Mill's Proof of the Principle of Utility

Ethics

Volumn 110, Issue 2, 2000, Pages 282-310

  Millgram, Elijah ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0040866025     PISSN: 00141704     EISSN: None     Source Type: Journal    
DOI: 10.1086/233270     Document Type: Review

References (50)

1. References to Utilitarianism are by page number in the standard edition of John Stuart Mill's Collected Works (Toronto and London: University of Toronto Press, Routledge & Kegan Paul, 1967-89), vol. X, followed by chapter and paragraph for the convenience of readers to whom the standard edition is not easily accessible; so the reference above is to p. 234, chap. 4, pars. 2 and 3. Other references in the text will be to the standard edition by volume and page number alone; vols. VII and VIII, which I will refer to frequently, are Mill's System of Logic. The quoted passage is presented as only half of the argument: it is supposed to show that happiness is desirable, but not yet that happiness is the only thing desirable. It will turn out, however, that the apparent division of labor between the two halves of the argument is at least misleading.
2. For a discussion of the early history of such criticism, see Steve Gerard, "Desire and Desirability: Bradley, Russell and Moore versus Mill," in Early Analytic Philosophy: Frege, Russell, Wittgenstein, ed. W. W. Tait (Chicago: Open Court Press, 1997), pp. 37-74.
3. I'm grateful to John Rawls for emphasizing to me Mill's interest in reaching the politicians.
4. Compare Skorupski's recent description of Utilitarianism: "It is written for the general reader. It is not a technical treatise of philosophy, like the System of Logic; neither is it a carefully polished piece of political argument, as On Liberty is" (John Skorupski, John Stuart Mill [New York: Routledge, 1989], p. 283). Skorupski takes the second sentence to follow from, or at any rate to be of a piece with, the first This, I will show, is a mistake. Utilitarianism is written for the general reader, and it is a "technical treatise of philosophy." This attempt to write for both audiences at the same time is not unique. The preface to Mill's Principles of Political Economy concludes with the following announcement: "Although [the writer's] object is practical, and, as far as the nature of the subject admits, popular, he has not attempted to purchase either of those advantages by the sacrifice of strict scientific reasoning. Though he desires that his treatise should be more than a mere exposition of the abstract doctrines of Political Economy, he is also desirous that such an exposition should be found in it" (II:xcii).
5. Compare Skorupski's recent description of Utilitarianism: "It is written for the general reader. It is not a technical treatise of philosophy, like the System of Logic; neither is it a carefully polished piece of political argument, as On Liberty is" (John Skorupski, John Stuart Mill [New York: Routledge, 1989], p. 283). Skorupski takes the second sentence to follow from, or at any rate to be of a piece with, the first This, I will show, is a mistake. Utilitarianism is written for the general reader, and it is a "technical treatise of philosophy." This attempt to write for both audiences at the same time is not unique. The preface to Mill's Principles of Political Economy concludes with the following announcement: "Although [the writer's] object is practical, and, as far as the nature of the subject admits, popular, he has not attempted to purchase either of those advantages by the sacrifice of strict scientific reasoning. Though he desires that his treatise should be more than a mere exposition of the abstract doctrines of Political Economy, he is also desirous that such an exposition should be found in it" (II:xcii).
6. My emphasis. As per usual, Mill adds a clause about the absence of pain; to keep the writing manageable, I'm going to suppress this rider from here on in.
7. Again, my emphasis. Mill is being disingenuous here: as he well knew, in the writing of earlier British moral philosophers, such as David Hume and Adam Smith, 'utility' had meant something very different, and the misunderstanding Mill is trying to correct was in fact quite natural. See Geoff Sayre-McCord, "Hume and the Bauhaus Theory of Ethics," Midwest Studies in Philosophy 20 (1996):280-98.
8. I should perhaps note that I am not here taking over Mill's technical sense of "nomenclature" (VIII:704-5).
9. Even as astute a reader of Mill as Skorupski falls into this trap; see n. 16 below.
10. "The truths known by intuition," he says, "are the original premises from which all others are inferred" - and by 'intuition' Mill does not mean what a late-twentieth-century philosopher would: he uses the word as a synonym for 'consciousness', i.e., for "what one sees and feels" (VII:6-7).
