¹ Throughout, I shall cite passages from the Critique of Pure Reason using the standard A/B pagination. I follow, with occasional modification, the translation of Norman Kemp Smith (London: Macmillan, 1929).

² By way of an attempt to ground the logical functions directly in unity of apperception, there have been attempts to ground the categories wholly in the latter. Most notably: Reich, Klaus, The Completeness of Kant's Table of Judgments, translation of revised edition (1948) by Kneller, Jane and Losonsky, Michael (Stanford, CA: Stanford University Press, 1992).Google Scholar For a critique of Reich's and some other recent approaches, see Brandt, Reinhard, The Table of Judgments, tr. Watkins, Eric, North American Kant Society Studies in Philosophy 4 (Atascadero, CA: Ridgeview Publishing Co., 1995).Google Scholar

³ Smith, Norman Kemp: ‘Judgment, its nature and conditions, is the real problem of the misnamed Analytic of Concepts … the two main divisions of the Analytic deal with one and the same problem (A Commentary to Kant's ‘Critique of Pure Reason,’ 2nd edition (London: Macmillan, 1923), p.343).Google Scholar Cf. Guyer, Paul, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, 1987), p.27.CrossRefGoogle Scholar Of course I agree that concepts are nothing apart from the exercise of ('determinant') judgement. I also agree with most commentators that this still permits a distinction between the latter and the third Critique's faculty of reflective judgement. Where I perhaps differ from most commentators is in holding that, while reflective judgement may be exercised independently of determinant judgement, any exercise of the latter does not simply presuppose, but is, an exercise of the former. See my ‘Unity of organism, unity of thought, and the unity of the Critique of Judgement’, The Southern Journal of Philosophy, supplement (1991 Spindel Conference) to Vol.30 (1992), esp. pp.152–3; also Longuenesse, Béatrice, Kant et le pouvoir de juger (Paris: Presses Universitaires de France, 1993), 208ff.Google Scholar As for the distinction between ‘doctrines’ in the first Critique, I note the following beyond what I argue in this paper. We may distinguish between a more ‘subjective’ and more ‘objective’ argument involving unity of apperception in the deduction. The former turns on an account of the very nature of judgement as an activity; by contrast, the System of Principles is only concerned with the truth of certain judgements. This may be regarded as grounding a distinction between separate ‘doctrines.’ (As for the Schematism, its location within the Analytic of Principles may rest on the fact that it not only builds on the deduction, but precisely on its objective side. I argue for this below.) I have emphasized the subjective side elsewhere: Matter in Mind: A Study of Kant's Transcendental Deduction (Bloomington: Indiana University Press, 1989).Google Scholar Here I focus on the objective side. (In any case, I do not assume that this way of drawing the subjective/objective distinction is what Kant himself has in mind in distinguishing two ‘sides’ of the deduction, Axvi–xvii.)

⁴ ‘This division is developed systematically from a common principle, namely, the faculty of judgment (which is the same as the faculty of thought) … Nor could we [in any other way] discover why just these concepts, and no others, have their seat [beiwohnen] in the pure understanding’ (A80–1/B106–7). This same reason of course explains why these concepts are ‘pure’ as opposed to empirical.

⁵ ‘[W]e are faced by the problem how these concepts can relate to objects which they yet do not obtain from any experience’ (A85/B117), for ‘they relate to their objects without having borrowed from experience anything that can serve in the representation of these objects’ (A86/B118). Kant also puts it a second way: How can such concepts relate to objects a priori? ‘The explanation of the manner in which concepts can thus relate a priori to objects I entitle their transcendental deduction’ (A85/B117). Apparently, the only way to legitimate our presumption that they relate to objects is by showing that they must: ‘Save through their original relation to possible experience, in which all objects of knowledge are found, their relation to any one object would be quite incomprehensible’ (A94/B126–7).

⁶ ‘The objective validity of the categories as a priori concepts rests, therefore, on the fact that, so far as the form of thought is concerned, through them alone does experience become possible. They relate of necessity and a priori to objects of experience, for the reason that only by means of them can any object whatsoever of experience be thought' (A93/B126).

