1 Korsgaard, Christine The Sources of Normativity, O'Neill, O. ed. (Cambridge: Cambridge University Press 1996), 123CrossRefGoogle Scholar. Henceforth I will refer to this work as SN.

2 See SN 122.

3 SN 122. In a note on this page, Korsgaard indicates that she is summarizing the argument she attributes to Kant in ‘Kant's Formula of Humanity.’ See Korsgaard, Christine ‘Kant's Formula of Humanity,’ Creating the Kingdom of Ends (Cambridge: Cambridge University Press 1996), 106–32CrossRefGoogle Scholar. Henceforth I will refer to this article as FH.

4 See Kant, Immanuel Grounding for the Metaphysics of Morals, Ellington, J. trans. (Indianapolis: Hackett 1993), 428–9Google Scholar, cited henceforth as GMS. I am referring to the Preussische Akademie (vol. IV) pagination of this work, which is included in the margins of the Ellington edition.

5 For example, David Cummiskey cites the regressive argument, largely in the form in which Korsgaard presents it, as the Kantian defense of the claim that if we take there to be fully justified actions, then we must hold humanity to be an end-in-itself (unconditionally valuable). Cummiskey seems to endorse this defense. See Cummiskey, Kantian Consequentialism (New York: Oxford University Press 1996), 62–80CrossRefGoogle Scholar.

6 FH 117

7 Before she begins the regressive argument, Korsgaard says: ‘In the argument for the Formula of Humanity as I understand it, Kant uses the premise that when we act we take ourselves to be acting reasonably and so we suppose that our end is, in his sense, objectively good’ (FH 116 ). At the beginning of the argument, Korsgaard asks: ‘Suppose that you make a choice, and you believe what you have opted for is a good thing. How can you justify it or account for its goodness?’ (FH 121) Korsgaard appears to use the terms ‘good end’ and ‘objectively good end’ interchangeably. I will employ the simpler term ‘good end.'

8 FH 110

9 See FH 111 and Korsgaard, ‘Motivation, Metaphysics, and the Value of the Self,’ Ethics 109 (1998), 55.CrossRefGoogle Scholar

10 FH 110-11

11 See FH 111 and FH 113.

12 For evidence that Korsgaard is using ‘humanity’ as a synonym for rational nature, see FH 110-14. For evidence that she equates rational nature with the power of rational choice, see FH 123. At FH 123 Korsgaard says that ‘humanity is the power of rational choice.'

13 See FH 119-24. For a summary of the regressive argument, see Korsgaard, Christine ‘Two Distinctions in Goodness,’ Creating the Kingdom of Ends (Cambridge: Cambridge University Press 1996), 256–62CrossRefGoogle Scholar. Henceforth I will refer to this article as TDG. See also Korsgaard, ‘Aristotle and Kant on the Source of Value,’ Creating the Kingdom of Ends (Cambridge: Cambridge University Press 1996), 239–43CrossRefGoogle Scholar.

14 For evidence that this is the notion of a sufficient condition that Korsgaard has in mind, see FH 122.

15 See, for example, Hill, Thomas E. Jr. ‘Humanity as an End in Itself,’ Ethics 91 (1980) 84–90CrossRefGoogle Scholar, and Cummiskey, Kantian Consequentialism.

16 FH 115

17 See FH 115, 120, and 122.

18 At FH 116, Korsgaard says: ‘If one's end cannot be shared, and so cannot be an object of the faculty of desire for everyone, it cannot be good, and the action cannot be rational.’ At FH 122, she says that ‘the good must be a consistent, harmonious object of rational desire.’ See also FH 123 where Korsgaard says that ‘the good is a consistent, harmonious object shared by all rational beings.’ We find in Korsgaard's discussion of happiness confirmation that, according to her, a consistent, harmonious object is something such that one person's realizing it does not preclude others from doing so. Everyone's happiness does not form a consistent harmonious object, Korsgaard says. Moreover, she clearly implies that it fails to form such an object for the reason that some people's being happy precludes others from being happy. (See FH 122 and III in this paper.)

