In the first chapter I spoke of actions that we judge to be right and what ought to be done as being “reasonable,” or “rational” … and I contrasted the motive to action supplied by the recognition of such reasonableness with “non-rational” desires and inclinations. This manner of speaking is employed by writers of different schools, and seems in accordance with the common view and language … . For we commonly think that wrong con- duct is essentially irrational, and can be shown to be so by argument; and though we do not conceive that it is by reason alone that men are influenced to act rightly, we still hold that appeals to the reason are an essential part of all moral persuasion … . On the other hand it is widely maintained that, as Hume says, “Reason, mean- ing the judgment of truth and falsehood, can never of it- Sidgwick on self be any motive to the Will”; and that the motive to action is in all cases some Non-rational Desire, including under this term the impulses to action given by present Moral Motivation pleasure and pain. It seems desirable to examine with some care the grounds of this contention before we pro- ceed any further (ME 23).1 So opens Methods I.III. Sidgwick holds that moral judgments are claims about what it is reasonable to do. In Methods I.I he
Robert Shaver 1 “ME” refers to Henry Sidgwick, The Methods of Ethics, seventh edition (Indianapolis: Hackett, 1981). References to the first edition (London: Mac- millan, 1874) are abbreviated as “ME (1)” and to the second edition (London: Macmillan, 1877) as “ME (2).” Other abbreviations refer to the following works by Sidgwick: BB = “Bentham and Benthamism in Politics and Ethics,” in Miscellaneous Essays and Addresses (London: Macmillan, 1904); CE = “Can- ons of Evidence in Psychical Research,” Proceedings of the Society for Psychi- cal Research 6, 1889; DEP = The Development of European Polity, second edi- ton (London: Macmillan, 1913); E = “Eton,” Macmillan’s Magazine, February 1861; FT = “The Feeling-Tone of Desire and Aversion,” Mind n.s. 1, 1892; GE = “Green’s Ethics,” Mind 9, 1884; LE = Lectures on the Ethics of T. H. Green, Mr. Herbert Spencer, and J. Martineau (London: Macmillan, 1902); MS = “The Morality of Strife,” in Practical Ethics (New York: Oxford University Philosophers’ Imprint Press, 1998); OHE = Outlines of the History of Ethics, fourth edition (London:
Manitoba.
Robert Shaver Sidgwick on Moral Motivation writes of “rational procedure,” “rational precepts,” the “end Sidgwick holds (i) and (ii), rejecting (iii) [on one reading of of reasonable human action,” and “reasonable conduct.” (iii)]. Hume, on Sidgwick’s and the received view, holds (ii) When we fail to do what we see to be right, we are acting on and (iii), rejecting (i). I shall sometimes make reference to “irrational springs of action” or “non-rational inclinations” the debate over Smith as a way to illuminate Sidgwick. (ME 1-6). He also holds that these judgments about what it Second, when I write of moral judgments’ motivating, I is reasonable to do can motivate. He must, then, respond to intend either of two possibilities: the agent takes the moral Hume’s argument that reason cannot motivate. Most of judgment to be backed by his all-things-considered judg- Methods I.III is this response. ment about what he has most reason to do, and so desires to I shall clarify Sidgwick’s claim that moral judgments are perform, and performs, the action enjoined by the moral judgments about what it is reasonable to do (I). I then give judgment; or the agent does not take the moral judgment to two interpretations of his view of moral motivation (II-III): be backed by his all-things-considered judgment, and so has insofar as one is rational, moral beliefs cause desires; or inso- merely a desire (or “some motivation”) to perform the ac- far as one is rational, one has a standing desire to act rightly, tion, which is defeated by desires backed by other reasons. and moral beliefs combine with this desire to motivate. Sidgwick does not hold that moral judgments need always These views differ from the accounts often given by anti- be the winning motivation, even in rational agents. Humeans that rely on entities such as “besires”—accounts which there is no need to adopt (IV). I then give Sidgwick’s I argument against Hume, defending the view that judgments The topic is Hume’s objection that if moral judgments are about what is reasonable can motivate (V), and I reply to the claims about what is reasonable, they cannot motivate. But following objections (VI): the argument is question-begging; first the position Sidgwick is defending should be sketched. Hume can escape the argument by endorsing a standing- Following Smith, one can distinguish rationalism as a desire model of motivation; anti-Humean views cannot ac- count for obvious slack between belief and motivation; and conceptual position (that when I make a moral judgment, I Sidgwick’s strategy wrongly puts the burden on Hume. I am making a claim about what it is reasonable to do) from close with an objection I do not answer: Hume might ex- rationalism as a substantive position (that there are require- plain moral motivation as the noncognitivist does. ments of rationality that match up with moral require- 3 Two preliminaries. First, Sidgwick can be seen as offer- ments). By “rationalism” below, I mean rationalism as a ing a version of Michael Smith’s “moral problem”: conceptual position. Sidgwick does not give an explicit argument for rational- (i) Moral judgments are beliefs about what it is ism. He finds rationalism “in accordance with the common reasonable to do. view and language …. For we commonly think that wrong (ii) Moral judgments sometimes motivate. conduct is essentially irrational, and can be shown to be so (iii) Beliefs alone do not motivate.2 by argument” (ME 23). Rationalism is the position we need
2 See Michael Smith, The Moral Problem (Oxford: Blackwell, 1994) p. 12. For the same problem, see David McNaughton, Moral Vision (Oxford: Black- 3 Smith, Problem pp. 63-5. well, 1988) p. 23. 2
