¹ See my ‘Self-legislation in Kant's moral philosophy’, Archiv für Geschichte der Philosophie, 86 (2004), 257–306.Google Scholar

² For a sampling of Kant's attitudes about religious enthusiasm and religious observances, see Religion 6: 172-5,194–202; and a few of Kant's letters 10: 175-80, 191-A. See also Kühn, Manfred, Kant: A Biography (Cambridge: Cambridge University Press, 2001)CrossRefGoogle Scholar.

³ Larmore, Charles, ‘Objectively good’, review of The Invention of AutonomyGoogle Scholar, by Schneewind, J. B., The New Republic, 219 (28 September 1998), 42–5.Google ScholarHare, R. M., ‘The simple believer’, in Essays on Religion and Education (Oxford: Clarendon, 1992), p. 30Google Scholar ; and Sorting Out Ethics (Oxford: Clarendon Press, 1997), p. 27Google Scholar.

⁴ Frederick C. Beiser, unpublished commentary on Lara Denis's ‘Kant's criticism of atheism’, APA Pacific Division Meetings, 30 March 2001. Beiser claims that Kant advocated, for political reasons, a moral religion which he considered to be inconsistent with core principles of the critical philosophy articulated in the first Critique. See also Enlightenment, Revolution, and Romanticism (Cambridge, MA: Harvard University Press, 1992)Google Scholar, ch. 2. On similar grounds, Yovel has concluded that, according to Kant, ‘the postulate of the existence of God tells us nothing of God, only of man and the world, and the Kantian theory of the Deity remains strictly humanistic’, political and historical. Yovel, Yirmiahu, Kant and the Philosophy of History (Princeton: Princeton University Press, 1980), esp. pp. 118–21, 275-7.Google Scholar See also n. 15 below.

⁵ Wood, Allen, ‘Religion, ethical community and the struggle against evil’, Faith and Philosophy, 17 (2000), 498–511CrossRefGoogle Scholar (hereafter ‘REC’) ; and Hare, John E., ‘Kant on recognizing our duties as divine commands’, Faith and Philosophy, 17 (2000), 459–78CrossRefGoogle Scholar (hereafter ‘KRD’). See also , Wood'sKant's Ethical Thought (Cambridge: Cambridge University Press, 1999)Google Scholar (hereafter KET) and ‘Rational theology, moral faith, and religion’, in The Cambridge Companion to Kant, ed. Guyer, Paul (Cambridge: Cambridge University Press, 1992), pp. 394–416CrossRefGoogle Scholar (hereafter ‘RTM’) ; Hare, John E., God's Call: Moral Realism, God's Commands and Human Autonomy (Amsterdam: Eerdmans, 2001)Google Scholar (hereafter GC) and The Moral Gap: Kantian Ethics, Human Limits, and God's Assistance (Oxford: Clarendon Press, 1996)Google Scholar (hereafter MG).

⁶ Despite some significant difference in their perspectives, Hare does seem sympathetic to some of Wood's social emphases (see n. 45 below) and appreciation for aspects of the Enlightenment and Wood seems sympathetic to Hare's rejections of constructivist ethical anti-realism. KET, 157, 374-5, nn. 4-5.

⁷ There is also a substantial disagreement between Wood and Hare about the role of revelation in Kant's religion and moral theism which I do not take up here. See , Wood, ‘Kant's deism’, in Rossi, P. J. and Wreen, M. (eds), Kant's Philosophy of Religion Reconsidered (Bloomington: Indiana University Press, 1991), pp. 1–21Google Scholar ; , Hare, MG, 38–53Google Scholar ; cf. Quinn, Philip L., ‘Kantian philosophical ecclesiology’, Faith and Philosophy, 17 (2000), 512–34CrossRefGoogle Scholar ; and O'Brien, Kelli S., ‘Kant and Swinburne on revelation’, Faith and Philosophy, 17 (2000), 535–57CrossRefGoogle Scholar.

