1 Prolegomena §60 IV 365.7. I cite Kant's individual works with the initials of their (German) titles. I also cite the first Critique by the usual designations of its two editions, ‘A’ and ‘B.’ Occasionally I cite Kant's works by the volume, page, and line numbers of Immanuel Kant, Kants Gesammelte Schriften, Königlich Preußische (now Deutsche) Akadernie der Wissenschaften (Berlin: G. Reimer, now De Gruyter 1902-), usually referred to as ‘Akadernie Ausgabe.’ The volume and page numbers from this edition have been carried over into all recent translations. Translations have been revised without notice.

2 The systematic character of Kant's metaphysics is analyzed by Buchdahl, Gerd in Kant and the Dynamics of Reason (London: Blackwell 1992)Google Scholar. I have criticized his positive account elsewhere (‘Buchdahl's “Phenomenological” View of Kant: A Critique,’ Kant-Studien [forthcoming]). Here I contend that ‘noumenal causality’ is a coherent notion. Buchdahl's positive account is based on its not being coherent. Thus the present essay aims to remove his interpretation's point of departure.

3 Commentar, II 52. ‘Double Affection’ was developed by Adickes, Erick (Kants Lehre von der doppelten Affektion unseres Ich als Schlüssel zu seiner Erkenntnistheorie [Tübingen: J.C.B. Mohr 1929], esp. 46–59).Google Scholar

4 See, e.g., Gram, Moltke (‘The Myth of Double Affection,’ in Werkmeister, W.H. ed., Reflections on Kant's Philosophy (Gainesville, FL: Florida State University Press 1975) 29–63Google Scholar. Another problem is that Adickes's view requires inserting a substantive sense of ‘appearance in itself’ (e.g., 49) into Kant's view, where it has no place.

5 In defending this stronger interpretation of sensory affection by things in themselves, I in part defend the position taken by Paton, Herbert J. (Kant's Metaphysics of Experience 2 vols. [London: George Allen & Unwin 1936))Google Scholar. The view I set out and defend is also very close to the main strand of Allison's, Henry view of transcendental affection (Kant's Transcendental Idealism: An Interpretation and Defense [New Haven, CT: Yale University Press 1983], ch. 11)Google Scholar. Allison recognizes most of the positive grounds I offer for my interpretation. However, his view is ambiguous, if not equivocal, for there are passages where he retreats to a ‘dual description’ view like Prauss's or Buchdahl's. I shall defend the stronger, more metaphysical strain in Allison's view, and I shall defend it against some objections he does not address. When Allison summarizes Kant's view and attempts either to avoid or to respond to criticisms of Kant's idealism, he tends to disregard the ontological aspect of Kant's forms of intuition and to reduce Kant's distinction between things in themselves and appearances to a ‘methodological’ matter of different descriptions, empirical and non-empirical, of one set of things. See, e.g., Kant's Transcendental Idealism, 240-1, 250 (last ¶); also Allison, H. ‘Transcendental Idealism: The “Two Aspects” View,’ in Ouden, Bernard den and Moen, Marcia eds., New Essays on Kant (New York: Peter Lang 1987), 170Google Scholar; and Kant's Theory of Freedom (Cambridge: Cambridge University Press 1990), 44. These ambiguities are largely resolved in his Idealism and Freedom (Cambridge: Cambridge University Press 1996), ch. 1.

6 Prol §13 Anm. II, §32 (IV 288.34-289.14, 314.33-315.6). The accuracy and reliability of the Prolegomena can be contested on points of important detail, but I shall show below that these passages do in fact accurately summarize Kant's view in the first Critique.

7 The question is this: ‘On what ground rests the relation of that in us which is called representation to the object?’ (X 130.6-8; Kant Selections, Beck, Lewis White ed. and trans. [New York: Macmillan 1988], 81)Google Scholar.

8 On Kant's sensationism see George, Rolf ‘Kant's Sensationalism,’ Synthese 47 (1981) 229–55, esp. 230, 246CrossRefGoogle Scholar.

