NOTES CHAPTER ONE 11.Bentham's precursors generally spoke of "pleasure' or "happiness'. Hume used "utility', though more narrowly than Bentham. See the discussion in Sidgwick 1907, bk. 4, ch. III, sect. 1, n. 2. Bentham seems to have given "utility' its wide modern sense. Mill claims the dubious credit for the term "utilitarian'; see Utilitarianism, ch. II, n. to para. 1. It avoids, he says, "tiresome circumlocution'; but that hardly justifies its own bulk, ugliness, and inaccuracy. It could have been bulkier; Bentham once thought of proposing "eudaimonologian'; see Baumgardt 1952, p. 505. 12.J. S. Mill, Utilitarianism, ch. II, para. 2. See also J. Bentham, An Introduction to the Principles of Morals and Legislation, ch. I, sects. 2 and 3. Neither of them was aware that the words "happiness' and "pleasure' are too vague and slippery to be much of an explanation. And Bentham, having used "pleasure' and "pain' to explain "utility', immediately turned around and undid whatever specificity the ordinary use of those terms lent in adding, "By utility is meant that property in any object, whereby it tends to produce benefit, advantage, pleasure, good, or happiness . . .' (sect. 3). 13.Jones 1964, pp. 655--6. 14.Sidgwick 1907, esp. pp. 111--12, 127--9, 396--8. For a good discussion of Sidgwick's account see Schneewind 1977, pt.2, esp. chs. 11--12. 15.Nozick 1974, pp. 42--5. 16.I talk here in terms of "seeming to value', which is enough to make the retention of mental states look puzzling, but I shall go on to argue that this appearance is not misleading; see ch. I sect. 6; ch. II sects. 3, 4, 6; ch. III sect. 7; ch.IV sects. 3, 5, 6. 17.Sidgwick is well aware of this line of thought, but his objection to it is uncharacteristically thin. See Sidgwick 1907, pp. 398--402. 18.But for acute criticism from economists, see Vickers 1975; Sen 1980--1, sects.3 and 6; Hahn and Hollis 1979, Introd. 19.See Morton 1980, pp. 135--8, where, following Freud, he contrasts the "object' and the "aim' of a desire: "It isn't as if the desire says {"I want that{', and then to satisfy it one has to produce just that. Rather, it says, {"I want something, and here's what it must do for me{'.' (p.139.) 10.Hobbes sees something like this as characteristic of desire generally: "...there is no such finis ultimus, utmost aim, nor summum bonum, greatest good, as is spoken of in the books of the old moral philosophers ... Felicity is a continual progress of the desire, from one object to another; 314 NOTES TO PAGES 12--14 the attaining of the former, being still but the way to the latter. The cause whereof is, that the object of man's desire, is not to enjoy once only ...but to assure for ever ... I put for a general inclination of all mankind, a perpetual and restless desire of power after power, that ceaseth only in death.' (Hobbes Leviathan ch.XI.) Tibor Scitovsky discusses this feature of consumer desire in Scitovsky 1976. 11.That is, I do not think that we can stop with faults of fact and logic. Richard Brandt disagrees. See his account, worked out with exemplary rigour, of rational desires as desires that "survive maximal criticism by facts and logic'; Brandt 1979, p. 10, but see chs. II--VII passim. What he says about the criticism of desires seems to me correct as far as it goes but incomplete, and thus it puts too much stress on information, as if all that criticism of desires needs, besides logic, is facts, the more the better. 12.See e.g. Kenny 1973, p. 61. 13.I owe the example to Derek Parfit. 14.Criticizing desires is not far short of what we mean by making sense of life. There has been a long tradition of complaint against utilitarianism that it puts happiness where the meaning of life belongs. The latest is David Wiggins, in Wiggins 1976, esp. p. 332. But he takes the notion of "utility' to be more closely tied to Hume-like accounts of action and deliberation than it need be. And he overlooks the role that global desires and plans of life play in "utility'; they shift the concept far closer to what Wiggins means by "meaning' than to what he means by "happiness'. 15.See Glover 1977, pp. 63--4. 16.I prefer "informed' to "rational', a word sometimes used with much the same qualification in mind. "Rational' is often used of adapting means to ends, and seems to me to fit less well both simple tastes and deliberation about ends. "Informed' seems to me to have more scope. (But, then, "rational' would also become a technical term, and one could stipulate a broad sense for it too.) 17.This stark view still gets attributed to utilitarianism, e.g. in Wiggins 1976, p.340. 18.This builds enough strength into the notion of "informed' for us not to need to resort to formulas such as "ought to desire' or "have reason to desire'. For instance (to use an example I am about to introduce) suppose my doctor tells me that I shall die if I do not lay off drink. Well, I might listen, agree, but not even desire to lay off it. Can I not be informed but foolish? So how could well-being be fulfilment merely of informed desires? "Informed' requires more than listening and believing. I should also have to realize fully what is at stake. I should have to appreciate what losing those extra years of life means; and I should have to keep in mind just how pleasant a drink is. It is only desires formed in response to that amount of appreciation that are informed. Or take a different sort of case. Suppose a recluse accepts all the information that we give him about the joys of human companionship and even decides 315 NOTES TO PAGES 14--15 that he would be better off being more gregarious. Yet he might, because his desire for solitude is very stubborn, accept all the information and not change his desires. Again, informed, it seems, but foolish---or, rather, unfortunate. But we should describe the case more fully. What might make his desire so stubborn? A number of things could, so let us take one possible explanation: people frighten him and to avoid the anxiety he shuns company. Then, it is not that he does not desire companionship; it is just that he also desires not to be anxious. In his particular case, the two desires are incompatible and the second is greater (and might remain greater when he is fully informed) than the first. Weakness of will is yet another kind of case, which I talk about in the text. 19.The stress on "informed' desire may seem to make the desire account incurably anthropocentric, not extendible to animals, and so in that respect distinctly inferior to a mental state account. Indeed is any account of well-being acceptable in moral and political theory if it is not so extendible? My concern is with the notion simply of human well-being. Still, I do not see any obstacle to an informed desire account's applying to animals too. It does not take human powers of intellect for a desire to be "informed'. It needs only that the desire be shaped by appreciation of the nature of its object. Pains, even for animals, cover a wide spectrum, from physical (being stepped on) to psychological (being terrified). Pains get their status as pains not solely from their phenomenological feel but also from their status in a life---that they are to be avoided, to be stopped, and so on. So part of their status as pains brings in desire and action. It may seem that it brings in desire only in a very extended sense of the term. But then no informed desire account limited only to humans will be plausible without also using a similarly extended sense of desire. The relevant sort of desire does not have to be held antecedently to its fulfilment (a human can enjoy something, want to have it continue or return, that he never knew he would enjoy, or even knew existed). The relevant sort of desire does not have to be conscious (one can be made better off by something, e.g. relaxation, that one has never focused sharply enough to form a wish for). And so on. 20.I owe the example to Roslind Godlovitch. 21.If this is the best interpretation of "strength', then many objections to utilitarianism are wide of the mark---for instance, objections to its reducing all preferences, from ideals at one end of the spectrum to tastes at the other, to a single scale of "intensity' (see, e.g., Sen 1980, p. 211 and Sen and Williams 1982, p. 8). Utilitarianism is not committed to such reductions. And whether enough tastes can swamp ideals depends entirely upon the structure that informed desires display. There is nothing in the informed desire account either that commits one to a reduction of all prudential values to a single substantive super-value called "utility' (see ch. II sect. 4) or even that rules out certain forms of incommensur- ability (see ch. V). 316 NOTES TO PAGES 16--20 22.Richard Brandt thinks not, and he makes a strong case. See Brandt 1979, ch.VII, sect. 3, esp. pp. 146--8 and ch. XIII, sect. 1; also Brandt 1982, esp. sects.8--10. For discussion of Brandt's charge see Sen 1980--1, sect. 4. For a rich discussion of this and related issues see Elster 1982. 23.See ch. VII sect. 5. The trouble that Brandt has in mind arises from a person's having conflicting rankings over time with no super ranking to resolve them. But the intrapersonal case of two rankings, mine now and mine later, is formally similar to an interpersonal case of two rankings, mine and yours. There is no super ranking in the latter case either, but we need not fall back on a mental state account. Another way is to go beyond simple ordinal rankings, either, say, by appeal to the test of potential Pareto improvement or by the adoption of some sort of cardinal scale of utility, with comparability. Unless such ways are closed, which I think they are not, Brandt's worry may be met. These too are issues that I shall return to when I talk about measurement (ch. VII). 24.The first example is Derek Parfit's, the next two are L. W. Sumner's, in whose book, Abortion and Moral Theory, there is a very good discussion; see Sumner 1981, esp. p. 183. 25.I shall not go on listing troubles for the desire account, though there are others. For instance, it has difficulty distinguishing between selfish and selfless action. An act of self-sacrifice has to be intentional; so it will be the object of an informed desire; so it will, when fulfilled, increase the agent's welfare. Self-sacrifice is transmuted, by this philosophers' stone, into self-interest. See Overvold 1980, Overvold 1982, Brandt 1982. This is an instance of the more general trouble that I have mentioned: informed desire spreads to objects outside the bounds of well-being. An adequate solution to the general problem ought to solve this one too. 26.I model this account on one defended in Sumner, 1981 ch. 5, sect. 21; but I have also benefited from long discussions with Sumner during his leave in 1982--3, when he was a valuable addition to the Oxford philosophical community. 27.It is not that this is a utilitarian's only hope, though I think that it is the only good one. Amartya Sen has recently asked whether we really need regard these various accounts of "utility' as exclusive alternatives (Sen 1980--1, pp. 203--4). What instead we should do, he suggests, is to regard "utility' as a plural notion, a vector of many different considera- tions. Suppose someone prefers bitter truth to comforting delusion, but is palmed off with the latter. We feel sorry for him. Then later he learns that it was all a deception. We feel sorrier for him. Why not take our sympathy as an indicator of the person's utilities? Bare desire fulfilment and experienced desire fulfilment are both relevant, and to insist on a choice between them, Sen says, seems "arbitrary and uncalled for'. Similarly for the other hard choices we have been considering, for example, the choice between actual and informed desires. 317 NOTES TO PAGES 22--23 But the doubt about Sen's eclectic approach is that if one adopted it one would have to supply some weighting for these various vectors when they merge in decision, and the weighting would have to be neither arbitrary nor left to haphazard intuitions. In this respect, the non-eclectic approaches are superior. The informed-desire account, for instance, encompasses Sen's various vectors and attaches weight to them in a manner motivated by the spirit of the notion of well-being. When the person we alluded to learns that it was all deception, his experienced hurt counts too, Sen says. But so it does on a desire account, because few want to be hurt. Similarly, for the choice between actual and informed desires; a plausible desire account gives weight to both. If the lacuna in Sen's notion of "plural utility' is properly filled, it is likely to lead in the end to one of the other accounts. It may not be that all elements that he has in mind for his "plural utility' will be accommodated, but perhaps all that the notion of "well-being' should be made to carry will be. CHAPTER TWO 11.The introduction of this restriction on the desire account marks a break with decision theory. F. P. Ramsey, in his famous paper "Truth and Probability' (in Ramsey 1931, p. 173), which laid the foundations of decision theory, cautioned: "It must be observed that this theory is not to be identified with the psychology of the Utilitarians, in which pleasure had a dominating position. The theory I propose to adopt is that we see things that we want, which may be our own or other people's pleasure, or anything else whatever ...' The break with "the psychology of the Utilitarians' seems to me right, but the "anything else whatever' seems to me to go too far in the opposite direction. 12.See Feinberg 1980, esp. pp. 173--6. 13.Aristotle took the notion of posthumous harm seriously, even though his notion of eudaimonia is loftier and more immune to accidents of fortune than "well-being' is. " ... if we deny that a dead man is happy ... even this admits some dispute; for it is popularly believed that some good and evil---such as honours and dishonours, and success and disasters of his children and descendants generally---can happen to a dead man, in as much as they can happen to a live one without his being aware of them' (Nicomachean Ethics, 1100a15--22). For his own cautious, unclear conclus- ion, see bk. 1, ch. XI. 14.I borrow the example of slander as a posthumous harm from Joel Feinberg; see his excellent discussion in Feinberg 1980, ch. 3. 15.So I believe now. But I began some years ago so convinced of the opposite that I took it as a test of adequacy that an account avoid this conclusion. I cannot find any conclusion on this subject that is entirely comfortable. A radical one that is tempting, but still not comfortable, is to decide that these issues (and the issues surrounding the Experience 318 NOTES TO PAGE 23 Requirement generally) put so much pressure on our notion of well-being that it fragments. Perhaps we need two notions of life's going well: one with and one without the Experience Requirement. This would open up new possibilities: perhaps the former notion fits prudential value theory better, and the latter fits moral theory better (I owe this suggestion to L. W. Sumner). See the illuminating discussion in Parfit 1984, ch. 8, esp. sect. 59. But to return to my conclusion: does the general account solve the specific problem of the desire account's undermining the distinction between selfish and selfless action? (see ch. I, n. 25.) Well, there are different kinds of case. I might not for a moment want to give up my pleasure or contentment for someone else but think that, on moral grounds, I ought to. Now, the desire account does not have to be tied to any particular theory of action. It is compatible with the view that, whatever the state of one's desires, one might recognize a moral reason to act contrary to one's prudential reasons and, because moral reasons trump prudential reasons, act in accord with the stronger (moral) reason. In that case, on anyone's account of selfless action, one has acted selflessly. Or, to move to a different kind of case, a father might take his children's welfare so much to heart that his forgoing pleasures for their sake enhances not only their well-being but his too (e.g. Pe$1re Goriot). It is not, in such cases, that he wants his children's welfare for the selfish reason that it will enhance his own; on the contrary, he wants his children's welfare. If he is prudentially shrewd, he might also recognize that his life will be better, in purely prudential terms, by his living for certain other persons. But even having that level of awareness is compatible with his acting not for his own benefit but for his children's. Thus, even on the desire account, he can sacrifice certain of his own personal goods---his own pleasure, peace of mind, contentment---for his children. And that is enough for us to go on calling this a selfless action. The fact that his aim of helping his children becomes so central to his life as no longer to be separable from what makes his own life successful ought not to upset this description. One can, of course, define a "selfless' action so that only acts that actually make one worse off count, and then the mere fact that his own well-being happens also to be enhanced does upset this description. But one could instead define "selfless' so that acts done (motive now entering the account) to help another, at considerable cost to oneself, can count. The second definition is better, and more in accord with how we ordinarily speak. The first would mean that we could not call "selfless' the act of a saintly person who had so grown in moral stature that he regarded moral failures as making his life shabby, disappointing, weak, worse in purely prudential terms, and who also thought that in the present circumstances morality required that he give up much personal benefit in order to help others. He too might be wise enough to realize that his own life is better for him personally by his acting morally, but that should not be enough to make his act selfish. 319 NOTES TO PAGES 24--26 Furthermore, it would be a mistake to try to explain the distinction between selfless and selfish actions by going over to an account of well-being that tied the notion to one's own states, to the exclusion of other persons'. That would make the line between one's own well-being and other persons' unrealistically sharp. A father might sacrifice himself for his children solely out of duty, or he might, like Goriot, do it with a sympathetic merger of his well-being with his children's. It is true that the fact that this sympathetic merger can occur makes the desire account messy. But the concept of well-being seems to be similarly messy. This issue is part of a larger one, the relation of prudence and morality; see ch. II nn. 12 and 38, ch. IV sect. 4. 16.See Barry 1965, pp. 11--15, 61--6, 71--2, 295--9 and Dworkin 1977, pp. 234 ff. (also p. 277) for arguments that they should. For further discussion of some of the issues see Williams 1973. 17.See Williams 1973, pp. 105--6, for an argument that they should. 18.John Harsanyi, in Harsanyi 1977, p. 62, says they should. 19.Brandt 1967 defines "utility' that way; see esp. pp. 29, 34. See also Brandt 1979, p. 203. J. S. Mill seems to have thought so too. "Of two pleasures, if there be one to which all or almost all who have experience of both give a decided preference, irrespective of any feeling of moral obligation to prefer it, that is the more desirable pleasure.' (Utilitarianism, ch. II para. 5, my italics; see also ch.II para. 8.) 10.The question of restricting "utility' to self-regarding desires is part of the question of what we should take as the argument in utility functions. Kenneth Arrow, in Arrow 1963, takes the relevant preferences to be preferences between alternative social policies. That leaves it open whether people rank them on a purely self-regarding basis or on a more mixed self-regarding plus other-regarding basis. For some theoretical purposes are the relevant preferences self-regarding, and for other purposes not? And when the relevant preferences are the mixed sort and preferences between social states as a whole can enter, how can an individual rank them rationally, when his ranking is supposed to be part of the rational calculation necessary to rank social states rationally? For a good discussion, see Brandt 1967. 11.I borrow the useful terms "ethical push' and "ethical pull' from Nozick 1981, pp. 401 ff. 12.The complications can be great. If my desire to be moral becomes part of my own well-being, it looks as if my well-being becomes my ground to be moral. But that turns morality into a self-regarding exercise, which means that things have clearly gone wrong. A person does not, in general, see his desire to be moral as a kind of desire that enters his well-being---that misunderstands the kind of (non-prudential) value one is dealing with. It could be only in a very special, limited, secondary sense that a desire to be moral could be part of one's well-being: namely, to the extent that this goal becomes, as for most of us it does to some 320 NOTES TO PAGE 26 extent, part of the quality of one's life from one's own personal point of view---free from guilt and loss of self-respect. Of course, the guilt and loss of self-respect would have their real source in something non-prudential: the requirements of morality, the authority of moral reasons for action. But though they have their source outside prudence, they take a prudential toll. Then there is this complication. Most of us want to be moral, but occasionally some of us want even more to do what we know is incompatible with it. For instance, there are standards of impartiality that many of us accept but do not live up to. It is not always a matter of our succumbing to temptation; sometimes it is a matter of our never having formed serious intentions in the first place. Plato may be right that if we were fully enough informed, we should not behave like this. Still, some persons are like this, perhaps from ignorance. So we must allow that it is possible for a person's desire to be moral to be only one among many and to have a less than supreme place in his preferences. That gives us one at least partially "informed' ordering. However, it is also true that, all things considered, most of us should prefer to see justice done rather than injustice, to have impartial distributions rather than biased ones, to act morally rather than immorally. But now we have a second "informed' ranking, incompatible with the first. Which gives us the utility of a person who holds both? Imagine someone who prefers retaining certain of his privileges to having justice done, yet at the same time preferring acting morally to acting immorally. That seems to me an accurate description of a not uncommon situation. But there need not be inconsistency here. We can say that such a person recognizes justice as a stronger reason for action than self-interest; here we refer to a hierarchy of reasons. But a reason can be, for him, the stronger reason and still not be his stronger desire; here we refer to a hierarchy of desires. To accept this description of the case, I have said before, does not commit one to holding that not all failures in morality are failures in knowledge. That can be left an open question. What it does commit one to saying is that, although Jones may not be able to see something as a reason for action unless it connects in some way with his desires, the connection need not be an identical ordering between strength of reason and strength of desire. Jones can have both a hierarchy of reasons for action and an inconsistent hierarchy of desires. However, although the two hierarchies can be inconsistent, the inconsistency is not internal to the concept of utility. A desire can be "informed' if the object of desire is not repudiated when its nature is understood. Utility does not require the satisfaction of a person's fully informed preference ordering, unless it is, or will in time be made, actual. Only actual desires should be satisfied. In short, there are two hierarchies, but utility uses only one. 13.Anyone who takes this line is, of course, building up a lot of problems for later. Can this wide notion of "well-being' ever be the one appropriate to moral theory? Does not the inclusion of the whole gamut of shabby 321 NOTES TO PAGES 27--29 desires (sadism, envy, spite, intolerance, etc.) just show that moral theory has to employ a narrower notion of well-being, say one centred on basic human needs, where the sordid side of human nature does not get a look-in? (see ch.III). And is not right prior to good? Do we not have to tailor our demands for our own good to fit within the bounds of what right allows? "An individual who finds that he enjoys seeing others in positions of lesser liberty understands that he has no claim whatever to this enjoyment', John Rawls correctly remarks in Rawls 1972, p. 31 (see Pt. Three). 14.I should repeat, to avoid terribly easy confusion, that my concern is with an account of "well-being', not an account of motivation or action, in which the opposition between reason and desire (cognition and appetite, etc.) also figures largely. Though the two accounts bear a great deal on one another, they are not the same. 15.See Aristotle, Metaphysics 1072a29, "we desire the object because it seems good to us, rather than the object seems good to us because we desire it'. See also Norman 1971, ch. III; Wiggins 1976; Wiggins 1978--9; Wollheim 1979; Platts 1980; McDowell 1978; McDowell 1979; Bond 1983. 16.Thomas Nagel seems to say that one can; see Nagel 1980, esp. p. 90. 17.I owe this example to Garrett Thomson, and I have benefited generally from long discussions with him. 18.E. J. Bond, whose approach seems to me too homogeneous, speaks as if, where there is perception of value, the perception "generates' a desire for it. The word "generates' certainly suggests priority, but it is hard to find anything the understanding of which would require, or even suggest, priority. See Bond 1983, p. 44. In an earlier paper, Bond offered the following argument (Bond 1979, esp. p. 56). Valuing must be prior to desiring, because something's being valuable serves to explain and not to justify an informed desire. How could it, unless it were prior? For instance, I may want something simply to spite someone else; or I may just want it---I do not know why; or I may, more respectably, want something because I value it. Here we seem to have three explanations of desire, only the last of which brings in value, and which brings it in to explain desire and so cannot be identical with the desiring. However, if I want something to spite someone else, and it will actually do the trick, then it is of some value to me. We should not confuse "value' in the (prudential) sense with which we are now concerned and "value' in an altogether loftier (possibly moral) sense, where anything as nasty as spite is doubtfully of value. My proposal is that in Freud's case seeing an option as valuable and desiring it on properly understanding what it is like come to the same. And that seems to hold in this case of spite. The second explanation---that people sometimes just want things and do not know why---is not relevant. I have an impulse to sing; I just feel like it. But that is a very dubious case of informed desire. Informed desires require awareness of, and are a 322 NOTES TO PAGES 29--32 response to, the nature of the object of the desire. An impulse does not seem to be an informed desire in the right sense. 19."Seeing in a favourable light' is a phrase from McDowell 1978, see esp. pp. 15--17. McDowell rightly objects to the view that desire makes things valuable by blindly fixing on them as its object. But something like that happens sometimes. More importantly, he makes understanding prior to desire. But, in cases like the present one, insisting on a priority either way makes desire and understanding too independent. And making them independent to that degree makes it hard to explain the special sort of understanding involved in "seeing in a favourable light'. Certainly, that desire is not blind does not mean that it has to be subordinate to understanding. 