1 Stich, S. P., The Fragmentation of Reason (Cambridge, MA: MIT Press, 1990), pp. 1–3Google Scholar.

2 There may be some cases in which people have epistemic duties imposed by their office, e.g., policemen, investigators, etc., but these are special cases which do not seem to me to serve as useful models for ordinary epistemic agents.

3 See Stich, The Fragmentation of Reason, for a sceptical answer.

4 Doubts about the importance of etiological recapitulation of evidentiary relations to the concept of knowledge and justification can be forcefully raised by considering the example of the gypsy lawyer from Lehrer's, KeithKnowledge (London: Oxford University Press, 1974), p. 122. Though Alston cites what appears to be this discussion on p. 103 he does not rebut Lehrer's treatment of the example there, nor does he discuss it elsewhere in this collectionGoogle Scholar.

5 See Essay 7.

6 One is reminded here of Moore's infamous attempt to give directions for finding sense data.

7 C is an INUS condition for p) iff C is not by itself either necessary or sufficient for p but forms part of a set of conditions that are jointly sufficient for p?.

8 See Essays 1, 2 and 6.

9 Sellars, W., “Empiricism and the Philosophy of Mind,” in his Science, Perception and Reality (London: Routledge & Kegan Paul, 1963), pp. 127–96. See esp. pp. 186ffGoogle Scholar.

10 It is “justification” rather than “knowledge” that is at issue here because of the unveridical nature of the belief in case 3.

11 See Essays 8 and 9.

12 Alston makes this same point in Essay 3.

13 For those unconvinced of this in the case of Sellars see “The Lever of Archimedes,” part of The Carus Lectures, published in the Monist, 64 (01 1981): 3–36; see esp. pp. 10ff.); for those unconvinced of this in the case of Descartes see my unpublished manuscript, “Science and Metaphysics: A Reconstruction of Cartesian Philosophy,” Dalhousie University, 1992, chap.8CrossRefGoogle Scholar.

14 I qualify the version of foundationalism deployed by Manifest Image in this way to distinguish it from some recent variants in which versions of reliabilism are counted as a version of foundationalism. The central difference is that while classical foundationalism accepts a doctrine of the given, reliabilist foundationalism does not, a consequence of which is that while classical foundationalism i s a theory of conditions for the acceptance of propositions, reliabilist foundationalism in general will not be.

15 There is considerable controversy about the question whether the central objects of epistemology should be acts of acceptance of beliefs and, if the latter, whether beliefs come in degrees. I abstract from these large and difficult issues here. For a good introduction to this controversy see Jeffrey, R., “Dracula Meets the Wolfman: Acceptance vs. Partial Belief,” in Induction, Acceptance and Rational Belief, edited by Swain, M. (Dordrecht: Reidel, 1970), pp. 157–85CrossRefGoogle Scholar.

16 I choose a warrant principle relying on the notion of intuitive obviousness for purposes of illustrating problems with mixing coherentist and foundational styles of justification across levels. I choose this illustration partly because it can be formulated in a way that makes iteration formally an unproblematic matter. I do not mean to suggest, however, that the concept of “intuitive obviousness” operates in an entirely uniform way across levels nor that iteration itself is without substantive problems. I also rely on this notion partly because I think that it will figure in any adequate treatment of foundational epistemic practices. However, the general thesis I will be defending below is that there are reasons not to mix coherentist and foundationalist styles of justification across levels, a thesis that applies to any acceptance-rule conception of foundationalism.

17 See Principles I, section 66ff. (CSM I, 216) where Descartes speaks of “clear knowledge” of sensations. CSM abbreviates The Philosophical Writings of Descartes, Vols. land 2, translated by Cottingham, J., Stoothoff, R. and Murdoch, D. (Cambridge: Cambridge University Press, 1984.)Google Scholar

18 Meditation IV: CSM 11,41.

19 In fact I think that there are other options, but my purpose here is to see whether we can mix styles of method/methodology across levels when the choice is restricted to one between coherentism and foundationalism. Also see the last paragraph, below.

20 This practice does not compromise the status of FWP as a foundational warrant principle since FWP assigns warrant to propositions on the basis of a condition that is not a belief. Nor does this practice compromise the status of the propositions thus warranted as foundationally warranted, since the inference involving FWP is employed to decide which propositions to accept, not to decide which epistemic status is enjoyed by those propositions.

21 Indeed, this rationale is the one operative in our alternative argument for Inter-nalism offered above in Part II, Section 5.

22 One reason for taking a cautionary approach to letting both atomic and conditional statements acquire foundational warrant is the danger of inconsistency, a danger that would not seem to exist if only atomic statements were assigned this status. Clearly this is a complication that needs to be sorted out in any foun-dationalist program.