STRONG FOUNDATIONALISM: HAS IT BEEN REFUTED?

by Maxwell J. Goss

 

Introduction

            In his book Epistemology: Becoming Intellectually Virtuous, W. Jay Wood surveys and critiques a number of traditional approaches to the problem of epistemic justification.[1] He is particularly critical of foundationalism, which, he maintains, cannot answer the skeptic and which lacks the resources to ground such common-sense intuitions as our belief in the reliability of the senses. He describes strong foundationalism in particular as “excessive in the extreme,”[2] and, in a section called “Problems with Strong Foundationalism,”[3] raises no less than sixteen separate objections to the project. His objections[4], while varied, center mainly on two foundationalist concepts, basic beliefs and basing relations, and can be grouped into three categories: the problematic nature of basic beliefs, the role of “background assumptions” in forming basic beliefs, and the legitimacy of basing relations other than strict deduction and induction. Wood’s criticisms, if conclusive, would devastate foundationalism. Given the growing opinion that the foundationalist project is either dead or “mortally ill,”[5] we must ask:  Can such criticisms be answered? The purpose of this essay is to defend strong foundationalism against Wood’s most serious objections, showing that rumors of its demise are greatly exaggerated.

           

            Given the varying usage of terms in the literature, it is useful to start with some definitions. Wood describes the three elements composing the foundationalist account of knowledge: basic beliefs, those self justifying beliefs which together form the base or foundation on which the “edifice” of knowledge is built;

nonbasic beliefs, whose justification is derived from basic beliefs; and basing relations, which specify how nonbasic beliefs derive their justification from basic beliefs. On this account, then, any justified belief that does not qualify as basic must ultimately be predicated (by means of a legitimate basing relation, such as deduction or induction) on one or more belief that does qualify. The various versions of foundationalism agree that an epistemic regress (that is, the giving of reasons for what one believes), if it would be fruitful, cannot go on forever, but must terminate at a set of beliefs that itself requires no justification: Knowledge requires foundations.[6]

            Strong foundationalism is distinguished from “weak” or “moderate” varieties by its insistence on a strict set of criteria for basicality. It regards as properly basic only those beliefs that are “unshakably certain,”[7] and has traditionally limited this class to three (closely related) types of belief. First, a belief qualifies as basic if it is self-evidently true, that is, if one grasps its truth immediately and without proof, simply by virtue of understanding the terms in question. For example, the so-called “laws of logic” (such as the law of identity) are self-evident; to understand them is to see that they are true. Second, an incorrigible belief, a belief about which it is impossible to be mistaken, is basic. Beliefs about one’s internal states (such as “I have a headache”) are typically classed as incorrigible. The subject is immune from error in such cases, as there is no “gap” between appearance and reality. Third, a belief is basic if it is evident to the senses, directly known by means of the subject’s sight, touch, taste, smell, or hearing.

            Here Wood distinguishes between Locke’s account of “sensitive knowledge” and what he identifies as the positivist account. Locke holds the deliverances of the senses to be reliable enough indicators of the external world to be used as epistemic starting points, so that the subject may consider them certain unless proven otherwise, while the early Russell, C.I. Lewis, and others take the narrower view that reserves

certainty to beliefs concerning the subject’s internal perceptual states.[8] It is the difference between “I see a table” and “I am appeared to in a table-esque manner,” or “I currently have a visual sensation of a brown patch shaped like¼” The former view does not, strictly speaking, belong to strong foundationalism, though it may be admitted by moderate foundationalism. Strong foundationalism does not regard as basic any belief about which it is possible to be mistaken, and so restricts the third criterion to beliefs “gleaned¼ from the immediate contents of [one’s] perceptual consciousness.”[9] It is possible, in fact, to view this narrower interpretation of “evident to the senses” as marking a subset of the beliefs that satisfy the incorrigibility criterion, though it is sometimes helpful to consider it as a separate category.