20.See MacIntyre 1981, p. 62: "To have understood the polymorphous character of pleasure and happiness is of course to have rendered these concepts useless for utilitarian purposes'. 21.A somewhat similar conclusion is reached, though by a different argument, in Narveson 1967, see esp. ch. III. 22.It is not that "utility' provides a complete account of value, because "value' is used very broadly---for instance, of the moral categories "right' and "duty'. But it does cover part of value's domain: what enhances a person's own well-being. 23.It is often, even usually, understood that way. See e.g. Rawls 1982, p. 175: " ... the one rational good is satisfaction of desire or preferences, or, more generally, the satisfaction of the most rational ordering of desires and preferences'. Rawls' calling "utility' a "good' suggests that it is a substantive value. (Also he implausibly equates "utility' with "the satisfac- tion of the most rational ordering of desires'; "well-being' can be plausibly represented by an account that blends actual and ideal desires.) Discussion of the sense in which Aristotle's eudaimonia is an end is helpful here. See e.g. Hardie 1965, Ackrill 1980, McDowell 1980. 24.Since utility is not a substantive value at all, we have to give up the idea that our various particular ends are valuable only because they cause, produce, bring about, are sources of, utility. On the contrary, they are the values; utility is not. "Utilitarianism regards them [i.e. a person's attachments, aims, plans, etc.] as worthless in themselves and valuable only to the extent of their effects on utility. They are not any more important than what happens to be caught in the impersonal metric of utility', Sen and Williams 1982, pp. 5--6. We also have to give up the view that utilitarianism is reductionist. Talk about "utility' places all prudential values in a structure (of informed desires). Sen and Williams also speak of "this high degree of assimilation' in regarding "all interests, ideals, aspirations, and desires as on the same level' (p. 8). When they speak of "the same level', they must have in mind reduction to a common substantive value. But the structure of informed desires is not that at all. So most of the things thought to follow from this "reduction' do not follow: e.g. "the neat model of 323 NOTES TO PAGES 32--34 maximizing one homogeneous magnitude' (p. 16), the conception of "utilitarian rationality' (pp. 16--17). See also Bernard Williams' essay, "Conflicts of Values', in Williams 1981, p. 78: "The most basic version of the idea that utility provides a universal currency is that all values are versions or applications in some way of utility ... Indeed, in this version, it is not clear that there is more than one value at all, or, consequently, real conflicts between values.' 25.See Scanlon, 1975 pp. 656--8, for nearly the same principle of division (he attaches "solely' to "depends' in the explanation of "subjective'). 26.I come back to these questions in ch. III sect. 7 and ch. IV sect. 5. The label "objective-list theory' I take from Parfit 1984, see pp. 466, 493, 499. Parfit defines the objective-list theory so that it is committed to there being facts about value independent of, and determining, desires ("desired because valuable, not valuable because desired'); see p. 499. So, on that definition, it cannot merge with the informed-desire account; they are no longer both able to agree to the compromise position about the relation of value and desire set out in sect. 3. 27.e.g. Rawls 1972, sect. 15, but also sects. 64, 78; Scanlon 1975. 28.R. M. Hare mounts a similar attack on talk about objective and subjective moral judgments; see Hare 1976b and Hare 1981, ch. 12. 29.See, e.g., Brandt 1979, ch. VI, esp. pp. 110--15. In order to bring out how demanding a requirement the word "informed' is meant to mark, think of a person, apparently fully informed of all relevant facts, who none the less has some (to us) crazy aim in life---say, counting the blades of grass in various lawns. We protest that no one is the least bit interested in the information, and he replies that he knows. We insist that the information is of no use to anyone, and he replies that he knows that too. We can find no logical error that he is committing, nor any matter of fact that he denies. Still, he would not, on the standard I am using, be fully informed. It is, of course, hard to fix the boundaries of the "factual'. But, besides that, one could not say that his desire was formed by appreciation of the nature of the object (my sense of "informed'). Prudential values are reasons for action. Reasons are not idiosyncratic; they must be generally intelligible as reasons. Suppose that we got the blade counter to admit that the time he spent pursuing his aim could be spent doing other things, that certain other things that he could do would enhance his health or give him more enjoyment, while his crawling around on his hands and knees in all weathers is ruining his chest and is very boring, and so on. Then we could see his blade counting under no description that made it intelligible as desirable. Nor could he. He could go on reporting, as a bare matter of fact about his psychology, that he wanted to count blades of grass. But he could not say that it was his aim or end in life, because those terms (at least the uses that are relevant here) imply something more than that bare psycho- logical fact---namely, that blade counting makes his life valuable. His desire to count blades of grass would not be a desire formed by a 324 NOTES TO PAGES 34--35 perception of the nature of the object of desire. The perception of what it is to count blades of grass would not be, could not intelligibly be, what formed his desire. See later discussion in ch. IV sect. 3, ch. VIII sect. 3. 30.This point is well developed in Slote 1983, ch. 2, but Slote makes it into more of a point against having life plans than I think it is. 31.Here my direction is contrary to Derek Parfit's in Parfit 1984, ch. 15. He argues that if one adopts his Reductionist account of personal identity, then "It becomes more plausible, when thinking morally, to focus less upon the person, the subject of experiences, and instead to focus more upon the experiences themselves.' This would allow utilitarians to counter the objection that they treat the group as a super-person with the reply that a person is a mini-group. The Reductionist account points in that direction. But prudential value theory, which also enters import- antly, in the end has whole lives, or long stretches, as its subject, and if the assessment gives long-term goals such as accomplishment and deep personal relations dominant place among substantive values, then we cannot focus on experiences. Of course the fact that prudential assessment has to take whole lives as its subject does not, on its own, show that: the assessment could take that form but come up with the substantive answer that a life that strings together lots of certain sorts of experiences is best. It is just that it looks as if the substantive answer is not that. See also ch. XII n. 3. 32.The case is defended in Sen 1979. 33.There are also doubts about aggregation that arise from time preferences. Some periods matter to us more than others. I do not want to consider our tendency to discount the future just because it is future; there seems to me little justification for that. But a year at the prime of life does seem to matter more than a year of callow youth or of senility. So why think that the best life can be found simply by summing the value of its several periods? But why not? The quantities that are added already allow for the lower value of callow youth or senility. The fact that many desires in childhood and senility are intense does not matter. Intensity is not the sort of strength relevant to utility. And the objects of these intense desires will often not weigh heavily as utilities. The periods that characteristically matter less are characteristically of lower utility. There is no reason here for "greatest sum' and "most valuable' to come apart. These matters are well discussed in Slote 1983, ch. 1. Of course, my conclusion needs to be backed by a fuller account of the measurement of utility, especially of how one compares fulfilment of desires in one period with fulfilment in another, when the preferences may be greatly changed (see the discussion of Brandt's objection to desire accounts, ch. I sect.5; also chs. VI and VII). My view is roughly this: there is a more stable, authoritative set of values behind the desires that grip us at various times of our life. They are the ones we all use in saying 325 NOTES TO PAGES 35--38 that certain periods of life---childhood, senility---matter less than equally long periods of the prime of life. But the same values are available for utilitarian calculation. 34.The locus classicus is the Nicomachean Ethics, and the best modern work has been done in interpretations of it; see this ch. n. 23. Aristotle said that we agree on what to call the principle---eudaimonia---but as to what it means, he observed, "opinions differ', NE I.iv.2. 35.See Ackrill 1980 for an account of eudaimonia as a substantive inclusive end. For the distinction between a "dominant' and an "inclusive' end, see Hardie 1965. 36.That is the excellent semantic analysis of "good' in Ziff 1960, ch. VI, see esp. p. 218. His account seems to me substantially correct and in need of only a minor qualification. He intends his account to apply to all non-deviant uses of "good', except for those he quite plausibly regards as special cases (such as "a good mile off'), which probably grew out of the central use but which a semantic analysis of the central use need not cover. But there are uses of "good' that are non-deviant and not special that Ziff's Semantic Analysis should cover but does not. For instance, there are many uses of "good' where it means roughly "has the defining characteristics to a high degree', as in the phrase "a good Freudian slip'. Here no interests need be relevant. Although it is possible to have interests that Freudian slips satisfy (an author of a psychology textbook might), still it is also possible for a Freudian slip to be a good one simply because it is all that it should be. But the respects in which Ziff's account fails do not matter for our purposes. All the uses of "good' that are of interest to moral theory are the sort that Ziff's account fits. 37.See, e.g., R. B. Perry, Realms of Value, excerpt quoted in Brandt 1961, p. 265. 38.Of course, that is not the end of doubts. I have followed the utilitarian assumption---actually it is far more widespread---that we have a notion of good or valuable life independent of our notion of what it is right or wrong to do. Indeed, utilitarians bring this notion to moral theory and define a standard of right and wrong in terms of it. But there is a tradition that holds that good has status only in the context of a theory of right, or that the concept of right is prior to the concept of good. See Kant, Critique of Practical Reason, pt.1, bk. 1, ch. II; Rawls 1972, esp. pp. 30--3, 395--6, 560. But the desire account qualifies the independence of good and right in certain ways. One thing that people with mature values take as an aim is acting morally. Ethical push comes to incorporate much of ethical pull. Moreover, when a person faces a conflict between his own interest and his moral obligation, the latter wins: moral reasons trump prudential reasons. But there still remains a conception of a valuable life independent of, not subordinate to, right. That immoral desires make no claim on fulfilment, that they have no authority in determining action, which seems to be the nerve of Rawls's claim, does not mean that they 326 NOTES TO PAGES 38--41 cannot make one better off. And I doubt whether even an ideal person, whose values are perfectly developed, will find that ethical push and ethical pull entirely coincide. But certainly with the mass of humanity they do not. In thinking about others (say the inept sadist who can rise to no richer life), we might well decide that their welfare is greater for their acting wrongly. Similarly, in thinking about our own lives, we may decide that, though we have no sufficient reason to act immorally, it would increase our well-being to do so. See the discussion in Slote 1983, pp. 122--3. I discuss these issues more fully in ch. IV sect. 4 and ch. VIII sect. 2. 39.J. Bentham, An Introduction to the Principles of Morals and Legislation, ch.I, sects. 1 and 3. 40.J. S. Mill, Utilitarianism, ch. II, para. 5. 41.Mill's references to "utility' in On Liberty are well in accord with the desire account. "I regard utility as the ultimate appeal on all ethical questions; but it must be utility in the largest sense, grounded on the permanent interests of man as a progressive being'; see ch. I, para. 11. Note the link between utility and interests, moreover to permanent interests (a notion behind which there is clearly a full conception of substantive prudential values, about which people can be mistaken). Note also the stress on progressive beings: there is ample evidence in On Liberty that Mill, like Aristotle, saw activity as central to a good life, not static states of mind or possession of goals, but forms of life that change as one's values mature. For a good discussion of the importance of activity to Mill, see Gray 1983, ch. IV, sects. 1 and 2, esp. p. 72. 42.Sidgwick 1907, pp. 111--12; see generally bk. I, ch. IX; bk. III, ch. XIV; bk. IV, chs.I--III. CHAPTER THREE 11.These points are argued in Rawls, 1972 sects. 6, 15, 28, 60, 68; Rawls 1975a; Rawls 1975b; Rawls 1982; Scanlon 1975; Scanlon 1982; Dworkin 1981; Feinberg 1973, ch. VII sect. 5; Wollheim 1975; and by many others. 12.Both Richard Wollheim and Amartya Sen hold eclectic positions; see Wollheim 1975, Sen 1980, Sen 1980--1. 13.For a good discussion of the semantic differences, see Wiggins 1985. See also Braybrooke 1968; Richards 1971, ch. 3 sect. 3; White 1971, p. 114; Feinberg 1973, ch.VII sect. 5; Wollheim 1975. 14.I owe the example to Garrett Thomson. 15.This is an analysis defended, e.g., in Barry 1965, ch. III sect. 5A; and attacked, e.g., in Miller 1976, ch. IV sect. 2 and in Wiggins 1985. 327 NOTES TO PAGES 42--44 16.A distinction is often drawn, for the whole class of needs, between instrumental and absolute (those not instrumental to a further end). But "instrumental' fits statements such as "an element needs a free electron in order to conduct electricity' poorly; a free electron is neither an instrument for bringing conduction about (it is the instrument of the conduction) nor a means to that end (which "instrument' suggests). And "absolute', as I shall shortly argue, fits basic human needs poorly. For the distinction roughly as I draw it, see Braybrooke 1968, p. 90; Feinberg 1973, pp. 111--12. 17.Miller 1976 takes the tough line that they are not elliptical, because what is needed in these cases is part of the relevant end and so the means-end model does not fit; ch. IV, sect. 2. For instance, if we say that humans need satisfying human relations in order to be happy, what we give in the last clause is the whole of which the satisfying relations are a part, not the end to which they are a means. But the part- whole model does not fit all basic needs. It is true that the means-end model has to be stretched beyond simple cause-effect relations to fit all the cases to which it is usually applied, and the interpretation of the means-end model gets difficult with the relation of ends that have more final ends. Still, if the model does not fit at all, talk about "need' is improper. The question whether basic need statements are elliptical is connected with the question whether the sense of the word "need' changes from instrumental to non-instrumental cases. I do not know of any successful arguments that it does (but see Wiggins 1985 for an arresting one). What is at issue is particularly obscure. The best case for saying that there is a separate non-instrumental sense of "need' also allows that the content of the "in order to u' clause, with its mention of health or harm or functioning, gets incorporated into the explanation of the meaning of the word "need'. The case for saying that there is no separate non-instrumental sense allows that health or harm or func- tion are usually taken as read. Both, therefore, give health or harm or functioning central place in the explanation of the sense of the whole sentence. And that is what matters for moral theory. Nothing of importance to morality rests on there being one sense or two. 18.Is this even an account of "well-being'? Perhaps basic needs are indeed the best index, for moral contexts, of how well a person's life goes, but perhaps they take in too little of a person's life to be called, with its all-encompassing air, "well-being'. But well-being does not always suggest how life goes as a whole. We are happy using "welfare' in political contexts and confining it to some conception of basic provision. However, for discussions of "needs' and "welfare' that point in the opposite direction, see Richards 1971, pp. 37--9. 19.See esp. Rawls 1972, sects. 15, 25, 60. 10.See Miller 1976, pp. 136--8; Benn and Peters 1959, p. 146; Feinberg 1973, p.110; Wiggins 1985. 328 NOTES TO PAGES 44--48 11.See Feinberg 1973, p. 110: " ... what is needed to live a minimally decent life by the realistic standards of a given time and place and what is only added {"gravy{' ... ' 12.See Benn and Peters 1959, p. 146: "Basic needs are thus fairly precisely determinate because they are related to norms set by conditions already very widely enjoyed. They are the needs for precisely those things that most other people have got.' 13.It is not that "basic need' is essentially contestable, which all normative notions are often thought to be. Nor is it just that it is vague. Vagueness can be cured by stipulation; essential contestability may be incurable but seems not fatal. I am saying that "basic need' is too indeterminate to do the work assigned to it, that it cannot be made determinate by pointing out its links to notions indeterminate in much the same way, and that arbitrary stipulation undermines its status as a moral notion. I am not saying that the indeterminateness of these key notions ("harm', "health', "proper function', "minimum provision', "human right', etc.) cannot be overcome. There are ways, not arbitrary ones either---one being to fit these notions inside a broadly utilitarian, desire-based moral theory and to let utilitarian considerations determine where it would be best to fix the boundaries. But that is just another way of undermining need accounts. 14.If my crippled son has a happy disposition and leads a productive life and I have another son who is an unhappy neurotic, I might have good moral reason to help the second before the first. 15.We get closer if we bring in rights. The best case for offering the cripple the wheelchair or nothing rests on the nature and extent of rights. We have a right to bodily integrity---roughly, to those basic physical capacities necessary to live a recognizably human existence. But what if all the cripple's aims in life are sedentary and he is already living a recognizably human existence? Anyway, how do we fix what is a "recognizably human existence'? (That is the problem of the minimum acceptable level all over again.) And is freedom from depression a right? Could the cripple trade the wheelchair for psychotherapy? We are not yet very close to moral bedrock. See the discussion of rights later, ch. XI. 16.For an example of an account of well-being that gives health this sort of priority, see Schwartz 1982. 17.One reason why it is not important is that those are desires that society would want to stamp on anyway. Another is that the strategy backfires. If the society is welfare egalitarian, then anyone who develops expensive tastes to skew allocation in his favour will get more goods than his mates (to bring him up to their level of welfare) but not more welfare (either than they get or than he would otherwise have got). In fact---this is why his strategy backfires---he will be worse off for his strategems than he would otherwise have been, because everyone's welfare level will go down, his included, as a result of his expensive tastes. See discussions in Rawls 1975b and in Dworkin 1981, Part I, to the second of which I owe the point about the strategy's backfiring. 329 NOTES TO PAGES 48--52 18.John Rawls develops this line of thought in Rawls 1975b. 19.This is what is wrong with the very similar argument that stresses not the moral irrelevance of these chance factors but the moral relevance of our responsibility for the desires we end up with. It is appropriate to hold us responsible for many of the desires we form, and not for most of the needs that assail us. If I form expensive tastes, it is perfectly reasonable to ask me, even to give me every incentive, to give them up and choose again. Perfectly reasonable, so long as it is tastes that are in question. But then it distorts the issue to concentrate just on expensive tastes. There is a lot that falls outside the class of "basic needs', and so in the class of "mere desires', that are nothing like tastes. It is far from reasonable, if a person has responsibly chosen a life centred on accomplishment, understanding, and beauty---because these are what would make something worthwhile out of his life---to tell him to give up these aims because they are more demanding on society's resources than many other persons' approach to life. Once one realizes how much ends up in the class of "mere desires' besides "expensive tastes', the case for ignoring "mere desires' is much less appealing. 20.Again it is John Rawls who makes this point; see Rawls 1982. 21.See Rawls 1972, p. 527; also Rawls 1982, p. 172 and Scanlon 1975. 22.Difference in scale has not received enough attention. In A Theory of Justice Rawls confines attention to "social justice', but he thinks that the Original Position, constructed to generate social principles, is the device for generating most moral principles (see Rawls 1972, pp. 120, 130, 139, 255). In The Republic Plato makes similar disturbing shifts in scale: "Justice can be a characteristic of a single individual or of a whole city, can it not? ... We may therefore find justice on a grander scale in the larger thing, and easier to get hold of' (368E). 23.For an important version of a flexible need account, see Scanlon 1975; for his account of "urgency' see esp. pp. 660--1. 24.This weakens another line of thought responsible for a lot of sympathy for need accounts. We all think morality must work with a hierarchy of values. Trumping of some sort goes on and has to be explained. And how can it be explained except through some such device as basic needs? See e.g. Braybrooke 1968, p. 93, where he gives as a "fundamental principle' of morals and prudence: "Matters of need take precedence over matters of preference'; also for defence of a more qualifying version of the principle (the trumps are not just basic needs but "bad, entrenched, non- substitutable needs'), see Wiggins 1985. Certainly values form a hierarchy. But the need account is not the only way to generate the hierarchy, nor in any obvious way the best. Basic needs do not trump mere desires. The trumps would have to come out of a more flexible need account: important needs trump mere desires. But "important' is "harmful if denied', which is "damaging to prudential values if denied'. So we are back to informed desires which not only form a hierarchy but also contain resources for introducing much complexity into it. For instance, the desire 330 NOTES TO PAGES 53--56 account may end up restricting the range of informed desires that have moral weight on some such grounds as: (1) life goes better if persons do not inhibit their natural attachments and commitments, if, in effect, they accept some diminution of impartiality; (2) many effects on well-being are too slight to warrant the costs of moral sanctions; (3) on the large social scale knowledge of individual conceptions of the good is so unreliable and assessment of their value so contentious they ought in large measure to be ignored, etc. Certainly the class of my informed desires is larger than the class of my interests that you have morally to take into consideration. But there are many explanations of that. 25.See ch. II sect. 4. 26.The notion of "human nature' will have the consequences that the need account draws from it only if it already has a large normative component. Not all ages would have accepted the inferences from "human nature' that most of us nowadays will. This creates problems for those who believe that the application of "objective' is confined exclusively to values that are all-purpose means to any conception of the good life. Perhaps they cannot get around the difficulty. In any case, there are other ways of applying "objective'. 27.For defence of the neutrality of moral standards, see Rawls 1972, p. 527; Dworkin 1978; Dworkin 1981; Scanlon 1975; and very many others. 28.Thomas Nagel's argument in Nagel 1980 gets very close to this; see esp. pp. 122--3. For comments on it, see above ch. II sects. 3--4. 29.See the discussion above, ch. II sects. 3--4. 30.See above ch. II sect. 4. CHAPTER FOUR 11. The contrast that John Rawls draws between his view (justice as fairness) and perfectionism, on the one hand, and utilitarianism, on the other, is certainly implausible. He says that his own view "shares with perfectionism the feature of setting up an ideal of the person that constrains the pursuit of existing desires. In this respect justice as fairness and perfectionism are both opposed to utilitarianism.' See Rawls 1972, p. 262. Yet they are not opposed to utilitarianism on the informed-desire account. The only versions of utilitarianism they would, in this way, be opposed to are ones that had no "ideal of the person' and no way of criticizing "existing desires', which would be pretty inadequate even as versions of utilitarianism. 12.This distinction might already set off alarm bells. Do not most of the historically important forms of perfectionism pass imperceptibly 331 NOTES TO PAGES 57--60 from "prudential' to "moral' concerns? Can any account of purely "prudential' perfectionism get far without bringing in "moral' matters? Is there any---especially any perfectionist---notion of "prudence' indepen- dent of the notion of what is "morally right'? My drawing a distinc- tion between "prudential' and "moral' perfectionism does not beg any of these questions. The distinction is not meant to be beyond revision, and I come to these reasonable qualms later. 13.The transition from the notion of a perfect (teleios) life to that of achieving an end (telos) was made easy for a Greek by the etymological connection of the words. In English, however, "perfect' comes from Latin perficere, the roots of which are per plus facere, i.e. thoroughly made, completed. "Perfect', in its contemporary use, is "flawless', and there is no longer any close connection with the realiza- tion of an end. I take these points from Passmore 1971, p. 20. Later, therefore, we shall have to look at the merits of a single ideal form of life independently of Aristotle's teleological metaphysics. 14.We see here, that when Aristotle asks whether there is one task or end, he wants to know not whether there is a single one but whether there is a unique one. But that is puzzling. There may be rational beings other than humans somewhere in the universe, and then rationality would not be unique to humans. Also he ends up saying that our final end is pure rationality, theoretical contemplation, a life, as Aristotle remarks, in a way more divine than human. This is a curious culmination for an argument that makes so much of the notion of what is unique to humans: what is unique to them is something not unique to them but what they share with the gods and possess on their own only in a debased form. This is a plain contra- diction, the best way of dealing with which is simply to drop all claims of what is unique to humans and concentrate instead on their single end. The claims about uniqueness are, anyway, not essential to the argument. Aristotle's concern seems really to be the search for the one peak to what he sees as the hierarchical structure of human functions, and it would not matter if some other species turned out to have a peak of the same shape. 