Robert Shaver Sidgwick on Moral Motivation arguments to reject. irrational in the sense of being subject to rational criticism. This might seem far-fetched. “Irrational” has a range of Indeed, Sidgwick distinguishes between moral judgments meanings.4 On a narrow account, irrationality requires in- and claims such as “the food is disagreeable” by noting that ternal inconsistency: I believe what I think I have decisive “the peculiar emotion of moral approbation is … inseparably reason to reject, or I do what I think I have decisive reason bound up with the conviction …that the conduct approved not to do.5 It is unlikely that wrong conduct always involves is ‘really’ right—i.e., that it cannot, without error, be disap- this sort of irrationality. On a wide account, irrationality re- proved by any other mind” (ME 27). “Reason” rather than quires only that one is subject to rational criticism.6 Sidg- “moral sense” is appropriate because “the term Sense sug- wick holds that I can be subject to rational criticism without gests a capacity for feelings which may vary from A to B internal inconsistency. For example, in rejecting the view without either being in error ... and it appears to me funda- that there are only hypothetical imperatives, he writes that mentally important to avoid this suggestion” (ME 34). Ra- “[w]e do not all look with simple indifference on a man who tionalism can be thought of as claiming that moral judg- declines to take the right means to attain his happiness, on ments, unlike simple reports or expressions of feeling, are no other ground than that he does not care about happiness. subject to rational criticism, although the criticism need not Most men would regard such a refusal as irrational” (ME 7). consist in showing an internal inconsistency.8 After arguing that various axioms, such as the superven- ience of the moral on the non-moral, pass his tests for high- II est certainty, Sidgwick comments that “I regard the appre- Sidgwick seems to claim that belief and rationality are suffi- hension … of these abstract truths, as the permanent basis of cient for moral motivation. For example, “[w]hen I speak of the common conviction that the fundamental precepts of the cognition or judgment that ‘X ought to be done’ ... as a morality are essentially reasonable” (ME 383). Reasonable- ‘dictate’ or ‘precept’ of reason to the persons to whom it re- ness lies in passing epistemic tests for clarity, self-evidence, lates, I imply that in rational beings as such this cognition and consistency within and between people; internal consis- gives an impulse or motive to action” (ME 34). These cogni- tency is not enough. Similarly, Sidgwick often describes cus- toms, biases, prejudices, and privileges as being “irrational,” irrationality. But Sidgwick goes on to distinguish two “grades” of inconsis- where this seems best glossed as “against the balance of rea- tency: “Our impulses to action may be such as not to conflict, and yet not har- sons.”7 All this suggests that for Sidgwick, wrong conduct is monized or systematized: or they may be actually opposed and conflicting.” As an example of the latter, Sidgwick suggests “when desiring an end we decline to
4 For a useful discussion, see T. M. Scanlon, What We Owe to Each Other take the necessary means … or if aiming generally at a kind of objects or results, we shew ourselves arbitrarily indifferent to a particular individual or instance.” (Cambridge, MA: Belknap, 1998) pp. 25-33. 5 As an example of lack of system, Sidgwick suggests maxims “that cannot be Sidgwick sometimes uses “irrational” in this way (e.g., ME 5, 23, 24, 37, bound together and firmly concatenated by means of some one fundamental 236, BB 160, PD 670, LE 23, 29). principle” (ME [1] 25-6). “Inconsistent,” then, turns out to mean something 6 For this account, see Derek Parfit, Reasons and Persons (Oxford: Oxford like “subject to rational criticism.” University Press, 1984) p. 119. 8 For a very different, non-rationalist reading of Sidgwick, see David Brink, 7 ME 269-70, PPE 390, OHE 34-5, E 297, CE 47, DEP 154. In the first edi- “Sidgwick’s Dualism of Practical Reason,” Australasian Journal of Philosophy tion of the Methods, Sidgwick writes that “one meaning of ‘irrational’ as applied 66, 1988. For a defence of the rationalist reading against Brink, see my Rational to conduct is ‘inconsistent.’” This might seem to favour the narrow account of Egoism (Cambridge: Cambridge University Press, 1999) pp. 78-9. 3
Robert Shaver Sidgwick on Moral Motivation tions “in so far as they relate to conduct on which any one is desire; the difference between Sidgwick and Hume is not deliberating ... are accompanied by a certain impulse to do over whether desires are present, but rather over how reason the acts recognised as right” (ME 77). If Sidgwick did not influences desire. Similarly, Sidgwick describes Hume as hold that belief and rationality are sufficient for motivation, holding that “the motive to action is in all cases some Non- there would be no need for an argument against Hume. rational Desire” (ME 23). The “non-rational” qualification Sidgwick does not, however, hold that beliefs motivate suggests that Sidgwick allows other, “rational” desires to directly, without desires. Cognitions give or are accompa- play a role in motivation. (For Sidgwick, an “irrational de- nied by “motives” or “impulses.” When reviewing Methods sire” is “a desire directed to an action which we at the same I.III, Sidgwick writes of “the impulse that prompts us to time judge to be bad or incompatible with our highest good: obey the dictates of Reason” (ME 39). Later he claims that e.g. desire of the pain of another human being, in the case of “we expect that the judgments of moral goodness or bad- malevolent impulse—of escape from a post of duty and ness, passed either by the agent himself or by others, will— danger, in the case of a cowardly impulse” [LE 23; also LE by the fresh motive which they supply on the side of vir- 29, 31]. It is a desire “impelling us to volitions opposed to tue—have an immediate practical effect in causing actions to our deliberative judgments” [ME 24]. Non-rational desires be at least externally virtuous” (ME 426-7). A desire is a “felt are desires that produce actions “without our having judged impulse ... to actions tending to the realisation of what is de- such actions to be either right or wrong” [ME 24]. Rational sired” (ME 43n2). “Motives” are “desires that we feel for desires are presumably desires directed to actions which we some of the foreseen consequences of our acts” (ME 60; also judge to be good. One can choose “without rational choice— 202-3, 362-3). Throughout, Sidgwick uses “motive,” “im- i.e. not as good” [LE 31].) pulse,” and “desire” interchangeably (e.g., ME 43ff, 362ff).9 Second, Sidgwick accepts Green’s description of him as This picture is reinforced by passages elsewhere. holding that “the opposition between Desire, Reason, and First, after giving Hume’s account of how reason can Will” lies “‘between different desires, of which reason how- interact with desire, Sidgwick objects that “the account just ever ... supplies the object to the one, while some irrational given of the influence of the intellect on desire and volition appetite is the source of the other; the will being the arbiter is not exhaustive” (ME 25). This implies that for Sidgwick, which determines the action according to the rational or irra- as for Hume, reason has a role in motivation by influencing tional desire’“ (LE 22). Rational action is action on the basis of (rational) desire.
