[1] See, e.g., Alvin Plantinga, Warrant: The Current Debate (New York: Oxford University Press, 1993), Ch. 1.

[2] William P. Alston, "Epistemic Desiderata," Philosophy and Phenomenological Research 53 (1993): 527-51.

[3] Cf., e.g., Alvin I. Goldman, Epistemology and Cognition (Cambridge, MA: Harvard University Press, 1986), p. 118.

[4] Here one might object that reliabilism does not claim that the premises of the above-described argument support reliabilist-justified beliefs; but it seems I am saying that reliabilism claims just that. This objection represents a misunderstanding of my view. My view is that, however a reliabilist might want to characterize the supports relation that is alleged to hold between a reliable process and the resulting belief, and whatever precisely the reliabilist might want to say is doing the supporting, it must be the case that a description of the thing doing the supporting must logically support the proposition believed. Unless that's the case, the reliabilist has not offered an account of epistemic support, the kind of support that properly informs an account of epistemic justification. This seems to me to be a quite minimal constraint; a stronger constraint, one that would pose serious problems for both foundationalism and reliabilism, would have it that the supporters themselves must logically support the justified beliefs. For further discussion, please see the discussion of the "supports constraint" in Ch. 2, Sect. VII.

[5] Herbert Feigl, "De Principiis Non Disputandum...? On the Meaning and the Limits of Justification," in Max Black, ed., Philosophical Analysis (Ithaca, NY: Cornell University Press, 1950), pp. 119-56.

[6] Nicholas Rescher, in Methodological Pragmatism (Oxford: Basil Blackwell, 1977), offers a solution to the problem along these lines (see p. 23f).

[7] Ibid., p. 122.

[8] Though in fact theoretically inclined scientists and lawyers are concerned about, if not defending their standards in a way that might satisfy philosophers, at least clearly enunciating their standards of theoretical acceptability, or of evidence. In such clarification of standards there is a process of weeding out the bad standards that lends some positive epistemic value to those that are left.

[9] Apropos this claim: Ernest Sosa proposes what he calls an "argumentative account of justification," according to which for a belief to be justified is for a subject to justify, or have justified, it, which means offering "considerations or reasons in its favor." Though one would think the point deserves more discussion, he argues rather briefly that this "dictionary definition" of `justification' should be adopted, and so epistemologists should let that word go; instead we should talk in terms of `appropriate' and `apt' beliefs, which have positive epistemic value without reasons, justifications, etc. The implication for my project is that, while one might concede (1a) and (2a), one need not concern oneself with the Problem of Meta-Justification at all. This is an interesting proposal; I will advance a similar proposal in Ch. 4. See "Methodology and Apt Belief" and "Equilibrium in Coherence?" in Sosa, Knowledge in Perspective: Selected Essays in Epistemology (Cambridge: Cambridge University Press, 1991), pp. 253ff, 260f.

[10] Cf. William P. Alston, "Epistemic Circularity," in Epistemic Justification: Essays in the Theory of Knowledge (Ithaca, NY: Cornell University Press, 1989), p. 327; and elsewhere in Alston's recent work. The only significant difference between this formulation and Alston's is that the argument form above has a consequent in terms of p being justified for S, whereas in Alston's formulation p is said to be true. (So each (nb) would read, "pn is true at tn.") As we shall see shortly, this is ultimately an unimportant difference.

[11] One solution to the PMJ (viz., particularism ) rejects this demand. That solution will be evaluated (and rejected) in due course.

[12] Obviously, a more sophisticated reply would be to specify a standard which has it that if some non-(na) related condition is met, then (nb) holds. This sort of reply will be very important in coming to grips with the PMJ, but for purposes of simplicity epistemic circularity may most easily be introduced via the simpleminded reply; no questions will be begged thereby.

[13] Though as we shall see, some solutions to the PMJ (viz., meta-coherentism and regressism) attempt to do something like this.

[14] Cf. Ch. 2, Sect. VI ("The Arbitrary Case Argument"). Moreover, the conclusion would follow by inductive generalization from all the (BP)'s taken together; so the other premises would not be needed.

[15] "Truth-linked" is Goldman's term; see Alvin I. Goldman, Epistemology and Cognition (Cambridge, MA: Harvard University Press, 1986), p. 3 and elsewhere. Epistemic qualities are "truth-linked" when they are either necessary or sufficient conditions for truth (or probable truth). Goldman is, of course, not the only epistemologist who speaks of truth-linked qualities (or standards, which may also be called `truth-linked'). Laurence Bonjour, for example, says: "The basic role of justification is that of a means to truth, a more directly attainable mediating link between our subjective starting point and our objective goal" (The Structure of Empirical Knowledge [Cambridge, MA: Harvard University Press, 1985], p. 7). More generally, at least amongst realists not of the Putnam variety, most epistemologists do speak of central epistemic properties as truth-linked (which, it is often said, explains the desirability of those properties).

[16] And some have suggested getting rid of the term `circularity' in this context; e.g., Sosa: "Admittedly the circle involved is not the ordinary sort where we reach the conclusion only by circling back to the premises, the conclusion itself lying among the premises. Our inductive justification of the reliability of a faculty P is more like a spiral than a circle" ("The Coherence of Virtue and the Virtue of Coherence," in Knowledge in Perspective, op. cit., p. 197).

[17] In "Reliability, Justification, and the Problem of Induction," Midwest Studies in Philosophy, IX (1984): pp. 555-67; esp. p. 558.

[18] Ibid., p. 558.

[19] William P. Alston, The Reliability of Sense Perception (Ithaca and London: Cornell University Press, 1993), p. 15.

[20] Cf. Wesley C. Salmon, The Foundations of Scientific Inference (Pittsburgh: University of Pittsburgh Press, 1966), p. 15.

