narrower field of view. By the principle of specification, every species has subspecies in it. A logical horizon is made up of smaller horizons (subspecies), never of points without extent (individuals). Now take several horizons, meaning genera each fixed by its own concept. We can think of a common horizon drawn around them, from which we survey them all as from one center.

B 687 That common horizon is the higher genus, and so on, until at last the top genus gives the universal and true horizon. It is fixed from the viewpoint of the supreme concept, and it takes in under itself all manifoldness, as genera, species and subspecies.

Homogeneity brings me up to this highest standpoint. Specification brings me down to all the low ones and to their greatest variety. In this way nothing in the whole extent of all possible concepts is empty, and nothing can be found outside it. So when we assume that all-embracing field of view, divided through and through, a principle follows: non datur vacuum formarum. That is, there are not several different original and first genera, cut off from each other, so to speak, with an empty gap between them. All the manifold genera are only divisions of a single highest, universal genus. And from this principle comes its immediate consequence, datur continuum formarum. That is, all the differences among species border on one another. There is no jumping from one to another. One gets there only by way of all the smaller steps of difference. Put briefly, no species or subspecies is the very nearest to another (in the concept of reason). Intermediate species always remain possible.

B 688 Their difference from the first and from the second is smaller than the difference between those two.

So the first law guards against running wild into a variety of different original genera, and it recommends sameness of kind. The second law sets a check on this leaning toward agreement. It commands us to tell the subspecies apart before we turn with our general concept to the individuals. The third law unites the first two. Even in the greatest variety, it prescribes sameness of kind through the stepwise passage from one species to another. This shows a sort of kinship between the branches, in that they all grew from one trunk.

This logical law of the continuity of species, continui specierum (formarum logicarum), rests on a transcendental one (lex continui in natura). Without that, the prescription would only lead the understanding astray in its use, because it might take a road exactly opposite to nature’s. So this law has to rest on pure transcendental grounds, not empirical ones. If it were empirical, it would come later than the systems. In fact it first brought forth what is systematic in our cognition of nature. Nor are there, hidden behind these laws, any designs for running a test with them as mere trials.

B 689 It is true that this connection, where it fits, gives a strong reason to hold the unity we thought up hypothetically to be well founded, and so the laws are useful in that respect too. But one can plainly see that the laws judge the thrift of fundamental causes, the manifoldness of effects, and the resulting kinship among the members of nature to be reasonable in themselves and fitting to nature. So these principles recommend themselves directly, and not merely as devices of method.

It is easy to see, though, that continuity of forms is nothing but an idea. No matching object can be shown for it in experience. One reason is that the species in nature really are divided off from one another, and so must in themselves make up a quantum discretum. If the stepwise progress in their kinship were continuous, they would have to hold a real infinity of links in between any two given species, and that is impossible. The other reason is that this law can’t be put to any definite use in empirical work. It doesn’t give even a faint sign of affinity that would tell us by what, or how far, to look for the series of degrees in their differences. It gives nothing but a general indication that we have to look for them.

B 690 Suppose we rearrange the principles just listed, putting them in the order that suits their use in experience. The principles of systematic unity would then stand roughly like this: manifoldness, kinship and unity. Each of these is to be taken as an idea, at the top degree of its completeness. Reason takes for granted the cognitions of the understanding, which are applied first of all to experience, and it looks for their unity according to ideas that reach much further than experience can. Kinship among the manifold, with its variety left intact, under one principle of unity, concerns not only things but even more their bare properties and forces. Say experience (not yet fully corrected) gives us the course of the planets as circular, and then we find differences. We suspect those differences in whatever can change the circle, by a constant law, through all the endless intermediate degrees into one of the deviating orbits. In other words, the motions of planets that aren’t circles will come more or less close to the properties of a circle, and they land on the ellipse. Comets show orbits that vary still more. As far as we can observe, they don’t even come back around in a circle. But we guess at a parabolic course, which is related to the ellipse. When the long axis of the ellipse is stretched very far, a parabola can’t be told apart from it in any of our observations.

B 691 So, led by those principles, we arrive at a unity of kind among the shapes of these orbits. Through that we go on to the unity of the cause behind every law of their movement, which is gravitation. From there we push our conquests further. We try to explain all the varieties and seeming departures from those rules by the same principle. In the end we even add more than experience could ever confirm. By the rules of kinship, we think up hyperbolic comet orbits, paths on which these bodies would depart from our solar system entirely. Traveling from sun to sun, they join in their course the more distant parts of a world system that has no bounds for us, and that is held together by one and the same moving force.

What is remarkable about these principles, and what alone concerns us here, is this. They seem to be transcendental. They contain mere ideas that the empirical use of reason is to follow, and that use can follow them only, so to speak, asymptotically, that is, by merely approaching them without ever reaching them. Even so, as synthetic propositions a priori they have objective but indeterminate validity. They serve as a rule of possible experience, and in working on experience they are indeed used successfully, as heuristic principles.

B 692 Yet no one can produce a transcendental deduction of them. As we proved above, that is always impossible for ideas.

In the Transcendental Analytic we sorted the principles of the understanding into two groups. The dynamical ones are merely regulative principles of intuition. The mathematical ones are constitutive with respect to intuition. Even so, those dynamical laws are certainly constitutive for experience, because through them the concepts without which there would be no experience become possible a priori. Principles of pure reason, by contrast, can’t be constitutive even with respect to empirical concepts. Sensibility can supply no matching schema for them, so they can’t have an object in the concrete. Now suppose I give up that empirical use of them as constitutive principles. How can I still guarantee them a regulative use, and with it a measure of objective validity? And what meaning can that use have?

