Part 2 / Book 1 — CHAPTER II DEDUCTION OF THE PURE CONCEPTS OF THE UNDERSTANDING

First edition Subjective and Objective Deductions.—In dealing with the transcendental deduction, as given in the first edition, we can make use of the masterly and convincing analysis which Vaihinger[752] (building upon Adickes’ previous results, but developing an independent and quite original interpretation) has given of its inconsecutive and strangely bewildering argumentation. Vaihinger’s analysis is an excellent example of detective genius in the field of scholarship. From internal evidence, circumstantially supported by the Reflexionen and Lose Blätter, he is able to prove that the deduction is composed of manuscripts, externally pieced together, and representing no less than four distinct stages in the slow and gradual development of Kant’s views. Like geological deposits, they remain to record the processes by which the final result has come to be. Though they do not in their present setting represent the correct chronological order, that may be determined once the proper clues to their disentanglement have been duly discovered. That discovery is itself, however, no easy task; for the unexpected, while lending colour and incident to the commentator’s enterprise, baffles his natural expectations at every turn. The first stage is one in which Kant dispenses with the categories, and in which, when they are referred to, they are taken as applying to things in themselves. The last stage, worked out, as there is ground for believing, in the haste and excitement of the final revision, is not represented in the Prolegomena or in the second edition of the Critique, the author retracing his steps and resuming the standpoint of the stage which preceded it. The fortunate accident of Kant’s having jotted down upon the back of a dated paper the record of his passing thought (one of the few Lose Blätter that are thus datable) is the culminating incident in this philosophical drama. It felicitously serves as a keystone in the body of evidence supported by general reasoning.

Before becoming acquainted with Vaihinger’s analysis I had observed Kant’s ascription to empirical concepts of the functions elsewhere allotted to the categories, but had been hopelessly puzzled as to how such teaching could be fitted into his general system. Vaihinger’s view of it as a pre-Critical survival would seem to be the only possible satisfactory solution. For the view which I have taken of Kant’s doctrine of the transcendental object as also pre-Critical, and for its employment as a clue to the dating of passages, I am myself alone responsible.

The order of my exposition will be as follows:[753]

I. Enumeration, in chronological order, of the four stages which compose the deduction of the first edition, and citation of the passages which represent each separate stage.

II. Detailed analysis, again in chronological order, of each successive stage, with exposition of the views which it embodies.

III. Examination of the evidence yielded by the Reflexionen and Lose Blätter in support of the above analysis.

IV. Connected statement and discussion of the total argument of the deduction.

I. Enumeration of the Four Stages

(1) First Stage: That of the Transcendental Object, without Co-operation of the Categories.—This stage is represented by[754]: (a) II. 3 (from beginning of the third paragraph to end of 3) = A 104-10; (b) I. § 13 (the entire section) = A 84-92 (retained in second edition as B 116-24). a discusses the problem of the reference of sensations to an object, b that of the objective validity of the categories. b is therefore transitional to the second stage.

(2) Second Stage: That of the Categories, without Co-operation of the Productive Imagination.—This stage is represented by: (a) I. [§ 14] (with the exception of its concluding paragraph) = A 92-4 (retained in second edition as B 124-7); (b) II. (the first four paragraphs) = A 95-7; (c) II. 4 (the entire section) = A 110-14.

(3) Third Stage: That of the Productive Imagination, without Mention of the Threefold Transcendental Synthesis.—This stage is represented by (a) III.β (from beginning of seventh paragraph to end of twelfth) = A 119-23; (b) III. α (from beginning of third paragraph to end of sixth) = A 116-19; (c) I. § 14 (Concluding paragraph) = A 94-5; (d) III.δ (from beginning of sixteenth paragraph to end of section preceding summary) = A 126-8; (e) S(ummary) (in conclusion to III.) = A 128-30; (f) III.γ (from beginning of thirteenth paragraph to end of fifteenth) = A 123-6; (g) I(ntroduction) (from beginning of section to end of second paragraph) = A 115-16; (h) § 10 T(ransitional to the fourth stage) = A 76-9 (retained as B 102-4).

(4) Fourth Stage: That of the Threefold Transcendental Synthesis.—This stage is represented by: (a) II. 1-3 (from opening of 1 to end of second paragraph in 3) = A 98-104; (b) II. (the two paragraphs immediately preceding a) = A 97-8.

II. Detailed Analysis of the Four Stages

First Stage.—A 104-10; A 84-92 (B 116-24).

A 104-10; II. § 3.—This is the one passage in the Critique in which Kant explicitly defines his doctrine of the “transcendental object”; and careful examination of the text shows that by it he means the thing in itself, conceived as being the object of our representations. Such teaching is, of course, thoroughly un-Critical; and as I shall try to show, this was very early realised by Kant himself. The passages in which the phrase “transcendental object” occurs are, like the section before us, in every instance of early origin. It is significant that the transcendental object is not again referred to in the deduction of the first edition.[755] Though it reappears in the chapter on phenomena and noumena, it does so in a passage which Kant excised in the second edition. The paragraphs which he then substituted make no mention of it. The doctrine is of frequent occurrence in the Dialectic, and combines with other independent evidence to show that the larger part of the Dialectic is of early origin. That the doctrine of the transcendental object is thus a pre-Critical or semi-Critical survival has, so far as I am aware, not hitherto been observed by any writer upon Kant. It has invariably been interpreted in the light of the sections in which it does not occur, and, as thus toned down and tempered to something altogether different from what it really stands for, has been taken as an essential and characteristic tenet of the Critical philosophy. It was in the course of an attempt to interpret Kant’s entire argument in the light of his doctrine of the transcendental object that I first came to detect its absence from all his later utterances. But it is important to recognise that the difficulties which would result from its retention are quite insuperable, and would by themselves, even in the absence of all external evidence of Kant’s rejection of it, compel us to regard it as a survival of pre-Critical thinking. As Vaihinger does not seem to have detected the un-Critical character of this doctrine, it is the more significant that he should, on other grounds, have felt constrained to regard the passage in which it is expounded as embodying the earliest stage in the development of the deduction. He would seem to continue in the orthodox view so far as to hold that though the doctrine of the transcendental object is here stated in pre-Critical terms, it was permanently retained by Kant in altered form.

The doctrine of the transcendental object, as here expounded, is as follows:

“Appearances are themselves nothing but sensuous representations which must not be taken as capable of existing in themselves (an sich) with exactly the same character (in ebenderselben Art) outside our power of representation.”[756]

These sense-representations are our only possible representations, and when we speak of an object corresponding to them, we must be conceiving an object in general, equal to x.

“They have their object, but an object which can never be intuited by us, and which may therefore be named the non-empirical, i.e. transcendental object = x.”[757]

This object is conceived as being that which prevents our representations from occurring at haphazard, necessitating their order in such manner that, manifold and varied as they may be, they can yet be self-consistent in their several groupings, and so possess that unity which is essential to the concept of an object.

“The pure concept of this transcendental object, which in fact throughout all our knowledge is always one and the same, is that which can alone confer upon all our empirical concepts relation in general to an object, i.e. objective reality.”[758]

What renders this doctrine impossible of permanent retention was that it allowed of no objective existence mediate between the merely subjective and the thing in itself. On such teaching there is no room for the empirical object; and immediately upon the recognition of that latter phenomenal form of existence in space, Kant was constrained to recognise that it is in the empirical object, not in the thing in itself, that the contents of our representations are grounded and unified. Any other view must involve the application of the categories, especially those of substance and causality, to the thing in itself. The entire empirical world has still to be conceived as grounded in the non-empirical, but that is a very different contention from the thesis that the thing in itself is the object and the sole object of our representations. The doctrine of the transcendental object has thus a twofold defect: it advocates an extreme subjectivism, and yet at the same time applies the categories to the thing in itself.

But the latter consequence is one which could not, at the stage represented by this section, be appreciated by Kant. For, as we shall find, he is endeavouring to solve the problem of the reference of sense-representation to an object without assumption of a priori categories. It is in empirical concepts, conditioned only by a transcendental apperception, that he professes to discover the grounds and conditions of this objective reference. Let us follow Kant’s argument in detail. The section opens[759] with what may be a reference to the Aesthetic, and proceeds to deal with the first of the two problems cited in the 1772 letter to Herz[760]—how sense-representations stand related to their object. The exact terms in which this question was there formulated should be noted.

“I propounded to myself this question: on what ground rests the relation of that in us which we name representation (Vorstellung) to the object. If the representation contains only the mode in which the subject is affected by the object, it is easily understood how it should accord (gemäss sei) with that object as an effect with its cause, and how [therefore] this determination of our mind should be able to represent something, i.e. have an object. The passive or sensuous representations have thus a comprehensible (begreifliche) relation to objects, and the principles, which are borrowed from the nature of our soul, have a comprehensible validity for all things in so far as they are to be objects of the senses.”[761]

Thus in 1772 there was here no real problem for Kant. The assumed fact, that our representations are generated in us by the action of independent existences, is taken as sufficient explanation of their being referred to objects.

