ࡱ> %`bjbj.̟̟ha ppp z  8 dZ!\AV!!!!!! ! !UUUUUUU$Vh_Y>Uo;!!o;o;U!!UFFFo;!!UFo;UFFShDT!! / @TpSUV0AVSYDxY DTYDT|!*NF1 6a!!!UUOFX!!!AVo;o;o;o;$   Jason Stanley, Knowledge and practical interests Oxford University Press, Oxford, 2005, pp.XI + 191, ISBN 0-19-928803-8 (hb). This book is deep and genuinely interesting. It discusses the issue of knowledge in connection with practical concerns (or interests). The ideas it proposes are attractive and deserve critical discussion. There is every reason to suppose that this will become a classic in philosophy. Scholars and students will read it with great interest and will certainly capitalize on it, as this is a book that deserves respect and applause. First, there is the idea that knowledge is connected with action and practical interests. Surely this idea is of importance if one stresses that we tend to accumulate knowledge and then subsequently put it to use in action. In order to act, we base our actions on knowledge. (Yet, it must be conceded that we store knowledge also independently of practical interests; when I look around the room I am in, I may memorize what kind of items of furniture there are in the room independently of my wish to act; sometimes we are simply curious to know). The main idea of the book is that in situations in which stakes are high, we tend to seek more solid justification for our knowledge claims. This is plausible. I remember being a child. I saw the other children jump from a huge mass of stone to the next in the public gardens inside the Castle in Crotone. They were slightly older than I was and they were confidently jumping from one big mass of stone to the next, without worrying whether they would not have done it. Instead, I was worrying. I was not sure that I would have done it. And I did not act. Then, perhaps in a year or so, I jumped from that mass of stone to the next, like all the other boys and girls. Presumably in jumping I thought I knew that I would have done it. This situation is slightly different from the ones which are discussed in Stanleys book. Surely this is a situation in which practical interests are important. In the event of failing to make it to the other side, I would have fallen down and I could have broken a leg. This was what prevented me from jumping, when I was still wondering whether I would do it. However, unlike the situations discussed by Stanley, this is clearly one in which, if one stops too long in thinking of the risks incurred if one fails to have knowledge, then one is prevented from jumping. But one will not be a child for ever, and one attempts to jump when one is sure enough, even if one does not wonder any longer if one really knows that one can make it to the other side. When I jumped, in a way I knew that I could jump, even without having to wonder whether I knew that I knew that I could jump. The moral of my little story is that action is risky and that if one were to ponder all the risks involved in one action, one would never act. Suppose you are a surgeon and you have to operate. In order to operate, you have to know a number of things. For example, you have to know that you are using the right instruments. But the surgeon does not stop to look at the labels on his instruments. Doing that would take precious time. If the surgeon were to think too much of the stakes involved in the operation, he would probably not act at all, because in the context of these risks he would have to put in question every piece of knowledge he has. But he does not do this. If one were to do this, one would not act (at all) or one would not act efficiently. Now the opposite case, where the stakes are pretty low. I am sitting in an armchair reading Stanleys book. I am passively looking around in the room to see whether his theory works. I happen to see an armchair in front of me. I know that that is an armchair (I do not need to think I know that is an armchair; I am probably not contemplating a sentence at all in my mind in order to have the knowledge that that is an armchair). The stakes are pretty low. I am not going to act on the armchair. I am just reading Stanleys book. Suppose I were in a situation in which my practical interests made it crucial/important for me to know that that was a chair. (Suppose I receive a telephone call from Raffaella Carr the famous Italian TV showgirl - in which she says that if I have an armchair in my sitting room, I will win 60.000 euros). Would this change in the situation alter my mental panorama? Would I have to seek more reliable evidence to the effect that that is an armchair? It is not clear (to me) that by a change in the perception of the stakes I would, in such circumstances, stand up and touch the armchair to see if this is a solid armchair or, otherwise, the appearance of a chair. Yet, it is implicit in Stanleys argument that something of the sort should take place. My other doubt concerns the mental panorama of the knower. In what ways does the mental panorama of the knower change when he becomes aware of high stakes? Suppose that the knower is entertaining the thought I know that that is a barn (although I doubt that knowledge manifests itself in such thoughts; it is usually of the tacit sort). Suppose that the knower looks at the barn for an hour, having the same visual stimuli on his retina all the time. Supposedly, one hour is quite a lot of time and many things can change. In one hour, suppose that the knower will shift several times from a high stakes situation to a low stake situation. Now, I wonder whether his mental panorama has changed. Should we say that, if he had knowledge that that was a barn, that piece of knowledge was stored in a place in his brain and then subsequently removed, and then stored there again, and then removed? I think that this process of storing, erasing knowledge is implied by Stanleys theory. This seems to me have a high cost, since, by hypothesis, the alternation between high and low stakes could happen many times and we