¹ Moore, Andrew and Crisp, Roger, ‘Welfarism in Moral Theory’, Australasian Journal of Philosophy, lxxiv (1996), 598CrossRefGoogle Scholar.

² For an account of the distinction between person-affecting and impersonal principles, see Parfit, D., Reasons and Persons, Oxford, 1984, pp. 393 fGoogle Scholar.

³ For a general defence of this sort of conceptual analysis, see Jackson, F., From Metaphysics to Ethics: A Defence of Conceptual Analysis, Oxford, 1998Google Scholar.

⁴ Sen, Amartya, ‘Utilitarianism and Welfarism’, Journal of Philosophy, lxxvi (1979), 468Google Scholar.

⁵ For instance, Andrew Moore and Roger Crisp define welfarism partly in terms of what they call the ‘exclusiveness thesis’, according to which ‘individual well-being is the only good or value which has basic moral significance’(Moore and Crisp, 598). And Shelly Kagan suggests that we ‘call a view … that claims that well-being is all that matters … welfarism’ (Kagan, S., Normative Ethics, Boulder, Colorado, 1998, p. 48)Google Scholar. Wayne Sumner simply claims that ‘welfarism is monistic’ (Sumner, L. W., Welfare, Happiness, and Ethics, Oxford, 1996, p. 186Google Scholar).

⁶ Or, alternatively, that inequality has intrinsic disvalue. It can be argued that these are distinct views, but I need not distinguish between them here.

⁷ For this reason, Kagan explicitly claims that welfarists cannot be egalitarians (Kagan, p. 48).

⁸ The term ‘prioritarianism’ is derived from Parfit's so-called teleological priority view, according to which ‘Benefiting people matters more the worse off these people are’ (Parfit, D., Equality or Priority?, The Lindley Lecture, University of Kansas, 1991, p. 19Google Scholar). Unlike the teleological priority view, outcome welfare prioritarianism allows for the possibility that benefits to individuals who are not persons may contribute to the value of an outcome. Again unlike the teleological priority view, outcome welfare prioritarianism rather straightforwardly lets natural distributions fall under the scope of the principle. For instance, it implies that an outcome is better if, as a result of a natural lottery, a certain sum of benefits is distributed so that everyone has equal shares than it is if this sum is distributed so that some have greater shares than others.

⁹ How we should specify the prior level of welfare to which I refer depends on what we take the temporal unit of prioritarian concern to be. The temporal unit of prioritarian concern is the temporal part of an individual's life we should focus on when determining the value of a further benefit to her. For instance, we may focus on the individual's past, present, or life-time welfare level.

¹⁰ Sen, A., On Ethics and Economics, Oxford, 1987, p. 39Google Scholar. For a similar characterization see Ng, Yew-Kwang, ‘Welfarism and Utilitarianism: A Rehabilitation’, Utilitas, ii (1990), 171CrossRefGoogle Scholar.

¹¹ Blackorby, Charles, Donaldson, David, and Weymark, John A., ‘Social Choice with Interpersonal Utility Comparisons: A Diagrammatic Introduction’, International Economic Review, xxv (1984), 328 fGoogle Scholar.

¹² Hare, R.M., Freedom and Reason, Oxford, 1963, p. 30Google Scholar.

¹³ This requirement is also imposed on welfarist views in Blackorby, 329.

¹⁴ Thus, according to Sumner, Wayne, ‘Welfarists believe that the foundational values in an ethical theory are agent-neutral’ (Sumner, p. 185)Google Scholar.

₁₅ Sen, A., On Economic Inequality, 2nd edn., Oxford, 1997, p. 10Google Scholar; Blackorby, 332.

¹⁶ This, of course, does not imply that unrestricted outcome welfarism rules out views assigning extra weight to benefits that accrue to people who have been worse offin the past. To do this, we need only expand the outcome we assess so that it includes these past benefits. I elaborate the point that an outcome may have to be expanded to capture all relevant values in section VTI.

¹⁷ Even Sen does not properly distinguish them. In Sen, , On Ethics, he first introduces ‘“welfarism”, requiring that the goodness of a state of affairs be a function only of the utility information regarding that state’ (p. 39)Google Scholar, and then goes on to claim that ‘welfarism is the view that the only things of intrinsic value for ethical calculation and evaluation of states of affairs are individual utilities’(p. 40).

¹⁸ Thus, in one place Sen defines welfarism as follows: ‘The judgement of the relative goodness of alternative states of affairs must be based exclusively on, and taken as an increasing function of, the respective collections of individual utilities in these states’ (Sen, , ‘Utilitarianism’, 468)Google Scholar.

