Acting for the right reasons

Philosophical Review

Volumn 119, Issue 2, 2010, Pages 201-242

  Markovits, Julia ^a  

^a MASSACHUSETTS INSTITUTE OF TECHNOLOGY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 77949746314     PISSN: 00318108     EISSN: 15581470     Source Type: Journal    
DOI: 10.1215/00318108-2009-037     Document Type: Article

References (76)

1. Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. Mary Gregor (Cambridge: Cambridge University Press, 1997), 3 (4:390), emphasis in original. Parenthetical citations from Kant's work refer to the Royal Prussian Academy of Sciences edition of Kant's Gesammelte Schriften (Berlin: George Reimer, later Walter de Gruyter, 1900-), with volume and page numbers separated by a colon.
2. Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. Mary Gregor (Cambridge: Cambridge University Press, 1997)., 11 (4:398).
3. Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. Mary Gregor (Cambridge: Cambridge University Press, 1997)., 12 (4:398).
4. Itisimportanttodistinguish,inthiscontext,betweenactionswehaveinstrumental reasons to praise and actions that merit praise in their own right. Kant himself makes clear that he believes that an action performed from beneficent motives, such as sympathetic concern for others or the inclination toward honor deserves praise and encouragement because it fortunately lights on what is in fact in the common interest and in conformity with duty. Suchan actionis to be praised if praising it makes people more likely to perform similar actions because it is good that people perform them. But it is not to be esteemed as morally worthy because it may not be good in a person that she performs it. According to Kant, it may be merely accidental-fortunate-that a person acting on such motives does the right thing; the motive need not reflect a good will. Whether Kant is right to char-acterize sympathetic actions in this way will be discussed further below (see sec. 3). Thanks to an anonymous referee for the Philosophical Review for pressing me to be more precise on this point. I will also return, in discussing the problem my account of moral worth raises for utilitarianism, to the question of whether a distinction between actions we have instrumental reasonsto praise andactions thatmerit praise intheir own rightmustbe acknowledged (see sec. 4, n. 65).
5. We might think an action performed out of respect for the moral law is one per-formed whenever it is believed to be right, regardless of whether it actually is right. I set this thought aside here because it seems to me less promising than the version of the Motive of Duty Thesis I focus on above. I come back to this alternative version in sec. 3.
6. For an argument for the claim that action performed from the motive of duty involves a kind of moral fetishism, see Michael Smith, The Moral Problem (Oxford: Blackwell, 1994), 71-76.
7. See Bernard Williams, "Persons, Character, and Morality," in Moral Luck (Cambridge: Cambridge University Press, 1981), 1-19, at 18.
8. Christine Korsgaard argues that this worry about Kant's account of moral worth stems from a misreading of it. She maintains that the appropriate bearers of moral value a reactions, which, shesays, include both the act performed and the end to which it was performed. She takes this account of actions to make the Motive of Duty Thesis less unpalat-able. She writes, The idea that acting from duty is something cold, impersonal, or even egoistic is based on the thought that the agent's purpose or aim is in order to do my duty rather than in order to help my friend or in order to save my country, or whatever it might be. But that is just wrong. Sacrificing your life in order to save your country might be your duty in a certain case, but the duty will be to do that act for that purpose, and the whole action, both act and purpose, will be chosen as one's duty. (Korsgaard, Acting for a Reason, in The Constitution of Agency [Oxford: Oxford University Press, 2008], 207-29, at 218-19)
9. This response seems to me unsatisfactory. As I hope will become clear, even in Kors-gaard's version, the thesis retains the tension with other aspects of a Kantian approach to ethics that I'll describe in a moment and remains vulnerable to other objections and counterexamples-particularly those provided by apparently morally worthy agents who act in ignorance of the moral status of their actions-that I will discuss in sec. 2. But more relevantly here, it does not, in any case, entirely avoid the kind of one thought too many worry that the Motive of Duty Thesis inspired. That's because it's not clear how the motive of duty could, on Korsgaard's view, relate to the action of, say, doing something in order to helpone's friend, except by providing a more fundamental motive: I did something to help my friend in order to do my duty. The duty, and not the friend, remains the primary target of one's attention. (I'll have more to say about such chains of motivating reasons later on.)