[21] Van Cleve, ibid. In any case, however, rule circularity is clearly a different concept from epistemic circularity.

[22] For a favorable discussion of requiring justified belief that "the premises are `properly connected' to the conclusion," for good (justified belief-conferring) arguments, see Richard Feldman, "Good Arguments," in Frederick F. Schmitt, ed., Socializing Epistemology: The Social Dimensions of Knowledge (Lanham, MD: Rowman & Littlefield, 1994), pp. 159-188.

[23] One might well imagine a third account of epistemic circularity, on which the necessary condition is this: one must already have justified belief in the conclusion in order to be justified in believing the inference from the premises to the conclusion. Notice that this might be hard to distinguish from rule circularity.

[24] Note that this is not to require awareness or knowledge of any inference rule licensing the inference.

[25] In other words, one's belief in the conclusion is supposed to be based on one's belief in the premises (at least). What precisely this "basing" amounts to is a topic I will not broach here.

[26] Van Cleve, ibid.

[27] On what exactly this term `vicious' means, see the end of the present section.

[28] Another example:

John has a justified belief.
Therefore, someone has a justified belief.

[29] Notice, again, that nothing in this discussion depends on what my premises are; so it doesn't matter whether I assert, dogmatically, that John has a belief or, absurdly, that the moon is made of green cheese.

[30] Op. cit.

[31] It is also worth pointing out that the arguments in Ch. 2 (which are intended to show that epistemic circularity is vicious ) might apply just as well to arguments of form (A); this would require some demonstration, but if correct, it is further evidence that arguments of form (A) ought to be considered epistemically circular.

[32] Raised by Marshall Swain in communication.

[33] Perhaps the epistemic circularity of arguments for such claims could be used to support the view that they are principles of common sense and not in need of argument; cf. Ch. 4, Sect. III. This is an interesting prospect I regret not having the time to explore here.

[34] For an explanation of the meaning of `doxastic practice', see Ch. 4, Sect. II.

[35] But note that I will continue to use `support' when applied to ordinary beliefs in the ordinary sense.

[36] See, for example, James W. Cornman's definition of `self-justifying': "Foundational versus Nonfoundational Theories of Empirical Justification," in George S. Pappas and Marshall Swain, eds., Essays on Knowledge and Justification (Ithaca, NY: Cornell University Press, 1978), p. 230.

[37] Descartes' criterion of truth may be better regarded as a "mutually supported" standard: see just below.

[38] Notice that we might get a start on pinning down the notion of properly basic j-standards by substituting "need not be" for "is not" here.

[39] Thus Sextus Empiricus: "[W]e do not allow them to adopt a criterion by assumption, while if they offer to judge the criterion by a criterion we force them to a regress ad infinitum" (Outlines of Pyrrhonism II.20, trans. R. G. Bury [Buffalo, NY: Prometheus Books, 1990], p. 101).

[40] The standard in virtue of which the premise(s) are justified probably cannot be the same as the standard in virtue of which S is justified in inferring the conclusion from the premises. But the minimum number of licensing standards has no bearing on explaining the Meta-Regress Argument. The claim that there must be some licensing standards is one that will be examined in Ch. 3.

[41] This assumes that this is an exhaustive list of possibilities generated by (1) and (2): (3a), (3b), (4a), (4b), and (4c). The assumption seems plausible in light of common experience with regress arguments.

[42] See Roderick M. Chisholm, "The Problem of the Criterion," in The Foundations of Knowing (Minneapolis: University of Minnesota Press, 1982), p. 66.

[43] See esp. "Methodology and Apt Belief," in Knowledge in Perspective, op. cit., pp. 248ff.

[44] Robert Nozick, Philosophical Explanations (Cambridge, MA: Harvard University Press, 1981), pp. 278-9.

[45] Reprinted in William P. Alston, Epistemic Justification: Essays in the Theory of Knowledge (Ithaca and London: Cornell University Press, 1989).

[46] Reliability, op. cit.,p. 17.

[47] See Alston's "Belief-forming Practices and the Social," in Frederick F. Schmitt, ed., Socializing Epistemology: The Social Dimensions of Knowledge (Lanham, MD: Rowman & Littlefield, 1994), p. 40. This point was brought out 36 years earlier by Henry E. Kyburg, Jr., in his reply to Braithwaite's attempt (to be discussed anon) to give an inductive argument for induction. Comparing induction with "the Oblonsky Method of tea-leaf reading," Kyburg says, "Both methods are in the same logical boat." See "R. B. Braithwaite on Probability and Induction," British Journal for the Philosophy of Science 11 (Nov. 1958): 203-20, esp. 207-8.

[48] Ibid., p. 43.

[49] Reliability, op. cit., p. 138f., and "Belief-forming Practices," op. cit., pp. 40-1 and 43.

[50] Richard B. Braithwaite, Scientific Explanation: A Study of the Function of Theory, Probability and Law in Science (Cambridge: Cambridge University Press, 1955); see esp. Chapter VIII.

[51] Ibid., pp. 259-61.

[52] In a passage that is very interestingly similar to the MRA; ibid., p. 274.

[53] It is induction by simple enumeration that will take us from the claim that a policy has been effective in the past to the claim that it is effective, period.

[54] Ibid., p. 275.

[55] Ibid.

[56] Viz., if we must regard a rule licensing any inference at all as among the premises of the argument, we may as well add the rule explicitly to those premises; but then there would have to be a second rule that licenses the original premises plus the first rule. So the second rule is added to the premises; but then there would have to be a third rule, and so on.

[57] Ibid., p. 277.

[58] Ibid., p. 281.

[59] Ibid., p. 283.

[60] Ibid., p. 282.