Still, there is a problem. The objects in question are also objects of intuitional apprehension. So how without further ado can we hold that conditions upon judgements about objects represent features of the objects themselves? Kant himself seems aware that more is required:

The concepts which thus contain a priori the pure thought involved in every experience, we find in the categories. If we can prove that by their means alone an object can be thought, this will be a sufficient deduction of them, and will justify their objective validity. But since in such a thought more than simply the faculty of thought, the understanding, is brought into play, and since this faculty itself, as a faculty of knowledge that is meant to relate to objects, calls for explanation in regard to the possibility of such relation, we must first of all consider, not in their empirical but in their transcendental constitution, the subjective sources which form the a priori foundation of the possibility of experience. (A96–7)

As this suggests, a full response to the problem of the deduction calls for attention to its ‘subjective’ side. My own view, elaborated elsewhere (see n. 3), is that this needs to explain, not merely that certain intellectual structures are necessary for experience, but how it is possible to regard those structures as ‘entering into’ or ‘shaping’ experience precisely by informing the very intuitions involved in experience. In turn, this calls for a view according to which non-conceptual anticipations regarding the advance of possible consciousness — essential to the ‘objective’ account proposed here in section 2 — are as directly incorporable into intuitional states as are the sensations that they contain.

⁷ ‘The explanation of the manner in which concepts can [my emphasis] thus relate a priori to objects I entitle their transcendental deduction …' (A85/B117; cf. A96–7, B159).

⁸ Strictly, ‘How [the categories] make experience possible’ and ‘what are the principles’ refer to the schematism and system of principles, respectively.

⁹ ‘[I]f category and sensuous intuition are really heterogeneous, no subsumption is possible; and if they are not really heterogeneous, no such problem as Kant here refers to will exist’ (Smith, Kemp, Commentary, p.334).Google Scholar Jonathan Bennett takes a somewhat different line to the same effect, emphasizing the difficulty of formulating an intelligible problem, specific to the categories, for Kant to wrestle with in the first place: Kant's Analytic (Cambridge: Cambridge University Press, 1966), pp.148ff.Google Scholar As for a problem regarding the possibility of subsumption under concepts in general: ‘[its resolution] has never been better presented than by Kant himself’ precisely in his account of understanding as a faculty of rules in the first place (Bennett, , Kant's Analytic, p.144Google Scholar; cf. Smith, Kemp, Commentary, pp.334–5).Google Scholar

¹⁰ Some commentators see the system of principles — to its advantage — as not really bearing at all on the notion of unity of apperception, at least as it seems primarily to function in the deduction. Thus Guyer (who sees Kant as torn between various strategies in the deduction): ‘[S]ince, as we saw, there is no successful argument that the unity of apperception requires a priori rules which is independent [of the general principle of the analogies of experience] … the analogies cannot depend on a prior proof that the unity of apperception, more abstractly conceived, requires a priori rules …’ (Claims of Knowledge, pp.209–10; cf. p.27). Others dilute the deduction's concern with unity of apperception in the first place. Even in that case, however, the deduction tends not to fare well. For Strawson, for example, what it does is ‘too general and too obscure to have more than a precarious hold on our minds'; it is up to the principles, with ‘definite and acceptable arguments,’ to save the day: Strawson, P. F., The Bounds of Sense (London: Methuen & Co., 1966), p.18.Google Scholar

¹¹ Cf. Bird's, Graham claim that ‘unity of apperception’ refers to two different things for Kant. Of course we are in agreement that one of them is a certain sort of ‘unity or identity of the person.’ Bird describes the other as ‘a unity or identity in concepts or language’ or ‘a certain unity of language in accordance with which we name and describe what is presented to our senses’ (Kant's Theory of Knowledge (London: Routledge & Kegan Paul, 1962), pp.120–1Google Scholar). In some ways, my own approach to the second notion might be regarded as more on the ‘metaphysical’ side.

¹² Cf. A106–8, B132–3.

¹³ The rest of the present section speaks to this point.

¹⁴ Perhaps the most frequently cited passage to this effect is: ‘all combination — be we conscious of it or not, be it a combination of the manifold of intuition, empirical or non-empirical, or of various concepts — is an act of the understanding’ (B130).