10. In my discussion of other versions of the Right Reasons Thesis, I have taken the thesis to be about necessary conditions for the moral worth of actions. Here I expand the thesis to state necessary and sufficient conditions for the moral worth of actions. There is, of course, considerable debate about what it takes for agents to be morally responsible for their actions, and it is very plausible that an action can have moral worth only if it is one for which the agent is morally responsible. So the Coincident Reasons Thesis provides suf-ficient conditions for the moral worth of actions only if meeting the conditions for moral worth established by the thesis also entails meeting the conditions for moral responsibil-ity. It is my view that only agents who can be held morally responsible for their actions have moral reasons-that is, moral reasons apply to agents only if they've met the condi-tions for moral responsibility. Since the Coincident Reasons Thesis attributes worth to an agent's actions only when the agent is motivated to act by the moral reasons that apply to her, conditions for moral responsibility will have been met whenever the conditions for moral worth specified by the thesis have been met.
11. Kant, Groundwork, 3-4 (4:390).
12. See Kant, Groundwork, 3-4 (4:390)., 4: 390 (in the preface) and Kant's discussion of acting from duty at 4: 397-98 of sec. 2 (all quoted above).
13. Philip Stratton-Lake, Kant, Duty, and Moral Worth (London: Routledge, 2000), 60. Stratton-Lake considers a thesis that is similar to the Coincident Reasons Thesis, phrased not as a condition for the moral worth of actions but rather for the moral worth of agents: according to what he calls the Symmetry Thesis, the reason why a good-willed per-son does an action, and the reason why the action is right, are the same. Ibid., 16. He extracts this statement of the thesis from Christine Korsgaard, Creating the Kingdom of Ends (Cambridge: Cambridge University Press, 1997), 60. Stratton-Lake goes on to reject this thesisastoo strongasitstandsandinanycaseacceptsaversionofthethesisonlyasanec-essary, and not as a sufficient, condition for moral worth, for reasons I will come to later on (in sec. 3). But he maintains that some thesis like it must be accepted by Kantians.
14. Although all morally worthy (good-willed) actions will on this view be rational- that is, actions the agent has conclusive reasons to perform and that the agent performs for those reasons-not all rational actions will count as morally worthy. Some rational actions-actions that lack moral significance (my choice of the most appropriate spoon with which to eat my breakfast cereal, for example) are simply morally neutral. The Coin-cident Reasons Thesis speaks of morally justifying reasons and actions I morally ought to perform precisely to exclude morally insignificant actions from consideration-they fall outside of its scope.
15. As Stratton-Lake also notes, in relation to his symmetry thesis. See Stratton-Lake, Kant, Duty, and Moral Worth, 20, and my n. 12, above. As he points out, Korsgaard seems to accept both. My argument here mirrors his.
16. At most, the fact that I ought to φ provides me with a derivative reason to φ: a reason that adds no normative weight to my primary reasons for φing and does no work in explaining why φing is right.
17. Mark Twain, The Adventures of Huckleberry Finn (New York: Oxford University Press, 1996), 271-72.
18. Jonathan Bennett's discussion of the case of Huck Finnin" The Conscience of Huckleberry Finn," Philosophy 187 (1974): 123-34
19. first focused philosophers' attention on the example. Nomy Arpaly, in her book Unprincipled Virtue: An Inquiry into Moral Agency (New York: Oxford University Press, 2003), also appeals to the Huck Finn case, which she discusses at length, in her defense of a condition on morally worthy action similar to the one I propose here, in the form of the Coincident Reasons Thesis. Arpaly thinks her simi-larcondition, which shecalls Praise worth in essas Responsiveness to Moral Reasons (72), tells only part of the story about when actions have moral worth. I willd is cussher proposed further condition and the examples she thinks show it to be necessary in a moment. See chap. 3 of Unprincipled Virtue for a nuanced discussion of the problem of moral worth.