The section of the Critique under consideration shows that Kant had come to realise the inadequacy of this explanation quite early, indeed prior to his solution of the second and further question which in that same letter is spoken of as “the key to the whole secret” of metaphysics. On what grounds, he now asks, is a subjective idea, even though it be a sense impression, capable of yielding consciousness of an object? In the letter to Herz the use of the term representation (Vorstellung) undoubtedly helped to conceal this problem. It is now emphasised that appearances are nothing but sense representations, and must never be regarded as objects capable of existing in themselves, with exactly the same character, outside our power of representation. Now also Kant employs, in place of the phrase “in accord with,” the much more definite term “corresponding to.” He points out that when we speak of an object corresponding to our knowledge, we imply that it is distinct from that knowledge. Consciousness of such an object must therefore be acquired from some other source than the given impressions. In other words, Kant is now prepared to withdraw his statement that “the passive or sensuous representations have an [easily] comprehensible relation to objects.” In and by themselves they are purely subjective, and can involve no such concept. The latter is a thought (Gedanke), a concept (Begriff), additional to, and distinct from, the given impressions. Its possibility, as regards both origin[762] and validity, must be “deduced.”

There then results this first and very peculiar form of the transcendental deduction. That part of it which persists in the successive stages rests upon an explicitly developed distinction between empirical and transcendental apperception. Kant teaches, in agreement with Hume, though, as we may believe, independently of his direct influence, that there is no single empirical state of the self which is constant throughout experience.[763]

“The consciousness of the self, according to the determinations of our state in inner perception, is merely empirical, and always in process of change.... That which has to be represented as of necessity numerically identical cannot be thought as such through empirical data. There must be a condition which precedes all experience, and renders experience itself possible, if a transcendental pre-supposition of this kind is to be rendered valid.... This pure, original, unchangeable consciousness I shall name transcendental apperception.”[764]

Kant would seem to have first developed this view in a quite crude form. The consciousness of the self, he seems to have held, consists in its awareness of its own unceasing activities. As consciousness of activity, it is entirely distinct in nature and in origin from all apprehension of sense impressions.[765] This teaching is a natural extension of the doctrine of the Dissertation,[766] that such pure notions as those of possibility, existence, necessity, substance, cause, are “acquired by attending to the actions of the mind on the occasion of experience.” Kant would very naturally hold that consciousness of the identity and unity of the self is obtained in a similar manner. Such, indeed, is the teaching of the section before us.

“No knowledge can take place in us ... without that unity of consciousness which precedes all data of intuitions, and in relation to which all representation of objects is alone possible.”[767] “It is precisely this transcendental apperception that constructs out of (macht aus) all possible appearances, which are capable of coexisting in one experience, a connection of all these representations according to laws. For this unity of consciousness would be impossible if the mind could not become conscious, in the knowledge of the manifold, of the identity of the function whereby it combines it synthetically in one knowledge. Thus the mind’s original and necessary consciousness of the identity of itself is at the same time a consciousness of an equally necessary unity of the synthesis of all appearances according to concepts, i.e. according to rules.... For the mind could not possibly think the identity of itself in the manifold of its representations, and indeed a priori, if it did not have before its eyes the identity of its action....”[768]

That is to say, the self is the sole source of all unity. As a pure and original unity it precedes experience; to its synthetic activities all conceptual unity is due; and by reflection upon the constancy of these activities it comes to consciousness of its own identity.

“...even the purest objective unity, namely that of the a priori concepts (space and time), is possible only through relation of the intuitions to [transcendental apperception]. The numerical unity of this apperception is therefore the a priori condition of all concepts, just as the manifoldness of space and of time is of the intuitions of sensibility.”[769]

To this consciousness of the abiding unity of the self Kant also traces the notion of the transcendental object. The latter, he would seem to argue, is formed by analogy from the former.

“This object is nothing else than the subjective representation (of the subject) itself, but made general, for I am the original of all objects.”[770] “The mind, through its original and underived thinking, is itself the pattern (Urbild) of such a synthesis.”[771] “I would not represent anything as outside me, and so make [subjective] appearances into objective experience if the representations were not related to something which is parallel to my ego, and so in that way referred by me to another subject.”[772]

These quotations from the Lose Blätter would seem to contain the key to Kant’s extremely enigmatic statement in A 105, that “the unity which the object makes necessary can be nothing else than the formal unity of consciousness in its synthesis of the manifold of its representations,” and again in A 109, that “this relation [of representations to an object] is nothing else than the necessary unity of consciousness.”[773]

But this does not complete the sum-total of the functions which Kant is at this stage prepared to assign to apperception. It mediates our consciousness of the transcendental object in still another manner, namely, by rendering possible the formation of the empirical concepts which unify and direct its synthetic activities. This is, indeed, the feature in which this form of the deduction diverges most radically from all later positions. Space and time are, it would seem, regarded as being the sole a priori concepts.[774] The instruments through which the unity of apperception acts, and through which the thought of an object becomes possible, are empirical concepts. Such general concepts as “body” or “triangle” serve as rules constraining the synthetic processes of apprehension and reproduction to take place in such unitary fashion as is required for unitary consciousness. The notion of objectivity is specified in terms of the necessities which these empirical concepts thus impose.

“We think a triangle as object in so far as we are conscious of the combination of three straight lines according to a rule by which such an intuition can at all times be generated. This unity of rule determines the whole manifold and limits it to conditions which make the unity of apperception possible; and the concept of this unity [of rule] is the representation of the object.... All knowledge demands a concept, ... and a concept is always, as regards its form, something general, something that serves as a rule. Thus the concept of body serves as a rule to our knowledge of outer appearances, in accordance with the unity of the manifold which is thought through it.... The concept of body necessitates ... the representation of extension, and therewith of impenetrability, shape, etc.”[775]

Such is the manner in which Kant accounts for our concept of the transcendental object. It consists of two main elements: first, the notion of an unknown x, to which representations may be referred; and secondly, the consciousness of this x as exercising compulsion upon the order of our thinking. The former notion is framed on the pattern of the transcendental subject; it is conceived as another but unknown subject. The consciousness of it as a source of external necessity is mediated by the empirical concepts which transcendental apperception also makes possible. And from this explanation of the origin of the concept of the transcendental object Kant derives the proof of its validity.[776] It is indispensable for the realisation by the unitary self of a unitary consciousness.

“This relation [of representations to an object] is nothing else than the necessary unity of consciousness, and therefore also of the synthesis of the manifold, by a common (gemeinschaftlich) functioning of the mind, which unites it in one representation.”[777]

Through instruments empirical in origin, and subjectively necessary, the notion of an objective necessity is rendered possible to the mind.

It is not surprising that Kant did not permanently hold to this view of the empirical concept. The objections are obvious. Such a view of the function of general concepts renders unintelligible their own first formation. For as they are empirical, they can only be acquired by conscious processes that do not involve them. That is to say, consciousness of objects follows upon a prior consciousness in and through which concepts, such as that of body, are discovered and formed. Yet, as the argument claims, general concepts are the indispensable conditions of unitary consciousness. How through a consciousness that is not yet unified can general concepts be formed? Also it is difficult to see how empirical concepts can be viewed as directly conditioned by, and as immediately due to, anything so general as pure apperception. These objections Kant must have come very quickly to recognise. This was the first part of his teaching to be modified. In the immediately succeeding stage,[778] so far as the stages can be reconstructed from the survivals in the Critique, the empirical concepts are displaced once and for all by the a priori categories.

The only sentences which can be regarded as possibly conflicting with the above interpretation are those two (in the second last and in the last paragraphs) in which the phrase “rules a priori” occurs. Even granting (what is at least questionable as regards the first) that the words are meant to be taken together, it does not follow that Kant is here speaking of categories. For contrary to his usual teaching he speaks of the concept of body as a source of necessity. If so, it may well, with equal looseness, be spoken of as a priori. That is indeed done, by implication, in the second and third paragraphs, where he speaks of a rule (referring to “body and triangle”) as making the synthesis of reproduction “a priori necessary.” Such assertions are completely inconsistent with Kant’s Critical teaching, but so is the entire section.

The setting in which the passage before us occurs has its own special interest.[779] When Kant, as it would seem, on the very eve of the publication of the Critique, developed the doctrine of a threefold synthesis culminating in a “synthesis of recognition in the concept,” he must have bethought himself of this earlier position, and have completed his subjective deduction by incorporation, probably with occasional alterations of phrasing, of the older manuscript. This procedure has bewildered even the most discerning among Kant’s readers; but now, thanks to Vaihinger’s convincing analysis, it may be welcomed as of illuminating interest in the historical study of Kant’s development.

I may here draw attention to the two important respects in which the positions revealed in this section continued to influence Kant’s later teaching: namely, in the emphasis laid upon the transcendental unity of apperception, and in the view of objectivity as involving the thought of the thing in itself.

The excessive emphasis which in this first stage is laid upon the transcendental unity of apperception persists throughout the later forms of the deduction, and, as I shall try to show, does so to the detriment of the argument. Though its functions are considerably diminished, they are still exaggerated; this is perhaps in part due to its having been in this early stage regarded as in and by itself the sole ultimate ground of unitary experience. There were, however, two other influences at work. Kant continued to employ the terminology of his earlier view, and in his less watchful moments was betrayed thereby into conflict with his considered teaching. But even more important was the influence of his personal convictions. He was irrevocably committed in his own private thinking to a belief in the spiritual and abiding character of the self; and this belief frequently colours, in illegitimate ways, the expression of his views. This is especially evident in some of the alterations[780] of the second edition, written as they were at a time when he was chiefly preoccupied with moral problems.