15.See e.g. Hardie 1965 for a discussion of "dominant end' and "inclusive end' and also Nagel 1972 for a discussion of an "intellec- tualist' and a "comprehensive' account of eudaimonia. 16.NE, x 7 1177a, b; 1178a. 17.NE, x 7 1177b; x 8 1178a. 18.ch. II sects. 3, 4. 19.This, and the following forms of moral perfectionism, can come in both hard and soft versions: the hard version would let this perfec- tionist principle trump all other moral principles (Nietzsche); the soft would make it one among many equally weighty principles (Rashdall; see below). 332 NOTES TO PAGES 60--66 10.Nietzsche 1874, sect.6. 11.Quoted in Clark 1975, p. 51. And the not-so-young Russell wrote to Gilbert Murray (27 April 1949), "Where democracy and civilisation conflict, I am for civilisation'; also quoted in Clark, p. 631. 12.Sometimes the Superman version becomes the more generous, though also more dangerous, Super Race or Super Class version, which says that it is not a few individuals who represent the supreme form of life, but a whole group---a natural aristocracy, say, or whites. Hastings Rashdall could write " ... the lower Well-being---it may be ultimately the very existence---of countless Chinamen or negroes must be sacri- ficed that a higher life may be possible for a much smaller number of white men'. Rashdall comes out with this appalling remark to illustrate a perfectly plausible principle, and the general tenor of his discussion is far more balanced and sympathetic than this unfortunate passage suggests. The principle is that, in certain circumstances, it might be right to purchase a high kind of life for a few with sacrifices on the part of the many. But he shows himself well aware of the intolerable conditions that present social inequalities produce. "I for one should certainly doubt whether, if I had the power, I could doom the world to a continuance of our present social horrors, although their removal might lead to the evanescence of research and specula- tion, {"sweetness and light{', full and varied exercises of faculties, and all the rest of it.' See Rashdall 1907, vol. 1, pp. 238--9. 13.The move to the Super Class or Super Race versions would just increase the problems. The empirical ground on which they are meant to stand is missing. It is just false that all the members of one race have greater capacities for knowledge or the appreciation of beauty than the members of another. Nor is there any natural aristocracy with a greater share of these capacities than some natural proletariat. There are only smooth gradations of degree in share, and often those fortunate in their share of one are unfortunate in their share of another. 14.For a defence of the occasion-by-occasion approach, see McDowell 1979. Martha Craven Nussbaum takes seriously the need to show in some detail how the approach works; see Nussbaum 1983--4. But even her rich, subtle discussion of a particular case of human relations leaves it entirely unclear how this approach could possibly answer questions such as the ones I am shortly to list in the text. 15.This is Philippa Foot's characterization in her article "Virtues and Vices' in Foot 1978, see esp. p. 8. 16.Leo Tolstoy, A Confession. 17.Russell 1953, p. 51. 18.An observation of Paul Edwards in Edwards 1967. 19.Radical argument---in contrast to the sort of definitional argument that I described earlier---is possible, but I think only in the sense that 333 NOTES TO PAGE 66 people do offer radical arguments, which need assessing. I do not think that any of them is, in the end, persuasive. "Radical' arguments, as I use the term, are sceptical in nature; they aim to show that the values that definitional deliberation establish are not really valuable at all (the nothing-is-valuable argument that I have just sketched) or that they lack the seriousness and importance in life that we ordinarily think of them as having. A further "radical' argument (besides the nothing-is-valuable argument) might therefore be a crude anti-realist argument that all values are just a matter of taste and so give no real reason to adopt this end or that. I see my argument in this chapter (and in this book) as neutral between realism and anti-realism on values. But it is not neutral when it comes to certain crude forms of anti-realism. A form that claims that all values are a matter of taste is not adequate to the phenomenology of deliberation about values or to our ordinary notion of "taste'; a form that claims that we have no reason to adopt one end rather than another uses "reason' in an indefensible way. This crude anti-realism is, I think, another unsuccess- ful "radical' argument. Deliberation about values that matters for prac- tical purposes answers the questions: What values are there? and How do they figure in my life? "Radical' arguments answer the second question: values matter much less than we ordinarily think. Since I think that they fail, in effect the only practical question that remains is, What values are there?, which the "definitional' argument seeks to answer. 20.Thomas Nagel, in his paper "The Absurd' in Nagel 1979, having accepted all of these doubts about backward steps, argues convincingly that there is indeed a real step back, but of a different kind (p. 15). It does not take us to a perspective from which life looks absurd. Absurdity does not enter in that direct way. It is, rather, a critical perspective on our ordinary value judgments. "The crucial backward step', he says, shows "that the whole system of justification and criticism, which controls our choices and supports our claims to rationality, rests on responses and habits that we never question, that we should not know how to defend without circularity ... ' (p. 15). And yet we do no disengage from life. We take our ambitions and commitments utterly seriously. That, Nagel submits, is just the absurdity. A situation is absurd when it contains a discrepancy between pretension and reality, and such a discrepancy inevitably appears in all human lives in "the collision between the seriousness with which we take our lives and the perpetual possibility of regarding everything about which we are serious as arbitrary or open to doubt' (p. 13). But it is not true that our values rest on responses and habits we never question. Our values are subject to all the kind of criticism that we have been talking of. And the critical perspective that we may adopt about our value judgments we may also adopt about other judgments---our scientific beliefs, our mathematics. In their case, too, criticism leads ultimately to questions that cannot be answered without 334 NOTES TO PAGES 68--70 circularity. Our values can be issued challenges that reduce us to silence. But what should our reaction to this be? Should we feel that it is absurd for us to go on, as we are bound to do, taking ourselves seriously? Should we feel this about the sciences and mathematics, and our carrying on, as we are bound to do, taking them seriously? Our reaction in those cases is not, and should not be, pessimism, despair, or a sense of our absurdity. It should be caution and open-mindedness. Nagel's step back is a real one, but it leads to modesty, not to absurdity. 21.This list may not be complete. But other candidates that I can think of, and items on other proposed lists, seem to me either not to belong, or to be subsumable under one of the headings already there. Is human life a value? (see Finnis 1980, pp. 86--7). No, unless it comes down to item (b): the features of a human existence. Good life is valuable, but it is not one of the substantive values; it is the sum of them, a valuable life. Is play? (see Finnis 1980, p. 87). Not, it would seem, on its own; it is part of enjoyment, and we need that broader category. Is the development and exercise of skills? (see Dworkin 1981, Part I). Well, not exactly under that description. There are rare but trivial Guinness Book of Records skills that have no value. The development and exercise of any skill is enjoyable, and of non-trivial skills a central part of accomplish- ment. And so on. 22.John Rawls thinks they do. He speaks of "the conception of the person' in Sidgwick's version of classical utilitarianism as " ... that of a container-person: persons are thought of as places where intrinsically valuable experiences occur ... Persons are, so to speak, holders for such experiences. It does not matter who has these experiences, or what is their sequential distribution among persons ... Assuming that no such experi- ences have a duration greater than a specified interval of time, one need not ask whether a person having a certain valuable experience in the present interval is the same person who had a certain valuable experience in a previous interval; the temporal sequence over intervals is no more relevant than distribution among persons within the same interval.' (See Rawls 1974--5, pp. 17--18.) This conception of the container-person collecting agreeable experiences which themselves are seen as temporary, discrete, and ephemeral, cannot be got out of any plausible account of prudential value. Indeed, any plausible account would undercut it. But Rawls is too quick to think that the inadequacy of one particular value theory shows something about the adequacy of utilitarianism as a moral theory. 23.The stronger line is argued in McDowell 1980. 24.Shakespeare, Measure for Measure, III.i, Claudio: Death is a fearful thing. Isabella: And shamed life a hateful. 25.We also rank natures, e.g. human nature with its capacities and the satisfactions and dissatisfactions arising from them against canine nature with its. Or human nature against New Man, a breed produced by 335 NOTES TO PAGES 71--79 genetic engineering to be contented slaves (see the discussion in Haksar 1979, pp. 3--4). 26.See references in ch. III n. 27. 27.See ch. III, sect. 7. 28.See Rawls 1972, p. 327, "... they [persons in the Original Position] are assumed to be committed to different conceptions of the good ... '; also Rawls 1980, p.539, "... justice as fairness assumes that deep and pervasive differences of ethical doctrines remain'; see also pp. 540--2. But why not assume instead that persons in the Original Position would (or could inoffensively be got to once they emerge from behind the Veil of Ignorance) agree on at least some goods beyond all-purpose means? Suppose that persons in the Original Position thought that they might turn out to have conceptions of the good that agree at certain points and that the points are quite central and important. Why would they not think it? Indeed, they might also reasonably think that substantive prudential value theory might show some values, besides all-purpose means, to be valuable to any human life. But then would they agree to a policy of strict value-neutrality? Would they not hedge their bets---"if that is the way things turn out, let us have some non-neutrality'? 29.See ch. II sect. 4. CHAPTER FIVE 11. This is argued in ch. II sect. 5. 12.e.g. Raz 1985--6, p. 117. 13.I discuss certain scales and their relative strength in ch. VI sect. 1. Since scales can be ranked from weak to strong (depending upon how much can be said about items as a result of their being measurable on the scale), we can speak loosely, as I do in the text, of weak and strong forms of incommensurability (the weaker the scale the stronger the form of incommensurability if two items cannot be measured on it). 14.This chapter bears the title of an article that I published a few years ago (Griffin 1977). Little of the article survives here; what does appears in nn. 28 and 34. 15.For an argument for incommensurability along these lines, see Thomas Nagel, "The Fragmentation of Value', in Nagel 1979. 16.Robert Nozick expresses the hope. See Nozick 1981, ch. V sect. 3, esp. the part entitled "Measurement of Moral Weight'. 17.As Ronald Dworkin claims. See Dworkin 1977, Introd. pp. xi--xv, chs. 6 and 7. 18.For my attempt at a substantive theory of rights, see ch. XI. 19.See chs. XII and XIII. John Searle claims, correctly I think, that it is a conceptual truth. But that promises essentially create obligations says 336 NOTES TO PAGES 79--80 nothing about the moral strength, if any, the obligations have. See Searle 1978, p. 86. 10.See ch. X sect. 2. 11.For the claim that certain moral conflicts lack the sort of rational resolution that would come from one obligation's outweighing another, see Williams 1973b, p. 173; Williams 1981, p. 74; and Wiggins 1976, p. 371. 12.Then we should get incommensurability of the sort that T. S. Kuhn may think holds between scientific theories with radically different paradigms. I speak of what Kuhn may think, because, in talking about "incommensurability', he sometimes speaks more strongly than he does at other times. I am here taking him at his strongest. Unfortunately, the powerful objections to the strong versions of Kuhn's claim about scientific theories cannot simply be shifted to moral theories without begging a lot of questions. See Kuhn 1970. For powerful criticism, see Newton-Smith 1981, chs. V--VII. For a claim that moral reasons exhibit just this form of incommen- surability, see MacIntyre 1981. MacIntyre suggests that "moral incommen- surability' arises from our now using concepts from incompatible moral traditions. Hence, when rights are matched against utility, and both against justice, "...it is not surprising that there is no rational way of deciding which type of claim is to be given priority or how one is to be weighed against another. Moral incommensurability is itself the product of a particular historical conjunction' (p. 68). MacIntyre describes an import- ant possibility, though his particular instance (rights, justice, and utility being from incompatible traditions) is unconvincing. 13.For example, Agamemnon's choice between the expedition and his daughter. That is Bernard Williams' example in Williams 1973b, p. 173, of which he says, "The agonies that a man will experience after acting in full consciousness of such a situation are not to be traced to a persistent doubt that he may not have chosen the better thing; but, for instance, to a clear conviction that he has not done the better thing because there is no better thing to have done'. See also Williams 1981, p. 74, and Wiggins 1976, p. 371. 14.Furthermore, the difference between the two values that we are invited to try to compare is often greatly exaggerated. John Rawls, in Rawls 1982, pp. 179--80, invites us to consider different conceptions of the good life, one affirming aesthetic values, attitudes of contemplation towards nature, gentleness (Friends of the Earth), the other valuing risk, excite- ment, competition. "These conceptions of the good are incommensurable because their final ends and aspirations are so diverse, their specific content so different, that no common basis for judgement can be found.' If by a common basis Rawls means yet another substantive value to which they can be reduced, he is probably right that one cannot be found. But then it seldom can---even in the most ordinary, humdrum cases. If that is our standard, incommensurabilities crop up everywhere. But the basis of comparison could be different in kind: different amounts of prudential 337 NOTES TO PAGES 81--82 value, where "prudential value' is not itself yet another prudential value. In any case, the "diversity' of these two conceptions is overdone. True, gentleness and the spur of competition (to simplify a bit) are both valuable (or rather what they bring to one's life is); anyone would want to have them both in his life. No doubt sometimes one can add elements of the one only at the (painful) cost of the other, but that is partial incompatibility, not incomparability. Competition is valuable because it spurs us on to better things; gentleness is valuable because it opens life to responses and relations that are valuable. They do not come from "aspirations' that are such "diverse' approaches to life; they come from aspirations that are, or should be, universal and to some extent combined in everyone's life. There is a scale on which we can, and do, rank them: contribution to a good life. Certainly sometimes we can say that the contribution that a certain sort of spur of competition would make to our lives will be less than the contribution that non-competitiveness in that area would make. There may well be limits to, breaks in, the rankings, but that is a different matter. 15.There are several names in currency for each ordering. By a weak ordering I mean one generated by a relation that is reflexive, transitive, and complete. By a strong ordering I mean a weak ordering plus anti-symmetry (i.e. no two objects can occupy the same place in an ordering); by a partial ordering a weak ordering minus completeness. For a good summary of different kinds of orderings, see Sen 1970, ch. 1, esp. pp. 8--9. I am greatly indebted to John Broome for a discussion of these matters. 16.See Kant, Grundlegung, in Paton 1961, p. 102. 17.Charles Taylor develops a similar contrast; he speaks of the "incommen- surably higher', e.g. "Integrity, charity, liberation and the like stand out as worthy of pursuit in a special way, incommensurable with other goals we might have, such as the pursuit of wealth, or comfort...'; the latter we may pursue or not as we choose, while the former we must pursue or be "open to censure'. See Taylor 1982, pp. 135--7. 18.Or is the point about the replaceable and the irreplaceable slightly different? On the face of it, commensurability implies replaceability. But clearly some objects are uniquely valuable, beyond substitution or compensation. Without such objects certain central human emotions would not even make sense. One could not grieve for a loved spouse or child, if every lost object could be replaced. This is a theme nicely developed in Nussbaum 1984. None of that, I think, can be gainsaid. We can suffer losses, and losses of the very things we value most, that cannot be replaced no matter how favourably life subsequently goes. But we must steer clear of a type-token confusion. The prudential value is deep loving relationships to particular persons. It can be realized, of course, only in a relationship to this or that particular person. The particular person cannot be replaced, but the prudential value can have a new instantiation. If a particular loved person dies, he is dead, and his life cannot be lived by 338 NOTES TO PAGES 82--89 anyone else. But when a child dies, the parents might have another child whom they would not otherwise have had. Or when his wife dies, a man might remarry. And these new relationships can enrich one's life as much as the old. It does not destroy grief that one can love again, or that the new love be as a valuable as the old. Indeed, an inability to love again itself destroys grief by focusing not on what is lost but on one's own suffering. The irreplaceability of individuals is not the incommensurability of values. 19.Charles Taylor has this sort of clash of world views in mind in his discussion of incommensurability in Taylor 1982, sect. 2. 20.Ronald Dworkin uses the word "trump' in describing the relation of rights to utility (e.g. liberty in relation to level of prosperity); see this ch. note 7. John Rawls uses the language of "lexical ordering' in describing the relation of liberty to, e.g., prosperity; see Rawls 1972, esp. sect. 8. 21.Rawls admits this; he thinks that liberty has priority in a certain range, viz. after "a certain level of wealth has been attained'. See Rawls 1972, sects. 26 and 82, esp. pp. 152, 542--3. 22.See, e.g., Popper 1966 vol. 1, pp. 284--5: "...there is, from the ethical point of view, no symmetry between suffering and happiness... Instead of the greatest happiness for the greatest number, one should demand, more modestly, the least amount of avoidable suffering for all'. Popper has in mind their moral weight, and we are now interested in their prudential weight; still, one would expect the same asymmetry there. For general discussion of these issues, see Griffin 1979. John Rawls approvingly quotes Petrarch's remark that a thousand pleasures are not worth one pain, and correctly observes that Petrarch thereby "adopts a standard for comparing them that is more basic than either'; see Rawls 1972, p.557. 23.I go into these points more fully (and discuss negative utilitarianism generally) in Griffin 1979. 24.I argue this in ch. XI sect. 9. See also the excellent Hart 1975, where this line of argument is powerfully developed. 25.Rawls, of course, thinks that it is. He thinks that I should be reluctant to surrender the (small) liberty to gain the (great) prosperity because I should be putting at risk any capacity to pursue what most matters to me---the central part of my life plan. But it may be that I would (surrender some political liberty) because it would help me to pursue what I most value (the relationships with people I love who live in a country without those political liberties). To know this, I should have to be this side of the Veil of Ignorance, but then I am. For a more general discussion, more closely tied to the terms of Rawls' argument, see Hart 1975, which raises powerful doubts about it even in these tighter terms. 26.I shall return to this also in ch. XI; see esp. sect. 9. 27.I owe many of the thoughts expressed here to Derek Parfit's pressing me for explanations and to his offering many himself. Readers of Parfit 1984 will hear echoes in my discussion of small step-by-step changes of the 339 NOTES TO PAGE 89 argument he presents for what he calls The Repugnant Conclusion (see his ch. XIX). My example and his are obviously very close. In one version (sect. 148) Parfit's argument is this. Consider a world, A+, of twenty billion people, half of them at an extremely high quality of life and the other half at an even higher one. Then consider another world, New A, with many extra groups of people at a very low quality of life but just high enough so that it would not have been better in itself if they had never existed. Now, New A is better than A+ in certain respects: first, there are twenty billion people all of whom are at a higher level than anyone in A+; second, New A is, Parfit argues, better in egalitarian terms. And since there is no respect in which New A is worse than A+, New A is better than A+. Now compare New A with New B. Though here the better-off group would lose, the worse-off group would gain far more. So unless we accept Maximax or some strict form of e$0litism, we cannot think that this outcome is worse. New B is better on any plausible view of beneficence, and the inequality in New B is better on any plausible view of equality. Therefore New B is better than New A. But we can continue on through the alphabet, from New B to New C, until we reach New Z, where there is an enormous population at a 340 NOTES TO PAGE 89 quality of life which, though just worth living, is gravely deprived, crimped, and mean---not much above the level where it would be in itself bad that lives are lived. That our reasoning carries us to New Z is The Repugnant Conclusion. But does it? Parfit canvasses, but rejects, the possibility of stopping the slide at an early stage: namely, by refusing, on e$0litist grounds, to accept that New B is better than New A. I agree with him that, whatever form of e$0litism might be acceptable (see my ch. IV), the one needed to stop the slide from New A to New B is not. But there is another possibility, confined entirely to the reasoning about beneficence. Parfit's argument seems implicitly to employ a totting-up conception of measuring well- being; it treats well-being as measurable on a single continuous additive scale, where low numbers, if added to themselves often enough, must become larger than any initial, larger number. But this seems not true in prudential cases, and it would seem likely that this incommensurability in prudential values would get transferred to interpersonal calculation. Perhaps it is better to have a certain number of people at a certain high level than a very much larger number at a level where life is just worth living. Then we might wish to stop the slide not necessarily at the move from New A to New B, or even from New B to New C, but at that point along the line where people's capacity to appreciate beauty, to form deep loving relationships, to accomplish something with their lives beyond just staying alive ... all disappear. (Parfit has a new discussion of this example in a forthcoming paper, "Overpopulation and the Quality of Life', in P. Singer (ed.), Practical Ethics, Oxford: Oxford University Press.) 28.Look closely at this one. A person might sacrifice his eyesight, or his hearing, or his touch for enough pleasure in return, but he would not sacrifice all three for any amount of pleasure. (This example comes from Tribe 1972, pp. 91--2; I discuss it more fully in Griffin 1977.) Well, would he not? It would obviously take an enormous number of pleasures even to approach the value of those three senses. And then our thoughts are likely to be distracted by extraneous considerations. For one thing, pleasures crammed in such number into our lives, diminishing marginal utility is likely to apply; the values of the pleasures might form some such series as 1, 1/2, 1/4, etc. The sum of that series, even if the series is infinite, never exceeds a small finite number---namely, two. But it is no threat to commensurability that certain infinite series sum to a small finite number---or, more generally, that some values, because they diminish with crowding, sometimes to nothing, can never be added in a way that will make them equal certain other values. And if we take "pleasure' broadly, to include a wide range of prudential values, thereby avoiding to some extent the effects of diminishing marginal utility, then it is doubtful that the pleasures can never outrank the value of these three senses. Would no amount of the satisfactions of intellectual or artistic accomplish- ment compensate a person for the loss of sight, hearing, and touch? Would a composer ordered to stop work on pain of those 341 NOTES TO PAGES 89--90 impairments not think it worth going on if he were sure to have a good amount of time working at the level of the deaf Beethoven? 29.I have argued for a form of pluralism in ch. II sect. 4; see also ch. IV sect. 3. 30.See MacIntrye 1981, p. 62: "For different pleasures and different happinesses are to a large degree incommensurable: there are no scales of quality and quantity on which to weigh them'. But all that this shows is that there is no commensurability of this kind. But this kind is not even what a utilitarian ought to be claiming. 31.John Rawls seems to think it does. "The weakness of hedonism reflects the impossibility of defining an appropriate definite end to be maximized'; see Rawls 1972, p. 560. But we do not need a single end such as pleasure, or any single substantive value at all, to get a single scale. That will be the argument of the next chapter. The policy of maximization does not stand or fall with hedonism. 32.See ch. II sect. 4. This is one of Bernard Williams' interpretations of incommensurability in Williams 1981, p. 77: "It is not true that for each conflict of values, there is some value, independent of any of the conflicting values, which can be appealed to in order to resolve that conflict'. Quite right. But this, too, is an extremely weak form of incommensurability (if a form at all). Utilitarians could, and should, accept it. 33.David Wiggins, in an argument that I am not sure that I follow, seems to think that we do. As I understand him, he denies that Aristotle held that there is a single substantive value (some "universal or all- purpose predicate of favourable assessment') and so denies that Aristotle thought that there was a common measure of value. See Wiggins 1978--9, esp. pp. 267--8. But the two are not the same, nor is the first a necessary condition of the second. 