1 Philosophical Explanations (Cambridge, MA: Harvard University Press 1981), 197-211

2 Elsewhere, I have provided a further way of arguing that there is no such entailment: ‘The Sceptic Is Absolutely Mistaken (As Is Dretske)/ Philosophical Papers 27 (1998) 29-43.

3 See, for example, Richard Feldman, ‘Fallibilism and Knowing That One Knows/ The Philosophical Review 90 (1981) 266-82,at 266-7 .

4 I will present a non-contextualist fallibilism about such knowledge. Stewart Cohen has responded to skepticism in a fallibilist way that is also contextualist: e.g., ‘How To Be a Fallibilist/ Philosophical Perspectives 2 (1988) 91-123. In a currently unpublished paper, I reject contextualist discussions of such matters.

5 The Significance of Philosophical Scepticism (Oxford: Clarendon Press 1984), 20-1

6 The Further Knowledge Condition does have an infallibilist instance, though Imagine Descartes's seemingly sensory evidence (for his belief that he is not dreaming) being so good as to provide all-but-infallible support for his belief. The Further Knowledge Condition entails that even this evidence is not enough. Yet for Descartes to improve it would, by hypothesis, result in his having infallibilist justificatory support for his belief.

7 For now, I ignore the Gettier problem, standardly taken to be the difficulty of deciding whether something (and if so, what) else is needed if a justified true belief is to be knowledge: Edmund Gettier, ‘Is Justified True Belief Knowledge?’ Analysis 23 (1963) 121-3.1 return to this issue in section VI. (I have argued elsewhere, though, for responses to the Gettier problem that allow us to retain the principle that a belief's being true and justified via good fallibilist evidence suffices for its being knowledge: ‘Actually Knowing,’ The Philosophical Quarterly 48 (1998) 453-69; ‘Knowing Failably,’ The Journal of Philosophy 96 (1999) 565-87; ‘A Fallibilist and Wholly Internalist Solution to the Gettier Problem,’ Journal of Philosophical Research 26 (2001) 307-24.)

8 The dreaming possibility described by Descartes supposedly involves one's internal experiences being phenomenologically indistinguishable in kind from the internal experiences one has when awake. So, one needs further knowledge (in addition to the knowledge of what internal experiences one is having) with which one can distinguish between one's having a dreaming experience like that and one's having an experience as part of being awake. In what follows, when I talk of the further knowledge being expected of one by the dreaming argument, this is what — on the skeptic's behalf — I have in mind.

9 This would be most clearly true of externalist elements in Descartes's supposed justification for his belief that he is not dreaming. For instance, the skeptic's idea could be that if Descartes is dreaming, then his belief that he is not dreaming is being formed too unreliably to be at all justified — regardless of whether he also has some introspectible experiences that would otherwise provide some internalist justificatory support for his belief.

10 Or, if that example is not sufficiently uncontroversial, suppose that you have a false belief that at least some of your present beliefs are at least somewhat justified. If this belief is false, then it is also wholly unjustified.

11 My discussion will be sufficiently general, therefore, to apply to what might seem to be significantly varied skeptical ways of thinking. For example, a skeptic could argue that Descartes's fundamental epistemic failing is that he lacks enough fallibilist justification to render his belief that he is not dreaming the best explanation of his apparently sensory data; and that skeptical way of thinking is not overtly infallibilist. However, insofar as being-the-best-explanation-of-that-data would be a justify catory feature of Descartes's belief, my analysis will imply that even this skeptic's underlying commitments are infallibilist.

12 Recall note 8's point about phenomenological indistinguishability.

13 Because the skeptic will deny that the non-skeptic can non-question-beggingly assume here that Descartes's dreaming is empirically possible, the skeptic should not be saying that the failing is one that happens in at least one empirically possible world, rather than just at least one possible world. This is why the skeptic has to rely here on an accusation only of the evidence's not entailing the belief's truth (as against its not nomically necessitating the belief's truth, for example).

14 Some skeptics might argue that Descartes fails to know that he is not dreaming, even given his introspectible evidence plus his having various other epistemic features. But this would not alter the analysis I am about to present. For those skeptics will still be saying that, given the extent to which Descartes remains dependent on his introspectible evidence, he falls prey to the dreaming possibility, thereby failing to know that he is not dreaming — no matter what epistemic help those other epistemic features would otherwise give him.

15 Different Cartesian skeptics about one's knowing that one is not dreaming might suggest alternative distillations of their thinking. I see this one as an instance of a way of reasoning that underlies, or is implied by, all others that might be suggested by various Cartesian skeptics, particularly given its use of the idea of phenomenological indistinguishability.

16 Again, the Gettier problem is being ignored until section VI. Epistemologists generally treat skeptical challenges and the Gettier problem as being quite separate issues, with proposed solutions to the one not being tested against the other. (Nevertheless, it is not so clear that this particular application of epistemological methodology is so warranted. See, for instance, Richard Kirkham, ‘Does the Gettier Problem Rest On a Mistake?’ Mind 93 (1984) 501-13, and my ‘Gettieristic Scepticism,’ Australasian journal of Philosophy 74 (1996) 83-97.)

17 Recall too that, as I am using these terms, the knowledge is fallible when its justification component is fallibilist — that is, when the justification is compatible with the falsity of the belief which is the knowledge in question.

18 Another version of the ensuing argument appears in my Good Knowledge, Bad Knowledge (Oxford: Clarendon Press 2001), 37-40.

19 Equivalently: He needs to know thathe is not/«si dreaming. For instance, he needs to know that he is not just dreaming there being a world around him, if he is to know that there is such a world.

20 My point here is not (contrary to what a skeptic might gratefully claim) that it is impossible to know that p before knowing that p. Obviously, that state of affairs is impossible; but so unilluminatingly impossible is it that the skeptic's requiring one to bring it about is incoherent on her part. (Bear in mind that she is purportedly demanding that one have some illuminating further knowledge if one is to know that p.)

21 The reasoning here is the same, mutatis mutandis, as earlier in this section. Its merely seeming to you that 2 + 3 = 5 is (1) its seeming to you that 2 + 3 = 5, along with (2) the fact that 2 + 3 ^ 5 . Therefore, your knowing that it does not merely seem to you that 2 + 3 - 5 is your knowing that not-[(l) and (2)] — your knowing that [not-(l) or not-(2)]. That is, it is your knowing that [It does not seem to you that 2 + 3 = 5, or 2 + 3 = 5]. In short, it is your knowing that E.

22 Plantinga, for one, argues that this is what Gettier cases show: Warrant and Proper Function (New York: Oxford University Press 1993), 32-7.

23 Because Descartes himself was an inf allibilist about knowledge, he should accept — as he did — that his evidence is inadequate if his belief that he is not dreaming is to be knowledge. However, this does not entail that fallibilists need interpret his epistemic situation in that way.

24 These were the conditions identified in section II as sufficing, if true, to show that one cannot know that one is not dreaming.

25 I have developed this conception much more fully in Good Knowledge, Bad Knowledge

26 This includes the person's meta-knowledge. Such knowledge, too, need not be i n f allibilist, as Feldman has explained in some detail: ‘Fallibilism and Knowing That One Knows.’ So, gradualist fallibilism can model, for instance, Descartes's knowing more, or his knowing less, fallibly that he knows that he is not dreaming.

27 I am grateful for the anonymous referees’ thoughtful and useful comments on this paper.