[61] Ibid., p. 290.

[62] As Black says, "Braithwaite is one of the very few writers who do not think that inductive justifications of induction are viciously circular. But I think he is unnecessarily restrained in his conclusions on the matter and my own treatment follows somewhat different lines." Max Black, "The Inductive Support of Inductive Rules," in Problems of Analysis: Philosophical Essays (Ithaca: Cornell University Press, 1954), p. 201n.

[63] Ibid., p. 197.

[64] Ibid., p. 196.

[65] Ibid.

[66] Ibid., p. 198.

[67] Ibid.

[68] Ibid., p. 204.

[69] Ibid.

[70] Ibid., pp. 204-5. Cf. Max Black, "Self-supporting Inductive Arguments," in Models and Metaphors: Studies in Language and Philosophy (Ithaca, NY: Cornell University Press, 1962), p. 209ff. Black writes: "I have no intention of claiming that the self-supporting argument can definitively establish or demonstrate that the rule is correct. Indeed, I do not know what an outright demonstration of the correctness or legitimacy of an inductive rule would be like" (p. 212).

[71] Cf. also Nozick, Philosophical Explanations, ibid., pp. 272-3, and Nicholas Rescher, Methodological Pragmatism: A Systems-Theoretic Approach (Oxford: Basil Blackwell, 1977), p. 28 and pp. 113-25, for similar sorts of "bootstrapping" proposals. As Rescher says, "Any experiential justification of a truth-criterion must pull itself up by its own bootstraps..." (p. 28).

[72] Ibid., p. 206. On p. 202, Black does acknowledge that "it is conceivable that circularity of a kind more subtle than any we have yet considered might arise in the satisfaction of the epistemic conditions for the legitimate use of our self-supporting arguments."

[73] In "Reliability, Justification, and the Problem of Induction," Midwest Studies in Philosophy IX (1984): 555-67. See esp. p. 558.

[74] Ernest Sosa has named a way of opting out of the circularity problem, which involves quelling this inquisitiveness: the avoidance strategy, meaning avoidance of the question of whether a given "faculty" is reliable. Sosa himself concludes that this strategy is not necessary to deal with the problem; but I simply think that it is impossible for actual philosophers to achieve. As much as they might talk about avoiding the question, their colleagues and their consciences will not let them. Moreover, as Sosa himself implies, this is a non-philosophical response to the problem and as such does not merit consideration – any more than "Up yours!" would merit discussion as a response to "How do you know that p?" See Ernest Sosa, "Reflective Knowledge in the Best Circles," Journal of Philosophy 94 (1997): 410-30; see esp. pp. 424-29.

[75] See, e.g., "Epistemic Circularity," op. cit., pp. 338-48.

[76] Op. cit., p. 342.

[77] Marshall Swain uses a term that might be more apt: `stopping-point'. In calling the claims in question `starting-points', I do not mean to imply that we actually begin philosophical inquiry or exposition with such claims.

[78] Though it will arise again in a different form in Ch. 4, when I advance my own view as to what one proper meta-epistemological starting-point is.

[79] Or, you might want to say, it does so openly. In Ch. 4, Sect. V, however, I argue that this sort of move cannot properly be described as "begging the question."

[80] Reliability, op. cit., p. 17.

[81] Ibid.

[82] Ibid.

[83] And apparently contradictory. Regardless of the reason, either all epistemically circular arguments are disqualified because they are epistemically circular, or some of them aren't. Statement (1) does not actually state that some epistemically circular arguments are acceptable, however. In any event, what follows will not turn on any claim that Alston held inconsistent views.

[84] "Belief-forming Practices and the Social," op. cit.

[85] Ibid., p. 43.

[86] A very similar reply to Sosa is elaborated below.

[87] Cf. Reliability, op. cit., p. 138.

[88] I.e., a virtuously circular argument would be one supporting a positive claim about the reliability of some obviously reliable practice, while a viciously circular argument would be one supporting a similar claim about an obviously unreliable practice. Of course, if we aren't sure about the reliability of a practice, we won't know whether the circularity of the argument is vicious or virtuous.

[89] Ernest Sosa, "Reflective Knowledge in the Best Circles," op. cit., p. 427.

[90] That this sort of move would itself result in a circularity is something that Alston foresaw: "Since even significant self-support exhibits epistemic circularity, I will refrain from taking it to be an independent reason for supposing the doxastic practice in question to be reliable. Because self-support requires assuming the practice in question to be a reliable source of belief, it provides evidence for reliability only on the assumption of that reliability; and that is hardly evidence in any straightforward sense" (Reliability, op. cit., p. 139).

[91] Op. cit., p. 428.

[92] Ibid., pp. 428-9.

[93] In other words, while many (e.g., philosophers of common sense ) would not have any quibbles with the basic notion of simply assuming that sense perception is reliable (although this claim would have to be refined), it is not nearly as uncontroversial or common-sensical to take for granted that epistemic circularity is not always vicious.

[94] A similar argument may be constructed using an Alston -style track record argument, where the two premises considered would be: (1a') S's belief that p1 meets conditions c at time t1; and (1b') p1 is true at t1. The conclusion would be modified in the obvious way.

[95] Interestingly enough, it is Sosa who, in another article, suggests this very sort of argument: "AR [an epistemically circular track record argument ] does seem anyhow viciously spiral, since it could only succeed as a sound and rationally persuasive inductive argument through the immediately patent existence of a deductive argument to the same conclusion, one that demands nothing more as premises than the inductive argument requires as premises or presuppositions (implicit assumptions) for its successful use as a means to rational persuasion" ("The Coherence of Virtue and the Virtue of Coherence," in Knowledge in Perspective: Selected Essays in Epistemology[Cambridge: Cambridge University Press, 1991], p. 202).