For reason, the understanding is an object just as sensibility is for the understanding. It falls to reason to organize into a system the unity of every possible empirical act of the understanding, just as the understanding ties together the manifold of appearances by concepts and brings it under empirical laws. But acts of the understanding without schemata of sensibility are indeterminate.

B 693 In the same way, the unity of reason is indeterminate in itself: it doesn’t settle under what conditions, or how far, the understanding is to link its concepts into a system. No schema in intuition can be found for the thoroughgoing systematic unity of all the understanding’s concepts. Still, an analogue of such a schema can be given, and it must be. It is the idea of a maximum: the greatest division and the greatest union of the understanding’s cognition under a single principle. For the greatest, the absolutely complete, can be thought determinately, because all restricting conditions, which give an indeterminate manifold, are left out. So the idea of reason is an analogue of a schema of sensibility. There is this difference, though. Applying concepts of the understanding to the schema of reason is not a cognition of the object itself (as it is when the categories are applied to their sensory schemata). It is a rule, a principle for making all use of the understanding hang together as a system. Now every principle that fixes a priori the thoroughgoing unity of the understanding’s use also holds, though only indirectly, of the object of experience. So the principles of pure reason will have objective reality with regard to that object too. But that is not in order to determine anything in it. It is only to indicate the procedure.

B 694 By that procedure, the understanding’s use in experience, empirical and determinate as it is, can come to agree with itself all the way through. It does so by being connected, as far as possible, with the principle of thoroughgoing unity, and derived from it.

I call all those subjective principles that come not from how the object is made but from the interest of reason, in view of a certain possible perfection in our cognition of that object, maxims of reason. So there are maxims of speculative reason that rest solely on its speculative interest, even though they may look like objective principles.

Treat purely regulative principles as constitutive, and as objective principles they may clash. Regard them only as maxims, and there is no true clash. There is only a different interest of reason, and that is what splits the ways of thinking apart. In fact reason has just one interest, and the quarrel between its maxims is only a difference between methods and a mutual limiting of them, each trying to satisfy that interest.

So for one would-be reasoner variety counts for more (that is the principle of specification), and for another unity counts for more (the principle of aggregation).

B 695 Each of them believes his judgment comes from insight into the object, yet he rests it on nothing but how strongly he is attached to one of the two principles. Neither rests on objective grounds. They rest only on the interest of reason, so they would be better called maxims than principles. I see discerning men quarreling over the characteristics of people, animals or plants, even of bodies in the mineral kingdom. Some, for example, suppose there are distinct national characters rooted in ancestry, or else definite, hereditary differences between families, races and so on. Others set their minds on the view that nature made all dispositions in this respect exactly the same, and that all difference rests on outward chance. When I see this, I only have to consider the nature of the object to grasp that it lies buried far too deep for either side to speak from insight into the nature of the object. It is nothing but the double interest of reason. One party takes one interest to heart, or merely affects to, and the other party takes the other. So it is the difference between maxims of nature’s manifoldness and maxims of nature’s unity. These can very well be reconciled. But so long as people mistake them for objective insights, they create not only dispute but also obstacles that hold up the truth for a long while, until some means is found.

B 696 It will be a means of uniting the conflicting interests and satisfying reason on the point.

The same goes for asserting or challenging the famous law of the continuous ladder of creatures, which Leibniz brought into circulation and Bonnet splendidly polished up. This law is nothing but a following of the principle of affinity, which rests on the interest of reason. Observation and insight into how nature is arranged could not have offered it as an objective claim. Experience can show us the rungs of such a ladder, but they stand much too far apart. What we imagine to be small differences are usually such wide gaps in nature itself that such observations count for nothing as designs of nature. This is especially so where there is a great variety of things, since it is always easy to come up with some resemblances and near-matches. By contrast, the method of hunting for order in nature under a principle like that is quite legitimate and excellent as a regulative principle of reason. So is the maxim of taking such order to have its basis somewhere in nature as a whole, though without settling where or how far. As such a principle, it goes much further than experience or observation could ever equal. Yet it determines nothing. It only marks out the way to systematic unity.

B 697 The Ultimate Aim of the Natural Dialectic in Human Reason

No idea of pure reason is ever dialectical in itself. Only their misuse can make a deceptive illusion arise from them. Our reason’s own nature sets them as tasks for us. Reason is the supreme tribunal for all rights and claims of our speculation, and it can’t possibly contain original deceptions and delusions. So we can presume they have a good and purposeful vocation in the natural makeup of our reason. The mob of would-be reasoners, though, shouts as always about nonsense and contradictions. They slander the government whose innermost plans they can’t penetrate, and to whose beneficial influence they should themselves owe their survival, and even the culture that puts them in a position to criticize and condemn it.

We can’t use an a priori concept with any certainty until we have carried out its transcendental deduction. The ideas of pure reason don’t allow a deduction of the kind the categories do. But suppose they are to have at least some objective validity, even an indeterminate one, and not stand for merely empty thought-things (entia rationis ratiocinantis).

B 698 Then a deduction of them has to be possible, even if it departs widely from the one we can carry out for the categories. That completes the critical work of pure reason, and we’ll now take it on.

