Kant's Theory of Knowledge

by H. A. Prichard

THE TRANSCENDENTAL DEDUCTION OF THE CATEGORIES

The aim of the Transcendental Deduction is to show that the categories, though a priori as originating in the understanding, are valid, i. e. applicable to individual things. It is the part of the Critique which has attracted most attention and which is the most difficult to follow. The difficulty of interpretation is increased rather than diminished by the complete rewriting of this portion in the second edition. For the second version, though it does not imply a change of view, is undoubtedly even more obscure than the first. It indeed makes one new contribution to the subject by adding an important link in the argument,[1] but the importance of the link is nullified by the fact that it is not really the link which it professes to be. The method of treatment adopted here will be to consider only the minimum of passages necessary to elucidate Kant's meaning and to make use primarily of the first edition.

It is necessary, however, first to consider the passage in the Metaphysical Deduction which nominally connects the list of categories with the list of forms of judgement.[2] For its real function is to introduce a new and third account of knowledge, which forms the keynote of the Transcendental Deduction.[3]

In this passage, the meaning of which it is difficult to state satisfactorily, Kant's thought appears to be as follows: 'The activity of thought studied by Formal Logic relates by way of judgement conceptions previously obtained by an analysis of perceptions. For instance, it relates the conceptions of body and of divisibility, obtained by analysis of perceptions of bodies, in the judgement 'Bodies are divisible'. It effects this, however, merely by analysis of the conception 'body'. Consequently, the resulting knowledge or judgement, though a priori, is only analytic, and the conceptions involved originate not from thought but from the manifold previously analysed. But besides the conceptions obtained by analysis of a given manifold, there are others which belong to thought or the understanding as such, and in virtue of which thought originates synthetic a priori knowledge, this activity of thought being that studied by Transcendental Logic. Two questions therefore arise. Firstly, how do these conceptions obtain a matter to which they can apply and without which they would be without content or empty? And, secondly, how does thought in virtue of these conceptions originate synthetic a priori knowledge? The first question is easily answered, for the manifolds of space and time, i. e. individual spaces and individual times, afford matter of the kind needed to give these conceptions content. As perceptions (i. e. as objects of perception), they are that to which a conception can apply, and as pure or a priori perceptions, they are that to which those conceptions can apply which are pure or a priori, as belonging to the understanding. The second question can be answered by considering the process by which this pure manifold of space and time enters into knowledge. All synthetic knowledge, whether empirical or a priori, requires the realization of three conditions. In the first place, there must be a manifold given in perception. In the second place, this manifold must be 'gone through, taken up, and combined'. In other words, if synthesis be defined as 'the act of joining different representations to one another and of including their multiplicity in one knowledge', the manifold must be subjected to an act of synthesis. This is effected by the imagination. In the third place, this synthesis produced by the imagination must be brought to a conception, i. e. brought under a conception which will constitute the synthesis a unity. This is the work of the understanding. The realization of a priori knowledge, therefore, will require the realization of the three conditions in a manner appropriate to its a priori character. There must be a pure or a priori manifold; this is to be found in individual spaces and individual times. There must be an act of pure synthesis of this manifold; this is effected by the pure imagination. Finally, this pure synthesis must be brought under a conception. This is effected by the pure understanding by means of its pure or a priori conceptions, i. e. the categories. This, then, is the process by which a priori knowledge is originated. The activity of thought or understanding, however, which unites two conceptions in a judgement by analysis of them—this being the act studied by Formal Logic—is the same as that which gives unity to the synthesis of the pure manifold of perception —this being the act studied by Transcendental Logic. Consequently, 'the same understanding, and indeed by the same activities whereby in dealing with conceptions it unifies them in a judgement by an act of analysis, introduces by means of the synthetical unity which it produces in the pure manifold of perception a content into its own conceptions, in consequence of which these conceptions are called pure conceptions of the understanding,'[4] and we are entitled to say a priori that these conceptions apply to objects because they are involved in the process by which we acquire a priori knowledge of objects.'

A discussion of the various difficulties raised by the general drift of this passage, as well as by its details,[5] is unnecessary, and would anticipate discussion of the Transcendental Deduction. But it is necessary to draw attention to three points.

In the first place, as has been said, Kant here introduces—and introduces without warning—a totally new account of knowledge. It has its origin in his theory of perception, according to which knowledge begins with the production of sensations in us by things in themselves. Since the spatial world which we come to know consists in a multiplicity of related elements, it is clear that the isolated data of sensation have somehow to be combined and unified, if we are to have this world before us or, in other words, to know it. Moreover, since these empirical data are subject to space and time as the forms of perception, individual spaces and individual times, to which the empirical data will be related, have also to be combined and unified. On this view, the process of knowledge consists in combining certain data into an individual whole and in unifying them through a principle of combination.[6] If the data are empirical, the resulting knowledge will be empirical; if the data are a priori, i. e. individual spaces and individual times, the resulting knowledge will be a priori.[7] This account of knowledge is new, because, although it treats knowledge as a process or act of unifying a manifold, it describes a different act of unification. As Kant first described the faculty of judgement,[8] it unifies a group of particulars through relation to the corresponding universal. As Formal Logic, according to Kant, treats the faculty of judgement, it unifies two conceptions or two prior judgements into a judgement. As Kant now describes the faculty of judgement or thought, it unifies an empirical or an a priori manifold of perception combined into an individual whole, through a conception which constitutes a principle of unity. The difference between this last account and the others is also shown by the fact that while the first two kinds of unification are held to be due to mere analysis of the material given to thought, the third kind of unification is held to be superinduced by thought, and to be in no way capable of being extracted from the material by analysis. Further, this new account of knowledge does not replace the others, but is placed side by side with them. For, according to Kant, there exist both the activity of thought which relates two conceptions in a judgement,[9] and the activity by which it introduces a unity of its own into a manifold of perception. Nevertheless, this new account of knowledge, or rather this account of a new kind of knowledge, must be the important one; for it is only the process now described for the first time which produces synthetic as opposed to analytic knowledge.

In the second place, the passage incidentally explains why, according to Kant, the forms of judgement distinguished by Formal Logic do not involve the categories.[10] For its doctrine is that while thought, if exercised under the conditions under which it is studied by Formal Logic, can only analyse the manifold given to it, and so has, as it were, to borrow from the manifold the unity through which it relates the manifold,[11] yet if an a priori manifold be given to it, it can by means of a conception introduce into the manifold a unity of its own which could not be discovered by analysis of the manifold. Thus thought as studied by Formal Logic merely analyses and consequently does not and cannot make use of conceptions of its own; it can use conceptions of its own only when an a priori manifold is given to it to deal with.