would therefore have to say that knowledge is stored and then removed, when we obviously have the intuition that, in this case, whatever it is that is in the mental panorama of the knower stays there and is never removed. Removing items of knowledge leaves traces in the memory. Should we then say that while the knower looks at that barn, while the high/low stakes situations alternate, the knower goes through a huge number of changes and also has a huge number of memories? Well, I doubt very much that human beings form knowledge in a way that is so sensitive to changes between situations. It is not even clear, in this situation, how memories of pieces of knowledge hook up to items of knowledge. Should the memories be formed when a piece of information, due to (a perception of) a high stake situation, is elevated to knowledge or otherwise when a piece of information is downgraded to mere belief. And are there memories of mere beliefs or are there just memories of items of knowledge? And given the sudden shifts from knowledge to (mere) belief and from (mere) belief to knowledge how many different types of memories should we have of the item of knowledge due to a constant perception for the length of an hour of a barn? Questions of this type are not absurd (This is not to imply that Stanley cannot answer them). On p. 10, Stanley writes: A standard use of knowledge attributions is to justify action. When I am asked why I went to the store on the left, rather than to the store on the right, I will respond by saying that I knew that the store on the left had the newspapers I wanted, but I did not know whether the store on the right did. When my wife asks me why I turned left rather than going straight, I reply that I knew it was the shortest direction to the restaurant. When it turns out that it was not a way to go to the restaurant at all, my wife will point out that I only believed that it was the shortest way to the restaurant. To say that an action is only based on a belief is to criticise that action for not living up to an expected norm; to say that an action is based on knowledge is to declare that the action has met the expected norm (p. 10). I tend to agree that successful action is based on knowledge, rather than on mere belief. However, action needs to be undertaken even when we have no guarantee that it will be successful. It is unfortunate that Stanley, in his discussion of knowledge, concentrates mainly on his bank examples (bank = financial institution), in which a claim to knowledge (X knows that the bank is open on Saturday) is sometimes true, sometimes false, depending on what is at stake for the putative knower. The most interesting cases of knowledge are those in which a piece of knowledge is transmitted from a person to another (through testimony, hearsay knowledge, etc.). And we often have to act (if we intend to act at all!) on the basis of this (volatile) kind of knowledge, in which we believe we know p because we believe that our informer is trustworthy. Now suppose that we obtain the information p through transmission; arguably we would have to say, like Stanley, that if the stakes are high, we do not know that p and hence we will tend not to act at all. However, if the situation is such that the stakes are high and there is urgency to act, we do not hesitate to act on the basis of the information that p (received through hearsay knowledge). Should this count as knowledge or not? Suppose I have a job competition and I am looking for a certain address. Inevitably, I may ask the people I meet on my way. Most of them will give me useful instructions, even if some may make a mistake and send me in the wrong direction. By asking a number of different people, counting on my confidence that, in general, people tend to be honest (and that society is based on honesty), I will be able to find my way, and if I find my way this way, then I have acted on beliefs which turned out to be knowledge only at the end, when my destination was reached and successful action guaranteed that my belief was knowledge. So, I am inclined to reject Stanleys persuasion that we act on knowledge and not on mere beliefs. The truth is that, in general, from the subjective point of view (with the exception of cases in which I obtain knowledge through perception), it is hard to differentiate belief from knowledge. What Stanley calls mere belief is belief dissociated from knowledge (which will never amount to knowledge because the proposition believed fails to be true). Surely, we tend not to act on mere belief, yet Stanley must admit that from the subjective point of view it is pretty hard to differentiate between mere belief and belief/knowledge that p. In order to differentiate mere belief from knowledge that p an objective stance is required it is not enough that a person will try to make up her mind about the status of a proposition. Instead, all the cases discussed by Stanley are cases in which a subjective stance is involved: and it is in such cases that it is hard to differentiate mere belief from knowledge. In fact, one should make the prcis that mere belief is a category that makes sense only from an objective point of view, since from a subjective point of view it makes little sense. Not even a mad man believes that when he believes that angels speak to him and tell him what to do, he is faced with mere belief. He certainly has reasons to believe that he believes that p and that he has knowledge that p. After all, are not his voices real enough to him? Returning to the more ordinary case of knowledge imparted through hearsay, should we believe or not believe what we are told? If we believe what we are told, we act, otherwise we dont. Suppose we act, should we think that we are acting on mere belief? Well, there are cases in which one acts on mere belief too. Suppose I receive a telephone call, and someone tells me that my best friend Nino has had an accident and has a broken leg, now being in the local