9 By mentioning ‘synthesis,’ I have raised a larger topic than I can explore here. Kant's views on synthesis have long been castigated as part of an ‘imaginary science’ of transcendental psychology, as Strawson put it (The Bounds of Sense, 32). Paul Guyer has argued persuasively that Kant's transcendental psychology is a legitimate enterprise, resting on ‘not a psychological claim but a basic constraint on any system for synthesizing data that are only given over time’ (‘Psychology and the Transcendental Deduction,’ in Forster, Eckart ed., Kant's Transcendental Deductions [Stanford: Stanford University Press 1989] 47–68, at 65; cf. 58, 67)Google Scholar. Guyer's analysis persuaded Strawson, who admitted that his original designation was ‘rude’ (‘Sensibility, Understanding, and the Doctrine of Synthesis: Comments on Henrich and Guyer,’ in Eckart Förster, ed., Kant's Transcendental Deduction 68-77, at 74-7). Kitcher, Patricia (Kant's Transcendental Psychology [Oxford: Oxford University Press 1990])Google Scholar, Brook, Andrew (Kant and the Mind [Cambridge: Cambridge University Press 1994])CrossRefGoogle Scholar, and Howell, Robert (Kant's Transcendental Deduction [Dordrecht: Kluwer 1992])CrossRefGoogle Scholar devote sustained attention to Kant's views on synthesis.

10 See Allison, (Kant's Transcendental Idealism), 115-22, 173–94Google Scholar; Paton, I 245-8, 260-2, 304-5, II 21-4, 31-2,42-65, 68-9; and Young, J. Michael (‘Functions of Thought and the Synthesis of Intuitions,’ in Guyer, Paul ed:, The Cambridge Companion to Kant [Cambridge: Cambridge University Press 1992] 101–22), 112-13CrossRefGoogle Scholar. Young's analysis makes plain both that there are two components to the meaning or the significance of categories, and that Kant is willing to speak, unqualifiedly though imprecisely, of concepts being ‘without sense … [or] meaning’ when not tied to sensory intuitions.

11 This is how Kant puts his point at 8150-51, III 119.4-24 and A244-24, IV 160.21-161.17. Robinson, Hoke (‘Two Perspectives on Kant's Appearances and Things in Themselves,’ journal of the History of Philosophy 32 [1994] 411–41)CrossRefGoogle Scholar proposes another view of the relation between appearances and things in themselves, in terms of two perspectives (God's and ours) on things, and according to which we project empirical objects. His positive proposal unfortunately remains very sketchy. For discussion see Allison, Idealism and Freedom, 12-16.

12 A ‘representational’ theory of perception, in the relevant sense, holds that the direct objects of our perception are mental representations, which indirectly represent the objects which (allegedly) cause them. Although the view goes back to the Stoics (at least; see Sextus Empiricus [Outlines of Pyrrhonism, Works Vol. 1. Rev. Bury, RG. trans. (Cambridge, MA: Harvard University Press 1933), II §72-75])Google Scholar, Locke is now the classic example.

13 Indeed, Buchdahl recognizes that he can only remove the support this passage would provide for a causal account of transcendental affection by reinterpreting Kant's apparently causal terminology in non-causal ways (73, 154). Buchdahl recognizes a wide array of passages in which Kant describes transcendental affection in causal terms (70-1, 129-31, 136, 153-4, 161, 164). Rather than attributing them to Kant's confusion about his own method (as Prauss did), Buchdahl seeks to retain those passages by showing that the apparently causal terms Kant uses in those passages(‘ground,’ ‘affect,’ and ‘determine’) do not necessarily carry causal connotations. Buchdahl is right that Kant occasionally uses the term ‘affect’ in a noncausal, purely logical sense, e.g., when Kant speaks of understanding affecting inner sense (B154, A152/B191, A77 /B102, B155, B191-92, A555/B583; Buchdahl, 139, 160-1). This does not, however, entail that Kant uses the terms ‘affect,’ ‘ground,’ or ‘determine’ in a non-causal sense when speaking of the stimulation of our sensibility by things in themselves.

