¹ Howard-Snyder, Frances, ‘The Rejection of Objective Consequentialism’, Utilitas, ix (1997)Google Scholar.

² Ibid., 241. As she observes (ibid., n. 1), if the world is indeterministic, objective consequentialism must be formulated in terms of objective probabilities of consequences. Like Howard-Snyder, I shall ignore this complication.

³ There is also a problem, not discussed by Howard-Snyder, of finding out what the alternatives are in a given situation.

⁴ Proponents of a ‘coarse-grained’ criterion for individuating actions would claim that these actions are numerically identical, while philosophers who advocate a ‘fine-grained’ criterion would hold that they are different. For an extensive discussion of these issues, and for references, see Goldman, Alvin I., A Theory of Human Action, Princeton, 1976, ch. 1Google Scholar.

⁵ This formulation of the relevant condition is rather crude. As Howard-Snyder points out, we should perhaps require only that you would probably succeed if you were to try. On the other hand, as she also stresses, to get a sufficient condition of ability we might have to add the assumption that your (probable) success would not be due to sheer luck. Also, we might want to put the condition in terms of choosing or wanting, rather than trying. None of these complications is important for our discussion.

⁶ For such a denial, see Feldman, Fred, Doing the Best We Can, Dordrecht, 1986, p. 25CrossRefGoogle Scholar. Cf. also Goldman's distinction between an ‘epistemic’ and a ‘non-epistemic’ sense of ‘ability’ (Goldman, p. 203).

⁷ ‘Bananas Enough for Time Travel?’, British Journal for the Philosophy of Science, xlviii (1997), 379Google Scholar.

⁸ Howard-Snyder, 246.

⁹ The opacity of ‘can’–contexts, together with the principle that ‘ought’ implies ‘can’, does not logically entail the opacity of ‘ought’–contexts. It might be claimed, for example, that ‘ought’ implies ‘can under every description’. If we grant Howard-Snyder's condition of ability, however, this claim is surely incorrect, since every action will have descriptions under which it is not performable by its agent.

¹⁰ I use the term ‘act-type’ for denoting things like moving one's arm, screaming, closing a door, etc. In contrast, a particular action, or ‘act-token’, is the instantiation of an act-type by a particular agent at a particular time. (See Goldman, p. 10, for a wellknown statement of this view.) According to this usage, producing the best consequences available is, arguably, an act-type.

¹¹ There are problems about the notion of an alternative, etc., which I here ignore.

¹² Howard-Snyder, 247.

¹³ Ibid.

¹⁴ Ibid., 246.

¹⁵ Ibid. Some might want to argue that a moral theory whose criterion of obligatoriness does not refer to act-types is likely to fail in ‘action-guidance’. This is not Howard-Snyder's point, however, since she grants that ‘objective consequentialists can deny action-guidance a crucial role in a moral theory’ (ibid., 241 f.).

¹⁶ I wish to thank Roger Crisp, whose comments on an earlier draft helped me clarify several points.