Notes

Introduction

1. In my view, Pyrrhonian skepticism requires a completely separate treatment from Cartesian skepticism. This is for a number of reasons, but the main issue is that it is not a paradox but rather a distinctive kind of ethical position (which means, among other things, that the scope of the doubt is significantly constrained). In contrast, I follow Conant (e.g., 2004; 2012) in thinking that Cartesian and Kantian forms of skepticism are related in important ways. Although I don’t have the space to address this issue here, I think Humean skepticism is hard to place with regard to other forms of skepticism precisely because it has elements of all three kinds of radical skepticism just listed. Henceforth, where I refer to “skepticism” or “radical skepticism” without qualification, then it is the particular Cartesian variety of this problem, as it is cast in the contemporary epistemological debate, that I have in mind.

2. That’s not to say that I think anti-luck epistemology is in doubt, only that its application to the problem of radical skepticism is limited. For some of my recent defenses of anti-luck epistemology, which is now embedded within a view of knowledge I call anti-luck virtue epistemology, see Pritchard (2007a; 2007c; 2009a; 2012a; 2012c; 2013c; forthcoming–a) and Pritchard, Millar, & Haddock (2010, chaps. 1–4).

3. Henceforth, “OC.” See, for example, Pritchard (2005g; 2011f; 2012e; 2014a; forthcoming–d).

4. See especially Pritchard (2012b). See also Pritchard (2007b; 2008b; 2009f).

5. Many thanks to all those people who have suggested alternative names (I’d thank everyone individually, but the list is far too large). Alas, in the end none of the names proposed had the crucial dual virtue of both being aesthetically pleasing and conveying the particular point that I wanted to make.

6. Which, note, is not to say that the skeptical problem is able to generate real radical skeptical doubt. This point will become much clearer later on in the book once we encounter Wittgenstein’s account of our hinge commitments.

7. I think this relates to Cavell’s (1988, chap. 6) famous remark, in the context of radical skepticism, about the “uncanniness of the ordinary.”

8. Note that this point is entirely compatible with Wittgenstein’s (1953, §124) oft-cited claim that philosophy “leaves everything as it is.” If this solution to radical skepticism is right, then there really is nothing amiss (on this front anyway) as regards our ordinary epistemic practices.

Chapter 1. Radical Skepticism and Closure

1. It is, of course, an interesting question how much affinity there is between Cartesian skepticism, as it is understood in the contemporary epistemological literature, and the kind of radical skepticism explored by Descartes himself, though I will not be getting into this issue here. For a useful discussion of this issue, see Luper (2011). See also Gaukroger (2011). As I noted in the introduction, there are, of course, other significant forms of radical skepticism. In the first instance I have in mind Pyrrhonian skepticism, which does not proceed by offering philosophical arguments as such, much less philosophical theses, but rather puts forward a piecemeal, and yet unrestricted, method of doubt. But there is also a broadly Kantian form of radical skepticism which I think is very important too. Very roughly, this is skepticism about the very idea that one could entertain contentful thoughts about an external world. (There is also Humean skepticism, though as I remarked in the introduction, this is harder to classify relative to the other main forms of radical skepticism. See Beebee (2011) for a useful recent overview of Humean skepticism). Although these forms of radical skepticism do feature in the contemporary debate, they do not nearly have the same profile as the broadly Cartesian form of radical skepticism that we will be concerned with in this book. I take the challenge posed by these forms of radical skepticism to be distinctive, and so deserving of a separate treatment. That said, if the anti-skeptical proposal that I set out in this work is sound, then it will at the very least make the task of dealing with the Pyrrhonian and Kantian skeptical challenges much easier (though I’m afraid the reader will have to take this on trust, as I won’t be arguing for this claim here). For some representative contemporary work on Pyrrhonian skepticism, see Bett (2011) and the essays collected in Machuca (2011). For some of my own engagements with this variety of skepticism, see Pritchard (2000b; 2011d; 2013b). For some helpful recent works that engage with both Pyrrhonian and Cartesian forms of radical skepticism, see Fogelin (1994), Williams (2001b), Gascoigne (2002), and Hazlett (2013). See Forster (2010; cf. Forster 2011) for an important recent work on Kantian skepticism, but note that his discussion of what he calls “veil of perception skepticism” doesn’t quite correspond to the kind of radical skepticism I have just described as Kantian. Indeed, this is a point on which this work has been critiqued—see, for example, Chignell & McLear (2010). For an influential treatment of Kantian skepticism more in keeping with my description of it, see McDowell (1994b). See also Conant (2004; 2012). For a general overview of the contemporary debate regarding radical skepticism, see Pritchard (2002c). For a recent annotated bibliography of the contemporary literature on radical skepticism, see Pritchard (2010b).

2. I first started referring to the problem posed by this form of radical skepticism in terms of epistemic angst in Pritchard (2005b, passim; 2005e; 2008c). Since completing this manuscript, Ernie Sosa has alerted me to the fact that in early work he used the phrase “meta-epistemic angst” to refer to a general epistemological outlook that he claims derives from Roderick Chisholm’s writings. See Sosa (1991, 209n).

3. Note that I use the term “perceptual” here in the broad sense that it includes all sensory perception (i.e., and not just visual perception), including proprioception.

4. If you wish, add to the BIV scenario that the agent concerned has only recently become envatted. So construed, the kind of content externalist considerations put forward against BIV-based radical skepticism by Putnam (1981, chap. 1) and others are thereby neutralized. For further discussion of the relationship between content externalism and radical skeptical hypotheses, see Brueckner (2012).

5. This claim is actually logically stronger than we need, in that the following would suffice for our purposes:

(S₁1*) One does not know that one is not a BIV.

The version of the radical skeptical paradox that employs the stronger formulation is, however, more perspicacious, and hence we will stick with this formulation in what follows. The skeptic, after all, is not claiming that it is a mere incidental epistemic lack on our parts that we fail to know the denials of radical skeptical hypotheses, as if we could have this knowledge if only we were smarter, were more attentive, made more inquiries, etc. The skeptical claim is rather that this is an in principle epistemic lack on our part. See also notes 6, 9, and 10.

6. I noted above—see note 5—that (S₁1) could be replaced with a weaker claim, (S₁1*), without loss. With the weaker claim in play, however, it is also true that (S₁2) could be replaced with a weaker claim without loss as well:

(S₁2*) If one does not know that one is not a BIV, then one does not know E.

(S₁2*), after all, would suffice to generate the required logical tension between (S₁1*) and (S₁3). See also note 9.

7. This is essentially the formulation of the closure principle put forward by Williamson (2000a, 117) and Hawthorne (2005, 29). See also David & Warfield (2008).

8. As it happens, DeRose’s point was in fact targeted at the previous formulation of the closure principle (i.e., where knowledge is closed under knowledge entailment), but this doesn’t matter for our purposes.

9. As noted in notes 5 and 6, there is also a logically weaker version of this skeptical paradox available, which proceeds as follows:

THE RADICAL SKEPTICAL PARADOX (I*)

(S₁1*) One does not know that one is not a BIV.

(S₁2*) If one does not know that one is not a BIV, then one does not know that E.

(S₁3) One knows that E.

10. As noted above in note 5, all we actually need to generate the paradox is that one does not know the denials of skeptical hypotheses (as opposed to cannot).

11. See Stroud (1984) for a seminal contemporary discussion of the problem of radical skepticism qua paradox. As he famously put the point—see Stroud (1984, 82)—radical skepticism falls out of “platitudes” that “we would all accept,” but which are collectively inconsistent. Wright (1985; 1991) makes a similar point, and also expands on the dialectical implications of treating radical skepticism as a paradox rather than as a philosophical position. Consider, for example, this passage:

[T]he best philosophical paradoxes … signal genuine collisions between features of our thinking that go deep. Their solution has therefore to consist in fundamental change, in taking up conceptual options which may have been overlooked.… [T]he traditional sceptical arguments, in their strongest forms, are such paradoxes. (Wright 1985, 429–30)

We will be discussing the idea of radical skepticism as a paradox again in chapter 3 (§2), when we look at Wittgenstein’s distinctive response to this problem. See also Pritchard (2014c).

12. For more on undercutting and overriding anti-skeptical strategies, see Pritchard (2012b, pt. 3, §6). A very similar way of thinking about these issues is offered by Cassam (2007a). He argues that we should recast skeptical problems in terms of “how possible?” questions, and then distinguishes different levels of response to this problem. This is basically equivalent to recasting the radical skeptical argument into a paradox, as I explain in Pritchard (2009e). Moreover, Cassam then goes on to distinguish (among other things) between obstacle-overcoming and obstacle-removing anti-skeptical strategies, which is roughly equivalent to our distinction between overriding and undercutting anti-skeptical strategies. For another related, and influential, discussion of the dialectical situation with regard to the radical skeptical problem, see Williams (1991, chap. 1).

13. Another way of expressing this point—due to Schiffer (1996, 330)—is that if radical skepticism poses a genuine paradox then there is no “happy face” solution to this problem, only a “sad face” solution, because it would mean that there is indeed a “deep-seated incoherence” within our pretheoretical epistemological commitments.