hospital. There is something in the telephone call that alerts me to the possibility that it is a joke. Yet, I care a lot of my friend. I try to phone him to ascertain the truth but he does not answer me. The hospital is nearby. To allay my doubts I go to hospital and I ask if someone with a broken leg has been hospitalized in the last hour. Well, this is a case in which I have acted on mere belief, being myself sceptical about the content of the telephone call, but I have acted nonetheless, because the cost of acting was less than the cost of not acting (by acting I immediately found out the truth, while if I had not acted I would have doubts for several hours). Stanley may reply But it was not reasonable to act; you should not have acted. Yet, this is exactly a case in which the stakes are pretty high, I do not take a piece of information as knowledge, but I act nonetheless to allay my worries. Is not my state of mind important? Why should I be in anguish while I could easily come out of this state? Although I cannot say that everyone would have acted like me, my response from my point of view was pretty reasonable. Stanley claims that contextualism about knowledge is not plausible. First of all, he summarises the contextualists project through claims 1 and 2, provided below: Claim 1: Knowledge states are ordered according to epistemic strength; Claim 2: What determines the semantic value of instances of knows that p, relative to the context of use, is some collection of facts about the intentions and beliefs of the conversational participants in the context of use. Then Stanley proceeds to discuss two cases in which he believes contextualists to make the false predictions. In the Low attributor-High Subject Stakes contextualists make the prediction that the assertion Hannah knows that the bank is open on Saturday is true, while (for Stanley) it is false. In this situation, the attributor, Jill, does not perceive that Hannah is in a high stake situation. Since it is the speakers intentions that fix the semantic content of the assertion, the assertion Hannah knows that the bank is open on Sunday involves lower epistemic standards, thus it should count as true, while in fact Stanleys intuition is that it should count as false. Another situation in which the contextualist makes wrong predictions is ignorant high stakes. In this case Hannah is in a high stake situation, but does not know that. If it is Hannahs intentions that fix the semantic content, then the content of Hannahs assertion I know that the bank is open on Saturday should be as in the Low Stakes situation. So Low Stakes and Ignorant High Stakes situations should be on a par, which is not the case according to Stanley, as in Low Stakes Hannah knows that the bank is open on Saturday is true, while in Ignorant High Stakes it is not. I suppose this last case is problematic. Presumably, the contextualists might reply saying that the subjective point of view is what matters for action. Since Stanley has made a connection between knowledge and action, they are pleased to accept this connection and make use of this notion to explain that actions spring from a subjective point of view, thus making the case that in the Ignorant High Stakes situation the claim I know that the bank is open on Saturday is true. As a neutral reviewer, I have to say that this reply is not misguided. Stanley thinks that the best merit of contextualism about knowledge is to defeat philosophical scepticism: The sceptics error is to import facts about her own context into the interpretation of ordinary knowledge ascriptions, thereby coming to the conclusion that ordinary knowledge attributions have the same truth value as they would have, if uttered in the sceptics context. The putative feature of language-users explains the Sceptics fallacy. By appealing to semantic blindness, the contextualist seeks to accommodate the force of sceptical arguments, while divesting them of the threat they pose to ordinary epistemic practice (p. 29). I do not see great merit in such attempts to defeat the Sceptic. As far as I know the best attempts to deal with scepticism are in Grices analysis of phenomenal verbs (Grice 1989, 374-375; see the discussion of Grice in Turner (2002), as well as in Prices book Perception and in Strawson (1992). At least in those days, the Oxford School of Philosophy eliminated sceptical doubts once and for all (Why is it that Stanley does not seem to be aware of this?). Anyway, it is interesting to see how Stanley characterises contextualism in his own words: Contextualism is the doctrine that the relations denoted by know with respect to different contexts of use, come in higher and lower strengths. In particular, for this sort of contextualist, the word know has a content that is a function of the epistemic standards in the context. When Hannah finds out that she must deposit her check before the day is out, intuitively her evidence must satisfy a higher epistemic standard in order for her to know that the bank is open. The standard contextualist accounts for this by supposing that the word know changes its content in the new context. It expresses a relation that Hannah stands in to a proposition only if her evidence for that proposition satisfies the higher epistemic standards (p. 34). Contextualism about knowledge entails that the same sentence X knows that p, in different contexts, can express different semantic contents, these contents being differentiated in that, in one case, one would be faced with an epistemically stronger expression, and in the other case one would be faced with an epistemically weaker expression. If this was the case, then the predicate know would have to function like gradable adjectives and ought to be combinable with degree modifiers. In the same way in which we have (1) John is really tall, we would have to have (2) John really knows that p, where (2) would have to be epistemically stronger than John knows that p. Stanley shows that really