14 Kant provides a concise taxonomy of kinds of representation at A320=B376-77, III 249.37-250.14. In support of his view that Kant's discussion of transcendental affection of sensibility is simply (what Buchdahl calls) a ‘phenomenology’ of the fact that we happen to have certain sensations, Buchdahl repeatedly cites Kant's discussion in the Transcendental Aesthetic of the subjective basis of sensory qualities, viz., the idea that the ‘matter’ of sensation is a function of our subjective constitution (B44; Buchdahl, 118, 142, 158, 161). Buchdahl takes Kant's view to be that the sensory qualities we happen to find ourselves with are due exclusively to our own nature as sentient beings, and he takes this passage to concern transcendental affection. However, the fuller discussion in the first edition makes quite plain what is reasonably clear from the shorter discussion in the second edition from which Buchdahl quotes, namely, this passage concerns the ascription of secondary qualities to objects regarded as appearances (see A28-39, IV 34.35-35.8, regarding secondary qualities). Kant insists that there is nothing either necessary or a priori about the particular sensory qualities he mentions, and so there is nothing transcendentally ‘ideal’ about their analysis. This is in express contrast to space (and time), which though ‘subjective’ are necessary a priori conditions of our experience of objects. Because this passage concerns even those sensory affections which count as subjective, ‘secondary’ qualities, this passage in fact reiterates that the matter of sensation is not simply a function of our subjective constitution. It is, pace Buchdahl, also a function of the source of sensory affection (cf. Paton, I 139-40; see §IX below about the metaphysical status of empirical objects).

15 It has been suggested that Kant misstated his view by proscribing a ‘transcendental,’ rather than a ‘transcendent,’ use of the categories. While Kant might have made his meaning plainer to subsequent readers in this way, his use of the term ‘transcendental’ in this connection is drawn from the metaphysical tradition's concern with the ‘transcendental’ categories of ultimate reality and the attempt to use such transcendental metaphysical categories to prove the existence, e.g., of a divine first cause. Kant denies there is any such legitimate use of the categories, and he used the right term to make his meaning clear to traditional metaphysicians. As he says in connection with the distinction between phenomena and noumena, ‘The transcendental use of a concept in any sort of principle is this: that it is used in connection with [auf … bezogen wird] things in general and in themselves’ whereas the empirical use of a concept is in connection ‘merely’ with ‘appearances, that is objects of possible experience’ (A238=B298 III 204.10-14).

16 To be sure, he does say more about it in connection with his immediate topic, the Amphiboly of the Concepts of Reflection, but its range is much wider than that, since it concerns any and all judgments a priori about things. Prauss (75, n. 15) notes that Kant alludes to it in the Second Analogy (A190=B236, III 168.33-34) and in the resolution of the Third Antinomy (A295=B573, III 370.17).

17 For discussion of Kant's view of transcendental reflection, see Schnädelbach, Herbert (Reflexion und Discurs: Fragen einer Logik der Philosophie [Frankfurt am Main: Suhrkamp 1977]), 87–133Google Scholar. He points out that transcendental reflection is the very method Kant employs in the Critique of Pure Reason. Also see Dieter Henrich, ‘Kant's Notion of a Deduction and the Methodological Background of the First Critique,’ in E. Forster, ed., Kant's Transcendental Deductions, 29-46, at 40-6. Paton, Prauss, Buchdahl, and Allison recognize the importance of transcendental reflection to Kant's undertaking, though they do not investigate Kant's account of it in any detail. Paton refers to ‘transcendental knowledge’ (I 226-32), Allison refers to it only in passing (Kant's Transcendental Idealism, 67, 143-4,241,243, 275; d. 268), while Buchdahl (74, 79-83, 114) and Prauss (66-85, 213f.) develop their own versions of transcendental reflection.

18 A155-56=B195, III 144.17-20; the two marginal comments at issue were written at the beginning of Kant's chapter on the distinction of objects into phenomena and noumena (A235=B294-95). They are Reflexionen Nos. CIV and CVI from the Selbständige Reflexionen im Handexemplar der KdrV (A) (XXIII 34.14-17, 34.27). Kant's emendations are reproduced in the same volume of the Akademie edition, 46.18 (re: A147, IV 104.30), 48.14 (re: A247, IV 162.5-6), 48.16-17 (re: A247, IV 162.13-14), 48.25 (re: A251, IV 164.6), 49.17 (re: A259, IV 168.34-35), and 49.23 (re: A286, IV 183.36). Note, too, that Kant's claim that noumena can only be thought but not known, is made within and holds of the theoretical perspective.

19 Buchdahl's blanket way of speaking of the ‘activation’ of sensibility, understanding, and the categories tends to obscure Kant's important distinction between the passivity of sensibility and the spontaneity of thought. Though occasionally he remarks on Kant's distinction between the passivity of sensation and the activity of thought (128, 148, 162), typically he lumps them together and stresses the need for these conditions to be ‘activated’ (68, 82, 102 n. 33, 115, 155, 156, 285, 322-4).

