1 The first version of this paper was written at the invitation of the Program Committee for the Western Canadian Philosophical Association, and presented at the Fall 1991 Meeting of the Association. I am grateful to members of that audience and to subsequent audiences at Queen’s University (Kingston), the Southern California Philosophy of Science Group, the University of Pittsburgh, the University of Illinois (Urbana), and Harvard University, for their help in the paper’s evolution. I also want to thank an anonymous referee for CJP for an extremely penetrating discussion. Many of the comments, questions, and criticisms that I have received will be taken up in future work.

2 This argument is advanced by Harman, Gilbert in Thought (Princeton: Princeton University Press 1973)Google Scholar, Ch. 2; Goldman, Alvin in ‘What is Justified Belief?’ in Pappas, G. ed., Justification and Knowledge (Dordrecht: Reidel 1980)Google Scholar; Komblith, Hilary in ‘Beyond Foundationalism and the Coherence Theory,’ Journal of Philosophy 77 (1980) 597-612CrossRefGoogle Scholar; and me in The Nature of Mathematical Knowledge (New York: Oxford University Press 1983), Ch. 1. The locus classicus is now Goldman’s Epistemology and Cognition (Cambridge, MA: Harvard University Press 1986).

3 Although these beliefs are causally basic, they do not have the status traditionally associated with foundational beliefs, of being justificationally basic: whether they are justified depends on other facets of the subject’s belief system. In other words, the explanation of why they are there does not involve reference to other beliefs, but the explanation of why they are justified does. This point was pioneered in several important papers of Wilfrid Sellars, particularly in ‘Empiricism and the Philosophy of Mind’ (reprinted as Ch. 3 of Science, Perception, and Reality [London: Routledge 1963]). However, as the referee for this paper pointed out, my reliabilistic development of this point differs in important respects from Sellars’s articulation of it. My approach is closer to that offered by Komblith in ‘Beyond Foundationalism and the Coherence Theory.’

4 Although, on many accounts of perception, our beliefs about ordinary objects result from complex psychological processes involving inferences from unconscious beliefs. See, for example, the account offered by David Marrin Vision (San Francisco: Freeman 1982).

5 Here I have in mind the kinds of views about language acquisition developed by Noam Chomsky in Language and Mind (New York: Harcourt Brace Jovanovich 1972), and other writings.

6 For discussion of this example, and a brief survey of the relevant historical material, see my ‘Theories, Theorists, and Theoretical Change,’ Philosophical Review 87 (1978) 519-47.

7 It might be better to say that a conditionally reliable process has failed to be reliable: Priestley has engaged in a process that has a good chance of generating true belief if the background to which it is applied was itself reliably generated. I am grateful to the referee who has convinced me that this approach to the reliability of belief-generating processes needs more detailed discussion than I can give it here.

8 See the closing sentences of ‘Camap and Logical Truth’ in The Ways of Paradox (Cambridge, MA: Harvard University Press 1976).

9 Problems about the proper form of an epistemic society have recently been discussed by several authors. See, for example, Goldman, Alvin ‘Foundations of Social Epistemics,’ Synthese 73 (1987) 109-44CrossRefGoogle Scholar; Rescher, Nicholas Cognitive Economy (Pittsburgh: University of Pittsburgh Press 1989)Google Scholar; and Hull, David Science as a Process (Chicago: University of Chicago Press 1990)Google Scholar. I discuss the specific problems noted in the text in ‘The Division of Cognitive Labor; Journal of Philosophy 87 (1990) 5-23, ‘Authority, Deference, and the Role of Individual Reason,’ in McMullin, Eman ed., The Social Dimension of Scientific Knowledge (Notre Dame: University of Notre Dame Press 1992) 244-71Google Scholar, and, most systematically, in the final chapter of The Advancement of Science (New York: Oxford University Press 1993).

10 Contemporary versions of this argument descend primarily, I think, from Thomas Kuhn’s The Structure of Scientific Revolutions (Chicago: University of Chicago Press 1970 [1st ed. 1962]), esp. 170-1, 206-7. But the arguments themselves are far older, occurring in highly developed forms in the writings of nineteenth-century idealists. Contemporary versions of the arguments which relate them to post-Kantian discussions of epistemological problems can be found throughout the recent writings of Hilary Putnam; see, for example, the title essay in Realism with a Human Face (Cambridge, MA: Harvard University Press 1992).