            For the foundationalist, then, knowledge is like a building, with basic beliefs forming the foundation and nonbasic beliefs forming the superstructure. Like a physical building, the support goes in one direction and is only as sturdy as its foundation. According to both strong and moderate foundationalists, the foundation of a person’s noetic structure must be inherently justified, not depending for its justification on any other beliefs—else it would not be a foundation at all. In addition, the strong foundationalist maintains that the foundation cannot be open to doubt, that is, it must be composed of beliefs that are either self-evidently true, incorrigible, or (narrowly) evident to the senses. Wood takes issue with every element in this account of knowledge.

 

The Nature of Basic Beliefs

            Wood’s first main criticism centers on the foundationalist conception of basic beliefs as unnecessarily limited. “Why¼should we think that only beliefs that are self-evident or incorrigible or evident to the senses are properly basic?¼These criteria alone appear much too restrictive, for they fail to encompass many beliefs we hold without the benefit of argumentative support.”[10] In other words, the

criteria are simply unrealistic; they should be expanded to include other beliefs—for instance, the belief in the reliability of the senses[11]—that nearly everyone in fact “holds as basic.”

            It must be noted at the outset that no responsible foundationalist wants to jettison such common-sense beliefs. The strong foundationalist may very well regard many of them as generally trustworthy. He sees it as important, however, to divide the class of beliefs that most people take for granted into two categories, beliefs that may be refuted and beliefs that may not. A belief might be, loosely speaking, “basic,” in that the average person seldom calls into it question, simply trusting it without conscious reference to a more fundamental belief. Most of the time, for example, I have little reason to doubt that my senses do not deceive me in any important way; my belief in their reliability can be described as “basic,” at least in a psychological sense. Occasionally, however, I might be led to question this belief—say, by a mirage that seems to appear and disappear on the road. The apparently conflicting data delivered by my senses[12] show me that my belief in their reliability, which I had regarded as “basic,” is in fact refutable, open to question. This introduction of doubt sets me on an epistemic regress, whereby I seek to evaluate my belief in terms of something more fundamental. Such a regress will go on indefinitely until some stopping point can be found, namely, a set of beliefs so basic that it cannot sanely be called into question. Beliefs belonging to such a set cannot be refuted: they are basic, in the strictest sense of the word.[13]

            The distinction here is not arbitrary, as Wood seems to imply. Beliefs that are basic in the strict sense—or irrefutable—are of a different order from those that are not. Although we may hold a large number of beliefs as “basic” or privileged in some sense, it is clear that some of these are open to question

while others are not—and that only the latter may function as epistemic “stoppers.” They are not open to revision and therefore lie at the very bottom of any evidence tree. (If they were open to revision, they would not be at the bottom.[14]) Wood correctly points out that most people hold a great variety of beliefs without the benefit of argumentative support. Few of us, however, consider such beliefs as the reliability of the senses to be infallible. To treat these beliefs as basic in a strict sense is simply to sidestep the initial questions: By what standard are our beliefs to be evaluated? How do we know when these beliefs are mistaken? The strong foundationalist has good reason for admitting as basic in the strict sense only those beliefs that are self-evident, incorrigible, or (narrowly) evident to the senses: these are the only types of belief that are inherently justified, and therefore able to halt a regress. The burden of proof is on foundationalism’s detractors to produce another kind of belief that measures up to this criterion.

            This last point is critical in responding to another, related, objection raised by Wood. He follows Alvin Plantinga is questioning the epistemic status of the foundationalist criteria for basicality. What, he asks, is the status of the belief that only beliefs that are self-evident, incorrigible, or evident to the senses qualify as properly basic? This belief, he asserts, is clearly not self-evident, incorrigible, or evident to the senses, nor is it logically rooted in any belief that is.[15] If this assertion is true, it decisively refutes strong foundationalism, identifying a self-referential incoherence in its account of knowledge.

            Fortunately for foundationalists, the assertion is false. This may be seen by restating the strong foundationalist condition for basicality in another way: A belief is basic if and only if it cannot rationally be doubted.[16] It is hard to imagine any beliefs that would measure up to this “indubitability criterion,” other than those that are self-evident, incorrigible, or narrowly evident to the senses. If another class of belief

were to meet the criterion, the foundationalist would happily admit it as basic. It is certainly legitimate to ask whether the imposition of this condition is consistent: It would be ironic indeed if strong foundationalists baldly asserted it without justification. An analysis of this criterion for basicality, however, will show that it is not arbitrary at all.