does not really work as an intensifier (as is shown by *John knows that p but does not really know that p (really being used as an intensifier) (On this case Stanley is a bit reticent, but I would say that really has a plausible interpretation as It is a real fact that (very much in the same way as Obviously p means p and p is obvious); see Capone (1998)). Stanley is also right in saying that in John knows very well that p very well does not act as an intensifier, but as a discourse marker (Stanley does not suggest an interpretation of very well though). Stanley is also right in saying that an apparently comparative construction such as Hannah knows better than anyone else that she is poor is an idiomatic expression (my interpretation: If there is one who should know that she is poor, that is Hannah). However, there are expressions which Stanley ought to examine such as: I know for certain that p; I am absolutely sure that p (from Strawson, personal communication, 1994); These show that there are degrees of confidence in ones knowledge that p. There are also languages such as Italian in which a clitic expression serves to make the knowledge claim firmer (usually the clitic lo qualifies a proposition as shared knowledge; however, in contexts in which it is evident that this is not its function, it makes the knowledge claim more emphatic). Consider the following example: No, no, non che io so che Maria a Parigi, io lo so. (No, no, it is not just the case that I know that Mary is in Paris, I know that for certain). In Capone (1998; 2000), I provided cross-linguistic evidence that clitics in other languages such as Spanish, Portuguese, Polish, Modern Greek, Serbo-Croat, and Czech clitics work in a similar way. It is not amazing that ordinary language should find a way to express more or less confident knowledge and this is a fact known at least since the publication of Malcolm (1949) (also cited in Grice 1989). If there are really attested uses of know as a gradable epistemic term, then one could opt for an ambiguity hypothesis or for a contextualist approach, according to which the content of a knowledge attribution is fixed in context. The contextualist approach avoids the multiplication of senses and thus abides by Modified Occams Razor (see Grice 1989). In general, in order to posit ambiguity, one seeks the evidence of languages in which what correspond to two senses of a (single) word in another language are lexicalized differently (there are two lexemes each representing a different sense). The clitics in Italian and suchlike languages do not really attest to an ambiguity proper but to pragmatic processes going on in the language, where looser senses of sapere (kow) are circulated and make pressure on the language to change. Clitics are associated with the standard, not with the loose uses of sapere (know) (In addition, they can be used to make a firmer knowledge claim, say, by conversational implicature). Stanley denies that knowledge attributions are gradable and opposes the view by Cohen (1988) that knowledge is an indexical and, as such, one speaker may attribute knowledge to a subject while another speaker denies knowledge to that subject, without contradiction. He also opposes DeRoses (1992) view that the lack of contradiction is the key to the sense in which the knowledge attributor and the knowledge denier mean something different by know. Stanley wishes to contrast know with other expressions to which he has a contextualist stance, such as tall. However, he refutes automatic tests such as The Collective Description Test proposed by Cappelen & Lepore (2005). In general, he refuses to accept that there are tests allowing us to draw a line between context-sensitive and context-invariant expressions; in particular he takes issue with The Collective Description Test according to which if a predicate with a stable meaning is predicated of two different subjects it is possible to make use of a collective description and to attribute the predicate to that description on the assumption that the predicate has an invariant semantics. However, Stanley writes: Consider two sisters, Jill and Mary. John utters Jill loves her mother and Bill utters Mary loves her mother. One can conclude from the two premises offered that Each sister loves her mother. Since one can describe what Jill and Mary have in common by the verb phrase loves her mother, it follows that Cappelen & Lepores test falsely predicts that loves her mother is not context-sensitive (p. 50). Stanley thinks that the only way out for C & L is to posit that loves her mother is not context-sensitive but a case of controlled anaphora where her is bound by some higher operator. However, in case two utterances are proffered such as Jill is tall and Mary is tall we can draw the conclusion Both Jill and Mary are tall and the contextualist can reply that this too is a case of controlled anaphora in which a pronominal is bound by an operator. However, to put in practice what Stanley says we need a logical apparatus such as the following: Each of them is tall for e, where e ought to be an empty category such as a trace or pro or PRO. However, PRO occurs in non-finite clauses and is ungoverned. So e cannot be PRO. It cannot be pro because pro is not available in object position in English. For e to be a trace, there ought to be movement at logical form: However, a sentence such as (7) Each of them went to the cinema shows that Each of them does not work as an operator and does not necessitate movement from an extraction site at deep structure or at logical form. So we do not know what kind of empty category e is. Furtheremore, Cappelen & Lepore need not defend themselves in the way Stanley proposes. They may simply say that the Collective Description Test probably needs further constraints. A plausible constraint is that the substitution of the individual phrases by a collective description necessitate that no pragmatic interpretation work be done (instead, when