² Scheffler, Samuel, ‘Introduction’, Consequentialism and Its Critics, ed. Scheffler, Samuel, Oxford, 1988, p. 1Google Scholar.

³ ‘Good enough’ can be defined in different ways. On the absolute approach, an action is right if the value of its outcome is above a certain absolute level. On the comparative approach, an action is right if the value of its outcome is above a certain percentage of the value of the best outcome available. For more on these kinds of satisficing theories, see Carlson, Erik, Consequentialism Reconsidered, Dordrecht, 1995, pp. 13–19CrossRefGoogle Scholar, and Slote, Michael, Common-Sense Morality and Consequentialism, London, 1985Google Scholar.

⁴ This form of consequentialism is dubbed ‘factualism’ in Carlson, p. 20.

⁵ Roger Crisp has suggested that some particularists would deny this. They would claim that normative status can be determined by certain facts even though it does not supervene on these facts. Normative status also supervenes on facts that enable other facts to be right-making. For instance, the fact that an act brings pleasure might count as a right-making feature in one situation but what enables it to be right-making here is inter alia the fact that the act does not involve any cruelty. If the act had involved cruelty, then the fact that it also brings pleasure would no longer have been a right-making feature. But this means that particularists will accept NS if the relevant F-aspects include enabling conditions of various sorts.

⁶ Moore, G. E., Ethics, London, 1961 [1912], p. 84Google Scholar.

⁷ One way to adjust this condition to an indeterministic framework would be to assign to each action a risky prospect that consists of a probability distribution over a set of outcomes, each member of which might obtain if the action were performed. Let us say that two actions a and b match each other with respect to a probability-weighted distribution of F-aspects iff the outcomes associated with a and the outcomes associated with b can be paired off so that each resulting pair will consist of two outcomes that are exactly similar with respect to F-aspects and have the same probability of occurring. Normative Supervenience would then state that if there is a one-to-one correspondence between the alternative actions in two situations with respect to their probability-weighted distributions of F-aspects, then whatever normative status an act has in one choice situation is shared by its counterpart in the other choice situation.

⁸ To adjust this condition to an indeterministic framework, assign to each action a risky prospect, i.e. a probability distribution over some set of states of affairs, each member of which could obtain were the action performed. The value of a prospect depends exclusively on the outcome-values and outcome-probabilities. More exactly, if two risky prospects can be paired off so that each resulting pair will consist of two outcomes that have the same value and the same probability of occurring, then the prospects have the same value. Consequentialist Supervenience would now state that if there is a one-to-one correspondence between the alternative actions in two situations with respect their prospect-values, then whatever normative status an act has in one choice situation is shared by its counterpart in the other choice situation. This principle would thus express probabilistic consequentialism.

⁹ As a matter of fact, CS will also allow one feature of the situation to influence normative status: the number of alternative actions. For CS is only a constraint on situations involving the same number of alternative actions. It does not tell us how to relate the normative structures of situations with different numbers of alternatives. But should a consequentialist theory be completely insensitive to the number of alternative actions? Theoretically, we could imagine satisficing consequentialist theories that define ‘good enough’ by averaging the total outcome-value of all available outcomes over the number of alternative actions. In a way, even maximizing consequentialism is sensitive to the number of alternative options since whether there is an obligatory action in a situation depends on whether there is more than one action with optimal consequences. Nevertheless, it seems alien to consequentialism to claim that the sheer number of alternatives is normatively relevant. To meet this problem, we could add the condition that if the available outcomes in two situations have exactly the same value (i.e. the outcome-values do not vary within each situation, nor do they vary from one situation to another), then the options in these situations have the same normative status (i.e. the normative status does not vary within each situation, nor does it vary from one situation to another). This would rule out theories that pay too much attention to the number of alternative actions. But satisficing and maximizing theories would still be included.

¹⁰ More exactly, my formulation captures both satisficing theories that claim that ‘good enough’ should be defined absolutely and those that claim it should be defined comparatively. CS is not satisfied by theories that claim that what is ‘good enough’ may vary from one situation to the other depending on how difficult it is for the agent to perform the actions in the situation. But this seems acceptable. After all, these theories suggest that the normative status of an action is only partly determined by the value of outcomes; it is also determined by facts about how difficult it is for the agent to perform the action. For a defence of this kind of theory, see Eriksson, Björn, Heavy Duty. On the Demands of Consequentialism, Stockholm, 1994Google Scholar . As was pointed out to me by Gustaf Arrhenius, my formulation of consequentialism will also rule out theories that incorporate a notion of supererogation partly defined in terms of the agent's efforts and sacrifices.

¹¹ Moore, G. E., Principia Ethica, Cambridge, 1903, p. 147Google Scholar.

¹² That a deontological theory can thus be transformed into consequentialism has been pointed out by many authors. See, for instance, Carlson, pp. 40–3, and Danielsson, Sven, ‘Konsekvensetikens gränser’ (‘The Limits of Consequentialism’), in his Filosofiska utredningar, Stockholm, 1988, pp. 69–95Google Scholar.