11 The clearest versions of this line of reasoning emerge in the writings of some contemporary sociologists of knowledge, who have been much influenced by the anthropological work of E.E. Evans-Pritchard on the Azande, and by the studies of Mary Douglas. See, in particular, Bloor, David Knowledge and Social Imagery (London: Routledge 1976;Google Scholar Chicago: University of Chicago 1991 [2nd ed.]), and many of the essays collected in Hollis, Martin and Lukes, Stephen eds., Rationality and Relativism (Cambridge, MA: MIT Press 1982)Google Scholar.

12 This is the ‘pessimistic induction on the history of science’ (or what Hilary Putnam calls ‘the disastrous meta-induction’ in Meaning and the Moral Sciences [London: Routledge 1978], 25). The challenge to convergence is very clear in Kuhn, The Structure of Scientific Revolutions. Sharp formulations of the argument have been offered by Laudan, Larry Science and Values (Berkeley, CA: The University of California Press 1984)Google Scholar, Ch. 5; and Arthur Fine, The Shaky Game (Chicago: University of Chicago Press 1986), Ch. 7.

13 This argument has its roots in an interpretation of the later work of Wittgenstein, and in sociological reflections on the writings of Douglas, Mary (in particular Purity and Danger [London: Routledge 1966]Google Scholar). For developed versions, see Bloor, Knowledge and Social Imagery, and especially Shapin, Steven and Schaffer, Simon Leviathan and the Air-Pump (Princeton: Princeton University Press 1985)Google Scholar.

14 The original version of the pragmatist conception is due to C.S. Peirce, and is developed in some detail by Sellars, Wilfrid in Science and Metaphysics (London: Routledge 1967)Google Scholar, Ch. 5; the account in terms of ideally situated knowers is proposed by Putnam, Hilary (see Reason, Truth, and History [Cambridge: Cambridge University Press 1981), 55-6CrossRefGoogle Scholar, for a particularly crisp formulation, and numerous passages in Meaning and the Moral Sciences, and other subsequent writings).

15 For careful elaborations of accounts of truth along these lines, see Grover, Dorothy Camp, Joseph and Belnap, Nuel ‘A Prosentential Theory of Truth,’ Philosophical Studies 27 (1975) 73-125CrossRefGoogle Scholar, and Horwich, Paul Truth (Oxford: Blackwell 1990)Google Scholar. As Fine (The Shaky Game, Ch. 7 and 8) makes very clear, his ‘natural ontological attitude’ amounts to using a ‘no-theory’ of truth to counter skeptical and relativistic objections to claims about scientific knowledge.

16 Putnam (Reason, Truth, and History) is sensitive to the worry that his account may seem to ‘explain a clear notion with a vague one’ (56). But the root of the trouble is not vagueness. Rather it is the difficulty of understanding what we could mean by hailing a process as justificatory or as ideally justificatory independently of its propensity to deliver truth. The problem may be approached in another way by reflecting on Michael Dummett’s preferred account of mathematical truth in terms of proof. What singles out certain patterns as proof procedures? What gives them their special status? We are, I think, bound to ask these questions, and, once the explication of truth in terms of proof has been adopted, we are deprived of resources for answering them. (See Dummett, Truth and Other Enigmas [Oxford: Oxford University Press 1978]Google Scholar, and my The Nature of Mathematical Knowledge, 143.)

17 Fine has considerable fun with the idea of the ‘desk-thumping, foot-stamping shout’ that insists on the idea that the objects of commonsense and natural science REALLY exist (The Shaky Game, 129). Putnam attributes to his opponents appropriate distributions of capital letters (see ‘Realism and Reason; 123-40). Of course, it is incumbent on defenders of correspondence truth to do better than stamp their feet and sprinkle capital letters, and I shall try to show below how we can make some headway in articulating the intuitive feeling that something important is lost both in NOA and in Putnam’s internal realism.

18 I use the example of pictorial representations because it makes very immediate the idea of a relation of correspondence between representation and reality. However, it should not be thought that such representations have a special status because they lack the conventional characteristics of linguistic representations. To cite one important example, the collinearity of stations on the map of the London Underground does not correspond to the collinearity of the actual stations. (Despite this, the map has been used with enormous success by millions of people, and is, subject to its conventions, absolutely exact!) Thus I dissent in several ways from Paul Horwich’s attempt to show that a deflationary account of truth can accommodate the ‘correspondence intuition’ (see his Truth, 115-17).