In the third place, there is great difficulty in following the part in knowledge assigned to the understanding. The synthesis of the manifold of perception is assigned to the imagination, a faculty which, like the new kind of knowledge, is introduced without notice. The business of the understanding is to 'bring this synthesis to conceptions' and thereby to 'give unity to the synthesis'. Now the question arises whether 'the activity of giving unity to the synthesis' really means what it says, i. e. an activity which unifies or introduces a unity into the synthesis, or whether it only means an activity which recognizes a unity already given to the synthesis by the imagination. Prima facie Kant is maintaining that the understanding really unifies, or introduces the principle of unity. For the twice-repeated phrase 'give unity to the synthesis' seems unmistakable in meaning, and the important rôle in knowledge is plainly meant to be assigned to the understanding. Kant's language, however, is not decisive; for he speaks of the synthesis of the manifold as that which 'first produces a knowledge which indeed at first may be crude and confused and therefore needs analysis[12]', and he says of the conceptions which give unity to the synthesis that 'they consist solely in the representation[13] of this necessary synthetical unity'.[14] Again, 'to bring the synthesis to a conception' may well be understood to mean 'to recognize the synthesis as an instance of the conception'; and, since Kant is speaking of knowledge, 'to give unity to the synthesis' may only mean 'to give unity to the synthesis for us', i. e. 'to make us aware of its unity'. Moreover, consideration of what thought can possibly achieve with respect to a synthesis presented to it by the imagination renders it necessary to hold that the understanding only recognizes the unity of the synthesis. For if a synthesis has been effected, it must have been effected in accordance with a principle of construction or synthesis, and therefore it would seem that the only work left for the understanding is to discover the principle latent in the procedure of the imagination. At any rate, if the synthesis does not involve a principle of synthesis, it is impossible to see how thought can subsequently introduce a principle. The imagination, then, must be considered to have already introduced the principle of unity into the manifold by combining it in accordance with a conception or principle of combination, and the work of the understanding must be considered to consist in recognizing that the manifold has been thereby combined and unified through the conception. We are therefore obliged to accept one of two alternatives. Either the understanding merely renders the mind conscious of the procedure of a faculty different from itself, viz. the imagination, in which case the important rôle in knowledge, viz. the effecting of the synthesis according to a principle, is played by a faculty different from the understanding; or the imagination is the understanding working unreflectively, and the subsequent process of bringing the synthesis to a conception is merely a process by which the understanding becomes conscious of its own procedure. Moreover, it is the latter alternative which we must accept as more in accordance with the general tenor of Kant's thought. For the synthesis of the imagination is essentially the outcome of activity or spontaneity, and, as such, it belongs to the understanding rather than to the sensibility; in fact we find Kant in one place actually saying that 'it is one and the same spontaneity which at one time under the name of imagination, at another time under that of understanding, introduces connexion into the manifold of perception'.[15] Further, it should be noted that since the imagination must be the understanding working unreflectively, and since it must be that which introduces unity into the manifold, there is some justification for his use of language which implies that the understanding is the source of the unity, though it will not be so in the sense in which the passage under discussion might at first sight lead us to suppose.

We can now turn to the argument of the Transcendental Deduction itself. Kant introduces it in effect by raising the question, 'How is it that, beginning with the isolated data of sense, we come to acquire knowledge?' His aim is to show (1) that knowledge requires the performance of certain operations by the mind upon the manifold of sense; (2) that this process is a condition not merely of knowledge, but also of self-consciousness; and (3) that, since the manifold is capable of entering into knowledge, and since we are capable of being self-conscious, the categories, whose validity is implied by this process, are valid.

Kant begins by pointing out[16] that all knowledge, a priori as well as empirical, requires the manifold, produced successively in the mind, to be subjected to three operations.

1. Since the elements of the manifold are as given mere isolated units, and since knowledge is the apprehension of a unity of connected elements, the mind must first run through the multiplicity of sense and then grasp it together into a whole, i. e. into an image.[17] This act is an act of synthesis; it is called 'the synthesis of apprehension' and is ascribed to the imagination. It must be carried out as much in respect of the pure or a priori elements of space and time as in respect of the manifold of sensation, for individual spaces and times contain a multiplicity which, to be apprehended, must be combined.[18] The necessity of this act of synthesis is emphasized in the second edition. "We cannot represent anything as combined in the object without having previously combined it ourselves. Of all representations, combination is the only one which cannot be given through objects,[19] but can be originated only by the subject itself because it is an act of its own activity."[20]

2. Since the data of perception are momentary, and pass away with perception, the act of grasping them together requires that the mind shall reproduce the past data in order to combine them with the present datum. "It is plain that if I draw a line in thought, or wish to think of the time from one midday to another, or even to represent to myself a certain number, I must first necessarily grasp in thought these manifold representations one after another. But if I were continually to lose from my thoughts the preceding representations (the first parts of the line, the preceding parts of time or the units successively represented), and were not to reproduce them, while I proceeded to the succeeding parts, there could never arise a complete representation, nor any of the thoughts just named, not even the first and purest fundamental representations of space and time."[21] This act of reproduction is called 'the synthesis of reproduction in the imagination'.[22]

Further, the necessity of reproduction brings to light a characteristic of the synthesis of apprehension. "It is indeed only an empirical law, according to which representations which have often followed or accompanied one another in the end become associated, and so form a connexion, according to which, even in the absence of the object, one of these representations produces a transition of the mind to another by a fixed rule. But this law of reproduction presupposes that phenomena themselves are actually subject to such a rule, and that in the manifold of their representations there is a concomitance or sequence, according to a fixed rule; for, without this, our empirical imagination would never find anything to do suited to its capacity, and would consequently remain hidden within the depths of the mind as a dead faculty, unknown to ourselves. If cinnabar were now red, now black, now light, now heavy, if a man were changed now into this, now into that animal shape, if our fields were covered on the longest day, now with fruit, now with ice and snow, then my empirical faculty of imagination could not even get an opportunity of thinking of the heavy cinnabar when there occurred the representation of red colour; or if a certain name were given now to one thing, now to another, or if the same thing were called now by one and now by another name, without the control of some rule, to which the phenomena themselves are already subject, no empirical synthesis of reproduction could take place."

"There must then be something which makes this very reproduction of phenomena possible, by being the a priori foundation of a necessary synthetical unity of them. But we soon discover it, if we reflect that phenomena are not things in themselves, but the mere play of our representations, which in the end resolve themselves into determinations of our internal sense. For if we can prove that even our purest a priori perceptions afford us no knowledge, except so far as they contain such a combination of the manifold as renders possible a thoroughgoing synthesis of reproduction, then this synthesis of imagination is based, even before all experience, on a priori principles, and we must assume a pure transcendental synthesis of the imagination which lies at the foundation of the very possibility of all experience (as that which necessarily presupposes the reproducibility of phenomena)."[23]

In other words, the faculty of reproduction, if it is to get to work, presupposes that the elements of the manifold are parts of a necessarily related whole; or, as Kant expresses it later, it presupposes the affinity of phenomena; and this affinity in turn presupposes that the synthesis of apprehension by combining the elements of the manifold on certain principles makes them parts of a necessarily related whole.[24]

3. Kant introduces the third operation, which he calls 'the synthesis of recognition in the conception',[25] as follows:

"Without consciousness that what we are thinking is identical with what we thought a moment ago, all reproduction in the series of representations would be in vain. For what we are thinking would be a new representation at the present moment, which did not at all belong to the act by which it was bound to have been gradually produced, and the manifold of the same would never constitute a whole, as lacking the unity which only consciousness can give it. If in counting I forget that the units which now hover before my mind have been gradually added by me to one another, I should not know the generation of the group through this successive addition of one to one, and consequently I should not know the number, for this conception consists solely in the consciousness of this unity of the synthesis."

