Book reviews

Epistemic contextualism (EC) is the view that what's expressed by claims of the form ‘S knows [/doesn't know] that p’ depends on such factors as the interests, purposes, and expectations of the speaker. While EC has “been met with overwhelming scepticism by a vast majority of epistemologists and philosophers of language” (1), according to Michael Blome-Tillmann, this is largely owing to shortcomings of extant versions of the view.1 Blome-Tillmann's primary aim here is to present a contextualist semantics for ‘knows’ that's superior to other such proposals—one, moreover, that sheds light on central epistemological issues.The preferred semantics for ‘knows’, which Blome-Tillmann has been developing and defending over several recent papers, is Presuppositional Epistemic Contextualism (PEC), the main inspirations for which are found in the work of David Lewis and Robert Stalnaker. Working, like Lewis, within a relevant-alternatives framework, Blome-Tillmann proposes that(L*) x satisfies ‘knows p’ in context C ↔ x's belief that p is properly based andx's evidence eliminates all ¬ p-worlds, except for those that are properly ignored in C. (31)Blome-Tillmann accepts the Lewisian idea that actuality is never properly ignored, which secures the factivity of ‘knows’ (33, 92–95). But he rejects Lewis's Rule of Attention, whereby the mere mentioning of ineliminable skeptical possibilities suffices for their not being properly ignored, with the result that “knowledge” can easily be made to “vanish.” In its place, he puts the Rule of Presupposition:If [possibility] w is compatible with the speakers' pragmatic presuppositions in C, then w cannot be properly ignored in C. (20)Together with (L*), this Rule yields the core idea of PEC, and the one that does most of the theoretical work in the book: namely, that a necessary condition for the satisfaction of ‘knows p’ in C is that the subject's evidence eliminates all the ¬p-worlds compatible with what is pragmatically presupposed in C (34).Finally, the notion of (pragmatic) presupposition is borrowed from Stalnaker's work on linguistic communication:x pragmatically presupposes p in C ↔ x is disposed to behave, in her use of language, as if she believed p to be common ground in C (26),where, again following Stalnaker, ‘common ground’ refers to those things that, for purposes of the conversation, all parties accept (that is, treat as true), and believe that all accept, and believe that all believe that all accept, and so forth (23).With the basic ideas of PEC laid out (chapter 1), Blome-Tillmann applies it to the skeptical puzzle (chapter 2), responds to a host of objections, both to PEC specifically (chapter 3) and to EC per se (chapter 4), and illustrates PEC's bearing upon a number of further puzzles—including Gettier Cases, the lottery puzzle and inductive knowledge (chapter 5), and Moorean thought and transmission failure (chapter 6). As well, Blome-Tillmann indicates how PEC might accommodate the influential recent thesis that evidence = knowledge—or, as suits the semantic character of PEC, that ‘evidence’ = ‘knowledge’.As this quick sketch suggests, this is an ambitious book. It's also a very good one, and an important contribution to the current debate over the semantics of knowledge attributions. The writing is clear and illuminating, and even when one disagrees with Blome-Tillmann, his arguments reward close attention. Contextualists are lucky to have him; noncontextualists will benefit from consideration of his careful work on the position's behalf; and even those epistemologists and philosophers of language not invested in the debate over EC will find plenty of interest in the handling of the issues addressed.Here, and partly by way of highlighting some of the book's central claims and arguments, I'll register some critical concerns. In doing so, I'll be focusing on the book's first four chapters, leaving aside the interesting discussions of the final two. Briefly, though: the response to the Gettier problem (136–40) is similar to Lewis's, the general handling of the lottery puzzle follows Stewart Cohen's (149), the problem of induction (at least in its “knowledge” version) is treated as an instance of the general skeptical argument (on which, see below) (152), and (as Blome-Tillmann notes; see 186n42) the basic moves made in the discussion of Mooreanism and transmission failure are available to noncontextualists. Blome-Tillmann's discussion of these matters is hardly just a rehearsal of things others have argued. But the heart of the book remains the presentation and defense of PEC (and EC), and the treatment of skepticism and similar puzzles, in chapters 1–4.About the latter, Blome-Tillmann, like other contextualists, takes one of the