14. Famously, Dretske (1970) and Nozick (1981) denied a version of the closure principle—essentially, they denied, respectively, the two cruder formulations of the closure principle offered above—though their reasons for denying this principle as they formulated it do not carry over to the more nuanced formulation of the principle that is at issue here—i.e., one that essentially involves a competent deduction on the part of the subject. For some useful critical overviews of the numerous grounds that have been offered in the literature for denying various closure-style principles, see Collins (2006), Kvanvig (2006), Brueckner (2010a), and Luper (2010). For a recent discussion of the merits of the closure principle, see the exchange between Dretske (2005a; 2005b) and Hawthorne (2005). Note that I am setting aside here the particular difficulties that beset so-called multipremise closure principles. See Hawthorne (2004, passim) and Collins (2006, §1) for discussion on this issue. We will be considering an anti-skeptical proposal (the entitlement reading of Wittgenstein’s On Certainty) that has some broad affinities with the “nonclosure” strategy in chapter 3, §5.

15. Although nothing depends on this here, this way of understanding epistemic internalism is explicitly along accessibilist, as opposed to mentalist, lines, where mentalism is the other dominant way of characterizing epistemic internalism in the contemporary literature. For two key defenses of accessibilism, see Chisholm (1977) and BonJour (1985, chap. 2). For the core defense of mentalism, see Conee & Feldman (2004). For some useful discussions of the debate between accessibilists and mentalists, see Steup (1999), Pryor (2001, §3), BonJour (2002), Pappas (2005), and Poston (2008). Note that henceforth I will take it as given, unless otherwise specified, that rational support meets this accessibilist requirement.

16. For a key defense of process reliabilism, see Goldman (1986), though note that Goldman wouldn’t endorse the crude rendering of the process reliabilist thesis that we have just offered in the main text. For a very useful recent survey of work on reliabilism, see Goldman (2008).

17. Note that in part 3 we will be questioning whether even by epistemic internalist lights it follows that one lacks a rational basis for believing (and thus knowing) that one is not a BIV.

18. I think the first usage of this label to describe this kind of anti-skeptical proposal is found in Pritchard (2002c).

19. For helpful discussion of Moore’s contributions to epistemology, see Baldwin (1990, chap. 9; 1993). For some useful discussions of neo-Mooreanism, see, for example, Sosa (1999), Black (2002; 2008), and Pritchard (2002d; 2005b; 2007b). For a survey of the literature on Mooreanism and neo-Mooreanism, see Carter (2012).

20. In the contemporary literature the closure principle is often contrasted with the so-called transmission principle—see, e.g., Davies (2004) and Wright (2004c)—and one might think that the latter is effectively identical with the closure_RK principle here formulated. There is a subtle difference, however. For whereas the closure_RK principle merely demands that competent deductions from rationally grounded knowledge generate rationally grounded knowledge, the transmission principle demands in addition that the very same rational basis for knowledge of the entailing proposition should be a rational basis for the entailed claim. This is a more specific thesis. For our purposes, however, all that matters is that the subject’s knowledge of the deduced proposition is no less rationally grounded, and we can bracket the issue of whether a particular rational basis has been “transmitted” across the competent deduction. It could be, for example, that the competent deduction itself transforms the rational basis for the agent’s knowledge of the entailing and entailed proposition. For more on this issue, see Pritchard (2014a). See also Klein (1981; 1995), whose discussion of closure-style principles is relevant here. Note that we will be examining the style of anti-skepticism that is based around a rejection of the transmission principle in its own right in chapter 3, §5.

21. We will be further discussing the possibility of denying the closure_RK principle on epistemic externalist grounds in chapter 3, §4. For further discussion of the relationship between epistemic externalism and closure in the context of radical skepticism, see Pritchard (2002b).

22. For further discussion of abduction more generally, including critiques of this form of reasoning, see Lipton (1991) and Douven (2011).

23. See Vogel (1990, 659) for an example of an abductivist response to the problem of radical skepticism that appeals to simplicity.

24. Indeed, if we are to weigh the simplicity of an explanation in ontological terms—i.e., in terms of how many entities it commits us to—then the skeptical hypothesis will presumably count as much simpler than the everyday scenario.

25. For example, it is often claimed that the way we typically evaluate testimony involves abductive inferences. See Adler (1994) and Lipton (1998).

26. See Lipton (1998) for a seminal account of the centrality of abduction to scientific reasoning.

27. BonJour (e.g., 1999) is the clearest articulation of such a rationalist approach to abductivism, though in offering this approach he appeals to a number of independently contentious claims. See Beebe (2009) for an excellent critical discussion of BonJour’s abductivism.

28. See Vogel (1990) for an influential defense of an abductive response to the problem of radical skepticism. See also BonJour’s (e.g., 1999) defense of this line of response, which is surveyed and discussed by Beebe (2009) and McCain (2012). For critical discussion of abductivism, see Fumerton (1992), Neta (2004b), and the exchange between Fumerton (2005) and Vogel (2005).

Chapter 2. Radical Skepticism and Underdetermination

1. Note that—as with, mutatis mutandis, our formulation of closure-based radical skepticism in chapter 1—(S₃1) is in fact formulated much more strongly than we actually need, for it would suffice for the skeptic’s purposes that we merely do not have this favoring epistemic support, rather than that we cannot have it. The stronger claim better captures what the skeptic is maintaining, however, in that the thesis in play is not merely that there is some incidental epistemic lack on our part, one that could potentially be rectified, such as by making further inquiries. The claim is rather that there is an in principle epistemic lack at issue. See also notes 3, 5, and 6.

2. Recall that we discussed abductive treatments of radical skepticism in chapter 1, and explicitly understood these proposals as attempting to demonstrate that one has a stronger rational basis for one’s everyday beliefs over radical skeptical alternatives. In doing so we were effectively treating them as being applied primarily against a radical skeptical claim like (S₃1). We noted that such proposals are highly problematic, however, and at best constitute overriding anti-skeptical strategies. Accordingly, we will set this style of anti-skepticism to one side here.

3. As with our formulation of (S₃1)—see note 1 above—this is actually stronger than we need in order to generate the skeptical paradox. In particular, it would suffice for the antecedent that one does not (as opposed to cannot) have a rational basis that favors one’s belief that E over the BIV hypothesis.

4. For some of the main discussions of underdetermination-style principles and their role in radical skeptical arguments, see Yalçin (1992), Brueckner (1994), Cohen (1998), Byrne (2004), Vogel (2004), and Pritchard (2005b, pt. 1; 2005f). See also Pritchard (2012d).

5. As noted in notes 1 and 3 above, there is actually a weaker formulation of this paradox available, which goes as follows:

THE RADICAL SKEPTICAL PARADOX (III*)

(S₃1*) One does not have rational support that favors one’s belief that E over the BIV hypothesis.

(S₃2*) If one does not have rational support that favors one’s belief that E over the BIV hypothesis, then one does not know that E.

(S₃3) One knows that E.

6. As noted above—see note 1—it would actually suffice for this paradox that one merely does not have this favoring rational support (i.e., as opposed to cannot).

7. For some of the key defenses of attributer contextualism, see DeRose (1995), Lewis (1996), and Cohen (2000). For some useful overviews of this position, see DeRose (1999), Brady & Pritchard (2005), Black (2006), and Rysiew (2011a; 2011b). Note that this kind of attributer contextualism is very different from the superficially similar view defended by Williams (1991). See Pritchard (2002c; 2002e) for a comparative discussion of these two types of contextualism. See also Black (2006). We will be looking at Williams’s proposal in its own right in chapter 4.

8. That’s not quite right, of course, in that in order to “talk past” someone one (usually, anyway) has to at least be talking with the person, and the contrast we just drew was between a skeptical conversational context and a normal, nonskeptical conversational content—i.e., two distinct conversational contexts. Still, the general idea that the radical skeptical problem on this view trades on a misunderstanding—i.e., on something purely semantic—should be clear.

9. Note that in what follows I will talk not of conversational contexts specifically (although this is the usual terminology employed by attributer contextualists), but rather more generally of contexts of epistemic appraisal (which leaves it open whether such contexts need to be specifically conversational). Nothing hangs on this shift of terminology for our purposes (indeed, arguably it makes the attributer contextualist position more palatable). See also note 12.

10. See Heller (1999) for a nonstandard version of attributer contextualism that in addition denies the closure principle. For a critique of this proposal, see Pritchard (2000a).

11. For a very honest response to this problem from a prominent attributer contextualist, see Cohen (2000), who argues that we must regard our beliefs in the denials of radical skeptical hypotheses as having an a priori status. As even Cohen concedes, however, this is not a comfortable dialectical move for the attributer contextualist to make. See also Cohen (1999). For more on the contrast between attributer contextualism and neo-Mooreanism as regards (roughly) the present formulation of the skeptical problem, see Pritchard (2005c; 2005d).

12. Note the potential importance of our terminological shift from expressing attributer contextualism in terms of conversational contexts to formulating the view in terms of contexts of epistemic appraisal (see note 9). After all, in considering radical skeptical hypotheses one might not actually be in a particular conversational context at all, at least where that is understood as something involving actual conversation between two or more people. Still, I think we can charitably interpret attributer contextualism such that whatever applies to conversational contexts will carry over—somehow, the details need not concern us here—into relevantly analogous nonconversational contexts of epistemic appraisal.