11. It might be thought that, at the terminus of the chain of justification, belief, feeling and justification all collapse into one. But Mill does in fact continue to distinguish the feeling from the belief, for reasons having to do with the role of general terms in observation. These issues can be put aside here.
12. The phrase is borrowed from Norman Kretzmann, "Desire as Proof of Desirability," in Utilitarianism with Critical Essays, ec. Samuel Gorowitz (Indianapolis: Bobbs-Merrill, 1971), pp. 231-41 (originally published in Philosophical Quarterly 8 [1958]: 246-58).
13. As far as the point at hand goes, Mill's use of proof as a technical term matches, and justifies his invoking, "the ordinary and popular meaning of the term" (207/1:5). But there is a further point, which we will soon come to, that stretches what is today the ordinary meaning a good deal.
14. Deductive validity notwithstanding, there are a couple of difficulties that are worth highlighting. In working the finger-exercise version of the first stretch of argument, the claim that happiness is the aggregate of the objects of noninstrumental desire is going to play a pivotal role, and the question is how we come by it. There seem to be two ways of getting there. The first is to read it off the subsequent argument, by noticing that Mill takes being an object of noninstrumental desire to be equivalent to being a component of happiness. But if this is what Mill is doing, then he seems to be entirely ignoring questions of the organization of the parts of a person's happiness into a whole. And I think that this is a genuine problem in Mill's view. The second way of getting the pivotal claim is to use the verbal equivalence of 'happiness' and 'pleasure', and then that of 'desiring a thing' and 'finding it pleasant'. The problem here is that pleasure has to come out being the object of desire; but as we ordinarily use the words, to find something pleasant is not the same as that thing's being pleasure. By our lights, there is a slide from treating pleasure as a propositional attitude to treating pleasure as the object of the attitude. Now, by the end of the passage that introduces the equivalence, it is clear that Mill takes himself to have introduced the terms in a way that bridges the apparent gap: having said that "to desire anything, except in proportion as the idea of it is pleasant, is a physical and metaphysical impossibility," he then says, by way of repetition, "that desire can [not] possibly be directed to anything ultimately except pleasure." But why would Mill want to use his words this way? I will towards the end of the argument return to the question of what substantive view underlies this aspect of Mill's vocabulary.
15. VII:176; compare VIII:760-61.
16. For instance: "Mill stands accused of committing glaring logical blunders. . . . such accusations . . . are not justified. . . . he explicitly states that a proof, in the commonly understood sense of the word, of the Utility Principle . . . cannot be given. So he is not claiming to present a deductively valid argument - which is what one must assume him to be doing to pin on him the familiar fallacies" (Skorupski, pp. 285-86).
17. VII:162; the full discussion appears in the previous section, at VII: 158-62. See n. 48, below.
18. VII:186-93. This may also seem hard to believe; after all, Mill would not have denied, of a syllogism like, 'All men are mortal; the Duke of Wellington is a man; so the Duke of Wellington is mortal,' that when the major premise is true, all its instances are true also. However, we now have a widely familiar analog of syllogistic reasoning as Mill understands it, and highlighting the analogy may make Mill's claim seem more reasonable. Think of the file compression utility (e.g., Zip or Compress) on your computer. When you compress and reexpand a file, you do not think of yourself as deriving new results, but as merely reformatting the file for more convenient storage or access. Now, in Mill's view, the actual inference to the Duke of Wellington's mortality is from particulars to particulars: from 'A was a man, and he died', and 'B was a man, and he died too', and so on, and 'The Duke of Wellington is a man', to The Duke of Wellington will die'. The major premise of our sample syllogism is not part of the inference proper, but is rather a way of storing the observed evidence in a way that makes it easy to keep track of and use; the syllogism is a method of extracting already available information, not of inferring new information.
19. VII:158-59; compare VII:66, or again, the Analysis: "We can afford . . . no aid to the reader in distinguishing [a pleasurable or painful sensation], otherwise than by using such expressions as seem calculated to fix his attention upon it" (James Mill, vol. 2, p. 190).
20. So while Mill did not think that deduction was properly understood as inference, he did not deny, but rather insisted on, the heuristic importance of deduction, and syllogistic deduction in particular. See, e.g., VII:196-99; VIII:665.
