¹ 20 January 1792 (11: 313; 315). References to Kant's works, with the exception of the Critique of Pure Reason, refer to the volume and page of the German Academy of Sciences edition, Gesammelte Schriften, ed. Wissenschaften, Königlich Preußischen Akademie der, later the Deutschen Akademie der Wissenschaften zu Berlin. 29 volumes (Berlin: Walter de Gruyter [and predecessors], 1902–).Google Scholar References to the Critique of Pure Reason follow the pagination of the first (A) and second (B) editions. Translations are my own, but I consulted the commonly used English translations. The following abbreviations are used: G=Groundwork for the Metaphysics of Morals; KpV=Critique of Practical Reason; KrV=Critique of Pure Reason; KU=Critique of the Power of Judgement; P=Prolegomena to Any Future Metaphysics.

² See KpV (5: 8n.) and G (4: 393).

³ B146 and B165–6 both say that empirical cognition is experience; Kant's wording at Bl and B128 suggests that empirical cognition is the genus of which experience is a species.

⁴ P §4 (4: 274).

⁵ Admittedly, Kant refers only to an analysis and a synthesis in this passage, and not analytic and synthetic arguments. Though perhaps not every analysis is happily conceived as an analytic argument, it seems fair to suppose that an analysis that drives towards a definite end (i.e. a purposeful analysis) should count as an analytic argument if anything does.

⁶ A prominent advocate of this interpretation is Patricia Kitcher. See especially Kitcher, Patricia, Kant's Transcendental Psychology (Oxford and New York: Oxford University Press, 1990)Google Scholar.

⁷ A prominent advocate of this interpretation is Strawson, P. F., in The Bounds of Sense: An Essay on Kant's ‘Critique of Pure Reason’ (London: Methuen, 1966)Google Scholar.

⁸ I wish to emphasize that it is not my aim to advance a sustained or detailed critique of either Kitcher's or Strawson's work. I shall be considering their respective positions largely in the abstract, and only in the context of considering – more deliberately than I think either has done – Kant's own conception of the method of the Critique.

⁹ Thus I think that we can best understand the starting point of the Deduction if we appreciate how it is related to the starting point of the entire book. There is widespread disagreement among commentators about the Deduction's starting point. According to P. F. Strawson, the problem with the idea that the starting point of the Deduction is experience concerns the ambiguity of the term: for if experience is taken in the considered Kantian sense (where experience is a mode of empirical cognition, and hence is already categorially structured), this threatens to trivialize the argument that ensues. Thus, Strawson and other commentators claim that the Deduction begins with some bare conception of self-consciousness. Karl Ameriks argues that the starting point of the Transcendental Deduction is empirical know-ledge, and that this does not doom the argument of the Deduction to triviality (‘Kant's Transcendental Deduction as a regressive argument’, Kant Studien, 19 [1978], 273–87Google Scholar). My account of the starting point of the Deduction is informed by Kant's conception of the Critique's method, and for this reason, perhaps, does not fall neatly into either camp, nor does it straightforwardly respond to the existing debate.

¹⁰ The issue of the proper method of metaphysics is the topic of Kant's 1764 ‘Prize Essay’ (Untersuchung iiber die Deutlichkeit der Grundsdtze der naturlichen Theologie und der Moral, 2: 273–302).Google ScholarPubMed Decades later, Kant is still working on the same topic: for Kant claims that the Critique is a ‘treatise on the method’ of a future metaphysics (Bxxii).

¹¹ I discuss the synthetic method at greater length in ‘Science and the Synthetic Method of the Critique of Pure Reason’, Review of Metaphysics, 59 (2006): 517–39Google Scholar.