⁸ See , Hare, GC, 89Google Scholar, 94, 96, 102; 105; cf. ‘KRD’, 465-6 ; , Wood, KET, 410Google Scholar n. 42, ‘REC’, 500, and Kant's Moral Religion (Cornell University Press, 1970) (hereafter KMR), pp. 26-7, 188Google Scholar . In a footnote to KMR, 188, Wood cites many passages, the least ambiguous of which is from the Religion (R, 6: 3). Wood says ‘it is not God's will, but only our own, which is conceived as the author of the moral law (the one whose will makes obedience to it obligatory.)’ (KET, 410). I have argued elsewhere that Kant insists that there is no author of the moral law in the strict sense, it has only legislators and/or ‘authors of obligation in accordance with’ it. See n. 1 above.

⁹ See also ‘Pölitz’, 28: 1002, 1141, 1241.

¹⁰ On certain ‘verificationist’ readings of Kant's theories of meaning and empirical knowledge, Kant's talk of ‘God's existence’ (die Existent or das Dasein Gottes) may seem illegitimate or without meaning. An adequate treatment of this important issue is beyond the scope of the present article. For present purposes, I simply note that I am following Kant's widespread usage throughout the Critical period, and that this provides some reason o t believe that an accommodation with his intended theories of meaning and cognition may be found. For further discussion, see Kühn, Manfred, ‘Kant's transcendental deduction of God's existence as a postulate of pure practical reason’, Kant Studien, 76 (1985), 152–69Google Scholar and Adams, Robert M., ‘Things in themselves’, Philosophy and Phenomenological Research, 57 (1997), 801–25.CrossRefGoogle Scholar In section 2, we will consider a separate argument, offered by Wood, against the claim that God's existence is postulated.

¹¹ Kant does not always clearly distinguish between these various kinds of priority. Especially in connection with the epistemic and ontological priority claims, this is attributable to the fact that he is concerned with a systematic account of knowledge. Of course, we could distinguish the different sorts of priority claim and identify several other possible positions on the relationship between theology and moral philosophy including positions that deny that either is prior in one respect or the other.

¹² On Kant's theory, God's existence does not give rise to any particular duties (cf. the theory of Christian Wolff). Kant argues that we have no duties to God, and that our only ethical duty with regard to God (to view our duties as God's commands) is a duty of a human to himself; cf. R, 6: 155 n.; MdS, 6: 486ff. We should view our duties as divine commands (which seems to presuppose God's existence) but this is supposed to be a rational commitment not a moral duty.

¹³ This is the second point made in the passage from the Religion's preface quoted above. It is perhaps not until the second Critique (1788) that Kant clearly endorses the claim that the moral law itself provides a sufficient, and the only morally worthy, motive (respect). In the ‘Praktische Philosophic Herder’ lecture notes from the early 1760s, Kant indicates that theology is needed to supply a special motive, critical for morality, a view that seems ot be preserved even in the ‘Moral Mrongovius II’ lecture notes, contemporaneous with the publication of the Groundwork.

¹⁴ Part of Kant's basis for this assumption rests upon the fact that we are finite rational beings who have needs. An extensive analysis of Kant's moral argument can be found in Wood, KMR. See also Hare, MG, ch. 3, pp. 69-96.

¹⁵ Kant's ‘moral argument’ occurs in a number of different variations in the corpus, sometimes a matter of separate, but compatible versions, other times changing in response to criticisms or in light of various other developments in his thought (for example, with respect to the nature of moral motivation). Yet, as Guyer has recently argued, the basic structure of the argument remains quite consistent. Guyer, Paul E., ‘From a practical point of view’ (hereafter ‘FPP’) in Kant on Freedom, Law, and Happiness (Cambridge: Cambridge University Press, 2000), pp. 333–71CrossRefGoogle Scholar ; at 337, 355. Many commentators have distinguished between a ‘worldly’, for example, social and immanent conception of the highest good and a ‘religious’, for example, transcendent and individual conception. One may be tempted to think that these two conceptions of the highest good are mutually exclusive and that the worldly conception came to supersede the religious conception in Kant's thought, perhaps in the third Critique. But the evidence suggests that both aspects of the highest good can be found in Kant's thought from the 1760s well into the 1790s and that Kant thought that the worldly conception includes and comprehends the religious conception, rather than superseding or rejecting it. See, for example, , Yovel, KPH, pp. 32Google Scholar, 72, 81-2,110; Velkley, Richard, Freedom and the End of Reason (Chicago: University of Chicago Press, 1989), esp. p. 153Google Scholar ; and Pogge, Thomas, ‘Kant on ends and the meaning of life’, in Reath, A., Herman, B., and Korsgaard, C., eds. Reclaiming the History of Ethics (Cambridge: Cambridge University Press, 1997).Google Scholar The Opus postumum is a different matter, which I briefly discuss at the end of section 2. For a defence of the claim that both are integral and complementary aspects of Kant's mature theory, see Mariña, Jacqueline, ‘Making sense of the highest good’, Kant-Studien, 91 (2000), 329–35CrossRefGoogle Scholar.