21. VII:15n.; this quotation is from the 1856 edition of The System of Logic, and was omitted in subsequent editions. Compare his criticism, at VII:93-97, of views of naming on which all truth comes out merely verbal, and his discussion "Of Propositions Merely Verbal" where he claims that "when any important consequences seem to follow . . . from a proposition involved in the meaning of a name, what they really flow from is the tacit assumption of the real existence of the objects so named" (VII:113).
22. Henry Sidgwick, The Methods of Ethics (Indianapolis: Hackett, 1981), p. 388.
23. Mill registers his awareness of this demand at VIII:951.
24. The omitted phrase is a qualification designed to avoid circularity in the establishment of a specifically moral criterion. This omission is repeated in the next passage. The passage is introduced as an explanation of difference in quality in pleasures, as opposed to quantity. But since, as we will see in a moment, quantitative comparisons work no differently, we can disregard the apparent restriction in scope. character; his character is shaped by his entire history; and this is not something we can observe (VII:865). Consequently, a criterion that appealed to psychological counterfactuals would be unusable: there is simply no way to determine what you would prefer, if . . . Mill's second problem is that there are many counterfactual selves - some dulled, some especially nasty, some overly tremulous, etc. - whose judgments are going to have to be ruled out as defective. It's hard to believe that this can be done without importing into the selection of counterfactual selves the full and robust set of evaluations that the use of the counterfactual selves was meant to certify. But if this is what is going on, this use of counterfactual selves is viciously circular: "to explain away the numerous instances of divergence from their assumed [moral] standard, by representing them as cases in which the perceptions are unhealthy" is "[a] striking instance of reasoning in a circle" (VIII:826). Notice that this is also a reason for not looking to the preferences of a carefully chosen coterie of right-thinking persons: appealing to the judgments of the phronimoi will not do. This is why Mill usually requires only experience of the options being compared. And I am accordingly inclined to think that his sole mention of "habits of self-consciousness and self-observation" is a slip on Mill's part.
25. A quick textual point, for those who are used to reading the passages in question as invoking counterfactuals. In English, counterfactuals are generally (although not necessarily) signaled by the subjunctive mood. These passages, amounting to about half a page of text, contain not even a single subjunctive, and the explanation cannot be stylistic: Mill's discussion of "Permanent Possibilities of Sensation" uses the subjunctive without hesitation (e.g., IX:184).
26. If this sounds implausible - why should I care about what other people prefer, if their preferences turn out not to match mine? - notice that the currently popular counterfactual technique has the same problem. Why should I care about what a counterfactual self would prefer, if his preferences turn out not to match mine? The difficulty is, however, perhaps especially acute for Mill, who in On Liberty had pointed out that others' experience may be an "unsuitable" guide if one has an "uncustomary" character, that "human beings are not . . . undistinguishably alike," "that people have diversities of taste, . . . [and] different conditions for their spiritual development," and that there "are . . . differences among human beings in their sources of pleasure, their susceptibilities of pain, and the operation on them of different physical and moral agencies" (XVIII:262, 270). I will get around to discussing the more general version of this problem in due course.
27. Kretzmann notices and highlights this move, but worries that he has "extended [Mill] and even departed from him" (Kretzmann, p. 241). One other worry that might arise at this point is whether the decided preference criterion is really usable: do we have to wait until all the votes are in before we can make up our minds? Notice, however, that this problem is very like another that utilitarianism already has. When assessing an action or rule, how far into the future do we need to look for the effects it will have on overall utility? The in-principle answer is, of course, to the end of time; but that cannot be the normal procedure.