As regards the other factor, the view adopted in regard to the nature of objectivity, there is ample evidence that even after the empirical concepts had been displaced by the categories Kant still continued for some time (possibly for several years in the earlier and middle ’seventies) to hold to his doctrine of the transcendental object. Passages which expound it in this later form occur in the Note on Amphiboly and throughout the Dialectic.[781] That this may not be taken for his final teaching is equally certain. The entire first layer of the deduction of the first edition, all the relevant passages in the chapter on phenomena and noumena, and some of those in the Dialectic, were omitted in the second edition; and nowhere, either in the other portions of the deduction of the first edition, or in the deduction of the second edition, or in any passages added elsewhere in the second edition, is such teaching to be found.

A brief statement of Kant’s doctrine of the transcendental object in its later form seems advisable at this point; it is required in order to complete and to confirm the interpretation which I have given of the earlier exposition. At the same time I shall endeavour to show that the sections in which the doctrine occurs, though later than the first layer of the deduction of the first edition, are all of comparatively early origin, and that they reveal not the least trace of Kant’s more mature, phenomenalist view of the empirical world in space.

We may begin with the passages in the chapter on phenomena and noumena. The meaning in which the term transcendental is employed is there made sufficiently clear.

“The transcendental employment of a concept in any principle consists in its being referred to things in general and in themselves.”[782]

That is to say, the term transcendental, as used in the phrase transcendental object, is not employed in any sense which would oppose it to the transcendent. In so far as the thought of the thing in itself is a necessary ingredient in the concept of objectivity, it is a condition of apperception, and therefore of possible experience; in other words, the thought of a transcendent object is one of the transcendental conditions of our experience. As Kant is constantly interchanging the terms transcendent and transcendental, such an explanation of the phrase is perhaps superfluous; but if any is called for, the above would seem to suffice. As we shall have occasion to observe,[783] other factors besides the a priori must be reckoned among the conditions of experience; and to both types of conditions Kant applies the epithet transcendental.

In the chapter on phenomena and noumena Kant enquires at considerable length whether the categories (meaning, of course, the pure forms of understanding, not their schematised correlates) allow of transcendental (i.e. transcendent) employment. The passages in which this discussion occurs[784] would seem, however, to be highly composite; many paragraphs, or portions of paragraphs, are of much later date than others. We may therefore limit our attention to those in which the phrase transcendental object is actually employed, i.e. to those which appear only in the first edition.

“All our representations are referred by the understanding to some object; and since appearances are merely representations, the understanding refers them to a something as the object of sensuous intuition. But this something, thus conceived (in so fern), is only the transcendental object; and by that is meant a something = x, of which we know, and with the present constitution of our understanding can know, nothing whatsoever, but which, as a correlate of the unity of apperception, can serve only for the unity of the manifold in sensuous intuition. By means of this unity the understanding combines the manifold into the concept of an object. This transcendental object cannot be separated from the sense data, for nothing then remains over through which it might be thought. Consequently it is not in itself an object of knowledge, but only the representation of appearances under the concept of an object in general which is determinable through the manifold of those appearances. Precisely for this reason also the categories do not represent a special object given to the understanding alone, but only serve to specify the transcendental object (the concept of something in general) through that which is given in sensibility, in order thereby to know appearances empirically under concepts of objects.”[785] “The object to which I relate appearance in general is the transcendental object, i.e. the completely indeterminate thought of something in general. This cannot be entitled the noumenon [i.e. the thing in itself more specifically determined as being the object of a purely intelligible intuition];[786] for I know nothing of what it is in itself, and have no concept of it save as the object of a sensuous intuition in general, and so as being one and the same for all appearances.”[787]

Otherwise stated, Kant’s teaching is as follows. The thought of the thing in itself remains altogether indeterminate; it does not specify its object, and therefore yields no knowledge of it; none the less it is a necessary ingredient in the concept of objectivity as such. The object as specified in terms of sense is mere representation; the object as genuinely objective can only be thought. The correlate of the unity of apperception is the thought of the thing in itself. This is what Kant is really asserting, though in a hesitating manner which would seem to indicate that he is himself already more or less conscious of its unsatisfactory and un-Critical character.

The phrase transcendental object occurs once in the second Analogy[788] and twice in the Note on Amphiboly.[789] The passage in the second Analogy may very well, in view of the kind of subjectivism which it expounds, be of early date of writing. By transcendental object Kant there quite obviously means the thing in itself. From the first reference in the Note on Amphiboly no definite conclusions can be drawn. The argument is too closely bound up with his criticism of Leibniz to allow of his own independent standpoint being properly developed. There is, however, nothing in it which compels us to regard it as of late origin; and quite evidently Kant here means by the transcendental object the thing in itself. The phrase substantia phaenomenon is not, as might at first sight seem, equivalent to the empirical object of Kant’s phenomenalist teaching. It is an adaptation of Leibnizian phraseology.[790] The second reference in the Note on Amphiboly occurs in a passage which may perhaps be of later origin;[791] but the transcendental object is there mentioned only in order to afford opportunity for the statements that it cannot be thought through any of the categories, that we are completely ignorant whether it is within or without us, and whether if sensibility were removed it would vanish or remain, and that it can therefore serve only as a limiting concept. We here observe it in the very process of being eliminated. As we shall find, Kant’s teaching is ill-expressed in the sections on Amphiboly; so much so that they could not be recast without seriously disturbing the balance of his architectonic. They were therefore allowed to remain unaltered in the second edition.

We may now pass to the Dialectic. The subjectivist doctrine of the transcendental object is there expressed in a much more uncompromising manner. Let us first consider the references to the transcendental object in the Paralogisms and in the subsequent Reflection. The phrase transcendental object occurs twice in the second Paralogism, once in the third, twice in the fourth, and three times in the Reflection;[792] and in all these cases there is not the least uncertainty as to its denotation. It is taken as equivalent to the thing in itself, and is expounded as a necessary ingredient in the consciousness of our subjective representations as noumenally grounded.

“What matter may be as a thing in itself (transcendental object) is completely unknown to us, though, owing to its being represented as something external, its permanence as appearance can indeed be observed.”[793] “We can indeed admit that something, which may be (in the transcendental[794] sense) ‘outside us,’ is the cause of our outer intuitions, but this is not the object of which we are thinking in the representations of matter and of corporeal things, for these are merely appearances, i.e. mere kinds of representation which are never to be met with save in us, and whose actuality depends on immediate consciousness just as does the consciousness of my own thoughts. The transcendental object is equally unknown in respect to inner and to outer intuition.”[795]

Here Kant at one and the same time distinguishes between, and confounds together, representation and its empirical object. What is alone clear is that by the transcendental object he means simply the thing in itself viewed as the cause of our sensations. In A 358 it is used in a wider sense as also comprehending the noumenal conditions which underlie the conscious subject.

“...this something which underlies the outer appearances and which so affects our sense that it obtains the representations of space, matter, shape, etc., this something viewed as noumenon (or better as transcendental object) might also at the same time be the subject that does our thinking....”

Similarly in A 379-80:

“Though the I, as represented through inner sense in time, and objects in space outside me, are specifically quite distinct appearances, they are not for that reason thought as being different things. Neither the transcendental object which underlies outer appearances, nor that which underlies inner intuition, is in itself either matter or a thinking being, but is a ground (to us unknown) of the appearances which supply to us the empirical concepts of the former as well as of the latter kind.”

The references in the Reflection on the Paralogisms are of the same general character and are equally definite.[796] A 390-1 has special interest in that it explicitly states that to appearances, taken as Kant invariably takes them throughout the Paralogisms in the first edition as mere subjective representations, the category of causality, and therefore by implication the category of substance, is inapplicable.

“No one could dream of asserting that that which he has once come to recognise as mere representation is an outer cause.”

We may now turn to the passages in the chapter on the Antinomies.

“The non-sensuous cause of our representations is completely unknown to us, and therefore we cannot intuit it as object.... We may, however, entitle the purely intelligible cause of appearances in general the transcendental object.... To this transcendental object we can ascribe the whole extent and connection of our possible perceptions....”[797]

Appearances can be regarded as real only to the extent to which they are actually experienced. Otherwise they exist only in some unknown noumenal form of which we can acquire no definite concept, and which is therefore really nothing to us. This, Kant declares, is true even of that immemorial past of which we are ourselves the product.

“...all the events which have taken place in the immense periods that have preceded my own existence mean really nothing but the possibility of extending the chain of experience from the present perception back to the conditions which determine it in time.”[798]

In other words, we may not claim that such events, empirically conceived, have ever actually existed in any such empirical form. A similar interpretation is given to the assertion of the present reality of what has never been actually experienced.