            The claim that revisable or dubitable beliefs cannot be epistemically basic is itself a basic belief—such beliefs are definitionally excluded from the class of basic beliefs. Even if a certain belief is considered broadly “basic” in that it is commonly held without argumentative support, it cannot be called basic in the strictest sense unless it is indubitable. Furthermore, it is clear that we do, and indeed must, partition our beliefs according to just such a distinction. For example, Wood sees no reason why my belief in the reliability of my senses should not be considered “properly basic.” The reason is that, while in practice I generally do not call this belief into question, it is still possible to do so in principle. The fact that my senses may be, and sometimes are, deceived (as in the case of the mirage) shows that my belief is open to revision.[17] It can rationally be doubted—and for that reason is, by definition, dependent for its credibility on those beliefs that cannot be doubted. Such a belief may, in fact, be very near the base of my evidence tree and in that sense more “basic” than many other beliefs; however, insofar as it is open to revision, it cannot be called basic in the narrowest sense. It is perfectly consistent for the foundationalist to maintain the indubitability criterion, and to judge the status of his beliefs according to it. Moreover, until someone shows how a belief that is neither self-evident, incorrigible, or evident to the senses can meet the criterion, it is both reasonable and useful to state it in its original form.

 

 

The Problem of “Background Assumptions”

            Wood’s second objection is that foundationalism cannot live up to its claim to halt an epistemic regress, due to its unwitting reliance on a variety of hidden “background assumptions.” He describes three types of assumption that supposedly compromise both the certainty and basicality of foundationalism’s basic beliefs. First, Wood claims that Descartes’s “I think, therefore I am” smuggles in, unargued, the existence of a unified self persisting over time. We would do better, according to Wood, to follow Russell, who “quipped that all Descartes was entitled to say is that ‘thoughts are being thunk!’”[18] Second, Wood defends Sellars’s “Myth of the Given” argument, which claims that even our seemingly basic perceptual beliefs, far from being independent and incorrigible, themselves “depend for their being intelligible on our being able to connect them correctly with (qualitatively similar) past experiences.”[19] In other words, one cannot recognize something as red without a prior experience of redness with which to compare it. Moreover, he cites Thomas Kuhn, Kant, and Wittgenstein among others to support his assertion that “all perception is theory-laden”—that a fallible, revisable conceptual scheme inevitably shapes the way we experience things like color. Third, Wood criticizes foundationalism’s alleged assumption of high access requirements. He claims that, for the foundationalist, it is not enough to possess basic beliefs; one must also “know that these stopping points are not arbitrary.” The demand cannot be met, he argues: such second-order knowledge proves highly elusive.

            The simplest way to respond to these criticisms is to point out that, as noted above, any belief that lies at the bottom of our epistemic structure is, by definition, basic. No belief that actually presupposes a certain set of background assumptions can be considered basic in the strict sense. Suppose it were true, as Wood claims, that various beliefs traditionally regarded as properly basic relied on a more fundamental framework of assumptions; in such a case, the solution would not be to discard the notion of basicality altogether, but rather to revise the set of beliefs we admit as basic. It seems that Wood, though he refutes

the basicality of a number of such beliefs, does not offer a serious challenge to the foundationalist conception of basicality itself, nor does he acknowledge the varied senses in which the term basic may be used. Nevertheless, if he is correct in asserting that most or all of the beliefs traditionally regarded as basic depend in turn on previously unseen background assumptions, then any version of strong foundationalism that relies on the foundational status of those beliefs must indeed return to the drawing board.

            Wood correctly argues that Descartes’s cogito assumes the existence of an “I” doing the thinking, but he considers this assumption gratuitous. The most we can conclude from the cogito, he suggests, is that thought is occurring (“thoughts are being thunk!”). There are two problems with this suggestion. First, it overlooks the fact that our certainty that thought is occurring is itself a significant claim about the world. It is contentful, and unshakeable. It has been asked: Why should something exist rather than nothing? It would seem possible that nothing should exist rather than something, yet our irrefutable certainty that thought is occurring closes off this logical option, as well as any conclusion that might be based on it. Second, Wood fails to notice what should be obvious, that thoughts require a thinker. Perhaps this thinker is not the continuous, “unified center of consciousness”[20] that I ordinarily suppose myself to be. However, something must be doing the thinking, and any thinker may, by grammatical convention, refer to the possessor of his thoughts in the first person singular. The certain existence of at least one such “I” is not without philosophical significance.