Stanley moves from Jill and Mary to Each sister he makes use of pragmatic interpretation identifying the referents of the proper names with a collective description, which adds an additional mode of presentation of the references (sisters)). Anyway, Stanley casts doubts on general tests for context-sensitivity and wants to compare the uses of a putative context-sensitive expression such as know and a widely-accepted context-sensitive expression such as possible. He compares two discourses, and finds the former non-acceptable and the latter acceptable, this being a proof (for Stanley) that know is not context-sensitive. Zoo (looking at a zebra in a normal zoo). I know that is a zebra. But can you rule out its being a cleverly painted mule? I guess I cant rule that out. So you admit that you dont know that is a zebra, and so you were wrong earlier? I didnt say I did. I wasnt considering the possibility that it could be a cleverly painted mule. Technology Its possible to fly from London to New York city in 30 minutes. Thats absurd! No flights available to the public today would allow you to do that. Its not possible to fly from London to New York City in 30 minutes. A. I didnt say it was. I wasnt talking about whats possible given what is available to the public, but rather what is possible given all existing technology. Well, what would Stanley say about the example (8)? I know and I do not know that Mary is honest (Higginbotham personal communication, 1994). Is this a permissible use or not? And what would be meant by that? Ordinary usage does not have any exact logic (to repeat an idea by Peter Strawson) and linguistic uses cannot be easily tamed by the logician. I doubt that a clear reply to the contextualist can be given by resorting to intuitions about usage. Given a range of uses, we have to establish to what extent they are literal or extended. So, through an investigation of usage, we may end up proceeding in circles. I think that conformance to rigorous philosophical methodology such as e.g. Modified Occams Razor can solve otherwise insuperable problems. What is important is to look for a unified semantic analysis of know at most allowing context to add features, not to change truth-values altogether. When the contextualist says that an assertion of X knows that p has different truth-values in different contexts in some ways he implies (albeit he is not aware that he implies that) that the expression is semantically ambiguous. Where should the logical forms that make the assertion X knows that p true in one context and false in another come from? A good point Stanley makes is that know is unlike an indexical whose character determines the value that the indexical takes on in context. There is no character associated with know determining a strong interpretation say in a Sceptical context and a weaker interpretation in a ordinary context. In the case of I it is the character of the expression that determines that the referent of I changes in accordance with the person who proffers the sentence containing I. There is nothing in the character of know implying that the content changes depending on the kind of context. After all, I imagine Peter Strawson smiling while in a conference full of sceptical minds he uses know in a perfectly ordinary way implying no strong epistemic meaning. This is enough to show that it is not the character of know to determine automatically its interpretation in context. At most one could claim that the speakers intentions determine the content of know. Even so, I wonder whether the position whereby know can determine either a strong or a weak meaning is sensible unless its ambiguity is presupposed. When the speaker uses know in its strong form, is he not using a knowledge attribution with a certain logical form? And where does this logical form come from? Unless one accepts the ambiguity view, the only way to make sense of the contextualist position is that both the strong and weak reading of know come from an undifferentiated concept, say know, which is neither epistemically strong nor epistemically weak. But I suppose that such an undifferentiated know is what Cappelen & Lepore have in mind in their Insensitive Semantics. Why not stop at that? Stanley discusses an assertion by Cohen (1999): Does knowledge come in degrees? Most people say no (though David Lewis says yes). But it doesnt really matter. For, on my view, justification, or having good reasons, is a component of knowledge, and justification certainly comes in degrees. So context will determine how justified a belief must be in order to be justified simpliciter (p. 60). Stanley takes issue with the idea that being a gradable expressions necessitates contextual interpretative augmentation. To prove his point he uses the following example: taller than 6 feet. This expression is gradable but not context-sensitive, according to Stanley. Hence he concludes that So gradability does not entail context-sensitivity. Well, I think that the moral Stanley draws from this particular example (9) is wrong. Suppose I am hiring bodyguards and I make the not too unreasonable assumption that I need bodyguards who are at least 185 cm tall, then when John, who is 175 cm tall, asks to be employed and I say to him (10) I need someone taller I certainly do not mean that anyone at all (who is taller) will do, but I mean that I need someone who is much taller. So, unlike Stanley, I think that contextual augmentation is involved in assessing the meaning of I need someone taller. Stanley argues that the contextualist needs the following: The metalinguistic Containment thesis For any context c, the word know expresses a relation that, relative to that context, contains as a component the property expressed by the word justified relative to c (p. 80). Stanley argues, however, that one cannot provide support for this thesis by appeal to the oddity of uttering instances of the schema X knows that p but X isnt justified in believing that