Critical Study of John Hawthorne's Knowledge and Lotteries and Jason Stanley's Knowledge and Practical Interests

Abstract

In two important recent books, John Hawthorne and Jason Stanley each argue that non-evidential factors, such as the cost of being wrong and salience of possible error, have a place in epistemological theorizing. This point is familiar from the work of epistemological contextualists, who emphasize non-evidential speaker factors: factors which, when present in a speaker's conversational context, affect the semantic content of her knowledge attributions. According to Hawthorne and Stanley, the appropriate focus is on the subject, rather than the speaker: when the relevant non-evidential factors are present in a subject they can affect whether the subject knows. This suggests a reorientation for epistemology, away from the standard “intellectualist” (Stanley's term) model, endorsed even by contextualists, according to which only evidential or more broadly truth-related factors (evidence, safety, sensitivity, reliability, etc.) bear on whether a subject knows. If Hawthorne and Stanley are right, then the contextualist program should give way to a program of anti-intellectualist invariantism, or to use a more common label, subject-sensitive invariantism (SSI).1 The books explore in insightful and original fashion a range of important issues in epistemology, all focused on the concept of knowledge: the lottery puzzle, skepticism, the knowledge rule of assertion, the relation of knowledge to objective chance, epistemic closure, etc. Here, we limit ourselves to the relative merits of SSI. Stanley claims that the only plausible argument for contextualism is based on intuitions about certain stakes-shifting cases (like Keith DeRose's bank cases and Stewart Cohen's airport case). (84) This is because he takes the contextualist about knowledge attributions to be positing a sort of context-sensitivity which departs significantly from the familiar sorts. He painstakingly backs up this claim in chapters 2 and 3. In chapter 2, ‘know’ is shown not to be a gradable expression (it doesn't admit modifiers and has no associated comparative), and so the assumption that gradable expressions are context-sensitive is no evidence that ‘know’ is. Chapter 3 argues for two general points: first, our use of propositional anaphora (‘what I said’) and speech reports (‘I said that I knew’) doesn't follow the pattern displayed by familiar context-sensitive expressions (like modals, demonstratives, quantifiers, gradable expressions); second, we do not find the sorts of intra-discourse context-shifts that we find with the uncontroversial examples and which we would expect to find given the plausible assumption that context-sensitivity has its source in particular occurrences of elements in logical form.2 Stanley's conclusion is that if the intuitions in the stakes-shifting cases can be explained away adequately without resorting to contextualism, we would have little reason to accept contextualism.3 If Stanley is right about the dialectical situation, the ramifications for epistemology are significant. Contextualists have wanted to argue that, because these cases force us to acknowledge dramatic shifts in epistemic standards across contexts of use, we have good reason to think that even more dramatic shifts could be the source of the appeal of skeptical arguments. Without an independent reason to think that dramatic shifts occur, contextualists have no good answer to the charge that their response of skepticism is ad hoc, the sort of easy response available to solve almost any philosophical puzzle. With all this in mind, here are DeRose's original (1992) bank cases: Bank Case A (Low Stakes). My wife and I are driving home on a Friday afternoon. We plan to stop at the bank on the way home to deposit our paychecks. But as we drive past the bank, we notice that the lines inside are very long, as they often are on Friday afternoons. Although we generally like to deposit our paychecks as soon as possible, it is not especially important in this case that they be deposited right away, so I suggest that we drive straight home and deposit our paychecks on Saturday morning. My wife says, “Maybe the bank won't be open tomorrow. Lots of banks are closed on Saturdays.” I reply, “No, I know it'll be open. I was just there two weeks ago on Saturday. It's open until noon.” Bank Case B (High Stakes). My wife and I drive past the bank on a Friday afternoon, as in Case A, and notice the long lines. I again suggest that we deposit our paychecks on Saturday morning, explaining that I was at the bank on Saturday morning only two weeks ago and discovered that it was open until noon. But in this case, we have just written a very large and important check. If our paychecks are not deposited into our checking account before Monday morning, the important check we wrote will bounce, leaving us in a very bad situation. And, of course, the bank is not open on Sunday. My wife reminds me of these facts. She then says, “Banks do change their hours. Do you know the bank will be open