¹ All references to Kant's works are by the Akademie numbering of The Cambridge Edition of the Works of ltnmanuel Kant:

KpV The Critique of Practical Reason, trans. Mary J. Gregor, in Gregor, Mary J. (ed.), Practical Philosophy (Cambridge: Cambridge University Press, 1996).Google Scholar

GMS The Groundwork of the Metaphysics of Morals, trans. Mary J. Gregor, in Gregor, Mary J. (ed.), Practical Philosophy (Cambridge: Cambridge University Press, 1996).Google ScholarRGV Religion within the Boundaries of Mere Reason, trans. George di Giovanni, in Wood, Allen W (ed.), Religion and Rational Theology (Cambridge: Cambridge University Press, 1996).Google Scholar

² Schopenhauer, Arthur, On the Basis of Morality, trans. Bullock, Arthur Brodrick (New York: Macmillan, 1915), p. 69.Google Scholar

³ Ameriks, Karl, ‘Kant's deduction of freedom and morality’, Journal of the History of Philosophy, 19 (1981), 72.CrossRefGoogle Scholar See also Bittner, Rudiger, Moralisches Gebot oder Autonomie (Freiburg: Verlag Karl Alber, 1983), pp. 138–43;Google ScholarPrauss, Gerold, Kant iiber Freiheit als Autonomie (Frankfurt a.M.: Klostermann, 1983), p. 70.CrossRefGoogle Scholar

Dieter Henrich sees the fact of reason as being closer to the Hutchesonian ‘moral sense’ views of Kant's 1765 Observations on the Feeling of the Beautiful and of the Sublime: ‘The concept of moral insight and Kant's doctrine of the fact of reason’, in The Unity of Reason, trans. Edwards, Jeffrey, Hunt, Louis, Kuehn, Manfred, and Zoeller, Guenter (Cambridge, MA: Harvard University Press, 1994), pp. 55–89.Google Scholar

⁴ Ameriks, ‘Kant's deduction,’ p. 72. More recently, Ameriks has argued that the fact of reason involves not a dogmatic appeal to rational intuition, but rather a greater willingness to take the character of moral experience at face value: ‘the mere fact that many readers have responded so positively to his characterization of it implies that there is “something to be said for” the fact of reason … even if it cannot be “deduced” it is not thereby much worse off than most of what is asserted by philosophers.’ ‘”Pure reason of itself alone suffices to determine the will”’, in Interpreting Kant's Critiques (Oxford: Oxford University Press, 2003), p. 260. Ameriks is not setting the philosophical bar here very high.Google Scholar

⁵ G. W E Hegel, Lectures on the Philosophy of History, as quoted in Henrich, ‘The concept of moral insight’, p. 69.

⁶ See Ameriks, ‘Kant's Groundwork III argument reconsidered’, in Interpreting Kant's Critiques, pp. 226–48.

⁷ In his review of Schulz, published shortly before the Groundwork, Kant clearly presents the practical ‘presupposition’ of freedom not as a belief we must assume, but as something that characterizes the way in which a rational agent must act: ‘the most confirmed fatalist…must still, as soon as he has to do with wisdom and duty, always act as if he were free’ (8: 13, Kant's emphasis).

⁸ Ameriks does not consider such a purely regulative construal of the idea of freedom here, and so concludes that there is no sense in which a rational agent ‘must’ art under the idea of freedom that could support any of Kant's moral conclusions. See ‘Kant's Groundwork III argument reconsidered’, p. 247.

⁹ Ameriks faults Kant here for failing to rule out this sort of compatibilist response (‘Kant's Groundwork III argument’, p. 243). On my view, Kant has no need to rule out this response, because he is not yet trying to show that we must believe or assume that we are transcendentally free at all. At this point in the argument, Kant is only trying to show the role that the concept of a transcendentally free agent plays in determining the proper norms of rational deliberation.