¹⁹ Scanlon espouses such an impersonal version when he claims that ‘fairness and equality do not represent ways in which individuals may be better off. They are, rather, special morally desirable features of states of affairs or of social institutions’ (Scanlon, T. M., ‘Rights, Goals, and Fairness’, Consequentialism and its Critics, ed. Scheffler, S., Oxford, 1988, p. 81Google Scholar). Likewise, Temkin endorses an impersonal view of the badness of inequality in Temkin, L. S., Inequality, Oxford, ch. 9Google Scholar. (But for a criticism of his reasons for so doing, see Rabinowicz, Wlodek, ‘The Size of Inequality and Its Badness’, Theoria, lxix (2003)Google Scholar.)

²⁰ Broome opts for a person-affecting version in Broome, J., Weighing Goods, Oxford, 1991, ch. 9Google Scholar.

²¹ For a presentation (and criticism) of this idea, see Temkin, pp. 91 f. For a more general criticism of measures of inequality that focus on groups, see pp. 101 f.

²² Perhaps, strictly speaking, the claim that welfarists consider benefits good is too strong. Consider frustrationism, i.e. the view that while the frustrationof a preference has negative value, the satisfaction of a preference has zero value (for a defence of frustrationism, see Fehige, C., ‘A Pareto Principle for Possible People’, Preferences, ed. Fehige, C. and Wessels, U., Berlin, 1998)CrossRefGoogle Scholar. According to frustrationism, there really is no such thing as a benefit, unless that includes the mere avoidance of harm. Nevertheless, it may seem that frustrationism could figure in a welfarist view in which, say, an outcome is better the lower a sum of preference-frustration it includes. Perhaps. I am not sure about this. But note that unlike outcome welfare egalitarianism, such a view will be sensitive to increases in individual welfare (where these amountmerely to the avoidance of harm) whatever the level at which they occur. Thus, there is a clearsense in which such a view is considerably more responsive to increases in welfare than outcome welfare egalitarianism.

²³ See, e.g., Dworkin, Ronald, ‘What is Equality? Part 1: Equality of Welfare’, Philosophy and Public Affairs, x (1981), 188 fGoogle Scholar., and Broome, p. 178.

²⁴ This version differs from standard versions of the pareto principle, since standard versions are formulated in terms of preferences rather than benefits: see e.g. Broome, p. 152. It corresponds to Broome's principle of personal good:see Broome, p. 165.

²⁵ From now on, I shall not explicitly point out whether it is the all things considered or the ceteris paribus version I have in mind. I trust it will be clearfrom the context.

²⁶ Pareto outcome welfare egalitarianism thus implies what Parfit calls ‘moderate teleological egalitarianism’ (Parfit, , Equality, p. 30)Google Scholar. According to moderate teleological egalitarianism, a better outcome cannot be reached by levelling down, because the increase in equality will be outweighed by the decrease in benefits. This, however, does not imply that there is no levelling down objection to be raised against moderate teleological egalitarianism. As Parfit points out, such egalitarians are committed to the view that it can be in one respect better to level down. For instance, they are committed to the view that G is in one respect better than F. G is better as regards equality. Tern kin attempts to defuse this objection in Temkin, ch. 9. I argue that he does not succeed in Holtug, Nils, ‘Egalitarianism and the Levelling Down Objection’, Analysis, lviii (1998)Google Scholar, and Holtug, Nils, ‘Good for Whom?’, Theoria, lxix (2003)Google Scholar.

²⁷ Its plausibility, however, is questioned by some welfare-egalitarians.For instance, Dennis McKerlie suggests that there is no particular reason why egalitarians should weight equality and increases in benefits such that the pareto principle is always satisfied; see McKerlie, Dennis, ‘Equality’, Ethics, cvi (1996), 287Google Scholar.

²⁸ In H: [(10−4)+0+(10−4)]/3. In I:[0+(11−5)+(11−5)]/3.

²⁹ Thus, if a certain principle satisfies the egalitarian summative condition, it will also satisfy the egalitarian pareto condition, although a principle may satisfy the latter condition but not the former.

³⁰ Suppose we assume anonymity in our definition of welfarism. Then if there was a permutation of benefits over a-c that would render H pareto-superior, the claim that I was better would be ruled out. But there is no such permutation.

³¹ Broome, p. 69.

³² See e.g. Temkin, ch. 6.

³³ Sumner, p. 217.

³⁴ Sen explicitly makes a similar assumption. He takes the value of alternative states of affairs to be functions only of ‘the respective collections of individual utilities in these states’ (Sen, , ‘Utilitarianism’, 468Google Scholar, my emphasis).