34.Misunderstanding of just this point---of the nature of this scale and of its quantitative attribute---bedevils discussion of incommensurability. Ber- nard Williams complains, "In cases of planning, conservation, welfare, and social decisions of all kinds, a set of values which are, at least notionally, quantified in terms of resources, are confronted by values which are not quantifiable in terms of resources: such as the value of preserving an ancient part of a town, or of contriving dignity as well as comfort for patients in a geriatric unit' (in Williams 1973c, pp. 102--3). Part of his worry is the way cost-benefit analysts actually arrive at their social recommendations, and who can fail to share it? What is easily measured gets prominent place; what is hard to measure either gets left out (e.g. liberty, autonomy---see the standard cost-benefit analysis procedure of leaving judgements about "justice' to "politicians') or gets hasty and crude treatment (e.g. human life---see virtually any cost-benefit discussion of the subject, though see also subtle treatments by E. J. Mishan, T. C. Schelling, and John Broome). But the other part of Williams' concern is to deny that these values are commensurable. But that some values lend 342 NOTES TO PAGE 90 themselves to quantification in terms of resources and others do not shows nothing. That is not the scale of prudential value; it is, at best, a scale for a certain (limited) range of values. If we are able to enhance the dignity of some geriatric patients by giving them greater privacy or enhance their comfort by giving them better heating, how should we decide which to do? We could use the ranking of the patients themselves---so long as they understood the options---and provide them with whichever one they want more. If Williams' point is that these two values cannot be got on the same scale, then he is wrong. The patients, when informed, can rank them---and in the strong sense. If his point is, rather, that although they can be got on the same scale their value is not determined by their place there, then would he say that, even if the patients want the heating more, we decision-makers might determine the privacy to be more valuable and give them that instead? Anyway, I think that Williams takes altogether too dim a view of what it is for values to be "quantifiable in terms of resources'. It is not, he says, "an accidental feature of the utilitarian outlook that the presumption is in favour of the monetarily quantifiable... It is not an accident, because (for one thing) utilitarianism is unsuprisingly the value system for a society in which economic values are supreme...' (p. 103). But to make money the common measure is not to make it the supreme value. To think so would be doubly mistaken. First, the true scale of well-being employs no supreme value, so money could not be it. And, second, whenever---in certain ranges---money functions as the common measure, it is used transparently. If I want to preserve the old stone wall and the copse (see ch. VI sect. 3) but should prefer a certain amount of money in compensation, then I should prefer what that money would give me (perhaps greater beauty) to what the old stone wall and copse give me (beauty). I discuss these points further in Griffin 1977, esp. sect. 3. 35.Isaiah Berlin sees monism and commensurability as more closely linked than they seem to me to be. But the labels "monism' and "commensur- ability' can bear so many different interpretations that it is not easy to be sure. See Berlin 1969, p. l: "If the claims of two (or more than two) types of liberty prove incompatible in a particular case, and if this is an instance of a clash of values at once absolute and incommensurable, it is better to face this intellectually uncomfortable fact than to ignore it, or automatically attribute it to some deficiency on our part which could be eliminated by an increase in skill or knowledge; or, what is worse still, suppress one of the competing values altogether by pretending that it is identical with its rival---and so end by distorting both. Yet, it appears to me, it is exactly this that philosophical monists who demand final solutions---tidiness and harmony at any price---have done and are doing still.' Later (pp. lv--lvi) he speaks with approval of "those who are aware of the complex texture of experience, of what is not ... capable of computation', with whom he contrasts those who think "that we can uncover some single central principle... a principle which, once found, will govern our lives---this ancient and almost universal belief, on which 343 NOTES TO PAGES 91--95 so much traditional thought and action and philosophical doctrine rests, seems to me invalid, and at times to have led (and still to lead) to absurdities in theory and barbarous consequences in practice'. But this argument seems to me to overlook an important point: "computation' on a single scale does not need a substantive "single central principle'. 36.This is why it is puzzling that Berlin connects the defence of pluralism and incommensurability with the defence of liberalism. See his "Introduc- tion' to Berlin 1969; see also Bernard Williams' discussion of Berlin's views in "Conflicts of Values' in Williams 1981, and in his introduction to Berlin 1978. CHAPTER SIX 11. See, e.g., Tibor Scitovsky's categorical claim in Scitovsky 1976, p. 134 (see also p. 90): "...satisfaction or happiness cannot be measured, of course...' For a more cautious view, along much the same line, see Williams 1973c, pp. 101--2, which argues that although well-being, explained in some very crude hedonist way, might be measurable, the more plausible the explanation the less measurable well-being is. But then Kyburg 1984, p. 236, replies: "It is often held that certain qualities, despite the fact that we speak of them in comparative terms, are not amenable to measurement. I have in mind such qualities as beauty, virtue, value, and the like, which are either regarded as {"subjective{' (and thus not fit subject matter for science), or as absolute and transcendent (and thus not fit subject matter for science). On the view of measurement that I have presented, there seems to be no reason why such qualities cannot in principle be measured.' In general, it is students of measure- ment who are least sceptical about the measurement of such attributes as "utility', "preference', "well-being', and "prudential value'. 12.Stevens 1959, p. 19. 13.See Suppes and Zinnes 1963, p. 14. 14.See, e.g., Siegel 1956. Despite the large number of scales that are of practical interest, Stevens' definition over-corrects our over-narrow ideas about measurement. Not all rules assigning numbers should be called "measurement'. The one-to-one correlation of numbers and National Health Service patients, though a rule-governed assignment of numbers, probably should not. A patient's number functions too much like his name, and it seems better to restrict the term "measurement' to attributes that are in some sense quantitative (on which see below, sect. 2). This point is more fully argued in Torgerson 1958, p. 17. 15.This account of the general form of a proof of measurability is developed in the following places: Scott and Suppes 1958, Suppes and Winet 1955, Suppes and Zinnes 1963. 16.What an attribute must be like in order to be "quantitative' is partly a matter for stipulation. My requirements are relatively undemanding 344 NOTES TO PAGES 96--98 ---merely that (@EA)(@EB) (AaB |v AaB |v A=aB) where "a', "a' and "=a' mean "has more of the attribute than', "less of it than', and "same amount as'. On that definition length, weight, and hardness all count as quantities, but so do value, intelligence, funniness, and even beauty. For a more demanding requirement (the existential quantifiers of my definition become universal quantifiers) see Ellis 1968, ch. II sect. A. In contrast to both of these definitions, one of the pioneers of modern measurement theory, N. R. Campbell, has argued that proper "quantities' appear only when we get beyond ordinal scales to cardinal ones. See Campbell 1938, pp. 141--2: "Nobody, I think, has ever proposed to measure happiness or virtue by assigning numerals to represent its degree. But it is not uncommon to find mention of a {"quantity of happiness (or virtue){'; indeed Benthamite hedonism would be meaningless unless there were such a thing as {"quantity of pleasure{'. But the word {"quantity{' implies, or ought to imply, much more than a mere order . . . For all typical {"quantities{', mass, volume, cost and so on, are truly additive properties . . . happiness and virtue are not necessarily additive.' 17.See ch. V sect. 3. 18.I borrow this suggestion from John Broome, to whom I am indebted for discussion on many issues in this chapter. 19.That, it seems to me, is exactly what happens to "utility' in the course of its cardinalization by appeal to levels of discrimination. This form of cardinalization goes back to Borda and Edgeworth and has been developed more recently by Goodman and Markowitz and others. See Goodman and Markowitz 1952; for a good general discussion, see Sen 1970, ch. 7. 2. It is based on three assumptions: first, that each person is able to discriminate only a finite number of levels of his utility; second, that he is indifferent between options on the same level; and third, that the difference between one level and the adjacent ones is the same at all levels. With those assumptions, we can derive a cardinal scale unique up to a linear transformation. But although it is a cardinal scale, is it any longer a scale of utility? The discriminations a person can make depend to some extent upon what his experience has been, what training has happened to come his way. Suppose a person has drunk good, middling, and poor whisky, and has made those three discriminations. Then his experience broadens and he encounters whiskies that fall between those categories, resulting, let us say, in his introducing two categories that he did not have before. It is very hard to think that, just because his discriminations have increased, his utility from the best has thereby increased. It does not help to base the scale not on the discriminations that a person actually makes but on the discriminations that, with full experience, he could make. Some things permit finer discriminations than others. Suppose that, with my ear and my palate both at the peak of training, I make finer discriminations in my enjoyment of music than in my enjoyment of whisky. I discriminate, say, twenty levels with music, and only five with whisky. But, let us suppose, I enjoyed the best whisky as 345 NOTES TO PAGES 98--102 much as the best music. But then, if you improve my lot by giving me the music I find best instead of the merely passable music I now have to put up with, you increase my utility by nineteen units. If instead you give me the best whisky instead of the merely passable stuff I am now drinking, you increase my utility only four units. Yet you would increase it either way to levels that I equate. The trouble seems fundamental. This approach to the cardinal measurement of utility seems to be more a measure of a person's proneness to or capacities for discrimination, or of the extent to which the subject lends itself to discriminations, than a measure of his utility. 10.In ch. I sect. 4. 11.A theory of error is a central part of an adequate theory of measurement. How much variation in reports of quantities can there be, and in what circumstances, before one begins to wonder whether there is a scale at all? For a good discussion of these issues, see Kyburg 1984, esp. ch. IV sect. 4. 12.Well-being would seem to be a prime candidate for measurement on a ratio scale. There is a natural zero: a state with neither value nor disvalue, a life neither worth living nor worth not living, a condition to which one is indifferent. We should have to define "negative value' carefully, so as to distinguish between merely the loss of benefit (where zero could be treated as the status quo) and the presence of a state that makes life worse (where we have the natural zero referred to above). We should also have to see whether the global preferences we formed about when life becomes worth living agree with the piecemeal judgments we make about individual features that had value and disvalue (in the second of the senses above). Could we get the result that a life that contained a net balance of individual values over disvalues was still not quite worth living? Would this show that we had got the individual values wrong, or that we ask for a certain surplus before we think life as a whole to be worth living? If the latter, then where is natural zero? 13.Von Neumann and Morgenstern 1953. 14.The word "mystical' is L. J. Savage's; see Savage 1954, p. 94; he goes on to say that "the probability-less idea of utility has been completely discredited in the eyes of almost all economists...' (p. 96). See also Morgenstern 1972, p.1181: "All value or utility is expected value or utility, expected with certain probabilities...' 15.But not by all that much. Incommensurabilities, even of the weaker form, restrict the scope of all additive scales, probabilistic or not. Most important, the greatest restriction on the scope of the simple cardinal scale in my example comes, not from diminishing marginal utilities, but from the fact that it must measure informed desire. We cannot just ask people to compare packages of probabilities and pay-offs, because we should then be getting at their actual desires, the values that in fact they attach to objects, not at their informed desires, the values that the objects will prove to have to them. It is possible to apply a scale of utility taken 346 NOTES TO PAGE 106 in conjunction with probability, such as one satisfying the von Neumann- Morgenstern axioms, to informed preferences. But the great increase in scope comes from applying it, as is more commonly done, to the packages of utilities and probabilities between which we observe people choosing. Then, instead of imposing the tough requirement that orderings be informed, we need merely ask that they satisfy what economists call "consumer rationality'---for example, that "at least as good as' be a weak ordering. (Some economists, though, recently have been jacking up the standards for "consumer rationality'. E.g. Gravelle and Rees 1981, pp. 7--8, describes a rational decision-taker as one who (1) sets out all the feasible alternatives, (2) takes into account whatever information is readily available or worth collecting, (3) ranks the alternatives in order of preference, (4) chooses the alternative highest in this ordering. The second requirement is a big step towards what I have been calling informed preferences.) But this scale would not, in the interesting sense, be wider than mine, because it would not any longer be measuring the same quantity as mine. This scale would measure an empirical quantity: how valuable something seems to a certain person (see, e.g., Davidson, Suppes, and Siegel 1957). But what a person actually wants cannot be what we are after; a person can want what will be of no advantage to him, even perhaps of harm to him, and the notion that we are after must be kept close to what is in a person's interest. My scale measures a normative quantity: how valuable something is to a person's life. There are, broadly, two different conceptions of utility in play in modern discussions, each suiting a different theoretical setting. There is utility-e ("e' for "empirical') and utility-n ("n' for "normative'). Utility-e is the conception needed in a theory explaining how people, when subject to minimum restrictions of rationality, decide and act. It is the conception appropriate to the empirical parts of decision theory and of economics. Utility-n is the conception needed in a theory that aims to set standards for how people ought to decide and act. It is the conception appropriate to the more normative parts of decision theory, to welfare economics, to most of social choice theory, and to morality. The common, empirical von Neumann- Morgenstern scale gets its scope from its needing the presence of only subjective utility ("utility-e') and subjective probability, which can be found in very many situations of decision and action. But since moral theory needs a different conception of well-being, we have to go a different route. So, it is not that the quantity we want to measure would not allow us to cardinalize with reference to probabilities. It is, rather, that given the restrictions needed simply to get at our quantity, it is doubtful that we gain much scope by doing so. CHAPTER SEVEN 11. ch. I sects. 1, 2 and 6. 12.ch. I sect. 3. 347 NOTES TO PAGES 106--113 13.See Shapley and Shubik 1974, ch. 4 sect. 5, p. 51: "When we move to larger groups, or a whole society, the concept of group utility takes on value overtones, becoming an expression of social welfare more than just a device to explain or predict the group's actions. Social welfare is (or ought to be) the optimand for planners, legislators, and other policy makers ... ' The important move, however, is not from a small to a large scale, but from empirical theories that aim to explain decision, choice, and action to normative theories that aim to set standards for decision, choice, and action. 14.See Rawls 1972, sects. 15 and 49; and Rawls 1982. 15.ch. III sects. 4--7. 16.Parfit 1984, App. I. 17.ch. III sect. 7, ch. IV sect. 5. 18.ch. III sect. 7. 19.ch. I sect. 5. 10.Robbins 1938. 11.The proposal I have in mind is made, e.g., by John Harsanyi, Amartya Sen, Kenneth Arrow, and Donald Davidson. They all develop the proposal in different ways, but they have this common starting point. See Harsanyi 1976, ch. II; Harsanyi 1977, ch. 4; Sen 1973, pp. 14--15, 45; Arrow 1977; Arrow 1978; Davidson 1986. There is a similar proposal in Hare 1981, chs. 5 and 7. I have published a version of this chapter applied specifically to Hare's argument in a collection of essays about his recent work; see Griffin (forthcoming). 12.See Harsanyi 1976, pp. 50, 79--80; Harsanyi 1977, p. 59; Arrow 1977, p.220; Hare 1981, p. 128. 13.See esp. Harsanyi 1977, pp. 58--9. 14.Rawls 1982 develops a picture of how a well-ordered society regulated by an ordinal utilitarianism with comparability might use a deep utility function (Rawls calls it "a shared highest-order preference') in making interpersonal comparisions (sects.6 and 7). And he goes on to suggest that the conception of a citizen who will decide by appeal to a deep utility function "is psychologically intelligible only if one accepts, as Sidgwick did, a hedonist account of the good as a basis of an account of the rational judgment of individuals' (p. 181). 15.See ch. II sect. 4, ch. IV sects. 3 and 5. 16.Rawls thinks that resort to the deep utility function has some particularly unlikeable results. If we assume that everyone has a "shared highest-order preference', he says, then we no longer see persons as having any particularly strong commitments to more determinate forms of the good. We think of them as ready to entertain any new convictions or aims, to abandon any past attachments and loyalties, if greater well-being can be bought that way. This results, Rawls thinks, in the loss of "the 348 NOTES TO PAGES 114--117 distinctiveness of persons': persons have no conception of how to lead a life that is peculiarly theirs; they become "bare persons' (see Rawls 1982, esp. pp. 180--1). Rawls describes a monster (no individuality) who, not surprisingly, behaves odiously (sheds loyalties at the sight of a bigger pay-off). I doubt that the consequences are quite so horrific, but the issue is worth debating only if the most plausible form of desire account is likely to include a deep utility function, which it is not. 17.ch. IV sect. 3. I am indebted to Michael McDermott for helpful, wide-ranging discussions about comparability, especially about compar- isons using objective conceptions of well-being. 18.ch. IV sect. 5. 19.The desires relevant here have to be at least minimally informed; that is, one has to know enough about what one is getting to rule out one's ending up no better off or even worse off. But it does not have to be a fully informed desire, that is, one such that if one knew more one would not want anything different. Desires that are minimally informed will at least register on the positive side of the scale of prudential value. 20.R. M. Hare rightly stresses that to form preferences one does not need antecedent values, tastes, or attitudes. Tastes and attitudes change; we often form new preferences. Many preferences---and preferences central to prudence and morality---come about because we go out and gain understanding of the nature of the options before us with the aim of giving shape to our inchoate and unenlightened desires. However, none of this helps the model of the judger's preference. Even if I form an entirely new preference, it must still be an expression of my values or tastes or attitudes---not, of course, necessarily antecedent ones, possibly ones that only now, for the first time, come into existence. Still, they must be mine, and not everyone need have them. Without them, I should have no preference. Of course, I have---we all have---a general vacuous preference to have a more, rather than a less, valuable life. But this general preference just takes the question of what is going on in comparisons back one stage. See Hare 1981, pp. 125--6. 21.Rawls is right, therefore, that interpersonal comparisons of well-being need some sort of identity of preferences or values (see Rawls 1982, p. 173): you and I, in making interpersonal comparisons, will have to agree in the prudential values we apply in assessing the lives of the persons we compare. But this sort of agreement on values is nothing like the agreement represented by a deep utility function. It does not "destroy individuality' or create "bare persons' for us to have to agree about what makes life good, any more than would our having to agree on what makes an act morally right. Nor does it involve odious behaviour (shedding loyalties, abandoning attachments): to grow in one's understand- ing of what is prudentially valuable involves shedding something but it is scarcely odious to do so. No doubt people will in fact have different conceptions of what makes life prudentially valuable. And given the difficulty of making such judgments, they should; it would be high- 349 NOTES TO PAGES 117--124 handed to try artificially to homogenize conceptions. But the same is true of conceptions of moral right and wrong. This will mean that some disputes about interpersonal comparisons of well-being will be difficult to settle. But the informed desire account does at least give some framework, some criteria, to the dispute: judgments about prudential value have to be backed by defensible claims about what people, when informed, aim at. 22.This is Mill's view too: "If there be one to which all or almost all who have experience of both give a decided preference' (my italics). See J. S. Mill, Utilitarianism, ch. II para. 5. 23.This conception of preference is widespread: probably dominant in economics and still strong in philosophy. See, e.g., Layard and Walters 1978, p. 124: "Our basic theory assumes first that, for all the alternative consumption bundles he could conceivably face, the individual has a preference ordering. This reflects his tastes. . . . from the opportunities available to him he does the best he can, best being defined according to his tastes' (my italics). Also Arrow 1984, section entitled "Choice Under Static Conditions': "The utility theory of choice states that the choice in any given situation depends on the interaction of the externally given obstacles [i.e. income and prices] with the tastes of the individual . . . The utility theory asserts, more precisely, that the tastes can be represented by an ordering according to preference of all conceivable alternatives' (his italics). 24.See, e.g., Mackie 1981, p. 157: "... anything worth calling utilitarian- ism in an exact sense is committed to maximization based on calculations involving interpersonal comparisons ... such maximizing calculations may be morally objectionable in that they neglect questions of distributive justice or the rights of individuals, and in any case they are quite impracticable' (my italics). Or John Rawls 1982, p. 170: "... given the circumstances of justice in which citizens have conflicting conceptions of the good, there cannot be any practical agreement on how to compare happiness as defined, say, by success in carrying out plans of life, nor, even less, any practical agreement on how to evaluate the intrinsic value of these plans. Workable criteria for a public understanding of what is to count as advantageous in matters of justice, and hence as rendering some better situated than others in the relevant interpersonal comparisons, must, I believe, be founded on primary goods, or some similar notion.' See also Rawls 1972, sects. 15, 49. 25.See the discussion of whether the informed-desire account is "subjective' or "objective' in ch. II sect. 4. See also ch. III sect. 7. 26.e.g. Rawls 1972, sects. 16, 25, 49, 60; Rawls 1975a; and Rawls 1975b; also Scanlon 1975. 27.See ch. III sect. 5, and Part III passim. 28.See ch. III sect. 6. 350 NOTES TO PAGES 127--129 CHAPTER EIGHT 11.Not that the Requirement of Psychological Realism is easy to formulate. The rough idea that surfaces from time to time in philosophical discussion is this: moral principles must be such that our accepting them will, conjoined with normal human motivation, determine action---not neces- sarily unfailingly but at least often enough to give the principle a point. But what difference would it make if what we needed to get persons actually to live up to a moral principle went beyond "normal' human motivation? If we convincingly threatened someone with roasting alive if he did not live up to a certain principle, he would be likely to do it, as self-sacrificing as it might sometimes prove to be. Hell and Purgatory show that. But what if, instead, we merely whispered into his ear each night during his childhood, "Unless you live up to that principle, you'll roast alive'? If the whispers produced compliance, would the principle have met the Requirement? Does it matter to a moral principle that no one would follow it without fear (e.g. of ostracism) or pleasure (e.g. from praise)? The Requirement becomes too stiff if "normal human motivation' is taken to be "human motivation without threats or inducements from others'. But there are limits, both practical and moral, to how far fear, pleasure, and so on can be used to produce compliance. One important moral limit is autonomy: part of the dignity of life is choosing one's own path through it. Given those limits, there can be principles so demanding that compliance simply will not be forthcoming. So the Requirement must be something like this: moral principles must be such that one's believing or accepting them will, conjoined with threats and inducements that both morality and the facts of human psychology permit, determine action. One might, after all, autonomously subject oneself to threats and inducements in order to keep oneself on the straight and narrow. This formulation of the Requirement makes its application much more complicated and its use as a stick to beat certain moral theories much less easy than is usually thought. I am grateful to Philippa Foot for discussion of these issues. 12.For recent discussion of the reduction to self-interest, see Quinton 1973, pp. 5--10; Williams 1985, ch. 3; and Kavka 1985. 