¹⁵ This seems to be the general aim behind all five versions of the deduction formulated by Wolff, Robert Paul: Kant's Theory of Mental Activity (Cambridge: Harvard University Press, 1963).Google Scholar By contrast, while finding them in the deduction as well (pp.130–4), Bennett dismisses several attempts of this sort as aside from the main trove of ‘some good things’ (Kant's Analysis, p.100) to be found there; all of the latter seem to me to dilute the deduction's main concern with unity of apperception.

¹⁶ Kemp Smith says ‘united in’.

¹⁷ This has an affinity with Allison's approach, in so far as Allison emphasizes that the doctrine of unity of apperception is only meant to bear on the ‘I think’ that is able to accompany all my representation ‘if the representation is to function as a representation, that is, to represent some object’ (Allison, Henry E., Kant's Transcendental Idealism (New Haven: Yale University Press, 1983), p.137).Google Scholar However, Allison's approach turns on the demand for unity only among particular manifolds of representations in the representation of objects; the present approach demands, as integral to the representation of objects, a connection between any such particular representations and all possible representations.

¹⁸ In what follows, I am concerned with the notion of unity of apperception in (1) only in so far as it bears on Kant's concern with (2'). It seems clear that Kant also has an independent interest in (1), in connection with a highly original, if not altogether clear, view regarding the sense and reference of the first-person singular pronoun. Going beyond the position developed in Matter and Mind (ch.6), I have developed my view on this in: ‘Self as matter and form: some reflections on Kant's view of the soul’, in Klemm, D. and Zoeller, G. (eds.), Figuring the Self: Subject, Absolute, and Others in Classical German Philosophy (Albany: State University of New York Press, 1997).Google Scholar I argue that the purely ‘formal’ notion of self that is embodied in Kant's doctrine of transcendental unity of apperception is not incompatible with identification of the ‘transcendental subject of one's thoughts’ with an ‘I of inner sense’ — thus with some sort of ‘object' after all, though neither a purely noumenal subject nor an ordinary human being qua human.

¹⁹ Since ‘all necessity … is grounded in a transcendental condition’, there must ‘be a transcendental ground of the unity of consciousness in the synthesis of the manifold of all our intuitions’ (A106); here, presumably by way of anticipation, there is already a reference to a unity comprising all our intuitions.

²⁰ Since at A103 Kant had held that the combination of a manifold of representations has as its necessary condition a synthesis in accordance with concepts, and at A105 that the reference of representations to an object has as its necessary condition a synthesis in accordance with rules and thus concepts, Carl suggests that it is the necessity of these conditional relations that is to be transcendentally grounded (Carl, Wolfgang, Die Transzendentale Deduktion der Kategorien in der ersten Auflage der Kritik der reinen Vernunft: Ein Kommentar (Frankfurt: Vittorio Klostermann, 1992), p.176).Google Scholar

Guyer, (Claims of Knowledge, pp. 106ff.Google Scholar, 103ff.) suggests and criticizes two alternatives. One, defended by Henrich, concerns the synthetic truth that, necessarily, for any representational state it is possible to have, it is possible to combine it with others in a consciousness of onself as having them; this requires a broadly ‘Cartesian’ certainty of the numerical identity of the self. (Cf. Henrich, Dieter, Identität und Objektivität (Heidelberg: Carl Winter, 1976), pp. 84ff.Google Scholar, and ‘The identity of the subject in the transcendental deduction’, in Schaper, Eva and Vossenkuhl, Wilhelm (eds.), Reading Kant (Oxford: Basil Blackwell, 1989), pp.250–80.Google Scholar) The other relies on Kant's ‘own definition of an object as the expression of a necessary connection’. Guyer dismisses it because it begs the question against the empiricist's rejection of necessary connections among appearances. My own approach is, at least broadly, of this second type.