“Moreover, in outcome it is a matter of indifference whether I say that in the empirical progress in space I can meet with stars a hundred times farther removed than the outermost now perceptible to me, or whether I say that they are perhaps to be met with in cosmical space even though no human being has ever perceived or ever will perceive them. For though they might be given as things in themselves, without relation to possible experience, they are still nothing for me, and therefore are not objects, save in so far as they are contained in the series of the empirical regress.”[799] “The cause of the empirical conditions of this process, that which determines what members I shall meet with and how far by means of such members I can carry out the regress, is transcendental and is therefore necessarily unknown to me.”[800]

Such is the form in which Kant’s pre-Critical doctrine of the transcendental object survives in the Critique.[801] It contains no trace of the teaching of the objective deduction of the first and second edition or of the teaching of the refutation of idealism in the second edition. It closely resembles Mill’s doctrine of the permanent possibilities of sensation, and is almost equally subjectivist in character. As already noted,[802] it also lies open to the further objection that it involves an illegitimate application of the categories to things in themselves. As Kant started from the naïve and natural assumption that reference of representations to objects must be their reference to things in themselves, he also took over the current Cartesian view that it is by an inference in terms of the category of causality that we advance from a representation to its cause. The thing in itself is regarded as the sole true substance and as the real cause of everything which happens in the natural world. Appearances, being representations merely, are wholly transitory and completely inefficacious. Not only, therefore, are the categories regarded as valid of things in themselves, they are also declared to have no possible application to phenomena. Sense appearances do not, on this view, constitute the mechanical world of the natural sciences; they have a purely subjective, more or less epi-phenomenal, existence in the mind of each separate observer. It was very gradually, in the process of developing his own Critical teaching, that Kant came to realise the very different position to which he was thereby committed. The categories, including that of causality, are pre-empted for the empirical object which is now regarded as immediately apprehended; and the function of mediating the reference of phenomena to things in themselves now falls to the Ideas of Reason. The distinction between appearance and reality is no longer that between representations and their noumenal causes, but between the limited and relative character of the entire world in space and time and the unconditioned demanded by Reason. But these are questions whose discussion must meantime be deferred.[803]

I may now briefly summarise the evidence in favour of the view that the doctrine of the transcendental object is a pre-Critical or semi-Critical survival and must not be taken as forming part of Kant’s final and considered position. (I) Of the six sections in which the phrase transcendental object occurs, three[804] were omitted in the second edition, and in the passages which were substituted for them it receives no mention. There are various reasons which can be suggested in explanation of the retention of the other three[805] in the second edition. The Note on Amphiboly was too unsatisfactory as a whole to encourage Kant to improve upon it in detail. The other two are outside the limit at which Kant thought good to terminate all attempts to improve, whether in major or in minor matters, the text of the first edition.[806] To have recast the Antinomies as he had recast the Paralogisms would have involved alterations much too extensive. Also, there were no outside polemical influences—or at least none acting quite directly—such as undoubtedly reinforced his other reasons for revising the Paralogisms. (2) Secondly, the transcendental object is not mentioned in the later layers of the deduction of the first edition, nor in the deduction of the second edition, nor in any passage or note added in the second edition. That Kant should thus suddenly cease to employ a phrase to which he had accustomed himself is the more significant in view of his conservative preference for the adapting of familiar terminology to new uses. It can only be explained as due to his recognition of the completely untenable character of the teaching to which it had given expression. As the object of knowledge is always empirical, it can never legitimately be called transcendental. (3) Thirdly, the general teaching of the passages in which the phrase transcendental object occurs is by itself sufficient proof of their early origin. They reveal not the least trace of the deepened insight of his final standpoints. As we know, it was certain difficulties involved in the working out of the objective deduction that delayed the publication of the Critique for so many years; and the sections which deal with these difficulties contain Kant’s maturest teaching. In them he seems to withdraw definitely from the positions to which he had unwarily committed himself by his un-Critical doctrine of the transcendental object. I now pass to the second section constitutive of the first stage.

A 84-92=B 116-24, I. § 13.—Just as in II. § 3 Kant deals solely with the first of the two questions formulated in the letter of 1772 to Herz—the reference of sense-representations to an object,—so in I. § 13 he raises only the second—that of the objective validity of intellectual representations (now spoken of as pure concepts of understanding, or pure a priori concepts, and only in one sentence as categories). And just as in the former section he carries the problem a step further, yet without attaining to the true Critical position, so in this latter he still assumes that it is the application of these pure concepts to real independent objects, i.e. to things in themselves, which calls for justification. We must again consider the exact terms in which this problem is formulated in the letter to Herz.[807]

“Similarly, if that in us which is called a representation, were active in relation to the object, that is to say, if the object itself were produced by the representation (as on the view that the ideas in the Divine Mind are the archetypes of things), the conformity of representations with objects might be understood. We can thus render comprehensible at least the possibility of two kinds of intelligence—of an intellectus archetypus, on whose intuition the things themselves are grounded, and of an intellectus ectypus which derives the data of its logical procedure from the sensuous intuition of things. But our understanding (leaving moral ends out of account) is not the cause of the object through its representations, nor is the object the cause of its intellectual representations (in sensu reali). Hence, the pure concepts of the understanding cannot be abstracted from the data of the senses, nor do they express our capacity for receiving representations through the senses. But, whilst they have their sources in the nature of the soul, they originate there neither as the result of the action of the object upon it, nor as themselves producing the object. In the Dissertation I was content to explain the nature of these intellectual representations in a merely negative manner, viz. as not being modifications of the soul produced by the object. But I silently passed over the further question, how such representations, which refer to an object and yet are not the result of an affection due to that object, can be possible. I had maintained that the sense representations represent things as they appear, the intellectual representations things as they are. But how then are these things given to us, if not by the manner in which they affect us? And if such intellectual representations are due to our own inner activity, whence comes the agreement which they are supposed to have with objects, which yet are not their products? How comes it that the axioms of pure reason about these objects agree with the latter, when this agreement has not been in any way assisted by experience? In mathematics such procedure is legitimate, because its objects only are quantities for us, and can only be represented as quantities, in so far as we can generate their representation by repeating a unit a number of times. Hence the concepts of quantity can be self-producing, and their principles can therefore be determined a priori. But when we ask how the understanding can form to itself completely a priori concepts of things in their qualitative determination, with which these things must of necessity agree, or formulate in regard to their possibility principles which are independent of experience, but with which experience must exactly conform,—we raise a question, that of the origin of the agreement of our faculty of understanding with the things in themselves, over which obscurity still hangs.”[808]

The section before us represents the same general standpoint as that given in the above letter. Here, too, it is the validity of the a priori concepts in reference to things in themselves that is under consideration. The implication of Kant’s argument is that the categories, being neither determinable nor discoverable by means of experience, will only apply to appearances if they determine, or rather reveal, the actual non-experienced nature of things in themselves. These pure concepts, it is implied, owing to their combined a priori and intellectual characteristics, make this inherent claim. Either they are altogether empty and illusory, or such unlimited validity must be granted to them. Kant, that is to say, still holds, as in the Dissertation, that sense-representations reveal things as they appear, intellectual representations things as they are.

“We have either to surrender completely all claims to judgments of pure reason, in the most esteemed of all fields, that which extends beyond the limits of all possible experience, or we must bring this Critical investigation to perfection.”[809]

The pure concepts, unlike space, “apply to objects generally, apart from the conditions of sensibility.”[810] But here also, as in the letter to Herz, the strange and problematic character of such knowledge is clearly recognised.

Kant’s discussion of the concept of causality in A 90 may seem to conflict with the above contention—that it is its applicability to things in themselves which Kant is considering. But this difficulty vanishes if we bear in mind that here, as in the Dissertation, there is no such distinction as we find in Kant’s later more genuinely phenomenalist position, between the objects causing our sensations and things in themselves.[3] The purely intelligible object, supposed to remain after elimination of the empirical and a priori sensory factors, is the thing in itself. The objects apprehended through sense are real, only not in their sensuous form.

There are two connected facts which together may perhaps be taken as evidence that I. § 13 is later than II. 3 b. Intellectual concepts are reinstated alongside the a priori concepts of space and time. Kant has evidently in the meantime given up the attempt to construe the former as empirical in origin. That that attempt was earlier in time would seem to be proved by the further fact, that the a priori concepts are here viewed as performing the same kind of function as that ascribed in II. 3 b to concepts that are empirical. They are conditions of the “synthetic unity of thought.”[811] This view of the function of concepts is certainly fundamental and important, and Kant permanently retained it from his previous abortive method of ‘deduction.’ But it was a long step from the discovery of the distinction between empirical and a priori concepts to its fruitful application. That involved appreciation of the further fact that the two problems, separately stated in the letter to Herz and separately dealt with in II. 3 b and in I. § 13—the problem of the relation of sense-representations, and the problem of the relation of intellectual representations, to an object,—are indeed one and the same, soluble from one and the same standpoint, by one and the same method of deduction, namely, by reference to the possibility of experience. Only in and through relation to an object can sense-representations be apprehended; and only as conditions of such sense-experience are the categories objectively valid. Relation to an object is constituted by the categories, and is necessary in reference to sense-representations, because only thereby is consciousness of any kind possible at all.

That this truly Critical position had not been attained when I. § 13 was written,[812] is shown not only by its concentration on the single problem of the validity of a priori concepts, but also by its repeated assertion that representations can be consciously apprehended independently of all relation to the faculty of understanding. The directly counter assertion appears, however, in the sections (I. § 14, II.: first four paragraphs) which immediately follow in the text of the Critique—indicating that in the period represented by these latter the revolutionary discovery, the truly Copernican hypothesis, had at last been achieved. They constitute the second stage, and to it we may now proceed.

Second Stage.—A 92-4 = B 124-7; A 95-7; A 110-14.

A 92-4, I. § 14 (with the exception of the concluding classification of mental powers).—This section makes a fresh start; it stands in no necessary relation to any preceding section. The problem is still formulated, in its opening sentences, in terms reminiscent of the letter to Herz; but otherwise the standpoint is entirely new, and save for the wording of a single sentence (A 93: “if not intuited, yet”), is genuinely Critical. The phrase “possibility of experience” now appears, and is at once assigned the central rôle. The words “if not intuited, yet” in A 93 may possibly have been inserted later in order to tone down the flagrant contradiction with the preceding paragraphs. In any case, even this qualification is explicitly retracted in A 94.