13.There is another problem with this way of making morality not alien. Is it enough just to place morality inside me (i.e. inside my self-interest)? The only plausible case for reducing morality to self-interest has to appeal to what happens in the long run. But why is it enough to get morality inside me? Why should I care about what happens to me much later? Is the difference between now and later in some respects like the difference between me and others? This line of thought was pursued in Sidgwick 1907, pp. 418--19, and much further and with great power in Parfit 1984, Pt. II. 14.For a good discussion, see Ullmann-Margalit 1977. 15.See ch. X sect. 2, ch. XII sect. 6, ch. XIII sects. 5--6. 351 NOTES TO PAGES 129--137 16.See ch. X sect. 2, ch. XIII sect. 5. 17.This is Bernard William's objection in Williams 1985, pp. 45--6. 18.This argument is developed in Williams 1985, ch. 3. 19.ch. IV sect. 1. 10.See ch. IV sect. 3. 11.See sect. 5 below. 12.Nozick 1981, pp. 401--2; see also my discussion earlier at ch. IV sect. 2. 13.See Hobbes 1969, sect. 25; D. Hume, A Treatise of Human Nature, bk. III Pt. I. For contemporary examples, see references to Williams and Foot in n. 22. 14.See his article "Internal and External Reasons' in Williams 1981. 15.Williams 1981, pp. 101--2. 16.See ch. I sect. 3. 17.See Williams 1981, pp. 103--4. 18.Williams 1981, pp. 103--5. 19.See ch. II sect. 3, ch. IV sect. 3. 20.I go into this sort of deliberation more fully in ch. IV sect.3. 21.I argue all this more fully in ch. II sect. 3. 22.Thus Bernard Williams argues that in the case of external reasons "... the new motivation [must] be in some way rationally arrived at, granted the earlier motivations. Yet at the same time, it must not bear to the earlier motivations the kind of rational relation which we considered in the earlier discussion---for in that case an internal reason statement would have to be true in the first place. I see no reason to suppose that these conditions could possibly be met.' See Williams 1981, p. 109. He contrasts the sorts of deliberation already discussed in connection with internal reasons with some form (not entirely clear) of "rational' deliberation. But the deepest, most fateful deliberation seems to fall between his two categories---more radical than the first sort but not as totally cut off from human desires, it would seem, as the second. There is the same sort of flaw, it seems to me, in Philippa Foot's similar argument. "I am, therefore, putting forward quite seriously a theory that disallows the possibility of saying that a man ought (full unsubscripted {"ought{') to have cares other than those he does have e.g. that the uncaring, amoral man ought to care about the relief of suffering or the protection of the weak. In my view we must start from the fact that some people do care about such things, and even devote their lives to them ... These things are necessary but only subjectively and conditionally necessary, as Kant would put it.' See her "Morality as a System of Hypothetical Imperatives', in Foot 1978, pp. 169--70. A moral person, she says, is one who cares "about others, and about causes such as liberty and justice' (p. 166). But this is just what cries out for explanation, and Foot leaves it unexplained. What should I care about? 352 NOTES TO PAGES 137--140 What is worth the sort of devotion and commitment that morality requires? Certainly not everything. Foot says later, a little more gener- ously, "I believe that a reason for acting must relate the action directly or indirectly to something that the agent wants or which is in his interest to have ... ' (p. 179). But what is that last clause, with its use of that idealized notion of interests, doing there? It opens up an enormous amount. Reasons for acting are not just a matter of a person's actual cares, concerns, aims, desires, and so on, but also of what, whether he realizes it or not, is in his interest. But then why can he not also see other considerations, not necessarily connected with his interests, as reasons? Once we ask, What should I care about?, and we see just how complex the relation betwen reasons and desires actually is, the distinction between "subjective' and "objective' gets dropped as a poor fit. 23.See ch. II sect. 4; ch. VII sect. 4. 24.Bernard Williams is a case in point; see Williams 1981, p. 109. 25.The rest of this paragraph recapitulates arguments from ch.II sect. 3; here, as there, a reader of John McDowell will hear echoes of some of his claims (see ch. II n. 19). 26.The authority problem that Bernard Willaims has with internal reasons spreads to what he says about "character' and "integrity' (see his "Persons, Character and Morality' and "Utilitarianism and Self- Indulgence' in Williams 1981 and see also Williams 1973a, sect.3, 5). We need, but he does not supply, some understanding of how we are to assess character. He often refers to "reflection' ( e.g. Williams 1985, ch. IX but see also the Index), but he says little, especially little in the way of concrete examples, about what it does. His view is simply this. Any fully developed person has "commitments' or "ground projects' that he "takes seriously at the deepest level, as what his life is about' (Williams 1973a, p. 116), projects which are "closely related to his existence and which to a significant degree give a meaning to his life' (Williams 1981, p. 12). He cannot abandon such projects just because the reports from the utility network indicate that, as chance or the actions of others have stacked things, utility would be maximized by his doing so. That would be to ask him to commit a kind of suicide, to abandon himself. It would be to attack "the value of integrity' (Williams 1973a, pp. 96--9). But what does Williams mean by "integrity' here? Whereas "integrity' as ordinarily used, to mean something like "honesty' or "general upright- ness', does name something valuable, "integrity' as Williams uses it, in its more literal etymological sense of "wholeness', does not. Persons often actually have as their "ground project', as what their life is about, morally hideous, or merely shabby or shallow, ambitions. To ask a person whose life is centred on resentment or revenge or vanity or one-upmanship to "abandon' himself may be exactly what he needs. Some persons most need, and all of us to no small degree would benefit from, some well chosen "disintegration' and "reintegration'. Even if everyone's values were 353 NOTES TO PAGES 140--142 perfectly respectable, there would be a more general difficulty. Williams needs to explain "integrity' in a way that makes it, at once, valuable in some non-question-begging way and yet incompatible with acting, utilit- arian fashion, as reports from the utility network demand. That is no easy job. There are, it is true, other interpretations of "wholeness' to fall back on. We might, for instance, take it to mean a person's being all of a piece---not hypocritical, insincere, self-deceived, and so on. "Integrity', so interpreted, would clearly be something worth protecting, but it would not be incompatible with utilitarianism. Or we might take "wholeness' to require commitment that involves the whole of a person; "wholeness' could then be seen as a kind of "wholeheartedness' such that a person who takes, say, truthfulness as one of his central values must see certain features of particular situations as morally dominant and certain others ---especially utilities---as morally minor. Here integrity is incompatible with utilitarianism but not valuable in a non-question-begging way; it is simply the rejection of the utilitarian approach to values. Williams' emphasis on character and integrity connects with his preference for "thick' ethical concepts, such as "loyal', "courageous', "generous', "grateful', over thin abstractions, such as "good', "right' (Williams 1985, esp. pp. 129, 140, 200). Character manifests itself in virtues and vices, and different characters, even different good ones, manifest different ones. Not only may persons properly bring different views of things to the job of sizing up a particular situation, but also the sizing up itself is best conducted not by looking for abstract good-making features but by sharpening our eye for all the detail that bears on assessment. But thick concepts have no small authority problems of their own. Which loyalties count? How much? 27.See Parfit 1984, Pt. II passim. 28.A good recent attempt, in the Kantian tradition, to derive substantive moral conclusions from formal features of rationality is Darwall 1983. Darwall parlays the impersonality of rationality into a modern contractu- alist conception of equal respect. So the success of his derivation turns partly on the acceptability of that contractualist conception. He finds in rationality a requirement that a rational principle be "self-supporting', i.e. that it would be rational according to it to act on it (see pp. 218--20, 228--30). Then by adding certain other features of formal rationality he goes on to derive two richer requirements on a rational principle, namely, in order of increasing richness, (1) that it would be rational according to it to choose all agents to act on it when this choice is made from an impartial standpoint (a move to a form of Kantian universalization), (2) that it would be rational to choose that all act on it, were that choice made from a perspective in which one is motivated by a concern for one's ability rationally to pursue ends, but is denied any information about oneself in particular (a move to something very like Rawls' Original Position). I discuss the requirement that a principle be self- supporting at the end of ch. X sect. 2. I discuss the acceptability of the contractualist's conception of equal respect in ch. IX sects. 3 and 354 NOTES TO PAGES 143--145 4 and have more to say about Kant's Categorical Imperative test in ch. X sect. 4. 29.See Hare 1952, ch. 11, Hare 1963, ch. 4. 30.Hare is well aware that his view is open to the charge of linguistic conservatism. The view in question is that from the logic of key moral terms we can derive the universality and prescriptivity of moral judg- ments and from universal prescriptivity we can derive a form of impartiality, broadly utilitarian, that will determine which principles and dispositions should guide action. But what if we chose to adopt different moral terms, or change our present moral terms in just the way that would prevent the derivation of universal prescriptivity? To the charge that he is making us the captive of our present conceptual scheme, Hare replies (Hare 1981, ch. 1.5) that he would be if he concentrated on "secondarily evaluative words', but he concentrates instead on words like "must' and "ought', which do not encapsulate any particular moral commitment. But the real worry, I think, is different from the one he here meets, namely that even "ought' and "must' encapsulate particular commitments as to what morality, and moral reasoning, is. Hare's answer to this deeper worry is, I believe, that if we alter the meanings of the most basic evaluative words, we shall alter the questions we ask, and that if we want to answer those questions then we are stuck with those concepts. But the worry lingers. We might, in the course of building a moral theory, revise certain features of those key terms, so that we should no longer be asking exactly the same questions as at the start, nor entirely new ones either. Could there be, initially, any ground for saying that such partial revision will not, or should not, take place? Pressures to change concepts usually mount only in the course of theory-building. So can one do more than build one's theory always sensitive to the need for conceptual revision? This, of course, allows Hare a further response, one to which I am sympathetic. He can say that when the theory is complete one finds at the end no need to revise the universal prescriptivity of "ought' and "must'. But this can be claimed only at the close of theory-building, so it cannot be the source of the theory. There is also the more familiar, much discussed problem for Hare's view: Can we derive from the logic of the key terms more than a pretty indeterminate form of universalizability and prescriptivity, nothing deter- minate enough to yield substantive moral principles? But I think that there is this prior problem, too. For further discussion of Hare's views, see J. L. Mackie's objections in "Rights, Utility and Universalization' and Hare's response in "Reply to J. L. Mackie' both in Frey 1984. 31.See ch. II sect. 5. 32.This is a question that Derek Parfit asks in the course of his discussion of personal identity; see Parfit 1984, ch. 15. He suggests that his reductionist view of "person' points towards saying the first. In what follows I give considerations that point towards saying the second. But it 355 NOTES TO PAGE 146 is not clear that these views are contrary. There is an ambiguity in the expression "the whole of one person's life'. Parts of a life, such as experiences that we value, can be contrasted to a whole life in different ways. "A whole life' can be interpreted temporally (as Parfit does): one's life from start to finish (not a period in it). Or it can be interpreted (as I do): everything that now, or over some period, constitues one's life (not an aspect of it). 33.It is not that there are no reasons to hold back from a policy of maximization in the case of one person. I shall mention three that have considerable weight. First, might it not be better to follow a policy, not of maximizing one's own well-being, but ensuring that one does not have any period that is very bad, even if the total would thereby be less? I have already given my answer in ch. II sect. 5. A second reason to hold back is what might be called a holistic approach to evaluation, namely, the view that the value of the whole may not always be the value of the parts. (I owe this point and the following example to Paul Seabright.) For example, suppose a person who has had seventy years of very good life has an accident, and imagine each of the alternative results. In one, he is killed outright. In the other, he lives on, diminished and in some pain, for another ten years, but years which on their own would be worth having. Now if maximizing means adding, then there is no question but that the second outcome is better: seventy good years plus ten years of positive value. Yet if evaluation is sometimes holistic there is a question. Perhaps the life as a whole would be better if he is killed outright; perhaps the person would rationally prefer it. The holistic evaluation should not really make us hold back. The example counts against maximizing in the sense of aiming at the greatest sum of the value of the parts; it is really an objection to a rather crude aggregative approach to evaluation. But it does not count against maximizing in the sense of making one's life as valuable as possible. On the contrary, it uses that conception of the maximum in favouring outright death. Even as an objection to an aggregative approach, it is unconvincing. How could outright death be the rational choice? Ex hypothesi, the seventy year old thinks both that the extra ten years would be worth having but that, added to the seventy very good years, they would make the life as a whole worse. One can see the value categories that might be appealed to: a person may value living, and dying, with style; he may treat his whole life as a kind of aesthetic object. But that we have these value categories does not mean that they can coherently be used here. Although not everyone agrees, beauty, for instance, has a value only by being appreciated and so adding to the value of persons' lives; it is implausible that in this example what is added to persons' lives justifies outright death. A third possible reason for not maximizing is the powerful one that it is simply beyond our powers. Perhaps the most that we can reasonably 356 NOTES TO PAGE 146 hope to do is what economists call "satisficing' (see Herbert Simon's essays collected in Simon 1982, esp. "A Behavioral Model of Rationality'). The maximizing strategy might be stated: (1) identify all options, (2) evaluate them, (3) choose the best. But one can seldom identify all the options. Even a simple choice between keeping or breaking a promise does not, for a maximizer, consist in only two options: one can break a promise in practically innumerable ways. And one cannot decide at a certain point that further work identifying the options would cost more than it would be worth; to decide that would require investigation, which itself might cost more than it would be worth, though knowing that would itself require investigation, and so on. There is often no such thing as the "best' outcome; in many situations that is a concept without an application. One can identify a "best', or at least "equal best', option if the set of options is fairly small, but real life seldom provides such sets. Perhaps in real life the rational strategy is satisficing: (1) fix an aspiration level, (2) start enumerating the options, (3) evaluate them as one goes along, (4) accept the first to be at or above the aspiration level. Take this case. I follow the satisficing strategy and come upon option A, which meets my aspiration level. But soon it becomes clear to me that option B is better (I might have learned this without any work on my part). It would now be irrational for me not to choose B. I have a reason to choose B (it is better) and no reason not to (it costs me nothing). Satisficing is not a rational strategy in the strong sense that maximizing appears to be: it would often be plainly irrational. Anyway, how do I fix my aspiration level? For instance how do I fix an asking price for my house? My aspiration level would itself be irrational if I had a very good chance of getting something much better without much cost to me. So the only rational way to fix an aspiration level involves doing, as best one can, the complex trade-off between further costs, further benefits, and the probabilities involved. What seems to be at fault is the particular characterization of the maximizing strategy given just above. On that characterization, maximizing requires enumeration of all the options, and the impossibility of doing that undermines the whole strategy. But instead of reacting to this by going over to a satisficing strategy, one could look for a better characterization of the maximizing one. For example: (1) identify the options, stopping when one guesses that the cost of further identification will exceed the benefit (the same sort of calculation that one needs to fix the aspiration level), and then proceeding as before. For a good discussion of these issues, see Pettit 1984. There are many further arguments for satificing, centring on the consequence that the use of a rational strategy has for an agent's character or temperament. Spontaneity is valuable; satisficing gives it some space, while maximizing gives it virtually none (see Elster 1982, ch. 2). Satisficing fosters a valuable moderation, while maximizing encourages discontent, obsessive perfectionism, sometimes even greed (see Pettit 1984). These, and several other cases, need careful consideration. Let me just state my belief that they do not make a strong case for satisficing. For 357 NOTES TO PAGES 146--151 instance, there are good reasons to be content with being merely fairly well off, without trying to squeeze out of life every last bit of comfort or security; that over-ambitious strategy would often involve pursuing one value at the expense of a second, greater one that relentless striving for the first excludes. These arguments for satisificing often trade on an overly narrow conception of the values that can enter a maximizing strategy. The issues that arise here connect with important issues about the self-defeating quality of certain maximizing strategies and about the difference between local and global consequentialist perspectives. What would maximizing deliberation be like in practice? Might maximization, even if it were accepted as the rational strategy, have only a limited role in our actual decision procedures? I come to these issues in ch. X sect. 2. 34.e.g. in Suppes 1967. 35.See John Rawls' illuminating account of this variety of decision theory in Rawls 1972, pp. 552 ff.; see also pp. 441 ff. 36.I have argued this earlier at ch. II sect. 4. 37.e.g. A. Sen and B. Williams in their introduction to Sen and Williams 1982, pp. 16--18. 38.J. S. Mill, Utilitarianism, ch. IV, esp. paras.3 and 9. The sentence I use to express Mill's conclusion comes from para. 3, where he has not yet argued that only happiness is a good. But when, in para. 9, he draws his final conclusion, it is stated too compactly to show what he has in mind, namely what he says in para. 3 strengthened with only: only general happiness is a good to the aggregate. 39.Mill 1972, p. 1414. 40.Utilitarianism, ch. V, n. to para. 36. 41.Also the n. to ch. V, para. 36. 42.ch. IV, paras.3 and 9. 43.Economic theory, which gets its framework from utilitarianism, may suffer from a tendency to the same conflation. The social welfare function traverses the same ground: individual good, to uncontentious forms of the common good, to contentious forms, and finally to standards for action. So it incorporates the whole of moral philosophy. The risks of conflation are even greater in economics; the term "social welfare function' often includes voting procedures (e.g. majority rule), on the ground that they are devices for determining society's choices. But voting procedures often have purely practical justifications: we need a solution (better rather than worse, but sometimes any solution rather than continued conflict); solutions of conflicts of interest should promote social stability, etc. So the best voting procedure may not produce an individual's best or a group's best or the morally best. It is yet another standard. 44.See F. Hutcheson, An Inquiry Concerning Moral Good and Evil (1725), sect. III, para. 8: "that action is best, which procures the greatest happiness for the greatest numbers'. C. Helve$0tius, De l'esprit (1758), 358 NOTES TO PAGE 151 discours II, ch. XXIII: "Les principles d'une bonne morale ... doivent toujours e$2tre appuye$0s ... sur ... l'inte$0re$2t public, c'est a$1 dire celui du plus grand nombre.' C. Beccaria, Dei Delitti e delle Pene (1754, English trans. 1770): "la massima felicita$1 divisa nel maggior numero'. J. Priestley, An Essay on the First Principles of Government (1758), p. 17, Part I, of liberty: "The good and happiness of the members, that is of the majority of the members of any state is the great standard'. For a thorough discussion of the history of the formula up to Bentham, see Baumgardt 1952, pp. 35--59. 45.See his A Fragment on Government (1776), Preface, sect. 2 and passim. On Bentham's dropping the formula in favour of simply "the greatest happiness', see Baumgardt 1952, p. 505. See also Bentham's note of 1822 to his use of "the principle of utility' in An Introduction to the Principles of Morals and Legislation (1789), ch. I, sect. 1; the gloss in the footnote also supports a simple "the greatest happiness' reading. 46.Mill speaks of "The Greatest Happiness Principle', which could mean "the greatest happiness of the greatest number', but it need not; it could also mean simply "the greatest total happiness' without regard to number. "The greatest happiness of the greatest number' was used twice in Sidgwick 1907, pp. 420 and 499, as one version of the principle of utility, but this is only a lapse; it is clear (see pp. 415--16) that Sidgwick understands the principle of utility as requiring simply the greatest total utility possible. 47.e.g. Baier 1958, p. 192; Britton 1959--60; Popper 1966, vol. II p.285; Rescher 1966, esp. pp. 8 and 25 but passim; Narveson 1967, p. 126; Franklin 1968, p. 162; Wolff 1968, p. 7; Rawls 1972, p. 22 n.9; Williams 1973c, p. 96; Dworkin 1977 p. 23; article on "Utilitarianism' in Flew 1979; MacIntyre 1981, p. 62. 48.This point is hardly original. The first to make it was a nineteenth- century economist (Edgeworth 1881, pp. 117--18); more recently much the same point has been made by game theorists (von Neumann and Morgenstern 1953, p. 11) and general linguists (Jespersen 1924, p. 246). But the point should not be overdone. It does not show, for instance, as Peter Geach thinks it does, how to "shoot down' utilitarianism, because the fault is entirely removed simply by shifting to the formula "the greatest happiness', which most utilitarians use anyway. See Geach 1977, pp. 91--4. 49.Von Neumann and Morgenstern 1953, p. 11. 50.Of course, double dependent superlatives can be coherent so long as one or both are hyperbolic. In a letter to The Times some years ago Mr James Raimes wrote in support of multiple forms of art (such as lithography) as against rare forms (such as Old Master paintings), and he ended with a call for "the very best art for the most people at the lowest price possible'. But it is likely that Mr Raimes did not mean any of his three superlatives strictly; what he wants is good art at modest prices for large numbers of people to enjoy. And no doubt what he meant by "good 359 NOTES TO PAGES 152--159 art' would cover examples that might still differ a bit between themselves in degree of goodness. The trouble comes in formulae in which the dependent functions are to be strictly maximized: every degree of variation is meant to count. So these various ways of avoiding the trouble do not help such formulae as "the greatest number of radios assembled in the shortest possible time'. The Times, 9 March 1968. 51.I am sure that many persons will be unconverted by this argument. They will still think that the formula is all right and that our job is just to find words to express what we have always really had in mind in using it. Perhaps one of the superlatives is eliminable, or not a strict maximum. Well, the likely candidate for elimination is "the greatest number', but that just leaves the formula "the greatest happiness', and if that is what one means, one should say it. Or suppose that it is not quite so simple, that the superlatives are lexicographically ordered. Well, the principle might then go: first maximize happiness but, if there are ties, maximize number. Maximization of number would have a modest role to play, although it might get some appeal from seeming to introduce considera- tions of justice, and I shall discuss it further from this point of view in a moment. Or suppose that both maxima are eliminable and that what we really have in mind is a rate: the act with the highest utility is, we should then say, the one which produces the greatest amount of happiness per person affected by the act. But the trouble with all these suggestions is that, although they give coherent standards, they are not plausible as interpretations of what the original formula means. And if these new standards are meant, then they should be stated, and the original formula simply scrapped. 52.Von Neumann and Morgenstern 1953, pp. 10 ff. 53.For an explanation of the Prisoner's Dilemma, see ch. X n. 1. 54.See Frankena 1963, p. 34; if we "understand the principle of utility as enjoining us to promote the greatest good of the greatest number', it "thus becomes a double principle ... it has become a combination of the principle of utility with a principle of justice'. See also Rescher 1966, p. 25: "the principle of utility is a two-factor criterion ({"greatest good{', {"greatest number{'), and ... these two factors can in given cases work against one another'. 55.For instance, we could produce such a distribution by upsetting an equal distribution where everyone is well off, even thereby lowering the total, so long as we benefit more persons by the change than we harm. The case against finding in "the greatest number' an adequate account of justice is well, and much more fully, argued in Rescher 1966, pp. 25--8, from which I draw my points here. 56.Rawls 1972, p. 27; see also Nozick 1974, pp. 32--3. 57.e.g. Bernard Williams and Philippa Foot; see the references in n.22. 58.ch. II sect. 5. 59.Pt. Two, esp. ch. V. 360 NOTES TO PAGES 159--165 60.chs. IX and XI. 61.Sidgwick 1907, pp. 404, 506--9; the second passage cited is the close of the book, where Sidgwick unflinchingly faces the possibility of a funda- mental contradiction in our apparent intuitions of what is reasonable in conduct. 