main selling points of EC to be the resolution of the skeptical puzzle it offers (1). Thus, given PEC, insofar as speakers in “quotidian contexts” (36) presuppose the denial of skeptical hypotheses, those possibilities may be properly ignored. However, in “philosophical contexts” (50), where we consider skeptical possibilities, the “unsteady” among us can be made to cease to pragmatically presuppose that they do not obtain (42). When that happens, ‘know(s)’ picks out a more demanding relation and denials that S ‘knows’ are liable to come out true. In this way, we're able to preserve the truth of our ordinary claims to ‘know’, while explaining the sense that the skeptic's claims might be right too. (They sometimes are!) Essentially the same explanation is given of shifty intuitions about the truth-values of some more pedestrian ‘knowledge’ ascriptions, such as those arising in the well-known Bank and Airport Cases (36–37).That certain seemingly contradictory ‘knowledge’ attributions and denials don't actually conflict is a central deliverance of EC. Why would we think that they do? Why would we make mistakes about the contents of various ‘knowledge’ claims? Also, geared as it is toward the use of epistemic terms in speech, what sort of explanation can (P)EC offer of what happens in “solitary contexts,” where a thinker is engaged in no conversation? Blome-Tillmann's response to both these matters involves seeing the subject as “being in a covert conversation with herself: thought is a limiting case of communication, one in which the common ground collapses into the set of propositions the thinker accepts, believes himself to accept, etc.” (63). So PEC applies without difficulty to solitary subjects. Further, subjects engaged in actual (public) conversations are “members of two different contexts: firstly, a public context of utterance in which the subject converses with other speakers and, secondly, a private context of thought” (43). Often these contexts include the very same presuppositions. But when they don't, competent speakers can fail to see that “‘knows’, as uttered in their public context of conversation, is sensitive to what is presupposed in that public context of conversation and not to what is presupposed in their private context of thought” (47).One familiar complaint against EC is that it doesn't really engage with epistemological issues, skepticism included. Responding to the charge of “epistemological irrelevance,” Blome-Tillmann notes that philosophical contexts are not inevitably skeptical, so philosophers (including contextualists) can hold that speakers often speak truly when asserting that they ‘know’ some ordinary proposition (50–51). And, to the objection that EC concerns only ‘knowledge’, not knowledge, Blome-Tillmann replies that insofar as the philosophical context is nonskeptical, theorists can safely employ the object language and report that people often speak truly when they assert that they know (51–52).However, this doesn't seem to address certain versions of the relevant complaint. According to Kornblith (2000) and Bach (2010), skepticism doesn't concern what happens in special, “philosophical” contexts, where certain possibilities of error are brought to our attention, with us perhaps being made to take them seriously and it becoming harder to ‘know’. Rather, it's the view that we lack knowledge, even in ordinary contexts, and by ordinary standards, as we have no good reason to suppose that skeptical possibilities do not obtain. Perhaps PEC, together with the idea that ‘knowledge’ = ‘evidence’, has the consequence that ‘evidence’ (‘justification’, ‘good reason’) is context sensitive too, and that our ‘evidence’ in ordinary contexts excludes skeptical possibilities (153–54). But, with skepticism understood aright, it's clear that what's doing the antiskeptical work isn't the contextualist semantics but the idea that skeptical possibilities are epistemically irrelevant—that knowledge doesn't demand their elimination—an idea that doesn't require EC, much less PEC.Where the contextualist semantics does clearly do work is in accounting for some shifty and apparently conflicting intuitions about the truth-values of various ‘knowledge’ ascriptions. As we saw above, Blome-Tillmann's account of what goes on here, including why “competent speakers are sometimes mistaken about the semantic properties of the expressions at issue” (47), makes essential use of the idea of a “private context of thought.” But this account prompts natural concerns.First, like the assimilation of thought to communication itself, the application of Stalnakerian ideas about the latter to the phenomenon of individual reflection seems, at best, rather forced. “What is most distinctive about [presupposition],” Stalnaker (2002, 701) writes, “is that it is a social or public attitude: one presupposes that ϕ only if one presupposes that others presuppose it as well.” True, thought involves presuppositions (63), but it needs to be argued that this is presupposition, not just in the “pretheoretical” sense of assuming or taking for granted (22), but in the “technical” (27) sense that Stalnaker, and PEC, deploy. So too, there may be “mental correlates” of speech acts such as assertion (64n2), but it's hardly obvious that Stalnaker's account thereof should be applied to them as well (ibid.). Indeed, we've just noted one reason for thinking it shouldn't.Second, the proposed account of how we can end up having false epistemological judgments seems to get the explanatory order backward. Blome-Tillmann says that “those who are persistent will pragmatically presuppose ¬ sh [sh = some skeptical hypothesis] in their private context of thought and will thus have non-sceptical intuitions,” and “those who are indecisive will be unclear as to whether they pragmatically presuppose ¬ sh in their private context of thought and will therefore oscillate between sceptical and non-sceptical intuitions,” independently of whether ¬ sh is pragmatically presupposed in their respective contexts of utterance (45). But surely it's more plausible to think that the mentioned beliefs (“intuitions”) precede and explain the relevant presuppositions.2 For this reason too, the proposed error theory, and the proposed treatment of “solitary contexts,” seems problematic.Independently of questions about the specific error theory Blome Tillmann proposes, the suggestion that there are changes in the content of ‘know’ “which seem entirely hidden from competent speakers” (98) has struck many as implausible, since we generally recognize such contextual shifts. One of the book's highlights is Blome-Tillmann's discussion in response: That we might have difficulty recognizing the context sensitivity of ‘know’, he argues, is made plausible by the fact that the term exhibits a unique combination of features. Its context sensitivity, like that of ‘flat’ or ‘summer season’, is less obvious than that of “core indexicals” such as ‘I’ and ‘here’ (103–10). However, ‘know’ is not linked to a scale of epistemic strength and so is not, like ‘flat’, gradable (118–24). Further, Blome-Tillmann argues, challenges to ‘knowledge’ ascriptions “are never merely challenges of the truth of one's ‘knowledge’-ascription, but rather also challenges of what one assumed to be the common ground in one's conversation” (127). (I claim to ‘know’ that a zoo creature is a zebra. You ask, “But can you rule out the possibility that it's a cleverly painted mule?” If I attempt a clarificatory reply—“No, but all I meant was . . . ”—I implicitly grant that the creature's not being a cleverly painted mule is not part of the common ground, and my initial ‘knowledge’ claim is undermined [128].) Blome-Tillmann thinks that this helps explain why, while they naturally occur in connection with other context-sensitive expressions (“Kansas is flat as States go”; “There's nothing worth watching on T.V.”), there is a conspicuous lack of clarifying devices for ‘knows’ (128–29). Relatedly, we seem unable, within a ‘knowledge’ ascription, to make the relevant context sensitivity itself explicit: “Given our presuppositions, I know that p,” for instance, looks defective (129). Nor is this anomalous, since “there are a large number of indexicals [specifically, what Kaplan calls ‘pure indexicals’] the context-sensitivity of which cannot be clarified or made explicit by appending additional linguistic material” (129): “given where we are, it's hot here,” say, looks ill-formed too.However, it's doubtful that this will, or should, satisfy the objector. If correct, Blome-Tillmann's account of the “linguistically exceptional” (135) character of ‘knows’ would explain why its context sensitivity is hard to spot, and why it cannot itself be made explicit in uses of ‘knows’. Still, once spotted, there is nothing—challenge situations, perhaps, aside—preventing us from making explicit what various ‘knowledge’ claims actually express, either in generic contextualist terms or along the lines of (L*). That done, it would be obvious, for instance, that certain surface-contradictory claims aren't really conflicting. What's not clear is why—with, as Bach (2010, 110) puts it, everything now out on the table—the contextualist semantics would still be met with such resistance3—especially if, as the argument for EC requires, our sensitivity to that semantics is what guides the relevant linguistic behavior and intuitions.4My thanks to Kent Bach and Geoff Pynn.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.