"The word 'conception'[26] might itself lead us to this remark. For it is this one consciousness which unites the manifold gradually perceived and then also reproduced into one representation. This consciousness may often be only weak, so that we connect it with the production of the representation only in the result but not in the act itself, i. e. immediately; but nevertheless there must always be one consciousness, although it lacks striking clearness, and without it conceptions, and with them knowledge of objects, are wholly impossible."[27]

Though the passage is obscure and confused, its general drift is clear. Kant, having spoken hitherto only of the operation of the imagination in apprehension and reproduction, now wishes to introduce the understanding. He naturally returns to the thought of it as that which recognizes a manifold as unified by a conception, the manifold, however, being not a group of particulars unified through the corresponding universal or conception, but the parts of an individual image, e. g. the parts of a line or the constituent units of a number, and the conception which unifies it being the principle on which these parts are combined.[28] His main point is that it is not enough for knowledge that we should combine the manifold of sense into a whole in accordance with a specific principle,[29] but we must also be in some degree conscious of our continuously identical act of combination,[30] this consciousness being at the same time a consciousness of the special unity of the manifold. For the conception which forms the principle of the combination has necessarily two sides; while from our point of view it is the principle according to which we combine and which makes our combining activity one, from the point of view of the manifold it is the special principle[31] by which the manifold is made one. If I am to count a group of five units, I must not only add them, but also be conscious of my continuously identical act of addition, this consciousness consisting in the consciousness that I am successively taking units up to, and only up to, five, and being at the same time a consciousness that the units are acquiring the unity of being a group of five. It immediately follows, though Kant does not explicitly say so, that all knowledge implies self-consciousness. For the consciousness that we have been combining the manifold on a certain definite principle is the consciousness of our identity throughout the process, and, from the side of the manifold, it is just that consciousness of the manifold as unified by being brought under a conception which constitutes knowledge. Even though it is Kant's view that the self-consciousness need only be weak and need only arise after the act of combination, when we are aware of its result, still, without it, there will be no consciousness of the manifold as unified through a conception and therefore no knowledge. Moreover, if the self-consciousness be weak, the knowledge will be weak also, so that if it be urged that knowledge in the strictest sense requires the full consciousness that the manifold is unified through a conception, it must be allowed that knowledge in this sense requires a full or clear self-consciousness.

As is to be expected, however, the passage involves a difficulty concerning the respective functions of the imagination and the understanding. Is the understanding represented as only recognizing a principle of unity introduced into the manifold by the imagination, or as also for the first time introducing a principle of unity? At first sight the latter alternative may seem the right interpretation. For he says that unless we were conscious that what we are thinking is identical with what we thought a moment ago, 'what we are thinking would be a new representation which did not at all belong to the act by which it was bound to have been gradually produced, and the manifold of the same would never constitute a whole, as lacking the unity which only consciousness can give it.'[32] Again, in speaking of a conception—which of course implies the understanding—he says that 'it is this one consciousness which unites the manifold gradually perceived and then reproduced into one representation'.[33] But these statements are not decisive, for he uses the term 'recognition' in his formula for the work of the understanding, and he illustrates its work by pointing out that in counting we must remember that we have added the units. Moreover, there is a consideration which by itself makes it necessary to accept the former interpretation. The passage certainly represents the understanding as recognizing the identical action of the mind in combining the manifold on a principle, whether or not it also represents the understanding as the source of this activity. But if it were the understanding which combined the manifold, there would be no synthesis which the imagination could be supposed to have performed,[34] and therefore it could play no part in knowledge at all, a consequence which must be contrary to Kant's meaning. Further if, as the general tenor of the deduction shows, the imagination is really only the understanding working unreflectively,[35] we are able to understand why Kant should for the moment cease to distinguish between the imagination and the understanding, and consequently should use language which implies that the understanding both combines the manifold on a principle and makes us conscious of our activity in so doing. Hence we may say that the real meaning of the passage should be stated thus: 'Knowledge requires one consciousness which, as imagination, combines the manifold on a definite principle constituted by a conception,[36] and, as understanding, is to some extent conscious of its identical activity in so doing, this self-consciousness being, from the side of the whole produced by the synthesis, the consciousness of the conception by which the manifold is unified.'

Hitherto there has been no mention of an object of knowledge, and since knowledge is essentially knowledge of an object, Kant's next task is to give such an account of an object of knowledge as will show that the processes already described are precisely those which give our representations, i. e. the manifold of sense, relation to an object, and consequently yield knowledge.

He begins by raising the question, 'What do we mean by the phrase 'an object of representations'?'[37] He points out that a phenomenon, since it is a mere sensuous representation, and not a thing in itself existing independently of the faculty of representations, is just not an object. To the question, therefore, 'What is meant by an object corresponding to knowledge and therefore distinct from it?' we are bound to answer from the point of view of the distinction between phenomena and things in themselves, that the object is something in general = x, i. e. the thing in itself of which we know only that it is and not what it is. There is, however, another point of view from which we can say something more about an object of representations and the correspondence of our representations to it, viz. that from which we consider what is involved in the thought of the relation of knowledge or of a representation to its object. "We find that our thought of the relation of all knowledge to its object carries with it something of necessity, since its object is regarded as that which prevents our cognitions[38] being determined at random or capriciously, and causes them to be determined a priori in a certain way, because in that they are to relate to an object, they must necessarily also, in relation to it, agree with one another, that is to say, they must have that unity which constitutes the conception of an object."[39]

Kant's meaning seems to be this: 'If we think of certain representations, e. g. certain lines[40] or the representations of extension, impenetrability, and shape,[41] as related to an object, e. g. to an individual triangle or an individual body, we think that they must be mutually consistent or, in other words, that they must have the unity of being parts of a necessarily related whole or system, this unity in fact constituting the conception of an object in general, in distinction from the conception of an object of a particular kind. The latter thought in turn involves the thought of the object of representations as that which prevents them being anything whatever and in fact makes them parts of a system. The thought therefore of representations as related to an object carries with it the thought of a certain necessity, viz. the necessary or systematic unity introduced into the representations by the object. Hence by an object of representations we mean something which introduces into the representations a systematic unity which constitutes the nature of an object in general, and the relatedness of representations to, or their correspondence with, an object involves their systematic unity.'[42]

Certain points, however, should be noticed. In the first place, Kant is for the moment tacitly ignoring his own theory of knowledge, in accordance with which the object proper, i. e. the thing in itself, is unknowable, and is reverting to the ordinary conception of knowledge as really knowledge of its object. For the elements which are said, in virtue of being related to an object, to agree and to have the unity which constitutes the conception of an object must be elements of an object which we know; for if the assertion that they agree is to be significant, they must be determinate parts or qualities of the object, e. g. the sides of an individual triangle or the impenetrability or shape of an individual body, and therefore it is implied that we know that the object has these parts or qualities. In the second place, both the problem which Kant raises and the clue which he offers for its solution involve an impossible separation of knowledge or a representation from its object. Kant begins with the thought of a phenomenon as a mere representation which, as mental, and as the representation of an object, is just not an object, and asks, 'What is meant by the object of it?' He finds the clue to the answer in the thought that though a representation or idea when considered in itself is a mere mental modification, yet, when considered as related to an object, it is subject to a certain necessity. In fact, however, an idea or knowledge is essentially an idea or knowledge of an object, and we are bound to think of it as such. There is no meaning whatever in saying that the thought of an idea as related to an object carries with it something of necessity, for to say so implies that it is possible to think of it as unrelated to an object. Similarly there is really no meaning in the question, 'What is meant by an object corresponding to knowledge or to an idea?' for this in the same way implies that we can first think of an idea as unrelated to an object and then ask, 'What can be meant by an object corresponding to it?'[43] In the third place, Kant only escapes the absurdity involved in the thought of a mere idea or a mere representation by treating representations either as parts or as qualities of an object. For although he speaks of our cognitions,[44] i. e. of our representations, as being determined by the object, he says that they must agree, i. e. they must have that unity which constitutes the conception of an object, and he illustrates representations by the sides of an individual triangle and the impenetrability and shape of an individual body, which are just as 'objective' as the objects to which they relate. The fact is that he really treats a representation not as his problem requires that it should be treated, i. e. as a representation of something, but as something represented,[45] i. e. as something of which we are aware, viz. a part or a quality of an object. In the fourth place, not only is that which Kant speaks of as related to an object really not a representation, but also—as we see if we consider the fact which Kant has in mind—that to which he speaks of it as related is really not an object but one and the same object to which another so-called representation is related. For what Kant says is that representations as related to an object must agree among themselves. But this statement, to be significant, implies that the object to which various representations are related is one and the same. Otherwise why should the representations agree? In view, therefore, of these last two considerations we must admit that the real thought underlying Kant's statement should be expressed thus: 'We find that the thought that two or more parts or qualities of an object relate to one and the same object carries with it a certain necessity, since this object is considered to be that which prevents these parts or qualities which we know it to possess from being determined at random, because by being related to one and the same object, they must agree among themselves.' The importance of the correction lies in the fact that what Kant is stating is not what he thinks he is stating. He is really stating the implication of the thought that two or more qualities or parts of some object or other, which, as such, already relate to an object, relate to one and the same object. He thinks he is stating the implication of the thought that a representation which in itself has no relation to an object, has relation to an object. And since his problem is simply to determine what constitutes the relatedness to an object of that which in itself is a mere representation, the distinction is important; for it shows that he really elucidates it by an implication respecting something which already has relation to an object and is not a mental modification at all, but a quality or a part of an object.