62.See ch. II sect. 5 and ch. V. CHAPTER NINE 11. Those familiar with the writings of Ronald Dworkin will recognize these expressions. He uses "equal concern and respect' to mark a "funda- mental and axiomatic' right from which not only more particular rights but also the general authority of collective goals are to be derived (see Dworkin 1977, esp. pp. xiv--xv but also pp. 180--3, 272--8). I take his expression and split it into two parts of very different significance. I think that what he means by "equal concern and respect', at least judging by its fundamental place in his argu- ment, is what I mean here by "equal respect'. One of the points that I want to argue throughout this Part is that there is no single moral "principle of equality', that equality of different things matters at different points in moral theory, and that we have to keep them straight. 12.See ch. X sect. 3. 13.ch. VIII sect. 2, but see also the discussion of ethical pull in ch. VIII sect. 5. 14.Some utilitarians (e.g. Mill) have tried to ground obligations in our desires or ends. But the failure of this attempt is not also a failure of utilitarianism. It is a mistake to think that because utilitarianism is essentially teleological in one way (viz. in maintaining that right and wrong are determined by how possible actions promote our ends) it has to be teleological in every way (e.g. in maintaining that what makes this, or anything, an obligation is that it does promote human ends). It is a mistake to see utilitarianism and Kantianism as systematically opposed. It is possible to produce a powerful combination of a substan- tially utilitarian theory with a largely Kantian conception of obligation, as R. M. Hare's work clearly shows. My own account of obligation is less close to Kant's than Hare's is. 15.As do, e.g., G. J. Warnock in Warnock 1971 and J. L. Mackie in Mackie 1977, ch. 5. Both Warnock and Mackie limit their claims to "narrow morality' ("a system of a particular sort of constraints on conduct---ones whose central task is to protect the interests of persons other than the agent and which present themselves to an agent as checks on his natural inclinations or spontaneous tendencies to act') in contrast to "broad morality' ("a general all-inclusive theory of conduct: ... whatever body of principles [someone] allowed ultimately to guide or 361 NOTES TO PAGES 165--171 determine his choices of action'); see Mackie 1977 pp. 106--7. My objections will also come from "narrow morality'. 16.Warnock 1971, pp. 19--26. 17.Mackie 1977, p. 111. 18.Mackie 1977, p. 116. 19.Warnock 1971, p. 76; Mackie 1977, p. 114. 10.For Rawls' criticism of the utilitarian conception of impartiality, the Ideal Observer, as an undesirable impersonality, see Rawls 1972, pp. 188, 190; for his argument that the Ideal Contractor conception provides the superior interpretation of impartiality, see pp. 189--90. 11.e.g. in John Mackie's unpublished lectures, Justice and Rights; see also R. M. Hare's discussion in Hare 1975 passim but esp. pp. 89--95, 101--7. 12.e.g. in Diggs 1982, pp. 112--14. 13.Utilitarianism, ch. V, n. to para. 36. 14.See, e.g., Rawls 1972, pp. 36, 44 ("the aggregative-distributive dicho- tomy') and Rawls 1974--5, p.19 (utilitarianism "puts no value on the distribution of good'; it "gives no weight to distribution ... '). See also Sen 1973, p. 23 ("utilitarianism ... is much too hooked on the welfare sum to be concerned with the problem of distribution ... '). See also Williams 1973, p. 142 (an implication of utilitarianism is that "questions of equitable and inequitable distribution do not matter'). 15.For the first two remarks see Mackie 1978. For the third see Rawls 1972, pp. 27, 187, and Nozick 1974, pp. 32--3. 16.For an example of someone who takes "separateness' as a reason for adopting a particular view about impartiality, see Nozick 1974, pp. 32--3. 17.See Rawls 1972, p. 255 (where the Original Position is equated with "the point of view of noumenal selves'); Rawls 1980, pp. 549--50. For contrast of Ideal Observer and Ideal Contractor, see Rawls 1972, pp. 188--90. 18.Rawls 1972, see p. 60 for their preliminary statement and pp. 302--3 for their final and full statement. 19.Rawls 1972, p. 3. 20.Rawls 1972, p. 4. 21."Thousands Who Need Not Die Each Year', Sunday Times, 27 March 1983. 22.Rawls 1978, p. 47. On the limits in scope of the two principles of justice, see p. 49: "The first principles of justice as fairness are plainly not suitable for a general theory'; see also his n. 4. 23.Rawls 1972, p. 88; see in general sect. 14. 24.Rawls 1972, p. 83. Rawls also limits his attention to a society composed of healthy people, on the ground that the special needs of the ill and the handicapped need special treatment and are likely to distort our 362 NOTES TO PAGES 171--173 judgment about the normal. See Rawls 1975a, p. 96. But this exclusion does not help with the examples either. For further discussion of this exclusion, see ch. XI n. 23. 25.Rawls 1972, p. 4. 26.Rawls 1972, p. 218. 27.Rawls 1972, p. 248. 28.Rawls 1972, p. 277. 29.Rawls 1972, p. 178. 30.Rawls 1972, sect. 31. 31.Rawls 1972, p. 449. 32.Rawls 1978, p. 48. 33.Rawls 1972, p. 573. 34.There is another way that Rawls' contractualism is not rich enough. I have been speaking of Rawls' treatment of justice as fairness, but there are comparable troubles for his account of rightness as fairness. A good example to consider is that of promising (see Rawls 1972, pp. 344 ff.). For example, should long-term promises be binding, even if entered into in good faith, and even if the circumstances have not changed? Lady Diana Cooper reports: "She [Lady Curzon] told me an amazingly characteristic fact about George [Lord Curzon]. On marriage he made her sign a pledge that, in case of his death, she would never remarry.' (Quoted in Zeigler 1983, p. 118.) But suppose, years later, with her husband dead, Lady Curzon wanted to remarry and deeply regretted that promise of her distant, youthful self. Is she obliged to keep it? How would appeal to contractualism settle the matter? There are at least two ways one can see the institution of promising: one where all promises (so long as they were made by a person capable of understanding their terms, not repudiated by the promisee, etc.) are binding; another where all such promises are binding so long as they are not very long-term. Both institutions are compatible with Rawls' two principles of justice. How are we to choose between them? The Original Position tells us nothing, or nothing relevant. What one must appeal to is one's conception of what would be right. And this conception is not, therefore, rightness as fairness, because it is detached from the whole contractarian apparatus. 35.Rawls 1972, pp. 102--3. 36.Rawls 1972, pp. 141--2. 37.Rawls 1972, pp. 119--20; see also Rawls 1980, p. 536, where "best deal' considerations are said not to be the only ones but are thereby acknowledged to be important ones. 38.Rawls 1972, p. 103. 39.Rawls 1972, p. 176--7. 40.Rawls 1972, p. 177. 363 NOTES TO PAGES 173--185 41.Rawls 1980, p. 560. 42.That formulation is derived from Scanlon 1982, p. 110. Scanlon sees contractualism not as an expression of the moral point of view but as that very closely connected thing, an account of what kind of judgment moral ones are. So I here alter his focus. Still, to give an account of the moral point of view is to give an indication of what morality is and so adopt a view about the nature of moral judgments. They are different matters, but close. 43.Scanlon 1982, p. 111. 44.Scanlon 1982, p. 112. 45.See B. J. Diggs' discussion of the compromise element in his own contractualism in Diggs 1982, pp. 103--4. 46.Gauthier 1982a; see also Gauthier 1977 and Gauthier 1982b. 47.Gauthier 1982a, pp. 160--1; see also Gauthier 1985. 48.Diggs 1981. Although Diggs' version of the "moral judge' can be developed with the capacious conception of "reason' that I use in the text, it seems likely that Diggs himself would keep it tighter. The conception of the moral judge is derived from what Diggs calls "the Imperative of a moral social morality', namely, "Join others wherever possible, (1) in acting in ways that each person together with others can reasonably and freely subscribe to as a common morality and (2) in treating each person in ways consistent with the person's developing and freely exercising his capacity as a rational being to govern himself' (p. 276). The judge acts in the spirit of the Imperative, so he seeks a decision that all parties can reasonably subscribe to (p. 281). The judge's decisions, as Diggs explains rationality (pp. 277--8), will therefore probably be like the decisions of Gauthier's arbitrator. See also Diggs 1982. 49.See Sen 1973, pp. 15--23. Sen subsequently presented a second, still weaker version of the Weak Equity Axiom in Sen 1975, p. 285. He eventually rejected the Axiom as too strong in Sen 1980. 50.I discuss this further in Griffin 1981. 51.See ch. X sect. 3. 52.See ch. V sect. 6. 53.Rawls 1972, pp. 141--2. The distortion to the moral point of view that enters with unanimity also enters with certain conceptions of equality, namely those that see equality as requiring that we include each person's point of view separately and thus that we find the assessment that is least unacceptable to the person to whom it is most unacceptable. For an example, see Thomas Nagel's paper, "Equality', in Nagel 1979, esp. p. 123. 54.As, e.g., John Harsanyi does; see Harsanyi 1976, Part A, but esp. pp. 4 and 14. 55.This is Rawls' vision; see the subtle discussion in Rawls 1972, sect. 81. 364 NOTES TO PAGES 185--191 56.See Mackie 1977, ch. 6 sect. 2. 57.Captain Oates was a member of Scott's polar expedition of 1911--12. Finding himself weakened beyond hope of survival, he walked out into the cold to die, with the words "I am just going outside and may be some time.' 58.For a rich development of this line of thought, see Nagel 1980, esp. p. 90. I have borrowed (immediately below) the example of wanting to be an accomplished pianist from him too; see p. 122. 59.ch. VII sect. 4. 60.For discussions of the analogy with secondary qualities, sympathetic and unsympathetic respectively, see McDowell 1985 and Williams 1985, ch. VIII, esp. pp. 149--52. 61.Nagel uses the analogy with secondary qualities to bolster up this distinction, which I think it does not do. See Nagel 1980, pp. 80--4. 62.There is a good survey of the difficulty that major moral theories have in accommodating the distinction between obligation and supererogation in Heyd 1982, Pt. I. 63.Two recent attempts to find relatively short answers of both these kinds are Williams' argument about "integrity', which he uses to support not only a strong personal moral perspective but also something approaching a deontologist's conception of responsibility (see his remarks about "negative responsibility' in Williams 1973a, sects. 3 and 5; but see also "Persons, Character and Morality' and "Utilitarianism and Self- Indulgence' in Williams 1981 and my earlier discussion at ch. VIII sect. 3, esp. n. 26); and Nagel's treatment of the "obscure topic of deontological constraints' (see Nagel 1980, pp. 126--35). 64.There are, no doubt, many ways of bringing out the force of deontological reasons. See for example Nagel's suggestive remarks about the centrality of intention in the deontological picture; Nagel 1980, pp. 131--3. He does not deny moral role to outcomes, however caused. But he thinks that sometimes, on moral grounds, we will not intentionally do (e.g. torture) what would have a better outcome, and that this shows that deontological constraints also have moral weight. He then tells us where "the strength of the deontological view lies'. He asks, "what is the essence of aiming, what differentiates it from merely producing a result know- ingly'? The difference, he answers, is that action intentionally aimed at a goal is guided by that goal; hence, action aimed at evil, even that good may come, is guided by evil. But the essence of evil, he says, is that it should repel us: "That is what evil means'. So to aim at evil is to swim "head on against the normative current'. But this does nothing to justify deontological constraints. The metaphor of swimming head on against the normative current does not even fit; we do evil that good may come, so there is a surface current in one direction and an undertow in the opposite. But the confusing normative eddies can be explained in many different ways (for example, one does not need the deontologist's 365 NOTES TO PAGES 191--192 perspective to account for the repulsion that a decent person would feel at torturing), and one would not be tempted by Nagel's explana- tion, unless one already believed deontological constraints to have independent weight. Indeed Nagel may well accept this; he may see his remarks more as an exposition of than as an argument for deonto- logy. 65.See ch. X sect. 5, ch. XIII sects. 2, 6. CHAPTER TEN 11. The story is this: each of two prisoners finds himself in a situation where he correctly reasons that if: Self interest therefore counsels him to confess. Confessing is better for him whatever the other prisoner does. If the other confesses, it is better for him to confess; if the other stays silent, it is better for him to confess. Now the other will reason similarly. So they will both confess, thereby getting ten years each, while if they had both stayed silent they would have got only two. Despite the contrived nature of this example, dilemmas of this form are important in real life. One feature of the prisoners' situation that makes the grip of their dilemma particularly tight is that nothing affects the con- sequences except what they do in this situation. In real life, of course, what a person does in one case can affect his reputation or set an example, and thereby affects how things will go for him in the future. So, if we treat lack of influence on the future as essential to a Prisoner's Dilemma, then there are very few two party Prisoner's Dilemmas in real life. But if we take a less purist line and do not treat it as essential, then there are many of them. One important 366 NOTES TO PAGE 193 case is nuclear arms' control. Every time a new round of escalation is possible, the United States and the Soviet Union would each have its best result if the other did not produce the new arms while it did, its worst result if it did not while the other did, its second best if neither did, and its third best if both did. But each is better off producing the new arms whatever the other does. So each produces them, thereby getting its third best result. And once we move to many-party cases, especially situations where numbers are very large, as they are in modern societies, then the reputation one gains or the example one sets often plays no role at all, and unfortunately pure Prisoner's Dilemmas abound. There is the following depressing one. Each of us is best off himself if others help those in distress while he does not, worst off if he does while others do not, second best if all do, and third best if none do. But each is better off not helping, whatever the others do. There are good discussions of both the formal features of Prisoner's Dilemmas and their possible real-life instances in Parfit 1984, ch. II and Axelrod 1984, ch. I. The Prisoner's Dilemma is an abstract specification of one kind of situation in which the pattern of pay-offs is determined by the joint actions of the participants. There are other kinds, many of which also have important real-life instances. The whole range of situations is studied in the Theory of Games (see e.g. von Neumann and Morgenstern 1953, Luce and Raiffa 1957, and Brams 1976), and it is a test of a moral or political theory that it can come up with good advice in all of these situations. 12.It is a contentious matter whether any kind of utilitarian reasoning, even the kind that I imagine here in which the agent ignores the effects of what groups do and concentrates on his own action in isolation, would lead to the conclusion that I ought to burn logs. Mancur Olson has argued, as I do here, that the conclusion "holds true whether behaviour is selfish or unselfish, so long as it is strictly speaking {"rational{'. Even if the member of a large group were to neglect his own interests entirely, he still would not rationally contrib- ute towards the provision of any collective or public good, since his own contribution would not be perceptible.' (Olson 1965, p. 64) To which Brian Barry has replied, "This is surely absurd. If each contribu- tion is literally {"imperceptible{' how can all the contributions together add up to anything?' (Barry 1978, p. 32.) The most directly pertinent discussion that I know of is in Regan 1980 (but see also the closely related discussion in Parfit 1984, sects. 28--9). Regan thinks that a person who argues as I do is committed to "logically inconsistent assumptions. On the one hand, he is committed to the assumption that a single crossing never makes a difference to the state of the grass. From this it follows that there is no difference (as far as the state of the grass is concerned) between the consequences of no one's crossing and one person's crossing ... [or] between the consequences of one person's crossing and two persons' crossing ... and 367 NOTES TO PAGE 193 so on. Since the relation of {"there being no difference between the consequences of ... {' is transitive, we can conclude that there is no difference (as far as the grass is concerned) between the consequences of no one's crossing and the consequences of everyone's crossing. But this is inconsistent with another premiss of the argument, the premiss that the overall consequences of everyone's crossing are much worse than the overall consequences of everyone's walking around' (pp. 59-- 60). But the relation "there being no perceptible difference between the consequences of ... ' is non-transitive. And the value at stake with lawns and buildings is aesthetic; perception is necessary for change in the sort of value that here concerns a utilitarian. Besides, not every single act (crossing the lawn, burning a log fire) has to make, over time, an actual difference. It may be that four or five crossings/fires have an effect that the lawn/buildings will entirely recover from (the grass will revive in a few hours; the next rain will wash a certain amount of deposit off the buildings) and that we shall get back to the status quo ante. But a hundred crossings/fires, let us say, will do permanent perceptible damage---though, since the grass will revive a bit and the buildings have some of the deposit washed off by the next rain, the damage might be done by only ninety-five of the crossings/fires. And from the fact that the damage of n crossings is indistinguishable from the damage of n + 1, and so on, it does not follow that the damage from n crossings is indistinguishable from the damage from n + 50. A series of imperceptible changes can grow into a perceptible change, and so a change in value, even when any individual one is reversed in short order. Regan is well aware that the important relation might be thought to be "there being no perceptible difference ... ' But of this he says that there must be some point at which a single crossing either produces a recognizable change in the grass or at least changes the likelihood that the persons around will enjoy the grass less. "If this were not so, then the whole string of changes together could make no difference' (p. 61). But this is just to assert that there could not be a series of minute changes each of which, singly, made no perceptible difference but enough of which, in aggregate, did. And there do seem to be cases. I might do something (a hair's-breadth turn of the screw) that produced such a minute change in your nerve endings that you noticed no change, but were there a hundred such changes you would be in noticeable pain. There is nothing in this claim to give rise to Wang's paradox: it is not a matter of saying that the addition of a lot of "no differences' produces "some difference'. Each succeeding state differs from the previous one in minute physiological structures, and enough such physiological changes rise to consciousness. Furthermore, the eventual rise to consciousness is possible even when the move from one physiological state to its immediate successor has no effect not only on consciousness but also on the probability of 368 NOTES TO PAGES 194--201 being in any particular state of consciousness. One does not have to resort to Regan's differences in probability to avoid paradox; the physiological story can do that without them. If utilitarianism can avoid the embarrassment of the problem of moral outlook, it cannot be because there will always be some minute but genuine disutility somewhere along the line. Sometimes there is, and sometimes not. 13.For an example of a contractualist's derivation of a rule against cruelty, see Richards 1971, ch. 10 sect. I; for Richards' strained discussion of cruelty to animals, which never once mentions their pain, see pp. 182--3. 14.See discussion in ch. IX n. 34. 15.An intriguing historical question is how act utilitarianism ever became the dominant interpretation. Bentham, Mill, and Sidgwick were pretty certainly not act utilitarians, on this tight interpretation. When did this interpretation edge the others out? Was it with G. E. Moore? See, e.g., Moore 1966, p. 121: "... it must always be the duty of every agent to do that one, among all actions which he can do on any given occasion, whose total consequences will have the greatest intrinsic value.' For recent examples of what looks like the tight interpretation, usually (not surprisingly) by authors who do not find much to say for it, see Hodgson 1967, p. 1; Williams 1973a, p. 128; Brandt 1979, ch. XIV, sect. 2 and ch. XV sect. 3; Regan 1980, p. 12; etc. 16.For excellent discussions, see Adams 1976 and Parfit 1984, ch. I sect. 6. 17.See the discussion in Diggs 1982, pp. 103--4. 18.See Bales 1971. 19.ch. VIII sects. 2 and 4. 10.The main sources are Hare 1976a and Hare 1981, ch. 2 sect. 1 and ch. 3. 11.The chief difference between Hare's proposal and mine, besides the one I now go on to talk about, is that I suggest not only two decision procedures but also separating decision procedures from the criteria of right and wrong (I mentioned this separation earlier and return to it briefly in a moment). So I have, in a sense, three levels. And this may still not be enough. The two decision procedures that I have in mind are for individuals. Perhaps a political decision proced- ure (a procedure for governments) would be different again. Perhaps the practical decision procedure of an intelligent, scrupulous individual need not have restrictions quite as tough as the decision procedure that we should want for government agencies. Hare might say that a political decision procedure can be accommodated on the intuitive level without extensive rebuilding. So it can: "acceptance utility' can be used of "acceptance by an individual', "acceptance by persons generally in society' (two interpretations which Hare gives it in Hare 369 NOTES TO PAGES 201--206 1981), but also of "acceptance by members of the judiciary/by the police/by all agents of the government/etc.'. But this gives scope for considerable complexity. Still, what seems to me entirely right is Hare's insistence on a multi-level structure in moral thought. For more on this see ch. XI sect. 9. 12.cf. Aristotle, Nicomachean Ethics, II.1, II.4; also III.5, X.9; and Politics Eta 13.1332a42--1332b3. 13.Parfit 1984, Pt. I. 14.Sect. 2. 15.Sect. 10. 16.p. 49. 17.Is there not strong "local' colour to the way Parfit defines "success- fully following a theory'? "Say that someone successfully follows Theory T when he succeeds in doing the act which, of the acts that are possible for him, best achieves his T-given aims' (p. 53). This seems to assume that decisions are made occasion by occasion. And it is possible to some extent to live this way. But one important decision we make is whether indeed to live this way, a decision which itself then determines the set of "acts that are possible'. That set is not a datum, not fixed by human (or even one individual's) psychology. It is an agent's key, strategic decisions that importantly determine the set. 18.Sects. 9, 17. 19.Bernard Williams has a similar worry about the psychological implica- tions of a multi-level view, see Williams 1985, pp. 107--8 and my discussion later at ch.XI sect. 9. 20.Related issues arise in what Amartya Sen called an Assurance Game. See Sen 1967 and Sen 1974. See also the informative exchange between Kurt Baier and Sen: Baier 1977 and Sen 1977; and the discussion in Elster 1979, ch. I sect.4, ch. III sect. 7. 21.See, e.g., Kurt Baier, in Baier 1977, who identifies three central problems of Prisoner's Dilemmas: the (1) isolation, the (2) co- ordination, and the (3) assurance problems (pp. 197--8). 22.The sort of global consequentialism I am discussing here should be distinguished from Donald Regan's rigorously worked-out proposal of "Co-operative Utilitarianism', which says: "What each of us ought to do is to co-operate with whoever else is willing to co-operate in producing the best possible consequences, given what the non-co-operators are doing' (see Regan 1980, p. 124; that Regan sees this formula as a complete test of moral right and wrong is clear from his various summaries of co-operative utilitarianism on pp. x, 11, 124, 135--6, and 211). This seems to me to make co-operation far too central to morality. Regan's criterion fits some of morality poorly; it lacks scope. It gives an odd and roundabout justification for a ban on baby- battering. And it gives the wrong verdict in the case where co- 370 NOTES TO PAGES 207--210 operation is well under way, to the benefits of which I can add little, but where I can do more with some private enterprise of my own. And I think it misses some points about fairness. No doubt we should produce the best consequences by our all co-operating and showing up for votes unless there are votes spare, no long-term damage will be done by skiving, etc. But that tells us nothing about issues of fairness, of the sort that I discuss in the next section, that turn not on the quality of consequences but on who can legitimately be let off contributing. 23.This way of speaking is, of course, very loose. Before I choose what to do, all consequences are hypothetical; and after I have chosen, only one set of consequences will be actual. In that sense, I never choose between actual consequences. So by "actual' consequences I mean what would actually happen if I were to act in a certain way (and "hypothetical' consequences must be understood accordingly). But this looseness of speech does not matter, because it is eliminable. This looseness of speech is well pointed out in Singer 1982. 24.The problem is raised in Diamond 1967, and the example I use is adapted from his. My thoughts have been much clarified through conversation with John Broome and by reading an early version of Broome 1984. 