9 Usually for Sidgwick a desire is something “felt.” It is a state “in which Third, against Leslie Stephen, Sidgwick claims that “the we are conscious of an impulse to get out of the present state into a future one” feelings that normally cause action are not pleasures and (FT 98). But Sidgwick does not always require an occurent feeling: “not un- 10 frequently during long processes of work for remote ends, the desire of the end, pains as such, but desires and aversions.” It is, for exam- while remaining sufficiently strong to supply the requisiste impulse to action, ceases to have a perceptible character as feeling; we only infer its presence from 10 Sidgwick does not explain the “normally.” Perhaps he is thinking of re- the actions that it stimulates, and from the satisfaction that follows on the at- flex actions, which may be caused by a feeling that is not a desire or aversion. tainment of some intermediate end which has no significance for us except as a A similarly passing comment, restricted to voluntary actions, occurs in the step towards the ultimate end” (FT 99). The feeling of desire “may temporarily Methods: “I shall use the term Motive to denote the desires of particular results, cease to be present in … consciousness,” though it “will normally recur at brief believed to be attainable as consequences of our voluntary acts, by which desires intervals during the process” (ME 55). we are stimulated to will those acts” (ME 363). 4
Robert Shaver Sidgwick on Moral Motivation ple, aversion that “normally accompan[ies] the moral judg- Smith’s characterisation of “the Humean theory” claims ment that an act is wrong.” Stephen holds that if this aver- that motivating reasons are constituted by belief-desire pairs sion is not an aversion to the pain of remorse or pain caused and that “for any belief and desire pair that we imagine, we to others, the moral judgment “cannot, any more than any can always imagine someone having the desire but lacking other operation of the intellect, move the will, apart from the belief, and vice versa.”12 This does not rule out holding emotion.” Sidgwick replies that “[t]his seems to me a rather that beliefs (sometimes) cause desires with the pair then mo- subtle psychological question; but at any rate I should say tivating. Indeed, this is Smith’s own account of moral moti- that in the case of moral judgments I am frequently not con- vation.13 scious of any emotion other than what is implied in the mere Smith’s characterisation does not fit Hume. For Hume consciousness that the judgment carries with it a motive” would lose his main anti-rationalist argument were he to (RS 573). Sidgwick holds that desires are (at least normally) agree that moral beliefs sometimes cause desires (or, more needed for action and that moral judgments carry desires precisely, “motivating passions”), without the contribution with them. of pre-existing desires. He could not argue that rationalists Fourth, those who hold that beliefs motivate directly usu- fail to capture the connection between moral judgment and ally give motivation by duty as their example. But Sidgwick motivation.14 Nor could he argue that, since reason is “inac- seems uninterested in taking motivation by duty as motiva- tive,” it cannot conflict with the passions.15 tion without desire. In his argument against the view that I Sidgwick must refute Hume’s anti-rationalist argument. desire only my own pleasure, he notes as a counter-example To do so, it is sufficient to argue that moral beliefs some-
“the love of virtue for its own sake, or desire to do what is right as such” (ME 52). Elsewhere he sometimes characteris- 12 Smith, Problem p. 7; also 119-20. 13 Smith, Problem pp. 73, 160, 179, 180, 181, 193, 213, 214. This fits es this motivation as “the desire ... of doing what is right as Smith’s argument for the Humean theory, which simply identifies motivating such, realising duty or virtue for duty’s or virtue’s sake,” “a reasons with desires (Problem 116). The argument says nothing about the pure desire to act rightly” (ME 204, 206). He even character- causes of desire. (Smith sometimes interprets the Humean theory as implying that the belief-desire pair is the “source” of motivation [Problem 93]. He ises Kantian moral motivation as “pure Regard for Moral glosses the Humean theory as denying that beliefs “posses[s] the causal power to Law, Desire to do Right as Right” (ME [1] 346).11 produce behaviour” [Problem 103]. When presenting the case against identify- ing valuing with believing, he treats it as an apparent consequence of the Humean theory that “beliefs cannot produce actions” [Problem 137]. But these claims must be read in light of Smith’s account of moral motivation: the point III must always be merely that beliefs without desires cannot produce actions.) 14 David Hume, A Treatise of Human Nature, second edition, ed. L. A. It is important to see precisely what Sidgwick rejects in Selby-Bigge and P. H. Nidditch (Oxford: Clarendon, 1978) pp. 457-8; An En- Hume, given that he agrees that desires are needed for moti- quiry Concerning the Principles of Morals, third edition, ed. Selby-Bigge and Nidditch, pp. 172-3, 294. For the same point, see Parfit, “Reasons and Motiva- vation. tion” Proceedings of the Aristotelian Society sup. 71, 1997, 105-6; Elizabeth
Radcliffe, “Hume on the Generation of Motives: Why Beliefs Alone Never Mo- 11 The corresponding passage in the seventh edition reads “pure regard to tivate,” Hume Studies 25, 1999, 101-3, 117; Barry Stroud, Hume (London: duty or choice of Right as Right” (ME 366; also GE 185). There Sidgwick also Routledge and Kegan Paul, 1977) p. 157. For an opposed reading of Hume, see describes Kantian motivation as “disinterested obedience to Law as such, with- Rachel Cohon’s forthcoming Hume’s Virtue: Feeling and Fabrication, chs. 2-3. out the intervention of sensible impulses” (ME 513). 15 Hume, Treatise pp. 413-5. 5
Robert Shaver Sidgwick on Moral Motivation times cause desires, without the contribution of pre-existing model: we have “a determination to pursue whatever con- desires. (Call this the “causal” model.)16 Since Sidgwick duct may be shown by argument to be reasonable,” a “genu- takes moral beliefs to be beliefs about justifying reasons, he ine impulse to conform to the right rules of conduct,” and “a would describe one whose moral beliefs fail to cause the disinterested aversion to unreason” (ME 5, 101, 447). The relevant desire as imperfectly rational; rationality lies (in passages concerning a “desire to do what is right as such” 17 part) in the functioning of this causal mechanism. His cited earlier also suggest this model (although they are typi- claim, then, is that moral beliefs and rationality are sufficient cally offered as descriptions of Stoic or Kantian views, rather for motivation. Hume must deny the possibility of the 18 than as Sidgwick’s own account of ordinary moral motiva- causal mechanism. I take this, rather than the conceptual tion). But Sidgwick’s wording usually suggests the causal separability of particular beliefs and desires, to be the issue. model: that the cognition “gives an impulse or motive” or is Sidgwick may be tempted by a different position. He “accompanied by” an impulse or “carries with it a motive” might claim that the needed desire is produced by the belief or “supplies” a “fresh motive” suggests that the impulse is because, insofar as one is rational, one has a standing desire not already present.20 More importantly, if Sidgwick held to do what one takes to be right. (Call this the “standing de- the standing-desire model, there would be little issue with sire” model.)19 Some passages suggest the standing-desire Hume over whether reason can motivate. Hume and Sidg-