Kant continues thus: "But it is clear that, since we have to do only with the manifold of our representations, and the x, which corresponds to them (the object), since it is to be something distinct from all our representations, is for us nothing, the unity which the object necessitates can be nothing else than the formal unity of consciousness in the synthesis of the manifold of representations." [I. e. since the object which produces systematic unity in our representations is after all only the unknown thing in itself, viz. x,[46] any of the parts or qualities of which it is impossible to know, that to which it gives unity can be only our representations and not its own parts or qualities. For, since we do not know any of its parts or qualities, these representations cannot be its parts or qualities. Consequently, the unity produced by this x can only be the formal unity of the combination of the manifold in consciousness.[47]] "Then and then only do we say that we know the object," [i. e. we know that the manifold relates to an object[48]] "if we have produced synthetical unity in the manifold of perception. But this unity would be impossible, if the perception could not be produced by means of such a function of synthesis according to a rule as renders the reproduction of the manifold a priori necessary, and a conception in which the manifold unifies itself possible. Thus we think a triangle as an object, in that we are conscious of the combination of three straight lines in accordance with a rule by which such a perception can at any time be presented. This unity of the rule determines all the manifold and limits it to conditions which make the unity of apperception possible, and the conception of this unity is the representation of the object=x, which I think through the aforesaid predicates of a triangle." [I. e., apparently, 'to conceive this unity of the rule is to represent to myself the object x, i. e. the thing in itself,[49] of which I come to think by means of the rule of combination.']

In this passage several points claim attention. In the first place, it seems impossible to avoid the conclusion that in the second sentence the argument is exactly reversed. Up to this point, it is the thing in itself which produces unity in our representations. Henceforward it is we who produce the unity by our activity of combining the manifold. The discrepancy cannot be explained away, and its existence can only be accounted for by the exigencies of Kant's position. When he is asking 'What is meant by the object (beyond the mind) corresponding to our representations?' he has to think of the unity of the representations as due to the object. But when he is asking 'How does the manifold of sense become unified?' his view that all synthesis is due to the mind compels him to hold that the unity is produced by us. In the second place, the passage introduces a second object in addition to the thing in itself, viz. the phenomenal object, e. g. a triangle considered as a whole of parts unified on a definite principle.[50] It is this object which, as the object that we know, is henceforward prominent in the first edition, and has exclusive attention in the second. The connexion between this object and the thing in itself appears to lie in the consideration that we are only justified in holding that the manifold of sense is related to a thing in itself when we have unified it and therefore know it to be a unity, and that to know it to be a unity is ipso facto to be aware of it as related to a phenomenal object; in other words, the knowledge that the manifold is related to an object beyond consciousness is acquired through our knowledge of its relatedness to an object within consciousness. In the third place, in view of Kant's forthcoming vindication of the categories, it is important to notice that the process by which the manifold is said to acquire relation to an object is illustrated by a synthesis on a particular principle which constitutes the phenomenal object an object of a particular kind. The synthesis which enables us to recognize three lines as an object is not a synthesis based on general principles constituted by the categories, but a synthesis based on the particular principle that the three lines must be so put together as to form an enclosed space. Moreover, it should be noticed that the need of a particular principle is really inconsistent with his view that relation to an object gives the manifold the systematic unity which constitutes the conception of an object, or that at least a [Greek: hysteron proteron] is involved. For if the knowledge that certain representations form a systematic unity justifies our holding that they relate to an object, it would seem that in order to know that they relate to an object we need not know the special character of their unity. Yet, as Kant states the facts, we really have to know the special character of their unity in order to know that they possess systematic unity in general.[51] Lastly, it is easy to see the connexion of this account of an object of representations with the preceding account of the synthesis involved in knowledge. Kant had said that knowledge requires a synthesis of the imagination in accordance with a definite principle, and the recognition of the principle of the synthesis by the understanding. From this point of view it is clear that the aim of the present passage is to show that this process yields knowledge of an object; for it shows that this process yields knowledge of a phenomenal object of a particular kind, e. g. of a triangle or of a body, and that this object as such refers to what after all is the object, viz. the thing in itself.

The position reached by Kant so far is this. Knowledge, as being knowledge of an object, consists in a process by which the manifold of perception acquires relation to an object. This process again is a process of combination of the manifold into a systematic whole upon a definite principle, accompanied by the consciousness in some degree of the act of combination, and therefore also of the acquisition by the manifold of the definite unity which forms the principle of combination. In virtue of this process there is said to be 'unity of consciousness in the synthesis of the manifold', a phrase which the context justifies us in understanding as a condensed expression for a situation in which (1) the manifold of sense is a unity of necessarily related parts, (2) there is consciousness of this unity, and (3) the consciousness which combines and is conscious of combining the manifold, as being necessarily one and the same throughout this process, is itself a unity.

Kant then proceeds to introduce what he evidently considers the keystone of his system, viz. 'transcendental apperception.'

"There is always a transcendental condition at the basis of any necessity. Hence we must be able to find a transcendental ground of the unity of consciousness in the synthesis of the manifold of all our perceptions, and therefore also of the conceptions of objects in general, consequently also of all objects of experience, a ground without which it would be impossible to think any object for our perceptions; for this object is no more than that something, the conception of which expresses such a necessity of synthesis."

"Now this original and transcendental condition is no other than transcendental apperception. The consciousness of self according to the determinations of our state in internal sense-perception is merely empirical, always changeable; there can be no fixed or permanent self in this stream of internal phenomena, and this consciousness is usually called internal sense or empirical apperception. That which is necessarily to be represented as numerically identical cannot be thought as such by means of empirical data. The condition which is to make such a transcendental presupposition valid must be one which precedes all experience, and makes experience itself possible."