20. A defender of the Motive of Duty Thesis might be tempted to respond to Huck's case by saying that Huckis not really doing something he thinks is wrong because he doesn't really understand what wrong means. When he describes his action of protecting Jim as wrong to himself, he must mean something like forbidden by society, and we all know that that is not what wrong means. (Such an interpretation will be particularly tempt-ing to those philosophers, including, for example, Christine Korsgaard, who are skeptical about the very possibility of clear-eyed akrasia.) But I don't find this description of Huck's circumstance splausible. Huckmay indeed think that whenever some actiscondemned by society it is wrong, and he may even (though the book surely leavesthi sunder determined) believe that such acts are wrong in virtue of being condemned by society. But thinking that wrong means condemned by society is quite another thing. I think Huck means wrong by wrong-after all, he responds to his judgment that he is acting wrongly with a terri-ble attack of guilty conscience-a response to wrongdoing that will be familiar to many of us. Of course, nothing really turns on the correct interpretation of Huck's case-what matters is that a case of the kind I am imagining, in which someone acts rightly while really believing he is acting wrongly, is psychologically imaginable. Moreover, the defender of the Motive of Duty Thesis must establish more than that Huck doesn't believe himself to be acting wrongly in protecting Jim to credit his action with moral worth-he Hucktakes himself to beacting rightly (whatever term he would use instead) - to show that he acts from the motive of duty. And this is surely incredible.
21. As I note above, Philip Stratton-Lake has pointed out the appeal of a view he calls thesymmetrythesis,whichheattributestoKorsgaard,andwhichclaimsthatgood-willed agents act for right-making reasons (in Kant, Duty, and Moral Worth; see n. 12, above). And as I also note above, Nomy Arpaly has defended a thesis very like the Coincident Reasons Thesis, which she calls Praiseworthiness as Responsiveness to Moral Reasons (in Unprin-cipled Virtue; see n. 16, above). Arpaly takes her thesis to provide a necessary, but not suf-ficient, condition for the moral worth of actions, and Stratton-Lake thinks his similar the-sis identifies neither necessary nor sufficient conditions for moral worth, for reasons I will discuss.
22. According to Stratton-Lake, while morally worthy actions are indeed motivated by right-making reasons, their agents also act from a secondary or regulating, dispositional motive of duty: they will be disposed to be motivated by right-making reasons (on the pri-mary level) only if they judge their actions to be right. See Stratton-Lake, Kant, Duty, and Moral Worth, 60-67. This shares the significant disadvantage of the more traditional interpreta-tion of Kant's Motive of Duty Thesis in that it fails to recognize Huck Finn's decision to protect Jim as worthy. Arpaly's version of the same basic idea, which I discuss next, avoids this problem.
23. Arpaly, Unprincipled Virtue, 87.
24. Kant, Groundwork, 3-4 (4:390).
25. Stratton-Lake, Kant, Duty, and Moral Worth, 65.
26. I have intentionally phrased this as a necessary but not sufficient condition: the nature of an action may not always tell us something about the character of its agent, even when that agent's motive nonaccidentally results in an action of that type. Some motives might nonaccidentally result in right actions without thereby endowing those actions with moral worth. This might be the case, for example, if there were a necessarily existing God, who necessarily rewarded all right (but not necessarily worthy) actions in an afterlife and punished wrongones. If the existence of sucha God was known, then selfish motives might nonaccidentally produce right actions, but such actions would still not beworthy, or reflect well on the character of the agent-and the agent would not be performing them forright-making reasons.
27. Like that suggested by Stratton-Lake.
28. Arpaly, Unprincipled Virtue, 88-89.
29. What if the reason my friends should call me is not that doing so will make me happy but rather that remembering to call is the kind of behavior that is constitutive of loyalty, which is valuable in itself? Doesn't it then make a difference whether my friend is fair-weather or die-hard-whether, for example, she remembers my birthday reliably or not? In this case it may well matter, but it matters in away that is easily accommodated by the Coincident Reasons Thesis. My friend has reason to be loyal, and being loyal is not some-thing she can do intermittently, when the thought happens to strike her. Being loyal is not something you can accomplish with a single act. So where loyalty is required, a fair-weather friend does not just fail to do the right thing for the right-making reasons; she doesn't do theright-theloyal-thing at all. Thanks to Daniel Markovits for pressing meon this point.