¹⁰ Henry Allison argues that Kant worries about the circle because, although he has established that rational agents must take themselves to be bound by the moral law, he has not yet shown why we are entitled to think of ourselves as rational agents in the first place. However, Allison never explains why that problem would produce the sort of circularity that Kant worries about. That is, it is unclear why our grounds for thinking ourselves to be rational agents might seem to Kant to be exclusively if covertly moral. A morally committed person must see herself as a rational agent, but so too must someone committed to pursuit of her own happiness, or any other goal other than acting from the impulse of the moment. See Allison, , Kant's Theory of Freedom (Cambridge: Cambridge University Press, 1990), pp. 218–21.CrossRefGoogle Scholar

¹¹ See Korsgaard, Christine, ‘Skepticism about practical reason, in Creating the Kingdom of Ends (Cambridge: Cambridge University Press, 1996), pp. 311–34.CrossRefGoogle Scholar

¹² See Schönecker, Dieter, Kant: Grundlegung 111: Die Deduction des kategorischen Imperativs (Freiburg: Karl Alber, 1999), p. 325.Google Scholar

¹³ See Reath, Andrews, ‘Hedonism, heteronomy, and Kant's principle of happiness, in Agency and Autonomy in Kant's Moral Theory (Oxford: Clarendon Press, 2006), pp. 33–67.CrossRefGoogle Scholar

¹⁴ See Kant's Logic, sec. 92, p. 139. Schönecker observes that, as Kant presents it, the two ‘arcs’ of the circle do not meet logically. We take ourselves to be under moral laws because [weil] we are free, but we take ourselves to be free in order to [um] think of ourselves under moral laws. The problem then is not just one of the logical relations between the steps in the argument, but with the likely background commitments of those ready to accept it. See Schönecker, pp. 333–9.

¹⁵ See Allison, Kant's Theory of Freedom, p. 226.

¹⁶ See Allison, Kant's Theory of Freedom, p. 226. For a defence of the appeal to the ‘Superiority des ontologischen Status der Verstandeswelt’ see Schnönecker, pp. 364–79, and McCarthy, Michael, ‘The objection of circularity in Groundwork III’, Kant-Studien, 76 (1985), 28–42.CrossRefGoogle Scholar

¹⁷ Korsgaard, ‘Morality as Freedom’, in Creating the Kingdom of Ends, pp. 168–9.

¹⁸ Beck, Lewis White, ‘The fact of reason: An essay on justification in ethics’, in Studies in the Philosophy of Kant, (Indianapolis: Bobbs-Merrill Co., 1965), p. 209.Google Scholar

¹⁹ Rawls, John, ‘Themes in Kant's moral philosophy’, in Forster, Eckart (ed.), Kant's Transcendental Deductions (Palo Alto, CA: Stanford University Press, 1989), pp. 81–113.Google Scholar See also Neiman, Susan, The Unity of Reason (Oxford: Oxford University Press, 1994).Google Scholar

²⁰ Cf. Ameriks, ‘Kant's deduction’, and Allison, Kant's Theory of Freedom, p. 245.

²¹ Cf. Ian Proops, ‘Kant's legal metaphor and the nature of a deduction’, Journal of the History of Philosophy, 41 (2003), 209–29CrossRefGoogle Scholar: ‘[Kant's] point is that the moral law is a datum of reason, not of sense’ (p. 228). Yet here Kant is clearly contrasting ‘fact of reason’ not with ‘fact of sense’, but with ‘datum of reason’. This contrast would be surprising if the fact were, as Proops claims, a special kind of phenomenological evidence that we can be motivated by pure reason.

²² See Allison, Kant's Theory of Freedom, pp. 236–8.

²³ Both Allison and Proops argue that the fact of reason is just our awareness, when confronted with moral questions, of a non-empirical source of motivation within us that we are entitled t o take to be an exercise of pure practical reason. Supposedly, the purity of this motivation reveals itself in the phenomenology of the feeling of respect, as something utterly distinct from empirical forms of pain, pleasure and desire.

This interpretation would seem to conflict with Kant's insistence that the moral law not only does not need a deduction, but also that it cannot be given one. If this reading is correct, then Kant has ready to hand a deduction of the moral law that is closely analogous t o the transcendental deduction of the categories. This deduction would begin with a particular kind of experience (moral motivation independent of sensible feeling) and derive the moral law, as a binding principle of reason, as a necessary condition for such experience.

