1 See Davidson's interesting discussion in ‘How is Weakness of the Will Possible?', inJ. Feinberg, ed., Moral Concepts (Oxford: Oxford University Press 1970).

2 For finer–grained deliberation, such points of view may be subdivided, as when trying to decide what she has reason to do from the point of view of making money, the surgeon reflects that while from the point of view of cash flow there is reason to operate, in view of possible malpractice suits, from the point of view of long–term capital structure there is reason to refuse to operate. It is in this way that one can reconstruct a distinction between reasons-on-balance and prima facie reasons even within some point of view.'

3 The point here is not that there can be reasons for action which do not function as reasons for all agents. The distinction between prima facie reasons and reasons-on-balance is orthogonal to that between reasons as expressing abstract communal norms and reasons as expressing concrete individual motivations.

4 The Moral Point of View (New York: Random House 1965), 38-9

5 See for instance Kurtzman's, D. ‘Ceteris Paribus Clauses: Their Illumination and Elimination,’ American Philosophical Quarterly, 10 (1973) 35–42.Google Scholar

6 ‘Because I have reason on balance to do it,’ is of course never a reason for acting, neither in the prospective context of deliberation nor in the retrospective context of appraisal. Citing reasons is always citing prima facie reasons. But this does not reduce the notion of a reason-on-balance to empty formality. Such citation Justifies an action only when it is further claimed, at least implicitly, that the considerations cited outweigh possibly countervailing ones not mentioned. It is this second claim which secures the connection to action-guidance and gives separate content to the notion of reasons-on-balance.

7 This issue has been addressed in a somewhat different way in the context of conditionalization in deontic logics. See Chellas, B. ‘Conditional Obligation,’ in Stenlund, S. ed. Logical Theory and Semantic Analysis (Dordrecht: Reidel, D. 1974)Google Scholar. For a comprehensive review of related literature, see the bibliography in Bernardo, G. di ed., Logica Deontica e Semantica (Bologna: Societa Editrice il Mulino 1977) 349–447.Google Scholar

8 ‘The Finality of Moral Judgements’ Philosophical Review, 82 (1973) 364-70

9 Ibid.

10 Ibid.

11 Foot, P. ‘Are Moral Considerations Overriding?', in her Virtues and Vices (Oxford: Blackwell's 1978) 181-8Google Scholar

12 For the best of these, together with a survey of the others, see Anderson, A. and Belnap, N. Entailment, Vol. I (Princeton: Princeton University Press 1975)Google Scholar especially Part I.

13 Of course there is a dual notion of negative relevance corresponding to that stated above, which holds Just in case the background beliefs alone do not entail not-q, but those beliefs together with p do entail not-q.

14 A more sophisticated account would assign probabilities to directly relevant considerations during the first stage and require a maxim stated in terms of the expected return along the dimension defining the perspective. the arguments we are concerned with can all be approached equally well in the simpler interpretation according to truth values, however.

15 See my ‘Freedom and Constraint by Norms,’ American Philosophical Quarterly, 16 (1979) 187-96.

16 The sort of significance captured in such a semantics concerns the appropriateness of inferences and other performances, including assertions about what ought to be done. Perspectival schemes for practical inference are thus more akin to assertibility semantics than truth-conditional semantics (see my 'Truth and Assertibility,’ Journal of Philosophy, 73 [1976) 137-49). So invoking such schemes does not beg the question against a non-cognitivist rendering of action appraisals.

17 This is one of the points of the discussion of duty in the first section of the foun. dations of the Metaphysics of Morals.