A 95-7.—The same standpoint appears in the first three paragraphs of Section II. The categories are “the a priori conditions on which the possibility of experience depends.”[813] By the categories alone “can an object be thought.”[814] The further important point that only in their empirical employment do the categories have use and meaning is excellently developed.

“An a priori concept not referring to experience would be the logical form only of a concept, but not the concept itself by which something is thought.”[815]

A 110-14, II. 4.—In this section also the argument starts afresh, indicating (if such evidence were required) that, like I. § 14, it must have been written independently of its present context. But the argument is now advanced one step further. The categories are recognised as simultaneously conditioning both unity of consciousness and objectivity.

“There is but one experience ... as there is but one space and one time....” “The a priori conditions of a possible experience are at the same time conditions of the possibility of objects of experience”[816] “...the necessity of these categories rests on the relation which our whole sensibility, and with it also all possible appearances, have to the original unity of apperception....”[817]

Now also it is emphasised that save in and through a priori concepts no representations can exist for consciousness.

“They would then belong to no experience, would be without an object, a blind play of representations, less even than a dream.”[818] They “would be to us the same as nothing.”[819]

The wording is still not altogether unambiguous, but the main point is made sufficiently clear.

These paragraphs are the earliest in which traces of a genuine phenomenalism can be detected. The transcendental object, one and the same for all our knowledge, is not referred to. ‘Objects’ (in the plural) is the term which is used wherever the context permits. The empirical object is thus made to intervene between the thing in itself and the subjective representations. But the distinction between empirical objects and subjective representations on the one hand, and between empirical objects and things in themselves on the other, is not yet drawn in any really clear and definite manner.

A similar phenomenalist tendency crops out in Kant’s distinction[820] between objective affinity and subjective association.

“The ground of the possibility of the association of the manifold, so far as it lies in the object, is named the affinity of the manifold.”

None the less Kant’s subjectivism finds one of its most decided expressions in A 114.

Third Stage.—A 119-23 = III. β; A 116-19 = III. α; A 94-5 = I. § 14 C(oncluding paragraph); A 126-8 = III. δ; A 128-30 = S(ummary); A 123-6 = III. γ; A 115-16 = III. I(ntroduction); A 76-9 (B 102-4) = § 10 T(ransition to fourth stage).

A 119-23, III. β (from the beginning of the seventh paragraph to the end of the twelfth). The doctrine of objective affinity already developed in the above sections is now made to rest upon a new faculty, the productive imagination. As Vaihinger remarks, the wording of this section would seem to indicate that it is Kant’s first attempt at formulating that new doctrine. He has not as yet got over his own surprise at the revolutionary nature of the conclusions to which he feels himself driven by the exigencies of Critical teaching. He finds that it is deepening into consequences which may lead very far from the current psychology and from his own previous views regarding the nature and conditions of the knowing process and of personality. As evidence that this section was not written continuously with II. 4, [821a] we have the further fact that though the doctrine of objective affinity is dwelt upon, it is described afresh, with no reference to the preceding account. Also, the empirical processes of apprehension and reproduction, already mentioned in A 104-10, are now ascribed to the empirical imagination which is carefully distinguished from the productive.

III. α repeats “from above” the argument given in III. β “from below.” It insists upon the close connection between the categories (first introduced in II. 4[821]) with the productive imagination of III. β.

Vaihinger places III. δ next in order, on account of the connection of its argument with III. α.[822] But it dwells only upon the chief outcome of the total argument, viz. that the orderliness of nature is due to understanding. That productive imagination is not mentioned, is taken by Vaihinger to signify Kant’s recognition that it can be postulated only hypothetically, and that as doctrine it is not absolutely essential to the strict deduction.

S summarises the entire argument, and in it “pure imagination” receives mention.

Within this third stage III. γ is subsequent to the above four sections. For it carries the doctrine of productive imagination one step further. In III. β, III. α, and S, productive imagination has been treated merely as an auxiliary function of pure understanding.

“The unity of apperception in relation to the synthesis of imagination is the understanding; and the same unity with reference to the transcendental synthesis of the imagination is the pure understanding.”[823]

It is now treated as a separate and distinct faculty. So far from being a function of understanding, its synthesis “by itself, though carried out a priori, is always sensuous.”[824] It is

“one of the fundamental faculties of the human soul.... The two extreme ends, sensibility and understanding, must be brought into connection with each other by means of this transcendental function of imagination.”[825]

In this section there also appears a new element which would seem to connect it with the next following stage, namely, the addition to the series, apprehension, association, and reproduction, of the further process, recognition. As here introduced it is extremely ambiguous in character. It is counted as being empirical, and yet as containing a priori concepts. This decidedly hybrid process would seem to represent Kant’s first formulation of the even more ambiguous process, which corresponds to it in the fourth stage.

In III. I recognition is again mentioned, but this time in a form still more akin to its treatment in the fourth stage. It is not recognition through categories, but, as a form in apperception, is the

“empirical consciousness of the identity of the reproductive representations with the appearances by which they were given.”[826]

In all other respects, however, the above six sections agree (along with I. § 14 C) in holding to a threefold division of mental powers: sensibility, imagination, and apperception. This third stage is thereby marked off sufficiently clearly from the second stage in which pure imagination is wanting, and from the fourth stage in which it is dissolved into a threefold a priori synthesis.

In both I. § 14 C and in III. I the classification which underlies the third stage is explicitly formulated. Their statements harmoniously combine to yield the following tabular statement:

1. The synopsis of the manifold—a priori through sense, i.e. in pure intuition.

2. The synthesis of this manifold—through pure transcendental imagination.

3. The unity of this synthesis—through pure original transcendental apperception.

At this point Vaihinger adds to the above section the earlier passage § 10 T.[827] It is even more definitely than III. γ and III. I transitional to the fourth stage. It must be classed within the third stage, as it holds to the above threefold classification. But it modifies that classification in two respects. First, in that it does not employ the term synopsis, but only speaks of pure intuition as required to yield us a manifold. The term synopsis, as used by Kant, is, however, decidedly misleading.[828] His invariable teaching is that all connection is due to synthesis. By synopsis, therefore, which he certainly does not employ as synonymous with synthesis, can be meant only apprehension of external side-by-sideness. It never signifies anything except apprehension of the lowest possible order. Kant’s omission of the term, therefore, tends to clearness of statement. Secondly, the classification is also modified by the substitution of understanding for the unity of apperception. Apperception is, however, so obscurely treated in all of the above sections, that this cannot be regarded as a vital alteration. What is new in this section, and seems to connect it in a curious and interesting manner with sections in the fourth stage, is its doctrine of

“a manifold of a priori sensibility.” “Space and time contain a manifold of pure a priori intuition.”[829]

That is, in this connection, an entirely new doctrine. In all the previous sections of the deduction (previous in the assumed order of original writing) the manifold supplied through intuition is taken as being empirical, and as consisting of sensations. Kant here also adds that the manifold, “whether given empirically or a priori,”[830] must be synthesised before it can be known.

“The spontaneity of our thought requires that this manifold [of pure a priori intuition] should be run through in a certain manner, taken up, and connected, in order that a knowledge may be formed out of it. This action I call synthesis.”

Fourth Stage.—A 98-104; A 97-8.—As already noted, there are in Kant two persistent but conflicting interpretations of the nature of the synthetic processes exercised by imagination and understanding, the subjectivist and the phenomenalist.[831] Now, on the former view, imagination is simply understanding at work. In other words, imagination is merely the active synthesising side of a faculty whose complementary aspect appears in the logical unity of the concept. From this point of view the transcendental and the empirical factors may be taken as forming a single series. The transcendental and the empirical processes will vary together, some form of transcendental activity corresponding to every fundamental form of empirical activity and vice versa. Such an inference only follows if the subjectivist standpoint be accepted to the exclusion of the phenomenalist point of view. But since Kant constantly alternates between them, and never quite definitely formulates them in their distinction and opposition; since, in fact, they were rather of the nature of obscurely felt tendencies than of formulated standpoints, it is quite intelligible that an inference derived from the one should be drawn even at the very time when the other is being more explicitly developed. This, it would seem, is what actually happened. When we come to consider the evidence derivable from the Reflexionen and Lose Blätter, we shall find support for the view that after January 1780, on the very eve of the publication of the Critique, while the revolutionary, phenomenalist consequences of the Critical hypothesis were becoming clearer to him, he unguardedly allowed the above inference to lead him to recast his previous views in a decidedly subjectivist manner. The view that transcendental imagination has a special and unique activity altogether different in type from any of its empirical processes, namely, the “productive,” is now allowed to drop; and in place of it Kant develops the view that transcendental functions run exactly parallel with the empirical processes of apprehension, reproduction, and recognition. Accordingly, in place of the classification presented in the third stage, we find a new and radically different one introduced into the text, without the least indication that Kant’s standpoint has meantime changed. It is given in A 97:

A. Synopsis of the manifold through sense.

B. Synthesis.

1. Synthesis of apprehension of representations in [inner] intuition.

2. Synthesis of reproduction of representations in imagination.

3. Synthesis of recognition of representations in the concept.

And Kant adds in explanation that “these point to three subjective sources of knowledge which make the understanding itself possible, and which in so doing make all experience possible, in so far as it is an empirical product of the understanding.” What, now, are these three subjective sources of knowledge? They certainly are not those classified in the table of the third stage. A roughly coincides with its first member; consequently B 1 is left without proper correlate. B 2 is altogether different from the previous synthesis of imagination, for in the earlier table transcendental imagination is regarded as being solely productive, never reproductive.[832] It is now asserted to be reproductive—a contradiction of one of his own most emphatic contentions, which can only be accounted for by some such explanation as we are here stating. Nothing is lacking as regards explicitness in the statement of this new position. “...the reproductive synthesis of imagination belongs to the transcendental acts of the soul, and, in reference to it [viz. to the reproductive synthesis], we will call this power too the transcendental power of the imagination.”[833] Lastly, even B 3 does not coincide with the pure apperception of the other table. B 3 is more akin to the recognition which in the third stage is declared to be always empirical. In any case, it is recognition in the concept; and though that may ultimately involve and condition transcendental apperception, it remains, in the manner in which it is here developed by Kant, something very different. But this is a point to which we shall return. There is an added complication, running through this entire stage, which first requires to be disentangled. The transcendental syntheses are declared to condition the pure representations of space and time no less than those of sense-experience.