tomorrow?” Remaining as confident as I was before that the bank will be open then, still, I reply, “Well, no. I'd better go in and make sure.” (913) How do these cases support contextualism?4 It seems (premise 1), you speak the truth in both cases. But (premise 2) you have the same evidence and generally the same truth-related factors regarding the proposition in question. Therefore, (conclusion) the sentence ‘I know that the bank will be open on Saturday’ must differ in semantic content in these two contexts of use, due to a difference in content associated with ‘know’: contextualism is true.5 Stanley grants both premises but denies the conclusion. The conclusion follows only on the assumption of intellectualism—only on the assumption that, if the content of ‘I know that the bank is open on Saturday’ is the same in the two cases, a difference in the sentence's truth-value across the cases requires a difference in the evidence for the bank is open on Saturday. But perhaps intellectualism is false: even if ‘I know that the bank will be open on Saturday’ has the same content across the cases, it could be true in one case but false in the other because the practical facts differ. SSI can handle the bank cases just as well as contextualism. DeRose's bank cases are cases of first-person present-tense knowledge attributions in which the attributor is aware of his practical interests. SSI delivers the same results as contextualism for these cases. The same goes for first-person versions of the lottery puzzle. The contextualist says that the sentence ‘I don't know my ticket is a loser’ is true in a context in which the possibility that one wins is salient, even though it wasn't true in a previous context when that possibility wasn't salient (e.g., because, as in Hawthorne's example, one is thinking about whether one will have enough money to go on Safari next year). For the contextualist, the change in truth-value is due to the association of a higher epistemic standard in the later context than in the former, and this association is made possible by the content-shifting power of salience. The SSIist, piggybacking off the contextualist, may say that salience destroys knowledge. When the possibility of your winning wasn't salient, you knew you would lose. Now that it is, you don't know. And the same case could be made for skeptical arguments. The SSIist might say that the salience of the possibility that one is a brain in a vat strips one of much of one's empirical knowledge. These anti-intellectualist manoeuvres incur an explanatory burden, for it is unclear how salience of possible error, or raised practical stakes, could destroy knowledge. But the view does at least match contextualism in its claims about the truth-values of first-person present-tense knowledge attributions and denials. Moreover, the SSIist doesn't make the vindication of our intuitions about cases dependent on delicate use/mention issues, such as the distinction between the question of whether what is said in uttering ‘I know that p’ is true and the question of whether the subject knows that p. It appears, then, that we have a prima facie case for thinking that either contextualism or SSI is true. But which? One consideration that might seem to favor contextualism is the fact that SSI predicts the truth of odd-sounding temporal and modal embeddings. It sounds wrong to say “I don't know that p, but, assuming p is true, I used to know that p,” or “If I didn't have so much at stake, then if p were true, I'd know it was.” But SSIists seem committed to the potential truth of both of these utterances. Is this a reason to favor contextualism over SSI? Stanley suggests (pp. 107–13) that contextualists might find themselves committed to similarly peculiar temporal and modal embeddings, depending on how they explain the source of context-sensitivity. If so, these oddities would be no reason to favor contextualism over SSI. He considers in particular David Lewis's (1996) contextualism, according to which knowledge-ascribing sentences contain a context-sensitive quantifier under semantic analysis. Under Lewis's account, ‘S knows that p’ is true in a context iff ‘S has evidence which rules out all possibilities in which not-p’ is true in the context, with ‘all possibilities’ contextually restricted to those that are not being properly ignored in the context, where attention to a possibility is one way of not being properly ignored. Because the best semantic account of quantifiers holds that they express properties, it would seem that Lewis is committed to thinking that the relevant property is not being properly ignored by A, where A is the attributor. But then we should expect that a sentence such as ‘If A were properly ignoring more possibilities than he is, S would know that p, if p were true’ to be acceptable, which of course it isn't. We have a reservation about Stanley's argument. Consider his example of ‘every bottle’. In a context of use, this expresses a property, e.g., being a bottle on such and such shelf. Which property it expresses depends on what is salient to the attributor or