¹⁶ In KMR, Wood argues that the moral argument leads to or reveals ‘an unwelcome conclusion about the person himself as a moral agent’, that the agent is logically committed to abandoning the pursuit of the highest good and thus to becoming a scoundrel. This is what he calls the ‘absurdum practicum’ argument. Wood also seems to hold that these considerations are not best taken to lead to a ‘theoretical error’ or ‘any inconsistency in judgments’. If these considerations did lead to an inconsistency in judge-ments, Wood argues, then the argument of the Analytic of the second Critique that the moral law is valid must be ‘incomplete’, which would be inconsistent with Kant's apparent claims to the contrary. KMR, 25-30. I believe Kant thought both conclusions followed from these considerations and that the ‘practical absurdity’ itself typically emerges from, or is associated with, a conflict of judgements. As for the alleged ‘incompleteness’ of the argument of the Analytic, see n. 29 below.

¹⁷ , Wood, KMR, 132–3Google Scholar . God's existence seems to be a ‘physical or metaphysical condition … of the possibility of the highest good’ (KpV, 5: 143). Our inability to rationally endorse another condition may itself be due to limitations of our finite nature (KpV, 5: 145). Of course, Kant recognizes that if one mistakenly believes that virtue is its own reward, he need not presuppose the existence of God in order to intend or promote the highest good (Refl. 6287, 18: 557). But since Kant thinks that such a belief is untenable, he does not consider this to be a rational way out. Why does Kant think the possibility of the highest good requires the actual existence (as opposed to the mere possibility) of God? As Guyer has helpfully explained, the answer lies in Kant's conception of real, as opposed to merely logical, possibility (‘FPP’, 345-50). As Hare emphasizes, the possibility of a ‘kingdom of ends’ makes more specific demands upon the executive and judicial functions that God must discharge, see n. 45 below.

¹⁸ Kant's mature position is that these rewards are not a necessary motive or incentive for morality and that a good will's dutiful actions are not motivated by the happiness she hopes for, though this happiness is necessarily part of her end. (Of course an impure will might be motivated by these rewards, not realizing that God rewards moral worth, not mere legality.) Even before Kant articulates respect for the moral law as the incentive of pure practical reason, that is, when he still sees religion as a moral motive, he denies that rewards are the basis of this religious motive. On the motive/end distinction see , Wood, KMR, 51ffGoogle Scholar . 167; , Hare, MG, 75ffGoogle Scholar.

¹⁹ I think there are two interrelated, yet distinguishable claims here: first, our acceptance of the moral law as a divine command is necessary to guarantee the internal coherence of our moral maxims; second, without the existence of this God (or a functional equivalent), practical reason's apparent demands would in fact be illusory and/or incoherent.

²⁰ On the developmental relation between the Groundwork's ‘deduction’ and the ‘fact of reason’, in the second Critique, see Ameriks, Karl, Kant's Theory of Mind (Oxford: Clarendon Press, 1982/2000), ch 6Google Scholar ; and ‘Pure reason of itself alone suffices to determine the will, in Höffe, Otfried (ed.), Imtnanuel Kant: Kritik der praktischen Vernunft (Klassiker Auslegen, 26; Berlin: Akademie Verlag, 2002), pp. 99–114CrossRefGoogle Scholar.