            What about the claim that my supposedly basic beliefs are intelligible only with reference to a) my ability to relate them to a body of prior experiences (Sellars’s “Myth of the Given”) or b) a particular theoretical framework (Kuhn’s claim that “all perception is theory-laden”)? If either argument is taken to its conclusion, knowledge becomes deeply problematic. Fortunately, neither is compelling.

            Wood follows Sellars in arguing that perceptual beliefs rely on comparison. I cannot be said to recognize a particular sensation until I have collected a certain number of similar experiences against which to judge it. The sensation does not translate to knowledge, then, without some point of reference gained

from prior experience. This account, however, reveals a serious flaw: If knowledge cannot be acquired without prior knowledge, then how can the process ever get started in the first place? At first glance, knowledge would appear impossible to acquire, something which neither Sellars nor Wood wants to affirm. There are two ways to avoid this conclusion. Either we are born with some kind of “innate ideas” which form the initial reference points for judging our subsequent experiences, or it is possible to distinguish (at least) two types of knowing, comparative and perceptual. The first alternative, famously discredited by Locke, is, again, something both Sellars and Wood would likely prefer to avoid. The second alternative is exactly what the foundationalist assumes. It is quite possible to have a certain perceptual experience (i.e., to be aware of a “sense-datum”) without comparing it to other similar experiences—otherwise, there could be no perceptual experience at all. It is true, of course, that if I report an experience of, say, redness, I must possess some point of comparison to which I am referring (namely, the memory of some prior experience of redness).[21] It is not necessary, however, to describe such a comparative report as properly basic. The direct experience is basic, but understanding the experience in comparative terms may be nonbasic. The comparison is predicated on a set of basic beliefs, including my current experience of redness and my current memory of past experiences of redness. Such a distinction is a hallmark of foundationalism, and it renders Wood’s argument pointless. Alternatively, denying the distinction renders it incoherent.

            A careful definition of the term basic, therefore, is critical. This applies equally to Wood’s assertions concerning the inevitable influence of a person’s theory-laden conceptual framework. He writes that certain philosophers “have shown us that perceptual reports, even of so primitive a sort as ‘I see red,’ are embedded in larger theoretical and linguistic frameworks¼which invest them with the meaning they have for us.”[22] To be sure, it is entirely possible that our concepts shape the way we interpret our experience (sense or otherwise), and that they frequently do so apart from our conscious reflection. My conceptual pre-commitments may, for example, lead me to attend to certain experiences and not others, or

to associate particular experiences and not others, or to draw certain conclusions from my experiences. They may even lead to mistakes that are quite difficult to untangle. However, such interpretation always takes place at the level of nonbasic rather than basic beliefs. A basic belief is, by definition, immune to interpretive mistakes. It is epistemically prior to any conceptual framework. Such a framework may affect how I treat a particular perception, but it can have no influence on the initial perception itself.[23] Wood has not succeeded in showing that one’s basic beliefs are intelligible only with reference to a pre-existing body of experience or theoretical framework. His failure to distinguish between different senses of the term basic (as seen, for instance, in his assertion that “the likelihood of error with respect to foundational beliefs exceeds zero”[24]) renders his criticism ineffectual.