²⁴ Herman, Barbara reaches a similar conclusion in ‘Rethinking Kant's hedonism’, in Byrne, Alex, Stalnaker, Robert and Wedgwood, Ralph (eds), Fact and Value: Essays on Ethics and Metaphysics for Judith Jarvis Thomson (Cambridge, MA: MIT Press, 2001), pp. 129–54.Google Scholar

²⁵ Kant here defines self-love in its most general sense as ‘this propensity to make oneself as having subjective determining grounds of choice into the objective determining ground of the will in general’ {KpV 5:74).

²⁶ Kant does claim that there is something strange about the idea of a command t o be happy, because such a command ‘would be foolish, for one never commands of someone what he unavoidably wants already’ (KpV 5:37). However, this need not mean that a purported command to be happy would not really be a command at all. In general, foolish or unnecessary commands do not, in virtue of such flaws, count as merely suggestions.

²⁷ See Wood, Allen, ‘Self-love, self-benevolence, and self-conceit’, in Engstrom, Stephen and Whiting, Jennifer (eds), Aristotle, Kant, and the Stoics: Rethinking Happiness and Duty (Cambridge: Cambridge University Press, 1996)Google Scholar; and Reath, Andrews, ‘Kant's theory of moral sensibility: Respect for the moral law and the influence of inclination’, in Agency and Autonomy in Kant's Moral Theory (Oxford: Clarendon Press, 2006), pp. 8–32.CrossRefGoogle Scholar

²⁸ ‘“Nil fecerit, esto: hoc volo, sic jubeo, sit pro rationae voluntas.” Imperat ergo viro.’ Juvenal and Persius, Satires, trans. Ramsay, G. G. (London: William Heineman, 1930), pp. 100–1 (11. 221–6).Google Scholar

Proops sees this allusion as suggesting not ‘willfulness and violence’ on the part of pure reason, but as rather a reminder of reason's ‘originally lawgiving’ status. This disregards the fact that, here, the wife does not properly have any such prior status, but is instead making herself into the head of the household by sheer chutzpah.

²⁹ Henrich, Dieter, ‘Kant's notion of a deduction and the methodological background of the first Critique’, in Forster, Eckart (ed.), Kant's Transcendental Deductions (Palo Alto, CA: Stanford University Press, 1994), p. 83.Google Scholar

³⁰ This is not to deny Kant's claim that a person can have an evil ‘disposition’ that systematically subordinates morality to self-love (RGV 6: 36). My position only entails that such a disposition could only be sustained through a continuous process of self-deception, what Kant calls ‘a certain perfidy on the part of the human heart in deceiving itself as regards its own good or evil disposition’ (RGV 6: 38).

³¹ My position bears some resemblance to that of Pawel Lukow's ‘The fact of reason: Kant's passage to ordinary moral knowledge’, Kant- Studien, 84 (1993), 204–21. Lukow sees the fact as a picture of how the moral law could fit into the rest of our moral psychology. This portrayal is not to justify morality to us, but to bring us to a certain kind of recognition of ourselves as morally committed beings. On the reading I have offered such recognition is as much a matter of self-constitution as it is of self-discovery.Google Scholar

³² In the terms of contemporary meta-ethics, Kant's account is constructivist in the very broad sense that the ultimate grounds of morality are to be found in basic features of the way we conceive of ourselves as agents, rather than in something distinct from or prior to such self-conceptions (see Rawls, John, ‘Kantian constructivism in moral theory’, in Darwall, Stephen, Gibbard, Allan and Railton, Peter (eds), Moral Discourse and Practice (Oxford: Oxford University Press, 1997), pp. 247–66).Google Scholar However, Kant's position differs from those forms of constructivism that hope to understand recognizably moral thought in terms of how some non-moralized form of practical reasoning would be exercised in some special sort of discursive context. For Kant, morality defines a distinctive type of practical reasoning, which cannot be reduced to or modelled on some kind of non-moral form of practical reason being exercised under special conditions (e.g. Rawls, Gauthier, Korsgaard). In this, Kant's mature position bears a strong affinity to certain kinds of non-reductive realism in ethics, such as those of Dworkin, Nagel and Scanlon.

For helpful comments and discussion, I would like to thank Hugh Chandler, Christine Korsgaard, Rahul Kumar and T. M. Scanlon as well as two anonymous referees for Kantian Review.