16 For the causal model, see Parfit, “Reasons and Motivation” 105; Radcliffe wick would agree that motivation requires a pre-existing de- 101-3, 117; Rachel Cohon, “Hume and Humeanism in Ethics,” Pacific Philoso- sire. The issue between them would instead concern phical Quarterly 69, 1988, 103, 105-6, 108; Jonathan Dancy, Moral Reasons whether reason just supplies “ideas of actual or possible (Oxford: Blackwell, 1993) pp. 8-9, 12. Stroud may suggest it but is probably better read as identifying desires with a disposition that need not be “before the facts,” since Sidgwick thinks reason supplies moral judg- mind.” See Stroud, pp. 168, 179. ments and that moral judgments do not convey ideas of ac- 17 I put the point in terms of perfect rationality (or “one feature of rational- ity”) because one might argue that when my belief about what I have reason to tual or possible facts (ME 25). This is a possible reading of do is false, it is sometimes more rational for me to have a desire opposed to this Methods I.III, but it makes Sidgwick’s concern with motiva- belief than for me to have a desire consistent with this belief. (For this point di- tion at the start of the chapter misleading.21 rected against Smith, see Nomy Arpaly, “On Acting Rationally Against One’s Best Judgment,” Ethics 110, 2000, 493-501.) One can agree but still claim (as Arpaly admits) that having a desire opposed to my belief about what I have jus- impulse or desire to do what they judge to be right …. It is … one motive tifying reason to do is a falling off from perfect rationality. among others” with “in most human beings the belief that a certain course of ac- 18 For controversy over Hume’s actual position, see Radcliffe, and Cohon, tion is right … is pro tanto a motive for doing it …. We are perfectly familiar Virtue chs. 2-3. In the context of the dispute with Sidgwick, Hume might admit with this motive” (C. D. Broad, Five Types of Ethical Theory [London: Rout- the causal mechanism but deny that this mechanism is (even partially) constitu- ledge and Kegan Paul, 1930] pp. 146, 179). tive of rationality, perhaps by invoking an account of rationality that is inappli- 20 This is hardly decisive, however. As one referee notes, a particular moral cable to the fit between beliefs and desires. On such an account, weakness of belief such as “one ought to be brave” could be said to give or supply a fresh will would not count as irrationality. This, however, makes his disagreement motive in the sense that it gives direction to the standing desire to do what one with the rationalist semantic—what the rationalist, and indeed most of us, con- takes to be right. sider rationality is sufficient, when combined with beliefs about reasonable ac- 21 Hastings Rashdall, after quoting much of Methods I.III, also suggests both tion, to motivate. the causal and standing-desire models, but like Sidgwick seems to favour the 19 For a standing-desire model as a rationalist reply to Hume, see Jonathan former. He writes, for example, that the “recognition of the thing as right is ca- Harrison, Hume’s Moral Epistemology (Oxford: Clarendon, 1976) pp. 13-15. pable of producing an impulse to the doing of it”; the desire to do what we see to In his initial exposition, C. D. Broad seems to attribute the model to Sidgwick, be our duty “may be created by the Reason which recognizes the rightness”; “an although his later gloss is more ambiguous: compare “[h]uman beings have an act may be done … from a desire which is created solely by our conviction that 6
Robert Shaver Sidgwick on Moral Motivation The causal model may represent a change in position. In the standing desire, then on this model I might, with perfect the first edition of the Methods, in reply to the contention that rationality by my own lights, make a moral judgment with- “Reason cannot be a spring of action, as it must always be out the desire to act on it, just as I might believe there are an- Feeling that stimulates the Will,” Sidgwick claims that chovies on a pizza without being rationally compelled to de- sire the pizza. This could not appeal, since for Sidgwick (in no one is competent or really concerned to maintain that all editions), if I think x ought to be done, then it follows that the apprehension of duty is a state of consciousness which occurs without any emotional element. Hence we I think I have a justifying reason for doing x. I could not need not ask whether a mere cognition can act upon the then, with perfect rationality by my own lights, make the Will and prompt to action. It is enough if it be granted judgment without the desire. That would be to hold that I that there exists in all moral agents as such a permanent have a justifying reason to do x without desiring to do x—a desire ... to do what is right or reasonable because it is failure of perfect rationality.23 (Perhaps this explains why such: so that when our practical reason recognises any course of conduct as right, this desire immediately im- Sidgwick cut the first-edition passage.) pels us ... towards such conduct .... We may assume Sidgwick could, however, adopt an augmented version then as generally admitted that the recognition of any of the standing-desire model. He could add that it is part of action as reasonable is attended with a certain desire or the concept of a rational agent that she have a desire to do impulse to do it: and that in this sense the Reason may what is right.24 For a rationalist, this is very plausible: since 22 be affirmed to be a spring of action (ME [1] 22, 27, 28). thoughts about what is right are thoughts about what is ra- tional, the claim becomes that it is part of the concept of a ra- This explains moral motivation by finding it unnecessary to tional agent that she have a desire to do what is rational. hold that moral judgments sometimes motivate by them- One worry about combining rationalism and a standing- selves. Moral agents have a standing desire to do what they desire model is raised by J. L. Mackie. Against Jonathan take to be right. If this is satisfactory, the argument of the Harrison’s suggestion that rationalists can answer Hume by later editions, given in V below, is unnecessary. Not surpris- adopting a standing-desire model, Mackie objects that ingly, it is absent from the first edition; a rough version of it appears, and the quoted passage from the first edition dis- [i]t is not merely that it is linguistically odd to use words appears, in the second edition (ME [2] 25-31). like “right” and “wrong” with no prescriptive force—to Sidgwick had good reason to abandon the standing- say, for example, “X is right and Y is wrong, but of desire model as it is stated in the first edition. For if I lack 23 For a similar argument, see Smith, Problem pp. 62, 150-2. 