            Wood concludes his account of the alleged problems associated with background beliefs by raising the problem of “access requirements.”[25] He argues that, if the foundationalist account of knowledge is to be accepted, then it is not enough simply to possess strong foundations; one must also be able to say why these foundations support his other beliefs. It is necessary, in other words, that one be aware of “the properties or features making these and not some other foundational beliefs the ones by which certain knowledge is gained.”[26] This demand for second-order knowledge, Wood argues, inevitably involves one in a vicious justificatory regress.[27]

            The answer to this objection is straightforward: Wood’s account of foundationalism’s access requirements is simply mistaken. Earlier he describes what he understands to be the high access

requirements of Descartes’s foundationalism: “Put briefly, in order to know, one must know that one knows and be able to show that one knows.”[28] Another way to put this is to say that first-order knowledge (simply knowing something) is not enough; one must also possess second-order knowledge (knowing that one knows something). Such second-order knowledge, however, is by no means essential to contemporary versions of strong foundationalism. Wood understands this, as is clear from his remarks in one of his endnotes: “Being self-reflectively aware that one possesses knowledge, justifiably believing this to be the case, knowing this to be the case and showing this to be the case are all distinctive requirements, many or all of which do not appear in some versions of foundationalism. Failure to keep these separated leads to what William Alston calls ‘levels confusion in epistemology.’”[29]

            In his critique of strong foundationalism, Wood involves himself in a “levels confusion” of his own. His failure to distinguish between various versions of foundationalism seems to reveal a lack of sympathy for his subject. High access requirements are not, in fact, among foundationalism’s “background assumptions.” It is entirely possible, on foundationalist terms, to be fully justified in one’s belief without being able to explain one’s justification. Wood’s criticism misses the mark.

           

Basing Relations

            Finally, Wood’s criticizes the “notion of basing relations contained in classical foundationalism—most notably the idea that all nonbasic beliefs can be inferred from basic beliefs via deduction or induction.”[30] He objects vigorously to such a restriction, citing several additional types of inference relation that he deems valid. These include explanatory inference, mutual support or “concurrence,” and cognitive spontaneity.

            In defense of explanatory inference, Wood provides a fairly elaborate account of the “oft-told story of the discovery of Neptune.” He points out that it was irregularities in the orbit of Uranus that led J.C.

Adams to posit the existence of a planet whose size and position matched that of Neptune—yet “independent empirical confirmation of Neptune’s existence” was not made until much later. Surely, Wood suggests, Adams was justified in his belief in Neptune, even though he had no direct visual evidence for its existence. This belief was not reached by means of strict deductive or inductive reasoning, but by a reasoning to the best explanation.[31]

            Of course, both strong and weak foundationalists can and do make use of explanatory inference.[32] Wood does not appear to be arguing against any actual contemporary foundationalists, but rather against a kind of caricature that, as he puts it, “insist(s) on deduction and induction as the sole basing relations.”[33] Again, he does not seem to grasp just what features are essential to strong foundationalism. His comment concerning the “empirical confirmation” of Neptune is revealing in this regard. Were the observations of Uranus’s orbital irregularities something other than empirical? He seems to suggest that a foundationalist would not deem them empirical enough. Moreover, he claims that many of our justified beliefs, such as Adams’s belief in the existence of Neptune, cannot be “grounded in anything remotely like the kind of certainty that strong foundationalists insist upon.”[34] This claim, apart from being false, reflects what appears to be a deep misunderstanding of strong foundationalism. A foundationalist surely could consider Adams’s belief in Neptune justified, even though it fell short of absolute certainty (as does ours). Adams’s hypothesis can actually be understood as an instance of Bayesian reasoning that is quite compatible with strong foundationalism.[35] Furthermore, it is difficult to deny the foundationalist claim that Adams’s belief

was grounded in (i.e., justified in terms of) a number of “ironclad” certainties, including his sense-data and his knowledge of mathematics.

 

            Wood also thinks that “mutual support” should be included as a legitimate basing relation. He argues that, just as the individual strands of a rope are strengthened when woven together, so may many of our individual beliefs be strengthened when taken together as a set. Our beliefs thus “achieve a strength when combined that exceeds what they have in isolation.”[36] In one sense, this assertion is both true and

consistent with foundationalism. Independent lines of evidence often do combine to make a stronger, sometimes dramatically stronger, case for the truth of a proposition than any of the lines taken in isolation. It can be shown, for instance, that the testimony of two or three independent witnesses, each of whom is only moderately reliable, can (providing they agree) yield a very high degree of confirmation when taken together.[37] This should not be surprising. However, there is another, more fundamental, sense in which Wood’s assertion is false. Just as a rope cannot be stronger than the sum of the strength of its individual strands, so the concurrence of several lines of evidence cannot be stronger than the total of their strength taken in isolation. The reason for this lies in the nature of the basing relation, which is merely a means of transferring epistemic “strength” from one belief to another; it can confer no such strength in itself. To think that it might is to mistake an argument’s basing relations for its bases, its structure for its content. So-called “mutual support” does not in itself constitute a valid basing relation, anyway; it only describes the combined effect of several other lines of support.