“This synthesis of apprehension also must be executed a priori, i.e. in reference to representations which are not empirical. For without it we could not have the a priori representations either of space or of time, since these can be generated only through the synthesis of the manifold which sensibility presents in its original receptivity. Thus we have a pure synthesis of apprehension”[834] “...if I draw a line in thought or desire to think of the time from one noon to another, or merely represent to myself a certain number, I must, firstly, apprehend these manifold representations one after the other. But if the preceding representations (the first parts of the line, the antecedent parts of time or the units serially represented) were always to drop out of my thought, and were not reproduced when I advance to those that follow, no complete representation, and none of all the aforementioned thoughts, not even the purest and first basal representations of space and time, could ever arise.”[835]

This, as Vaihinger remarks, is a point of sufficient importance to justify separate treatment. But it is introduced quite incidentally by Kant, and obscures quite as much as it clarifies the main argument.

It is convenient to start with the second synthesis. Kant’s argument is much clearer in regard to it than in regard to the other two. He distinguishes between empirical and transcendental reproduction. Reproduction in ordinary experience, in accordance with the laws of association, is merely empirical. The de facto conformity of appearances to rules is what renders such empirical reproduction possible;

“...otherwise our faculty of empirical imagination would never find any opportunity of action suited to its capacities, and would remain hidden within the mind as a dead, and to us unknown power.”[836]

Kant proceeds to argue, consistently with his doctrine of objective affinity, that empirical reproduction is itself transcendentally conditioned. The form, however, in which this argument is developed is peculiar to the section before us, and is entirely new.

“If we can show that even our purest a priori intuitions yield no knowledge, save in so far as they contain such connection of the manifold as will make possible a thoroughgoing synthesis of reproduction, this synthesis of the imagination must be grounded, prior to all experience, on a priori principles; and since experience necessarily presupposes that appearances can be reproduced, we shall have to assume a pure transcendental synthesis of the imagination as conditioning even the possibility of all experience.”[837]

In the concluding paragraph Kant makes clear that he regards this transcendental activity as being exercised in a twofold manner: in relation to the empirically given manifold as well as in relation to the a priori given manifold. How this transcendental activity is to be distinguished from the empirical is not further explained. I discuss this point below.[838]

The argument of the section on the synthesis of apprehension, to which we may now turn back, suffers from serious ambiguity. It is not clear whether a distinction, analogous to that between empirical and transcendental reproduction, is being made in reference to apprehension. The actual wording of its two last paragraphs would lead to that conclusion. That, however, is a view which would seem to be excluded by the wider context. Kant is dealing with the synthesis of apprehension in inner intuition, i.e. in time. By the fundamental principles of his teaching such intuition must always be transcendental. Empirical apprehension can only concern the data of the special senses. The process of apprehension referred to in the middle paragraph must therefore itself be transcendental.

But it is in dealing with the synthesis of recognition that the argument is most obscure. It is idle attempting to discover any possible distinction between an empirical and a transcendental process of recognition. For the transcendental process here appears as being the consciousness that what we are thinking now is the same as what we thought a moment before; and it is illustrated not by reference to the pure intuitions of space and time, but only by the process of counting. It may be argued that empirical recognition is mediated by transcendental factors—by pure concepts and by apperception. But unless we are to take transcendental recognition as synonymous with transcendental apperception, which Kant’s actual teaching does not seem to justify us in doing, such considerations will not enable us to distinguish two forms of recognition. Apart, however, from this difficulty, there is the further one that the concepts in and through which the recognition is executed are here described as being empirical. The only key that will solve the mystery of this extraordinary section, hopelessly inexplicable when viewed as a single continuous whole, is, it would seem, the theory of Vaihinger, namely,[839] that from the third paragraph onwards (already dealt with as forming the first stage of the deduction) Kant is making use of manuscript which represents the earliest form in which his explanation of the consciousness of objects was developed, with the strange result that this section is a combination of the latest and of the earliest forms of the deduction. While seeking to make out a parallelism between the empirical, conscious activities of imagination and understanding on the one hand, and its transcendental functions on the other, he must have bethought himself of the earlier attempt to explain consciousness of objects through empirical concepts conditioned by transcendental apperception, and so have attempted to expound the third form of synthesis by means of it. As thus extended it involves a distinction between transcendental and empirical apperception, and upon that the discussion, so far as it concerns anything akin to recognition, altogether turns. But there is not the least further mention of recognition itself. As transcendental, it cannot be taken as the equivalent of empirical apperception; and as a synthesis through concepts, can hardly coincide with pure apperception. The title of the section, “the synthesis of recognition in the concept,” is thus no real indication of the astonishing fare prepared for the reader. The doctrine of a threefold synthesis seems to have occurred to Kant on the very eve of the publication of the Critique. The passage expounding it may well have been hurriedly composed, and when unforeseen difficulties accumulated, especially in regard to recognition as a transcendental process, Kant must have resolved simply to close the matter by inserting the older manuscript.

III. Evidence yielded by the “Reflexionen” and “Lose Blätter” in support of the above analysis.

The evidence, derived by Vaihinger from the Reflexionen and Lose Blätter, briefly outlined, is as follows.[840] (1) In the Reflexionen zur Anthropologie relevant passages are few in number, and represent a standpoint very close to that of the 1770 Dissertation. Imagination is treated only as an empirical faculty.[841] Recognition, which is only once mentioned,[842] is also viewed as merely empirical. The understanding is spoken of as the faculty through which objects are thought.[843] The categories are not mentioned, and it is stated that the understanding yields only ideas of reflection. “All knowledge of things is derived, as regards its matter, from sensation—the understanding gives only ideas of reflection.”[844] So far, these Reflexionen would seem to coincide, more or less, with the first stage of the deduction. They contain, however, no reference to transcendental apperception; and are therefore regarded by Vaihinger as representing a still earlier standpoint.

(2) In the Reflexionen zur Kritik der reinen Vernunft there is a very large and valuable body of relevant passages. No. 925 must be of the same date as the letter of 1772 to Herz; it formulates its problem in practically identical terms.[845] Nos. 946-52 and 955 may belong to the period of the first stage. For though the doctrine of the transcendental object as the opposite counterpart of the transcendental subject is not mentioned, the spiritualist view of the self is prominent. In No. 946 it is asserted that the representation of an object is “made by us through freedom.”

“Free actions are already given a priori, namely our own.”[846] “To pass universal objective judgments, and to do so apodictically, reason must be free from subjective grounds of determination. For were it so determined the judgment would be merely accidental, namely in accordance with its subjective cause. Thus reason is conscious a priori of its freedom in objectively necessary judgments in so far as it apprehends them as exclusively grounded through their relation to the object.”[847] “Transcendental freedom is the necessary hypothesis of all rules, and therefore of all employment of the understanding.”[848] “Appearances are representations whereby we are affected. The representation of our free self-activity does not involve affection, and accordingly is not appearance, but apperception.”[849]

It is significant that the categories receive no mention.

Almost all the other Reflexionen would seem to have originated in the period of the second stage of the deduction; but they still betray a strong spiritualist bias.

“Impressions are not yet representations, for they must be related to something else which is an action. Now the reaction of the mind is an action which relates to the impression, and which if taken alone[850] may in its special forms receive the title categories.”[851] “We can know the connection of things in the world only if we produce it through a universal action, and so out of a principle of inner power (aus einem Prinzip der inneren Potestas): substance, ground, combination.”[852]

These Reflexionen recognise only the categories of relation,[853] and must therefore be prior to the twelvefold classification. There is not the least trace of the characteristic doctrines of the third and fourth stages of the deduction, viz. of the transcendental function of the imagination or of a threefold transcendental synthesis. The nature of apprehension is also most obscure. It is frequently equated with apperception.