conversational participants, but the property expressed itself is not about salience to a certain attributor. Thus, ‘John bought every bottle but wouldn't have if I had been thinking about a different grocery store’ won't come out true. Similarly, it seemed to us that one could revamp Lewis's view by proposing that which property ‘all possibilities’ expresses in a context of use depends on what the attributor is properly ignoring but the property expressed isn't itself about the attributor and his ignorings. So, perhaps SSI is committed to the peculiar sounding embeddings in a way that contextualism isn't. Nonetheless, there are ways to soften the blow. As Hawthorne (177n.40) remarks, anyone who thinks barn façade cases are Gettier cases must say that in many cases in which you look at a barn it is true to say, “I know that's a barn, but if unbeknownst to me there were a lot of barn facades around then even if this one was a barn I wouldn't know it was.” The defeasibility of knowledge, too, generates odd-sounding predictions like this: “I don't know that Mary is sick, now that her boyfriend told me she was faking it, but if he is misleading me, then if he hadn't told me anything, I would know (having seen her coughing earlier today).” If we can live with these embeddings, perhaps we can live with SSI's temporal and modal embeddings as well. SSIists, however, must further part company with contextualists on third person knowledge attributions. When Mary and John, whose evidence is the same as Smith's, utter not only ‘We don't know’ but also ‘And neither does Smith,’ the intellectualist contextualist will say that Mary and John speak the truth about Smith if they speak the truth about themselves. The SSIist will of course have to insist that this move is in general illicit, because, e.g., less might be at stake for Smith than for Mary and John or fewer counterpossibilities of error salient to him. Yet we seem to make this move on a regular basis. It would be very odd for Mary and John to wonder, “We don't know whether the plane stops in Chicago, but if it does, then Smith knows it does.” The SSIist has to explain away mistakes about third-person knowledge attributions and denials. Keith DeRose (2004) has called this the problem with SSI. Hawthorne and Stanley are well aware of it, and propose their own solutions. The core idea behind both proposals, spelled out most clearly in Stanley (who is trying to improve upon Hawthorne's account (see pp. 162–66)), is that we treat others as if they were in our situation, having our stakes (or finding salient possibilities we find salient). Stanley puts it this way: our real aim is to deny that if they were in our situation they'd know. Indeed, they wouldn't. But what is strictly said, that they don't know, is false. (101–4) This proposal is committed to an error theory about some of our knowledge attributions. But, as Hawthorne and Stanley would be quick to point out, contextualism has a similar problem because as we noted above, contextualists place heavy weight on the use/mention distinction. Mary and John are happy to assert not only ‘Smith doesn't know’ but ‘What Smith said is false. He doesn't know any more than we do.’ Equally worrisome, Stanley notes that contextualists have to strain, more than SSIists, to accommodate intuitions about other third person cases—cases in which the attributor has lower stakes than the subject. Stanley calls such cases Low Attributor-High Stakes and claims that contextualism delivers the wrong result about them. Intuitively, Stanley claims, the speaker's attribution of knowledge to the subject is false, whereas contextualism predicts its truth. Contextualism in his view also than SSI in certain person cases, such as Stakes which a subject thinks her stakes are but they are He that the weight of the versions of the stakes-shifting SSI over contextualism. does not that contextualism is false. Therefore, there is a way to accommodate without Stanley's intuitions about all the versions of the stakes cases, by for a contextualist of it is part of the for knowledge attributions that that p’ only when the subject is to as if p in of the So, when a attributor says of a subject knows that p,” the attributor that intellectualist contextualism, it will have to to This takes of all but the subject has stakes, but we have the the attributor when she says, doesn't know that handle this case, we the contextualist when one's stakes that one a certain standard to know, in one's speech context associated with having an epistemic that at least that so that if a subject doesn't it with to p, the subject doesn't with to p in one's speech possibility for all the stakes-shifting cases is to with a view which would also have the over contextualism that it doesn't different intuitions depending on whether the say or he said is rather than doesn't The possibility of these of up the to the epistemological question of whether intellectualism is true from both the semantic question of whether knowledge-ascribing sentences in content across contexts of use and the question of whether knowledge-ascribing