28. Mill has just given a prediction of the future state of society. Now we saw that Mill does not think that individual psychologies are predictable (n. 27, above). But he thinks that problems of individual psychological prediction can be sidestepped when large groups of people are being studied. The political scientist "can get on well enough with approximate generalizations on human nature, since what is true approximately of all individuals is true absolutely of all masses" (VII:603). Mill had been impressed by the new science of statistics (VIII:932) and was quite willing to believe that even if you could not predict the counter-factual preferences of individuals, under certain circumstances predicting the preferences of the majority of the experientially privileged was well within the realm of possibility (VIII:846-47; cf. VIII:873, 890). In that case, the reader might be wondering, why not appeal to the counterfactual preferences of the experientially privileged as a group? Why the reliance on actual preferences? Mill does not actually discuss this option, but he would have had reasons for not availing himself of it. First of all, the desired social science (the science of character, which Mill dubbed "Ethology") was never more than a gleam in his eye; barring special circumstances, such as the argument we now have on hand, we don't for the present have any way of predicting the preferences of a group. Second, prediction in the social sciences has its limits, and of precise predictions of the history of society, extended arbitrarily far into the future (similar to those we have come to expect from astronomy), Mill thought, "there is . . . no hope" (VIII:877). Small errors snowball, and so the predictions of a "Deductive" science like Ethology would need to be continually checked against and corrected by observation - in this case, observation of the actual preferences of the experientially privileged. Since the actual preferences are not, on Mill's account of the social sciences, dispensable, he may have preferred to cut out, as so much wasted motion, the detour through counterfactuals.
29. Mill sometimes suggests that the economic arrangements of the future will also work to mute the conflict between the alternatives; e.g., in the Principles of Political Economy, he speaks of a time when "civilization and imprcvement have so far advanced, that what is a benefit to the whole shall be a benefit to each individual composing it," having in mind in this case the more equitable distribution of gains in productivity (III:768).
30. There's a nicety here that deserves pointing out. The criterion of desirability introduced in chap. 2 avoids circularity by specifying that the preferences consulted shall be those given "irrespective of any feeling of moral obligation to prefer" one item rather than another (211/2:5). (See n. 24, above.) In order to prevent this condition from excluding the preferences of the acculturated utilitarian majority for the general happiness over all else, Mill argues that the "strengthening of social ties . . . leads [each individual] to identify his feelings more and more with [others'] good . . . He comes, as though instinctively, to be conscious of himself as a being who of course pays regard to others. The good of others becomes to him a thing naturally and necessarily to be attended to" (231-32/3:9, emphases Mill's). Future utilitarians, thinks Mill, will not want one thing but feel constrained by the moral law to prefer another. Their desire for the general happiness will be a native preference, one that can thus serve as grist for the "decided preference" criterion.
31. For a lengthier and much more thorough rendition of Mill's views on this point, see his essay, "Auguste Comte and Positivism" (X:263-368).
32. There is in fact a more likely way for the contingency of value to exhibit itself, and one which I suspect worried Mill a good deal. If A's being preferable to B is a matter of the majority of the experientially privileged preferring A to B, then A's being preferable to B depends on there being individuals having experience of the different options: if there are no such individuals, or perhaps not enough of them, then there is no fact of the matter as to which is preferable. The applicability of the decided preference criterion requires that there be people who try things out - who conduct, in Mill's phrase, "experiments of living" (XVIII:261) - and I am inclined to think that this was one of Mill's reasons for insisting so strongly on the liberty to try things out. (Notice the awkward tradeoff: in order for there to be a standard of preferability, people have to, and hence have to be encouraged to [XVIII: 269], do what will prove to be the incorrect thing.)
33. See VII:227, 231, 237-61, 277.
34. The sentence can read on the model of sentences like: "To the scientific community, the theory of relativity is true." Norman Kretzmann takes Mill's argument to lean in a pragmatist direction (Kretzmann, p. 239), and it is in this vicinity that I find his suggestion most evocative. At this point we are able to say what Mill has in mind in his response to Spencer. That "the truths of arithmetic are applicable to the valuation of happiness" (258n./5:35n.) is to be read off the preferences of the experientially privileged. (They are not always applicable: higher and lower pleasures are incommensurable, and this fact is also to be read off the preferences of the experientially privileged.) The processes that will make future persons into natural utilitarians will, Mill further argues, also create persons who take it for granted "that the interests of all are to be regarded equally" (231/3:9).