It makes a big difference whether something is given to my reason as an object outright, or only as an object in the idea. In the first case my concepts aim at determining the object. In the second, what I have is really just a schema. No object is granted to it directly, not even hypothetically. All it does is help us represent other objects indirectly, by relating them to this idea, according to their systematic unity. Take the concept of a highest intelligence. It is a mere idea. That is, its objective reality is not supposed to consist in its relating straight to an object (taken that way, its objective validity couldn’t be justified). What it gives us is a schema for the concept of a thing in general, built to meet the demands of the largest possible unity of reason. Its only purpose is to secure the greatest systematic unity in the empirical use of our reason. We do this by deriving the object of experience as if from its ground or cause, namely the imagined object of the idea. Then, for example, the rule says this.

B 699 Regard the things in the world as though they had their existence from a supreme intelligence; in this way the idea is really only a heuristic concept, not an ostensive one. It doesn’t tell us what an object is like. It tells us how, with its guidance, we should look for the constitution and connection of the objects of experience in general. Now suppose we can show the following. There are three kinds of transcendental ideas: psychological, cosmological and theological. None of them relates directly to any object corresponding to it, or to its determination. Yet every rule for the empirical use of reason, on the assumption of such an object in the idea, leads toward systematic unity, always extends our cognition of experience, and can never run against it. If we can show that, then it is a necessary maxim of reason to proceed by such ideas. And this is the transcendental deduction of all the ideas of speculative reason. They are not constitutive principles that would stretch our cognition to cover more objects than experience can supply. They are regulative principles, serving to bring the manifold of empirical cognition in general into systematic unity. Within its own boundaries, that cognition is developed and put right further than the bare use of the principles of the understanding could manage without such ideas.

B 700 Let me make this clearer. Let us take the ideas just named as principles. First, in psychology: we should connect the mind’s appearances, actions and receptivity, following the thread of inner experience, as though the mind were one simple substance. It would last (at least through life) with a personal identity, while its states keep changing, and the states of the body would count only as outer conditions of those. Second, in cosmology: we have to trace the conditions of natural appearances, both inner and outer, in an inquiry that can never be finished, and we should proceed as though the series went on forever in itself and had no first or topmost member. This doesn’t mean denying that there are first grounds of appearances outside all appearances, grounds that are purely intelligible. But we may never bring them into explanations of nature, since we don’t know them at all. Third and last, in theology: whatever can belong to the context of possible experience, we have to regard as though it formed an absolute unity, one that is thoroughly dependent and still conditioned within the world of sense. And at the same time we have to regard the sum of all appearances (the world of sense itself) as though it had a single supreme, all-sufficient ground outside its own range. That ground is a reason that is, so to speak, independent, original and creative.

B 701 With this reason in view, we steer every empirical use of our reason toward its widest reach, as though the objects themselves had come from that original model for all reason. What this means is as follows. We are not to trace the soul’s inner appearances back to a simple thinking substance. We are to derive them from one another, in line with the idea of a simple being. We are not to trace the order of the world and its systematic unity back to a highest intelligence. Instead the idea of a supremely wise cause supplies us with the rule that tells reason how to satisfy itself best when it connects causes and effects in the world.

Only the cosmological idea gives us any trouble. When reason tries to treat that one as objective, it runs into an antinomy. Nothing at all stops us from taking the other ideas as objective and hypostatic. (The psychological and theological ideas contain no such antinomy.) No contradiction lies in them, so how could anyone challenge their objective reality? He knows no more about their possibility, to deny it, than we do to affirm it. Still, it isn’t enough, for assuming something, that nothing positive stands against it. We are not allowed, on the mere credit of a speculative reason that likes to wrap up its business, to take thought-beings that go beyond all our concepts (though they contradict none) and introduce them as real and determinate objects.

B 702 So they must not be accepted in themselves. Their reality should count only as the schema for the regulative principle of systematic unity in all our cognition of nature. They should be laid down only as analogues of real things, and not as such things in themselves. From the object of the idea we remove the conditions that confine our concept of the understanding. But those same conditions are the only thing that makes it possible for us to have a definite concept of any thing at all. And now we think of a something. We have no concept whatever of what it is in itself. But we do think of it as standing in a relation to the sum of appearances, and that relation is analogous to the one the appearances have to each other.

So assuming such ideal beings doesn’t actually stretch our cognition past the objects of possible experience. What it stretches is only the empirical unity of those objects. It does that by way of the systematic unity, for which the idea furnishes the schema. For that reason the idea counts as a regulative principle and not as a constitutive one. We may posit a thing that corresponds to the idea—a something, a real being. That doesn’t amount to saying we want to widen our cognition of things by means of transcendent concepts. The being is put at the base only in the idea, and not in itself, so it is there merely to express the systematic unity.

B 703 That unity is meant to be a guide for the empirical use of reason. In doing so we settle nothing about what grounds that unity, or about the inner nature of the being on which, as its cause, the unity depends.

The transcendental concept of God, the single definite concept that bare speculative reason provides, is deistic in the strictest sense. In other words, reason doesn’t even vouch for such a concept being objectively valid. It supplies only the idea of a something. All empirical reality rests its highest and necessary unity on that something. We can’t think of it except by analogy with a real substance that is the cause of everything according to the laws of reason. That holds on one condition: that we do try to think of it as a particular object at all. The alternative is to be content with the bare idea of reason’s regulative principle and put aside, as excessive for the human understanding, the completion of all the conditions of thought. But that cannot be reconciled with the goal of perfect systematic unity in our cognition, a goal on which reason at least imposes no limits.