¹³ Sidgwick, Henry, Methods of Ethics, 7th edn, Indianapolis, 1981 [1907], p. 382Google Scholar. Actually, Sidgwick makes the stronger claim that I am bound to have this aim not just as a moral agent but as a rational agent. However, I am not committed to this stronger claim. Note also that in subscribing to Sidgwick's credo I am not suggesting that consequentialism should be construed as a decision method for deliberating agents. I take ‘aim at good generally’ in the credo metaphorically, suggesting that the normative status of the agent's actions depend on all valuables he can bring about, which, of course, is not the same thing as to say that the agent should maximize value.

¹⁴ I am here indebted to Eric Carlson who argues that the idea of non-discrimination lies at the heart of consequentialism. See Carlson, pp. 56 f.

¹⁵ See, for instance, Carlson, pp. 42 f., and Danielsson, p. 77.

¹⁶ This is not the same thing as saying that the truth of a claim about intrinsic value cannot vary from one agent to another. I take the distinction between agent-relative and agent-neutral value to be silent on the question about truth relativism in ethics. Furthermore, I do not want to deny that what is intrinsically good for an agent can vary from one agent to another if ‘intrinsically good for an agent’ is understood as ‘part of the agent's well-being’. The claim is that what is intrinsically good period cannot vary in this way.

¹⁷ Zimmerman, Michael J., ‘Review of Changing Preferences: A Study in Preferentialism, by Krister Bykvist’, forthcoming in Theoria, 12 fGoogle Scholar.

¹⁸ An action is immediately performable by an agent at a time t iff the agent has an ability at t to perform the action and the action starts at t (or slightly later than t). See Carlson, p. 77.

¹⁹ This kind of normative indeterminacy is argued for in Carlson, pp. 104 f., and Carlson, Erik, ‘Consequentialism, Alternatives, and Actualism’, Philosophical Studies, xcvi (1999)Google Scholar.

²⁰ Arguably, this theory is still a form of act-consequentialism since it assigns normative status to actions in virtue of the outcome-value of individual actions. Consequently, it seems wrong to assume that indirect theories cannot be act-oriented.

²¹ More exactly, an action is maximal for an agent if it is performable by the agent but not a proper part of any other action performable by the agent. For more on this notion, see Goldman, Holly, ‘Doing the Best One Can’, ed.Goldman, A. I. and Kim, J., Values and Morals, 1978Google Scholar.

²² Minimalism is defended in Sobel, J. H., ‘Utilitarianism and Past and Future Mistakes’, Nous, x (1976)Google Scholar, and in his ‘Utilitarian Principles for Imperfect Agents’, Theoria, xlviii (1982)Google Scholar, and maximalism in Goldman, ‘Doing the Best One Can’. For a thorough discussion of minimalism and maximalism, see Consequentialism Reconsidered, ch. 7.

²³ I defend a modified version of maximalism in ‘Alternative Actions and The Spirit of Consequentialism’, forthcoming in Philosophical Studies.

²⁴ More generally, this principle states that if you cannot do what you ought to do without doing action a, then you ought to do a in the prerequisite sense. For more on the notion of prerequisite ought, see Consequentialism Reconsidered, pp. 142 f.

²⁵ Zimmerman raises this criticism in Zimmerman, ‘Review of Changing Preferences’, 13 f.

²⁶ See, Consequentialism Reconsidered, p. 88.

²⁷ The morally relevant sense of ‘can’ is the sense that figures in the principle ‘ought implies can’.

²⁸ This comes close to the neo-utilitarian theory advocated in Feldman, Fred, Doing the Best We Can, Dordrecht, 1986CrossRefGoogle Scholar. Possibilism is also defended in Zimmerman, Michael J., The Concept of Moral Obligation, Cambridge, 1996, ch. 6CrossRefGoogle Scholar.

²⁹ Earlier versions of this paper were presented at the conference Decision Theory and Moral Philosophy: Some Common Themes, held at Department of Philosophy, Lund University in 1999Google Scholar, to the audience at Wlodek Rabinowicz's seminar in Moral Philosophy, held at Lund University in 1999, and at ISUS 2000, held at Wake Forest, NC. For helpful comments on earlier versions of this paper, I would like to thank an anonymous referee of this journal, Gustaf Arrhenius, Johan Brännmark, Erik Carlson, Roger Crisp, Sven Danielsson, Anandi Hattiangadi, Magnus Jiborn, Elinor Mason, Edward McClennen, Philippe Mongin, Jan Odelstad, Ingmar Persson, Wlodek Rabinowicz, Toni Rønnow-Rasmussen, Wayne Sumner, Peter Vallentyne, Jan Österberg, and, most particularly, Michael Zimmerman.

I am grateful for financial support provided by the Bank of Sweden Tercentenary Foundation.