30. See Richard Henson, "What Kant Might Have Said: Moral Worth and the Overdetermination of Dutiful Action, "Philosophical Review 88 (1979): 39-54, at48. Hensonuses the comparison to battle-citations in a different context.
31. Imagine that Paul, who is taking a difficult math examination, works out the answer to each question perfectly. His achievement would be no less epistemically worthyi fit were true of him that, had the lovely Linda been seated at the next desk, he would have been so distracted that he would have gotten every question wrong.
32. Such as that suggested by Arpaly and Stratton-Lake.
33. Stratton-Lake denies that verdictive normative beliefs can give us reasons in this way (see Kant, Duty, and Mora lWorth, 12, 20). I'm not sure from what argument he takes this conclusion to follow. It is not, it seems to me, entailed by the conclusion that the fact that I ought to φ is not a normative reason to φ. Nor does it seem to me to follow from StrattonLake's other claims that a verdictive moral consideration cannot be cited in support of itself (20) because no verdict constitutes evidence for itself (19). Even if we accept that my belief that I ought to φ is never an epistemic reason to believe I ought to φ, it may nonetheless be a normative reason for me to φ.
34. See, for example, Korsgaard's response in The Sources of Normativity (Cambridge: Cambridge University Press, 1996), 257 to G.A. Cohen'sexample of the Mafioso, who feels bound by his gangster's code of strength and honor to perform merciless acts:
35. I want to say of the Mafioso whatI saidof the Knight... who felt himselfto be obligated to fight a duel. There is a sense in which these obligations are real-not just psycholog-ically but normatively. And this is because it is the endorsement, not the explanations and arguments that provide the material for the endorsement, that does normative work.
36. Broome calls such a condition enkraticin, for example, Does Rationality Consist in Responding Correctly to Reasons? Journal of Moral Philosophy 4 (2007): 349-74, in sec. 4 (hereafter, Does Rationality Consist?). Hetalks about wide-versusnarrow-scopeversions of the requirement in, for example,
37. "Wide or Narrow Scope?" Mind 116 (2007): 359-70.
38. Broome adds the qualifier, if [we] believe [we] will [doit] if and only if [we] intend to [do it]. See Does Rationality Consist?
39. See, for example, Broome, "Normative Requirements," Ratio 12 (1999): 404;
40. Broome, Wide or Narrow Scope?; Nico Kolodny, Why Be Rational? Mind 114 (2005): 509-63.
41. If anything, there is something particularly sinister about agents who act terribly wrongly while claiming to have right on their side.
42. My defense of the Coincident Reasons Thesis as a sufficient condition for the moral worth of actions has the perhaps surprising implication that someone's actions could have moral worth even if she has no normative beliefs at all. (As Krister Bykvist has pointed out to me; it's interesting to note how very un-Kantian this worry is: the ideal Kantian agent is, after all, usually accused of overrationalizing-of having one thought too many.) Is this grounds for objecting to the thesis? One might worry that, if, as I've argued, acting worthily does not require awareness of the reasons for which one acts as reasons, or even any normative beliefs at all, then many examples of nonrational animal action will suddenly qualify as morally worthy. A quick response to the worry posed by nonrational animals is that the Coincident Reasons Thesis does not in fact entail that their actions have moral worth: since such ani-mals lack the higher rational capacities required for moral responsibility, no normative reasons apply to them, so their motivating reasons cannot coincide with their normative reasons (see n. 9). This response, however, feels unsatisfactorily glib since part of what is at issue is what kind of rational feature such animals' actions lack (the absence of which exempt them from moral obligations and their actions from moral worth). While I feel there is much more to be said about the matter, for example, I suspect that the lower ani-mals also don't properly act for motivating reasons, in the way in which I understand the term (see my discussion in sec. 3), I set these further issues aside here.