24 One might read the first edition as saying just this by supposing that a certain course is intrinsically right or reasonable.” Against Kant, he requires Sidgwick makes it part of the concept of a moral agent that she desires to do desires for motivation. See Rashdall, The Theory of Good and Evil, second ed. what is reasonable. Sidgwick, however, goes on to appeal to “experience” to (London: Oxford University Press, 1924) v. i, pp. 104, 106, 121, 106n2 (also show the presence of the desire to do what is reasonable, suggesting that he 140-1). takes the desire to be conceptually independent of the concept of a moral agent 22 J. B. Schneewind also notes that Sidgwick seems to have changed his po- (ME [1] 27)). Alternatively, one might note that Sidgwick sometimes runs to- sition (though Schneewind says very little about what the post-first-edition posi- gether the cognition that x ought to be done with the impulse or desire to do x tion is). See Schneewind, Sidgwick’s Ethics and Victorian Moral Philosophy (e.g., ME 39), perhaps because he thinks the cognition causes a desire, as in the (Oxford: Clarendon, 1977) p. 235; for Schneewind’s terse comment on the later causal model. On this interpretation, the desire is not, as the long quotation view, see p. 207. seems to suggest, independent of the cognition. 7
Robert Shaver Sidgwick on Moral Motivation course it is entirely up to you whether you prefer what is that she have a desire to do what is right. If Sidgwick ac- right to what is wrong.” … What is more important … is cepted this standing-desire model, his argument against that Clarke, Butler, and many others are concerned to Hume’s view of moral motivation in Methods I.III would be defend the metaphysical view which is represented by 27 the way in which such moral terms combine a descrip- unnecessary. tive logic with a prescriptive force, namely that there are objective requirements or categorical imperatives in the IV nature of things. Harrison’s suggestion would abandon 25 Today, unitary states that are both belief- and desire-like this claim. are the usual opponents of the Humean theory.28 (Mackie, Since the standing-desire model makes claims only about quoted above, also sees them as the historical opponent.) motivation, Mackie must, in arguing that Harrison loses pre- But Sidgwick need not attack Hume by subscribing to such scriptivity, be identifying prescriptive force with motivating unHumean states. force. But rationalists should deny this identification. Con- One reason for favouring unHumean states is phenome- sider an analogy: say that given a conditional and its conse- nological: one might take oneself to be motivated to act on quent, I infer the antecedent. It is up to me whether I do so moral judgments without noticing any desire. The Humean can avoid this by rejecting a phenomenological conception of in the sense that I may or may not be motivated to do so. 29 But it is not up to me whether I do so in the sense that doing desire. More importantly, this does not seem to be Sidg- so would be a mistake. Rationalists, then, can answer Mack- wick’s thought: when he discusses the phenomenology of ie by accepting that motivation is contingent on a standing acting on moral judgments, he never stresses the absence of desire without accepting that justification is contingent on a desires, and usually notes their presence. 26 standing desire. 27 Smith claims that a standing desire to do what is right would make non- I conclude that Sidgwick opposes Hume’s view of moral derivative moral concerns into derivative ones, derived from this desire, and motivation by adopting the causal model, but that he could hence would give a psychologically false and morally unattractive picture of moral motivation. One might think that Sidgwick has similar reservations: also do so by adopting a standing-desire model, augmented “there is a certain excellence in services springing from spontaneous affection by the claim that it is part of the concept of a rational agent which does not attach to similar acts done from pure sense of duty” (ME 345). But Sidgwick’s worry seems to concern action on the basis of any moral con-
ception of the situation; it is not a worry about the particular description of 25 J. L. Mackie, Hume’s Moral Theory (London: Routledge and Kegan moral motivation given by the standing-desire model. For Smith’s argument, Paul, 1980) p. 55. See also pp. 47-8. see Smith, Problem pp. 75-6; “The Argument for Internalism: A Reply to 26 Stephen Darwall gives a different argument against Sidgwick’s holding a Miller,” Analysis 56, 1996, 180-3; “In Defense of The Moral Problem,” Ethics standing-desire model. He thinks Sidgwick’s reason for rejecting non-normative 108, 1997, 112-15. For some of the (numerous) replies to Smith, see Terence D. analyses of moral judgments is that moral judgments, unlike non-normative Cuneo, “An Externalist Solution to the ‘Moral Problem’,” Philosophy and Phe- judgments, are “intrinsically action guiding.” If so, Sidgwick cannot hold that nomenological Research 59, 1999; Russ Shafer-Landau, “Moral Judgment and the action-guidingness comes from a standing desire rather than from the moral Moral Motivation,” Philosophical Quarterly 48, 1998; Sigrun Svavarsdottir, judgment itself. However, as noted, Sidgwick’s rejection of non-normative “Moral Cognitivism and Motivation,” Philosophical Review 108, 1999. analyses relies on pointing out specific problems encountered by specific analy- 28 I have in mind not only besires but also, for example, Jonathan Dancy’s ses—for example, apparently opposed judgments turn out to be consistent— motivating representations that are belief-like in that their essence is not to mo- rather than on any appeal to intrinsic action-guidingness. See Darwall, “Intui- tivate, but desire-like in that they “motivate in their own right.” See Dancy pp. tionism and the Motivation Problem,” in Philip Stratton-Lake ed., Ethical Intui- 23-4 (and, in general, ch. 2). tionism (Oxford: Clarendon, 2002) pp. 269-70. 29 See Smith, Problem pp. 105-9. 8