13. This is the reason why attributer contextualism doesn’t face the problem of “abominable conjunctions”—see DeRose (1995)—which we saw in chapter 1 (§1) afflicted views that deny the closure principle. Recall that abominable conjunctions concern the assertion of sentences that simultaneously affirm an instance of one’s everyday knowledge while also denying that one has knowledge of the denial of a skeptical hypothesis that is clearly inconsistent with that everyday knowledge. According to attributer contextualism, the key to resolving this problem is to realize that the introduction of the skeptical hypothesis in the second conjunct effectively ensures that the relevant context of epistemic appraisal is a skeptical one, where high epistemic standards are in play. So while attributer contextualists hold that there are everyday contexts of epistemic appraisal relative to which the assertion of the first conjunct expresses a truth, there is no single context of epistemic appraisal relative to which the assertion of the conjunction as a whole expresses a truth.

14. Actually, I think the idea that the “conversational air” can simply clear in this way is far from straightforward. This is the so-called problem of epistemic descent, as explained in Pritchard (2001a). I will be setting this concern to one side in what follows, however.

15. I believe the first to make this point was Schiffer (1996). Note that this is now part of a cluster of related (semantic/linguistic) objections that are made to attributer contextualism. See, for example, Stanley (2005, chap. 2) on whether “knows” is a gradable adjective like “flat” or “tall” as (some canonical versions of) attributer contextualism suggest. There is also the related issue of whether attributer contextualism correctly describes our everyday practices of asserting knowledge ascription sentences. For further criticism in this direction, see Pritchard (2001a; 2005a). For a helpful overview of the contemporary literature regarding both issues, see Rysiew (2011b, esp. §4).

16. We will return to this issue in chapter 4 when we explore inferential contextualism, and in chapter 6 when we will examine rational support contextualism.

17. One finds variants of this particular charge against the attributer contextualist response to radical skepticism in a number of places. See, for example, Feldman (1999; 2001) and Kornblith (2000).

18. For a related critique of the attributer contextualist treatment of radical skepticism, see Sosa (2011, chap. 5). Note that, for now, I have set aside the distinctive kind of attributer contextualism that is proposed by Neta (2002; 2003) and the related contrastivist position put forward by Schaffer (2004; 2005). Neta’s view is distinctive in being a specifically evidential version of the view, one that allows agents to have factive evidential support for their empirical beliefs relative to everyday contexts of epistemic appraisal. Schaffer’s proposal is that knowledge is to be understood as an essentially contrastive notion, with the relevant contrast determined by the context of epistemic assessment. I will be exploring these proposals in chapter 6 in the context of epistemological disjunctivism (where Neta’s view will be described as rational support contextualism).

19. That’s not quite right, of course, in that all we have actually shown is that what is in effect an instance of the one principle in the context of the skeptical problem entails a corresponding instance of the other (which, given that we are interested in these principles only in terms of their application to the skeptical problem, is really all we need for our purposes). But it ought to be clear nonetheless from the argument just given that the general point that the closure_RK principle entails the underdetermination_RK principle holds. Here is the argument, where p is some arbitrary proposition and q depicts a scenario known by the subject to be incompatible with p (note that I’ve kept the original numberings for the claim, even despite the shift in content):

(P1) S has rationally grounded knowledge that p. [Premise]

(P2) S has rationally grounded knowledge that not-q. [From (P1), Closure_RK]

(P3) S’s rational support favors her belief that p over q. [From (P1), (P2)]

20. Of course, in principle at least, it is possible to reject the underdetermination_RK principle even while leaving intact a weaker restricted version of this principle, which nonetheless generates the same radical skeptical puzzle. Put more precisely, then, what is required is not merely a rejection of the underdetermination_RK principle, but rather a rejection of this principle that specifically blocks the skeptical puzzle generated by this principle. That is, one needs to expressly argue that rationally grounded knowledge of a quotidian proposition, such as that one is presently seated at one’s desk, is compatible with a lack of favoring rational support for this belief over skeptical alternatives like the BIV scenario. (The same point applies, mutatis mutandis, to responses to the closure_RK-based skeptical paradox that proceed by rejecting the closure_RK principle.)

21. There are a number of puzzles concerning the apparent rationality of believing incompatible propositions, including known to be incompatible propositions, such as the original (Kyburg) lottery puzzle and the preface paradox. So, to take the example of the preface paradox, it seems that I can have a rational basis for believing that there are some mistakes in my book, while also having a rational basis for believing each and every proposition in that book. But note that the incompatibility in question here is not in direct conflict with the rational ground or rational ground* principles. In order to generate a direct conflict one would need to add additional claims—such as that where one has an individual rational basis for belief in a series of propositions one thereby has a collective rational basis for believing the set of propositions as a whole (this is known as the “agglomeration principle” in the literature)—but these additional claims are themselves contentious. In any case, I will be setting this broader issue to one side in what follows. The locus classicus for discussions of the lottery paradox is Kyburg (1961). For further discussion of this paradox, see Sorensen (2002; 2011, §3). The locus classicus for discussion of the preface paradox is Makinson (1965). For two helpful overviews of this paradox in a specifically epistemological context, see Sorensen (2002; 2011, §4). See also Christensen (2004) and Kaplan (2013) for two very different accounts of the ramifications the rejection of the agglomeration principle has for our understanding of the requirements of rationality. I discuss the relevance of preface-style paradoxes to a certain form of disagreement-based skepticism in Pritchard (2013b).

22. Strictly speaking, the negation of (P3*) is the claim that it is not the case that S has rational support for the belief that she is a BIV, rather than what we have here (viz., that S does not have rational support for the belief that she is a BIV). Although generally speaking one should be wary about importing negations into claims in this way, it ought to be clear that nothing rides on our doing so in this particular case. If readers prefer, they can suppose that there is an intermediate step between (P4*) and (P5*), whereby we move from the strictly formulated negation of (P3*) to the formulation offered in (P5*).

23. As noted above—see note 19—the foregoing doesn’t strictly speaking demonstrate that these two principles are logically equivalent, in that all that has actually been shown is that an instance, in the context of the skeptical problem, of the one principle is logically equivalent to a corresponding instance of the other principle. It is, however, really only the equivalence between the two principles in the skeptical context that matters to us. Moreover, and again as before, the general claim nonetheless holds (in this case the logical equivalence), and holds on the same grounds. So, for the purists, here is the entailment from the underdetermination_RK principle to the rational ground* principle, where (as before) p is some arbitrary proposition and q depicts a scenario known by the subject to be incompatible with p (and again, as before, note that I’ve kept the original numberings for the claim, even despite the shift in content):

(P1*) S has rationally grounded knowledge that p. [Premise]

(P2*) S’s rational support favors her belief that p over the alternative that q. [From (P1*), Underdetermination_RK]

(P3*) S has rational support for believing that q. [Assumption for Reductio]

(P4*) S’s rational support does not favor her belief that p over the alternative that q. [From (P3*)]

(P5*) S does not have rational support for believing that q. [From (P2*), (P3*), (P4*), RAA]

And here is the entailment from the rational ground* principle to the underdetermination_RK principle:

(P1**) S has rationally grounded knowledge that p. [Premise]

(P2**) S does not have rational support for believing that q. [From (P1**), Rational Ground*]

(P3**) S has rational support for her belief that p. [From (P1**)]

(P4**) S’s rational support favors her belief that p over the alternative that q. [From (P2**), (P3**)]

24. Indeed, one can think of the insularity of reasons thesis as the epistemic aspect of the more general idea of the “veil of perception,” an idea that also has a metaphysical component, concerning the very nature of our perceptual experiences. Like the metaphysical aspect of this problem, which maintains that our perceptual experiences fall short of direct contact with an external world, the insularity of reasons thesis highlights a sense in which the rational standing of our perceptual beliefs regarding the external world also falls short, in that even in the best case it is compatible with widespread falsity in one’s beliefs about the nature of that world and one’s relationship to it.

25. The loci classici in this regard are Lehrer & Cohen (1983) and Cohen (1984). For a helpful general discussion of the new evil demon intuition and its epistemological significance, see Littlejohn (2009). See also Bach (1985) and Engel (1992).

Chapter 3. Wittgenstein on the Structure of Rational Evaluation

1. The foregoing gloss on Moore’s general line with the skeptic, while broadly accurate, nonetheless misses out a number of subtleties regarding Moore’s position, particularly with regard to his “Proof of an External World” (Moore 1939). For a more detailed discussion of Moore’s contributions to epistemology, see Baldwin (1990, chap. 9). For a recent survey of contemporary work on Moore’s proof, see Carter (2012).

2. Indeed, as we will see, Wittgenstein would similarly claim that the game of believing presupposes certainty too.

3. There are obvious affinities here with the anti-skeptical use of the principle of charity made by Davidson (e.g., 1977; 1983). I will be commenting on these affinities in chapter 4, §5.