p. Ok. I think we agree with that. But what about the following kind of statement? (11) John knows that p; he isnt justified in believing that p, relative to c, but he is justified in believing that p relative to c. I agree that this is not the kind of statement one would make in everyday discourse, but I imagine that a contextualist would find this acceptable in the context of philosophical disputes. I suppose that (11) would support a view according to which justified in believing that p is context sensitive. However, at this point (and Stanley does not stress this) the contextualist is under pressure to make his proposal more explicit. Does his proposal amount to accepting that justified in believing that p contains a hidden indexical, say a 0 PP? Then there should be at least two of such hidden indexicals 0 and 0. Say 0 corresponds to to degree x and 0 corresponds to to degree y, where y is different from y and x involves a greater degree of justification. At this point, Stanley can capitalize on such a detailed account, by saying that, after all, for what he knows, there could even be, say, many different PPs, that is gradations of justification. Is it implausible that a language should be so accurate in distinguishing between degrees? By an analogous reasoning, one could also have a huge number of gradations but this is contrary to our intuitions. Alternatively, the contextualist should posit just one hidden indexical, say 0, 0 taking on specific (different) values in context. Ok, things for Stanley do not change much, as he is always in a position to say that, given the possibility of a high number of gradations, a claim to knowledge is justified relative to a context, but not relative to others. Given the possibility of many gradations, how should we chose the gradation that is most suitable for a context? How do we know which gradation the putative knower or attributor had in mind? (The reader may wonder whether I have abandoned the persuasion that there can be various expressions of knowledge, involving different degrees of confidence. Well, suppose I accept that X knows that p has got a univocal semantics, then I could accept that the standards of justification are increased by combining know with for certain or for sure, and all I mean by using these adverbials is that my justification enables me to exclude every chance that not p. The problem of too many gradations here does not arise, as each gradation is matched by a different type of expression and the adverbials that combine with know are not infinite). Having rejected contextualism about knowledge claims, Stanley puts forward his theory about interest-relative invariantism. He says: Bare interest-Relative Invariantism (henceforth IRI) is simply the claim that whether or not someone knows that p may be determined in part by practical facts about the subjects environment (). The advocate of IRI simply proposes that, in addition to whatever ones favoured theory of knowledge says about when x knows at a time t that p, there is a further condition on knowledge that has to do with practical facts about the subjects environment (p. 84). What characterises the text above is its objectivism. There is no talk of the knowers intentions or of the attributors intentions. It does not really matter for Stanley whether the knower is aware of his environment. Stanley writes: () practical facts about a subjects environment at a time t might make it the case that the subject must have stronger evidence than usual in order to know a proposition p at that time than she must possess in order to know that proposition at other times, where strength of evidence is measured in probabilistic terms (p. 84). The question that turns up in the readers mind at this point is: Why is this the case? In fact, under contextualist assumptions Stanleys considerations follow for free; given that in different contexts, the sentence X knows that p expresses different propositions, in a context where the stakes are higher it is more likely that the speaker utters X knows that p having in mind a stronger justification relation. But all this has been rejected by Stanley. Stanley makes a prcis on p. 86: So according to the advocate of IRI, a knowledge ascription such as Herman knows at 1:30 pm on September 24, 2004 that Hillary Clinton is a Democrat expresses the same proposition relative to every context of use, and hence is not context-sensitive. There is no specifically epistemic sense in which knowledge-attributions are context-sensitive. According to IRI, there is a univocal knowledge relation, which is sensitive to the subjects practical situation at the putative time of the knowing (p. 86). Now we are left with a sense of mystery. Why is it that the evidence must be greater when a knowledge claim is made in a high stake situation than in a low stake situation? The answer provided by Stanley has to do with the knowers practical interests: The theory I will consider is one according to which knowing a true proposition requires a subject who believes it to possess a sufficient level of evidence for that proposition, where sufficiency is measured in terms of some kind of probability. The basic idea is that, the greater the practical investment one has in a belief, the stronger ones evidence must be in order to know it, where strength of evidence is measured probabilistically (p. 88). Connecting the issue of the strength of ones evidence with ones practical interests is not a way of defending or justifying IRI, since the same type of connection militated in favour of contextualism. So the mystery is not resolved. Can the solution lie in a definition of the verb know? Perhaps the author is able to excogitate an invariant semantics for know that explains why X knows that p is true in one situation but not in another. Consider the definition of know he provides on p. 89: Knows () if and only if (1) p is true at w; (2) p is not a serious epistemic possibility for x at w; (3) if p