Robert Shaver Sidgwick on Moral Motivation Since the phenomenological argument for unHumean erations can provide desire-independent justifying reasons— states is the most prominent, more should be said. Consider desire-independent constraints. These can be and feel au- David McNaughton’s statement of the argument: thoritative in that they limit the options I can choose while remaining rational.32 Reflection on what it is like to realize that one is morally Jonathan Dancy gives a different argument. We should required to act in a certain way will ... support [the] con- favour unHumean states over the causal model since on the tention that a moral belief can be sufficient to motivate an agent .... Once we are aware of such a requirement causal model, our choice of action seems to be constrained by that rec- ognition .... In particular, the claims that morality makes sometimes beliefs need the independent contribution of on us appear to be quite independent of our desires— desire, and sometimes they don’t. This seems to mean they may even conflict with what we want. On the be- that there are two sorts of beliefs, Humean and non- lief-desire theory no sense can be made of this concep- Humean. But that cannot be quite right, because the tion of a moral demand. Moral ... considerations can very same beliefs which on one occasion need an inde- only provide an agent with reason to act if he has ap- pendent non-cognitive supplement if there is to be an ac- propriate desires .... [Thus Humeans] must claim that tion, can elsewhere figure in a state which needs no such the apparent authority of moral demands is an illusion.30 supplement at all. Beliefs are not carved up into two sorts, those that can motivate alone and those that need This is unconvincing for two reasons. some help. It is the same sort of belief all the time. But First, unHumean states are not needed to avoid depend- in that case how is it that the extra provided by a desire can on other occasions be provided by a belief? ence on antecedent desires; the causal model allows for mo- tivation without antecedent desires. Kant himself often Dancy concludes that the anti-Humean should not concede seems to hold that respect is a feeling caused by a moral be- that desires are ever needed: the causal model “will be un- lief; the feeling causes a desire (or “interest”) which then mo- able to retain its insistence that purely cognitive motivation 31 tivates. is possible if it once yields to the pressure to admit Humean Second, Sidgwick can, without embracing unHumean motivation as a possibility.”33 states, say that moral demands are independent of my de- The causal model can be defended. On it, one need not sires in that moral demands concern what it is reasonable to agree that the same beliefs sometimes need an independent do, and what it is reasonable to do need not depend on what desire to motivate and sometimes do not. Nor need one I desire. (My desires are exclusively relevant only if the best agree that the extra is provided by more beliefs. The beliefs normative theory is the instrumental theory.) Moral consid- that do not need independent desires, but instead cause de- sires, are beliefs about what it is reasonable to do. They
30 McNaughton pp. 48-9. cause desires not everywhere, but only insofar as one exhib- 31 For evidence and discussion, see Mark Timmons, “Kant and the Possibil- ity of Moral Motivation,” Southern Journal of Philosophy 23, 1985, pp. 377-98; 32 For this second strategy against McNaughton, see W. K. Frankena’s Richard McCarty, “Kantian Moral Motivation and the Feeling of Respect,” (much earlier) “Obligation and Motivation in Recent Moral Philosophy,” in Journal of the History of Philosophy 31, 1993, pp. 421-35; McCarty, “Motiva- Frankena, Perspectives on Morality, ed. K. E. Goodpaster (Notre Dame: Uni- tion and Moral Choice in Kant’s Theory of Rational Agency,” Kant-Studien 85, versity of Notre Dame, 1976) pp. 51-60. 1994, pp. 15-31. 33 Dancy p. 21. 9
Robert Shaver Sidgwick on Moral Motivation its that feature of rationality that requires consistency be- V tween one’s beliefs about what it is reasonable to do and Sidgwick’s reply to Hume begins by noting that “[e]very one’s desires; rationality is the extra. The beliefs that need one, I suppose, has had experience of what is meant by the independent desires to motivate—such as “there are ancho- conflict of non-rational or irrational desires with reason.” vies on the pizza”—do not cause desires because the ration- For example, we feel “bodily appetite prompting us to in- ality of the believer does not combine with their content to dulgences which we judge to be imprudent, and anger cause a desire. More generally, although it might be myste- prompting us to acts which we disapprove as unjust or un- rious to carve beliefs into two sorts according to their effects kind” (ME 23-4). Sometimes, our judgment that we ought simpliciter, there is nothing mysterious about carving beliefs not to act imprudently or unjustly stops us from acting im- into two sorts according to how their different contents react prudently or unjustly. On the face of it, this is an example of 34 in the presence of another property (rationality). reason motivating. I conclude that there is no case for adopting unHumean Sidgwick notes that there is a Humean tactic for handling states, and so no reason to reject Sidgwick’s anti-Humean this potential counter-example. The Humean understands position on the ground that it does not involve such states. the moral judgment as describing facts about either conse- quences of action or means to ends that combine with pre- existing desires. Thus
34 One referee rightly notes that in places, Sidgwick does not mention ra- tionality; beliefs seem to do all the work. (i) In “The Morality of Strife,” Sidg- (1) I desire to hit Jones. wick writes “[s]uppose, for instance, that every one who is liable to drink too (2) I hold that I ought not to hit Jones. much had clearly present to his mind, in the moment of temptation, the full Therefore I do not hit Jones. amount of harm that his insobriety was doing to his bodily health, his reputation, his means of providing for those dependent on him; some, no doubt, would drink all the same, but the great majority of those not yet in bondage to the un- becomes (say) natural craving would draw back. Suppose, again, that any one who is wronging a neighbour saw, as clearly as any impartial judge or friend would see, the viola- tion of right that he is committing; surely only a thoroughly bad man would per- (1) I desire to hit Jones. sist in his wrong-doing” (MS 49). Sidgwick does not mention rationality here, (2)’ I believe that not hitting Jones is a necessary but one can read the reference to “bondage” as suggesting that the belief moti- vates only given the absence of the practical irrationality that addiction brings. means to satisfying my desire to use Jones’s This does not cover the case of the man wronging his neighbour. But Sidgwick swimming pool this summer. adds that, thoroughly bad men aside, one wrongs “from a mixture of intellectual (3) I desire to use Jones’s swimming pool more muddle with passionate impulse or selfish negligence” (MS 49). This is com- patible with thinking that wrong-doing is the result of either mistaken moral be- than I desire to hit Jones. liefs or the sort of practical irrationality suggested by “passionate impulse” and Therefore I do not hit Jones. “negligence.” (ii) In “Unreasonable Action” (discussed briefly in VI), Sidgwick again emphasises that practical irrationality stems from one’s beliefs’ not being clearly present to the mind. But again, this is consistent with holding that ra- On the Humean view, “the experience which is commonly tionality plays a role; irrationality lies in not keeping one’s belief clearly present described as a ‘conflict of desire with reason’ is ... more to the mind. Moreover, Sidgwick does claim that irrational action is possible even when the belief is clearly present to the mind; here something beyond the properly conceived as merely a conflict among desires and belief is needed as an explanation (UA 139-42). 10