(3) The Lose Blätter aus Kants Nachlass (Heft I.) contains fragments which also belong to the second stage of the deduction, but which would seem to be of somewhat earlier date than the above Reflexionen.[854] They have interesting points of contact with the first stage. Thus though the phrase transcendental object does not occur in them, the object of knowledge is equated with x, and is regarded in the manner of the first stage as the opposite counterpart of the unity of the self.[855] These fragments belong, however, to the second stage in virtue of their recognition of the a priori categories of relation. There is also here, as is in the Reflexionen, great lack of clearness regarding the nature of apprehension; and there is still no mention of the transcendental faculty of imagination. Fragment 8 is definitely datable. It covers the free spaces of a letter of invitation dated May 20, 1775.[856] Fragment B 12[857] belongs to a different period from the above. This is sufficiently evident from its contents; but fortunately the paper upon which it is written—an official document in the handwriting of the Rector of the Philosophical Faculty of Königsberg—enables us to decide the exact year of its origin. It is dated January 20, 1780. The fragment must therefore be subsequent to that date. Now in it transcendental imagination appears as a third faculty alongside sensibility and understanding, and a distinction is definitely drawn between its empirical and its transcendental employment. The former conditions the synthesis of apprehension; the latter conditions the synthetic unity of apperception. It further distinguishes between reproductive and productive imagination, and ascribes the former exclusively to the empirical imagination. In all these respects it stands in complete agreement with the teaching of the third stage of the deduction. The fact that this fragment is subsequent to January 1780 would seem to prove that even at that late date Kant was struggling with his deduction.[858] But the most interesting of all Vaihinger’s conclusions has still to be mentioned. He points out that at the time when this fragment was composed Kant had not yet developed the doctrine characteristic of the fourth stage, namely, of a threefold transcendental synthesis. Moreover, as he observes, the statement which it explicitly contains, that reproductive imagination is always empirical, is inconsistent with any such doctrine. The teaching of the fourth stage must consequently be ascribed to an even later date.[859]

(4) The Lose Blätter (Heft II.), though almost exclusively devoted to moral and legal questions, contain in E 67[860] a relevant passage which Reicke regards as belonging to the ‘eighties, but which Adickes and Vaihinger agree in dating “shortly before 1781.” On Vaihinger’s view it is a preliminary study for the passages of the fourth stage of the deduction. But such exact dating is not essential to Vaihinger’s argument. It is undoubtedly quite late, and contains the following sentence:

“All representations, whatever their origin, are yet ultimately as representations modifications of inner sense, and their unity must be viewed from this point of view. A spontaneity of synthesis corresponds to their receptivity: either of apprehension as sensations or of reproduction as images (Einbildungen) or of recognition as concepts.”

This is the doctrine from which the deduction of the first edition starts; it was, it would seem, the last to be developed.[861] That we find no trace of it in the Prolegomena, and that it is not only eliminated from the second edition, but is expressly disavowed,[862] would seem to indicate that it had been hastily adopted on the very eve of publication, and that upon reflection Kant had felt constrained definitively to discard it. The threefold synthesis can be verified on the empirical level, but there is no evidence that there exist corresponding transcendental activities.

IV. Connected Statement and Discussion of Kant’s Subjective and Objective Deductions in the First Edition

Such are the varying and conflicting forms in which Kant has presented his deduction of the categories. We may now apply our results to obtain a connected statement of the essentials of his argument. The following exposition, which endeavours to emphasise its main broad features, to distinguish its various steps, and to disentangle its complex and conflicting tendencies, will, I trust, yield to the reader such steady orientation as is necessary in so bewildering a labyrinth. In the meantime I shall take account only of the deductions of the first edition,[863] and from them shall strive to construct the ideal statement to which they severally approximate. Any single relatively consistent and complete deduction that is thus to serve as a standard exposition must, like the root-languages of philology, be typical or archetypal, representing the argument at which Kant aimed; it cannot be one of the alternative expositions which he himself gives. Such reconstruction of an argument which Kant has failed to express in a final and genuinely adequate form must, of course, lie open to all the dangers of arbitrary and personal interpretation. It is an extremely adventurous undertaking, and will have to be carefully guarded by constant reference to Kant’s ipsissima verba. Proof of its historical validity will consist in its capacity to render intelligible Kant’s own departures from it, and in its power of explaining the reasons of his so doing. Its expository value will be in proportion to the assistance which it may afford to the reader in deciphering the actual texts.

Our first task is to make clear the nature of the distinction which Kant draws between the “subjective” and the “objective” deductions. This is a distinction of great importance, and raises issues of a fundamental character. In regard to it students of Kant take widely different views. For it brings to a definite issue many of the chief controversies regarding Critical teaching. Kant has made some very definite statements in regard to it; and one of the opposing schools of interpretation finds its chief and strongest arguments in the words which he employs. But for reasons which will appear in due course, adherence to the letter of the Critique would in this case involve the commentator in great difficulties. We have no option except to adopt the invidious position of maintaining that we may now, after the interval of a hundred years and the labours of so many devoted students, profess to understand Kant better than he understood himself. For such procedure we may indeed cite his own authority.

“Not infrequently, upon comparing the thoughts which an author has expressed in regard to his subject, whether in ordinary conversation or in writing, we find that we can understand him better than he understood himself. As he has not sufficiently determined his concept, he has sometimes spoken, or even thought, in opposition to his own intention.”[864]

Let us, then, consider first the distinction between the two types of deduction in the form in which it is drawn by Kant. In the Preface to the first edition,[865] Kant states that his transcendental deduction of the categories has two sides, and assigns to them the titles subjective and objective.

“This enquiry, which is somewhat deeply grounded, has two sides. The one refers to the objects of pure understanding, and is intended to expound and render intelligible the objective validity of its a priori concepts. It is therefore essential to my purposes. The other seeks to investigate the pure understanding itself, its possibility and the cognitive faculties upon which it rests. Although this latter exposition is of great importance for my chief purpose, it does not form an essential part of it. For the chief question is always simply this,—what and how much can the understanding and Reason know apart from all experience? not—how is the faculty of thought itself possible? The latter is as it were a search for the cause of a given effect; and therefore is of the nature of an hypothesis (though, as I shall show elsewhere, this is not really so); and I would appear to be taking the liberty simply of expressing an opinion, in which case the reader would be free to express a different opinion.[866] For this reason I must forestall the reader’s criticism by pointing out that the objective deduction, with which I am here chiefly concerned, retains its full force even if my subjective deduction should fail to produce that complete conviction for which I hope....”

The subjective deduction seeks to determine the subjective conditions which are required to render knowledge possible, or to use less ambiguous terms the generative processes to whose agency human knowledge is due. It is consequently psychological in character. The objective deduction, on the other hand, is so named because it deals not with psychological processes but with questions of objective validity. It enquires how concepts which are a priori, and which as a priori must be taken to originate in pure reason, can yet be valid of objects. In other words, the objective deduction is logical, or, to use a post-Kantian term, epistemological in character.

It is indeed true, as Kant here insists, that the subjective deduction does not concern itself in any quite direct fashion with the Critical problem—how a priori ideas can relate to objects. “Although of great importance for my chief purpose, it does not form an essential part of it.” This, no doubt, is one reason why Kant omitted it when he revised the Critique for the second edition.[867] None the less it is, as he here says, important; and what exactly that importance amounts to, and whether it is really true that it has such minor importance as to be rightly describable as unessential, is what we have to decide.

Though empirical psychology, in so far as it investigates the temporal development of our experience, is, as Kant very justly claims, entirely distinct in aim and method from the Critical enquiry, the same cannot be said of a psychology which, for convenience, and on the lines of Kant’s own employment of terms, may be named transcendental.[868] For it will deal, not with the temporal development of the concrete and varied aspects of consciousness, but with the more fundamental question of the generative conditions indispensably necessary to consciousness as such, i.e. to consciousness in each and every one of its possible embodiments. In the definition above given of the objective deduction, I have intentionally indicated Kant’s unquestioning conviction that the a priori originates independently of the objects to which it is applied. This independent origin is only describable in mental or psychological terms. The a priori originates from within; it is due to the specific conditions upon which human thinking rests. Now this interpretation of the a priori renders the teaching contained in the subjective deduction much more essential than Kant is himself willing to recognise. The conclusions arrived at may be highly schematic in conception, and extremely conjectural in detail; they are none the less required to supplement the results of the more purely logical analysis. For though in the second edition the sections devoted to the subjective deduction are suppressed, their teaching, and the distinctions which they draw between the different mental processes, continue to be employed in the exposition of the objective deduction, and indeed are presupposed throughout the Critique as a whole. They are indispensably necessary in order to render really definite many of the contentions which the objective deduction itself contains. To eliminate the subjective deduction is not to cut away these presuppositions, but only to leave them in the obscure region of the undefined. They will still continue to influence our mode of formulating and of solving the Critical problem, but will do so as untested and vaguely outlined assumptions, acting as unconscious influences rather than as established principles. For these reasons the omission of the subjective deduction is to be deplored. The explicit statement of the implied psychological conditions is preferable to their employment without prior definition and analysis. The deduction of the second edition rests throughout upon the initial and indispensable assumption, that though connection or synthesis can never be given, it is yet the generative source of all consciousness of order and relation. Factors which are transcendental in the strict or logical meaning of the term rest upon processes that are transcendental in a psychological sense.