is a serious practical question for x at t, then p has a sufficient low epistemic probability, given xs total evidence; (4) x believes at t that p on the basis of non-inferential evidence or believes that p on the basis of competent inference from propositions that are known by x at t (p. 89). This definition does not say anything about the comparative dimension of knowledge, and specifically it does not require that justification be greater in high stakes than it should be in low stakes. However, it has a comparative dimension by (logical) implication. All knowledge situations in which the stakes are not high do not require the application of point (3); in such cases, point (3) of the definition is skipped (is not applied), hence the truth of a knowledge claim which in a different situation (high stakes) would be deemed false. There is the question of who has to judge whether [if p is a serious practical question for x at t, then p has a sufficient low epistemic probability, given x total evidence]. Stanley on p. 101 (High Attributor-Low Subject Stakes) clearly says that the knower need not be aware of the high stakes involved in the situation. It appears, therefore, that the truth of a knowledge-claim will be relativised sometimes to the knower, sometimes to the attributor. If it is relativised to the attributor, I suppose that Stanley can no longer claim that this is a theory of the mental panorama of the knower. This is evident, when he talks about consequences of what one knows. Is it not obvious that if one knows that p, and p entails q, then one knows that q? Well, this is what Stanley has to deny. In fact, he writes: Suppose that p is not a serious practical question for x at t. But then at t, x infers a consequence q from p, and q is a (or thereby becomes) a serious practical question for x at t, according to the above characterization. Realizing that q follows from p would then change xs goals and interests. If so, then p becomes a serious practical question for x at t. So competently inferring a conclusion is in fact a way not only of gaining knowledge, but also of undermining knowledge (94). Now, while Stanleys discovery is probably very interesting and surprising, it shows that his definition of knowledge does not make it ideal for capturing the knowers mental panorama. It appears that one should always revise ones items of knowledge when he encounters different situations thus one is always revising ones own mental states. It is very hard to call a mental state a mental state under such a theory (this is not to say that Stanley is not making some illuminating points about mental reality; if I am wrong, he is). In a way, acceptance of clause (3) is like requiring that the proposition believed should come under a mode of presentation. Of course, we know that knowledge is semantically opaque, but now we are subordinating the content of the knowledge claim to the perception of a mode of presentation and we are saying that, due to semantics, if one is in context a X knows that p is true, while if one is in context b, X knows that p is false. The definition, however, is not complete, because it is not clear whose mode of presentation must be taken into account: the speakers or the knowers? Unless Stanley adopts a disjunction and says that either the speakers or the attributors mode of presentation is involved, there are problems as his discussion of High Attributor/Low Subject Stakes makes clear (pp. 98- 102). I should note in passing that clause (3) of the definition [if p is a serious practical question for x at t, then p has a sufficient low epistemic probability, given x total evidence] is ambiguous: for x at t does not mean from xs point of view at t, x is not the experiencer but the person affected. There is in the sentence a hidden understood implicit argument from x or the attributors point of view. O.k., some amendments are needed, like the ones I propose or alternative to the ones I propose. It is straightforward that the definition explains why in a high stakes situation one need not know that p, while one knows that p in a low stakes situation. Stanley, however, thinks that his view has some advantages over the contextualist in connection with the High Attributor/Low Subject Stakes. Let us see what he says in reply to contextualists. A contextualist way of tackling the situation is the following: Hannah is in high stakes and is aware that she does not know that p. Since she does not know that p, she is not able to assert that p. However, if she were to say that Bill knows that p, she would have to say that p by implication, since X knows that p entails p. Hence Hannah cannot assert that Bill knows that the bank will be open on Saturday. However, the contextualist does not successfully explain why Hannah can assert that Smith does not know that the bank will be open. Just because a proposition is not assertible, this does not mean that its negation is assertible. Well, Stanley must concede that, according to the contextualist too, Hannah knows that she does not know that p (in the strong interpretation of know presumably). Furthermore, she knows that Bill does not know (in the strong interpretation of know) that the bank will be open on Saturday given what Bill says in reply to his question (and given what he says, in response to her question whether the bank will be open on Saturdays); and she knows that Bill does not know that the bank is open on Saturdays in the strong sense of know because his answer does not eliminate all chances or most chances that p is false. It is true that the contextualist cannot say that Hannah knows that Bill does not know that the bank will be open on Saturday independently of Bills reply, but this is also true of Stanleys explanation, since if Bill intends not to give a true or exhaustive reply to Hannah and he in fact knows (independently of what he says) that the bank is open on Saturdays, then Stanleys explanation miserably fails. Let us now examine