Robert Shaver Sidgwick on Moral Motivation aversions; the sole function of reason being to bring before ments are judgments about what is reasonable that some- the mind ideas of actual or possible facts, which modify in times motivate.37 the manner above described the resultant force of our vari- ous impulses” (ME 25). VI Sidgwick rejects this account of the conflict. Moral or I shall consider four objections. prudential judgments “have some ... influence on volition,” (A) There is an obvious objection. The Humean could re- but “cannot legitimately be interpreted as judgments re- fuse to admit Sidgwick’s example of a motivating moral specting the present or future existence of human feelings or judgment as a case of reason motivating. For he would re- facts of the sensible world.” Sidgwick’s argument for this ject Sidgwick’s identification of moral judgments with judg- negative claim consists in “showing the inadequacy of all at- ments of reasonableness. If this identification were rejected, tempts to explain” moral or prudential judgments that do the Humean could agree that moral judgments can motivate not take them to be different from “all notions representing without accepting that reason can motivate. facts of physical or psychical experience” (ME 25). Hence The objection can be put differently. Sidgwick’s motive Sidgwick argues that, say, “one ought to tell the truth” can- for rejecting the view that reason cannot motivate is his wish not be analysed by any such representative claim. He pro- to maintain that moral judgments are judgments of reason- ceeds by examining candidate analyses such as “truth-telling ableness (given that he also believes that moral judgments is an efficient means to some end” or “truth-telling excites a motivate). It begs the question, in defence of the belief that feeling of approbation in me” or “not truth-telling will be moral judgments are judgments of reasonableness, to as- penalised by God.” He rejects each. The instrumentalist an- sume that moral judgments are judgments of reasonable- alysis leaves out evaluations of ends; the subjectivist analysis ness. leaves out disagreement; the voluntarist analysis leaves out Sidgwick has a reply. For he can run his argument with- questions of the rightness of the penaliser (ME 26-31).35 out supposing that moral judgments are judgments of rea- The argument thus runs as follows: Humean analyses of sonableness. Rather than offer moral judgments as motivat- (2) in the manner of (2)’ are unsatisfactory;36 hence the Hum- ing, he can offer judgments of reasonableness as motivating. ean cannot account for some cases in which reason seems to ([2] could be replaced by “I hold that hitting Jones is unrea- motivate; hence we should reject the claim that reason does sonable.”) He can then argue against the Humean strategy not motivate; hence we can keep the claim that moral judg- of handling potential counter-examples to the claim that rea- son cannot motivate by analysing judgments of reasonable- 35 I leave aside whether these famous arguments succeed. My focus is the role these arguments play against Hume and in favour of a particular account of ness as specifying means to desired ends or facts that cause moral motivation, rather than the soundness of the arguments themselves. further desires. The same arguments that block this analysis 36 G. C. Field, who directs Hume’s argument against Sidgwick, objects that “the alternatives given [by Sidgwick] obviously do not nearly exhaust the possi- for moral judgments are applicable to judgments of reason- ble rival accounts.” He goes on to analyse judgments of goodness in terms of desires everyone would have given full information. Sidgwick would reply that 37 For a different reconstruction, see John Deigh, “Sidgwick on Ethical no such analysis is plausible unless it contains normative concepts, but I put Judgment,” in Essays on Henry Sidgwick, ed. Bart Schultz (Cambridge: Cam- aside that (well-worn) debate here. See G. C. Field, Moral Theory (London: bridge University Press, 1992) pp. 242-3, 251-8. I criticise Deigh in “Sidg- Methuen, 1966) (first published 1921) p. 61 and chapter 11. wick’s Minimal Metaethics,” Utilitas 12, 2000, 275 n50. 11
Robert Shaver Sidgwick on Moral Motivation ableness. If “x is reasonable” is analysed as “x is an efficient Moreover, one can say this without adding that rational means to some end,” we cannot claim that ends are reason- agents must have a desire to do what is right. One can al- able; if “x is reasonable” is analysed as “x excites approba- low, as the rationalist does not, that I can fail to be motivated tion in me,” parties disagreeing over reasonableness do not by my moral beliefs without any irrationality by my own really disagree; if “x is reasonable” is analysed as “acts other lights. than x will be penalised by God,” we cannot say that some However, if a Humean moves to this model to avoid divine penalties are reasonable.38 Sidgwick can conclude analysis of moral judgments, the objection that rationalists that since the Humean strategy of handling potential coun- cannot capture moral motivation is lost. For on this model, ter-examples fails, there is no objection to thinking that rea- thoughts such as “I ought not to hit Jones” can motivate, son can motivate, and so no objection to thinking that if mor- whatever gloss one gives to them, by combining with a desire al judgments are claims about what is reasonable, moral to do what one thinks is morally right. The analysis of the judgments can motivate. moral judgment is irrelevant to this explanation of motiva- (B) A Humean might object that no analysis of moral tion (as it is not, for example, on an account according to judgments is needed to explain Sidgwick’s cases. Hume ar- which moral judgments describe or express one’s desires.) gues that we confuse “calm desires” such as “the general The rationalist merely gives a particular gloss—thinking “I appetite to good, and aversion to evil, consider’d merely as ought not to hit Jones” is thinking “hitting Jones is unrea- such” with the calm sensation of reasoning.39 For Hume, the sonable.” And if there is no objection to the rationalist gloss, “general appetite to good” is a desire for pleasure, and so there seems no reason to deny that failure to be motivated moral judgments would need to be connected to pleasure, by moral beliefs manifests irrationality. perhaps by the sort of analysis Sidgwick rejects. But the (C) Anti-Humeans face a standard objection: two people strategy of combining a standing desire with a moral judg- can have the same beliefs yet different motivation, and so ment shows how one could avoid any analysis: moral moti- their beliefs alone cannot be what motivate them.40 vation can be explained by citing a desire to do what one Sidgwick can allow, however, that the same beliefs can takes to be right. For example, 40 For the problem, see Dancy pp. 12, 22; Cohon, “Hume and Humeanism” 111-12. Dancy’s reply is that a belief can motivate in one case but not in an- other, without forcing one to say that a desire must explain this difference and so (1) I