"Now no cognitions[52] can occur in us, no combination and unity of them with one another, without that unity of consciousness which precedes all data of perception, and by relation to which alone all representation of objects is possible. This pure original unchangeable consciousness I shall call transcendental apperception. That it deserves this name is clear from the fact that even the purest objective unity, viz. that of a priori conceptions (space and time) is only possible by relation of perceptions to it. The numerical unity of this apperception therefore forms the a priori foundation of all conceptions, just as the multiplicity of space and time is the foundation of the perceptions of the sensibility."[53]

The argument is clearly meant to be 'transcendental' in character; in other words, Kant continues to argue from the existence of knowledge to the existence of its presuppositions. We should therefore expect the passage to do two things: firstly, to show what it is which is presupposed by the 'unity of consciousness in the synthesis of the manifold'[54]; and secondly, to show that this presupposition deserves the title 'transcendental apperception'. Unfortunately Kant introduces 'transcendental apperception' after the manner in which he introduced the 'sensibility', the 'imagination' and the 'understanding', as if it were a term with which every one is familiar, and which therefore needs little explanation. To interpret the passage, it seems necessary to take it in close connexion with the preceding account of the three 'syntheses' involved in knowledge, and to bear in mind that, as a comparison of passages will show, the term 'apperception', which Kant borrows from Leibniz, always has for Kant a reference to consciousness of self or self-consciousness. If this be done, the meaning of the passage seems to be as follows:

'To vindicate the existence of a self which is necessarily one and the same throughout its representations, and which is capable of being aware of its own identity throughout, it is useless to appeal to that consciousness of ourselves which we have when we reflect upon our successive states. For, although in being conscious of our states we are conscious of ourselves we are not conscious of ourselves as unchanging. The self as going through successive states is changing, and even if in fact its states did not change, its identity would be only contingent; it need not continue unchanged. Consequently, the only course possible is to show that the self-consciousness in question is presupposed in any experience or knowledge. Now it is so presupposed. For, as we have already shown, the relation of representations to an object presupposes one consciousness which combines and unifies them, and is at the same time conscious of the identity of its own action in unifying them. This consciousness is the ground of the unity of consciousness in the synthesis of the manifold. It may fairly be called transcendental, because even a conception which relates to space or time, and therefore is the most remote from sensation, presupposes one consciousness which combines and unifies the manifold of space and time through the conception, and is conscious of the identity of its own action in so doing. It may, therefore, be regarded as the presupposition of all conceiving or bringing a manifold under a conception, and therefore of all knowledge. Consequently, since knowledge is possible, i. e. since the manifold of representations can be related to an object, there must be one self capable of being aware of its own identity throughout its representations.'

At this point of Kant's argument, however, there seems to occur an inversion of the thought. Hitherto, Kant has been arguing from the possibility of knowledge to the possibility of the consciousness of our own identity. But in the next paragraph he appears to reverse this procedure and to argue from the possibility of self-consciousness to the possibility of knowledge.

"But it is just this transcendental unity of apperception[55] which forms, from all possible phenomena which can be together in one experience, a connexion of them according to laws. For this unity of consciousness would be impossible, if the mind in the knowledge of the manifold could not become conscious of the identity of the function whereby it unites the manifold synthetically in one knowledge. Consequently, the original and necessary consciousness of the identity of oneself is at the same time a consciousness of an equally necessary unity of the synthesis of all phenomena according to conceptions, i. e. according to rules which not only make them necessarily reproducible, but thereby determine an object for their perception, i. e. determine the conception of something in which they are necessarily connected. For the mind could not possibly think the identity of itself in the manifold of its representations, and this indeed a priori, if it had not before its eyes the identity of its action which subjects all synthesis of apprehension (which is empirical) to a transcendental unity, and first makes possible its connexion according to rules."

The argument seems indisputably to be as follows: 'The mind is necessarily able to be aware of its own identity throughout its manifold representations. To be aware of this, it must be aware of the identity of the activity by which it combines the manifold of representations into a systematic whole. Therefore it must be capable of combining, and of being conscious of its activity in combining, all phenomena which can be its representations into such a whole. But this process, from the point of view of the representations combined, is the process by which they become related to an object and so enter into knowledge. Therefore, since we are capable of being conscious of our identity with respect to all phenomena which can be our representations, the process of combination and consciousness of combination which constitutes knowledge must be possible with respect to them.' Thus the thought of this and the preceding paragraph seems to involve a circle. First the possibility of self-consciousness is deduced from the possibility of knowledge, and then the possibility of knowledge is deduced from the possibility of self-consciousness.

An issue therefore arises, the importance of which can be seen by reference to the final aim of the 'deduction', viz. the vindication of the categories. The categories are 'fundamental conceptions which enable us to think objects in general[56] for phenomena'[57]; in other words, they are the principles of the synthesis by which the manifold of sense becomes related to an object. Hence, if this be granted, the proof that the categories are applicable to objects consists in showing that the manifold can be subjected to this synthesis. The question therefore arises whether Kant's real starting-point for establishing the possibility of this synthesis and therefore the applicability of the categories, is to be found in the possibility of knowledge, or in the possibility of self-consciousness, or in both. In other words, does Kant start from the position that all representations must be capable of being related to an object, or from the position that we must be capable of being conscious of our identity with respect to all of them, or from both?

Prima facie the second position is the more plausible basis for the desired conclusion. On the one hand, it does not seem obvious that the manifold must be capable of being related to an object; for even if it be urged that otherwise we should have only 'a random play of representations, less than a dream'[58], it may be replied, that this might be or might come to be the case. On the other hand, the fact that our representations are ours necessarily seems to presuppose that we are identical subjects of these representations, and recognition of this fact is the consciousness of our identity.

If we turn to the text for an answer to this question, we find that Kant seems not only to use both starting-points, but even to regard them as equivalents. Thus in introducing the categories[59] Kant begins by appealing to the necessity for knowledge that representations should relate to an object.

"Unity of synthesis according to empirical conceptions would be purely contingent, and were these not based on a transcendental ground of unity, it would be possible for a confused crowd of phenomena to fill our soul, without the possibility of experience ever arising therefrom. But then also all relation of knowledge to objects would fall away, because knowledge would lack connexion according to universal and necessary laws; it would be thoughtless perception but never knowledge, and therefore for us as good as nothing."

"The a priori conditions of any possible experience whatever are at the same time conditions of the possibility of the objects of experience. Now I assert that the above mentioned categories are nothing but the conditions of thinking in any possible experience, just as space and time are the conditions of perception requisite for the same. The former therefore are also fundamental conceptions by which we think objects in general for phenomena, and are therefore objectively valid a priori—which is exactly what we wished to know."

The next sentence, however, bases the necessity of the categories on the possibility of self-consciousness, without giving any indication that a change of standpoint is involved.

"But the possibility, nay, even the necessity, of these categories rests on the relation which the whole sensibility, and with it also all possible phenomena, have to original apperception, a relation which forces everything to conform to the conditions of the thoroughgoing unity of self-consciousness, i. e. to stand under universal functions of synthesis, i. e. of synthesis according to conceptions, as that wherein alone apperception can prove a priori its thorough-going and necessary identity."

Finally, the conclusion of the paragraph seems definitely to treat both starting-points as really the same.[60] "Thus the conception of a cause is nothing but a synthesis (of the consequent in the time series with other phenomena) according to conceptions; and without such a unity, which has its a priori rule and subjects phenomena to itself, thorough-going and universal and therefore necessary unity of consciousness in the manifold of sense-perceptions would not be met with. But then also these perceptions would belong to no experience, consequently they would have no object, and would be nothing but a blind play of representations, less than a dream."