19 This is very clear in the writings of Daniel Dennett (see, in particular, The Intentional Stance [Cambridge, MA: Bradford Books 1987]), but the fundamental conception goes back at least to Wilfrid Sellars’s seminal essay ‘Empiricism and the Philosophy of Mind’ (Ch. 5 of Science, Perception, and Reality [London: Routledge 1963]).

20 The point here is akin to one that is familiar from discussions of explanation and reduction. We cannot explain the systematic predominance of male births among humans by recounting the details of innumerable events of copulation, fertilization, and embryonic growth. For further discussion, see Sober, Elliott ‘Equilibrium Explanation,’ Philosophical Studies 43 (1983) 201-10CrossRefGoogle Scholar, and my ‘Explanatory Unification and the Causal Structure of the World,’ in Kitcher, P. and Salmon, W. eds., Scientific Explanation (Minneapolis: University of Minnesota Press 1990) 410-505Google Scholar, esp. 426-30. In the context of discussions of truth, the point was forcefully made by Field, Hartry ‘Tarski’s Theory of Truth,’ journal of Philosophy 69 (1972) 347-75CrossRefGoogle Scholar. Field denied that a mere list of assignments of referents to expressions was sufficient for the theory of reference, and, derivatively, for the theory of truth. Setting the theory of truth in the context of explaining successful behavior enables us to see that understanding systematic success demands more than simply conjoining claims like ‘Ophelia believes that the path leads to the willows and the path does lead to the willows.’ We need to identify the generic property — correspondence to reality — that is shared by the pertinent representations.

21 In arriving at this formulation, I have been helped by discussion with David Bakhurst.

22 The problems generated by Putnam’s ‘permutation’ arguments (’Realism and Reason,’ and Ch. 2 of Reason, Truth, and History) are deep and difficult, but I think that my approach to realism can at least help to make clearer what the realist wants to claim. Similarly, I believe that it enables us to scrutinize more thoroughly Fine’s charge that realism’s defense of abductive reasoning rests on a fallacy. Cashing these promissory notes must wait for other occasions.

23 Realists are often accused of having failed to understand the achievements and the problems of Kant’s epistemology. Realism’s WORLD is thus made to seem noumenal in the least charitable interpretation of that term. But if we keep our gaze firmly on the analogical move to realism, this interpretation cannot be sustained.

24 For an extremely lucid formulation of these points, see Fraassen, Bas van The Scientific Image (Oxford: Oxford University Press 1980), 39-40CrossRefGoogle Scholar.

25 For use of this example, see Barnes, Barry and Bloor, David ‘Relativism, Rationalism, and the Sociology of Knowledge,’ in Hollis, M. and Lukes, S. eds., Rationality and Relativism (Cambridge, MA: MIT Press 1982) 21-47Google Scholar, esp. 38 ff. Barnes and Bloor rely on Bulmer, R. ‘Why is the Cassowary not a Bird?’ (Man, new series 2 [1967] 5-25)CrossRefGoogle Scholar.

26 Strictly speaking, realists need not claim this much. Those with nominalistic inclinations in systematics may want to contend for the independent existence of the individual entities classified (the organisms) while denying that there is an objectively correct way of assorting them. But most contemporary systematists would want to go further.

27 The strategy I deploy here derives from the seminal work of Saul Kripke and Keith Donnellan on reference. I discuss the case of Priestley at some length in ‘Theories, Theorists, and Theoretical Change,’ and elaborate the account of Fresnel in chapter 5 of The Advancement of Science.

28 This is a point that has been emphasized by Dudley Shapere. See his Reason and the Growth of Knowledge (Dordrecht: Reidel 1984).

29 See, for example, Quine, W.V. ‘Natural Kinds,’ in Ontological Relativity and Other Essays (New York: Columbia University Press 1970)Google Scholar, and Rescher, Nicholas A Useful Inheritance: Evolutionary Aspects of the Theory of Knowledge (New York: Rowman and Allanheld 1989)Google Scholar. A somewhat more extensive presentation of the argument of the text is given in my ‘The Naturalists Return,’ esp. 101), and a full-dress version in Stich, Stephen The Fragmentation of Reason (Cambridge, MA: Bradford Books 1990)Google Scholar.

30 A watershed article that summarizes a wealth of problematic instances is Shapin’s, Steven ‘History of Science and its Social Reconstructions,’ History of Science 20 (1982) 157-211CrossRefGoogle Scholar.

31 Once again, I only indicate the outline of a line of argument. For far more detail — but still not enough — see Chapters 6 and 7 of The Advancement of Science.