1 Act utilitarianism is, of course, sensitive to what people believe the past to be like. It is not, however, sensitive to the truth of such beliefs. I discuss this issue at length in ‘Teleology, Consequentialism, and the Past,’ The Journal of Value Inquiry 22 (1988), 89-101.

2 See, for example: Mabbott, J.D. ‘Punishment,’ in Gorovitz, Samuel ed., Mill: Utilitarianism (Indianapolis: Bobbs-Merrill 1971), 88-98Google Scholar; Nozick, Robert Anarchy, State, and Utopia (New York: Basic Books 1974)Google Scholar chs. 3 and 7; Scheffler, Samuel The Rejection of Consequentialism (Oxford: Clarendon Press 1982)Google Scholar, chs. 1-3; and Williams, Bernard A Critique of Utilitarianism, in Smart, J.J.C. and Williams, Bernard Utilitarianism: For and Against (Cambridge: Cambridge University Press 1973)Google Scholar.

3 Note that relative to a given set, Pareto optimality is necessary, but not sufficient, for total welfare maximization.

4 For ease of readability here and below I express people's welfare in terms of numbers. Note, however, that I only make use of the ordinal properties (equals, is greater than, and is less than) of these numbers for a given person. Welfare is not assumed to be cardinally measurable nor interpersonally comparable.

5 In fact rights constrained Paretianism does not even presuppose that welfare is cardinally (reflecting intensity) – as opposed to ordinally – measurable.

6 Note that the set of actions relative to which Pareto optimality is to be assessed is the set of actions that make each person (taken one by one – not collectively) at least as well off as some (not necessarily the same for all) RCP candidate. The universal quantifier (for each person) precedes the existential quantifier (there is some RCP candidate). Thus, there need not be some (particular) RCP candidate relative to which everyone is at least as well off (i.e., actions need not be weakly Pareto superior to some RCP candidate).

7 For further discussion of the axiological/non-axiological distinction, see Ch. 2 of my dissertation ‘The Teleological/Deontological Distinction’ (University of Pittsburgh 1984), and my paper ‘The Teleological/Deontological Distinction,’ Journal of Value Inquiry 21 (1987), 21-32. I there argue that it is a more fundamental classificatory distinction than the traditional teleological/non-teleological distinction(s).

8 Given greater space I would defend a modified version of rights based Paretianism. The modified version would allow – but not require – agents (1) to waste their own welfare when it is at no cost to anyone else (so actions need not be Pareto optimal, if the only person who could be made better off is the agent); and (2) to sacrifice their own welfare for the benefit of someone else (so agents can choose to have their welfare drop below their minimum entitlement). For the present purposes, however, I shall ignore these modifications.

9 In Collective Choice and Social Welfare (San Francisco: Holden-Day 1970), Amartya Sen proves that in a certain sense it is not possible to consistently combine Pareto-optimality with certain rights considerations. In ‘How to Combine Pareto Optimality and Rights Considerations’ (unpublished) I use rights based Paretianism as an example of how Pareto-optimality can be combined with important rights considerations. Although this in no way undermines the validity of Sen's important theorem, it provides insight concerning the plausibility of the conditions on which Sen's theorem rests.

10 Here I assume that welfare is understood in a narrow, self-regarding sense according to which the welfare of the owners is not affected by the mere fact that my friend obtains a copy of their program without paying.

11 In ‘Rights and Agency,’ Philosophy and Public Affairs 11 (1982), 3-37, and ‘Evaluator Relativity and Consequential Evaluation,’ Philosophy and Public Affairs 12 (1983), 113-32, Amartya Sen develops a very different way of making an axiological theory sensitive to rights considerations. He does this by using a teleological (goal maximizing) theory, where the goal is sensitive to rights considerations. His theory, I would argue, is not appropriately sensitive to welfare considerations. Nonetheless, the motivation for his theory is similar to that for mine.

12 I have focused on rights based Paretian theories of the permissibility of actions. Similar theories can be formulated for assessing the permissibility of other sorts of objects (e.g., social institutions, basic social structures, etc.). For example, a rights based Paretian theory of the permissibility of basic social structures might judge a structure permissible just in case it is Pareto optimal relative to the set of social structures that are weakly Pareto superior to some appropriately specified social structure.

13 Note, however, that rights based Paretianism is subject to the following objection: consider a situation in which by making one person slightly worse off than his/her minimum welfare entitlement an agent could make millions of people very significantly better off. In such a situation (1) rights based Paretianism seems to say that it is wrong to make the one person slightly worse off, and (2) that seems like the wrong answer.

It is not automatic, however, that rights based Paretian theories will say that such an action is wrong. It all depends on what rights the theory is based on. Rights generate constraints that apply under specified conditions. An action type that violates the constraints under normal conditions may not violate the constraints under catastrophic conditions. Thus, for example, under normal situations, killing someone violates the constraints, but killing someone in self-defense does not. Likewise killing one person to save the lives of millions may not violate the constraints, if the constraints are appropriately specified. So, some – but not all – rights based Paretian theories have the objectionable implication. The question is whether the most plausible such theory has the implication.

Furthermore, even if the most plausible rights based Paretian theory has this feature, having this feature may be the least of several evils. Any theory that recognizes the separateness of persons by placing limits how one person can be sacrificed for the welfare of others will be subject to counter-examples of the above sort. And theories that do not recognize the separateness of persons are subject to an opposite kind of counterexample. Prohibiting the sacrifice of the one for the many in extreme cases is an objectionable feature, but overall (considering a wide range of cases) doing so may give better answers than not doing so.

14 Rights based utilitarianism should not be confused with rights constrained utilitarianism, which was introduced in Section 2. The latter is not axiological, and does not satisfy the monotonicity condition, because it absolutely prohibits violating the constraints.

15 Inspiration for this project comes both from Nozick's remarks in Anarchy, State, and Utopia, Ch. 4 on permitting constraint violations provided proper compensation is given, and from David Gauthier’s discussion in Morals by Agreement (Oxford: Oxford University Press 1986) of the importance of Pareto optimality and of using rights-like reference points to ensure a minimum welfare entitlement. I have benefited from critical comments on earlier versions of this paper made by David Braybrooke, Daniel Chan, Peter Danielson, Dave Davies, David Gauthier, Bruce Hunter, Tom Hurka, Shelly Kagan, Jan Narveson, and an anonymous referee for this journal.