[96] Nonetheless, some argument for the even more basic claim, that justified beliefs necessarily have justifiers, is given below in Ch. 3, Sect. VII, "Strangely Justified Beliefs."

[97] More precisely, the contents of a belief cannot render that same belief justified by standing in the relation of support to itself.

[98] This is similar to the problem that the foundationalist faces when asked, "We understand what it means for one belief to support another; but what does it mean for an experience or raw feel or what have you to support a belief?"

[99] Ernest Sosa, "The Coherence of Virtue and the Virtue of Coherence," Knowledge in Perspective: Selected Essays in Epistemology (Cambridge: Cambridge University Press, 1991), p. 202.

[100] See, e.g., Laurence Bonjour, The Structure of Empirical Knowledge (Cambridge, MA: Harvard University Press, 1985), pp. 93-101.

[101] Cf. what Paul Moser says about his own project of meta-justification. He says it "resembles the sort of justification for physical-object proposition characterized in §3.2 [roughly, an explanatory coherence account]. In both cases justification derives from propositions' having a sort of maximal explanatory power for one relative to a data base and to a set of competing propositions. Such an analogy between justification of physical-object propositions and meta-justification for epistemic principles is highly desirable; for both are, after all, species of epistemic justification" (Knowledge and Evidence [Cambridge: Cambridge University Press, 1989], pp. 261-2). This is not to deny that it is possible to accept coherentism and also reject meta-coherentism as I have defined it. Bonjour's view in The Structure of Empirical Knowledge, op. cit., may be an example.

[102] This precise point is not original. Cf. Sosa, "The Coherence of Virtue," op. cit., pp. 201-207, and Robert P. Amico, The Problem of the Criterion  (Lanham, MD: Rowman & Littlefield, 1993), pp. 66-9.

[103] William P. Alston, The Reliability of Sense Perception (Ithaca, NY: Cornell University Press, 1993), pp. 20-1; cf. p. 122.

[104] And if I am not mistaken this is basically what both W. V. Quine and Paul Moser have tried to do, although perhaps not specifically in reaction to anything like the Meta-Regress Argument.

[105] William P. Alston, Perceiving God: The Epistemology of Religious Experience (Ithaca, NY: Cornell University Press, 1991), p. 149.

[106] Nelson Goodman, Fact, Fiction, and Forecast, 4^th ed. (Cambridge, MA: Harvard University Press, 1983), p. 64. For argument that the Method of Reflective Equilibrium, on some construals, does involve epistemic circularity, see below.

[107] Nicholas Rescher, Methodological Pragmatism: A Systems-Theoretic Approach to the Theory of Knowledge (Oxford: Basil Blackwell, 1977), p. 103. Whether such methods of theory confirmation are usefully construed as theories of justification will receive due attention below.

[108] This claim will receive some support below. Moser, op. cit., p. 264, is one of the few people who uses something like a meta-coherentist justification of epistemic standards, and yet denies that it involves any circularity. He writes: "Explanatory particularism also avoids the problem of circularity by denying that we can justify our epistemic principles solely by what those principles countenance as knowledge or justification. On this method, the role of epistemic intuitions in meta-justification frees us from such circularity." As far as I can make out, this is a simple non sequitur. To this Amico, op. cit., pp. 100-1, effectively replies that Moser's approach is essentially the Method of Reflective Equilibrium applied to epistemic claims, and that this method requires further justification, which justification presumably leads either to regress or circularity.

[109] Or in terms of doxastic rather than propositional justification: If S believed that J1 on the basis of beliefs in each of the premises of Ai and that each of the component inferences involved in Ai is correct (and perhaps some other conditions were met), then S would be justified in believing that J1.

[110] One might well observe that Alston's "significant self-support" defense of epistemic circularity against the equal availability argument, discussed in Ch. 2, Sect. V, essentially involves arguing that the coherence of a doxastic practice's outputs does, in fact, mitigate the viciousness of epistemic circularity. Hence the following discussion may be regarded as a continuation of the considerations brought out in Chap. 2, Sect. V.

[111] John Rawls, A Theory of Justice (Cambridge, MA: Harvard University Press, 1971), p. 20.

[112] For discussion of the distinction between `narrow' and `wide' reflective equilibrium, see Norman Daniels, "Wide Reflective Equilibrium and Theory Acceptance in Ethics," Journal of Philosophy 76 (1979): 256-82.

[113] Op. cit., pp. 63-4. All italics in original.

[114] Moser, op. cit., p. 261.

[115] In a related context, Susan Haack draws a similar distinction: "I do not, of course, deny the genetic point, that the codification of valid forms of inference, the construction of a formal system, may proceed in part via generalisation over cases – though in part, I think, the procedure may also go in the opposite direction.... But I do claim that the justification of a form of inference cannot derive from intuition of the validity of its instances" ("The Justification of Deduction," Mind 85 [1976]: 118). Cf. also Daniels, op. cit., p. 259n.

[116] Those making this mistake include R. M. Hare, "Rawls' Theory of Justice – I," Philosophical Quarterly 23 (1973): 144-55; and Peter Singer, "Sidgwick and Reflective Equilibrium," The Monist 58 (1974): 490-517. Apparently some have thought MRE is committed to foundationalism based on their (false) impression that MRE involves the positing of some epistemically privileged set of beliefs (probably intuitions about particular cases, rather than principles), from which the epistemically less-privileged principles ultimately get their warrant. On that, see Michael R. DePaul, "Reflective Equilibrium and Foundationalism," American Philosophical Quarterly 23 (1986): 59-69, as well as Daniels, op. cit., pp. 264ff.

[117] DePaul, op. cit., p. 58.