19 FH 122

20 TDG257

21 Korsgaard, ‘Aristotle and Kant on the Source of Value,’ 227. See also TOG 259.

22 SN 122. To my knowledge, in her interpretive writings such as FH Korsgaard does not explicitly assert that the regressive argument supports this claim.

23 See TOG 258

24 One might argue here that Fran must be falsely characterizing her own ends. She says that she aims to be the sole victor in the accordion competition and to be the first and only wife of her co-worker Ron. But, accurately described, her ends must be to marry well and to be the best accordion player she can be. Accurately described, her ends do meet Korsgaard's third criterion. In response, I do not see why we should conclude that Fran must be falsely characterizing her own ends. Some agents surely do have ends such as that of being the sole winner in a competition, rather than, say, doing their best in it. In real life, it is accurate for these agents to describe some of their ends in such a way that the ends do not fulfill Korsgaard's third criterion. It seems perfectly coherent to imagine someone who can accurately describe all of her ends in this way. Of course, I am not insisting that there really are such people.

25 Korsgaard might respond that though Fran takes herself to have reasons for her actions, she is rationally compelled to acknowledge that she fails to have any. An agent really has a reason to perform an action only if it aims at an end such that his realization of it would not preclude anyone else from realizing it. Yet why must Fran embrace this conception of a reason? What, precisely, would be irrational in her failure to do so? If Korsgaard showed that Fran was rationally compelled to embrace this conception, then she would have grounds for concluding that, given the nature of her ends, Fran would be bound to complete normative skepticism. To my knowledge, however, Korsgaard does not show this. To my knowledge, there are two places in which Korsgaard might be construed to offer support for the claim that an agent really has a reason to perform an action only if it aims at an end such that his realization of it would not itself preclude anyone else from realizing it. The places are Lecture 4 of SN, especially 131-45 and Korsgaard, ‘The Reasons We Can Share: An Attack on the Distinction Between Agent-Relative and Agent-Neutral Values,’ Creating the Kingdom of Ends (Cambridge: Cambridge University Press 1996), 275–310CrossRefGoogle Scholar. Discussion of these very difficult works is beyond the scope of this paper. But it is worth noting that Lecture 4 of SN presupposes the success of Korsgaard's arguments in Lecture 3, arguments I criticize in V below. For criticism of Lecture 4 and of ‘The Reasons We Can Share,’ see Politas, Vasilis ‘Reasons and Values,’ International Journal of Philosophical Studies 5 (1997) 426–47Google Scholar. Even if, contrary to my opinion, Korsgaard succeeds in defending the claim in question, she would, I think, still need to face my second objection to her suggestion that not having good ends in her sense amounts to embracing complete normative skepticism.

26 This example stems from Gaut, Berys ‘The Structure of Practical Reason’ in Ethics and Practical Reason, Cullity, Garrett and Gaut, Berys eds. (Oxford: Oxford University Press 1997), 174.Google Scholar