²¹ More exactly, we would presumably also include representation of a manifold of possible ways of having gotten to the given appearance in the first place. For a rather different approach that nevertheless likewise turns on the representation of possible ways of proceeding from, and having arrived at, a given representation, see Melnick, Arthur, Space, Time, and Thought in Kant (Dordrecht: Kluwer Academic Publishers, 1989).CrossRefGoogle Scholar

Apart from the immediate context, that something like this is in question also seems evident from the following culmination of the more ‘systematic’ presentation of Kant's argument:

Now if this unity of association had not also an objective ground which makes it impossible that appearances should be apprehended by the imagination otherwise than under the condition of a possible synthetic unity of this apprehension, it would be entirely accidental that appearances should fit together into a connected whole [Zusammenhang] of human knowledge … There must, therefore, be an objective ground … upon which rests the possibility, nay, the necessity, of a law extending to all appearances, to regard them all as data of the senses that must be associable in themselves and subject to universal rules of a thoroughgoing connection in their reproduction. (A121–2)

²² There are presumably considerations that bear here on the two-step development of the argument of the B-deduction, in contrast with the A-version. Since I do not think that this feature of the B-deduction affects my point, I abstract from the two-step character of its argument.

²³ See ‘Introduction’ to the third Critique, section 4. Unlike some commentators, I do not take this passage as reason for concluding that Kant never intended to argue, in the second analogy, for a view of causal determination in terms of laws: cf. Buchdal, Gerd, Metaphysics and the Philosophy of Science (Cambridge: MIT Press, 1969), pp.500–5Google Scholar, 516–19, 641–5.

²⁴ There may thus seem to be an ambiguity. Is the ‘whole’ in question inner sense, or rather the ‘possibility of experience’ that Kant identifies with transcendental unity? Kant sees a special connection between the two: ‘For the original apperception stands in relation to (bezieht sich auf) inner sense (the whole (Inbegriff) of all representations), and indeed a priori to its form’ (A177/B220). See section 3 for more on the connection between unity of apperception and inner sense.

²⁵ In this passage, unity of apperception is described as the ‘form’ of the whole in question:

But there is an alternative method, namely, to investigate the possibility of experience as [of] a knowledge wherein all objects — if their representation is to have objective reality for us — must finally be capable of being given to us. In this third (in diesem Dritten), the essential form of which consists in the synthetic unity of the apperception of all appearances, we have found a priori conditions of complete and necessary determination of time for all existence in the [field of] appearance, without which even empirical determination of time would be impossible. In it we have also found rules of synthetic unity a priori, by means of which we can anticipate experience. (A217/B265)

Use of the term ‘unity of apperception’ both for a special sort of whole and for the a priori form of that whole is unsurprising, given similar double-duty done by the term ‘form of intuition’ in Kant.

²⁶ Synthetic unity of the manifold of intuitions, as given a priori (als a priori gegeben), is thus the ground of the identity of apperception, which precedes a priori all my determinate thought' (B134; Kemp Smith, following Vaihinger, substitutes hervorgebracht for gegeben). ‘Thus unity of the synthesis of the manifold … is given a priori (a priori… gegeben) as the condition of the synthesis of all apprehension …’ (B161)

²⁷ Some examples — translations from Opus postumum, ed. E. Förster, tr. E. Förster and M. Rosen (Cambridge: Cambridge University Press, 1993); pagination also indicated from vol. 22 of the Akademie-Ausgabe of Kant's gesammelte Schriften, ed. Prussian (later German) Academy of Sciences (Berlin: Walter de Gruyter & Co., 1900–8) — ‘[T]here is only one space and one time. Through this representation, all objects of empirical representation are connected into an absolute whole — all are representations through which the subject constitutes itself according to its possibility (by synthetic a priori propositions)’ (p.171, 22.12); ‘Space and time are only subjective forms of sensible intuition, which contain the axioms: There is only one space and one time, in which an infinite aggregate of perceptions can be coordinated with one another into a system’ (p.172, 22.29); ‘Space and time are intuitions with the dynamic function of positing a manifold of intuition as appearance (dabile); thus also an aspectabile, as appearance, which precedes all apprehensive representation (perception as empirical representation with consciousness) and is thought synthetically a priori, according to a principle as thoroughly determining (intuitus quern sequitur conceptus) in which the subject posits itself in the the collective unity of the manifold of intuition’ (p. 179, 22.44); ‘Space and time in intuition are not things but the acts of the power of representation positing itself, through which the subject makes itself into an object’ (p.193, 22.88). Some additional passages in the Critique which compare unity of apperception with space and time are A110, A123–4, A117n., A127.