20 Much havoc has been inadvertently wrought in the literature by uncritically assuming that any Kantian use of concepts must count as application of those concepts, where application is a matter of subsuming intuitions of particulars under schematized categories. Straw son made this mistake, if implicitly in his assertion of Kant's alleged principle of significance. Most recently it is made explicitly by Patt, Walter (‘“Things in Themselves” and “Appearances,”’ in Funke, Gerhard ed., Akten des 7. internationalen Kant-Kongress Vol. II.l [Bonn: Bouvier 1991] 149–57), 151, 152Google Scholar. Also see, e.g., Butts, Robert E. (‘The Methodological Structure of Kant's Metaphysics of Science,’ in Butts, Robert E. ed., Kant's Philosophy of Physical Science [Dordrecht: Reidel 1986]), 165CrossRefGoogle Scholar, who states this as point 4 of the supposed ‘Central Tenets of Kant's Programme’; his later puzzlement about whether moral agents are just useful fictions is an ineluctable result of this mistake (175 n. 6). Butts disregards Kant's view that practical reason has primacy over theoretical reason.

21 In connection with the distinction between phenomena and noumena, and in support of his general thesis, that the categories and the principles defined in their terms can only be given ‘real’ definitions by specifying the conditions under which objects can be given to which those concepts and principles can be applied, Kant says the following about the category of causality: ‘If I omit time, in which something follows on something else according to a rule, I would find nothing more of the concept of cause in the pure category than that it would be something on the basis of which the existence of something else can be inferred. However, on that basis not only could cause and effect not be distinguished, but also because this capacity to infer [Schließenkönnen] of course at once requires conditions about which I know nothing, the concept would have absolutely no determination for it application to any object’ (A243=B301, III 206.10-17). Notice that Kant does not say here that, abstracting from time, there would be no difference between cause and effect; he does not retract his asymmetrical characterization of the category of causality as ‘The relation of causality and dependence (cause and effect)’ (A80=B106, III 93.8-12). He only says that the inference from one thing to the existence of another would be symmetrical. That is, we could as well infer the existence of the cause from the effect, if only we could identify either one of them.

22 A383, IV 240.1-3; 8419-20,421, III 274.9-15, 274.36-275.3; KdU §89, V 460.20-32; see Westphal, Kenneth R. ‘Kant's Critique of Determinism in Empirical Psychology,’ in Robinson, Hoke ed., Proceedings of the 8th International Kant Congress Vol. II pt. 1 (Milwaukee: Marquette University Press 1995) 357–70Google Scholar. The view I urge here is quite close, if not identical, to that advocated by Rescher, Nicholas (‘On the Status of “Things in Themselves,”’ Sythese 47 [1981] 289–300)CrossRefGoogle Scholar, though I have tried to give more substance to Kant's transcendental basis for thinking of the sensuous manifold ‘as the product of a mind-external reality that somehow impinges upon our mind ab extra’ (Rescher, 293). My view is also very close to Thompson, Manley (‘Things in Themselves,’ Proceedings and Addresses of the American Philosophical Association 57 [1983] 33–48)CrossRefGoogle Scholar, except that I give more credence to the idea that at the transcendental level there is a role in Kant's theoretical philosophy for regarding things in themselves as causes of our sensory manifold. In part this is because the location of an effect is determined by the location of what is acted upon, not necessarily the erstwhile location of the cause (contra Thompson, 42). This can be admitted while agreeing with Thompson that things in themselves play no role for Kant in explaining particular experiences (43). (Thompson's insistence that a cause must be spatially located in order to be effective at that location is an instance of transcendental illusion; see below, 231.) Finally, my view is similar to Baldner's, Kent (‘Causality and Things in Themselves,’ Sythese 77 [1988] 353–73)CrossRefGoogle Scholar. However, I give more credence to the genuinely causal connotations of Kant's locutions about things in themselves as causes of appearances, in part because my case is based less on a general view of intentionality and more on the specifics of Kant's own transcendental reflection on sensibility. Also, I have misgivings about Baldner's equating things in themselves with the intentional objects of our experience, but this cannot be gone into here.