4. Although the “hinge” metaphor is the dominant symbolism in the book, it is accompanied by various other metaphors, such as the following: that these propositions constitute the “scaffolding” of our thoughts (OC, §211); that they form the “foundations of our language-games” (OC, §§401–3); and also that they represent the implicit “world-picture” from within which we inquire, the “inherited background against which [we] distinguish between true and false” (OC, §§94–95). Henceforth, I will follow most commentators of On Certainty in making use of the hinge metaphor. As will become apparent in due course, however, this metaphor may not be the most apt, in that it implies that we can change our hinge commitments at will (just as we might change the hinges on a door). At least on my reading of Wittgenstein, in contrast, there is nothing remotely optional about our hinge commitments.

5. There is, of course, far more to be said about Austin’s treatment of skepticism, though this is not the place to undertake such an exploration. For some helpful recent discussions of Austin’s epistemology, see the exchange between Travis (2005a) and Millar (2005), and also Travis (2008), Kaplan (2011a; 2011b), and Lawlor (2013). See also Longworth (2012, §3).

6. I further discuss this aspect of Wittgenstein’s thinking, in the context of the very different treatments of radical skepticism offered by Austin (1961) and Stroud (1984), in Pritchard (2011f, §1; 2014c). I return to this contrast between Wittgenstein’s approach to radical skepticism and that associated with ordinary language philosophy below—see chapter 6—as part of a critique of rational support contextualism and attributer contextualism.

7. This is sometimes known as “Russell’s Hypothesis”—see Russell (1921, 159).

8. A further worry about the Wittgensteinian account of the structure of rational evaluation is that it might license epistemic relativism. For on this view can’t there be distinct epistemic systems that incorporate different hinge commitments and thus generate opposing rational beliefs? But if that is possible, then how is a disagreement among two parties who adhere to these respective epistemic systems to be rationally resolved? We will be exploring this issue in its own right in the next chapter (§3) in the context of the inferential contextualist view that is advocated by Williams (1991) and that is inspired by Wittgenstein’s remarks on hinge propositions in On Certainty. For an important recent discussion of the relationship between Wittgenstein’s stance on hinge commitments and epistemic relativism, see Williams (2007), to which Pritchard (2010c; cf. Pritchard 2009b) is effectively a response. See also Kusch (2010) and the significant new treatment of this issue offered by Coliva (2010, passim).

9. Just to be clear: in calling this view, and other subsequent views, a “reading” I am not suggesting that it is being explicitly put forward as an interpretation of Wittgenstein’s On Certainty. Often, as in this case, such “readings” are merely inspired by the text.

10. For further discussion of the epistemic externalist development of Wittgenstein’s account of the structure of rational evaluation, see Pritchard (2001b; 2005g; 2011f; forthcoming–d). For an influential contextualist variant of the externalist reading, see Williams (e.g., 1991). I discuss Williams’s view further below—see especially chapter 4, §3 and §7.

11. Note that DeRose’s own “abominable conjunctions” were specifically targeted at putative counterexamples to the closure principle for knowledge rather than the closure principle for rationally supported knowledge (i.e., the closure_RK principle).

12. For a key defense of process reliabilism, see Goldman (1986). For a very useful recent survey of work on reliabilism, see Goldman (2008).

13. For a prominent defense of virtue epistemology explicitly cast along epistemic externalist lines, see Greco (1999; 2000; 2003; 2007). See also Sosa (1991; 1998; 2007; 2009; 2011) for a defense of an influential version of virtue epistemology that is broadly speaking in the epistemic externalist camp. See Axtell (1997) for a helpful critical overview of epistemic externalist and epistemic internalist versions of virtue epistemology. For more on (basis-relative) modal conditions on knowledge, see Nozick’s (1981) defense of a sensitivity condition, and Sosa’s (e.g., 1999) defense of a safety condition. I explore the prospects of such modal conditions on knowledge—including specifically safety-based theories of knowledge—in a number of places. See Pritchard (2002d; 2005b, passim; 2007a; 2008d; 2009c, passim; 2009d; 2012c; 2013c; forthcoming–a).

14. The foremost exponent of this kind of proposal is Wright. See, especially, Wright (2004c; cf. Davies 2004). See also Wright (1985; 1991; 2000; 2002; 2003a; 2003b; 2004a; 2004b; 2007; 2008a; 2008b). For some helpful discussions of Wright’s proposal, see Davies (2004), Coliva (2012), Pryor (2012), Williams (2012), and Zalabardo (2012). Note that this notion of epistemic entitlement is very different from that defended by Burge (1993; 2003) and Peacocke (2003). See Altschul (2011) for a useful survey of these three different conceptions of epistemic entitlement.

15. The entitlement strategy is often formulated in terms of the rejection of the “transmission” principle for knowledge rather than in terms of the rejection of what we are terming the closure_RK principle. As noted in chapter 1—see note 20—the two principles are subtly different, with the closure_RK principle being if anything the (marginally) logically weaker principle of the two. Given their similarity, and the fact that of the two the closure_RK principle is the logically weaker if either is, for our purposes nothing of consequence is lost by focusing on the closure_RK principle.

16. In earlier work—see, for example, Pritchard (2005g)—I pressed such a claim myself. See also Jenkins (2007) and Pedersen (2009).

17. Even Williamson (2000a), who holds that knowledge cannot be analyzed in terms of belief plus some other conditions (such as truth and an epistemic condition like justification), nonetheless holds that knowledge entails belief. See Williamson (2000a, §1.5). For a useful taxonomy of different ways of thinking about belief, see Stevenson (2002).

18. In this regard, consider Radford’s (1966) famous example of the diffident schoolboy who knows the answer to the question he is asked, but who, it is claimed, doesn’t believe it because he doesn’t think he knows it. Even if we grant that there is knowledge in the absence of belief here (which is of course contentious), it remains that the schoolboy is not at all agnostic about the truth of the target proposition—of all the options available, it is explicit to the example that he is inclined toward regarding a very specific answer as being the correct one. The point of the case is not that the schoolboy is agnostic about whether the answer he gives is correct (since that would undermine the claim that the schoolboy has knowledge), but rather that he doesn’t have the kind of confidence in this regard that we would usually associate with having knowledge.

19. See in particular Wright (2004c, 194) for his discussion of “rational trust” in this regard. (The “rational” element in “rational trust” will be explored in more detail below.) See also Wright (2008b).

20. The avowedly transcendental response to radical skepticism offered by Sosa (2011, chap. 8) has some commonalities with Wright’s entitlement proposal, particularly to the extent that one is meant to be able to derive a rational basis for rejecting skeptical hypotheses from the fact that endorsing them would lead to incoherence. As such, it suffers from many of the problems that we have seen beset Wright’s account, as I explain in Pritchard (forthcoming–b). For a more sympathetic treatment of Sosa’s proposal, see Neta (forthcoming).

21. In the contemporary literature, the entitlement reading is often contrasted with a proposal known as dogmatism. See, for example, Pryor (2000; 2004; 2012). See also Wright (2007) for a response to dogmatism on behalf of the main proponent of the entitlement reading, and Williams (2013) for a helpful critical appraisal of both Wright’s view and the alternative dogmatist proposal. Since dogmatism is not offered as a view that is inspired by Wittgenstein’s On Certainty, I have not explored it here. I will, however, be discussing dogmatism in part 3—see chapter 6, §3—where I will be critically contrasting it with an epistemological disjunctivist account of perceptual knowledge.

22. This type of proposal has been recently expounded in some detail in Moyal-Sharrock (2004). See also the highly influential earlier work by McGinn (1989; cf. McGinn 2008; 2011) which takes a similar line. See also Wright (1985), Stroll (1994; 2005) and Phillips (2005). For a very useful critical discussion of the nonpropositional reading, see Coliva (2010).

23. As a number of commentators have noted (e.g., Williams 2004d), we need to carefully distinguish between Wittgenstein’s remarks in the first notebook that makes up On Certainty (i.e., up to §65), and the rest of the text. This is because the first notebook seems more concerned with Moore (1925) than with the later Moore (1939). This is significant for our purposes because the first Moore text is effectively dealing with idealism, and thus with the assertion of claims which purport to refute this philosophical thesis, while it is the latter text which is focused on the kind of everyday certainties which we now think as distinctively “Moorean.” Wittgenstein’s line in the first notebook regarding Moore’s rejection of idealism is very clearly that Moore’s assertions in this respect are nonsense. In particular, the claim is not that Moore has taken ordinary empirical expressions which have a legitimate use in normal contexts and employed them out of context, but rather that he is putting normal terms to an entirely illegitimate theoretical use. Here is Wittgenstein:

“A is a physical object” is a piece of instruction which we give only to someone who doesn’t yet understand what “A” means, or what “physical object” means. Thus it is instruction about the use of words, and “physical object” is a logical concept. (Like colour, quantity, …) And that is why no such proposition as: “There are physical objects” can be formulated.

Yet we encounter such unsuccessful shots at every turn. (OC, §36; cf. §§35 & 37)

In saying that physical objects exist, as if this were a normal empirical claim, Moore is thus simply making a meaningless assertion. Moreover, notice that in contrast with an assertion like “I have two hands,” Wittgenstein’s contention is that there is no conversational context where this assertion expresses an ordinary empirical proposition. Given these clear differences between the target of the first notebook of On Certainty and the later notebooks, I think we would do well to bracket the “nonpropositionalism” of the first notebook. In particular, one might agree with Wittgenstein that a statement like “There are physical objects” is meaningless and yet nonetheless be willing to argue that a statement like “I have two hands,” even when asserted in a Moorean fashion to make an anti-skeptical point, is nonetheless meaningful.