²¹ But cf. ‘Pölitz’, 28: 1010.

²² See KrV, A633/B661; KpV, 5: 122, 124-6, 132; ‘Pölitz’, 28: 1083.

²³ For a gloss on the meaning of ‘not as necessary’, see the end of this section and n.29 below.

²⁴ Moreover, how could it still command (‘Collins’, 27: 313) if it would lack the ability to command (KrV, A811/B839)? Kant's terminology on this point is neither uniform nor transparent, but if his position is not incoherent, he must be employing different senses of ‘command’, ‘validity’ and ‘worth’. What is needed, and offered in what follows, is an explanation for and interpretation of Kant's repeated assertion of such apparently contradictory claims, often in the same text. The persistence of this conflict seems to preclude a simple developmental explanation of it.

²⁵ That is, to believe in the existence (or possibility) of God (or to deny one of the premises of the argument other than those concerning the authority of the moral law and its demand that one promote the highest good).

²⁶ Just as one cannot know in advance that there is a ‘way out’ of the antinomies of theoretical reason (KrV, A421/B449). Scepticism is avoidable only if there is a rational way out (A507/B535).

²⁷ In this case, why the conflicting claims about it ‘commanding’ yet lacking authority and why the claims that respect for it is weakened, not obliterated? Because Kant thinks that we still could not reflectively avoid judging ourselves in the light of the law and admiring actions (especially the possible actions of beings who can act rationally in response to it) done out of rational respect for it.

²⁸ Even if the atheist's denial of (or agnosticism about) God's existence is not itself completely rational, it may constitute a defeater for his belief in the reality of moral obligation. For the claim that an irrational belief can constitute a defeater, see Plantinga, Alvin, Warranted Christian Belief (Oxford: Oxford University Press, 2000), pp. 364–6CrossRefGoogle Scholar ; and ‘Naturalism defeated’, unpublished manuscript, 1994 ; cf. Alston, William P., ‘Plantinga, naturalism, and defeat’, and Plantinga, Alvin, ‘Reply to Beilby's Cohorts’, in Beilby, J. (ed.), Naturalism Defeated? Essays on Plantinga's Evolutionary Argument Against Naturalism (Ithaca, NY: Cornell University Press, 2002), pp. 176–203, 204-75Google Scholar.

²⁹ This is one way in which the postulate of transcendental freedom and the postulate of God's existence differ. Freedom is supposed to be a necessary condition for the applicability of the moral law to any being. It is the ratio essendi of moral obligation in general (KpV, 5: 4 n., 125-6). But the postulate of God's existence is only ‘the condition … of the mere practical use of our pure reason’ (KpV 5: 4). The argument for the postulate of God's existence (i) involves the more specific demand that we promote the highest good (which depends upon the assumption that we are (or, at least, that there are) finite beings with needs) and (ii) that we do not live i n (indeed, cannot really conceive of) a world where the natural law could guarantee the proportionality of virtue and happiness. Finite rational beings in such a possible world could coherently intend the highest good (as required), regardless of whether or not there was (or could be) a God. Thus, while the moral law directly entails the first postulate, the second postulate can only be derived in conjunction with these additional claims about us and our world. This is at least part of what Kant has in mind when he claims that the postulate of God's existence is not ‘as necessary’ to assume as the moral law (and transcendental freedom) are (KU, 5: 450-1; cf. KpV, 5: 143 n.). It need not be the case that every rational agent to whom the moral law applies has a rational need for the postulate of God's existence like we do. This is part of the point of the ‘so far’ qualifications at R, 6: 3 and KpV, 5: 143.

³⁰ This also seems to imply that, if there really were no God (regardless of our beliefs), the putative authority of the moral law for us (and its command that we promote the highest good) would be an illusion, a position supported at KpV, 5: 114 and KU 5: 471 n., and in the passages cited after n.19 in the main text.

³¹ KET, 410, n. 41. cf. , Wood, ‘REC’, 501Google Scholar . See also n. 33, below. , Hare, ‘KRD’, 466Google Scholar ,475 n. 31. cf. GC, 104. In his publications, Hare has stopped short of explicitly insisting that, for Kant, belief in God's existence is required for religion. But neither does he deny this claim, which is easily inferred from his discussion. In correspondence, Hare has suggested that, for Kant, assertoric faith is not a necessary condition of having religion, but that religion rationally commits one to such a belief.