27 At this point, a defender of Korsgaard might object that I have overlooked an important ground she has for the claim that to avoid complete normative skepticism, an agent must hold there to be something unconditionally good from which the goodness of his good ends derives. This ground comes to light when we notice that the notion of good ends Korsgaard employs in the regressive argument does not come out of nowhere. Prior to the argument, Korsgaard establishes that if an agent takes there to be a categorical imperative, then he must agree that there are good ends in the robust sense. According to the defender, that she establishes this supports her notion that normative skepticism is the price of holding there to be nothing unconditionally good that makes good ends good. I agree that Korsgaard tries to show that if we assume there is a categorical imperative, then we must hold that there are good ends in the robust sense. (See, for example, TDG 260 and FH 116-17.) And, for the sake of argument, I grant that she succeeds. Still, I do not see how showing this supports Korsgaard's suggestion that denying there to be good ends (in her sense) amounts to embracing complete normative skepticism. It would support this suggestion only if denying there to be a categorical imperative would itself land one in complete normative skepticism. In the sense at work here, a categorical imperative would be a practical principle that was unconditionally binding on all rational beings. It would be a principle to which all such beings were bound to conform, regardless of their desires or particular natural constitutions. (For Kant's introduction of this notion of a categorical imperative, see GMS 414-20. Strictly speaking, imperatives do not, in Kant's view, apply to perfectly rational beings, e.g., God and angels, who necessarily act in accordance with unconditionally binding principles. See, for example, GMS 414.) Yet why should we believe that one could eschew the existence of such a principle only at the cost of finding oneself with no reason to do anything at all? Suppose the agent in our example, i.e., the one who thinks that the goodness of his good ends derives from their being objects of his reflective choice, denies that there is a categorical imperative. It is, I think, at least as unclear why this denial would push him into total normative skepticism as it is why his denial that he had ‘fully justified’ ends would push him into it. His having reasons to act, e.g., to pursue objects he has reflectively chosen rather than ones he has not, would not seem to depend at all on there being a practical principle binding on all rational beings- including angels and rational extraterrestrials, if they exist. To my knowledge, Korsgaard does not give us good grounds for thinking that if an agent denies that there is a categorical imperative (in the precise sense in question), he must be a normative skeptic. And unless we have good grounds for thinking this, the defender's objection lacks force.

28 Though here I will focus my criticism of the regressive argument on (v), I do not wish to suggest that I find nothing problematic in (vi)-(ix). I hope to give these steps the detailed attention they require on another occasion.

29 FH 121

30 GMS428

31 Of course, Korsgaard has at her disposal another means of showing that maximum species preservation fails to be unconditionally good: Kant's famous (and much criticized) Groundwork I argument that nothing except a good will can even be conceived of as unconditionally good. See GMS 393-5. For a recent criticism of Kant's argument, especially as a response to the kind of value realism I have discussed, see Berys Gaut, ‘The Structure of Practical Reason,’ 165-70.

32 In focusing on the possibility of a value realist posing this sort of question, I am following Berys Gaut. See Gaut, ‘The Structure of Practical Reason,’ 176. I do not wish to suggest that Gaut defends the environmentalist position I have mentioned. He does support a version of value realism, but not environmentalism.

33 See Gaut, The Structure of Practical Reason.'

34 FH 121

35 See FH 122.

36 FH 122

37 For this conception of happiness in Kant see, for example, GMS 399 and 405. In the Critique of Pure Reason, Kant says: ‘Happiness is the satisfaction of all our desires (Neigungen], extensively, in respect of their manifoldness, intensively, in respect of their degree, and protensively, in respect of their duration,’ Kant, Immanuel Critique of Pure Reason, Smith, N. Kemp trans. (New York: St. Martin's Press 1965), A806, B 834Google Scholar. In addition to a desire-satisfaction account of happiness, Kant offers a hedonistic account. See, for example, Kant, Immanuel Critique of Practical Reason, Beck, Lewis White trans. (New York: Macmillan 1993), 20Google Scholar (Preussische Akademie edition [vol. V] 22).

38 It is not logically impossible for everyone to be happy, even on Kant's desire-satisfaction account of happiness. For we can coherently imagine a world in which every person always gets what he wants. Of course, in this imagined world, no one person's satisfying any of his desires would preclude any other person from satisfying any of her desires.

39 Kant says that ‘an impartial rational spectator can take no delight in seeing the uninterrupted prosperity of a being graced with no feature of a pure and good will’ (GMS393).