¹² This point about the starting point of the Critique may be connected to its multivalent aspirations. Graham Bird, in his account of the Critique as a work of descriptive metaphysics, remarks: ‘Kant's descriptive metaphysics is a descriptive metaphysics of experience that includes science; it is a descriptive metaphysics of science, including psychology, and of ordinary experience’,‘Kant and Strawson's descriptive metaphysics’, in Glock, Hans-Johann (ed.), Kant and Strawson (Oxford: Clarendon Press, 2003), p. 77.Google Scholar Although this suggests a different relation of priority between experience and scientific cognition in the Critique from the one that I shall argue for in this paper, nevertheless it accurately canvasses much of the extent of the Critique's aspirations. So it may be that the Critique would have the resources to address all of this because it takes as given only the capacity of reason. By contrast, the analytic Prolegomena would provide us with an account only of certain kinds of scientific cognition; it could not promise any descriptive-metaphysics of experience as such, as perhaps the Critique does.

¹³ Kant's letter to Christian Garve of 7 August 1783 (10: 340). Kant twice refers to the Critique as a ‘whole new science’ in the Prolegomena (4: 279 and 261–2); and, in the closing passage of the Critique, Kant refers to the whole work as pure theoretical reason's ‘scientific and fully illuminating self-knowledge’ (A849/B877).

¹⁴ P (4: 261).

¹⁵ Metaphysical Foundations of Natural Science, preface (4: 467). See also A832/B860 and Jäsche Logic (9:139).

¹⁶ How we would draw out the analogy in the case of material sciences (e.g. mathematics and physics) is complicated somewhat by the priority of Critical philosophy as a formal science assessing the capacity of reason as such. For Kant's distinction between material and formal rational cognition, see G (4: 387).

¹⁷ Some of the characteristics of Kant's general conception of scientific method - particularly its way of understanding systematicity in teleo-logical terms - figure in the two short passages that serve as general introductions to the Critique's Transcendental Analytic and Analytic of Concepts (A64-6/B89-91). See also Bxxxvii-xxxviii.

¹⁸ Kant's Transcendental Psychology, p. 16.

¹⁹ Examples of these tasks include the following: attributing self-identity and empirical apperception (20, 145), making judgements about objects (142), and perceiving spatial and temporal arrays (142, 156 ff.).

²⁰ Kant's Transcendental Psychology, p. 18.

²¹ Ibid., p. 21.

²² Hume, David, Enquiries Concerning Human Understanding and Concerning the Principles of Morals (Oxford: Clarendon Press, 1975), pp. 12–14.CrossRefGoogle Scholar

²³ The idea is that Hume would need t o claim that it is not possible to have knowledge of some proposition, while his methodological resources only entitle him t o claim that w e d o no t actually have knowledge of it. To make a claim about our necessary ignorance thus requires a complete account of our cognitive capacity; which is to say that the Transcendental Dialectic's account of the incapacity or Unvermogen of reason must follow upon the Transcendental Analytic's account of its capacity or Vermögen.

²⁴ Despite possible appearances to the contrary, Kant's idea that Critical philosophy must begin with a conception of the form of the whole does not presuppose what is to be demonstrated. For to know the shape of the earth is not yet to know its magnitude; and to have some preliminary conception of human reason is not yet to have determined the bounds of its capacity.

²⁵ The Bounds of Sense, p. 15.

²⁶ Ibid., p. 52.

²⁷ Ibid., p. 50.

²⁸ Ibid., p. 15.

²⁹ Both quotations from Ibid., p. 16.

³⁰ The first edition preface focuses on the possible status of metaphysics as a science, while the second edition preface takes a broader view, and considers a range of examples of successful scientific cognition as a platform on which to diagnose the failure of metaphysics. I will focus entirely on the second edition, giving deliberate consideration to what we are supposed to learn about reason from Kant's presentation of these examples of successful scientific cognition. In the second edition preface, Kant also points to pure general logic as a successful science; but it has a different status from mathematics and physics since it is a formal science that abstracts from all content of know-ledge, and so does not itself provide knowledge of objects of some domain, or even indicate how such knowledge is possible.

³¹ Recognition of the end of human reason would allow us to recognize the elements, or functional parts, of human reason.