43. The problem with Huck's reasoning in cases like this is not just that it would lead him astray in other circumstances.As I argued in the previous section, counterfactual considerations like this don't tell us much about the moral worth of actions. The problem with relying on bad authoritiesis, as the Coincident Reasons The sishighlights, that actions can't be made right by the fact that they're recommended by these authorities: a bad authority's recommendation to do something is no reason to do it.
44. There is an important difference between relying on someone else's judgment about nonnormative questions and relying on her judgment about normative questions. Casesinwhichanotherperson'sverdictivemoraljudgmentthatsomeactIcanperformisright provides me with sufficient evidence that performing it would be best, and so with conclusive reason to perform it, without giving me any access to the underlying reasons on which she forms the judgment, will be very rare. In other words, morality places unusually strict limits on permissible deference to normative authorities, even accurate authorities. Usually, we must do our moral reasoning ourselves. This is an important feature of the practice of moral judgment that I can't explain satisfactorily here. But part of the explanation of the relative rarity of permissible deference to normative authorities, compared to other kinds of authorities, might lie in this: in order to be justified in allowing an authority's judgment to guide our actions, we must be in a good position to judge that the authority is a reliable one-that is, likely to be right about the subject matter of her expertise-without being in a position to form reliable beliefs about that subject matter ourselves. Doctors are a good example of authorities we can judge in this way-their track records, for example, allow us to judge their reliability
45. Stratton-Lake, Kant, Duty, and Moral Worth, 16-17.
46. See i Stratton-Lake, Kant, Duty, and Moral Worth, 17.
47. Should we instead think that while the moral worth of our actions is relative to our evidence, the rightness of our actions is not? Should we think, for example, that a doctor who offers her patient the course of treatment that her evidence (misleadingly) tells her is most likely to effect a cure acts worthily but not rightly (and not in response to genuine normative reasons)? That an agent who fails to offer aspirin to a friend who gives no out-ward indication of his discomfort is acting wrongly but not blameworthily? I think it would be a mistake to interpret such cases in this way: it seems clear, in the doctor case, that the doctor fulfills herobligations, actsasshe is morally required to act, and is responding to her normative reasons. And we should not say, in this headache case, that the agent is morally required to offer her silently suffering friend an aspirin. Rightness, moral requirements, and normative reasons go hand in hand: they all operate at the same level of evaluation. Of course, it would, in these cases, be fortunate if the doctor irresponsibly chose the more effective treatment, which her evidence did not support, and if our second agent offered her friend the aspirin anyway, on a whim, perhaps.
48. Why usually? I will discuss in the next section some cases in which it may not be possible for an agent to be motivated to perform an action by the normative reasons why it is right. Such cases, if they are possible, entail that it is sometimes not possible for an agent to perform the right action worthily. I will argue that such cases, if they exist, will be rare, on any plausible normative theory.
49. As is often the case with explanations, these kinds of explanations need not crowd out rationalizing explanations-both may be true at the same time.
50. It'snatural to think of rationalizing explanations as aspecies of causal explanation. Iwon'tconsiderthisproposalinanydetailhere.Butifmotivatingreasonsarecauses,then, if the Coincident Reasons Thesis is right, we should expect (1) that cases in which it is particularly difficult to identify the cause (at the level of motivating reasons) of an agent's actionwillalsobecasesinwhichit'smoredifficulttoassesstheaction'smoralworthand(2) that the irrelevance of counterfactuals about how an agent might have behaved in altered circumstances to determining the moral worth of her actions will be mirrored by the irrelevance of such counterfactuals to determining the motivating reasons that caused her to act. It seems to me that our intuitions about moral worth do falter where our intuitions about cause falter. And perhaps the most prominent defender of a causalist account of motivating reasons, Alfred Mele, does argue that appeals to counterfactuals cannot reliably reveal an agent's motivating reasons-see Mele, Motivation and Agency (Oxford: Oxford University Press, 2003), 51.
51. To say this is not to say that all motivating reasons are facts we take to be normative reasons, and certainly not to say that they are facts we take to be sufficient to justify our actions. We might sincerely offer a rationalizing explanation even when we believe our reasons for acting did not justify our action. I might offer as my reason for snapping at you the fact that your voice is rather shrill for this time in the morning, even if I know that the tone of your voice is no justification for snapping at you at all. And Huck could offer a sincere rationalizing explanation of his decision to protect Jim.