desire to hit Jones. must form part of the explanation in both cases. For (i) the absence of factors (2) I hold that I ought not to hit Jones. that would have blocked a causal chain are not typically seen as parts of the (3)’ I desire to do what I believe to be morally chain, and so need not enter into an explanation; and (ii) the explanation need not cite a desire at all—Dancy (like Cohon) gives, as examples of blocking fac- right more than I desire to hit Jones. tors, carelessness, inattention, despair, drunkenness, neurophysiological disor- Therefore I do not hit Jones. ders, and depression (Dancy pp. 22-6). (i) is controversial: one might see this as a merely pragmatic point concerning what we wish to focus on, and so as leaving open the possibility that desires still play a key role (even if in the back- ground). (ii) can be read as making the point Smith and Sidgwick make: moral 38 Allan Gibbard gives a similar argument, though in support of a noncogni- beliefs motivate practically rational agents. (ii) does not, however, justify think- tivist analysis of “x is reasonable.” See Gibbard, Wise Choices, Apt Feelings ing that beliefs can suffice for motivation without desire, since it is plausible to (Cambridge, MA: Harvard University Press, 1990) ch.1. think that depression and the like prevent motivation by preventing the desires 39 Hume, Treatise p. 417. usually caused by moral beliefs. 12
Robert Shaver Sidgwick on Moral Motivation cause different motivations. One agent might be irrational, even to find much discussion of whether beliefs can cause and so the belief either fails to produce a desire or produces desires.) If so, present-day Humeans need not resist Sidg- a desire that is defeated by other desires. Since Sidgwick’s wick’s description of cases of reason motivating as cases in claim is that what suffices for desire are moral beliefs and which beliefs cause desires. that feature of rationality that requires consistency between A different reply is that, if necessary, Sidgwick can run my beliefs about what it is reasonable to do and my desires, his argument without the starting assumption that there are the explanation of the failure of motivation is not restricted apparent cases of reason motivating. Sidgwick can reject, as to the belief. Sidgwick thinks that in most cases of irrational- he does, particular redescriptions of what the moral judg- ity, “seductive feeling” makes me temporarily forget my ment conveys; he can add that he and Hume agree that practical judgment, or makes me think it inapplicable to the moral judgments motivate; he can conclude that since the case at hand, or makes me pay attention to only some of the motivating power of moral judgments does not come from relevant reasons; in the latter two cases, I can be “obscurely (say) their telling us means to satisfy our desires, their moti- conscious” that I am making a mistake. These are all effects vating power seems to come simply from their being the sort on my beliefs. But Sidgwick does allow that in rarer cases, of judgment they are: they are judgments that cause desires my practical judgment is “clearly present in [my] conscious- insofar as one exhibits that feature of rationality that re- ness” yet I am not motivated (UA 139-42).41 quires consistency between one’s beliefs about what it is rea- (D) Sidgwick starts by assuming that there are apparent sonable to do and one’s desires. Moral judgments also seem cases of reason motivating. My judgment that I ought not to to be judgments about what is reasonable, given the way in act out of anger sometimes prevents me from so acting. which we argue over them. If so, moral judgments are ex- Sidgwick proceeds by rejecting Humean redescriptions of amples of reason motivating. these cases and concludes that reason does indeed motivate. This version of Sidgwick’s argument makes clear, how- But later generations, raised on a diet of Hume, will be chary ever, a final objection. One might explain the motivating about accepting any apparent cases of reason motivating. power of moral judgments as the noncognitivist does. Sidg- What Sidgwick gives as a Humean redescription will strike wick does not consider a noncognitivist analysis, although many now as the natural description: there is “merely a con- his rejection of a subjectivist analysis points the way: “truth- flict among desires and aversions.” telling excites a feeling of approbation” can become an ex- One reply is that what present-day Humeans insist on is pression rather than (or in addition to) a description of atti- that desires are needed for motivation. Unlike Hume, they tude, so that parties disagreeing over truth-telling are indeed do not insist that beliefs cannot cause desires. (It is hard disagreeing, by expressing opposed attitudes. There is no
need for Hume to concede that reason can motivate, since 41 Smith objects to anti-Humeans that “[i]t is a commonplace, a fact of ordi- nary moral experience, that when agents sufffer from weakness of the will they the apparent cases of reason defeating desire can be under- may stare the facts that used to move them square in the face, appreciate them in stood, as the noncognitivist understands them, as a clash of all their glory, and yet still not be moved by them” (Problem p. 123). Sidgwick desires. agrees that this is possible—as it seems McNaughton does not—but he doubts that, in most cases of weakness of will, the facts are appreciated in all their One reply would be to construct objections to noncogni- glory. For McNaughton, see pp. 128-31. 13
Robert Shaver Sidgwick on Moral Motivation tivism from Sidgwick. But whether Sidgwick would object to noncognitivism is not so clear. This may be surprising, given Sidgwick’s status in histories of ethics as a “non- naturalist” or “intuitionist” against whom the noncog- nitivists developed their alternative. But Sidgwick’s com- mitment is to rationalism rather than to cognitivism or “de- scriptivism.” Much depends on the success of noncognitiv- ism at capturing the features of moral argument Sidgwick wields against other analyses. Ambitious noncognitivists, such as Hare and Gibbard, might give Sidgwick all he wants.42 Introducing noncognitivism suggests a more comprehen- sive gloss on Sidgwick’s overall argument. Sidgwick and Hume agree that moral judgments motivate. They might motivate (a) by describing facts that combine with pre- existing desires; (b) by describing desires; (c) by being be- sires or the like; (d) by combining with a pre-existing desire to do what one takes to be right; (e) by causing desires, without the need for pre-existing desires; (f) by expressing desires. Sidgwick’s arguments against analyses of moral judgments, if successful, rule out (a) and (b). (c) is unneces- sary given the possiblity of (e). (d) does not raise a problem for rationalism. (e) is Sidgwick’s account. Whether Sidg- wick could instead adopt (f) depends on the possibility of ra- tionalist noncognitivism.43
42 For the same thought, see Scanlon pp. 59, 64. 43 Thanks to Shane Huson, Tim Schroeder, Sergio Tenenbaum, two anony- mous referees, and, especially, Joyce Jenkins for comments on earlier drafts. 14