³² See B147; cf. A4/B8. Kant suggests that mathematics only yields knowledge in so far as it is applicable to what can be given in empirical intuition. Thus, the proper object of mathematical cognition is nature, and mathematics is a science of nature.

³³ It encompasses considerably more because of the emphasis that is placed on scientific knowledge in Kant's articulation of the methodological framework of the Critique. Strawson overlooks this, perhaps even deliberately, as he downplays considerably the centrality of Kant's question about the possibility of synthetic a priori judging (see The Bounds of Sense, p. 43).

Incidentally, our general conception of experience shifts in Strawson's account. Initially, it is just supposed to be some general conception of experience of an objective world. But Strawson worries that the Transcendental Deduction would be question-begging if it took so rich a conception of experience as given. Thus, the Deduction begins with a more minimal conception of ‘inner experience’ (The Bounds of Sense, p. 92).

³⁴ Bird, Graham, ‘Kant's and Strawson's descriptive metaphysics’, p. 68.Google Scholar

³⁵ Ibid., p. 76.

³⁶ It is true that, in this passage, Kant only says that sensibility may contain a priori representations; he says nothing about the possibility of the understanding's containing a priori representations. The emphasis is placed on sensibility here, because it would go without saying (for Kant's eighteenth-century German reader, versed in Wolff and Baumgarten) that an account of fundamental a priori concepts would be part of a transcendental philosophy. The idea that sensibility could be constituted by certain a priori representations is a distinctively Kantian innovation, and for this reason he often points exclusively to his account of sensibility as the distinguishing feature of his transcendental idealism.

³⁷ This is to leave us with the a priori representations that we (i.e. whoever assents to the account in the front matter) suppose that sensibility must contain - i.e. ‘pure intuition and the mere form of appearances’ (A22/B36).

³⁸ To clarify: the strategy of isolation is the strategy of considering sensibility first in isolation from understanding, and then understanding in isolation from sensibility. The strategy of isolation is not the same as abstracting or isolating the formal element of representation from the material element (which Kant alludes to at B2). The distinction between sensible an d intellectual representation thus belongs to the governing presuppositions of the Critique. Perhaps Kant's appeal to it as a methodological principle lends it a status similar to that of the ‘Copernican’ hypothesis (Bxvi-xvii): the irreducible distinction between sensible and intellectual representation figures as a hypothesis, the test of which is to see if it allows us to succeed in metaphysics where others (who confuse the nature of sensible and intellectual representation, A271/B327) fail.

I do not wish, at least in the present context, to challenge the viability of Kant's strategy of isolation. Indeed, whether it is even possible to consider, in any meaningful way, each element in isolation from the other is a serious question. It is the crux of Hegel's complaints about Kant's theoretical philosophy.

³⁹ It is uncontroversial that the aim of the Transcendental Deduction is to account for the objective validity of the categories. However, the issue of whether any parallel treatment is given for space and time as the pure forms of intuition is rather controversial. Daniel Warren argues that the Metaphysical Exposition of Space (KrV§2), is concerned with the a priori origin of the representation of space; Kant is not, as Henry Allison and others have suggested, trying to establish its objective validity there (‘Kant and the apriority of space’, Philosophical Review, 107 (1998), 179-224). This leaves open the possibility that the Transcendental Expositions of space and time (KrV §3 and §5, respectively) effectively offer their transcendental deduction. Warren himself is non-committal on the issue, but seems to suggest that the demonstration of their objective validity may be postponed until the Transcendental Deduction chapter (see the final sentence of his paper, and 220 n. 53).

Kant is simply confusing on this issue. Just prior to the Transcendental Deduction, Kant claims that a transcendental deduction of space and time has already been provided; their ‘a priori objective validity’ has already been established (§13, A87/B119-120, see also A85/B118). But soon thereafter, Kant remarks that we see the need to demonstrate the objective validity of space (nothing is said about time) only once we consider the pure concepts of the under-standing, and recognize that a deduction is required for them (A88/B120). While this is not conclusive, it does suggest that there may be some problem with the idea that a transcendental deduction of space and time has already been given in the Transcendental Aesthetic.