            Finally, Wood points out that many of our beliefs are “cognitively spontaneous”; they simply arise apart from any conscious line of reasoning. Borrowing from the insights of cognitive science, he asserts that “we are psychologically so constituted that when confronted with a certain kind of perceptual display,

we immediately and noninferentially form a corresponding belief.”[38] This account of belief formation, unfortunately, fails to distinguish between what could be termed psychological and epistemic priority in forming beliefs. It is true that, most of the time, I am not conscious of how my perceptual beliefs are formed. I simply seem to find myself with these beliefs; I experience them as spontaneous. It does not follow, however, that the beliefs are not formed inferentially. The phrase “when confronted with a certain kind of perceptual display” is crucial. It suggests that a perceptual belief cannot be justified without a matching “perceptual display.” Wood, then, recognizes both that there is a necessary relationship between the two and that this relationship is one-way. What could this relationship be, if not an inferential one? The

inference relation is surely there, even if I am not aware of it psychologically. The very fact that I am able to judge my perceptual beliefs shows that they are at least tacitly inferred from certain perceptual displays; these beliefs can only be evaluated by distinguishing legitimate from illegitimate inferences. Psychological spontaneity, then, does not itself constitute a valid basing relation. Any validity it may possess derives entirely from its reliance on inference.

 

Conclusion

            It seems foundationalism is in no danger of bankruptcy. Neither the basic beliefs nor the basing relations that are essential to it have been undermined by Wood’s criticisms. One additional point is worth mentioning in conclusion. Many of Wood’s objections to strong foundationalism stem from an attempt to do an “end run” around the questions that give rise to epistemology in the first place. For instance, he may be right—as he asserts in several places—that we are justified in considering our senses to be generally reliable indicators of the external world. The justification of such a belief, however, can only be assessed with reference to a more fundamental set of beliefs, ones that are not open to doubt the way the belief in question is. To ignore the problem of how this justification is derived is to skip over what is both most vexing and most interesting to the epistemologist. Such a move leads to a certain arbitrariness in choosing

one’s epistemic starting points, as well as to an inability to offer any convincing reply to the skeptic. Wood raises many intriguing questions about the role of tradition, conceptual frameworks, and our cognitive makeup in the process of belief-formation, and is certainly correct in emphasizing the importance of the intellectual virtues in this process. However, the contributions of such factors simply cannot be assessed until we answer the prior question of justification.


[1] Jay Wood, Epistemology: Becoming Intellectually Virtuous (Downer’s Grove: InterVarsity Press, 1998).

[2] Wood, 91-92.

[3] Wood, 88-98.

[4] It should be pointed out that Wood does not claim to do more than report the objections of other critics, which do not necessarily reflect his own views. It seems clear, however, that he intends to persuade the reader of their validity, as the critique of foundationalism forms a critical line of argument in his case for virtue epistemology.

[5] Wood quotes Nicholas Wolterstorff, who writes: “On all fronts foundationalism is in bad shape. It seems to me that there is nothing to do but give it up for mortally ill and learn to live in its absence” (Wood, 203, n. 5).

[6] Wood, 84-85.

[7] Wood’s phrase, 85. Wood customarily casts his discussion in terms of what qualifies as “properly basic,” a phrase made famous by Alvin Plantinga. I will follow Wood’s use of this phrase, though I will reach different conclusions from Plantinga about what beliefs qualify as “properly basic.”

[8] Wood, 85-87. See also 204, n.12.

[9] Wood, 86.

[10] Wood, 89.