35. It might look like there are a couple of places to get off the boat here: you might object that the reasoning in question isn't actually practical, or, alternatively, that, although practical, it is merely means-end reasoning. But while one can always get out of the boat, it turns out that, in this case, there's no pier handy to step onto. Notice, first, that the two objections are not as different in spirit as they might at first glance seem to be. Since the reasoning was identified as practical by identifying its conclusion as practical, the first objection requires insisting that that conclusion does not give the agent a reason for action. Now this latter thought is normally underwritten by the idea that sentences like "I ought to do this" or "This is the correct preference" do not express reasons for action because they do not express desires. But the requirement that reasons for action be desires is in turn normally underwritten by instrumentalism. Because the point of correcting the agent's preferences was to pick out his reasons for action, let's abandon the first objection in favor of the second: the reasoning is conceded to be practical, but is entirely means-end, that is, correcting one's preferences is shown to be a way of satisfying some further desire. But now the reasoning so construed faces an uncomfortable dilemma. Either the desires from which it proceeds are corrected, or they are not. If they are unconnected, then we have the appearance of a pragmatic contradiction: the conclusion of the reasoning, recall, was that one should use in one's practical reasoning, not one's uncorrected, but corrected desires or preferences. If, on the other hand, they are corrected, then we have the question blatantly begged. To see that, consider the "black box" that corrects your preferences to mine. Asked why you should regard my preferences as those you should act on, rather than your own, I would not convince you by answering: because I would prefer it. The clean est way out of the dilemma would be to invoke desires or preferences that one was sure would not need correction, and so for which the uncomfortable choice does not arise. But the confidence required here would be inappropriate for human beings: to be sure (ahead of time, of any desires) that one will not be disappointed - so sure, as to be willing to make those desires a pivot of one's practical logic - is just to show a failure of imagination in the course of setting oneself up for a fall. Human beings can end being disappointed in anything.
36. The point is not that appeals to informed desires are in principle dispensable: taking the argument in that direction would bring us around to a conclusion I would not wish to endorse, that there are no properties that can be understood only via the reactions of their perceivers. (Such properties have traditionally been called "secondary qualities.") The point is, rather, that the explanation for taking the perception of some secondary quality to provide a reason for action will involve a noninstrumental pattern of practical inference. The problem of informed desires is an interesting and important topic in its own right, not merely an unworkable fix for instrumentalism. I do not here mean to be endorsing one or another approach to it, and in particular I should not be taken as recommending the decided preference criterion as a replacement for contemporary counterfactual-based devices.
37. Very quickly: To desire something, noninstrumentally, is to find the idea of it pleasurable, that is, to expect it to be pleasurable; and since the "decided preference criterion" is being used as a way of correcting desires so their objects are found pleasurable, we can read 'desired' as 'desired, subject to correction by the judgments of the experientially privileged'. Happiness means pleasure, and so now means something like: the aggregate of the objects of one's desires, as corrected by the preferences of the experientially privileged. So the premise of the stretch of argument that we are now marvering comes out as: the objects of desires corrected by the preferences of the experientially privileged are desired, when those desires are corrected by the preferences of the experientially privileged. Once again, the premise of the argument is a tautology. Now one's noninstrumental desires, corrected by the preferences of the experientially privileged, are what is desirable (by the "decided preference criterion"). So the conclusion, that each person's happiness is desirable for him, comes out meaning: the objects of one's desires, as corrected by the preferences of the experientially privileged (that is, when attained, happiness), are the objects of one's desires, as corrected by the preferences of the experientially privileged (that is, what is desirable). The conclusion is also, once again, a tautology; and is, once again, the premise of the argument repeated.
38. IX:468-69; the passage provides two examples.
39. See n. 21 above.
40. Mill, Analysis, vol. 2, pp. 191-92, for the analysis; notice the younger Mill's modification to the definition in the editorial footnote at pp. 194-95. For the desires for wealth, etc., see pp. 207-14. For the role of desires in action, see pp. 256-59, 262n.-64, 327-403. We can now tie up another loose end: the analysis of desire accounts for what we would see as the terminological slide that I highlighted in n. 14, above. Like his son, James Mill identifies, as a matter of terminology, pleasure (or more carefully, the representation of pleasure) with desire: "The terms . . . 'idea of pleasure,' and 'desire,' are but two names; the thing named, the state of consciousness, is one and the same" (pp. 191-92; cf. p. 327). A desire just is an idea of the sensation of pleasure, which idea may be associated with other ideas, most prominently, of causes of the sensation. It follows that understanding desires as (what we would nowadays call) propositional attitudes, i.e., attitudes whose objects are as it were logically integral to the mental state, is a mistake: "when the word desire is applied to the cause of a sensation . . . it is employed in a figurative, or metaphorical, not in a direct sense" (p. 351). What we would call the objects of desire are actually just the causes of the real objects, pain and pleasure: "The illusion is merely that of a very close association" (p. 192). This view, quite startling by today's lights, accounts for Mill's treating the moves from desiring X, to finding X pleasurable, to desiring pleasure (which happens to be associated with X) as traversing a series of innocuous synonymies.