Furthermore, the idea that a transcendental deduction of space and time has already been given in the Transcendental Aesthetic is unsatis-factory for the following methodological reasons. It should be non-controversial that at least the initial segments of the Transcendental Aesthetic are carried out under the auspices of the strategy of isolation - i.e. the Metaphysical Expositions of space and time (KrV §2 and §4). Now, if the Aesthetic is carried out entirely under the auspices of the strategy of isolation, then it never brings the understanding into view; and if this is the case, then (on the basis of the cooperation thesis) we can conclude that we could not yet have any account of sensibility as a cognitive capacity, and hence of its constitutive representations as objectively valid. But if the Aesthetic is not carried out entirely under the auspices of the strategy of isolation, then it is presumably in the Transcendental Expositions of space and time that the restrictions of that strategy are lifted (though perhaps just momentarily). Indeed, in the Transcendental Expositions, Kant endeavours to show that space and time are sources of synthetic a priori cognition (geometry and the general doctrine of motion). However, there is a problem with this, stemming from the general methodological restrictions that Kant imposes on the Critique when he distinguishes its method from the ‘analytic procedure’ of the Prolegomena: the Critique is a ‘whole new science, robbed of all help from other sciences’ (4: 279). It appears that the Transcendental Expositions rely on contraband resources; and even if they contain the transcendental deduction of space and time, their arguments must be regarded as parerga by the lights of the avowed method of the Critique.

⁴⁰ See the previous note.

⁴¹ This is because the preliminary idea of the whole is supposed to contain the conditions for determining the relation of the parts.

⁴² Thus, the Deduction begins: ‘The manifold of representations can be given in an intuition which is merely sensible, i.e. is nothing but receptivity, and the form of this intuition can lie a priori in our faculty of representation without being anything Trther than the mode in which the subject is affected’ (B129). What cannot be given through the senses is the ‘combination of a manifold’, for this is an ‘act of spontaneity’, and an ‘action of the understanding’ (B130). (Combination is deemed an Actus der Spontaneitat, then a Verstandeshandlung, and finally an Actus of the Selbsttdtigkeit of the representing subject.) Receptivity is associated with sensibility, and spontaneity is associated with the understanding.

Kant reminds the reader of the heterogeneity thesis at crucial junctures of the text: at the end of the front matter (thus introducing the Doctrine of Elements), at the outset of the Transcendental Logic, and within the Logic, at the outset of the Transcendental Deduction. Receptivity and spontaneity do not figure in the first invocation of the heterogeneity thesis at all; they are introduced in the second, where they figure in definitions of sensibility and understanding respectively (see A51/B75). And while sensibility and understanding do make an appearance in §15, particular emphasis is placed on the receptivity/ spontaneity pair.

⁴³ We are inside the framework of the Transcendental Logic, the account of the spontaneity of the mind; this is why the question concerns what it is for an intellect to have given representations, rather than what it is for a fundamentally receptive capacity to think. Having things the other way around would entail having the Transcendental Logic come before the Transcendental Aesthetic; this possibility was raised by J. S. Beck, which Kant mentions in his letter to J. H. Tieftrunk (13: 463).

⁴⁴ The cogito statement, on my reading, should not be identified with the principle of the synthetic unity of apperception, which alone introduces the crucial idea that all of a single subject's representations are subject to a necessary synthesis. This synthesis, as Kant argues only later in the Deduction, must be in accordance with the categories.

⁴⁵ ‘The highest principle of the possibility of all intuition in relation to sensibility was according to the transcendental aesthetic that all the manifold of sensibility stands under the formal conditions of space and time. The highest principle of it [sc. intuition] in relation to the understanding is: that all the manifold of intuition stands under the conditions of the original-synthetic unity of apperception’ (§17, B136).

⁴⁶ I would like to thank Markos Valaris, and two anonymous referees for this journal, for extremely helpful comments on earlier versions of this paper.