25.Kurt Baier, in his reply to an earlier version of this argument, objected not that the sort of concern for individuals in the principle of equal chances was anything so blatant as inconsistent with impartial maximiza- tion, but rather that it was "contrary' to its "spirit'. That earlier discussion was carried on in terms of "spelling out' utilitarianism, so let me use that vocabulary. Utilitarianism, Baier says, contains a conception of impartial- ity ("everybody to count for one, nobody for more than one') that implies "only that all persons matter equally, not that every person matters.' In fact he adds, it would be easier for a utilitarian, in applying his maximizing standard, if no particular person mattered at all; otherwise "he would be emotionally torn apart by those maximizations that require him to sacrifice one for the greater good of all'. He sums up: "Equal regard for persons here is tantamount to equal disregard'. (Baier 1985, pp. 121--2.) I find it difficult to pin down Baier's point. Perhaps it is this. When we tease out the spirit of the principle of utility, we find---to put it more bluntly than Baier does---that it is really pretty dreadful: no one matters very much; at least, a utilitarian will induce that attitude so that he will be able emotionally simply to get through life; a maximizer is unlikely to have even that modicum of concern for persons that the principle of equal chances represents. But this seems to me to pack far too much into the sort of "equal disregard for persons' that a utilitarian can be saddled with. Being torn apart need not be outside the emotional repertoire of a maximizer. And even if, for a quiet life, he turns himself into a detached bureaucratic deadbeat, there is no reason why part of his bureaucratic approach could not be to equalize chances in order to break ties. Strictly 371 NOTES TO PAGE 211 speaking, "equal disregard' means merely "everybody to count for one'. But that familiar sort of impartiality does nothing to direct us to reject the principle of equal chances. It does not, as I admitted in my paper, do anything to force us to accept it either. I claimed only that a utilitarian can parlay his deep notion of equal regard into that principle among others. I think that he gets attractive development of that underdeter- mined notion of "equal regard' if he does so. Baier's point then, if I am right, is close to the now familiar observation that utilitarianism ignores "the separateness of persons'. If it ignores it, why would it insist on it in a principle of equal chances? But the observation, though familiar, is distressingly obscure. "The separate- ness of persons' is sometimes used as a reason why utilitarians are wrong to transfer the intrapersonal maximizing standard to interpersonal cases. But the only clear sense to attach to the observation is that utilitarians do indeed make the transfer. The observation cannot, therefore, be a reason why the transfer is wrong. Modern contractualists, who are chief among those pressing this observation against utilitarians, go on to impose restrictions on the operation of utilitarian trade-offs which look attractive only up against the counter-intuitive permissiveness of utilitarianism. But once seen on their own, their restrictions are discovered to suffer from equally counter-intuitive strictness. The notion of "the separateness of persons' is far too obscure to help us decide whether the indisputable sense in which utilitarianism ignores the separateness of persons rules out its including a principle of equal chances. I agree with Baier that one cannot extract the principle of equal chances from the principle "everybody to count for one'. That would be to misunderstand what Bentham and Mill meant by it; they meant it is a mere spelling-out of the principle of utility. And, of course, there is a big difference between an equal right to happiness (as embodied in "every- body to count for one') and a right to equal happiness. And, true, utilitarianism embraces only the first. If I understand Baier correctly, then, he thinks that maximization displays a sort of moral insensitivity that rules out utilitarianism's being sensitive to equal chances. But we ought to drop the vague, rhetorical "equal regard', "equal disregard', and "individuals do (do not) matter'. There are ways in which utilitarianism allows individuals to matter, and ways in which it does not. "Everybody to count for one' means that in a way individuals do matter; they matter (weigh) equally in the calculus. Maximization means that in trade-offs which person benefits does not matter, only the sum. But that is part of the view about when trade-offs are justified. And individuals do not matter in the sense that utilitarian- ism offers no protections for individuals against the results of maximiza- tion. But what remains mysterious is why Baier thinks if utilitarianism excludes protection of individuals against the results of maximization (disregards persons in that sense) then it cannot include equal chances for individuals when there are equal maxima (regard persons in that sense). 26.There is, though, a special problem for my proposal that does not 372 NOTES TO PAGES 212--213 bother a Ross-like pluralism. My proposal makes the maximizing prin- ciple lexically prior to the two tie-breaking principles. And there is a general problem with any lexical ordering: why should there be such sharp cut-offs in relevance? Why should equal distribution or equal chance at it matter importantly until different total utilities appear on the scene and then not at all? Is there a coherent notion of equal respect that does not count equality of well-being in trade-off situations but does count it outside those situations? Is it not only coherent but also plausible? Equality crops up in different places all through moral theory. There are many possible principles of equality, several of them plausible and no doubt many of them to be accepted. One of the hard jobs in moral theory is to sort out these easily confused, but different, principles of equality and to get one's thinking about equality straight. To decide whether the conception of equality in this padded-out utilitarianism is acceptable requires a lot of work, especially in sorting out our intu- itions. It is hard to reject this conception of equality from the start as incoherent or as suffering from internal conflicts in the "spirit' of its parts. I suspect that it even has something to recommend it. But that is contrary to what I take to be accepted opinion, so needs a lot of argument. I give some of the argument at ch. IX sect. 3 and ch. XI sects. 5, 9. 27.It is this deep notion that would also animate ranking, say, taking turns above tossing coins. Suppose the situation will recur. You as the parent might always opt for a fifty per cent chance for each child, or you might, because this sort of randomizing could just result in a freak run of benefits for one of them in particular, prefer directing it once to one child and next to the other. The second approach does seem better---better because safer, and safer precisely in more surely producing the fairer or equal distribution that seems to be at the heart of this principle. Does this not show, after all, that there is a powerful independent principle of fairness? I do not think so. Utilitarianism's basic, original equipment is a commitment to equal distribution and a standard to govern deviations. The standard governing deviations does not sanction any in this case. So there is no justification here to deviate. This is what the deep conception of equal respect at work in impartial maximization would say. I am indebted here, and elsewhere in this section, to conversations with R. M. Hargrave. 28.Broome 1984 suggests that it is. 29.There are different kinds of free-riding. Sometimes my free-riding adds a real, though often very small, extra burden on each of the others (e.g. my taking my private car on the roads at rush hour, when the rest of you use buses; I may slow things up so little for each of you that you do not even notice; but I do slow things up a bit). Other times my free-riding adds no extra burden (e.g. when I have a log fire, and the next rain washes the deposit away). Still, in both kinds of case there is a steady burden that all the rest of you bear, viz. a benefit forgone. I can get my questionable justification going for my having the benefit ("no harm done') only because the rest of you are denying it to yourselves. 373 NOTES TO PAGES 213--217 30.This point has been made before; a good discussion is Sumner 1971, esp. p.109, but he holds that a fair procedure such as random selection stands outside utilitarianism as something which, though it might often be adopted on utilitarian grounds (e.g. if it were the only procedure that a group would consent to), might sometimes not be (e.g. if one could get away with loading the dice). 31.Here I just follow Derek Parfit's powerful reply to those who argue that in a nation-wide election the consequences of my act can never explain why I ought to vote; see Parfit 1984, pp. 73--5. 32.Expected benefit is actual benefit multiplied by the chance that one will produce it, i.e. the product of the average net benefit per citizen and two hundred million, divided by one hundered million. 33. Parfit 1984, ch. 3. 34.Nor does Parfit say that it is. I take it that his position is this. He considers the common claim that in Contributor's Dilemmas involving very many persons what each person does would make no difference. This, he says, is just false, and he goes on to show with great ingenuity that very often it does make a difference. But often is compatible with sometimes not. See Parfit 1984, esp. pp. 66--7. 35.It is certainly too weak to solve all problems of "fairness'. All that I am proposing is that it solves the free-rider problem. That still leaves other, central problems; e.g. a society with a small group of rich exploiters and a large group of poor exploited has a structural unfairness that is not removed just by giving all entrants into the society (say, babies at birth) an equal chance of being among the exploiters. I discuss this sort of structural injustice later in ch. XIII. 36.For a good discussion of how consequences enter the Categorical Imperative test, see Paton 1965, ch.VII sect. 4, ch. XIV sects. 4 and 5, ch. XVII sect. 4. For a contrary view, see Singer 1963, ch. IX sect. 3. 37.This is Paton's Formula I, or "The Formula of the Universal Law'; see Paton 1965, ch. XIV. For another good discussion of the test see Nell 1975, ch. V. 38.Since fairness is a "perfect duty', what ought to appear is a contradiction in conception, rather than a contradiction in the will; see Paton 1965, pp. 148, 171--2. But it is hard to see how to make a convincing case for the appearance of either sort of contradiction. 39.The much-discussed problems raised by talk of "moral relevance' I pass by. They are satisfactorily answered only by constructing a complete substantive moral theory, which then settles what has moral weight and what does not. 40.The case is elegantly stated by M. G. Singer; see Singer 1963, pp. 91, 189, 196, 209 and also Singer 1955, p. 375; Singer 1977, p. 406. I have 374 NOTES TO PAGES 217--226 somewhat changed his arguments but think that what I present is still in their spirit. 41.In Paton's classification, Formula II, "The Formula of the End in Itself'; see Paton 1965, ch. XVI. 42.There is textual support for this reading of Kant: e.g. Grundlegung, Paton's trans., Paton 1961, pp. 67--8 (marginal pagination); "... the man who has a mind to make a false promise to others will see at once that he is intending to make use of another man merely as a means to an end he does not share. For the man whom I seek to use for my own purposes by such a promise cannot possibly agree with my way of behaving to him, and so cannot himself share the end of the action.' 43.ch. IX sects. 3 and 4. 44.See Baier 1985, pp. 123--4. 45.Baier 1985, esp. pp. 125--9. 46.I must, however, acknowledge the difficulties of applying the broad principle of equal chances. If, when I walk up to the doctor, he takes me rather than you, his choice is in a way random; it is chance that I am injured in the arms and you in the legs. Chance distribution does not have to be man-made. When the principle of equal chances is observed, and when not, is hardly always easy to say. Clearly more needs to be said about what this principle amounts to, particularly about what is to count as having an equal chance. 47.ch. IX sects. 3 and 4. CHAPTER ELEVEN 11. See Thomson 1971. 12.Dworkin 1977, esp. Introduction, pp. xi--xv and chs. 6 and 7. 13.Nozick 1974, pp. 28--33. 14.For useful discussions on the relation of "rights' to "duties' see e.g. Arnold 1978; on the relation of "right' and "permission' see Nozick 1974, p.92; on the relation of "rights' and "entitlements' see McCloskey 1965, p. 117. 15.Hohfeld 1923. 16.e.g. Melden 1977, ch. VI; McCloskey 1975, pp. 413--14. 17.They are elements supported by Mill, and more recently hinted at in Berlin 1969, Introduction, p. lx; and appealed to, at least some of them, in Hart 1979. 18.See ch. IV sect. 3. 19.I am therefore distinguishing "autonomy' from "liberty'. They both often get lumped together under the heading "freedom', but I think that there are many reasons for keeping them separate. It helps to have separate names for two very different stages of agency: the early stage of 375 NOTES TO PAGES 226--235 choosing one's path through life (autonomy) and the later stage of not then being stopped by others from going down it (liberty). Also what is true of the one cannot simply be transferred to the other. On this account autonomy can be (and often is) attached to criteria of very different stringency, all the way from common-or-garden choosing for oneself (say, not dominated by Mother) up to one's choices being totally outside the causal nexus (non-heteronomous, in Kant's sense). 10.This positive freedom is suspect in many eyes. "Poverty can be evil', John Lucas writes, "and great poverty is a great evil: but it is a different evil from lack of freedom.' See Lucas 1966, p. 147. 11.This is close to Dworkin's position; see Dworkin 1977, p. 267. 12.See Rawls 1972, p. 27; Nozick 1974, pp. 32--3, and by now many others. 13.ch. IX sects. 3 and 4. 14.See esp. ch. IX passim, ch. X sect. 3; I summarize my remarks about equality in ch. XIII sect. 4. 15.It is, according to the United Nations Declaration of Human Rights. It is, according to the US Supreme Court in its decision Griswold v. Connecticut (381 US 479) and Roe v. Wade (410 US 113); see discussion in Wellman 1978. 16.Someone might hope that some of this indeterminateness could be dispelled by appeal to justice, that is, by appeal to the idea that rights are, roughly speaking, the claims that a just society would grant. But justice is no more basic than rights. The issues about rights before us now form a large portion of the stuff of a theory of justice, so there is no independent notion of justice to help us with these issues. 17.This is a good place for a brief further thought about what a substantive theory is. The theory grounds rights in, among other things, personhood. But to the sceptical eye this may look like no "grounding' at all. It may seem merely a shift from one normative notion to another. After all, the content of the notion of a "person', at least "person' as it figures here in this substantive theory, is not a purely semantic matter; most of us would single out autonomy and living out one's individual life plan as central to being a person, but we well know that other persons, especially in other ages, have not thought these features important. Even if it were a purely semantic matter, the semantics might just be reflecting a certain normative outlook. And if normative considerations enter in the choice of a concept of "person', why not let them enter earlier with the choice of what is to be a human right? Why the indirection? The appearance of grounding rights in a concept like personhood may, to the sceptical eye, look illusory; a substantive "theory' may seem to be no theory at all, because its explanans may seem as much in need of grounding as its explanandum. But this carries scepticism too far. True, not everyone uses this concept of personhood. In that sense it is not totally outside the normative circle. 376 NOTES TO PAGES 236--243 But that does not mean that the notion of rights is not explained by it. Explanation in morals does not fail to be explanation unless it employs notions either that are not normative or that everyone employs. It is enough if it grounds a vague, troublesome, criterionless explanadum in an explanans that is more definite, clearer, and in the use of which we are more sure. Also, think of the alternative. The sceptical spirit prompts us to make normative decisions at an earlier point, directly about human rights. Well, what would we then deem to be rights? What boundaries would they have? Are we to maximize their observance? If we are honest, we shall admit that we are baffled. We need help with all these questions, and the substantive theory is meant to give it. Of course, not everyone accepts the values in which rights are grounded (try convincing an Indian father in a culture in which arranged marriages are still standard that he is violating his child's right to autonomy). But this just shows that the debate about the existence of the right should be shifted back one stage to a debate about whether autonomy, all things considered, makes a better life. There is a lot to be said about that. 18.Mackie 1978, pp. 354--5. 19.See the useful discussion of Mill's view in Lyons 1976. 20.See e.g. Gewirth 1982, p. 160. 21.Rawls 1972, sect. 11. 22.See his "Rights, Utility, and Universalization', in Frey 1984. 23.There may well be doubt about this approach: perhaps trade-offs, especially the sort that I shall presently consider, are not central to explaining moral status. John Rawls' approach, in contrast, is to suggest that we get straight the principles that apply to normal people, with normal capacities and health, and attend only later to the special demands of the abnormal. Otherwise, our intuitions, he fears, will be distorted by the special urgency of the ill and handicapped (see Rawls 1975a, p. 96). His approach would be possible if it were possible to work out a notion of equal respect without determining what morality requires in these hard cases. And hard cases would arise even in a society of normal people. They do not arise simply from illness and handicap, but also arise when a society makes hard choices about how to balance demands of saving life, on the one hand, and promoting the constituents of a good life such as art and education, on the other. But we do not have determinate enough notion of equal respect to give us any answers, even in the relatively easy cases, independently of answers in the hard cases. 24.ch. IX sect. 3. 25.ch. IX sect. 3. 26.See ch. V. 377 NOTES TO PAGES 245--257 27.ch. X sect. 2. 28.For a good discussion of forms of "indirect ' utilitarianism, see Williams 1973. 29.See the helpful discussion in Scanlon (forthcoming). I have benefited from seeing a Ts. of that article and from discussion with Scanlon. 30.See ch. III sect. 5, ch. IV sect. 5. 31.ch. X sect. 2 and ch. IX sect. 3. 32.ch. X sect. 2. 33.ch. IX sect. 3. 34.I have discussed this more fully at ch. X sect. 2. 35.Warnock 1971, esp. ch. 2; Mackie 1977, ch. 5; see also Williams 1973a, pp. 107, 134--5 and Williams 1973c, p. 112. 36.See esp. Williams 1973a, pp. 106--10. 37.ch. X sect. 2. 38.ch. III sect. 3. 39.See above, nn. 2 and 3. CHAPTER TWELVE 11. "Analysis', however, is the first step. For good examples of it, see J. Feinberg 1970, esp. chs. 4 and 8, and Kleinig 1971. 12.This point is made in Kleinig 1971, p. 74. 13.For a good discussion of the link between desert and personal identity, see Parfit 1984, Pt. III esp ch. 15. Parfit discusses two general sorts of view about personal identity: Reductionist (that the fact of a person's identity over time consists simply in the holding of certain facts about physical and psychological continuity) and Non-Reductionist (that it consists in some further fact, e.g. the existence of a separately existing entity or ego); see sect. 79. Some Non-Reductionists think that only the deep further fact carries with it desert and thus that, if Reductionism is true, no one deserves reward or punishment. My own view is that, on the very general characterizations of the two view given above, Reductionsim is more plausible than Non-Reductionism, but that the most plausible version of Reductionism still leaves us with all the problems about desert that I am concerned with in this chapter (and the problems about distribution that I go on to in the next chapter). As Parfit says, reduced psychological connectedness would reduce responsibility (p. 326). Moral or religious conversion would seem to me a good example. But this particular sort of rupture in psychological connectedness has no clear bearing on personal identity. The notion of psychological continuity is extremely obscure. Suppose someone's values, beliefs, dispositions etc. change radically (Saul on the road to Damascus, a patient at a turning 378 NOTES TO PAGES 258--265 point in psychoanalysis, a wrongdoer who undergoes a moral conversion). Certainly there is rupture in many key sorts of psychological connected- ness. Yet it is often in such cases that one also feels most at one with one's own past; it is often only then that one lifts oneself out of the present moment and sees one's life as a single, comprehensible whole. These are complicated cases: in one sense there is especially strong unity of person, in another weak unity (at least, great change). The notion of psycho- logical continuity is too crude to lend the Reductionist much help in explaining what personal identity is reducible to. Also, some weakenings of psychological connectedness (e.g. moral conversion) weaken continuity of desert without weakening continuity of person. See also ch. II n. 31. 14.His prima facie obligation of "gratitude' is an example; see Ross 1930, pp. 21--3. Treating gratitude as an obligation raises problems as to whether morality allows some degree of partiality. A person usually builds up debts of gratitude to friends and family, because they have most opportunities to help. How much can these debts qualify equal respect? Where does gratitude end and favouritism begin? 15.Even if a community is disbanding, Kant says, and its members scattering to the four corners of the earth, they must execute the last murderer left in gaol, "for otherwise they might all be regarded as participators in the murder ... ' Kant 1887, p. 198. See also Kant 1965, pp. 101--2, 104--7. 16.See John Rawls' powerful dismissal of desert as a ground for distributing "income and wealth, and the good things in life generally'; Rawls 1972, sect. 48, esp. p. 310. See also D. A. J. Richards' contractarian theory of punishment, which, he says, has no place for desert; Richards 1971, pp. 127--32. 17.This is pointed out in Feinberg 1970, p. 82. 18."Gratitude is a debt, "tis true, but it differs from all other debts; for though it is always to be paid, yet it is never to be demanded.' Anon., "Characters and Observations', early eighteenth-century, quoted in Gross 1983, p. 199. 19.I am paraphrasing Lucas 1980, ch. 12, esp. pp. 203--4, where the case is beautifully put. 10.Ronald Dworkin distinguishes a distribution of resources that is "ambition-sensitive' from one that is "endowment-sensitive', and he thinks that equality allows the former but not the latter; see Dworkin 1981, p. 311. I earlier used what is in effect the same distinction in contrasting what creates no merit (endowment) from what does (doing better than par for one's endowments). But my suggestion now is that, though the disinction is important, it has no moral import: it supplies no moral reason for a distribution of resources to be ambition-sensitive. Not all that matters, even to how we ought to behave, matters morally. 11.See Feinberg 1970, p. 83. 12.Herbert Morris in his Introduction, Morris 1971, p. 1. 379 NOTES TO PAGES 266--269 13.See M. Buber's distinction between "real' and "neurotic' guilt, in Buber 1965, ch. VI; and a good discussion of similar topics in Taylor 1985, ch. IV. 14.Dostoevsky, in a letter to Katkov, editor of the monthly Russian Messenger, gave this first outline of Crime and Punishment: "My novel, besides, contains the hint that the punishment laid down by the law frightens the criminal much less than our legislators think, partly because he himself feels the desire to be punished.' 15.See The Genealogy of Morals, Second Essay, sect. 19, repr. in Nietzsche 1927, p. 706. 16.This is most brilliantly argued in The Possessed, in the suppressed chapter, "At Tihon's', which contains Stavrogin's confession. Stavrogin, overcome by a sense of emptiness, becomes increasingly violent. "I wanted to put powder under the four corners of the earth and blow it all up, but it didn't seem worth the effort.' He allows an intense but superficial attraction to lead him to rape a young girl, who out of shame kills herself. He feels a terrible need for punishment, which takes the form of violence directed against himself. "I conceived the idea of somehow crippling my life'; so he marries a lame servant in his lodgings. In time he writes a confession, for distribution to the police, the press, and all his acquaintances. He even comes to feel pity for the girl and wonders whether it is repentance. Tihon, whom he consults, knows that it is not. What Stavrogin regrets is not her death but how it haunts him. He looks to the publication of his confession to bring relief; it is his "last measure'. Tihon advises him against it. The confession reveals his contempt for the persons to whom it is addressed. He wants forgiveness from two or three selected persons, Stavrogin insists, "but by all means let everyone else hate me'. Tihon knows that Stavrogin will not be able to bear the ridi- cule that his extravagant gesture in confessing will attract. Tihon urges him to become a novice for a few years under a wise old monk whom he recommends. As it is, Tihon fears for him; he fears a new crime solely as a way to escape publishing the confession. 17.Although I borrow the label "atonement' from theology, I make what I call the Atonement View grimmer and less connected with spiritual growth than theologians do. The Christian doctrine of atonement is not a doctrine of pure payment of moral debts. It began as an account of a sinner's reconciliation with God: "at-one-ment'. So it too is completed only with a spiritual growth that overcomes the estrangement caused by sin. But, then, it naturally came to include the means of achieving this reconciliation, and expiation entered there. The word "atonement' is now used in popular speech more of the means than of the end, and I follow this use. See Dillistone 1968, esp. ch. I; for a good brief account see Dillistone 1983. See also Hodges 1955 and Moberly 1978, esp. Postscript. 18.Benjamin Whichcote, Moral and Religious Aphorisms, 1753. Also Aristotle, 380 NOTES TO PAGES 270--276 Rhetoric II, "We withdraw our wrath from the man who admits that he is justly punished'. 19.I find it especially hard to work out the consequences of the view I develop here for capital punishment. A lot has to be said, but there is what seems to me a striking argument against it: since I must respond to the wrongdoer as someone capable of repentance and reform, it seems that I may never, no matter what he has done, even if he has killed, respond by killing him. Execution seems hardly appropriate for someone who can reform, and even less appropriate for someone who has already reformed. 20.For the case for punishment's being a symbol, see Moberly 1968, esp. pp. 200 ff.; Feinberg 1970, ch. 5, "The Expressive Function of Punish- ment'. 21.See Lord Denning, in evidence reported in the Report of the Royal Commission on Capital Punishment, Cmd. 8932, 1953, para. 53, who describes punishment as "the emphatic denunciation by the community of a crime'. It is not. Denunciation is only a side-effect, and justified only if there are those who will listen, and only if it is the best form for the message to take, etc. See also Moberly 1968 and Feinberg 1970 on punishment as symbol. 22.See Axelrod 1984; see also Trivers 1971 and discussion in Elster 1979, p. 145. 23.The exception is Richards 1971; see pp. 127--32. But there the whole discussion takes roughly four pages, and although the brevity leaves it far from clear, it seems that he ends up adopting substantially the position that I argue for here. (There are differences, though: he says that punishment is justified only when it acts as a general deterrent or it ensures compensation. But this must be too strong a requirement. It would mean that a punishment that reformed but did not generally deter was not justified.) 