24. Though see Coliva (2010) for an insightful critique of the nonpropositional reading of On Certainty. See also my remarks in note 23 above.

25. Of course, there is also the further issue of how best to understand the nonsense that results on this view when one attempts to express one’s hinge commitment in the form of an assertion. Is it, for example, just plain nonsense, or is it somehow philosophically significant nonsense (if such a thing is possible)? For more on this issue, see the exchange between Conant (1998) and McGinn (2002).

26. For a development of a radically context-sensitive account of meaning which is in large part inspired by Wittgenstein’s work, see Travis (1989; 2001; 2006). See also Travis (2008), where the influence of Austin’s work is more in the foreground. For discussion of the anti-skeptical merits of this way of thinking about meaning, see Putnam (2001; 2012) and the exchange between Travis (2005a) and Millar (2005). See also Williams (2004a).

Chapter 4. Hinge Commitments

1. See note 8 for a qualification of the claim that our hinge commitments are in their nature unresponsive to rational considerations. Also, remember that we are treating the notion of belief in play, unless qualified, as knowledge-apt belief (i.e., the specific notion of belief which is a constituent part of knowledge).

2. Of course, exactly how one should understand the claim that beliefs are in their nature responsive to rational considerations is controversial—what it means, to use the parlance in the literature, for the telos of belief to be truth—but we do not need to get into that issue here. For our purposes we only need to note that a propositional attitude which is completely unresponsive to rational considerations, and which cannot be the result of a rational process, would not qualify as a belief—at least insofar as we are interested in the notion of belief that is component part of knowledge—and that claim is surely uncontroversial. For some of the key works on the debate about the relationship between truth and belief see Wedgwood (2002), Shah (2003), and Shah & Velleman (2005). See also Pritchard (2011c; 2014d) and the essays collected in Matheson & Vitz (2014).

3. This particular line of response was put to me by Crispin Wright.

4. Note that the fourfold taxonomy of readings of On Certainty offered in this chapter and the last is not meant to be exhaustive, but merely representative of some of the main lines of thought in this regard. For example, there is also the so-called therapeutic reading of On Certainty which is offered by Conant (1998), whereby the radical skeptic’s assertions are strictly senseless. Although this has some affinities with the nonpropositional reading which we discussed in the last chapter, many have distinguished these two proposals (e.g., Coliva 2010). For an intriguing feminist interpretation of Wittgenstein’s work as whole, including On Certainty—which is perhaps best described more broadly as an anti-modernist reading—see Tanesini (2004). For an excellent recent discussion of the various interpretations of On Certainty in the contemporary literature, see Coliva (2010). See also Brenner & Moyal-Sharrock (2005) and Pritchard (2011f; forthcoming–a). Note that the “inferential contextualist” reading of On Certainty offered by Williams (1991) will be discussed below in §3 and §7, and the “naturalistic” reading of On Certainty offered by Strawson (1985) will be discussed in §4.

5. I say “radically and fundamentally” here to stress both the extent and the depth of the error involved. Note that there are technical issues in play here regarding how best to “weigh” the overall truth of one’s beliefs. In particular, one cannot simply do so by counting the number of true beliefs that one has (insofar as one can make any sense of the idea counting beliefs in the first place), although it is often implicitly supposed by contemporary epistemologists that this is possible. For discussion of this point, which bears on the so-called trivial truths problem in epistemology, see Treanor (2014). See also Pritchard (2014d). For further discussion about how one is to “weigh” truth in one’s beliefs given that one cannot do it in purely numerical terms, see Treanor (2013). For two different perspectives, see Sider (2011) and Hazlett (2014).

6. This is a point that Wittgenstein emphasizes on a number of occasions. Consider, for example, this passage:

If my friend were to imagine one day that he had been living for a long time past in such and such a place, etc. etc., I should not call this a mistake, but rather a mental disturbance, perhaps a transient one.

Not every false belief of this sort is a mistake. (OC, §§71–72; cf. OC, §§54; 155–58)

7. In particular, we should be wary about construing our hinge commitments in such a way that they license epistemic relativism. I will be returning to this issue in the next section when we will be exploring the inferential contextualist view that is advocated by Williams (1991) and that draws its inspiration from Wittgenstein’s remarks on hinge propositions in On Certainty.

8. This is the extent to which one’s hinge commitments can be indirectly responsive to rational considerations. Rational considerations can lead one to change one’s beliefs, after all, and as one’s beliefs change so the personal hinge commitments that codify one’s (unchanging) über hinge commitment might alter. The point remains, however, that even one’s personal hinge commitments are not directly responsive to rational considerations in the way that one’s beliefs are.

9. Interestingly, notice that by the lights of the nonbelief reading, we can in fact capture a version of the universality of rational evaluation thesis that is harmless. For while we are rejecting the idea of universal rational evaluations—where this covers all the propositions we are committed to (i.e., that we regard as true)—we have also noticed that a significant body of our commitments are not to be thought of as (knowledge-apt) beliefs. Thus, insofar as we think of the universality of rational evaluation thesis as only applying to beliefs, then it can be endorsed, even by Wittgensteinian lights. Such a formulation of the universality of rational evaluation thesis would be entirely harmless from a skeptical point of view, however, in that one could not use it to motivate, via the closure_RK principle, the relevant bridging claim in the closure-based skeptical argument (because this bridging claim by its nature involves the rational evaluation of a hinge commitment). See also note 14 below.

10. Or, at least, whatever restrictions there are won’t be epistemic ones. There might be psychological restrictions on inferences of this kind, for example, in that at some point one is unable even to form the relevant beliefs since their contents have become too complex.

11. Stroud (1984) defends a simple version of the closure_RK principle*, which he attributes to Descartes. The advantage of this principle over closure-style principles that don’t entail iterativity for rationally grounded knowledge is that it can be used to generate radical skepticism that is based both on those radical skeptical scenarios that trade on massive error (such as the ones we are focusing on here, like the BIV scenario) and those radical skeptical scenarios that don’t trade on massive error (such as dreaming-based radical skepticism). For critical discussion of this type of closure principle in the context of Stroud’s proposal, see Wright (1991; cf. Pritchard 2001c). For further discussion of Stroud’s approach to radical skepticism in general, see Pritchard & Ranalli (forthcoming–a). For an influential recent treatment of dreaming-based radical skepticism, see Sosa (2007, chap. 1). For more on iterativity for knowledge, see Kelp & Pedersen (2011).

12. For further critical discussion of multi-premise closure-style principles, and their relation to their single-premise counterparts, see Hawthorne (2004, passim) and Collins (2006, §1). See also Pritchard (2013b) for a critical discussion of the status of multi-premise closure in the context of just the sort of “track-record”-style argument that is being envisaged here.

13. One might be tempted to classify our hinge commitments as aliefs. This is a kind of belief-like propositional attitude that has been most prominently defended by Gendler (2008a; 2008b). Very roughly, aliefs are spontaneous judgments that are (at least typically) in tension with what one actually believes. For example, at high altitude and looking down, and overcome with acrophobia, one might instinctively judge that one is in danger, even while being perfectly aware that one is entirely safe. The instinctive judgment that one is in danger is in this case an alief, whereas the judgment that one is safe is a belief. Given that our hinge commitments are entirely visceral, there is certainly a prima facie case for regarding them as aliefs. Ultimately, however, I think that it is probably best to not treat them as aliefs. For one thing, as just noted, aliefs are typically understood as being in tension with our beliefs, but of course our hinge commitments by definition cannot be in tension with what we believe. Moreover, and I think more importantly, there is nothing in the notion of an alief that means that it is in its nature unresponsive to rational considerations (though in practice they often are, as the acrophobia case just give illustrates). In contrast, this is a defining feature of our hinge commitments. For some useful critical studies of alief, see Nagel (2012) and Mandelbaum (2013). For a helpful overview of different ways of thinking about beliefs, see Stevenson (2002). See also chapter 7, note 9.

14. An alternative way of putting this point—as suggested in note 9 above—is that the universality of rational evaluation thesis is ambiguous in a crucial respect. Construed as a claim about the rational evaluation of one’s beliefs, specifically, it is not in conflict with Wittgenstein’s account of the structure of rational evaluation, but neither can it be used to generate the radical skeptical paradox. In order for it to have this import to the skeptical problem, it needs to be construed as a more general thesis about our commitments (i.e., not just our beliefs, but also our hinge commitments). But so construed the thesis is in conflict with the Wittgensteinian account of the structure of rational evaluation.

15. I have developed the nonbelief reading of Wittgenstein’s On Certainty in a number of works. See especially Pritchard (2012e; 2014a; forthcoming–d). For an application of this proposal to the epistemology of religious belief, see Pritchard (2011e; forthcoming–c). See also Coliva (2010) for the development of a similar reading of On Certainty that retains many of the main features of the non-propositional reading while nonetheless allowing that our hinge commitments can be cast along propositional lines. Since developing the nonbelief reading, I have come across remarks in Campbell (2001, 96) which might suggest a similar approach to our hinge commitments. In particular, in setting forth a particular account of delusions, he discusses these commitments and notes that they are “not normal factual beliefs,” which comes very close to making the point that I want to make that they are not to be thought of in terms of knowledge-apt belief.