The fact is that since for Kant the synthesis of representations in accordance with the categories, accompanied by the consciousness of it, is at once the necessary and sufficient condition of the relatedness of representations to an object and of the consciousness of our identity with respect to them, it seems to him to be one and the same thing whether, in vindicating the synthesis, we appeal to the possibility of knowledge or to the possibility of self-consciousness, and it even seems possible to argue, via the synthesis, from knowledge to self-consciousness and vice versa.

Nevertheless, it remains true that the vindication of the categories is different, according as it is based upon the possibility of relating representations to an object or upon the possibility of becoming self-conscious with respect to them. It also remains true that Kant vindicates the categories in both ways. For while, in expounding the three so-called syntheses involved in knowledge, he is vindicating the categories from the point of view of knowledge, when he comes to speak of transcendental apperception, of which the central characteristic is the consciousness of self involved, there is a shifting of the centre of gravity. Instead of treating representations as something which can become related to an object, he now treats them as something of which, as belonging to a self, the self must be capable of being conscious as its own, and argues that a synthesis in accordance with the categories is required for this self-consciousness. It must be admitted then—and the admission is only to be made with reluctance—that when Kant reaches transcendental apperception, he really adopts a new starting-point,[61] and that the passage which introduces transcendental apperception by showing it to be implied in knowledge[62] only serves to conceal from Kant the fact that, from the point of view of the deduction of the categories, he is really assuming without proof the possibility of self-consciousness with respect to all our representations, as a new basis for argument.

The approach to the categories from the side of self-consciousness is, however, more prominent in the second edition, and consequently we naturally turn to it for more light on this side of Kant's position. There Kant vindicates the necessity of the synthesis from the side of self-consciousness as follows:[63]

"[1.] It must be possible that the 'I think' should accompany all my representations; for otherwise something would be represented in me which could not be thought; in other words, the representation would be either impossible or at least for me nothing. [2.] That representation which can be given before all thought is called perception. All the manifold of perception has therefore a necessary relation to the 'I think' in the same subject in which this manifold is found. [3.] But this representation[64] [i. e. the 'I think'] is an act of spontaneity, i. e. it cannot be regarded as belonging to sensibility. I call it pure apperception, to distinguish it from empirical apperception, or original apperception also, because it is that self-consciousness which, while it gives birth to the representation 'I think', which must be capable of accompanying all others and is one and the same in all consciousness, cannot itself be accompanied by any other.[65] [4.] I also call the unity of it the transcendental unity of self-consciousness, in order to indicate the possibility of a priori knowledge arising from it. For the manifold representations which are given in a perception would not all of them be my representations, if they did not all belong to one self-consciousness, that is, as my representations (even though I am not conscious of them as such), they must necessarily conform to the condition under which alone they can stand together in a universal self-consciousness, because otherwise they would not all belong to me. From this original connexion much can be concluded."

[5.] "That is to say, this thorough-going identity of the apperception of a manifold given in perception contains a synthesis of representations,[66] and is possible only through the consciousness of this synthesis.[67] [6.] For the empirical consciousness which accompanies different representations is in itself fragmentary, and without relation to the identity of the subject. [7.] This relation, therefore, takes place not by my merely accompanying every representation with consciousness, but by my adding one representation to another, and being conscious of the synthesis of them. [8.] Consequently, only because I can connect a manifold of given representations in one consciousness, is it possible for me to represent to myself the identity of consciousness in these representations; i. e. the analytical unity of apperception is possible only under the presupposition of a synthetical unity. [9.] The thought, 'These representations given in perception belong all of them to me' is accordingly just the same as, 'I unite them in one self-consciousness, or at least can so unite them;' [10.] and although this thought is not itself as yet the consciousness of the synthesis of representations, it nevertheless presupposes the possibility of this synthesis; that is to say, it is only because I can comprehend the manifold of representations in one consciousness, that I call them all my representations; for otherwise I should have as many-coloured and varied a self as I have representations of which I am conscious. [11.] Synthetical unity of the manifold of perceptions, as given a priori, is therefore the ground of the identity of apperception itself, which precedes a priori all my determinate thinking. [12.] But connexion does not lie in the objects, nor can it be borrowed from them through perception and thereby first taken up into the understanding, but it is always an operation of the understanding which itself is nothing more than the faculty of connecting a priori, and of bringing the manifold of given representations under the unity of apperception, which principle is the highest in all human knowledge."

[13.] "Now this principle of the necessary unity of apperception is indeed an identical, and therefore an analytical, proposition, but nevertheless it declares a synthesis of the manifold given in a perception to be necessary, without which the thorough-going identity of self-consciousness cannot be thought. [14.] For through the Ego, as a simple representation, is given no manifold content; in perception, which is different from it, a manifold can only be given, and through connexion in one consciousness it can be thought. An understanding, through whose self-consciousness all the manifold would eo ipso be given, would perceive; our understanding can only think and must seek its perception in the senses. [15.] I am, therefore, conscious of the identical self, in relation to the manifold of representations given to me in a perception, because I call all those representations mine, which constitute one. [16.] But this is the same as to say that I am conscious a priori of a necessary synthesis of them, which is called the original synthetic unity of apperception, under which all representations given to me stand, but also under which they must be brought through a synthesis."[68]

Though this passage involves many difficulties, the main drift of it is clear. Kant is anxious to establish the fact that the manifold of sense must be capable of being combined on principles, which afterwards turn out to be the categories, by showing this to be involved in the fact that we must be capable of being conscious of ourselves as the identical subject of all our representations. To do this, he seeks to prove in the first paragraph that self-consciousness in this sense must be possible, and in the second that this self-consciousness presupposes the synthesis of the manifold.

Examination of the argument, however, shows that the view that self-consciousness must be possible is, so far as Kant is concerned,[69] an assumption for which Kant succeeds in giving no reason at all, and that even if it be true, it cannot form a basis from which to deduce the possibility of the synthesis.

Before, however, we attempt to prove this, it is necessary to draw attention to three features of the argument. In the first place, it implies a somewhat different account of self-consciousness to that implied in the passages of the first edition which we have already considered. Self-consciousness, instead of being the consciousness of the identity of our activity in combining the manifold, is now primarily the consciousness of ourselves as identical subjects of all our representations, i. e. it is what Kant calls the analytical unity of apperception; and consequently it is somewhat differently related to the activity of synthesis involved in knowledge. Instead of being regarded as the consciousness of this activity, it is regarded as presupposing the consciousness of the product of this activity, i. e. of the connectedness[70] of the manifold produced by the activity, this consciousness being what Kant calls the synthetical unity of apperception.[71] In the second place, it is plain that Kant's view is not that self-consciousness involves the consciousness of our representations as a connected whole, but that it involves the consciousness of them as capable of being connected by a synthesis. Yet, if it is only because I can connect (and therefore apprehend as connected) a manifold of representations in one consciousness, that I can represent to myself the identity of consciousness in these representations, self-consciousness really requires the consciousness of our representations as already connected; the mere consciousness of our representations as capable of being connected would not be enough. The explanation of the inconsistency seems to lie in the fact that the synthetic unity of which Kant is thinking is the unity of nature. For, as Kant of course was aware, in our ordinary consciousness we do not apprehend the interconnexion of the parts of nature in detail, but only believe that there is such an interconnexion; consequently he naturally weakened the conclusion which he ought to have drawn, viz. that self-consciousness presupposes consciousness of the synthesis, in order to make it conform to the facts of our ordinary consciousness. Yet, if his argument is to be defended, its conclusion must be taken in the form that self-consciousness presupposes consciousness of the actual synthesis or connexion and not merely of the possibility of it. In the third place, Kant twice in this passage[72] definitely makes the act of synthesis, which his argument maintains to be the condition of consciousness of the identity of ourselves, the condition of the identity of ourselves. The fact is that, on Kant's view, the act of synthesis of the representations is really a condition of their belonging to one self, the self being presupposed to be a self capable of self-consciousness.[73]