So suppose I assume a divine being. I then have not the faintest concept of how its highest perfection could be possible in itself, or why its existence is necessary.

B 704 Yet then I can do justice to all other questions about the contingent, and I can give reason the fullest satisfaction regarding the greatest unity it is to look for in its empirical use. I cannot give it any satisfaction regarding the assumption itself. This proves that reason’s speculative interest, and not any insight of its own, gives it the right to set out from a point so far beyond its sphere, to look at its objects from there as a complete whole.

Here we find a difference in the way of thinking, with one and the same assumption. It’s quite subtle, but it matters a lot in transcendental philosophy. I may have enough ground to assume something in a relative way (suppositio relativa) and still lack the right to assume it outright (suppositio absoluta). The distinction fits when we are dealing only with a regulative principle. We see that such a principle is necessary in itself, but we don’t see its source. We assume a highest ground for it merely so as to think its generality more determinately. For example, I might think of a being as existing that corresponds to a mere, and indeed transcendental, idea. Then I can never assume that this thing exists in itself. No concepts we have could reach it, none by which I can think any object determinately.

B 705 And the idea itself rules out the conditions under which my concepts would be objectively valid. Reality, substance, causality, even necessity in existence: these concepts have no meaning that determines any object, except in the use that makes empirical cognition of an object possible. So they can help explain how things in the world of sense are possible, but not how a whole world is possible. Its ground would have to sit beyond the world, and then it could never be an object of possible experience. All the same, I can accept a being like that, beyond our grasp and the object of nothing but an idea, relative to the world of sense, though not in itself. Suppose that behind my reason’s widest possible empirical use lies an idea (of a systematically complete unity, which I’ll discuss more exactly before long). Experience can never present this idea adequately in itself. Yet it’s indispensable if empirical unity is to come as near as possible to the highest degree. In that case I am not only entitled but compelled to realize the idea, that is, to give it an actual object. But the object is only a something in general, which I don’t know in itself at all. To it, as the ground of that systematic unity, I give properties only in relation to that unity.

B 706 And the properties I give it are only the ones analogous to the concepts of the understanding in empirical use. Following the analogy of the realities in the world (substances, causality, necessity), I will therefore think of a being that has all of this in the highest perfection. This idea rests only on my reason. So I can think of the being as an independent reason that causes the whole world by way of ideas of the greatest harmony and unity. I leave out every condition that limits the idea. My only purpose, sheltered by such a primal ground, is to make possible a systematic unity of the manifold throughout the world, and through that unity the widest possible empirical use of reason. To do this I look at every combination as though a supreme reason had arranged it, a reason of which ours is a weak copy. Then I think of this highest being solely through concepts that properly apply only in the world of sense. But I use that transcendental assumption in no way other than relatively, namely to supply the substratum for the greatest possible unity of experience. So I may quite properly picture a being, distinct from the world, using properties that belong only to the world of sense. I’m not asking to know this object of my idea as it may be in itself, and I’d have no right to ask. I have no concepts for that.

B 707 Risk them beyond the field of the senses, and reality, substance, causality, even necessity in existence, lose all meaning. They are empty labels for concepts, with no content at all. All I think is how a being stands to the largest possible systematic unity of the entire world, a being of which I know nothing in itself. My only purpose is to turn that being into the schema of the regulative principle that serves my reason’s widest possible empirical use.

Now consider the transcendental object of our idea. We can see that we can’t presuppose its reality in itself by means of the concepts of reality, substance, causality and the like, because those concepts don’t apply in the least to something wholly different from the world of sense. So reason’s supposition of a highest being as the supreme cause is merely relative. It is thought up for the benefit of the systematic unity of the world of sense. In the idea, it is a bare something, and we have no concept of what it is in itself. This also explains something else. Anything that exists and is given to the senses makes us need the idea of a primal being that is necessary in itself. And yet we can never form the faintest concept of that being or of its absolute necessity.

At this point we can set out plainly the outcome of the entire transcendental dialectic.

B 708 We can also determine precisely what the ideas of pure reason are finally aimed at. They become dialectical only by being misunderstood and handled carelessly. Pure reason is really occupied with nothing but itself, and it can have no other business. What is given to it is not objects, for the unity of the concept of experience. What is given to it is the understanding’s cognitions, for the unity of the concept of reason, which is to say, connection under one principle. The kind of unity reason has is the unity of a system. For reason, this systematic unity is not an objective principle that extends it over objects. It is a subjective maxim, whose job is to carry reason across the whole of possible empirical cognition of objects. Even so, the systematic connection that reason can give to the empirical use of the understanding does more than promote its extension. It also confirms its correctness. The principle of such a systematic unity is objective too, but in an indeterminate way (principium vagum). It isn’t a constitutive principle for determining anything about its direct object. It is a purely regulative principle and maxim. Its work is to advance and firm up reason’s empirical use without end (indeterminately), by opening new paths that the understanding does not know, and without ever running against the laws of empirical use in the least.

B 709 But reason can think this systematic unity only by giving its idea an object at the same time. No experience can give that object, because experience never provides an example of perfect systematic unity. This being of reason is a mere idea (ens rationis ratiocinatae). So it isn’t accepted outright, as something real in itself. It is taken as a basis only problematically, because no concept of the understanding can reach it. The purpose is to look at all connection among the things of the world of sense as though they were grounded in this reasoned being. The sole reason for doing so is to found systematic unity on it. Reason can’t do without that unity. For empirical cognition by the understanding it helps in every way, and it can never get in its way.