[11] It is possible here to distinguish between the claim that a) the class of basic beliefs should include the various deliverances of the senses (such as “there is a table”) and the more ambitious claim that b) the class of basic beliefs should include a proposition to the effect that the senses are reliable indicators of the external world. The latter claim concerns the senses themselves, while the former concerns only what they deliver to one’s awareness. Though Wood is not careful to distinguish the two, he suggests the more ambitious claim: “Consider one’s belief in the reliability of one’s senses or memory or consciousness; none of these beliefs is either self-evident or incorrigible or evident to the senses, yet they are held as properly basic by virtually everyone.” (Wood, 89).

[12] The data may be in apparent conflict because I know a puddle cannot both exist and not exist, and because the disappearance of the image occurred too quickly to be attributable to evaporation. It is important to note that the conflict here is between interpretations of the data rather than between the data per se.

[13] This analysis applies equally to the more modest claim that only the actual deliverances of the senses (as opposed to their general reliability) are to be considered basic (see note 12).

[14] This analysis points up a problem for moderate foundationalism, which admits fallible beliefs into the class of basic beliefs. In order to avoid circularity (and remain a foundationalist), it seems that one must sooner or later appeal to a belief that is irrefutable.

[15] Wood, 89.

[16] I take it that irrefutable and indubitable, while not strictly synonymous, are coextensive and hence more or less interchangeable when applied to basic beliefs.

[17] Plantinga writes, “I believe, for example, that I had lunch this noon. I do not believe this proposition on the basis of other propositions; I take it as basic; it is in the foundations of my noetic structure” (Alvin Plantinga, “Reason and Belief in God,” in Faith and Rationality, ed. Alvin Plantinga and Nicholas Wolterstorff [Notre Dame: university of Notre Dame Press, 1983], 59.) It may be true that, in most circumstances, he takes such a belief as more or less “basic.” However, were he given reason to doubt such a belief (for instance, by discovering that someone had slipped a mind-altering drug into his coffee), he could only judge its truth in terms of a yet more basic set of beliefs. What Plantinga calls his “foundations,” then, are themselves further analyzable in terms of levels of epistemic priority.

[18] Wood, 90.

[19] Wood, 90.

[20] Wood, 90.

[21] Unless, of course, I have just coined the word “red” to describe my new experience.

[22] Wood, 91.

[23] It may be argued that a particular theoretical framework may make a person’s mind selective in a certain way, leading him to retain certain data and forget other data. Even if such an assertion is true, it does not get around the fact that if, at a given moment, a person is having a particular experience, then that experience cannot, in that same moment, be called into question.

[24] Wood, 92.

[25] His line of reasoning in this section is less than clear, so it is not certain just why he subsumes access requirements under the problem of background beliefs.

[26] Wood, 91.

[27] On page 91, Wood supports his claim by putting forward a simplified version of Laurence BonJour’s “phi” argument: To accept as basic a certain claim B presupposes another claim K to the effect that “B has whatever features are required to make it and not some alternative the proper foundation for knowledge.” This argument does not, on inspection, render even second-order knowledge impossible for the epistemologist, because it mistakes an epistemic principle (K) for a premise.

[28] Wood, 87.

[29] Wood, 204, n. 13.

[30] Wood, 92-93.

[31] Wood, 92.

[32] Bertrand Russell, James Cornman, Alan Goldman, and Timothy McGrew all make explicit use of explanatory inference in their versions of foundationalism. Paul Moser also discusses it. See Bertrand Russell, The Problems of Philosophy (Oxford: Oxford University. Press, 1912); James Cornman, Skepticism, Justification, and Explanation (Dordrecht: D. Reidel, 1980); Alan Goldman, Empirical Knowledge (Berkeley: University of California Press, 1988); Timothy McGrew The Foundations of Knowledge (Lanham, Md.: Littlefield Adams, 1995); Paul Moser, Empirical Justification: Readings in Contemporary Epistemology (Dordrecht: D. Reidel, 1985).

[33] Wood, 93.

[34] Wood, 93.

[35] For the general form of such reasoning, see Colin Howson and Peter Urbach, Scientific Reasoning: The Bayesian Approach (LaSalle, Il.: Open Court, 1989), 96-102.

[36] Wood, 93.

[37] See Timothy McGrew, “How Foundationalists Do Crossword Puzzles,” Philosophical Studies 96 (1999): 333-50.

[38] Wood, 94 (emphasis his).