We may now turn to the first of the two main points to be considered, viz. the reason given by Kant for holding that self-consciousness must be possible. In the first paragraph (§§ 1-4) Kant appears twice to state a reason, viz. in §§ 1 and 4. What is meant by the first sentence, "It must be possible that the 'I think' should accompany all my representations; for otherwise something would be represented in me which could not be thought; in other words, the representation would either be impossible or at least for me nothing"? It is difficult to hold that 'my representations' here means objects of which I am aware, and that the thesis to be established is that I must be capable of being conscious of my own identity throughout all awareness or thought of objects. For the next sentence refers to perceptions as representations which can be given previously to all thought, and therefore, presumably, as something of which I am not necessarily aware. Again, the ground adduced for the thesis would be in part a mere restatement of it, and in part nonsense. It would be 'otherwise something would be apprehended with respect to which I could not be aware that I was apprehending it; in other words, I could not apprehend it [since otherwise I could be aware that I was apprehending it]', the last words being incapable of any interpretation. It is much more probable that though Kant is leading up to self-consciousness, the phrase 'I think' here refers not to 'consciousness that I am thinking', but to 'thinking'. He seems to mean 'It must be possible to apprehend all my 'affections' (i. e. sensations or appearances in me), for otherwise I should have an affection of which I could not be aware; in other words, there could be no such affection, or at least it would be of no possible importance to me.'[74] And on this interpretation self-consciousness is not introduced till § 3, and then only surreptitiously. On neither interpretation, however, does Kant give the vestige of a reason for the possibility of self-consciousness. Again, it seems clear that in § 4 'my representations', and 'representations which belong to me' mean objects of which I am aware (i. e. something presented); for he says of my representations, not that I may not be conscious of them—which he should have said if 'my representations' meant my mental affections of which I could become conscious—but that I may not be conscious of them as my representations. Consequently in § 4 he is merely asserting that I must be able to be conscious of my identity throughout my awareness of objects. So far, then, we find merely the assertion that self-consciousness must be possible.[75]

In the next paragraph[76]—which is clearly meant to be the important one—Kant, though he can hardly be said to be aware of it, seems to assume that it is the very nature of a knowing self, not only to be identical throughout its thoughts or apprehendings, but to be capable of being conscious of its own identity. § 6 runs: "The empirical consciousness which accompanies different representations is in itself fragmentary, and without relation to the identity of the subject." Kant is saying that if there existed merely a consciousness of A which was not at the same time a consciousness of B and a consciousness of B which was not at the same time a consciousness of A, these consciousnesses would not be the consciousnesses belonging to one self. But this is only true, if the one self to which the consciousness of A and the consciousness of B are to belong must be capable of being aware of its own identity. Otherwise it might be one self which apprehended A and then, forgetting A, apprehended B. No doubt in that case the self could not be aware of its own identity in apprehending A and in apprehending B, but none the less it would be identical in so doing. We reach the same conclusion if we consider the concluding sentence of § 10. "It is only because I can comprehend the manifold of representations in one consciousness, that I call them all my representations; for otherwise I should have as many-coloured and varied a self as I have representations of which I am conscious." Doubtless if I am to be aware of myself as the same in apprehending A and B, then, in coming to apprehend B, I must continue to apprehend A, and therefore must apprehend A and B as related; and such a consciousness on Kant's view involves a synthesis. But if I am merely to be the same subject which apprehends A and B, or rather if the apprehension of A and that of B are merely to be apprehensions on the part of one and the same subject, no such consciousness of A and B as related and, therefore, no synthesis is involved.

Again, the third paragraph assumes the possibility of self-consciousness as the starting-point for argument. The thought[77] seems to be this: 'For a self to be aware of its own identity, there must be a manifold in relation to which it can apprehend itself as one and the same throughout. An understanding which was perceptive, i. e. which originated objects by its own act of thinking, would necessarily by its own thinking originate a manifold in relation to which it could be aware of its own identity in thinking, and therefore its self-consciousness would need no synthesis. But our understanding, which is not perceptive, requires a manifold to be given to it, in relation to which it can be aware of its own identity by means of a synthesis of the manifold.' If this be the thought, it is clearly presupposed that any understanding must be capable of being conscious of its own identity.[78]

Further, it is easy to see how Kant came to take for granted the possibility of self-consciousness, in the sense of the consciousness of ourselves as the identical subject of all our representations. He approaches self-consciousness with the presupposition derived from his analysis of knowledge that our apprehension of a manifold does not consist in separate apprehensions of its elements, but is one apprehension or consciousness of the elements as related.[79] He thinks of this as a general presupposition of all apprehension of a manifold, and, of course, to discover this presupposition is to be self-conscious. To recognize the oneness of our apprehension is to be conscious of our own identity.[80]

Again, to pass to the second main point to be considered,[81] Kant has no justification for arguing from the possibility of self-consciousness to that of the synthesis. This can be seen from the mere form of his argument. Kant, as has been said, seems first to establish the possibility of self-consciousness, and thence to conclude that a synthesis must be possible. But if, as it is his point to urge, consciousness of our identity only takes place through consciousness of the synthesis, this method of argument must be invalid. It would clearly be necessary to know that the synthesis is possible, before and in order that we could know that self-consciousness is possible. An objector has only to urge that the manifold might be such that it could not be combined into a systematic whole, in order to secure the admission that in that case self-consciousness would not be possible.

Nevertheless, the passage under consideration may be said to lay bare an important presupposition of self-consciousness. It is true that self-consciousness would be impossible, if we merely apprehended the parts of the world in isolation. To be conscious that I who am perceiving C perceived B and A, I must be conscious at once of A, B, and C, in one act of consciousness or apprehension. To be conscious separately of A and B and C is not to be conscious of A and B and C. And, to be conscious of A and B and C in one act of consciousness, I must apprehend A, B, and C as related, i. e. as forming parts of a whole or system. Hence it is only because our consciousness of A, B, and C is never the consciousness of a mere A, a mere B, and a mere C, but is always the consciousness of A B C as elements in one world that we can be conscious of our identity in apprehending A, B, and C. If per impossibile our apprehension be supposed to cease to be an apprehension of a plurality of objects in relation, self-consciousness must be supposed to cease also. At the same time, it is impossible to argue from the consciousness of our identity in apprehending to the consciousness of what is apprehended as a unity, and thence to the existence of that unity. For, apart from the consideration that in fact all thinking presupposes the relatedness or—what is the same thing—the necessary relatedness of objects to one another, and that therefore any assertion to the contrary is meaningless, the consciousness of objects as a unity is a condition of the consciousness of our identity, and therefore any doubt that can be raised in regard to the former can be raised equally with regard to the latter.

