1 Lewis, David ‘Elusive Knowledge,’ Australasian Journal of Philosophy 74, 4 (1996) 549–67CrossRefGoogle Scholar. All page numbers otherwise unattributed are to this article. An interesting recent discussion of various contextualist theories, with many references, is to be found in DeRose's, Keith paper ‘Solving the Skeptical Problem,’ The Philosophical Review 104, 1 (1995) 1–52CrossRefGoogle Scholar.

2 Much as I agree with his anti-contextualisrn, and take his argument as having some force, I think we need something stronger than the central anti-contextualist argument put forward by Schiffer, Stephen in ‘Contextualist Solutions to Skepticism,’ Proceedings of the Aristotelian Society 96 (1996) 317–33CrossRefGoogle Scholar. Schiffer's point is that the contextualist way of dealing with skepticism presupposes that competent speakers of English are so unclear about the meanings of their own utterances, so misunderstand their context-dependence, that they largely do not realize that a denial of knowledge in a skeptical context has no bearing on the truth of our knowledge ascriptions in everyday contexts. He thinks the idea that the denials and the ascriptions pass each other by so completely, and that we are unaware of this and confused about it, ‘is not a semantic story to be taken seriously’ (328).

3 An Inquiry Concerning Human Understanding XII, ii (Indianapolis: Hackett 1977), 110

4 The inconsistency between the assertions of the skeptic and the consistent non-skeptic is of even less use to the contextualist, since it is easily explained in common sense ways making no use of different standards of knowledge. The skeptic typically asserts a skeptical conclusion for the quite simple reason that she has met skeptical arguments which she finds she cannot answer. On the other hand, the non-skeptic typically either has never met skeptical arguments, or has met them but believes they are in some way answerable.

5 Lewis takes a strong line on elimination. It is no use saying that our evidence is such that it is less probable that we are brains in vats than physical inhabitants of a physical world much as common sense take it to be. Our evidence leaves the brain-in-a-vat hypothesis at least a possibility. Lewis's sense of ‘possibility’ here is as close to that of logical possibility as makes no difference to our discussion.

6 Lewis-ignoring does not mean that one is ignorant of X; one may have heard of X before, but not currently have X in mind. One may have X in mind at a time, and later forget about it, and as a result then be ignoring it. Ignoring, for Lewis, is just the opposite of attending, where attending to X just means having X currently before one's mind.

7 The use of the words ‘your mind’ in (i) and (ii) is deliberate. It is the attentions and ignorings of those (including S) who attribute knowledge to S, not necessarily S herself, which are relevant. See Lewis, 561.

8 Though no real friend to skeptics about knowledge, Lewis surprisingly rather appears to be a skeptic about justified belief. Arguing that justification is not necessary for knowledge, he says: ‘What non-circular argument supports our reliance on perception, on memory and on testimony? And yet we do gain knowledge by these means’ (551). The question is rhetorical. He quite clearly means that we do not have such arguments to justify our reliance on perception, memory or testimony. (One may of course take justification to be a matter which does not involve supporting argument [but rather, say, reliability], but Lewis cannot take this line, as then he would not have the argument he takes himself to have for justification's not being necessary for knowledge.) I take skepticism about justified belief to be a far more serious matter than skepticism about knowledge, but our topic here is knowledge.

9 Not all arguments for skepticism involve reference to skeptical possibilities. For example arguments to the effect that foundationalism, coherence theories and infinite regresses are all untenable may seek to show that we have no justified beliefs and hence no knowledge.

10 Lewis develops his views on ‘rules of accommodation,’ rules governing the interpretation of words, and the standards by which they are used in a conversational context, in his ‘Scorekeeping in a Language Game’ Journal of Philosophical Logic 8 (1979) 339-59. See esp. 355.

11 The idea of a Rule of Attention for Flatness is not mentioned by Lewis (though I imagine he would not be averse to the notion); it was suggested to me by Peter Godfrey-Smith.

12 It is tempting to think that the change of judgment view deals better with the fact that, looking back, the man will not just admit that he no longer knows, but will admit that he never did. He will not say ‘I knew then (by standards appropriate to that context), though I do not know now (by standards appropriate to the current context).’ But Lewis's theory explains this too, since the man's current attention to an uneliminated possibility of error forces him to raise his standards in his retrospective judgment. (As noted earlier [n.7], it is the ignorings and attentions of the ascriber of knowledge that count—not those of the subject to whom knowledge is ascribed.)

13 I treat skeptical possibilities as logical possibilities, simply on the ground that they would not in general be regarded as empirical possibilities. To settle how they should be regarded we would need something more refined than the intuitive empirical/logical distinction with which we have been working. That in turn might necessitate more careful and qualified views on our retractions of know ledge claims in the face of newly attended-to possibilities.

14 My argument is to the effect that even those who are contextualists are left with a real contradiction here. Even if contextualism were true to the extent that there are different, contextually determined, standards of knowledge, it is not acceptable to say that there are two different standards operating in this case. This argument is against the Rule of Attention, and not against contextualism generally.

15 This paper has benefited greatly from very helpful comments given generously by David Lewis and Stephen Hetherington. I have also been helped by comments from the audience after presentations of earlier versions at the Australasian Association of Philosophy Conference in July 1998, and at the colloquium of the La Trobe University School of Philosophy in September 1998.