16. See especially Williams (1991). See also Williams (1988a; 1993; 2001b), and his earlier monograph, Williams (1977). For a useful symposium on Williams (1991), see the exchange among Rorty (1997), Vogel (1997), and Williams (1997). See also the discussion of this exchange offered by Pritchard & Ranalli (2013). I discuss Williams’s treatment of radical skepticism in a number of places—see, for example, Pritchard (2002e; 2005d; 2010c; 2011f).

17. I first described Williams’s view as inferential contextualism in Pritchard (2002e). It is interesting question how attributer contextualism and inferential contextualism relate to one another. One overarching difference between the two views is that whereas the former is primarily a semantic thesis, the latter is an epistemological thesis (at least insofar as this characterization includes the metaphysics of epistemology). So even if, for example, inferential contextualism turned out to entail a form of attributer contextualism, as is plausible, it would still be a mistake to conclude that the former is just a variant of the latter. I comment on a further difference between the two views in the main text below. For additional discussion of these two forms of contextualism, see Williams (2001a; 2004b; 2004c). See also Pritchard (2002c; 2002e).

18. For a different characterization of Williams’s conception of epistemic priority, see Ribeiro (2002).

19. A good point of comparison in this regard—a comparison that Williams himself draws—is the contextualist account of justification offered by Annis (1978).

20. Or, at the very least, Williams regards closure-based radical skepticism as simply being a variant of underdetermination-based radical skepticism rather than logically distinct, where the latter connects most directly with the “veil of perception,” and thus with issues concerning epistemic priority that are Williams’s focus. See, for example, Williams (1991, chap. 8) where he discusses the former form of radical skepticism in some detail. See also Williams (2010, 196).

21. This won’t come as a surprise to Williams, for he is quite explicit that his notion of a methodological necessity, while inspired by Wittgenstein’s remarks on hinge commitments in On Certainty, is not meant to be an interpretation of this notion. See Williams (1991, chap. 1).

22. Similar points apply to other putative methodological necessities of historical inquiry (or, for that matter, any other specific inquiry). Consider, for example, Williams’s claim that one such methodological necessity of this kind of inquiry concerns the general veracity of historical documentation. Notice that if there were a systematic deception in play with regard to all “official” testimony regarding the past, then that would almost certainly be in conflict with one’s über hinge commitment. A commitment to the absence of such a systematic deception is thus a plausible manifestation of one’s general über hinge commitment. It follows that one will tend to regard historical documentation as generally veracious. By casting the commitment in question as being specifically concerned with historical documentation, Williams makes it look as if this is a commitment that is peculiar to a particular context of inquiry. But closer inspection of the kind of commitment in play reveals that it is no such thing, but rather just the manifestation of the more general über hinge commitment.

23. We will return to the issue of a hierarchy of contexts as it applies to attributer contextualism in chapter 6, where we will be examining a related view called rational support contextualism.

24. That his conception of the structure of rational evaluation might lead to epistemic relativism of this kind is certainly a problem that Wittgenstein grapples with in On Certainty. Consider, for example, this famous passage:

Where two principles really do meet which cannot be reconciled with one another, then each man declares the other a fool and a heretic.

I said I would “combat,” the other man,—but wouldn’t I give him reasons? Certainly; but how far do they go? At the end of reasons comes persuasion. (Think of what happens when missionaries convert natives.) (OC, §§611–12)

I explore Wittgenstein’s treatment of epistemic relativism in more detail in Pritchard (2010c). Note that there are interesting issues here regarding Wittgenstein’s broader approach to the epistemology of religious belief—not just in On Certainty, but also as it figures in his other work, such as Wittgenstein (1966)—and how this interrelates both with his views about hinge commitments in On Certainty and with topics such as epistemic relativism. See Nielsen (1967) and Phillips (1976) for two influential accounts of Wittgenstein’s putatively fideistic approach to the epistemology of religious belief. I offer my own particular quasi-fideistic reading of Wittgenstein in this regard in Pritchard (2011e). See also Pritchard (forthcoming–c; cf. Pritchard 2000c), where I examine the relevance of Wittgenstein’s views in this regard in the context of the religious epistemology proposed by John Henry Newman (1979), an epistemology that was arguably a key influence on Wittgenstein’s thinking in On Certainty. See also Kienzler (2006), which offers a fascinating discussion of the influence of Newman on Wittgenstein, and Plant (2011), which offers a naturalistic reading of Wittgenstein’s epistemology of religious belief that he claims enables Wittgenstein to avoid charges of epistemic relativism.

25. Oddly, later on in the very same page cited here Williams seems to assert the very opposite of this claim and deny that there can be epistemic incommensurability. I discuss Williams’s ambivalent approach to these issues, including the logical tension between these two passages, at length in Pritchard (2010c). See also note 26.

26. It should be noted that Williams’s views on how inferential contextualism and epistemic relativism interrelate are complex. He has argued at length that the former is in fact the “antidote” to the latter (see Williams 2007), and yet as we have noted he also seems to endorse epistemic incommensurability, which is the very kind of thing that generates epistemic relativism. I explore Williams’s account of how inferential contextualism relates to epistemic relativism in detail in Pritchard (2010c). See also note 25.

27. Note that we will be returning to further consider Williams’s inferential contextualism below (§7), when we will be examining how it fares with regard to underdetermination_RK-based radical skepticism.

28. As Byrne (2004, 302) quips, the idea that radical skeptical doubts are idle in this sense, and hence can be ignored, “is a curious view for an erstwhile Waynflete Professor of Metaphysics, to say the least.” For more on Strawson’s response to radical skepticism, see the exchanges between Putnam (1998) and Strawson (1998a) and between Sosa (1998) and Strawson (1998b). One interesting exegetical question when it comes to Strawson’s later naturalistic response to radical skepticism is how it relates to his earlier treatment of this problem in Strawson (1959). For further discussion of this question, see Callanan (2011). See also Stern (2003) and Cassam (2008).

29. We should note an affinity between the anti-skeptical stance offered by Strawson and the entitlement reading of On Certainty that we explored earlier. In particular, we noted that the most prominent exponent of the entitlement reading explicitly advances this view as a response to a genuine problem posed by the radical skeptic. Here is the relevant passage from Wright again, arguing that the entitlement strategy (as we have described it)

concedes that the best sceptical arguments have something to teach us—that the limits of justification they bring out are genuine and essential—but then replies that, just for that reason, cognitive achievement must be reckoned to take place within such limits. The attempt to surpass them would result not in an increase in rigour or solidity but merely in cognitive paralysis. (Wright 2004c, 191)

Like Strawson’s naturalistic reading of On Certainty, the entitlement strategy concedes that the skeptical paradox is a bona fide paradox, but argues on independent theoretical grounds that there is a revisionary way of approaching matters that resolves the puzzle. In this sense, Strawson’s naturalistic interpretation of On Certainty is perhaps better grouped with the entitlement reading than the nonbelief reading.

30. For the twists and turns, see Pritchard (2013a). See also note 31.

31. Davidson’s initial inclination was to couch this anti-skeptical proposal in terms of an appeal to an “omniscient interpreter” (see Davidson 1977; 1983), but in light of various critiques of this way of expressing the argument, he retreated to the broadly transcendental version of his anti-skepticism that we will be exploring here. See especially Davidson (1999), which is a response to Stroud (1999). For some of the main critiques of the omniscient interpreter argument, see, for example, Vermazen (1983), Foley & Fumerton (1985), Brueckner (1986; 1991), Rorty (1986), Williams (1988b), and Ludwig (1992). For more on Davidson’s approach to radical skepticism more generally, see Pritchard (2013a) and Pritchard & Ranalli (2013).

32. Indeed, as noted above—see notes 9 and 14—there is a plausible rendering of the universality of rational evaluation thesis that is compatible with the Wittgensteinian account of the structure of rational evaluation, at least on the nonbelief reading, so there is a sense in which the proponent of this view can retain a commitment to this thesis.

33. Or, at the very least, he regards closure-based radical skepticism as simply being a variant of underdetermination-based radical skepticism, rather than a logically distinct radical skeptical paradox. See note 20.

Chapter 5. Epistemological Disjunctivism and the Factivity of Reasons

1. Note that epistemological disjunctivism should be kept apart from the more familiar form of disjunctivist view in the philosophy of perception, which is best classed as a metaphysical rather than epistemological thesis. It would take us too far afield to discuss metaphysical disjunctivism here, much less the logical relationship between this form of disjunctivism and its epistemological relative. I discuss these questions at some length in Pritchard (2012b, pt. 1), where I argue that epistemological disjunctivism doesn’t entail metaphysical disjunctivism. See also Pritchard (2008b) and Pritchard & Ranalli (forthcoming–b). This nonentailment claim has also been defended (at least in its most general form) by Snowdon (2005), Millar (2007; 2008b), Byrne & Logue (2008), and McDowell (2008, 382n). In particular, the claim in this regard is often that epistemological disjunctivism is compatible with a causal theory of perceptual experience, as defended, for example, by Grice (1961) and Strawson (1974). For further discussion of the logical connections between metaphysical and epistemological disjunctivism, see Haddock & Macpherson (2008b), Byrne & Logue (2009b), Fish (2009), and Soteriou (2009, esp. §2.4). See also Brogaard (2010) and Dorsch (2011). For some of the key defenses of metaphysical disjunctivism, see Hinton (1967a; 1967b; 1973), Snowdon (1980–81; 1990–91), and Martin (1997; 1998; 2002; 2004; 2006). See also the essays collected in Haddock & Macpherson (2008a) and Byrne & Logue (2009a). For an important recent exchange on metaphysical disjunctivism, see Hawthorne & Kovakovich (2006) and Sturgeon (2006).