[118] "Equilibrium in Coherence?" Knowledge in Perspective, op. cit., p. 262. Read literally, this is of course just nonsense. Wide equilibrium is a state of mind; pure coherentism is a theory of justification. On the one hand, obviously, Sosa does not mean that the state of mind "seems equivalent to" the theory; but on the other hand, it isn't clear what Sosa does mean, exactly.

[119] Op. cit., p. 59.

[120] Sosa, "Methodology and Apt Belief," in Knowledge in Perspective, op. cit., p. 247.

[121] Op. cit., p. 64.

[122] Goodman is not alone here; a very similar sort of hand-waving not uncommonly backs up otherwise very rigorous, complex theories of justification. For another example, see Robert Nozick, Philosophical Explanations (Cambridge, MA: Belknap Harvard, 1981), pp. 278-9, quoted above in Chap. 2, Sect. I.

[123] Rescher, Methodological Pragmatism, op. cit., p. 25ff. Rescher sometimes avoids the term "circle," preferring "cyclical" just as Sosa sometimes prefers the term "spiral." Very roughly, Rescher's procedure is to propose a method of inquiry, derive some alleged truths by its use, act on those truths, and then determine whether the action was successful; then, based on the success or failure of the action, the method is altered accordingly and the procedure is repeated. Since the method is evaluated according to the success of actions, this is supposed to be a "pragmatic" or "instrumental" justification.

[124] Ibid., pp. 101-2.

[125] Ibid., p. 103.

[126] Ibid.

[127] Ibid, p. 104.

[128] Thanks to Joe Salerno for this line of argument.

[129] It is a bit misleading, though common, to speak of the "circle" as large. Those steps of an argument that make the argument circular need not be very numerous in order for the "circle" to be considered "large." Loosely speaking, an entire argument might be said to be, or to contain, a "large" circle if the argument itself were quite large and complex, and the elements that make it circular are a relatively few in number. So the claim under examination in this section is that, if epistemically circular reasoning is embedded within a large, complex argument (a "system"), then the circularity is, or may be, virtuous.

[130] The following elaborates on Sosa, Knowledge in Perspective, op. cit., pp. 196, 202.

[131] The difference at issue between a "tight" and "large" circle is reflected in the difference between narrow and wide reflective equilibrium. An objection to the method of narrow reflective equilibrium that it is too conservative, and thus unreliable – is analogous to the objection to the first Podunk News argument, or the "tightly circular" track record argument for the reliability of sense-perception. Correspondingly, the proposal to widen MRE to include a broad variety of background assumptions (and thus hopefully to escape charges of conservativism) has as its analog the proposal to allow "large" circles as virtuous.

[132] Sosa, Knowledge in Perspective, op. cit., p. 202.

[133] Sextus Empiricus, Outlines of Pyrrhonism II.20., trans. R. G. Bury (Buffalo, NY: Prometheus Books, 1990).

[134] Roderick M. Chisholm, "The Problem of the Criterion," in The Foundations of Knowing (Minneapolis, MN: University of Minnesota Press, 1982), pp. 61-75.

[135] The following is very loosely adapted from "The Problem of the Criterion," ibid.

[136] The associated doctrine would be called "particularism" as well, viz.: that it is best to use instances of knowledge to arrive at a criterion of knowledge, without antecedently making reference to any such criterion to choose the instances.

[137] Again, "methodism" may also mean something like this: it is best to use an antecedently decided-upon criterion of knowledge to determine particular instances of knowledge, without supporting the criterion with any such instances.

[138] Chisholm, op. cit.

[139] The notion of explanation at work here will receive heavy use later on. Evidently, what is meant is not scientific or causal explanation, but meaning explanation. Nonetheless, the term `explanation' is apropos just considering the procedure required to produce the needed account, namely, abduction, also known as argument to the best explanation.

[140] G. E. Moore, "Proof of an External World," in Philosophical Papers (New York: Collier Books, 1959).

[141] Perhaps this standard would itself be derived from a more general (and perforce even more complex) standard about vision, or sense-perception. Plainly, one issue in deciding whether a certain standard "licenses" an epistemic claim is at what level of specificity the standard is to be formulated. A similar issue is much-discussed by reliabilists under the heading of "the generality problem."

[142] One could make a case, as well, that the particularist cannot posit the existence of justifiers, such as Moore's seeing his hand; that is, though of course Moore does see his hand, his seeing it does not stand in a relation of justification to his belief that he expresses by saying, "Here is one hand." After all, suppose that the strict particularist did want to say that Moore's seeing his hand justified his claim; then he would have to say that Moore's claim has a justifier, but there is no applicable standard of justification. That's at best a paradoxical conjunction of claims.

[143] Not coincidentally, Thomas Reid made just this criticism of Descartes' cogito. See Reid's Inquiry into the Human Mind, in William Hamilton, ed., Philosophical Works, Vol. I, 8e (Hildesheim, Germany: Georg Olms Verlagsbuchhandlung, 1967), Ch. I, Sect. III; p. 100.

[144] And in fact I have my own theory as to why certain terms ought, in philosophy, to be taken as primitive, but it would take us too far afield here to present it. Suffice it to say that such theories are possible, even if (rather surprisingly) they have not been proposed very often.

[145] John Locke, An Essay concerning Human Understanding, ed. Peter H. Nidditch, (Oxford: Clarendon Press, 1975), p. 591ff.

[146] Note that the two forms considered here are nonmodal and restricted to justification. For a discussion of various forms of epistemological skepticism, see George S. Pappas, "Some Forms of Epistemological Scepticism," in Pappas and Marshall Swain, eds., Essays on Knowledge and Justification (Ithaca, NY: Cornell University Press, 1978), pp. 309-16.