40 I do not believe Korsgaard would disagree with this characterization. See Korsgaard, ‘Kant's Analysis of Obligation: The Argument of Groundwork I.’ Creating the Kingdom of Ends (Cambridge: Cambridge University Press 1996), 60–1CrossRefGoogle Scholar. For discussion of this sort of interpretation of what Kant means by a good will, see Karl Ameriks ‘Kant on the Good Will; Grundlegung zur Metaphysik der Sitten: Ein kooperativer Kommentar, Höffe, O. ed. (Frankfurt: Klostermann 1989), 54–9Google Scholar.

41 I am here following Ameriks, ‘Kant on the Good Will,’ 53.

42 SN 122

43 SN 93

44 SN 93. See also SN 97.

45 This is nearly a direct quotation of SN 101.

46 See SN 111-13.

47 SN 91

48 I will not discuss what remains of Korsgaard's argument after step (4), but I do not wish to suggest that it is unproblematic. For criticism of (what amounts to) steps (6)-(8), see Skorupski, John ‘Rescuing Moral Obligation,’ European Journal of Philosophy 6 (1998) 348–50CrossRefGoogle Scholar. For a criticism of Korsgaard's move from (7) to (8), see Geuss, Raymond ‘Morality and Identity,’ in SN, 197Google Scholar.

49 For criticism of Korsgaard's argument from (1)-(3), see Cohen, G.A. ‘Reason, Humanity, and the Moral Law,’ in SN, 185Google Scholar.

50 SN 120-1

51 Korsgaard clearly believes that an agent must appeal to identity necessity in order to have sufficient grounds for maintaining his particular practical identity. At SN 129, for example, she says: ‘If we do not treat our humanity as a normative identity, none of our other identities can be normative, and then we can have no reasons to act at all.’ At SN 121, Korsgaard says: ‘our identity as moral beings- as people who value themselves as human beings - stands behind our more particular practical identities. It is because we are human that we must act in the light of practical conceptions of our identity, and this means that their importance is partly derived from the importance of being human. We must conform to them not merely for the reasons that caused us to adopt them in the first place, but because being human requires it.'

52 SN 125, italics mine.

53 See SN 16.

54 See SN 9-10.

55 SN 47

56 See SN 33.

57 See SN 33 and 35.

58 SN 40

59 See SN 40, in particular: ‘It is because we are confident that obligation is real that we are prepared to believe in the existence of some sort of objective values. But for that very reason the appeal to the existence of objective values cannot be used to support our confidence. And the normative question arises when our confidence has been shaken, whether by philosophy or by the exigencies of life. So realism cannot answer the normative question.’ Also see SN 47 and SN 90-1.

60 SN 40

61 SN 257-8

62 Unless the rescuer is able to justify his choice of his particular practical identity over other (incompatible) ones available to him, he cannot answer the normative question, as Korsgaard conceives of it. Let us suppose that the rescuer poses the normative question, i.e., asks himself what justifies the claims that morality makes on him. (See SN 9-10.) The rescuer holds that moral obligations stem from his particular practical identity .In posing the normative question, he is, in effect, asking himself why he ought to fulfill these obligations. Why ought he to risk his life in order to save others - instead, for example, of focusing on making a fortune for himself? In the rescuer's view, if he changed his particular practical identity to that of an entrepreneur, which is, he thinks, something that he could do, then he would not have the moral obligations that he now has. It is easy to see that if the rescuer fails to have a satisfactory answer to the question of why he should maintain his particular practical identity rather than adopt a different one such as that of an entrepreneur, then he also fails to have a satisfactory answer to the normative question. He has no adequate response to the question of what justifies the claims that, in his view, morality makes on him.

63 See SN 120-1.

64 I am grateful to David Cummiskey for discussion of this sort of response to my criticism of the practical identity argument.

65 See SN 112-13, especially the last paragraph in 3.3.7.

66 Material from sections IV and V was presented in a colloquium paper at the 1999 Pacific Division Meeting of the American Philosophical Association. Thanks to David Cummiskey for his helpful commentary there. I am also grateful for discussion with Berys Gaut, Thomas Pogge, and Michael Slote, as well as for comments from referees for the Canadian Journal of Philosophy.