2. Indeed, I know from personal correspondence that there are some features of epistemological disjunctivism, as I describe the view, that McDowell would not endorse. I discuss how epistemological disjunctivism is rooted in McDowell’s work in Pritchard (2012b, passim). See also Pritchard (2003; 2007b; 2008b; 2009e) and Neta & Pritchard (2007).

3. For a key defense of such an “accessibilist” conception of rational support, see Chisholm (1977). See also BonJour (1985, chap. 2) and also chapter 1, note 15. For further discussion of the epistemic externalism/internalism distinction in light of epistemological disjunctivism, see Neta & Pritchard (2007) and Pritchard (2007b; 2008b; 2009e; 2011b; 2012b, passim).

4. For further discussion of “good” and “bad” cases, including a taxonomy of cases of these kinds, see Pritchard (2011a; 2012b, pt. 1).

5. Or, at least, that what differences there are in the degree of reflectively accessible rational support available to the two subjects, they are indistinguishable differences.

6. A distinctive element of metaphysical disjunctivism—see note 1—is the idea that there is no common metaphysical component to one’s perceptual experience across pairs of good and bad cases (at least bar the trivial epistemological point that such experiences are ex hypothesi indistinguishable). It thus follows that the nature of one’s perceptual experience is fundamentally very different depending on whether one is in the good case or the bad case—hence, the disjunctivism. Similarly, epistemological disjunctivism maintains that the rational support for one’s belief is fundamentally very different (i.e., with no common component) depending on whether one is in the good case or the bad case, contra the new evil demon intuition.

7. See especially Pritchard (2012b, passim).

8. For some key defenses of this standard view about the relationship between seeing that p and knowing that p, see Dretske (1969, 78–139), Williamson (2000a, chap. 1), and Cassam (2007a; 2007b).

9. See Pritchard (2011a; 2012b, pt. 1) for a fuller version of this line of argument. Incidentally, McDowell (2002b, 277–79) appears to agree that seeing that p isn’t a form of knowing that p, and for similar reasons. See also Turri (2010) for a defense of a related view, to which French (2012) responds, and Sosa (2011, chap. 4). For further critical discussion of the putative entailment from seeing that p to knowing that p, see Ranalli (2014). For a recent critique of my response to the basis problem, see Littlejohn (2013).

10. See McKinsey (1991). For a recent set of discussions of this tension between first-person authority and content externalism, see Nuccetelli (2003). See also Pritchard (2002a).

11. For a more developed account of the access problem and why it is illusory, see Pritchard (2012b, pt. 1). See also Neta & Pritchard (2007).

12. “Undeniable” might be too strong, as a certain kind of radical metaphysical disjunctivism might well be willing to deny this claim. But recall that we are here considering only epistemological disjunctivism, and setting the status of metaphysical disjunctivism to one side (see note 1). More generally, we are aiming to defend, as best we can, an anti-skeptical proposal that does not appeal to any further contentious theses from outwith epistemology.

13. The zebra/cleverly disguised mule example is, of course, due to Dretske (1970).

14. I offer a more developed account of the distinction between favoring and discriminating epistemic support in Pritchard (2010d). See also Carter & Pritchard (forthcoming). For further discussion of how this distinction helps epistemological disjunctivism to avoid the distinguishability problem, see Pritchard (2012b, pt. 2).

15. Note that for ease of expression I have renamed the three claims that make up this skeptical paradox, compared with the formulation that was offered in chapter 2. Nothing of any consequence hangs on these changes.

16. McDowell motivates the idea that epistemological disjunctivism (our terminology rather than his) is rooted in our folk epistemic concepts by showing how the philosophical reasoning that rejects this proposal trades on a style of argument—the “highest common factor” argument—that he argues is dubious. See, especially, McDowell (1995; cf. McDowell 2002a). I sympathetically discuss McDowell’s treatment of this point in Pritchard (2012b, pt. 1).

17. That we can appeal to this distinction to deal with local skeptical scenarios is important to our understanding of the skeptical problem though, since with this distinction in place the case for some varieties of anti-skepticism is severely undermined. For more on this point, see Pritchard (2010d).

18. For more on this point, see the taxonomy of “good” and “bad” cases, and the different ways in which they bear on the subject’s rational position, as set out in Pritchard (2012b, pt. 1). See also Pritchard (2011a).

19. I think this point also explains why epistemological disjunctivism doesn’t face a version of the so-called paradox of dogmatism. Roughly, this paradox concerns the fact that if one has sufficient rational support to know that p, then why isn’t one in a position to thereby conclude that any evidence that not-p is misleading, and hence adopt a completely dogmatic stance with regard to p? One might think that epistemological disjunctivism faces an acute version of this problem, in that it allows that the rational support for one’s knowledge can be factive. It seems to follow that the epistemological disjunctivist cannot appeal to the idea that one’s rational support for p, while sufficient for knowledge that p, is nonetheless defeasible, which is a standard response to this puzzle. Once we spell out what is involved in being presented with evidence that not-p, however, this puzzle, as it specifically applies to epistemological disjunctivism anyway, disappears. On the one hand, if this just means the mere presentation of a not-p possibility, then there is nothing inherently suspect about the idea that our agent can continue to cite the factive rational support she has for the target proposition, and hence regard this error possibility as excluded. Effectively, she is treating this particular evidence for not-p as misleading (though “evidence” is not quite the right word, given that the error possibility isn’t rationally motivated). On the other hand, where the error possibility is rationally motivated, then our agent cannot simply continue to cite the factive rational support that she offered in support of her belief previously and must instead rationally engage with this error possibility. But in this scenario it’s also the case that she is no longer in possession of factive rational support, since the presence of a rationally motivated error possibility entails that the epistemic conditions are no longer paradigmatic. Either way, the possibility of factive rational support doesn’t generate the advertised dogmatic conclusion. For more on the paradox of dogmatism, see Harman (1973) and Kripke (2011). See also Sorensen (1988; 2011, §6.2).

20. For further discussion of some of the conversational constraints on appropriate knowledge claims that are relevant here, see Pritchard (2012b, pt. 3).

21. The mere presentation of a radical skeptical hypothesis thus doesn’t suffice to turn one’s epistemic situation into a suboptimal one (i.e., such that by epistemological disjunctivist lights factive rational support is no longer reflectively available). To argue otherwise is, I would contend, to concede far too much to the radical skeptic. For further discussion of the “good” cases in which one has reflectively accessible factive rational support for one’s beliefs by epistemological disjunctivist lights, see Pritchard (2012b, pt. 1). See also Pritchard (2011a).

22. I further explore these dialectical issues as part of a recent symposium on Pritchard (2012b)—see especially Goldberg (forthcoming), but also Littlejohn (forthcoming) and Neta (forthcoming), to which Pritchard (forthcoming–b) is a response. See also Zalabardo (forthcoming).

Chapter 6. Epistemological Disjunctivism and Closure-Based Radical Skepticism

1. Note that once we combine epistemological disjunctivism with a Wittgensteinian anti-skeptical proposal, which is what we will be doing in chapter 7, then we will have a further philosophical basis on which to critique attributer contextualism, in terms of its appeal to epistemically high-standards skeptical contexts of epistemic appraisal. Insofar as these contexts of epistemic appraisal are understood as attempting—whether directly or indirectly via a rational assessment of one’s hinge commitments—universal rational evaluations of one’s beliefs, then they are employing epistemic standards that are, by Wittgensteinian lights, inherently and fatally problematic. I return to this point in chapter 7. See also note 3 below.

2. Henceforth, when I refer to “attributer contextualism” without qualification, I will have the standard version of this view (as encountered in chapter 2) in mind.

3. Once we combine epistemological disjunctivism with a Wittgensteinian anti-skeptical proposal, which is what we will be doing in chapter 7, then we can motivate an even stronger point in this regard. According to the Wittgensteinian account of the structure of reasons, after all, there is something inherently dubious about the kind of epistemically high-standards contexts of epistemic appraisal employed by the skeptic in which doubt of our hinge commitments is in play. See chapter 7, §4.

4. This scenario is just a variant of a standard kind of case where one is in possession of misleading defeaters. It is an interesting question just what one’s epistemic position is in such a case, according to epistemological disjunctivism. It is clear that the subject no longer knows that p, since she doesn’t any longer believe that p, but that doesn’t settle what her epistemic position is in this regard. My own view, articulated at length in Pritchard (2011a; 2012b, pt. 1), is that in these conditions the subject continues to see that p, but that her seeing that p can no longer be part of the rational support reflectively available to her in support of her knowledge that p (and thus she is not the market for knowledge that p on this basis even if she did happen to believe that p). McDowell seems to agree—see, e.g., McDowell (2002b, 277–78; 2003, 680–81). See also note 5.