23 The idea that Kant believes there are two realms of objects, noumenal and phenomenal, where noumenal objects (somehow) cause phenomenal objects, has been dubbed a ‘two worlds’ view. I am as opposed to this interpretation of Kant as are Prauss, Buchdahl, and Allison. However, as I point out below, this is because ‘phenomenal objects’ are constructs, whereas noumena are real (see §IX).

24 Wood, Allen (‘Kant's Compatibilism,’ in Wood, Allen ed., Self and Nature in Kant's Philosophy [Ithaca: Cornell University Press 1984] 57–101)Google Scholar presents the most courageous and adequate attempt yet to make sense of the a-temporal causality apparently required by noumenal causality. In reply, Bennett, Jonathan (‘Kant's Theory of Freedom,’ in Wood, A. ed., Self and Nature in Kant's Philosophy, 102-12, at 102)Google Scholar contends that a ‘making to begin’ that is not itself a ‘happening’ is a contradiction. Cf. Howell, 56.

25 Kant is explicit that there is a ‘transcendental’ sense of things being ‘outside use; according to which those things are things in themselves that are distinct from us, though this involves no spatial determinations (A373, IV 234.21-23).

26 A297=B353, III 236.19-29; A396, IV 247.24-25. For discussion of transcendental illusion, see Grier, Michelle ‘illusion and Fallacy in Kant's First Antinomy’ KantStudien 84 (1993) 257–82Google Scholar. Allison notes that Kant's remark about a supposed object of a non-sensible intuition, whose ‘duration is not a time’ (B149 [III 118.23]) suggests ‘that Kant might not rule out all noumenal analogues to our sensible forms’ (Idealism and Freedom, 185 n. 18).

27 The most subtle and insightful analysis known to me of how Kant's view of space and time as a priori subjective forms of human intuition goes awry is in Sellars, Wilfrid Science and Metaphysics (London: Routledge & Kegan Paul 1968), 230–8.Google Scholar

28 Buchdahl responds to Strawson's objection somewhat differently (78-9).

29 Cf. A30=B45, B306, B307, B308-09, A254-55=B310, A287-88=B344-45, A491-92=B520- 21, A492-94=B521-22.

30 Recognizing that Kant relies on two ranges or kinds of properties of things does not, however, require ascribing to him a ‘two world’ ontology, which invites the fabrication of the doctrine of double affection.

31 Per Kant's remark about the given appearance of a house being an ‘Inbegriff’ of representations of apprehension. Robert Howell (38-40) and Hoke Robinson (419- 20) cite an array of passages in which Kant appears to identify appearances with representations. They hold that these passages are counter-evidence to Allison's dual aspect interpretation of Kant's distinction between appearances and things in themselves, and Howell contends that they provide evidence for an appearance theory and against the appearing theory of this distinction, which would include the interpretation developed here. These issues are complicated; they cannot — and need not — be fully disentangled here. It suffices for my purposes to show that an appearing theory provides a basis for a coherent interpretation of noumenal causality. If that is possible, that provides grounds for emphasizing those passages which favor an appearing theory. However, something much stronger can be said: If one keeps Kant's sensationism in mind, then it is clear that any ‘representations’ which can be ‘appearances’ of empirical objects or events must be generated by conceptually synthesizing complexes of sensations. The objects of such representations appear to us through those complexes. This entails that appearances and representations of appearances have a mind-dependent intentional existence and it entails that they are functions of the contents (the ‘objective reality’) of the concepts and sensations which generate them. (Kant's use of the term‘Inbegriff’ quoted above may well only refer to the content of the complex of concepts and sensations involved in representing the appearance of the house, and not to a numerical identity of the appearance of a house and the sensations and concepts which generate it.) This provides representations and appearances all the ‘mind-dependence’ Kant ascribes to them in the passages cited by Howell and Robinson. However, this does not require that what we are aware of as objects are our sensations or ‘ideas’ (in Locke's or Berkeley's senses) or complexes thereof. I think Kant may be guilty of hasty and misleading expression, but not of inconsistency or (sub specie Transcendental Idealism) falsehood. Also see Allison's response to Robinson on this count (Idealism and Freedom, 12-14).