can only be true relative to an (e.g., a or an epistemic Stanley is right that intellectualism is a for contextualism for in the of stakes-shifting cases. As Stewart in contextualist from his case of Smith and Mary and John, what is in the is a good enough reason for Smith to know, then it is a good enough reason for John and Mary to But there is no reason an anti-intellectualist to semantic or are to accommodate intuitions such as those in the cases. the are as as we know no one is on as they do the cases. these we are with intellectualist contextualism and each of which is committed to error about our to some of the cases at has to to accommodate those It will seem to some that all this and in the is an Consider what Hawthorne calls This is intellectualist invariantism with the claim that the standards for knowledge are enough so that a many of our knowledge claims are in fact true. it is that contextualism and SSI are committed to error of their own about the versions of the bank cases, there is now no from consideration of those cases, to invariantism just because it requires its own error theory to handle we to the invariantism, is a reason to that knowledge, or even is to salience or practical in the such factors are not truth-related or the used in standard Gettier misleading into the at we little in salience of possible error as a subject seems to us to be no other than the to the cases, to a between knowledge and salience of possible The situation is more for knowledge and This is the of our Hawthorne us to the are a for a lottery ticket that cost a in a ticket lottery with a and reason as I will the If I the ticket I will If I the I will a I to the This he claims, is The is that you don't know the is true. this lottery case to Hawthorne's A or so you bought a lottery ticket in a ticket the in a at I won't have enough money to go on Safari next the less as to the This is In Hawthorne's because you know the is true. like these are used to support Hawthorne's one is to use p as a in practical if and only if one knows that Hawthorne that it is to give the the difference between the and the other the lottery is that in the case the for the subject that he will is enough so as to be to the practical issues at he can the epistemic that is This knowledge out in favor of a about you can use p in practical just in case the that p is false is just in case it can be in your practical this the If this is so we give in and accept Hawthorne's is to appeal to the according to which that p is and for having epistemic or for being or for there being no epistemic for the subject that the epistemic of is the same in both cases or If it is the then by the you don't know that you will in the you don't know, in the lottery case, that you will the But if you don't that, then by you don't know that you won't have enough money to go on But the seems to be a of knowledge. Thus, Hawthorne that if the epistemic of is the same in the two cases, then skepticism So, skepticism the for you of winning between the cases, and so by the the two cases differ with to your knowledge that your ticket will lose. It follows that practical can affect knowledge in the way We can think of two ways to to Hawthorne's argument. one might argue that you know in the case isn't to in the case you do have a lottery and might be have a reason to think you might have enough money to go on Safari next In response Hawthorne (pp. argues that lottery cases can be at will for just about any of our claims to knowledge. If so, then there is about the case, because there is about the of that is a lottery one might question the We with this are the relevant to between If your epistemic for p is then, it should follow that you would be to accept a on p at any But you wouldn't be to do this for very many doesn't think he would on the of at any Thus, the seems to One can to this as Hawthorne does, by intellectualism about epistemic and so that how a proposition is for a person can with practical This seems to us If you me a on the proposition this will come up that doesn't seem to lower its and not to the that it will come up us the Is there any other reason to knowledge into the of certain of practical are and others not just on the If you an when a for his lottery he isn't to it, he will you that he doesn't know that the ticket will lose. If you the same in the he the rather than the given that he of on he will answer that he knows he won't have enough money to go on seem to knowledge its to their As Stanley A standard use of knowledge attributions is to When I I to the on the rather than the on the right, I will by that I knew that the on the had the I but I not know whether the on the right When my wife me I rather than I that I knew that it was the to go to the If we could have a view that knowledge relevant but doesn't us to the this might seem to just the right And it such a view is We can claim that knowledge is with a epistemic of error but only an This view at least the of Hawthorne's will of course to the Stanley as having rather than semantic The idea would be that there are cases