If we take this idea as a claim, or even just a premise, about some real thing to which we would credit the ground of the world’s systematic makeup, we have missed its point right away. What we really do is leave open what that ground is like in itself, because it slips beyond our concepts. We simply adopt an idea as a vantage point. From it, and only from it, can we spread that unity which reason needs so much and which does the understanding so much good.

B 710 In short, this transcendental thing is only the schema of the regulative principle by which reason spreads systematic unity over all experience, as far as it can. Such an idea has as its first object myself, considered purely as a thinking nature (soul). To find out what properties a thinking being has as it exists in itself, I have to ask experience. Even among the categories, there is none I can apply to this object, except where its schema is given in sensory intuition. But that route never leads me to a systematic unity of all the appearances of inner sense. So reason sets aside the concept drawn from experience (of what the soul actually is), which can’t carry us far. In its place reason takes up the concept of the empirical unity in all thinking. It thinks this unity as unconditioned and original, and so turns it into a concept of reason (an idea): the idea of a simple substance. This substance is unchanging in itself (personally identical), and it stands in community with other real things outside it. In a word, it is a simple, self-standing intelligence. Yet here reason has in view only one thing: principles of systematic unity for explaining the soul’s appearances. It wants us to treat all determinations as belonging to one subject, and all forces, as far as we can, as derived from one fundamental power. It wants us to see every change as belonging to the states of a being.

B 711 That being is one and the same throughout, and it abides. And it wants us to picture all appearances in space as altogether different from acts of thinking. The simplicity of the substance and the rest were meant merely as the schema of this regulative principle. They are not presupposed as if they were the real ground of the soul’s properties. Those properties might rest on quite different grounds, which we don’t know at all. Even if we wanted to grant these predicates of the soul outright, we still couldn’t cognize the soul in itself through them, since they amount to a mere idea that can’t be presented in any concrete case. Such a psychological idea can only bring advantage, so long as we are careful not to let it count as more than an idea, that is, as holding only relative to the systematic use of reason on the appearances of our soul. In that case laws drawn from experience about bodily appearances, a completely different sort of thing, don’t get mixed into the explanations of what belongs to inner sense alone. No empty hypotheses of the generation, destruction and rebirth (palingenesis) of souls are admitted either. We examine this object of inner sense in a pure state, uncontaminated by properties of another kind. What’s more, the inquiry of reason aims to reduce the explanatory grounds in this subject to a single principle, as far as it can.

B 712 A schema of this kind, treating the subject as though it were a real being, does that best, and in fact does it alone. The psychological idea can also stand for only one thing: the schema of a regulative concept. Suppose I even asked whether the soul is by nature spiritual. The question would make no sense at all. By a concept like that, I remove not only bodily nature but nature altogether, that is, all the predicates of any possible experience, and therefore every condition for thinking an object for the concept. Yet those conditions alone are what allow us to say that the concept has a sense.

Merely speculative reason has a second regulative idea, the concept of the world as such. Nature is, after all, the only given object for which reason needs regulative principles. Nature comes in two kinds: thinking nature and bodily nature. For the second kind we need no idea, if all we want is to think it in its inner possibility, that is, to determine how the categories apply to it. An idea means a representation that goes beyond experience, and none is even possible for bodily nature, because in dealing with it we are led only by sensory intuition. That is unlike the psychological basic concept (the I), which contains a definite form of thinking—its unity—a priori.

B 713 So for pure reason nothing remains except nature as a whole, together with the completeness of the conditions in nature under some principle. The series of these conditions, taken as an absolute totality as we trace their members, is an idea. In the empirical use of reason it can never be fully reached, yet it serves as a rule for how to go about it. When we explain given appearances (by going back or going up), we should act as though the series were endless in itself, in indefinitum. But where reason itself is taken as the determining cause (in freedom), that is, in practical principles, we should act as though what lay before us were not an object of the senses but an object of pure understanding. There the conditions can no longer be placed in the series of appearances. They can be placed outside it, and then the series of states can be looked at as if it started absolutely (through an intelligible cause). All of this proves that the cosmological ideas are only regulative principles. They are far from positing, constitutively as it were, a real totality of such series. The rest can be looked up in its own place, under the antinomy of pure reason.

The third idea of pure reason is reason’s concept of God. It holds a merely relative supposition of a being: the one, all-sufficient cause of all cosmological series.

B 714 As for the object of this idea, we have not the slightest ground to accept it outright (to suppose it in itself). What could bring us to believe in a being of the highest perfection, one that is by its nature absolutely necessary, or to assert such a being, just from its own bare concept? What could even give us the right? Nothing could, except the world, since only in relation to the world can this supposition be necessary. And there it shows clearly that this being’s idea, like every speculative idea, says only one thing: reason commands us to treat all connections in the world as governed by principles of systematic unity. That is, we are to look at it as though it all sprang from a single, all-embracing being, as the highest and all-sufficient cause. It is clear from this that reason has nothing in view here except its own formal rule for widening its empirical use. It never aims to widen it past every boundary of empirical use. So no constitutive principle for its use, directed to possible experience, lies hidden under this idea.

This highest formal unity rests on concepts of reason alone, and it is the unity of things in respect of their purposes. Reason’s speculative interest requires us to see every single arrangement in the world as if a supreme reason had intended it.