We may now pass to the concluding portion of the deduction. For the purpose of considering it, we may sum up the results of the preceding discussion by saying that Kant establishes the synthesis of the manifold on certain principles by what are really two independent lines of thought. The manifold may be regarded either as something which, in order to enter into knowledge, must be given relation to an object, or as something with respect to which self-consciousness must be possible. Regarded in either way, the manifold, according to Kant, involves a process of synthesis on certain principles, which makes it a systematic unity. Now Kant introduces the categories by maintaining that they are the principles of synthesis in question. "I assert that the above mentioned categories are nothing but the conditions of thinking in a possible experience.... They are fundamental conceptions by which we think objects in general for phenomena."[82] A synthesis according to the categories is 'that wherein alone apperception can prove a priori its thorough-going and necessary identity'.[83] In the first edition this identification is simply asserted, but in the second Kant offers a proof.[84]

Before, however, we consider the proof, it is necessary to refer to a difficulty which seems to have escaped Kant altogether. The preceding account of the synthesis involved in knowledge and in self-consciousness implies, as his illustrations conclusively show, that the synthesis requires a particular principle which constitutes the individual manifold a whole of a particular kind.[85] But, if this be the case, it is clear that the categories, which are merely conceptions of an object in general, and are consequently quite general, cannot possibly be sufficient for the purpose. And since the manifold in itself includes no synthesis and therefore no principle of synthesis, Kant fails to give any account of the source of the particular principles of synthesis required for particular acts of knowledge.[86] This difficulty—which admits of no solution—is concealed from Kant in two ways. In the first place, when he describes what really must be stated as the process by which parts or qualities of an object become related to an object of a particular kind, he thinks that he is describing a process by which representations become related to an object in general.[87] Secondly, he thinks of the understanding as the source of general principles of synthesis, individual syntheses and the particular principles involved being attributed to the imagination; and so, when he comes to consider the part played in knowledge by the understanding, he is apt to ignore the need of particular principles.[88] Hence, Kant's proof that the categories are the principles of synthesis can at best be taken only as a proof that the categories, though not sufficient for the synthesis, are involved in it.

The proof runs thus:

"I could never satisfy myself with the definition which logicians give of a judgement in general. It is, according to them, the representation of a relation between two conceptions...."

"But if I examine more closely the relation of given representations[89] in every judgement, and distinguish it, as belonging to the understanding, from their relation according to the laws of the reproductive imagination (which has only subjective validity), I find that a judgement is nothing but the mode of bringing given representations under the objective unity of apperception. This is what is intended by the term of relation 'is' in judgements, which is meant to distinguish the objective unity of given representations from the subjective. For this term indicates the relation of these representations to the original apperception, and also their necessary unity, even though the judgement itself is empirical, and therefore contingent, e. g. 'Bodies are heavy.' By this I do not mean that these representations necessarily belong to each other in empirical perception, but that they belong to each other by means of the necessary unity of apperception in the synthesis of perceptions, that is, according to principles of the objective determination of all our representations, in so far as knowledge can arise from them, these principles being all derived from the principle of the transcendental unity of apperception. In this way alone can there arise from this relation a judgement, that is, a relation which is objectively valid, and is adequately distinguished from the relation of the very same representations which would be only subjectively valid, e. g. according to laws of association. According to these laws, I could only say, 'If I carry a body, I feel an impression of weight', but not 'It, the body, is heavy'; for this is tantamount to saying, 'These two representations are connected in the object, that is, without distinction as to the condition of the subject, and are not merely connected together in the perception, however often it may be repeated.'"[90]

This ground for the identification of the categories with the principles of synthesis involved in knowledge may be ignored, as on the face of it unsuccessful. For the argument is that since the activity by which the synthesis is affected is that of judgement, the conceptions shown by the Metaphysical Deduction to be involved in judgement must constitute the principles of synthesis. But it is essential to this argument that the present account of judgement and that which forms the basis of the Metaphysical Deduction should be the same; and this is plainly not the case.[91] Judgement is now represented as an act by which we relate the manifold of sense in certain necessary ways as parts of the physical world,[92] whereas in the Metaphysical Deduction it was treated as an act by which we relate conceptions; and Kant now actually says that this latter account is faulty. Hence even if the metaphysical deduction had successfully derived the categories from the account of judgement which it presupposed, the present argument would not justify the identification of the categories so deduced with the principles of synthesis. The fact is that Kant's vindication of the categories is in substance independent of the Metaphysical Deduction. Kant's real thought, as opposed to his formal presentation of it, is simply that when we come to consider what are the principles of synthesis involved in the reference of the manifold to an object, we find that they are the categories.[93] The success, then, of this step in Kant's vindication of the categories is independent of that of the metaphysical deduction, and depends solely upon the question whether the principles of synthesis involved in knowledge are in fact the categories.

The substance of Kant's vindication of the categories may therefore be epitomized thus: 'We may take either of two starting-points. On the one hand, we may start from the fact that our experience is no mere dream, but an intelligent experience in which we are aware of a world of individual objects. This fact is conceded even by those who, like Hume, deny that we are aware of any necessity of relation between these objects. We may then go on to ask how it comes about that, beginning as we do with a manifold of sense given in succession, we come to apprehend this world of individual objects. If we do so, we find that there is presupposed a synthesis on our part of the manifold upon principles constituted by the categories. To deny, therefore, that the manifold is so connected is implicitly to deny that we have an apprehension of objects at all. But the existence of this apprehension is plainly a fact which even Hume did not dispute. On the other hand, we may start with the equally obvious fact that we must be capable of apprehending our own identity throughout our apprehension of the manifold of sense, and look for the presupposition of this fact. If we do this, we again find that there is involved a combination of the manifold according to the categories.'

In conclusion, attention may be drawn to two points. In the first place, Kant completes his account by at once emphasizing and explaining the paradoxical character of his conclusion. "Accordingly, the order and conformity to law in the phenomena which we call nature we ourselves introduce, and we could never find it there, if we, or the nature of our mind, had not originally placed it there."[94] "However exaggerated or absurd then it may sound to say that the understanding itself is the source of the laws of nature and consequently of the formal unity of nature, such an assertion is nevertheless correct and in accordance with the object, i. e. with experience."[95] The explanation of the paradox is found in the fact that objects of nature are phenomena. "But if we reflect that this nature is in itself nothing else than a totality[96] of phenomena and consequently no thing in itself but merely a number of representations of the mind, we shall not be surprised that only in the radical faculty of all our knowledge, viz. transcendental apperception, do we see it in that unity through which alone it can be called object of all possible experience, i. e. nature."[97] "It is no more surprising that the laws of the phenomena in nature must agree with the understanding and with its a priori form, that is, its faculty of connecting the manifold in general, than that the phenomena themselves must agree with the a priori form of our sensuous perception. For laws exist in the phenomena as little as phenomena exist in themselves; on the contrary, laws exist only relatively to the subject in which the phenomena inhere, so far as it has understanding, just as phenomena exist only relatively to the subject, so far as it has senses. To things in themselves their conformity to law would necessarily also belong independently of an understanding which knows them. But phenomena are only representations of things which exist unknown in respect of what they may be in themselves. But, as mere representations, they stand under no law of connexion except that which the connecting faculty prescribes."[98]

In the second place, this last paragraph contains the real reason from the point of view of the deduction[99] of the categories for what may be called the negative side of his doctrine, viz. that the categories only apply to objects of experience and not to things in themselves. According to Kant, we can only say that certain principles of connexion apply to a reality into which we introduce the connexion. Things in themselves, if connected, are connected in themselves and apart from us. Hence there can be no guarantee that any principles of connexion which we might assert them to possess are those which they do possess.