[147] Sextus Empiricus, Outlines of Pyrrhonism II.20, trans. R. G. Bury (Buffalo, NY: Prometheus Books, 1990).

[148] Reprinted in William P. Alston, Epistemic Justification: Essays in the Theory of Knowledge (Ithaca and London: Cornell University Press, 1989).

[149] There is a symbolic approximation of this proposal: $p(~sJBp & ~sJB~p). This is approximate because all or nearly all philosophers will, of course, say that there are some beliefs such that neither accepting or denying them is justified. In Chisholm's lingo, such beliefs should be withheld. So what the symbols leave out is the claim that it is a category mistake to apply the concept of justification to certain beliefs at all.

[150] This proposal will be elaborated in Section VI below.

[151] Or warranted, etc. As mentioned above, a number of different terms have been used to indicate some sort of non-deontological positive epistemic status. Perhaps the particular term of art here does not matter, although I will prefer "rational" in this discussion for reasons that will become clear.

[152] Or, alternatively, the source is constituted as rational on account of the rationality of the beliefs in which it issues. A topic we shall encounter below is the interdefinability of the rationality of doxastic practices and of beliefs.

[153] See Richard Foley, The Theory of Epistemic Rationality (Cambridge, MA: Harvard University Press, 1987).

[154] Ibid., p. 14.

[155] Cf. Alvin Plantinga, Warrant: The Current Debate (Oxford: Oxford University Press, 1993), p. 132ff and esp. 134-5. This theory of rationality borrows some ideas from Plantinga's account of warrant.

[156] I adopt here Alston's useful term; see, e.g., this gloss: "A prominent member of this group [of doxastic, i.e., belief-forming, practices] is the practice of going from sense experience (together, sometimes, with relevant background beliefs) to beliefs about things, events, and states of affairs in the immediate physical and social environment. … Clearly, what I am calling a `doxastic practice' is not a single belief-forming disposition, but some family, grouping, or system of individual dispositions, bound together in some important way. What binds the components together in the practice is some marked similarity in input, output, and/or function" (The Reliability of Sense Perception [Ithaca, NY: Cornell University Press, 1993], pp. 7-8).

[157] More than this can and will be said about the relation between `rationality' and `rational belief'.

[158] Supplanted, no doubt, by the concept of means-end rationality, which in epistemology is not particularly helpful by itself. A belief would be epistemically rational in this sense iff adopting the belief would be an adequate means to securing the end of truth. This account by itself is obviously inadequate, leaving as it does unanswered the question: "When is the adoption of a belief an adequate means to securing the end of truth?" Even as an account of the rationality of doxastic practices a doxastic practice is rational iff it is an effective means to getting true beliefs – the means-end concept is not very helpful. Here the crucial unanswered question is: "When is a doxastic practice an effective means to getting true beliefs?" Nonetheless, as we shall see below, a complete theory of epistemic rationality ought to be able to be construed as an answer to these questions; so one may regard the means-end conception of rationality as supplying an important desideratum against which different theories of epistemic rationality might be applied.

[159] See Thomas Reid, Essays on the Intellectual Powers of Man, in William Hamilton, ed., Philosophical Works, Vol. I, 8e (Hildesheim, Germany: Georg Olms Verlagsbuchhandlung, 1967), Essay VI. For evidence that Reid had the broad conception of reason discussed here, see p. 425: "It is absurd to conceive that there can be any opposition between reason and common sense. It [common sense] is indeed the first-born of Reason; and, as they are commonly joined together in speech and in writing, they are inseparable in their nature. We ascribe to reason two offices, or two degrees. The first is to judge of things self-evident; the second to draw conclusions that are not self-evident from those that are."

[160] But in the case of beliefs formed by good reasoning, then of course in order for those beliefs to be rational in the present sense, they would have to be supported, and it would be partly in virtue of such support that they would indeed be rational. The point here however is that there are some beliefs that can be rational – and thus have positive epistemic status, and thus might support nonbasic, justified beliefs – without themselves being supported by other beliefs.

[161] Of course, it may be possible to think up strange circumstances in which even these practices would be rational; but, obviously, such practices are usually irrational.

[162] Cf. Alston, Reliability, op. cit., pp. 138-40.

[163] "Def. An argument A for conclusion c (understood by S) is epistemically circular for S iff (i) if S were justified in believing that c on the basis of the set of S's beliefs about A, then S would be justified in believing each of A's premises and that A's inference is correct (= justifying beliefs about A); (ii) for at least one of the justifying beliefs about A, if S were to have a justified belief in it, then (it follows that) c would be true; and (iii) c is either a justification standard or an assertion that some doxastic practice is reliable."

[164] That such attempts have indeed failed is the conclusion of Ch. 3 of Alston's Reliability of Sense Perception, op. cit. – excellent work that I need not replicate.

[165] Thomas Reid, An Inquiry into the Human Mind, in Philosophical Works, op. cit., Ch. VI, Sect. XXIV, pp. 194-201.

[166] See, for just one example, Richard Foley, "Egoism in Epistemology," in Frederick F. Schmitt, ed., Socializing Epistemology: The Social Dimensions of Knowledge (Lanham, MD: Rowman & Littlefield, 1994).

[167] Another admitted point of vagueness: it suffers from a similar sort of generality problem that faces reliabilism. There are countless different ways of describing very closely-related doxastic practices. For instance, no doubt there are dozens of plausible substitutes for BSP as a basic sense-perceptual practice.

[168] "p is certain for S = Df For every q, believing p is more justified for S than withholding q, and believing p is at least as justified for S as is believing q" (Roderick M. Chisholm, Theory of Knowledge 3^rd ed. [Englewood Cliffs, NJ: Prentice-Hall, 1989], p. 12).