B 715 For our reason, applied to the field of experience, such a principle opens wholly new prospects. It lets us connect the things of the world by teleological laws and so arrive at their greatest systematic unity. Assuming a supreme intelligence as the sole cause of the whole world—though only in the idea, of course—can therefore always help reason and never harm it. Suppose we assume in advance that a maker had wise purposes for the shape of the earth (round, though somewhat flattened)*, for mountains, seas and so on. Following this path, we can make a great many discoveries. As long as we hold to the assumption as a purely regulative principle, even a mistake can’t hurt us. All that can follow, at worst, is this: where we expected a purposive connection (nexus finalis), we run into nothing but a mechanical or physical connection (nexus effectivus).

B 716 In such a case we only lose one more unity. We don’t ruin reason’s unity in its empirical use. This setback can’t strike the law itself, taken in its general and teleological meaning, at all. An anatomist can be shown to be wrong if he ties a limb of an animal body to a purpose that clearly doesn’t follow from it. But it’s completely impossible to prove, in any single case, that an arrangement of nature, whatever it is, has no purpose whatever. That’s why physiology (the doctors’ science) widens its very narrow empirical knowledge of the purposes of an organic body’s structure by a principle that pure reason alone supplied. It carries this so far that people quite boldly assume, with the agreement of everyone sensible, that nothing in an animal lacks a use and a good purpose. If this assumption were meant constitutively, it would go much further than our observations to date entitle us to go. We can see from this that it is only a regulative principle of reason. Its aim is the highest systematic unity, reached by way of the idea that the supreme cause of the world acts with purpose—as though this cause, being the highest intelligence, had made everything for the wisest purpose.

* The advantage of a spherical earth is well known enough. Few people know, however, that its flattening, as a spheroid, is the one thing that stops projecting lands, or smaller mountains perhaps pushed up by earthquakes, from steadily shifting the earth’s axis by a good deal in a fairly short time. They would do so if the swelling of the earth at the equator were not such a huge mountain that the thrust of any other mountain can never move it noticeably from its place relative to the axis. And yet no one hesitates to explain this wise arrangement by the equilibrium of the earth’s once-fluid mass.

B 717 But suppose we step away from this restriction of the idea to its merely regulative use. Reason is then led astray in all sorts of ways. It leaves the ground of experience, which has to hold the markers of its path, and ventures out beyond that ground into the incomprehensible and unfathomable. At those heights it inevitably gets dizzy, because from there it sees itself cut off completely from any use that fits with experience.

The first mistake that arises when the idea of a highest being is used not just regulatively but constitutively (against the nature of an idea) is lazy reason (ignava ratio)*. This name fits any principle that makes us see our inquiry into nature, wherever it stands, as absolutely finished.

B 718 Reason then goes to rest, as if it had done its job entirely. Even the psychological idea makes things very comfortable for reason when it is treated as a constitutive principle, first to explain the soul’s appearances and then even to carry our cognition of this subject out past all experience (to its state after death). But it also wrecks and ruins every natural use of reason that follows the guidance of experience. Consider the dogmatic spiritualist. He accounts for the unity of the person, which stays the same however its states change, by the unity of a thinking substance that he thinks he perceives directly in the I. Our concern for things that will happen only after our death he explains from the awareness that our thinking subject is immaterial, and the like. And so he excuses himself from any investigation of nature, through physical grounds of explanation, into what causes these inner appearances. He does it by skipping past the immanent sources of cognition in experience, as if by decree of a transcendent reason, for the sake of his own ease but at the loss of all insight. The damaging result is even easier to see in the dogmatism about our idea of a highest intelligence, and in the theological system of nature wrongly built upon it (physico-theology).

B 719 There, all the purposes that show up in nature (often made purposes only by ourselves) are used to make the search for causes very comfortable. Instead of looking for them in the general laws governing the mechanism of matter, we turn straight to the unsearchable decree of supreme wisdom. And we consider the work of reason finished when we spare ourselves its use. Yet that use of reason finds a guide nowhere except where the order of nature, and the series of changes under their inner and more general laws, hand one to us. We can avoid this mistake if we do not look at only a few parts of nature from the viewpoint of purposes, such as the way the solid land is distributed, its structure, and the nature and position of the mountains, or even just the organization in the plant and animal kingdoms. We must make this systematic unity of nature quite general, by relating it to the idea of a highest intelligence. Then we lay down a purposiveness according to general laws of nature as our basis. No particular arrangement is exempt from those laws. Some are simply marked out for us more clearly than others. And what we get is a regulative principle for systematic unity in a teleological connection. We do not fix that connection beforehand.

B 720 Only in expectation of it may we pursue the physical-mechanical connection under general laws. Only so can the principle of purposive unity extend the use of reason about experience in every case, without damaging it in any.

* The ancient dialecticians used this name for a fallacious argument that ran: Suppose your fate is to get well from this sickness. Then you will get well whether or not you call a doctor. Cicero says the way of arguing got its name because following it would leave no use of reason in life. That is the reason I give the same name to the sophistical argument of pure reason.

The second mistake comes from misreading the principle of systematic unity just described. It is the mistake of upside-down reason (perversa ratio, ysteron proteron rationis). The idea of systematic unity was meant only as a regulative principle for looking for that unity in the combination of things under general laws of nature. To the extent that any of it can be found by the empirical route, we are also to believe that we have come that much nearer to the completeness of its use, though we will never reach it. People do the opposite. They start by laying down, as a hypostasis, the reality of a principle of purposive unity. Because the concept of such a highest intelligence is in itself completely unfathomable, they define it in human terms (anthropomorphically). Then they impose purposes on nature, forcefully and like dictators, when they should properly look for them by way of physical research. The consequences are twofold. Teleology, which was supposed only to round out the unity of nature under general laws, now works to cancel that unity.