41. Ibid., vol. 2, pp. 357n., 378.
42. Ibid., vol. 2, pp. 209-11.
43. To be sure, one can see how the fact that human beings are desiring creatures might be made the starting point of an argument for the view we are considering. But the argument is not in Mill, and the mode of argument, as we will see in a moment, would not be one that Mill could admit.
44. See VII:567-77; also VII:306-14; VIII:833.
45. Foremost among further criteria that need to be satisfied in order for a mental transition to amount to an inference is, recall, that "the conclusion is . . . wider than the premises from which it is drawn" (VII:288). Mill never seems to have asked whether means-end reasoning, which he endorses, satisfies this condition, and it is unclear to me, in view of his discussion of parallel questions, how he could have given a satisfactory answer.
46. See, e.g., VII:193.
47. Even though sense-data are now out of siyle, taking a class of basic observations for granted is not. A familiar recent example is van Fraassen's identification of observation with "what the unaided eye discerns" (The Scientific Image [Oxford: Clarendon Press, 1980]), p. 59; see also pp. 16 ff., 56 ff., 214; for criticism, see essays in Images of Science: Essays on Realism and Empiricism, ed. Paul Churchland and Clifford Hooker [Chicago: University of Chicago Press, 1985]).
48. Even though sense-data are now out of siyle, taking a class of basic observations for granted is not. A familiar recent example is van Fraassen's identification of observation with "what the unaided eye discerns" (The Scientific Image [Oxford: Clarendon Press, 1980]), p. 59; see also pp. 16 ff., 56 ff., 214; for criticism, see essays in Images of Science: Essays on Realism and Empiricism, ed. Paul Churchland and Clifford Hooker [Chicago: University of Chicago Press, 1985]).
49. Anderson, for instance, adduces under this heading Mill's diagnosis of his nervous breakdown as an "experiment in living" (Elizabeth Anderson, "John Stuart Mill and Experiments in Living," Ethics 102 [1991]: 4-26). I think that it can be so understood, but not the way Anderson would like to: Mill himself was surprisingly unsuccessful in learning the lessons of his breakdown and partial recovery. He was never able to provide a satisfactory account of the importance in his own life of Romantic poetry; willing to surrender neither his instrumentalism nor his associationism, he was forced to treat the resistance of associations formed under the influence of poetry to analytical corrosion as a brute psychological fact. And while Mill is quite clear about his own motivational structure up until his breakdown, his view of himself after his recovery fails, quite strikingly, to match the ways he thought, felt, and lived. I hope to discuss these issues further elsewhere. There is another confusion in Anderson's paper that is common enough to be worth remark. Anderson thinks that "Mill's emphasis on the intrinsic desirability of gratifying the [higher] sentiments strongly suggests that he believed that dignity, beauty, honor, and so forth are values distinct from pleasure," and she takes Mill to be departing from "the basic premise of ethical hedonism, that pleasure is the sole respect in which things can be intrinsically valuable." She concludes that "Mill's conception of the good is not hedonistic but pluralistic and hierarchical." But this is just to lose track of Mill's terminology. For something to be desirable is, as we have seen, just to be desired (noninstrumentally, and subject to the correction of the experientially privileged) ; this is, merely as a matter of nomenclature, to be found pleasurable. Consequently, being pleasurable, in Mill's sense, is not a respect in which things can be found desirable. And if we tie the word 'hedonism' to Mill's word 'pleasure', whether Mill's conception of the good is "pluralistic and hierarchical" has nothing at all to do with whether it is "hedonistic."
50. That people want happiness is, we have seen, a verbal matter, and not a matter of experience at all. But for that reason, happiness is not a substantive account of what is wanted; the substantive account is to be given by empirical investigation of what people, and their experientially privileged stand-ins, actually desire.