[169] However, for reasons that will be made clear in later sections, I would define `certain' in terms of some variety of positive epistemic status other than justification.

[170] G.E. Moore, "A Defence of Common Sense," in Philosophical Papers (New York: Collier Books, 1959), pp. 32-59. See esp. pp. 33-4. His list begins: "There exists at present a living human body, which is my body. This body was born at a certain time in the past, and has existed continuously ever since, though not without undergoing changes; …" (p. 33).

[171] Thomas Reid, Essays, op. cit., pp. 434-41.

[172] Particularly in the text cited in footnote 26. Whether or not particular beliefs should be called `principles of common sense', it is clear that Reid regarded their truth as a matter of common sense: at op. cit., p. 425, he gives the name `common sense' to the first "office or degree of reason." The first "office of reason" comprises the faculties of judgment (or doxastic practices in Alstonian terminology) associated with perception, memory, and introspection, but not with reasoning (that's associated with the second degree of reason).

[173] Perhaps I could instead (though I will not) employ Richard Foley's account (in The Theory of Epistemic Rationality , op. cit.) of "propositions such that S's believing them gives him a reason, albeit not necessarily an indefeasible reason, to think that they are true" (p. 52), particularly because this is supposed to be an account of what makes basic beliefs rational.

[174] This example might prompt a question: is it not possible that there are some doxastic practices which are reliable, basic (not supportable except via self-support ), and which do not result in "obviously correct" beliefs? Perhaps "Christian Mystical Practice" as described by William P. Alston in Perceiving God: The Epistemology of Religious Experience (Ithaca, NY: Cornell University Press, 1991) could be cited as an example. I am willing to admit the mere possibility of such practices; at least, the topic bears discussion. And if so, can we bootstrap our way to a justification of the reliability of these doxastic practices (as I will propose, anon, that we can do for other doxastic practices)? Fortunately, the validity of the results of the present chapter do not appear to turn on that difficult question.

[175] Depending on the strength of the rationality asserted in the consequent of the standard, various standards may be associated with the same practice.

[176] P.F. Strawson, Introduction to Logical Theory (London: Methuen & Co., 1952), pp. 248-63.

[177] Ibid., p. 249.

[178] Ibid.

[179] Ibid., pp. 256-7

[180] Ibid., pp. 258-60.

[181] Wesley C. Salmon, "Should We Attempt to Justify Induction?" Philosophical Studies VIII (1957): 33-48.

[182] Ibid., p. 39.

[183] Cf. Nelson Goodman, Fact, Fiction, and Forecast, 4^th ed. (Cambridge, MA: Harvard University Press, 1983), pp. 64-5.

[184] Ibid.

[185] Again, see Nelson Goodman, ibid., on formulating deductive rules; cf. Max Black on bootstrapping inductive rules (as discussed above in Ch. 2, Sect. II).

[186] And obviously, I lack the space here to expound upon them.

[187] Obviously, I lack the space to engage any important issues surrounding the pragmatic view on this issue here; but if it should happen that a pragmatic view of truth should aid my reply to the meta-skeptic, so much the better for pragmatism and my solution.

[188] Thomas Reid, Essays, op. cit., Essay VI, Ch. V, advances as a "first principle of contingent truth" the following: "That the natural faculties, by which we distinguish truth from error, are not fallacious." (p. 447) (See this dissertation's frontispiece.) His brief commentary on this first principle (pp. 447-8) may be read as a summary of the views advanced in this chapter.

[189] I am afraid the rest of this chapter's discussion blurs a potentially important distinction: it is one thing to say that a given (properly basic) doxastic practice produces, more often than not, true beliefs; it is another to say that each belief formed by such a practice is probably true. Evidently, the latter is an inference I will have to infer from the former. The basic inference is reasonably straightforward: ceteris paribus, rational beliefs are probably true on the evidence that they are the results of rational practices. But, of course, there are other factors that might weigh into our assessment of the probability of any given belief. Fortunately, the success of the sort of arguments described in Section VII does not seem to require any more than a weak presumption in favor of the beliefs described as `rational'. So, while the present problem makes me nervous, it doesn't make me very nervous.

[190] Reliability, op. cit., pp. 123-4.

[191] Exactly how the declaration that I simply assume the PoR is correct constitutes a "reply" to meta-skepticism will be explored below.

[192] To be more precise, I claim that my belief (and anyone's so long as he is similarly positioned) in the PoR satisfies certain conditions in which it is impermissible to hold me to an obligation to provide an argument for the Principle. See just below.

[193] To be clear, again, I am not saying here specifically that one is justified in making such claims, but that one has no obligation to prove them in philosophical discourse.

[194] I owe this formulation of the point to George Schumm.

[195] This argument greatly resembles an argument Thomas Reid makes against an extreme skeptic: "Let scholastic sophisters entangle themselves in their own cobwebs; I am resolved to take my own existence, and the existence of other things, upon trust; … He must either be a fool, or want to make a fool of me, that would reason me out of my reason and senses. I confess I know not what a sceptic can answer to this, nor by what good argument he can plead even for a hearing; for either his reasoning is sophistry, and so deserves contempt; or there is no truth in human faculties – and then why should we reason?" (Reid, Inquiry, op. cit., p. 104) It also resembles later pragmatic arguments for the reliability of induction, e.g., Reichenbach's. See Wesley C. Salmon, The Foundations of Scientific Inference (Pittsburgh: University of Pittsburgh Press, 1967), pp. 52-4. However, I believe my argument differs from both of these arguments in the conclusion that is being urged.

[196] We might ascribe such a belief to most people de re, rather than de dicto. Most people have few beliefs about rationality per se.

[197] Cf. Reid, Essays, op. cit., IV.II, p. 425.