³² For Wood's evidence, see KET 317, 410, n. 41; ‘RTM’, 403-6. Among the evidence supporting these claims: R, 6: 153 n.; ‘Pölitz’, 28: 998, 1010; ‘Volckmann’, 28: 1137-8, 1151; ‘Vigilantius’, 27: 531; Refl., 6213, 6226, 6227, 6236, 6244 (18: 497-520). Sometimes when Kant says this theoretical understanding is sufficient for religion, he also means to imply that it is a sufficient theoretical prerequisite leading to moral theism, which itself does involve (assertoric) faith. See ‘Pölitz’, 28: 998, 1011. There is also Kant's somewhat puzzling remark that ‘The mere possibility of God's existence is already sufficient for moral religion; yet not as much as faith’ (Refl. 6226 (18: 515)). It is unclear how ‘sufficiency’ can come in degrees; but such a comment does reflect a certain fluidity in Kant's usage of the term ‘religion’.

³³ See also Kant's claim that ‘with respect to the object toward which our morally legislative reason bids us work, what is presupposed [by the concept of religion] is an assertoric faith, practical and hence free, that promises a result for the final aim of religion’ (R, 6:153-4 n.). In ‘RTM’, esp. pp. 404–5, Wood suggested that while ‘maybe’ Kant thinks that something less than assertoric belief is required by his argument, ‘Kant thinks … something stronger than [hopeful agnosticism about God's existence] would be preferable’ or ‘harmonizes better’ with morality. Although such qualifications are not articulated in Wood's more recent discussions (KET and ‘REC’), he has assured me in correspondence that he agrees that Kant asserts that there is a positive ‘harmony’ between assertoric faith and a moral disposition (or the coherent reflection upon such a disposition) and that such a harmony does not exist between mere religion and such a disposition (or reflection upon it). Wood also thinks, however, that in spite of Kant's claims to the contrary, the premises of the moral argument at best justify belief in God's possibility, not his existence. I do not adjudicate this issue here.

³⁴ , Guyer, ‘FPP’, 337Google Scholar ,355; and ‘The unity of nature and freedom: Kant's conception of the system of philosophy’, in Sedgwick, S. (ed.), The Reception of Kant's Critical Philosophy: Fichte, Schelling and Hegel (Cambridge: Cambridge University Press, 2000), pp. 19–53CrossRefGoogle Scholar (hereafter ‘UNF’), pp. 31, 41. This is not to deny that the third Critique marks a significant development in Kant's thought, as its theor y of reflective judgement and of teleology attests. But I see no radical shift in the third Critique on the points at issue here, or, as I suggested in n. 15, above, on the conception of the highest good.

³⁵ , Guyer, ‘FPP’, 364–5.Google Scholar

³⁶ Cf. the 1793 draft of the ‘Real Progress of Metaphysics’ (20: 298). The passage suggests we act as if we know it, thus supposing we cannot know it, but does not say ‘act as if it were true’, supposing it is not true.

³⁷ , Hare, ‘KRD’, 465–6; 475, n. 31.Google Scholar

³⁸ Guyer claims that the change on this point is ‘striking but not radical’ given what has come before, more a ‘change in emphasis than in doctrine’, especially in light of continuities in his overall conception of the ‘systematic unity of nature and freedom’. ‘UNF’, 47, 20. In the present context, it is sufficient that these statements represent a genuine change on the point at issue.