5. McDowell agrees with this point about how it is possible to be hoodwinked by radical skepticism into reducing the scope of the rational support available to one. See, in particular, McDowell (2012), where he approvingly discusses Clarke’s (1965) idea that it is possible, simply through an effort of attention on the part of the subject, for an agent to regard his otherwise normal perceptual experience as providing mere seemings rather than direct perception of objects. See also note 4.

6. I explore Neta’s proposal in more detail in Pritchard (2005c), where I also consider how this view relates to Williams’s inferential contextualism. We will return to consider rational support contextualism and inferential contextualism again in chapter 7—see §4—once we have a full description of our (part Wittgensteinian, part McDowellian) anti-skeptical proposal on the table.

7. See also Johnsen (2001). In addition, see the explanatory contextualism advanced by Rieber (1998), which is a form of attributer contextualism that explicitly incorporates a contrastive element. For a helpful critical discussion of the relative merits of contrastivism and the view that we are here calling rational support contextualism—albeit one that sides with the latter over the former—see Neta (2008b).

8. Schaffer (2004) in fact contrasts attributer contextualism with contrastivism, though note that he is building much more detail into his characterization of contrastivism than we are here.

9. The observant reader will spot that we are moving back and forth here between presenting contrastivism in terms of an ability to discriminate between specific objects (as in the second Schaffer quotation just offered) and in terms of being able to discriminate between one state of affairs, propositionally described, and an incompatible state of affairs, also propositionally described (as in the first Schaffer quotation). The second way of putting things is, I take it, Schaffer’s considered view, but for our purposes we can gloss over this distinction here.

10. Note that part of the case that Schaffer mounts in defense of contrastivism is linguistic—see, especially, Schaffer (2008)—but I take this to be controversial, and won’t be exploring it here. Our critique of the position will instead be focused on other aspects of the contrastivist proposal. See Rickless (forthcoming) for a helpful recent critical discussion of the linguistic case for contrastivism.

11. I critically evaluate contrastivism in more detail in Pritchard (2008a).

12. See also Pryor (2004; 2012). In addition, see Huemer’s (2001; 2007) defense of the closely related view known as phenomenal conservatism.

13. See McGrath (2013) for an insightful discussion of the role of defeaters with regard to the dogmatist position.

14. For some useful critical discussions of dogmatism, see Neta (2004a), White (2006), Wright (2007), Silins (2008), McGrath (2013), and Williams (2013). See also the essays collected in Tucker (2013), which covers both dogmatism and the related view known as phenomenal conservatism (see note 12).

15. Note that for ease of expression I have renamed the three claims that make up this skeptical paradox, compared with the formulation that was offered in chapter 2. Nothing of any consequence hangs on these changes.

16. This was the approach I took in Pritchard (2012b, pt. 3). As I explain in the introduction, the reason I took this approach wasn’t that I was convinced by it, but rather that for the purposes of that particular book it was vital to present epistemological disjunctivism in such a way that it did not depend on further philosophical claims, epistemological or otherwise, that were independently contentious.

17. See Schaffer (2007) for a heroic attempt at defending something like the closure principle for contrastivism. See also Kvanvig (2008) and Hughes (2013).

18. But see Hughes (2013), who argues that contrastivism can adequately defend closure-style inferences for knowledge, but that it cannot avoid the epistemically immodest consequences of such inferences. Since this is, as we have just noted, effectively the problem that epistemological disjunctivism faces in this regard, it would follow that contrastivism fares no better than epistemological disjunctivism as an anti-skeptical strategy even if we restrict ourselves to closure_RK-based radical skepticism.

Chapter 7. Farewell to Epistemic Angst

1. Or, at least, the universality of rational evaluation thesis has to go to the extent that it applies to our commitments in general, whether (knowledge-apt) beliefs or hinge commitments. As we saw in chapter 4—see notes 9, 14, and 32—insofar as one restricts this thesis to (knowledge-apt) beliefs, then it is entirely compatible with the Wittgensteinian account of the structure of rational evaluation, at least on the nonbelief reading. This is because the thesis would now be simply inapplicable to the rational evaluations of one’s hinge commitments. There is thus a sense in which the universality of rational evaluation thesis, construed in a very specific way, could potentially be squared with a Wittgensteinian epistemology.

2. One issue on which inferential and rational support contextualism arguably fares better than the biscopic proposal is with regard to so-called abominable conjunctions (see chapter 1, §1). Recall that this is a problem, due to De-Rose (1995), which concerns the oddity of simultaneously asserting both that one has knowledge of an everyday proposition and that one lacks knowledge of a denial of a skeptical hypothesis (even though the everyday proposition is clearly logically incompatible with the skeptical hypothesis). Although this problem is directly applicable to anti-skeptical positions that deny the closure principle (in all its various guises), it also applies to any view that, like contextualist proposals, wants to treat any claim to know that one is not the victim of a skeptical hypothesis as expressing a falsehood. Contextualists of all stripes can respond to this problem by maintaining that there is no single context of epistemic appraisal relative to which the assertion of an abominable conjunction expresses a truth. The first conjunct, if asserted in isolation, may well express a truth relative to an everyday context of epistemic appraisal. But by bringing in the skeptical hypothesis at issue in the second conjunct, one thereby changes the context of epistemic appraisal, such that while the second conjunct may well now express a truth (if asserted in isolation anyway), the first conjunct no longer does. In this way, contextualists can account for why any claim to know the denial of a skeptical hypothesis expresses a falsehood without having to deny either (i) the closure (or closure_RK) principle or (ii) that everyday knowledge claims express truths. The biscopic proposal doesn’t involve denying the relevant closure-style principles either, of course, but it does allow that one cannot have knowledge, including rationally grounded knowledge, of the denials of skeptical hypotheses, and so this problem still applies. Although the biscopic account cannot appeal to a shift in contexts of epistemic appraisal to explain away abominable conjunctions, it does have a diagnostic story to tell in this regard. After all, once one has become apprised of the Wittgensteinian account of the structure of rational evaluation, and in particular its epistemic implications with regard to our hinge commitments, one will no longer expect there to be an inferential route from rationally grounded knowledge of an everyday proposition to rationally grounded knowledge of the denial of a skeptical hypothesis. Accordingly, abominable conjunctions are now no longer puzzling, since they merely highlight the point that the scope of one’s rational evaluations doesn’t extend to one’s hinge commitments.

3. For some useful recent discussions of the contrastivist defense of a closure-style principle, see Schaffer (2007), Kvanvig (2008), and Hughes (2013). Relatedly, note that contrastivists will also need to offer a diagnostic story with regard to so-called abominable conjunctions (see note 2).

4. Or at least the skeptical problematic in the guises that we have engaged with. Recall from chapter 1 (note 1) that while we have here dealt with a prominent and fundamental type of radical skepticism, there are still other varieties remaining.

5. See Boult & Pritchard (2013) and Pritchard (2014a). I have recently become aware that Putnam (2006) also uses the analogy of vertigo in the context of radical skepticism when discussing Cavell’s (1979) treatment of the skeptical problem. While there are similarities between the approach I take to radical skepticism and that offered by Cavell, ultimately his stance on this topic is more concessive than mine. In particular, Cavell wants to claim that while the radical skeptical challenge is not bona fide, there is nonetheless a “truth in skepticism.” For some useful recent discussions of what Cavell has in mind in this regard, see Putnam (2006) and Shieh (2006).

6. Note that in earlier work I used the phrase epistemic angst as a cover-all term in this regard—see, for example, Pritchard (2005b, passim; 2005e; 2008c)—such that it referred both to a general epistemological anxiety brought on by the radical skeptical problematic and to what we are here calling epistemic vertigo. I now use this phrase so that it applies only to the former, and not also the latter.

7. I further develop my ideas about the general notion of risk, including as regards epistemic risk specifically, in Pritchard (2014b; 2015).

8. The idea that it is the detached perspective that generates radical skepticism is a recurring motif in the literature on this topic. For some of the key texts in this regard, see Clarke (1972), Cavell (1979), Stroud (1984), and Nagel (1986, chaps. 5–6; cf. Nagel 1970). Wittgenstein (1969) and Heidegger (1962) may also belong in this list, depending on one’s interpretation of them on this score. See Williams (1991) for an important, Wittgenstein-inspired, discussion of this way of thinking about radical skepticism. See Minar (1999; 2001a; 2001b) for some helpful discussions of Heidegger’s treatment of radical skepticism. Incidentally, Nagel is unusual in taking the detached perspective of philosophical reflection as being the source of a range of philosophical problems. For a discussion of the connection he draws between radical skepticism and the meaning of life, for example, see Pritchard (2010a).

9. The angst that remains might well be best thought of as a form of alief, which is a possibility that we rejected earlier as an account of the propositional attitude involved in our hinge commitments (see chapter 4, note 13). At the very least, the kind of propositional attitude involved in epistemic vertigo as just described is similar to the propositional attitude involved in acrophobia, and the latter is usually classified as an alief. For further discussion of aliefs, see Gendler (2008a; 2008b), Nagel (2012), and Mandelbaum (2013).