32 This passage should counter any temptation to read the last line of the above quote from the second Analogy differently. In this essay I defend what Howell, following Prichard, calls an ‘appearing theory’ of Kantian objects of empirical knowledge (Howell, 37-52). Howell argues that this interpretation faces an insuperable dilemma. I agree with Howell in taking the metaphysical aspects of Kant's idealism seriously, though I believe he has not taken them seriously enough: Recognizing the radical implications of Kant's doctrines of space and time as forms of intuition provides grounds for restricting some of the premises in Howell's objection in ways which dispel his dilemma. Howell argues that an appearing theory cannot consistently maintain that a thing in itself is identical with a thing as it appears to us in space and time, while also maintaining that things as they exist in themselves are nonspatiotemporal and unknown by us. Howell takes the example of someone (H) with forms of sensibility and judgment like ours who knows that a tree is conical. Kant accepts the judgmental or propositional form our kind of knowledge of the tree takes:

33 This is not to ascribe phenomenalism to Kant. ‘Phenomenalism’ requires the (essentially Cartesian) thesis that each ‘phenomenon’ (‘sensing strictly speaking,’ ‘idea,’ ‘sensory impression,’ ‘sense-datum,’ etc.) be exactly what it seems or appears to be. On Kant's view, only objects of judgments can be objects of consciousness. Kant's view doesn't require any sort of infallibility or incorrigibility, though it does require that if we're aware of an object at all, our judgment about it cannot be wholely false. (Cf. Kant's logic lectures: ‘All judgments accord with the laws of the understanding, all judgments of the understanding are thus true. Even in our mistaken judgments the understanding must always have done something, and thus it can never happen that everything in the judgment is false; instead there must always be something true in it. It is completely impossible that a person completely errs, if he judges’ (Logik Blomberg XXIV 84.25-31, my trans.].)

34 Schrader recognizes that Kant's theoretical and practical philosophies must be integrated (181-5). Unfortunately, his effort in this regard amounts to little more than insisting that Kant's locutions in his practical works must conform to Schrader's neo-Kantian view of Kant's theoretical philosophy.

35 Allison, Kant's Theory of Freedom, 39, 52, 55, 65, 241. I would, however, locate the needed revision somewhat differently. Kant himself argues against the possibility of deterministic psychology. Thus there is no problem granting Allison's ‘incorporation thesis.’ The problem lies instead in understanding how free (or, not known to be determined) psychological acts are to be coordinated with, and understood as causing, otherwise determined bodily behavior in space and time (see Westphal, ‘Kant's Critique of Determinism in Empirical Psychology’).

36 Kritik der praktischen Vernunft (hereafter ‘KdpV’) V, 95-97. Ameriks, Karl (‘Kant and Hegel on Freedom: Two New Interpretations,’ Inquiry 35 [1992] 219–32)CrossRefGoogle Scholar urges a revisionist, compatibilist reconstruction of Kant's views on freedom. He believes that Kant's noumenal metaphysics is coherent (219), but finds that Kant's moral theory doesn't sufficiently support recourse to incompatibilist noumenal freedom (227) and that relaxing causal determinism in the phenomenal realm would ‘cause a fundamental revision of Kant's transcendental account of experience’ (229). (Ameriks rejects Allison's claim that practical freedom involves indeterminism at the psychological level.) Against Ameriks, I suggest that incompatibilist freedom is equally fundamental to Kant's practical views, and none of Kant's theoretical arguments suffice to prove strict universal causal determinism, either. Consequently, revising Kant's belief in strict determinism needn't involve revising much if anything (apart from rhetoric) in his transcendental account of experience. Ameriks has yet to explain how his proposed (but still undisclosed) ‘sophisticated compatibilism’ is to deal with Kant's sharp argument in the second Critique against compatibilism überhaupt. In any event, unlike some other recent attempts at interpreting Kant's metaphysical views about freedom of the will in compatibilist terms, Ameriks' attempt clearly recognizes that such efforts involve reconstructing Kant's views, and not simply interpreting them. Most prominent among the recent compatibilist interpretations of Kant is Ralf Meerbote's work. Though I find this idea, as an interpretation of Kant, unconvincing, I cannot undertake its detailed criticism here. For discussion, see Allison, Kant's Theory of Freedom, 76-82.

37 Tugendlehre VI 433 note; Letter to Moses Mendelssohn of April 8, 1766 (X 69; Zweig, A. ed. and trans., Kant: Philosophical Correspondence [Chicago: University of Chicago Press 1967], 54)Google Scholar; and Kant's Note given in XII 380 (1^st ed., 406).