B 721 And reason also defeats its own aim, which is to use nature to prove that such an intelligent supreme cause exists. For suppose the highest purposiveness can’t be presupposed in nature a priori, that is, as part of its essence. Then what could direct us to look for it at all, and to climb its ladder toward the highest perfection of an author, a perfection that is absolutely necessary and therefore knowable a priori? The regulative principle asks us to assume systematic unity outright, as the unity of nature. That unity is not merely discovered by experience. It is assumed a priori, if still without being determined, and so as following from the essence of things. If I instead start by laying down a supreme ordering being, nature’s unity is in fact destroyed. It then becomes completely foreign to the nature of things, and accidental. And it can’t be known from nature’s general laws either. So a faulty circle appears in the proof, because we presuppose what should have been proved.

If we treat the principle that nature is a systematic unity as constitutive, when it is only regulative, we do nothing but confuse reason. The idea is laid down merely as the basis for a harmonious use of reason.

B 722 To presuppose it as a cause that really exists, a hypostasis, confuses reason too. The study of nature takes its own course, strictly along the chain of natural causes and by their general laws. It does follow the idea of an author. But the point is not to get from him the purposiveness it hunts for everywhere. The point is the reverse: to recognize his existence, as absolutely necessary, from that purposiveness, which we look for in what natural things essentially are and, if possible, in what all things in general essentially are. Whether this succeeds or not, the idea stays correct, and so does its use, as long as that use is kept within the terms of a merely regulative principle.

Complete purposive unity is perfection (considered absolutely). Suppose we don’t find this in the essence of the things that together make up the entire object of experience (that is, of everything we cognize with objective validity), and so in general and necessary laws of nature. How then could we infer from this, of all things, the idea of a supreme, absolutely necessary perfection of a primal being, the origin of all causality? The greatest systematic unity, and so purposive unity as well, is the school of human reason and even the foundation that makes its greatest use possible.

B 723 So the idea of this unity is inseparable from the essence of our reason. The same idea is therefore a lawgiver for us. It is then perfectly natural to assume a lawgiving reason to match it (intellectus archetypus), and to derive from that reason all the systematic unity of nature, which is the object our reason deals with.

When we dealt with the antinomy of pure reason, we claimed that every question pure reason poses has to be answerable. The excuse of the limits of our cognition, which is as unavoidable as it is fair in many questions about nature, can’t be allowed here. For these questions come to us not from the nature of things but from the nature of reason alone, and they concern only its inner makeup. We can now confirm that claim, bold as it looked at first sight, for the two questions in which pure reason has its greatest interest. That brings our examination of its dialectic to a full close.

Suppose we ask, regarding a transcendental theology,* a first question.

B 724 Is there anything distinct from the world that holds the ground of the world’s order and of how it hangs together under general laws? The answer is: no doubt. The world is a sum of appearances, so some transcendental ground must exist for it, that is, a ground that pure understanding alone can think. Second, we might ask whether this being is a substance, of the greatest reality, necessary, and so on. My answer is that the question has no meaning. Whatever category I reach for in trying to conceive such an object has an empirical use only. None has any sense unless it is applied to objects of possible experience, in other words to the world of sense. Outside that field, categories are just labels for concepts. We may admit such concepts, but we can’t understand anything through them. Third, we might ask whether this being, which is distinct from the world, can at least be conceived by analogy with the objects of experience. The answer is: certainly, but only as an object within the idea.

B 725 It is not an object in reality; it is one only as a substratum we do not know, beneath the order, purposiveness and systematic unity in how the world is set up. Reason has to make that the regulative principle of its study of nature. More than that, we can allow certain anthropomorphisms in this idea, without fear or reproach, as long as they serve that regulative principle. For it is always only an idea. It doesn’t refer directly to a being apart from the world. What it refers to is the regulative principle of the world being a systematic unity, and it does so only through a schema of that unity: a supreme intelligence that is, by wise purposes, the world’s author. The idea was not meant to say what the primal ground behind the world’s unity is in itself. It was meant to say how we are to put that ground to use, or rather its idea, relative to the systematic use of reason on the things of the world.

* What I said earlier about the psychological idea, and what it is properly for as a principle of the merely regulative use of reason, saves me the trouble of treating in particular the transcendental illusion that makes us picture that systematic unity of all the manifold of inner sense as a hypostasis. The procedure here is very close to the one the Critique follows with regard to the theological ideal.

But in this way, someone will go on to ask, can we assume one wise and all-powerful author of the world? Without any doubt. And more than that, we must presuppose one. But then we are enlarging what we know past the range of possible experience, aren’t we? No, not at all. All we have presupposed is a something.

B 726 As to what it is in itself, we have no concept at all (it is a merely transcendental object). But take the systematic, purposive order built into the world, which we have to assume whenever we study nature. In relation to that order we have thought of the unknown being solely on the pattern of an intelligence (an empirical concept). That is, with respect to purposes and the perfection that rests on them, we have given it exactly those properties which, under the conditions of our reason, can hold the ground of such a systematic unity. So, relative to the use our reason makes of it in dealing with the world, the idea is thoroughly well grounded. Now suppose we tried to grant it objective validity outright. We would then forget that what we are thinking of is only a being in the idea. We would be starting from a ground that no contemplation of the world can determine, and that would leave us unable to apply the principle in a way that fits the empirical use of reason.