in which one knows but isn't to on it, that one knows often or at least in some way that one is to on it Stanley is well aware of this and a to come up with a We that it is not easy to give such a (see our but we that he shifts the of As we above, many in our view will insist that intellectualism is a cost than positing or intellectualism us to the claim that you can or knowledge, not by or by but by or up a large check. And that seems Is there a more argument available for a When it to would deny that when can be to other It does not follow from the fact that your is even from the fact that you that you have a reason to that your won't But you know that your is you have an reason to that your won't that your is So, when it to we do to say that, when you know that p, p can be your your for course, when you know that p, p can to be your for some for the very reason that p has to do with if you know that your is that it is is not a reason you have for that is from the than But its to be a reason you in this case, is not the result of the for you that your is not the result of any other epistemic you have with to the proposition that your is When you know that p, in the for you that doesn't in the way of p being a reason you have for We can say very similar about the very there are facts such that, when you know to be true, they can be you have to do The fact that your will be by a which is the you and in the next seems to be such a that the of this fact would not that you have any reason to your (having no idea of the situation, you would be it in on the with your would your in it if be enough if that itself were not But if you know this then you do have a reason to your This is a of the situation. As with the situation, when you know that p, p might to be a reason you have for but not because of any epistemic of with to because for example, the for you that might know that your is But this is not a reason you have to your out of the way of an its to be a reason you have is not due to any epistemic with to the proposition that your is seems little to the practical and the cases in the way by the If the that doesn't in the way of p being a reason you have for for any then the that doesn't in the way of p being a reason you have for A, for any This other of the fact that, when we that p in the course of about what to we don't then p off from our about what to When we p, we can use p. But, noted that, when you know that p, then the that doesn't in the way of p being a reason you have for in with the a general not those by and more in a by Hawthorne and Stanley If you know that p, then the that doesn't in the way of p being a reason you have for or a problem for intellectualism if there are of for the that is the but are such that the that in the way of p being a reason one has but not the Consider again the airport Smith and Mary and John are with to the that the plane stops in According to the Smith and Mary and John all know that the plane stops in the that the plane doesn't stop in doesn't in the way of it being a reason they have say, In case, at For Mary and John, there is. and the assumption that Mary and John have knowledge, Mary and John have an reason not to checking with the a of the case is so that they should check with the given all that is at So, the had better not Mary and reason to just go and to Mary and John in just and How might the that the plane stops in is a reason that Mary and John have to the that, even if they have a very good reason to in their reason is by some other reason they The A proposition about the of error with the of the relevant or to use the a proposition to the that checking further A general intellectualist then, for is to make a first, having to do with of error can with about what is in fact the case and second, when the stakes are about of error can out about what is in fact the This is For this would seem to the sort of the one the plane stops the other it might And if it doesn't it the are I'd better check further to it stops in it This is not what we do when we we we do when we is the at and which is more the one the other Which is more the good or the that both that we don't facts the that those facts don't if we were facts would the one the plane is to the other it might not Which is more the fact that it is, or the that it In the intellectualist what seems to be a about having When you have a reason to or about when you have a reason to to that reason for in whether to This is not to say that, you have a reason to you should But it is to say that, when you have that you to it into with other you have without the relative of those In this you have are might not have as that it will that you will have to for the If you do then when whether to your you might have to the that it will how long you think have to for the if you have both as then you can use both in your without their to and have to So, to if I don't the Because this off the intellectualist there is us from on the of that if knowledge is with a of error, then knowledge can come and go with in the practical If Hawthorne's is false, so is