24.I am passing over all the difficulties with the notion of proportionality. Our judgments about the size of the punishment that "fits' the crime rest on the shakiest sorts of intuition. I say that this rough notion of proportionality has to be made less so. But can it be? Certainly the punishments that we have seen as "fitting' the crime have varied wildly over time---not so much in their relative positions (though even there too), but certainly in their absolute positions. See, e.g., The Boston Evening-Post, 18 Oct. 1742: "Capital Punishment' We hear from Hartford in Connecticut, that two Men were lately convicted there of counterfeiting the Bills of Credit on the Colony, and sentenced to be branded in the Forehead with a hot Iron, to have both their Ears cut off, and to kept in Prison during Life. If some such moderate Punishment were to be inflicted upon such Offenders in this Province, instead of Death, "tis tho't we would soon exceed any of our Neighbours in Convictions. 381 NOTES TO PAGES 276--283 25.See Bentham 1789 for several good arguments for limiting the degree of punishment: esp. ch. XIII, the discussion of punishment that is "unprofitable' or "needless', and ch. XIV passim. But Bentham's concern is effective social engineering, and he nowhere mentions the single most important determinant: the role of desert at the core of punishment. 26.Quinton 1954; Hart 1968, ch. I esp. sect. 2a. There is more point to insisting on what is and is not punishment than Hart allows here, and to insist on the difference does not burke any substantive problems; they can still arise, but in a form that helps make them and possible solutions clearer. 27.When philosophers speak of "retributivism', they are well aware that the name covers different views; see, e.g., Hart 1968, ch. IX, pt.2. The two most important ones, to my mind, are the Atonement View and the weaker Accountability View---both at the heart of retributivism, both moral views, both non-utilitarian, but differing between themselves in the weight they give to desert. The Accountability View is the more plausible, the chief doubt about it, which I shall come to in a moment, being whether it occupies a place different from the Repentance View. 28.I have a well-known nineteenth-century case in mind---the aftermath of the sinking of the Mignonette. See Simpson 1984. 29.That is the appearance; the reality is quite different. On the question of free will, Bentham was an agnostic, Mill a compatibilist. Bentham often sounds like a hard determinist. He accepted the mechanical account of action in psychological hedonism---hence the purple passage with which the Principles starts; "Nature has placed mankind under the governance of two sovereign masters, pain and pleasure. It is for them alone ... to determine what we shall do ... the chain of causes and effects ... [is] fastened to their throne ... every effort we can make to throw off our subjection, will serve but to demonstrate and confirm it' (ch.I sect. 1). (See also ch. XVII sect. 19 where Bentham remarks on the ambiguity of "free'; and n. to ch. XIII sect. 2, where the end of punishment is said to be "control' of action---even what Bentham calls "reformation' seems to be a form of social control.) But the purple passage, Bentham hastens to add, is only purple: "metaphor and declama- tion'. Bentham certainly accepts universal causation, but he never showed much interest in the question of free will. In a letter to George Wilson (8 July 1789), he wrote, "Entre nous I don't care two straws about liberty and necessity at any time. I do not expect any new truths on the subject: and were I to see any lying at my feet, I should hardly think it worthwhile to stoop to pick them up.' Baumgardt suggests that Bentham relegated the whole issue to the category of idle metaphysics, which ethics, being a descriptive study, could get along well without (Baumgardt 1952, p. 87--92, 395--9). See also the discussion of Bentham's account of motiva- tion in D. Lyons 1973, ch. I sect. 3. Mill too can sound pretty harshly determinist. Like Bentham, he was a psychological hedonist. And his discussions of sanctions (Utilitarianism, ch. 382 NOTES TO PAGES 285--286 III) and of virtue (ch. IV) at times makes human behaviour seem crudely mechanical. But Mill, who discussed the free will problem in some detail, was a soft determinist, and he believed universal causation to be compatible with moral responsibility. "When we say that all human actions will take place of necessity, we only mean that they will certainly happen if nothing prevents ... We are exactly as capable of making our own character, if we will, as others are of making it for us' (Mill 1843, bk. 6 ch. II sect. 3). The main sources are Mill 1843, bk. 6 chs. I--II; and Mill 1865, ch. XXVI. There is a good discussion in Ryan 1970, ch. VII. CHAPTER THIRTEEN 11. Perhaps this is not the most perspicuous way to define this particular dimension. It might be better to break it down further, say into three dimensions of size (principles centred on fine-grained interpersonal comparisons are possible in families, but not in modern nation-states), nature of motivation (families are moved by concern for each other, while parties to a contract are moved by self-interest), and relations to production (you and I can co-operate in production, but parents do not generally produce the family's goods jointly with their children, and you and I cannot produce goods jointly with persons who will be living a hundred years from now). 12.Hume 1751, sect. 3 pt.1; see also Hume 1738--40, bk. 3 pt.2 sect.2. 13.Hume 1751, sect. 3 pt.1. 14.Hume thought that the extremes were symmetrical, because he saw no ground for justice except self-interest. We accept the demands of justice, he thought, because in the long run it is in our interest to do so. But it is in our interest only so long as goods are tolerably plentiful. In dire scarcity the ground is removed from under all principles of justice. But Hume was wrong to think that self-interest is a rich enough ground for justice. It would give me no reason to accept any principle of justice that required me to defer to anyone who could not hit back: say, a subject people who will not make trouble in my lifetime, or a future generation who cannot. The question of the grounds of justice is different from the question of our psychological capacity always to act justly. In desperate straits all but the best of us are likely to get ruthlessly selfish. But rules of justice will still apply. David Miller discusses some anthropological evidence that in extremely poor societies where survival is in question notions of justice actually play little role; see Miller 1976, ch. VIII. 15.These are Robert Nozick's terms; see Nozick 1974, pp. 153--60. For similar distinctions see Matson 1983 ("top-down' and "bottom-up' prin- ciples of distribution) and Flew 1983 ("forward-looking' or "backward- looking' principles). John Rawls' discussion of "procedural justice' is also relevant; see Rawls 1972, sect. 14. 383 NOTES TO PAGES 287--289 16.Robert Nozick's concentration on justice in acquisition is a case in point; see Nozick 1974, esp. 150--3. 17.On "procedural' justice see Rawls 1972, sect. 14; on "process values' see Summers 1974--5, and Lucas 1980, ch. 4. 18.This is a central argument of Karl Marx in Capital. 19.This is a case argued in the justly influential Reich 1964. 10.See, e.g., Rawls 1972, p. 62: "All social values---liberty and opportun- ity, income and wealth, and the bases of self-respect---are to be distributed equally unless an unequal distribution ... is to everyone's advantage ... For simplicity assume that the chief primary goods at the disposition of society are rights and liberties, powers and opportunities, income and wealth ... ' This passage is quoted by Anthony Flew (his italics) who also quotes the crass collectivist assumption made in the blurb of Michael Harrington's The Twilight of American Capitalism: "A notable study which analyses reasons why sharp inequalities in sharing of the nation's wealth are inevitable outcomes of American capitalism'; see Flew 1983, p. 163. 11.See, e.g., Nozick 1984, p. 149: "... we are not in the position of children who have been given portions of pie by someone who now makes last minute adjustments to rectify careless cutting. There is no central distribution, no person or group entitled to control all the resources, jointly deciding how they are to be doled out.' Or Matson 1983, p. 107: "On the agreement model of justice the question of distributing anything hardly arises. The main idea of justice from the bottom up is that people are to keep what they produce unless they voluntarily exchange it for what others have made. It is no part of the agreement model that there will be a Master Distributor at all, distinct from the producers.' 12.Robert Nozick sometimes writes as if it can be; see Nozick 1984, ch. 7 sect. I. But Nozick seems to hold a more developed view: appropriation plus productive use (pp. 174--7). There is a good discussion of Nozick's views about acquisition in Lyons 1982, see esp. p. 363. 13.I use here the formula "as much and as good' rather than the somewhat more common "enough and as good'. Locke uses the second formula in introducing the Proviso (Locke 1690, sect. 27). He uses the first as well (sect. 36: "as good and as large a Possession'; sect.37: "the same plenty'). And when he uses the second, his gloss makes it clear that he means the first (sect. 33). Robert Nozick tries to turn the Lockean Proviso into something much weaker than my reading of it here. "The crucial point', he says, "is whether appropriation of an unowned object worsens the situation of others' (Nozick 1974, p. 175). He distinguishes a weaker and a stronger sense in which another's situation might be "worsened' (p. 176). The weaker is that an appropriator prevents another from freely using, without appropriation, what he could before; the stronger includes that condition but adds the further condition that an appropriator takes away another's opportunity to improve his situation. Nozick takes the proviso 384 NOTES TO PAGE 289 in the weaker sense: does any appropriation leave others worse off than they are at the time of the appropriation? This seems to me implausible both as an interpretation of Locke (see Locke 1690, sect. 34, where what matters is said to be not just leaving as good but as good for the others' improvement) and as a moral principle. Take this case. There is a continent where everyone ekes out an existence by subsistence farming. One isolated tribe hits on the idea of irrigation, and their yield soars. It was just a matter of time before a member of another tribe stumbled on these innovators. By chance I am the one who does, and I rush home and appropriate the most plentiful water supplies, which were previously unvalued and unowned. I then produce vast crops on my land, and the rest of you, because of what I can afford to pay you, come to work my lands for me. Eventually I buy your lands, by offering a trifle more than the value you are able to get out of them. I trade the agricultural surplus to other tribes and become even richer. According to Nozick's version of the proviso, I am justified: you are not worse off than you would otherwise have been. So you and your heirs are condemned to live only slightly above subsistence level, while I and my heirs, through no merit of ours, live in luxury. Nozick is aware of this sort of attack. His defence is to say that in this case you are not in fact as well off as you would have been but for my appropriation. If I had not stumbled on the innovating tribe, one of you would be likely to have done so eventually (p. 181). He founds upon this fact a limit on property rights---e.g. "a time limit on patents, as a rough rule of thumb to approximate to how long it would have taken, in the absence of knowledge of the invention, for independent discovery' (p. 182). But this defence is not strong enough. It underestimates how much a surge in wealth for one person on one occasion can affect opportunities on all further occasions. Suppose that it is likely that it would have taken a generation before one of you stumbled upon the innovating tribe (I am utterly incompetent at finding my way about and when I stumbled on the innovating tribe I was far more lost than any of you ever get). In the space of a generation I can become vastly rich, and my wealth can revolutionize my future prospects. I can go off and get educated, hire advisers, and so on, while all of you toil away as subsistence farmers. So even if I have control of the water for only a generation, that is long enough to make me far more likely than any of you to come upon the next thing that advances wealth, and, in virtue of that, even more likely to come upon the next after that, and so on. A process of enhancing my welfare at the expense of yours is now entrenched. Nozick's version of the Lockean Proviso looks plausible because it looks innocent: it gives me control over the value that I alone have added (what could be more reasonable than that?) so long as I do not thereby make the rest of you worse off than you would otherwise have been (what could be fairer than that?) But at each point in time the story to be told about "how you would otherwise have been' is determined by earlier unfavourable distributions. Once you miss out on one opportunity---even if solely 385 NOTES TO PAGES 289--291 because of my navigational incompetence---your future opportunities may well shrink, and go on shrinking. Then Nozick's version of the proviso looks far from innocent. 14.Locke 1690, e.g. sects.31, 32, 34. I am following the usual interpreta- tion of Locke and taking the provisos to be necessary conditions of justified appropriation. This has been vigorously disputed in Waldron 1979. He thinks that the proviso against waste is indeed a necessary condition but that the proviso of as much and as good is only sufficient. What seems to be to count against Waldron's reading is that Locke regarded my having as much and as good as you as one of my rights (for a not entirely clear mention of the right see sect. 36, for a perfectly clear one see sect. 46). It is true that there are passages where Locke mentions only the proviso against waste as a necessary condition (sect. 46). But that is because Locke is speaking there of a setting, before the advent of money, when there is such abundance that everyone may take as much as he can use (i.e. no waste). It is not that the proviso against waste is the only necessary condition, but rather that it is the only one relevant to a setting of plenty. 15.Nozick 1974, p. 175. 16.By J. R. Pennock, "Thoughts on the Right to Private Property' and by Jean Baechler, "Liberty, Property, and Equality', both in Pennock and Chapman 1980. Baechler lists (p. 273, p. 278 n. 1): \\Greek \\i>dio| \\sa+ i>dia \\Latin \\proprius \\proprietas \\German \\eigen \\Eigentum \\English \\proper \\property \\Italian \\proprio \\proprieta$1 \\French \\propre \\proprie$0te$0 17.The empirical evidence about the growth of property is well reviewed by L. C. Becker in "The Moral Basis of Property Rights' in Pennock and Chapman 1980. He also shows the impossibility of any quick move from this evidence to conclusions about property. Humans are, he grants, territorial, acquisitive, and egoistic; but they are also loving, co-operative, loyal, and capable of enduring bonds. And it is an important fact about humans that they are adaptable to change. Becker also points out that territoriality, as humans seem to display it, is as much a group phenomenon as an individual one; so if it has lessons for property, it may support group property rather than private property. 18.Some people think that it is. See Dunn 1984, p. 39: Dunn attributes to Locke the view that since labour is the source of ownership, "entitlement and merit are fused together ... those who possess more will be those who deserve to do so'. See also Miller 1976, ch. VIII sect. 4: Miller observes that in, for example, pure market societies that are growing and full of opportunities, desert is in fact regarded by the society as important; some such principle as "let each look after himself and be rewarded in 386 NOTES TO PAGES 291--295 proportion to the talents he displays' naturally arises in situations in which each can cope, if he will. See also Becker in Pennock and Chapman 1980, p. 100--1: he speaks there of the "labour-desert' justifica- tion which he characterizes, " ... when labour produces something of value to others ... then the labourer deserves some benefit for it.' 19.See ch. XII sect. 3. Lane 1979, p. 68 makes the point that desert creates a strong claim not for material reward but for recognition, and that pay, once a certain level is passed, is valued by the "meritorious' not for what it brings but for what it says about their achievement. 20.See ch. XI sect. 2. 21.See ch. III sect. 3. 22.So, for instance, Robert Nozick argues; see Nozick 1974, pp. 160--74. Also David Lyons suggests that we can get property rights out of, among other things, a liberty-right to do what is not harmful to others (a minimal right to land might be seen as a special case of such a right, although our actual, fuller, rights to land have to be fleshed out in other ways). But appropriations that have no detrimental effects on others are very rare; that test would be too strong. And if "not harmful to others' means that they would be no worse off than I am, then the appeal is to equality, not to liberty. See Lyons 1982. 23.ch. XI sects. 2 and 6. 24.This too is argued in Nozick 1974, pp. 268--71. 25.This is not the opinion of the drafters of eighteenth-century declara- tions of human, or natural, rights. Virginia's "A Declaration of Rights' of 12 June 1776 says: "That all men are by nature equally free and independent, and have certain inherent rights ... namely the enjoyment of life and liberty, with the means of acquiring and possessing property ... ' It is well known that in the following month the drafters of the Declaration of Independence of 4 July 1776 (wisely) displayed doubts about the fundamental status of property (chiefly because Jefferson thought it a civil rather than a natural right) and for Virginia's "life, liberty, and property' substituted "life, liberty and the pursuit of happi- ness'. However, Clause II of the French Declaration of 1789 restored property to a fundamental place: "The end of all political associations is the preservation of the natural and imprescriptible rights of man; and these rights are liberty, property, security, resistance of oppression.' Property has kept this place in the United Nations 1948 Universal Declaration of Human Rights, art.XVII: "Everyone has the right to own property alone as well as in association with others.' 26.This ch. sect. 2. 27.This ch. sect. 3. 28.ch. IX sect. 1. 29.A striking example of this is the convergence of R. B. Brandt's utility-based distributive rules and Ronald Dworkin's rights-based rules. 387 NOTES TO PAGES 296--301 They disagree (apparently---how much they really disagree depends on what Dworkin would eventually offer as his substantive theory of rights) over how equality figures at the deepest level, but from their different starting points arrive at astonishingly similar conclusions about the distributive rules that should be at work in society. See Brandt 1979, ch. XVI and Dworkin 1981. 30.See ch. III. 31.ch. III sect. 4. 32.ch. III sect. 4. 33.ch. III sect. 4. 34.ch. III sect. 5. 35.ch. III sect. 5; see also ch. IV sect. 5. 36.What is dubious is to ask such questions as Dworkin's, "Which among various conceptions of equality states an attractive political ideal, if any does?' There are many conceptions of equality, most of them state a political ideal attractive in one setting, or at one level, and unattractive at another. Or Sen's question, "Equality of what?', the only possible answer to which is, "Very many different things'. And the large variety of places where equality is relevant raises doubts about the generality of approach in such campaigns as Lucas' "Against Equality' and Flew's against The Politics of Procrustes. See Lucas 1965, Flew 1981. 37.ch. IX sect. 3. 38.ch. IX sect. 3. 39.There are those who would resist the accommodation of fraternity within prudential values. Fraternity, they say, is a different kind of good, a communal good---that is, a good the value of which is not the sum of values to the individuals involved. A communal good, in other words, has to be explained at least in part as a value not to individual members of the community but to the community itself. Other such values might be a sense of community, solidarity, esprit de corps, or national sense of purpose. But I do not think that there are any such things as communal goods, on this definition. The arguments that seem to support their existence confuse different theses. It is clear that one cannot explain certain values just by reference to how an isolated individual life goes. To explain fraternity requires bringing in the character of a group, relations between its members, and so on. Fraternity is clearly the sort of value that an individual cannot enjoy on his own. But one must not confuse a definitional thesis, a causal thesis, and a value thesis. The definitional thesis says that certain goods (e.g. fraternity) can only be defined in terms which make reference to groups, personal interaction, and so on. The causal thesis says that certain goods can occur only if persons interact in certain settings in certain ways. The value thesis says that certain values cannot be explained simply in terms of a sum of values in individual lives. For there to be communal values the third thesis has to succeed, but all 388 NOTES TO PAGES 302--304 that seem plausible are claims that turn out to support only the first or second thesis. Admittedly, fraternity is not an easy notion to pin down. It can be understood as I propose in the text as a kind of prudential value: a valuable sort of personal relation. But, no doubt, some will think that that interpretation leaves out something vital. They might think that it is neither merely a prudential value nor a communal value as defined just above but rather a moral value, a view about what equal respect requires, a view that is more powerful than the one I accommodate in the text in that it rules out any deviations from equal distribution. But if fraternity has such a strong consequence, then to defend its status as a value is to defend this strong interpretation of equal respect. And how will a defender of fraternity do that? (I am grateful to Jeremy Waldron, who is more of a friend of communal values than I am, for discussion of these issues.) 40.ch. X sect. 2. 41.ch. XI sects. 7--9. 42.It has been suggested that in a sense none do---that the notion of "distributive justice' does not apply in this setting. See Lucas 1980, p. 220: "The concept of distributive justice is applicable within the context of limited associations, with limited and definite aims held in common. Such aims give guidance how the fruits of common activities should be distributed.' He even doubts that a modern nation-state has limited and definite enough aims to allow "distributive justice' to be applied to it; so groups of such states clearly would not. " ... before questions of distribution arise, there has to be an association which produces the goods to be distributed, whose values justify some suitable basis for distribution.' But this is implausible. It assumes that there are no group-independent principles of distribution, that we can have no adequate idea of what each person is due apart from the aims of the co-operative group. But there are principles independent of the aims of the group. And the aims of co-operative groups do not readily generate any distributive principles. Their aims are generally selfish: each co-operator seeks greater benefit from co-operating than he can get without co-operating. In any case, the moral principles that apply to distribution inside a co-operative group- --e.g. the demand for equal provision expressed in the proviso of "as much and as good'---apply outside it too, e.g. to international cases. 43.See, e.g., Hardin 1977, p. 18: "India, for example, now has a population of 600 million, which increases by 15 million each year ... every one of the 15 million lives added to India's population puts an additional burden on the environment ... However humanitarian our interest, every Indian life saved through medical or nutritional assistance from abroad diminishes the quality of life for those who remain, and for subsequent generations. If rich countries make it possible, through foreign aid, for 600 million Indians to swell to 1.2 billion in a mere 28 years, as their current growth-rate threatens, will future generations of Indians thank us for hastening the destruction of their environment?' 389 NOTES TO PAGES 304--308 44.If someone thinks that every rich nation owes restitution to every poor nation, then (to isolate issues) take some science fiction. Let us say that we wake tomorrow to be greeted by peaceful extraterrestrials, who have had to abandon their planet when it froze over. 45.There are a lot of unsettled questions about what the proviso amounts to. Suppose two nations had as much and as good land, but then, through no one's fault, the land in one of them suffered serious erosion. Or suppose that two nations started with as much and as good, but one of them struck vast amounts of oil. 46.It is not much talked about in the context of famine, although the United Nations Universal Declaration of Human Rights, which is so profligate in its assignment of rights that it may not be a precedent of much interest, says that everyone has "the right to a standard of living adequate for the health and well-being of himself and his family, including food ...' (art. XXV). 47.Nozick 1974, ch. 7 esp. pp. 178--82. 48.This is a point well made by J. R. Lucas (Lucas 1980, ch. 13) and Michael Walzer (Walzer 1983, chs. 2 and 3). But they over-make it. The United States, during its period of highest immigration, managed to assimilate vast numbers of people from very different backgrounds. They all arrived looking for a better life, and that is probably the only quality that a co-operator needs to have. Also I should not myself over-make the point about the casual link between retention of goods and incentive to produce them. No doubt most governments of rich industrial nations could substantially raise income tax in order to help poor nations and not affect incentives at all. 49.ch. IX sect. 4. 50.To get a sense of just how much work has to go into making a set of property rights determinate, consider A. M. Honore$0's useful taxonomy of the rights and obligations (and also rules and liabilities) that make up our contemporary concept of property. Honore$0's taxonomy (slightly expanded by L. C. Becker) goes: 11. (claim) right to possess 12. (liberty) right to use 13. (power) right to manage 14. (claim) right to income 15. (liberty) right to consume or destroy 16. (liberty) right to modify 17. (power) right to alienate 18. (power) right to transmit 19. (claim) right to security 10. absence of term 11. prohibition of harmful use 12. liability to execution (i.e. confiscation for payment of debt) 13. residuary rules (i.e. rules governing reversion to another) 390 NOTES TO PAGES 308--312 These elements may be combined in a large number of ways. A particular instance of ownership may have some of them and not others, or have them only to a certain degree; or some of these elements may be held by one person and some by another; or all the elements may be held jointly by several people, etc. See Honore$0 1961 and L. C. Becker, "The Moral Basis of Property Rights', in Pennock and Chapman 1980. 51.ch. XI esp. sects. 8 and 9. 52.Here I follow the suggestion in Ryan 1984, pp. 161--2. 53.Pt. One. 54.ch. XI. 55.ch. XI sect. 9. 56.ch. III sect. 7. 57.ch. IX sect. 3. 58.See ch. X. 59.ch. XI. 60.ch. XI sect. 9. 61.e.g. Nozick, Flew, and Matson; see their works referred to in n. 5; in Nozick 1974, see esp. p. 238.