²⁸ Presumably, in thinking of the schemata as ‘intellectual’ and ‘resting upon an a priori rule’, Kant is thinking of the transcendental deter minations of time in question precisely as something already ‘determined’ with respect to the corresponding concepts.

²⁹ The categories contain ‘only the form of the thought of an object in general’ (A51/B75); they are ‘concepts of an object in general’ (B128; cf. B146, B158).

³⁰ Kemp Smith translates this as the ‘scope’ of time.

³¹ What follows is a conclusion regarding limitation of the categories within the bounds of experience, apart from which they are ‘merely functions of the understanding for concepts’ (A147/B187).

³² Of course it is also essential to the schematism chapter that we see a specific correlation with the clue chapter's logical functions of understanding in judgement. I am abstracting from this issue. My only concern is with the problem of ‘application’.

³³ Letter to Reinhold, 12 May 1789, tr. Zweig, Arnulf, Kant: Philosophical Correspondence 1759–99 (Chicago: University of Chicago Press, 1967, p.141)Google Scholar; also included as Appendix A to Allison, Henry E., The Kant-Eberhard Controversy (Baltimore, MD: The Johns Hopkins University Press, 1973).Google Scholar

³⁴ Of course by the same token it is arguable that appearances are not subsumable under empirical concepts. However, Kant can still open the schematism chapter with the legitimate observation that, at least in a certain respect, it is unproblematic that appearances are subsumable under empirical concepts. For example, at least the roundness thought in the concept of a plate is something that can be intuited (anschauen lässt) (A137/B176).

³⁵ Strictly, something more is also required, namely, an explanation of how the bringing-into-relation in question can be regarded as informing particular instances of intuitional apprehension. See n. 6, above.

³⁶ Kemp Smith says ‘simply’.

³⁷ Obviously, this is a form of ‘phenomenalism’. Here is a relevant passage from the section of the system of principles dealing with ‘the highest principle of all synthetic judgments': ‘That an object be given (if this expression be taken, not as referring to some merely mediate process, but as signifying immediate presentation (darstellen) in intuition), means simply (ist nichts anderes) that the representation through which the object is thought relates to actual or possible experience’ (A155–6/B195). I return to the point about phenomenalism at the end of section 4.

³⁸ It is of course only in so far as the principles of quantity and quality put forth in the system of principles are applicable to the very same objects to which the relational categories are applicable that there is anything in Kant's argument for the former that could not have been presented earlier in the Critique.

³⁹ Cf. Wolff, Kant's Theory: despite ‘great flourish and fanfare’, the section is ‘completely redundant’ with respect to the progress of Kant's argument, ‘merely a repetition of the Deduction principle’ (pp.226–7).

⁴⁰ Guyer, , Claims of Knowledge, p.179.Google Scholar I have added the bracketed letters.

⁴¹ According to Guyer (p.207), ‘[T]he real argument of the analogies, especially the second analogy's argument for a principle of universal causation, depends on the fundamental premise underlying the refutation of idealism’; apart from that, it is an attempt to show, in a manner completely independent of the deduction, ‘that certain principles are required in order to ground temporal determinations about objects’. While this line of argument was in fact Kant's original plan for the deduction, Guyer concludes ‘that the distinction between the “Analytic of Concepts” and the “Analytic of Principles” — that is, an initial defense of pure categories of the understanding, followed by a further argument for them in their “schematized” or temporally significant form — is artificial, and that the only part of even the “Analytic of Principles” which is really compelling is the theory of time-determination outlined in the “Analogies of Experience” as supplemented by the “Refutation of Idealism”’ (p.27).

⁴² Guyer says that the omission ‘makes it appear as though the principle of the analogies contains no advance at all over what has (supposedly) been established in the transcendental deduction’ (ibid., p.208). Although Kemp Smith does not find the revision to be of any special significance (p.355), he does claim that ‘in the Analytic of Concepts the temporal aspect of experience falls somewhat into the background, whereas in the Analytic of Principles it is emphasised’ (Commentary, p.343). Cf. Wolff, , Kant's Theory, p.239.Google Scholar Paton accords more with my own view in noting that the revisions ‘generally lay more stress upon the object’ (Kant's Metaphysic of Experience (London: George Allen & Unwin, 1936), vol. 2, p. 168n. 1); the elimination of reference to time was ‘in the interests of brevity’ (ibid., p.160).

