¹ With some qualifications, this is Kant's position. I take it that it also represents Hare's later theory, though Hare himself has, reasonably, been sceptical about distinctions between the ‘subjective’ and the ‘objective’. In a theory such as Hare's, moral principles and their particular consequences are taken to be prescriptions, and hence not possible objects of knowledge; their objectivity consists in their passing a certain test, which is itself said to be grounded in the nature of the moral point of view (in Hare's case, moral language). This raises, of course, a question of the relation between, on the one hand, a moral principle P, which cannot be an object of knowledge, and a claim of the form ‘P passes the test’, which presumably can be.

² A focus for recent discussion of these questions has been provided by ideas of convergence: see notably ‘Truth as predicated of moral judgments’ in Wiggins (1987). Wiggins’ view is criticised by Crispin Wright (1992, Ch.3), though it is unclear how far the notions of normativity that he uses avoid similar ideas. Contrary to some things that I have said earlier (particularly in ‘Consistency and Realism’ in my Problems of the Self (1972b)), I am now in sympathy with the aim of Wright's book, to give an account of truth itself that will have minimal substantive implications and will, so far as possible, leave epistemic and metaphysical issues to be discussed later. An adequate ‘minimalism’, as Wright argues, will need more than is offered by the ‘redundancy theory’; for a recent discussion of this, see Horwich (1990).

³ Both approaches go back to Plato, the first to the Republic and the second to the Theaetetus. For the criticism of ‘the examiner's stance’, which follows, see my ‘Knowledge and reasons’ (1972a).

⁴ Craig, 1990, 2. I put forward a sketch in this style in Descartes: the Project of Pure Enquiry, 1978, ch.2, having got the idea from the Australian philosopher Dan Taylor, who may have been influenced in this direction by John Anderson. Craig's rich and helpful development of the approach includes a convincing demonstration that attempts to define the third condition so as to produce sufficient and necessary conditions of knowledge are bound to fail.

⁵ I have discussed such concepts, and, to some extent, their possible relations to knowledge, in Ethics and the Limits of Philosophy, 1985, ch. 8.

⁶ There seems to be a serious problem in this area for the idea that David Wiggins has advanced in several publications (e.g. Wiggins, 1990), to the effect that with regard to some straightforward ethical judgments, as with some plain mathematical and factual judgments, ‘there is nothing else to think but that p’. Nothing else to think about what? If the question about which we are to have something to think is whether setting fire to the cat was cruel or not cruel, then we must think that it was cruel. But people need not have this thought if they are asked ‘is it all right to set fire to the cat?’ or ‘what did they do?’ Wiggins’ argument assumes that the concept of cruelty is always ethically to hand, but the problem is that it is not. With arithmetical examples and also, differently, plain factual examples, it is not like this.—I hope to discuss this matter more fully in a later paper.

⁷ As I implied in referring to the outstanding problems of truth (note 2) and of plain truth (note 6).

⁸ Susan Hurley, in Natural Reasons (1989), seems to be committed to this view.—The qualification ‘in ethical terms’ is important here. If there were a theory that understood cultural variety in terms of basic needs and physical circumstances, it would relate to that variety of human nature, but it would hardly advance the interests of a larger cognitivism, because the difference in circumstances would bear too remote an analogy to conditions of enquiry.