52. A paranoid parent who frequently rushes his perfectly healthy child to the doctor may mean well, but his actions are not fully worthy; in sec. 5, I show how the Coincident Reasons Thesis can explain why some well-meaning but wrong actions can be partially morally worthy.
53. Kant, Groundwork, 11 (4:398).
54. I won't tackle here the difficult problem of how to identify which of various incen-tives an agent has to perform an action serves as her motive, or the equally difficult prob-lem of what it is for an action to be overdetermined by an agent's motives. Both topics have received significant attention from Kant scholars writing about the motive of duty. See, for examples, Barbara Herman, "On the Value of Acting from the Motive of Duty," in The Practice of Moral Judgment (Cambridge, MA: Harvard University Press, 1993);
55. Marcia Baron, Kantian Ethics Almost without Apology (Ithaca, NY: Cornell University Press, 1995); as well as Stratton-Lake, Kant, Duty, and Moral Worth; and Henson, What Kant Might Have Said.
56. Although pleasure is the motive that interests me here, it is not, of course, the only possibleselfishmotiveanagentmighthavefordoinggood.Thereareanumberofreasons agents might care that they do good, rather than that good be done. They might, for example, want the reputation of a do-gooder. Or they might simply be too exclusively focused on their own virtue-and self-defeatingly so. As Hume pointed out, to suppose that a virtu-ous (that is, virtuous-making) action is one performed out of regard for its own virtue is to reason in a circle. Hume, A Treatise of Human Nature, ed. by L. A. Selby-Bigge and P. H. Nidditch (Oxford: Oxford University Press, 1978), 478. And in any case, the fact that performing some actis what avirtuousperson woulddois, like the factthatanactis right, nota reason to perform it but rather reflects the fact that there are other, conclusive, reasons to perform it. So the Coincident Reasons Thesis helps explain why acts motivated (solely) by anagent'sdesirefor herownvirtue aren'tworthy. ButIdon't mean to suggest that wanting not just that good be done but also to play a role in doing it is always a morally unattractive desire. Sometimes we might have good normative reasons to ensure that we be the ones to bring about some good-for example, when fairness demands that we contribute our share to a collective good. Thanks to Daniel Markovits and to an anonymous reviewer for the Philosophical Review for helpful discussions of these issues.
57. Here again, my account of moral worth departs from that offered by Nomy Arpaly. She asserts, in stark contrast to Kant, that the cold-hearted philanthropist is less praise-worthy than he would be if it were not for his cold-heartedness. Unprincipled Virtue, This is another example of a case in which Arpaly thinks a thesis like the Coincident Reasons Thesis cannot, on its own, account for the moral worth of actions.
58. Williams, "Persons, Character, and Morality," 18.
59. Thomas Nagel, The Possibility of Altruism (Oxford: Oxford University Press, 1970), 19.
60. Korsgaard, "Two Distinctions in Goodness," in Creating the Kingdom of Ends, 250.
61. Noninstrumentally valuable ends can sometimes be sources of noninstrumental (normative and motivating) reasons to perform actions even when those actions won't actually promote the ends in question (for example, when appearances are misleading). For example, the noninstrumental value of relieving someone's pain can give me a non-instrumental normative reason to offer a friend an aspirin even when she merely appears to have a headache, or a noninstrumental normative reason to give my child the medica-tions recommended by her (generally reliable) doctor even when the doctor's prescrip-tion won't in fact help. My end of relieving pain, when I value it for its own sake, can also provide me with noninstrumental motivating reasons to do these things. My (normative and motivating) reasons in each case will be the fact that my friend appeared to be in pain and the fact that the medication appeared to be most likely to effect a cure.
62. I take the term from Derek Parfit, who provides a sophisticated discussion of the ways in which utilitarianism, and other versions of consequentialism, might be self-defeating. See Reasons and Persons (Oxford: Oxford University Press, 1984), chap. 1, especially 5 and sec. 10. Parfit argues that the fact that a thesis is self-defeating may not tell againstitsplausibility.Iwillarguethat,iftheCoincidentReasonsThesisidentifiestheright necessary conditions for the moral worth of actions, the fact that a thesis is self-defeating may well tell against its plausibility.
