Jason Stanley, Knowledge and Practical Interests. Oxford: Oxford University Press, 2005. Pp. xi, 208. The thesis of this book is that what makes someone’s true belief a case of knowledge is partly determined by the believer’s practical interests. This thesis, which Stanley calls “interest relative invariantism” (henceforth, IRI) is incompatible with the “intellectualist” view that is most widely received in contemporary epistemology; the latter is the view that says that what makes someone’s true belief a case of knowledge does not depend upon any facts from the domain of practical rationality. Although it is not made completely explicit, a careful reading of Stanley’s book seems to reveal that his reasoning for IRI has the following overall structure. (1) Knowledge has a value that is distinct from the value of true belief, or justified true belief, or other epistemically valuable states that are not knowledge. (2) We can evaluate someone’s decision to act in a particular way by appeal to whether or not the person knew that a particular proposition that was a reason for her so acting was one that she knew to be true. (3) The best explanation for both (1) and (2) is that one may act on p if and only if one knows that p is true. (4) There is good reason to believe that one may act on p if and only if one knows that p is true. (From 3) (5) Whether one may act on p depends upon the costs of being wrong about p. (6) There is good reason to believe that whether one knows that p is true depends upon the costs of being wrong about p. (From 4, 5) (7) Our intuitions concerning whether someone knows that p is true are sensitive to the costs to that person of being wrong in believing that p. (8) Given 6, the best explanation of 7 is IRI: whether someone knows that p is true depends upon the costs of being wrong about p. (9) Therefore, there is compelling reason to believe that IRI is true. (From 6, 8) Step 1 states a widely held epistemological view that Stanley plausibly assumes to be true. He briefly makes the case for step 2 in his Introduction, where he discusses the use of knowledge attributions to justify and to criticize actions. He also defends step 3 in his Introduction, where he discusses the factors that obscure the knowledge-action link. (Steps 2 and 3 have both been defended at length in subsequent work.1) Step 4 follows from step 3. Step 5 is obvious and undisputed. Step 6 is plausibly inferred from steps 4 and 5. Stanley makes the case for step 7 in his Introduction, where he surveys a variety of hypothetical cases of knowledgeattribution and knowledge-denial – cases in which the costs of being wrong about a particular proposition vary significantly – and shows that our intuitions about the correctness of the knowledge-attributions or knowledge-denials in those cases vary systematically with those costs. So the first 7 steps of this reasoning are all defended, if at all, in the Introduction to Stanley’s book. John Hawthorne and Jason Stanley, “Knowledge and Action”, Journal of Philosophy 105 (2008): 571 – 90. 1 All of the rest of Stanley’s book – chapters 1 through 9 – is devoted to defending step 8. This is the step that most interests Stanley, and that has been most widely discussed. So Stanley’s task is to argue that, given the knowledgeaction link, IRI is the best explanation of our intuitions about such cases as the following: “Low Attributor-High Subject Stakes. Hannah and her wife Sarah are driving home on a Friday afternoon. They plan to stop at the bank on the way home to deposit their paychecks. Since they have an impending bill coming due, and very little in their account, it is very important that they deposit their paychecks by Saturday. Two weeks earlier, on a Saturday, Hannah went to the bank, where Jill saw her. Sarah points out to Hannah that banks do change their hours. Hannah utters, ‘That’s a good point. I guess I don’t really know that the bank will be open on Saturday.’ Coincidentally, Jill is thinking of going to the bank on Saturday, just for fun, to see if she meets Hannah there. Nothing is at stake for Jill, and shows nothing of Hannah’s situation. Wondering whether Hannah will be there, Jill utters to a friend, ‘Well, Hannah was at the bank two weeks ago on a Saturday. So she knows the bank will be open on Saturday.’” (Stanley 2005, 4) Why do we intuitively regard Hannah’s denial of knowledge to herself as true, and Jill’s attribution of knowledge to Hannah as false? Why is it that, in cases in which are exactly like this one, except that Hannah stands to lose very little by being wrong about the bank’s being open on Saturday, we would intuitively regard her denial of knowledge to herself as false, and Jill’s attribution of knowledge to her as true? Many different answers have been offered to these questions. Various philosophers have claimed (a) that these intuitive verdicts are the result of framing effects that operate on our psychology, and they reveal nothing very interesting about the nature of knowledge, or (b) that, when one stands to lose a great deal by being wrong about p, one should be less than knowledgeably confident that p, or (c) that, when one stands to lose a great deal by being wrong about p, one needs not only to know that p is true but also to know that one knows that p is true, or (d) that the intuitions above are revelatory of some semantic facts: either facts about the semantics of knowledge attributions, or facts about the structure of semantic facts quite generally. Stanley dispatches all of claims (a), (b), and (c) in a few pages of the Introduction, and devotes all of the rest of his book to criticizing claim (d). The main targets of Stanley’s book, then, are those who would resist step 8 of his reasoning by claiming that the best explanation of the fact stated in step 7 is a semantic explanation. Chapters 1 through 4 are devoted to stating and criticizing those “attributor contextualist” views2 according to which the semantic explanation concerns the semantics of knowledge attributions specifically, while chapter 7 states and criticizes those “relativist” views3 according to which the semantic explanation concerns the very structure of semantic facts themselves. Chapters 5 states See Stewart Cohen, “How to be a Fallibilist”, Philosophical Perspectives 2 (Atascadero, CA: Ridgeview, 1988); Keith DeRose, “Contextualism and Knowledge Attributions”, Philosophy and Phenomenological Research 52 (1992): 913 – 29 and “Solving the Skeptical Problem”, Philosophical Review 104 (1995): 1 – 51; David Lewis, “Elusive Knowledge”, Australasian Journal of Philosophy 74 (1995): 549 – 67. 3 See John MacFarlane, “The Assessment Sensitivity of Knowledge Attributions”, Oxford Studies in Epistemology 1 (2005): 197 - 233; Mark Richard, “Contextualism and Relativism”, Philosophical Studies 119 (2004): 215 – 42. 2 Stanley’s IRI view, chapter 6 argues that, while it has costs, they are less significant on balance than the costs incurred by attributor contextualism. And chapters 8 and 9 conclude by arguing that an IRI explanation of the data, unlike the attributor contextualist explanation, is not an application of a general schema that can be alltoo-easily applied to solve virtually any philosophical puzzle. Stanley’s book is full of detailed arguments that deserve extensive discussion, and I cannot devote such discussion to any one of those arguments within the very short space that has been mandated for this review. Fortunately, those detailed arguments have almost all been widely discussed already. The only observation that I would now like to add to that critical discussion is the following: Stanley’s most widely discussed arguments against attributor contextualism all take the form of noting that the semantic context-sensitivity posited by the attributor contextualist is without any clear precedent in natural language: therefore, the positing of such context-sensitivity is ad hoc. The attributor contextualist might claim that the context-sensitivity of “know that p” is just like the context-sensitivity of, e.g., “sufficiently justified in the belief that p” or “sufficiently sensitive in believing that p” or some such complex expression, but Stanley can justly complain that the obvious context-sensitivity in those expressions is carried by the context-sensitivity of “sufficiently”, and so is unlike the context-sensitivity that is alleged to be present in the semantically unanalyzable verb phrase “know that p”. I would like to point out, however, that IRI itself posits a metaphysical feature of knowledge – interest-relativity – that is without any clear precedent in the metaphysics of propositional attitudes. Of course, there clearly are some attitudes that are interest-relative: e.g., the relation of being rationally pleased that p. But the interest-relativity of such analyzable attitudes is carried by the interest-relativity of rationally, and so is unlike the interest-relativity that is alleged to be present in the metaphysically unanalyzable relation of knowledge. If attributor contextualism is ad hoc, then so too, for the very same reason, is IRI. Ram Neta University of North Carolina, Chapel Hill