[198] Again, cf. Reid, ibid.

[199] Reprinted in Epistemic Justification, op. cit., pp. 115-52.

[200] Ibid., p. 152.

[201] Op. cit., ch. 1.

[202] A sense that might require considerable work to tease out.

[203] See, for one example, Richard Swinburne, Faith and Reason (Oxford: Clarendon Press, 1989), p. 25. For another, see Alston, ibid., p. 152, and also Reliability, op. cit., pp. 120-1.

[204] Ludwig Wittgenstein, On Certainty, trans. Denis Paul and G.E.M. Anscombe, ed. G.E.M. Anscombe and G.H. von Wright (New York: Harper & Row, 1969). The proposition in question identifies a phenomenon that exercises Wittgenstein considerably in this, his last work.

[205] David Hume, A Treatise of Human Nature, 2nd ed., ed. L.A. Selby-Bigge and P.H. Nidditch (Oxford: Clarendon Press, 1978), I.IV.VII.

[206] Indeed, this formed the basis of Wittgenstein's critique of skepticism in On Certainty.

[207] On these latter points I may be differing with Wittgenstein, who doubted that it makes any sense at all to claim to doubt certain propositions. He appears to have held, in On Certainty, that a claim to deny some propositions should be taken as evidence that the person trying to make the denial doesn't even understand the proposition.

[208] Again I am in agreement with Alston, and also with Swinburne, op. cit., who writes: "It is because one set of inductive standards seem to me intuitively right and my use of them is not under the control of my will, that I trust that the resultant beliefs indicate how things are." (p. 26)

[209] One might wish to insist that there is no essential difference between these classes of beliefs: what it means to say that the belief that these doxastic practices are reliable is irresistible is precisely for the individual beliefs that result from these practices to be irresistible. I am not sure if anything important rests on drawing a distinction here (or failing to do so).

[210] But not legal responsibility.

[211] The point could perhaps be made simply by saying that it is false (rather than merely "odd," or "nonsense," or "wrong," etc.) to say either that S is, or that S is not, permitted to believe that p. In this case, one would have to be clear that the fact that it is false to say that a belief is obligatory does not imply that it is permitted. Exactly how to characterize the mistake – is it nonsense, merely "odd" or "strange,"false, etc.? – seems to make little difference. But for purposes of consistency, I will say my thesis is about a `category error': it is a category error to ascribe categories of permission and obligation to acts for which we are not responsible.

[212] Cf. Alston, op. cit., p. 120.

[213] See Ch. 1, Sect. I above, as well as Alston, op. cit., for further discussion of this point.

[214] See Alvin Plantinga, Warrant: The Current Debate, op. cit., ch. 1.

[215] Op. cit. and see Alston, "Epistemic Desiderata," Philosophy and Phenomenological Research 53 (1993): 527-51.

[216] Op. cit., p. 3e.

[217] Op. cit., p. 22e.

[218] Op. cit., p. 34e.

[219] Thus Foley: "[D]oes saying that from an epistemic point of view it is epistemically rational for a person S to believe the conclusions of arguments that are uncontroversial for him presuppose that S either now has or did have some kind of control over what he believes? I will argue in chapter 5 that although people ordinarily do have at least some kind of indirect control over what they believe, the answer to this question nonetheless is no. It can be epistemically rational for a person S to believe even that which, given his circumstances or given his limitations as a believer, he cannot believe. It also can be epistemically rational for S to believe that which, given his circumstances, or given his limitations as a believer, he cannot help but believe. Of course, in such cases S should not be blamed or praised for believing what he does" (Theory of Epistemic Rationality , op. cit., pp. 12-3).

[220] Here one may observe that I am suggesting we can give unproblematic arguments for r-standards after all; and then haven't we avoided epistemic circularity? Perhaps, but only to encounter it again: one may, after all, restate the problem of epistemic circularity in terms of reliability rather than justification. Suppose the conclusion is something to the effect that certain sorts of memories are reliable, and that the premises advert to those sorts of memory: the argument is circular despite not having a conclusion in the form of a j-standard. We can avoid this sort of circularity, however, by acknowledging that forming beliefs on the basis of certain kinds of memory experiences is just (part of) being rational, and resigning ourselves to assuming what we thought we should have proven.

[221] But this generalization does not appear to apply to Schumm's example of an epistemically circular argument.

[222] In any case we certainly want to avoid arguing for a j-standard that is nothing but a restatement in terms of justification of an r-standard that licenses belief in the premises. See just below.

[223] An argument due to Wilfred Sellars and Laurence Bonjour. As Louis P. Pojman presents it in What Can We Know? An Introduction to the Theory of Knowledge (Belmont, CA: Wadsworth, 1995), p. 97, the argument is intended to show that whatever features might be used to identify a given belief as basic can be used to show that the belief is not basic after all; hence, no belief is basic. Hence foundationalism, which has it that there are basic beliefs, must be rejected.

[224] But the supports constraint from Chapter 2, Section VII is, perhaps, one example of a proposition that one could rationally believe about justification without argument.

[225] For further suggestions on how this procedure might go, I refer the reader to Max Black and to the meta-coherentists, discussed in Chs. 2 and 3.

[226] Perhaps the latter principle itself is sophisticated enough to have been derived, historically, from an even simpler principle.

[227] Alston, Reliability, op. cit., pp. 123-4.

[228] Ibid., p. 124.

[229] Ibid., pp. 126-9.

[230] Ibid., p. 129.

[231] Ibid., p. 125.

[232] Ibid., pp. 130-33.

[233] Ibid., p. 131.

[234] In Reliability, op. cit., at least. As we saw in Ch. 2, Alston has since changed his view; he now believes that epistemic circularity is not vicious.