³⁹ Fascicle VII, Sheet X, page 1; Förster/Rosen translation, pp. 200-1. This and related subjects are explored extensively in the seventh and first fas-cicles (composed between 1800 and mid-1802), Förster/Rosen trans., pp. 200ff., see Forster's ‘Introduction’, pp. xxvii-xxviii. The earliest evidence I have found of Kant using the phrase ‘(virtual) divine commands’ ((tamquam or instar) göttliche Gebote), is in Refl. 8104 (19: 646, which Erich Adickes dated as after 1796). The phrase is also found in the 1797 Metaphysics of Morals (MdS, 6: 443, 487), and Refl. 8110 (19: 650, from 1800). See also Refl. 6360 (18: 690, from 1797). For discussion of Opus postumum, see Förster, Eckart, Kant's Final Synthesis: An Essay on the Opus Postumum (Cambridge, MA: Harvard University Press, 2000)Google Scholar (hereafter KFS), esp. ch. 5; and , Guyer, ‘UNF’, 43ffGoogle Scholar.

⁴⁰ , Förster, KFS, 143.Google Scholar

⁴¹ , Wood, KMR, 188Google Scholar . Second emphasis added. Cf. ‘REC’, 500-2, 507. ‘RTM’, 407. Most recently KET, 317-20, 410, n. 42.

⁴² Refl. 7092 (19: 247). See also KrV, A632/B660.

⁴³ ‘In order to understand the function and justification of “religion” in human life, we must turn to Kant's argument in the first part of Book III of the Religion’ ( , Wood, KMR, 188Google Scholar ; cf. ‘REC’, 502, 507). In KET, Wood draws significant attention to connections between Kant's ethical and religious thought, on the one hand, and his social, historical and anthropological thought, on the other. One excellent way to follow the developments in Kant's conception of society and history is in the recently published notes from his lectures on Anthropology. For discussion, see Jacobs, Brian and Kain, Patrick (eds), Essays on Kant's Anthropology (Cambridge: Cambridge University Press, 2003)CrossRefGoogle Scholar.

⁴⁴ , Wood, KMR, 191.Google Scholar

⁴⁵ Hare emphasizes the multiple, interconnected roles that God plays in making this ‘kingdom of ends’ possible by coordinating, maintaining and judging members of it. ‘KRD’, 468-71.

⁴⁶ Rousseau had claimed that ‘the object of laws is always general’, but even this is a claim about laws, not about commands, per se. Rousseau, Jean Jacques, Du contrat social; ou, principes du droit politique, II. vi (Amsterdam: Marc Michel Rey, 1762)Google Scholar.

⁴⁷ It is important to note again that, on Kant's mature account, anticipation of these rewards (or the proportionality of virtue and happiness) cannot be the agent's motive for complying with the commands. Genuinely ethical rewards presuppose a morally worthy motive. Ethical rewards and punishments may appear in the agent's end, but not the agent's motive. Likewise, the point of divine commands is not to induce agents to comply with the moral law. The point of divine commands is to promote the highest good (and enable us to promote it). Nonetheless, the promise of ethical rewards is part of what constitutes them as divine commands.

⁴⁸ But perhaps one could respond as follows: ‘in the case of a solitary individual, God may be needed as judge and executor, but not necessarily as legislator - why can't I be the legislator for myself? - while in the social case, a common legislator is needed, and only God can be that’. In reply, I would ask why a common legislator is needed in the social case. A common law may be a necessary part of a moral community, but if that law is necessary, why is a common legislator also needed? Also note that Kant's own claims about the incapacity of humans to serve as the legislature for a moral community actually hang on our inability to be judges or executors because of our limited knowledge and power. The operative assumption is that it is unhelpful in this context to separate the legislative, executive and judicial powers or functions.

⁴⁹ Of course, if Kant's moral argument for the existence of God fails and the ‘social’ argument Wood lays out can be salvaged from it, then the justification for the claims that we should regard our duties as divine commands would be that social argument. Along similar lines, see my comments on Reath's, Andrews ‘Legislating for a realm of ends: the social dimension of autonomy’, in my review essay of Reclaiming the History of EthicsGoogle Scholar, ed. Herman, Reath and , Korsgaard, Kantian Review, 3 (1999), 114–22Google Scholar.

⁵⁰ I would like to thank John Hare, Allen Wood, Karl Ameriks, Jeff Brower, Manfred Kühn, Reinhard Brandt, Andrew Chignell, Matt Caswell, the referees and editor of this journal for extensive comments on earlier drafts of this article and the Humboldt Foundation.