³⁵ Even if we are physicalists, we should claim that pains and pleasures are, if not directly mentioned in, then, in some suitable sense, entailed (or fixed) by full descriptions of relevant outcomes: see Jackson, ch. 1. To accept physicalism and deny such an entailment would amount to an implausible eliminative view of psychological states such as pains and pleasures.

³⁶ I shall not attempt to give a more precise account of what features should be included in a full description of an outcome. Note, however, that such a description may have to include dispositional properties, specified counterfactually. Suppose, for instance, we hold a dispositional theory of desires. Since a disposition need not be manifested, perhaps its presence is to be explained counterfactually in terms of how its bearer would behave, if suitably stimulated. Furthermore, if benefits depend on desires, we may need to include dispositional properties (desires) in the description of an outcome. However, dispositional properties such as desires are genuine properties of entities, so we need not worry about including them in our full description of outcomes. Being soluble is a dispositional property that sugar has, but it is a genuine property nevertheless.

³⁷ For the point that rational desires have such a ‘transworld’ status, see Bykvist, K., Changing Preferences:A Study in Preferentialism, dissertation, Uppsala, 1998, p. 43Google Scholar.

³⁸ Along such lines, Harsanyi claims that ‘a person's true preferences are the preferences he would have if he had all the relevant factual information, always reasoned with the greatest possible care, and were in a state of mind most conducive to rational choice’ (Harsanyi, J. C., ‘Morality and the Theory of Rational Behaviour’, Utilitarianism and Beyond, ed. Sen, A. and Williams, B., Cambridge, 1982, p. 55Google Scholar). And Brandt informs us that he ‘shall call a person's desire, aversion, or pleasure “rational” if it would surviveor be produced by careful “cognitive psychotherapy” for that person’ (Brandt, R., A Theory of the Good and the Right, Oxford, 1979, p. 113Google Scholar). Finally, Rawls writes: ‘our good is determined by the plan of life that we would adopt with full deliberative rationality if the future were accurately foreseen and adequately realized in the imagination … Here it is worth stressing that a rational plan is one that would be selected if certain conditions were fulfilled. The criterion of the good is hypothetical in a way similar to the criterion of justice’ (Rawls, J., A Theory of Justice, Oxford, 1971, p. 421Google Scholar).

³⁹ For instance, Griffin says: ‘Utility, we might try saying, is thefulfilment of desires that persons would have if they appreciated the true nature of their objects’ (Griffin, J., Well-Being, Oxford, 1986, p. 11)Google Scholar, and then adds: ‘Utility must, it seems, be tied at least to desires that are actual when satisfied’ (p. 11). On this approach, ‘Whatever the idealizing conditions adopted, their effect will be to screen out some of our actual desires’ (Sumner, , p. 130)Google Scholar.

⁴⁰ This requirement of actuality has been supported in the following manner (Griffin, p. 11). Suppose that if properly educated in literature, I would want to read Proust but, having had no such education, I have no desire for it. Since I do not have the necessary training, I will not get anything out of reading Proust, and so it may seem doubtful that I will benefit from doing so. To accommodate this claim, we can introduce the requirement of actuality. In fact, I find this argument for the ‘actuality’ requirement rather weak. If we take a person's rational desire for a state of affairs, s, to be his desire in apossible world in which he is rational, but for the world in which s obtains, the counterexample poses no problem. After all, in a world in which I was rational, my desire to read Proust would not extend to a world in which I could not appreciate him. However, what I am concerned with here is not so much whether this account of rational desires is plausible as whether it can be accommodated by the Containment Clause so as to give rise to a plausible characterization of welfarism.

⁴¹ For the claim that only actual desires can give rise to intrinsic value, see Rabinowicz, Wlodek and Österberg, Jan, ‘Value Based on Preferences: Two Interpretations of Preference Utilitarianism’, Economics and Philosophy, xii (1996), 10Google Scholar. For the claim that only actual desires can give rise to benefits, see Steinbock, B., Life Before Birth, Oxford, 1992, pp. 40 f., 71 fGoogle Scholar. I critically discuss this view in Holtug, Nils, ‘On the Value of Coming into Existence’, Journal of Ethics, v (2001)Google Scholar.

⁴² I would like to thank Krister Bykvist, Roger Crisp, Karsten Klint Jensen, Kasper Lipper-Rasmussen, Robert Pulvertaft, Wlokek Rabinowicz, Bertil Tungodden, and Peter Vallentyne for some very helpful comments on an earlier version of this article. Thanks are also due to the Danish Research Councils for financial support.