⁴³ Wolff, , Kant's Theory, p.240Google Scholar: the analogies are about to make good on the deduction's failure to ground the requirement of something stronger than merely associative connections among appearances.

⁴⁴ Kemp Smith suggests that the original formulation reflected, for the worse, Kant's ‘extremely misleading distinction between mathematical and dynamical principles’ (p.355n.).

⁴⁵ Contrary to Guyer, I do not regard Kant as supposing that the deduction has established the necessity of regarding appearances as objects or appearances of objects, but only as putting forth necessary conditions for such regard. In any case, not emphasizing the term ‘existence,’ but rather ‘appearance’ as opposed to ‘perception,’ Guyer regards the second formulation as felicitous at least as a counterweight to the first's suggestion that: ‘the basic principles for judgements that subjective representations signify independent objects and events have been established prior to the arguments for the analogies, and that what the latter are intended to do is only to ground certain special judgements, about certain special temporal determinations or aspects of objects or events already judged to be objective’ (Guyer, , Claims of Knowledge, p. 209).Google Scholar

⁴⁶ Cf. Smith, Kemp, Commentary, p.355Google Scholar: ‘The two proofs [of the general principle] repeat the main argument of the transcendental deduction, but with special emphasis upon the temporal aspect of experience'; apart from that, the references to time in the 2nd-edition proof ‘are too condensed to be intelligible save in the light’ of the proofs of the particular analogies. Cf. Wolff, , Kant's Theory, p.239Google Scholar; Guyer, , Claims of Knowledge, p.208.Google Scholar A. C. Ewing simply observes that the general principle and the arguments for it ‘cannot be understood’ apart from the arguments for the particular analogies (A Short Commentary on Kant's Critique of Pure Reason (Chicago: University of Chicago Press, 1938), pp.149–50); cf. Paton, , Kant's Metaphysic, p.171.Google Scholar

⁴⁷ For a recent non-epistemic interpretation of the second analogy, together with a classification of various possible approaches to it: Bayne, Steven M., ‘Objects of representations and Kant's second analogy’, Journal of the History of Philosophy, 32 (1994), 381–410.CrossRefGoogle Scholar

⁴⁸ Note that this falls short of the claim that experience (in a broad sense) is necessarily such that representation of appearances as objects is to any particular degree justified.

⁴⁹ I emphasize: a certain sort of phenomenalism. The view is phenomenal-istic to the extent that the notion of an object is explicated wholly in terms of (a) the introduction of a certain sort of structure into our anticipations regarding the apprehension of possible appearances and (b) the possibility of incorporating manifolds of anticipations, thus structured, within instances of intuitional apprehension. What is crucially not the case: it is not that the ‘items’ that are thereby being taken as objects (the appearances in question) are really items of some other sort altogether, e.g., sense data or ways of appearing. Nor is it that the judgements in question are really about our own anticipations, rather than objects. Apart from what is contained in the thought of them as objects, the ‘items’ in question are not items at all; they are simply the intentional objects of intuitional apprehension.

⁵⁰ For a discussion of the possibility that our representation of space must be an idea, but perhaps neither an idea of reason nor (like that of the sublime) an aesthetic idea, but rather some sort of ‘third thing’: Wohlfart, Günter, ‘Ist der Raum eine Idee?’, Kant-Studien, 71 (1980), 137–54.CrossRefGoogle Scholar And two final reminders: first, I have argued for a respect in which the schematism and the general principle of the analogies build on the result of the deduction. It should not be supposed that this exhausts the role of the deduction in the Critique. As noted earlier, there is an additional and crucial task to be accomplished by the deduction. I do not pursue it here since it has been the focus of my attention elsewhere; see n. 6, above. Second, I have been concerned with unity of apperception only in so far as I take it to be involved in a line of argument integrating the deduction, schematism, and system of principles. This is not meant to exclude its role in an independently valuable attempt to deal with the problem of the ‘I'; see n. 18, above.