This last phrase, ‘transcendental in a psychological sense,’ calls for a word of justification. The synthetic processes generative of experience are not, of course, transcendental in the strict sense. For they are not a priori in the manner of the categories. None the less they are discoverable by the same transcendental method, namely, as being, like the categories, indispensably necessary to the possibility of experience. They differ from the categories in that they are not immanent in experience, constituent of it, and cannot therefore be known in their intrinsic nature. As they fall outside the field of consciousness, they can only be hypothetically postulated. None the less, formal categories and generative processes, definable elements and problematic postulates, alike agree in being conditions sine qua non of experience. And further, in terms of Kant’s presupposed psychology, the latter are the source to which the former are due. There would thus seem to be sufficient justification for extending the term transcendental to cover both; and in so doing we are following the path which Kant himself willingly travelled. For such would seem to have been his unexpressed reasons for ascribing, as he does, the synthetic generative processes to what he himself names transcendental faculties.

This disposes of Kant’s chief reason for refusing to recognise the subjective deduction as a genuine part of the Critical enquiry, namely, the contention upon which he lays such emphasis in the prefaces both of the first and of the second edition,[869] that in transcendental philosophy nothing hypothetical, nothing in any degree dependent upon general reasoning from contingent fact, can have any place. That contention proves untenable even within the domain of his purely logical analyses. The very essence of his transcendental method consists in the establishment of a priori elements through proof of their connection with factual experience. Kant is here revealing how greatly his mind is still biased by the Leibnizian rationalism from which he is breaking away. His a priori cannot establish itself save in virtue of hypothetical reasoning.[870] His transcendental method, rightly understood, does not differ in essential nature from the hypothetical method of the natural sciences; it does so only in the nature of its starting-point, and in the character of the analyses which that starting-point prescribes. And if hypothetical reasoning may be allowed in the establishment of the logical a priori, there is no sufficient reason why it may not also be employed for the determination of dynamical factors. The sole question is as to whether the hypotheses conform to the logical requirements and so raise themselves to a different level from mere opinion and conjecture.[871] As Kant himself says,[872] though his conclusions in the subjective deduction may seem to be hypothetical in the illegitimate sense, they are not really so. From the experience in view of which they are postulated they receive at once the proof of their actuality and the material for their specification.

We may now return to the question of the nature of the two deductions. The complex character of their interrelations may be outlined as follows:

1. Though the subjective deduction is in its later stages coextensive with its objective counterpart, in its earlier stages it moves wholly on what may be called the empirical level. The data which it analyses and the conditions which it postulates are both alike empirical. The objective deduction, on the other hand, deals from start to finish with the a priori.

2. The later stages of the subjective deduction are based upon the results of the objective deduction. The existence and validity of a priori factors having been demonstrated by transcendental, i.e. logical, analysis, the subjective deduction can be extended from the lower to the higher level, and can proceed to establish for the a priori elements what in its earlier stages it has determined for empirical consciousness, namely, the nature of the generative processes which require to be postulated as their ground and origin. When the two deductions are properly distinguished the objective deduction has, therefore, to be placed midway between the initial and the final stages of the subjective deduction.

3. The two deductions concentrate upon different aspects of experience. In the subjective deduction experience is chiefly viewed as a temporal process in which the given falls apart into successive events, which, in and by themselves, are incapable of constituting a unified consciousness. The fundamental characteristic of human experience, from this point of view, is that it is serial in character. Though it is an apprehension of time, it is itself also a process in time. In the objective deduction, on the other hand, the time element is much less prominent. Awareness of objects is the subject-matter to which analysis is chiefly devoted. This difference very naturally follows from the character of the two deductions. The subjective enquiry is mainly interested in the conditions generative of experience, and finds its natural point of departure in the problem by what processes a unified experience is constructed out of a succession of distinct happenings. The objective deduction presents the logical problem of validity in its most striking form, in our awareness of objects; the objective is contrasted with the subjective as being that which is universally and necessarily the same for all observers. Ultimately each of the two deductions must yield an analysis of both types of consciousness—awareness of time and awareness of objects; a priori factors are involved in the former no less than in the latter, and both are conditioned by generative processes. Unfortunately the manner in which this is done in the Critique causes very serious misunderstanding. The problem of the psychological conditions generative of consciousness of objects is raised[873] before the logical analysis of the objective deduction has established the data necessary for its profitable discussion. The corresponding defect in the objective deduction is of a directly opposite character, but is even more unfortunate in its effects. The results obtained from the analysis of our awareness of objects are not, within the limits of the objective deduction, applied in further analysis of our consciousness of time. That is first done, and even then by implication rather than by explicit argument, in the Analytic of Principles. This has the twofold evil consequence, that the relations holding between the two deductions are very greatly obscured, and that the reader is not properly prepared for the important use to which the results of the objective deduction are put in the Analytic of Principles. For it is there assumed—a quite legitimate inference from the objective deduction, but one whose legitimacy Kant has nowhere dwelt upon and explained—that to be conscious of time we must be conscious of it as existing in two distinct orders, subjective and objective. To be conscious of time we must be conscious of objects, and to be conscious of objects we must be able to distinguish between the order of our ideas and the order of the changes (if any) in that which is known by their means.

Thus the two deductions, properly viewed in their full scope, play into one another’s hands. The objective deduction is necessary to complete the analysis of time-consciousness given in the subjective deduction, and the extension of the analysis of object-consciousness to the explanation of time-consciousness is necessary in order to make quite definite and clear the full significance of the conclusions to which the objective enquiry has led.[874]

One last point remains for consideration. Experience is a highly ambiguous term, and to fulfil the rôle assigned to it by Kant’s transcendental method—that of establishing the reality of the conditions of its own possibility—its actuality must lie beyond the sphere of all possible controversy. It must be itself a datum, calling indeed for explanation, but not itself making claims that are in any degree subject to possible challenge. Now if we abstract from all those particularising factors which are irrelevant in this connection, we are left with only three forms of experience—experience of self, experience of objects, and experience of time. The two former are open to question. They may be illusory, as Hume has argued. And as their validity, or rather actuality, calls for establishment, they cannot fulfil the demands which the transcendental method exacts from the experience whose possibility is to yield proof of its discoverable conditions. Consciousness of time, on the other hand, is a fact whose actuality, however problematic in its conditions, and however mysterious in its intrinsic nature, cannot, even by the most metaphysical of subtleties, be in any manner or degree challenged. It is an unquestioned possession of the human mind. Whether time itself is real we are not metaphysically certain, but that, whatever be its reality or unreality, we are conscious of it in the form of change, is beyond all manner of doubt. Consciousness of time is the factual experience, as conditions of whose possibility the a priori factors are transcendentally proved. In so far as they can be shown to be its indispensable conditions, its mere existence proves their reality. And such in effect is the ultimate character of Kant’s proof of the objective validity of the categories. They are proved in that it is shown that only in and through them is consciousness of time possible.

The argument gains immeasurably in clearness when this is recognised;[875] and the deduction of the first edition of the Critique, in spite of its contorted character, remains in my view superior to that of the second edition owing to this more explicit recognition of the temporal aspect of consciousness and to employment of it as the initial starting-point. Analysis at once reveals that though consciousness of time is undeniably actual, it is conditioned in complex ways, and that among the conditions indispensably necessary to its possibility are both consciousness of self and consciousness of an objective order of existence. Starting from the undeniable we are thus brought to the problematic; but owing to the factual character of the starting-point we can substantiate what would otherwise remain open to question.

As this method of formulating Kant’s argument gives greater prominence to the temporal factor than Kant himself does in his statement of the deductions, the reader may very rightly demand further evidence that I am not, by this procedure, setting the deductions in a false or arbitrary perspective. Any statement of Kant’s position in other than his own ipsissima verba is necessarily, in large part, a matter of interpretation, and proof of its correctness must ultimately consist in the success with which it can be applied in unravelling the manifold strands that compose his tortuous and many-sided argument; but the following special considerations may be cited in advance. Those parts of the Critique, such as the chief paragraphs of the subjective deduction and the chapter on Schematism, which are demonstrably late in date of writing, agree in assigning greater prominence to the temporal aspect of experience. This is also true of those numerous passages added in the second edition which deal with inner sense. All of these show an increasing appreciation of the central rôle which time must play in the Critical enquiries. Secondly, proof of the validity of specific categories is given, as we shall find,[876] not in the objective deduction of the Analytic of Concepts, but only in the Analytic of Principles. What Kant gives in the former is only the quite general demonstration that forms of unity, such as are involved in all judgment, are demanded for the possibility of experience. Now when proof of the specific categories does come, in the Analytic of Principles, it is manifestly based on the analysis of time-experience. In the three Analogies, for example, Kant’s demonstration of the objective validity of the categories of relation consists in the proof that they are necessary conditions of the possibility of our time-consciousness. That is to say, the transcendental method of proof, when developed in full detail, in reference to some specific category, agrees with the formulation which I have given of the subjective and objective deductions. In the third place, Kant started from a spiritualist standpoint, akin to that of Leibniz,[877] and only very gradually broke away from the many illegitimate assumptions which it involves. But this original starting-point reveals its persisting influence in the excessive emphasis which Kant continued to lay upon the unity of apperception. He frequently speaks[878] as if it were an ultimate self-justifying principle, by reference to which the validity of all presupposed conditions can be established. But that, as I have already argued, is a legitimate method of procedure only if it has previously been established that self-consciousness is involved in all consciousness, that is, involved even in consciousness of sequence and duration. And as just stated, the deductions of specific categories, given in the Analytic of Principles, fulfil these requirements of complete proof. They start from the time-consciousness, not from apperception.