63. The worry described here targets a simple, hedonist version of utilitarianism according to which the only thing that is valuable as an end is pleasurable experience. One might hold a more complicated utilitarian view according to which certain pleasur-able activities and relationships are themselves components of happiness, and so are valu-able as ends. This kind of utilitarianism would be less vulnerable to this aspect of the self-defeatingness charge, and so to the kind of worry I'm raising. But the problems raised by valuable practices, by the demandingness of applying the principle, and by Williams's worry about a debasing of the moral currency, which I go on to discuss, apply to this version of utilitarianismalso. See also Roger Crisp, MillonUtilitarianism (London: Routledge, 1997), 26-28, in which Crispargues that Millheld the simpler utilitarian view, notthe more complicated one.
64. Williams, Morality (Cambridge: Cambridge University Press, 1972), 96.
65. See, for example, Peter Singer, "Is Act-Utilitarianism Self-Defeating?" Philosophical Review 81 (1972): 94-104. My goal here is not to establish beyond doubt that utilitarianism is self-defeating but rather to suggest that if it is (and there are some grounds for thinking it is), this is a problem for utilitarianism. As I note, some very prominent defenders of util-itarianism have responded to the self-defeatingness charge by arguing it is on target but provides no objection to the theory, and it is this line of response that I am addressing.
66. Parfit, Reasons and Persons, 51.
67. John Stuart Mill, Utilitarianism, ed. by George Sher (Indianapolis: Hackett Publishing Company, 1979), 17.
68. Crisp, Mill on Utilitarianism, 144-45.
69. Although it's not clear to me that we can train ourselves to be good in this way.
70. Indeed, whatever normative moral theory turns out to be correct, there may be a possible world in which one cannot act rightly for the right (that is, right-making) reasons at all-that is, in which morally worthy action is impossible. Another of Stratton-Lake's examples illustrates this: Consider a world in which there is an omnipotent, evil demon whose aim is to stop good people from doing what they should in the light of the normative reasons why they should so act.... He achieves this by making it the case that if a good person ever acts from the normative reasons why she should so act, he will make it such that this action is wrong, and he tells them this. Every good person knows, therefore, that she cannot do the right thing from the normative reasons why this is right. For they know that if they are motivated to act in this way, then their actions will be morally wrong. (Kant, Duty, and Moral Worth, 18) Whatmakesthe truth of the thesis problematic from the utilitarian point of view is that if utilitarianism is correct, even the best world accessible to us from the actual world is like the world of the example, despite the absence of evil demons. The truth of utilitarian-ism combined with the truth of the Coincident Reasons The siswouldentail that we should, ideally, make ourselves, as far as is possible, into people whose actions lack moral worth-into people who don't care about or respond motivationally to what really matters.
71. According to some forms of utilitarianism, what we ought to do depends on our epistemic position: the right act is the one that maximizes expected utility, not the one that actually comes out best. It seems to me that these subjective forms of utilitarianism already concede the significance of motivating reasons since what makes them plausible is the thought that someone could be responding as well as possible to the importance of bringing about happy consequences while still, and blamelessly, failing to produce the utility-maximizing act.