38 In the third section of the ‘Religionslehre Pölitz’ (XXVIII 2,2, pp. 1091-1117; Wood, A.W. trans., Lectures on Philosophical Theology [Ithaca: Cornell University Press 1978], 131–159)Google Scholar. I thank Karl Ameriks for mentioning them to me in this connection.

39 I stress replacement; cf. the passages cited from the Prolegomena above, n. 6. Leibniz is surely among the ‘idealists’ from whose view Kant there distinguishes his. Adickes notes in passing that Kant's view has a monadological character (47), though he assumes that all things in themselves must be ‘spiritual.’ This is an overstatement. Just as there is no need to assume that every thing in itself or noumenon is free (seen. 45), there is also no reason to assume that they are all ‘spiritual’; at least some could be other sorts of abstracta. Adickes also greatly overestimates the admissible degree of mapping between phenomenal appearances to us and things in themselves (ibid.; cf. Howell, 56-7). On the Leibnizian background to Kant's Critical epistemology, see Cassierer, Ernst Kant's Life and Thought, Hagen, J. trans. (New Haven: Yale University Press 1981)Google Scholar, ch. II §4; Martin, Gottfried Kant's Metaphysics and Theory of Science, Lucas, Peter G. trans. (Manchester: Manchester University Press 1955)Google Scholar, Introduction and ch. 1; Buroker, Jill Vance Space and Incongruence (Dordrecht: Reidel 1981), esp. chs. 2 and 5CrossRefGoogle Scholar; and the remarks on Leibniz in Ameriks, Karl Kant's Theory of Mind (Oxford: Clarendon 1982)Google Scholar, and Laywine, Alison Kant's Early Metaphysics and the Origins of the Critical Philosophy (Atascadero, CA: Ridgeview 1993).Google Scholar

40 These two senses of ‘idealism’ and the exploitation of the resulting ambiguity by Leibnizians are discussed by Smith, Norman Kemp A Commentary to Kant's ‘Critique of Pure Reason’ (London: Macmillan 1923), 298–9Google Scholar, who follows Vaihinger, Hans ‘Zu Kants Widerlegung des Idealismus,’ in Strassburger Abhandlungen zur Philosophie: Eduard Zeller zu seinem siebenzigsten Geburtstage (Freiburg LB. and Tübingen: J.C.B. Mohr 1884) 85–164, at 107-11Google Scholar. I do not think that Kant's refutations of idealism are contradictory in the ways Vaihinger contends, but that is a topic requiring separate treatment.

41 A concept is ‘problematic’ if no theoretical grounds can be given to determine whether an object corresponds to it (A254-55=B310-ll, A286-88=B342-44, A771- 72=B799-800).

42 For discussion of Kant's theory of character (Gesinnung), see Allison, Kant's Theory of Freedom, 136-45Google Scholar. However, he treats ‘character’ as constitutive, whereas it is quite clearly a regulative construct (see Westphal, ‘Kant's Critique of Determinism in Empirical Psychology’).

43 KdrV A546-7=B574-5, III 370.33-371.14; KdU ‘Allgemeine Anmerkung zur Teleologie,’ V 484.7-19.

44 I do not say that Kant analyzes empirical objects in phenomenalist terms. Seen. 33 above.

45 Allison rightly points out (against Beck) that Kant's account of the noumenal ground of phenomena does not entail that every phenomenon is transcendentally free; a noumenal ground is only a necessary, not a sufficient, condition of transcendental freedom. Transcendental freedom is only ascribed on the basis of actions that can be understood only through the causality of reason (Kant's Theory of Freedom, 73-4); cf. A545=B573.

46 B xxi, B xxix-xxx, A633-34=B661-62, A640-41=B667-68, A776-77=B804-05, A804- 31=B832-59 (The Canon of Pure Reason); KdpVV 50-57.

47 I gratefully acknowledge that the initial research on this article was supported by an annual fellowship from the National Endowment for the Humanities (USA, 1992) and that revisions were made during a research fellowship from the Alexander von Humboldt-Stiftung (Germany, 1995). A preliminary draft was presented to the North New England Philosophy Association (1992). This paper has benefited from the comments of two anonymous referees for this journal.