But, someone will ask next, can I in this way make use of the concept and assumption of a highest being when I look at the world rationally? Yes. That is in fact what reason laid this idea down for. Then may I regard arrangements that look as if they had purposes as intentions?

B 727 May I do so by tracing them back to the divine will, although by way of particular dispositions placed in the world for that purpose? Yes, you may do that too. But it must not matter to you whether someone says divine wisdom arranged everything this way for its highest purposes, or says instead that the idea of highest wisdom is a guide for inquiring into nature and a principle of the systematic and purposive unity of nature under general laws of nature, even where we don’t perceive it. In other words, where you do perceive it, it has to be all one to you whether you say: God wisely willed it so, or nature wisely arranged it so. That is because the greatest systematic and purposive unity was precisely what your reason required, as a regulative principle for all study of nature, and it was just what entitled you to put the idea of a highest intelligence at the base, as a schema of the regulative principle. As much purposiveness as you find in the world by that idea, that much confirmation you have that your idea is legitimate. But the principle aimed at nothing except finding a necessary and the greatest possible unity of nature. So whatever we achieve of that unity, we will owe it to the idea of a supreme being. The general laws of nature are the very ones for which the idea was laid down.

B 728 Without contradicting ourselves, we cannot pass over them in order to treat this purposiveness of nature as contingent in its origin and hyperphysical. We had no right to assume a being above nature that has these properties. We could only lay down its idea, so as to see appearances as systematically tied together, by analogy with causal determination.

For the same reason we may also think of the world’s cause, in the idea, in a subtler anthropomorphic way (we couldn’t think anything of it without that): as a being with understanding, with pleasure and displeasure, and with a desire and a will that fit these, and so on. And we may go further and attribute infinite perfection to it, which is far more than anything empirical knowledge of the world’s order could entitle us to. For the regulative law of systematic unity tells us to study nature on the assumption that systematic and purposive unity, together with the greatest possible manifoldness, turns up everywhere, without end. Only a small part of this perfection will we ever spy out or attain. Even so, it belongs to the lawgiving of our own reason to look for this perfection everywhere and to suspect it everywhere.

B 729 Studying nature by this principle must always be to our advantage, and can never be to our harm. Under this conception, it is also clear that what I lay at the base is not the existence or knowledge of such a being, only its idea. So, strictly speaking, I derive nothing from this being, only from its idea, meaning from the nature of the things in the world as that idea would have it. A certain awareness, though undeveloped, of the right use of this concept of our reason seems also to have shaped the modest and fair language of philosophers of all ages. They speak of the wisdom and foresight of nature and of divine wisdom as expressions meaning the same. They even prefer the first expression for as long as the question concerns merely speculative reason. It holds back the arrogance of a bigger claim than we have a right to make. And it points reason back toward its own special field, which is nature.

So pure reason, which at the start seemed to promise nothing less than a widening of knowledge past all the boundaries of experience, contains, if we understand it correctly, nothing except regulative principles. These principles demand a unity greater than the understanding can reach in empirical use. They set the goal of its approach very far out.

B 730 In doing so they bring the understanding into the fullest harmony with itself, by means of systematic unity. But if we misunderstand them, and take them for constitutive principles of transcendent cognitions, through a shining yet deceiving illusion they breed mere persuasion and imagined knowledge, and with them endless contradictions and quarrels.

* * *

So every kind of human cognition begins with intuitions, moves on to concepts, and finishes with ideas. In all three of these elements we have sources of cognition a priori, and at first glance they seem to look down on the boundaries of all experience. Even so, a finished critique convinces us of this: when reason works speculatively, these elements will never carry it past the field of possible experience. It also tells us what this highest cognitive faculty is really meant for. It should use all its methods and principles only to follow nature to its very core, guided by every possible principle of unity, the unity of purposes being the foremost. It should never fly beyond nature’s boundary, because past that line all we have is empty space. In the Transcendental Analytic, it’s true, we examined critically every proposition that could extend our cognition beyond actual experience.

B 731 That examination gave us enough reason to be sure that these propositions never lead to more than a possible experience. Suppose we didn’t distrust even the clearest abstract and general doctrines, and suppose appealing, plausible prospects didn’t tempt us to cast off the constraint those doctrines put on us. Then we could certainly have skipped the exhausting questioning of all the dialectical witnesses that a transcendent reason brings forward to support its claims. For we knew beforehand, with total certainty, that whatever these witnesses assert may be honestly meant but has to be completely worthless, since it concerned news that no human being can ever get. Yet there is no end to the talk until we find the true cause of the illusion, which can mislead even the most sensible person. Besides, our transcendent cognition can be analyzed into its elements. That analysis is a study of our own inner nature, it is worth a good deal on its own account, and a philosopher is even obliged to carry it out. So it was necessary to track this whole effort of speculative reason, empty though it is, back to its first sources, and to do so in detail. That wasn’t the only point, though. The dialectical illusion in this case misleads not just in the judgment it produces.

B 732 It also draws us in through the interest we take in that judgment. It is natural, it always has been, and it will remain so forever. So it was also wise to write out, so to speak, the complete record of this trial and to place it in the archive of human reason, so that similar mistakes can be avoided in the future.