72. ThanksagaintoDanielMarkovitsforpushingmetocontendwiththisresponseon behalf of the utilitarian. A worry may lurk here for the larger strategy behind my argument. That strategy was to defend an account of moral worth on the basis of considerations whose appeal was supposed to transcend disagreements about first-order ethics, and then to show that that accounthadimplausiblefurtherimplicationswhenadoptedalongsideonefirst-ordereth-ical theory: utilitarianism. But I began by saying that a morally worthy action is one that merits praise in virtue of the positive light it sheds on the moral character of its agent. In explaining the notion of moral worth I had in mind, I emphasized (in n. 4) a distinction between an action that we have instrumental reason to praise because praising it will have good consequences, and an action that merits praise in its own right and irrespective of any consideration of the consequences of actually praising it. But is this a distinction a utilitar-iancouldrecognize?Wouldn'tautilitarianhave tounderstandtheideaofmeritingpraise solely in terms of the consequences of praising? If so, it seems the notion of moral worth I defend is from the outset one a utilitarian could not accept, and my defense of it no longer appears to be neutral between competing first-order moral theories. However, I don't think a utilitarian has to understand the idea of an action's mer-iting praise solely in terms of the consequences of praising it. Utilitarians can and surely would acknowledge that the question of whether or not someone is a good person is not settled by determining whether it would be useful to call him good, just as the question of whetherornotsomeoneiskind,orclever,ortalentedisnotsettledbydeterminingwhether it would be useful to think him so. (Indeed, it seems even less plausible to think that the appropriateness of applying such labels to agents depends only on the usefulness of our applying them than to think that the appropriateness depends only on the contributions such agents maketo total utility.) Theversion of utilitarianism I considerhere isa version of act-utilitarianism, and actutilitarians, of course, understand the rightness of actions interms of the value of their consequences. But they need not and should not understand any notion of appropriateness along these lines. It's worth noting that if they did understand appropriateness or meritonly in this way, they would beforced in to implausibly revisionary accounts not just of moral worth but also, for example, of epistemic justification (is it really appro-priate to believe a proposition only if doing so would be useful?), and of meaning (does a phenomenon really merit a certain description only if describing it that way would be useful?). I was helpfully pressed on this point by an anonymous reader for the Philosophical Review.
73. This actually isn't quite right because sometimes the fact that we ought to perform an action is overdetermined by the normative reasons. And it may also be possible for the fact that we're motivated to perform an action to be overdetermined by motivating reasons (a question I won't tackle here). In cases of overdetermination, there needn't always be perfect overlap between normative and motivated reasons for an act to be fully morally worthy. In cases of normative overdetermination, it may be okay not to be moti-vated by all the normative reasons justifying an act, so long as we're motivated by norma-tive reasons sufficient to justify the act (we may in some cases, perfectly reasonably, stop weighing reasons once we've determined the right way to act). And if there are cases of motivational overdetermination, it may be okay to have some nonmoral motivations for doing the right thing, so long as we're also fully motivated by the actual normative reasons justifying the act. Anotherclarificationthathasbeentosomeextentimplicitinmydiscussionabove: perfect overlap between motivating and justifying reasons requires that a fact providing a reason does the same amount of work in motivating us to φ as it does in justifying φing. Thanks to Roger Crisp and Joey Fishkin for raising some questions that pushed me to be more precise on these points.
74. A similar point applies to the case of supererogatory actions.
75. Stephen Kearns pointed out to me the significance of such cases. He also notes a difficulty such casespose for the way I have described worthy actions at various points in this essay. I can't say, in the case of wrong but partially worthy acts, that they're partly motivated by right-making reasons (a formulation I appeal to at some points in this essay) because in the case of wrong acts motivated by genuine normative reasons, those reasons won't, in the circumstances, be right making. Nor can I say of wrong but partly worthy acts (as I have of other morally worthy acts) that the reasons why they are performed and the reasons why they ought to be performed are (partly) the same. Because, again, in this case, the reasons motivating the act, though genuinely normative, are not reasons why the act ought to be performed since it's not the case that the act ought to be performed. As I hope my discussion in the text above bears out, I think these are merely issues of proper formulation and do not seriously undermine the thesis or the account it can provide of the partial moral worth of such actions. They still provide examples of actions in which motivating and normative reasons (partially) coincide. The normative reasons in this case are pro tanto and not winning reasons. It may not be correct to say that they are reasons the act ought to be performed (though this doesn't sound too terrible); but we might certainly say they are normative reasons to perform the act.
76. Even the force of the Kantian argument for the Coincident Reasons Thesis could, I think, be felt by some non-Kantians as well. The general Kantian thought that being good is a matter of being responsive to moral reasons, which the argument is intended to make precise, certainly has a broader appeal, not least because it sits so comfortably with the very plausible idea that intellectual, or epistemic, worth is a matter of responsiveness to epistemic reasons.