Stanleys positive proposal. He writes on p. 101: Here is an intuitively plausible account of what is occurring in High Attributor-Low Subject Stakes that is preferable to one that relies upon misjudging the likelihood of counterpossibilities. Let us suppose that High Stakes is a name for some person for whom the proposition p is a serious practical question. The account arises naturally from reflection upon the purposes someone in High Stakess situation would have in enquiring whether someone else knows that p. When High Stakes wants to know whether another person knows that p, this is the case presumably because High Stakes has an important decision to make, one that hinges upon whether or not p (this follows naturally from the fact that p is a serious practical question for High Stakes). So High Stakes interests lie in establishing p; that is, in acquiring information that will allow her to know that p. What High Stakes is interested in finding out, then, is whether, if that person had the interests and concerns High stakes does, that person would know that p. Since p is a serious practical question for High Stakes, she is not really worried about that persons own interests and concerns (p. 101). Now, there are two points to make: one could inquire whether another person knows that p regardless of ones interest in knowing whether p is the case or not. One may simply be interested in, say, Xs mental state, which in passing informs one that p is true. Now, it is clear that in this situation Stanleys explanation does not work. The other point to be made is that the explanation does not work because it is not about the knowers mental state. It works only on the basis of the attributors mental state but intuitively this is not how to define knowledge. Stanley believes that there is some promise to an interest-relative explanation of the grip of scepticism. It is very intuitive to suppose that, where ordinary action is concerned, sceptical scenarios are not serious practical questions. He writes: Suppose it could be argued that, relative to the evidence we ought to possess, the probability of a sceptical scenario occurring is sufficiently small (given perhaps some sufficiently objective sense of probability). If so, then sceptical scenarios will not have an effect on the warranted expected utilities of the action at my disposal. Then, despite the fact that the obtaining of a sceptical proposition might adversely affect the utility of some of my actions at my disposal, their extremely low probability would exclude them from being serious practical questions (p. 127). Well, I think we could also draw the following conclusion. The surgeon who must operate his patient may be paralysed by consideration of the seriousness of his practical question and may have to lower the standards of justification required to act efficiently. The real problem is how to find the right balance between too much sensitivity to a serious practical question and too little. Concluding, I am pleased that his book was written. Although I have criticised almost every part of it I do not write more because of severe space limitations My persuasion is that this is a very important book, which offers a serious alternative to contextualism. It is probably the assertions I have criticised most that will become important in philosophy. Although I do not agree with this anti-mentalistic picture of knowledge, it is possible that knowledge has more to do with the way the world is and what its stakes are than with the nave mental panorama of the knower. After all, in some cases we may fail to know that we know p (Williamson 2000). I applaud this extremely surprising and illuminating book. References Capone, Alessandro, 1998. Modality and discourse. D.Phil. thesis, University of Oxford. Capone, Alessandro, 2000. Dilemmas and excogitations: an essay on modality, clitics and discourse. Armando Siciliano, Messina. Capelen, Herman & Lepore, Ernest (2005). Insensitive semantics. A defense of semantic minimalism and speech act pluralism. OUP, Oxford. Cohen, Stewart, 1988. How to be a Fallibilist. Philosophical Perspectives 2, 91-123. Cohen, Stewart, 1999. Contextualism, Skepticism, and relativity. Philosophical perspectives 13, 57-89. De Rose, Keith, 1992. Contextualism and knowledge attributions. Philosophy and Phenomenological Research 52/4, 913-929. Grice, H.P., 1989. Studies in the way of words: Harvard University Press, Cambridge, Ma. Malcolm, Norman, 1949. Defending common sense. Philosophical review, January 1949. Price, H.H. (1961). Perception. Methuen, London. Strawson, P.F. (1992). Analysis and metaphysics. OUP, Oxford. Turner, Ken (2002). A note on the neo-Gricean foundations of Societal pragmatics. International Journal of Pragmatics, 1-19. Williamson, Timothy (2000). Knolwedge and its limits, Oxford: OUP. Biography Dr Alessandro Capone studied with Yan Huang and James Higginbotham at the University of Oxford, where he obtained a doctorate in linguistics (D.Phil). He carried out research and taught at the University of Messina. He carries out pure research in the Department of Philosophy of the University of Palermo. Among his publications, one can read Belief reports and pragmatic intrusion: the case of null appositives (Journal of Pragmatics 2008). Alessandro Capone Department of Philosophy Viale delle Scienze University of Palermo Email-address: alessandro.capone@istruzione.it  I would like to thank Jacob L. Mey, Yan Huang, Ernest Lepore, Leo Hoye for their feedback. .      1234]_y~T  7 8 9  K  ۿۯۦ󞖞~vnfnvv^Vh<mH sH hoxmH sH humH sH hfHmH sH hmH sH hsZmH sH h[H7mH sH hrmH sH hHJmH sH h|emH sH h w5mH sH h wh w5CJaJmH sH h whsZ5mH sH jhH+0J5UmH sH h wh w5mH sH h wh5mH sH h wh|e5mH sH !38 9 ^ _  @AEFZ$$0d]0^a$gd Z#$0d]0^a$gd|e 2]2^gd|eh ] { E M ^y?@jptw#5E6A 79;KvwиȸظظظظhKSmH sH h Z#mH sH h JmH sH hV+mH sH hE5mH sH h/_mH sH h` mH sH hzgmH sH hsZmH sH h mH sH hE9mH sH h|emH sH hoxmH sH ;w|(2 w!;#B#R#Y##!$X$Y$Z$''((*y*z***++E+F+,,X1[1^1d1e1i1122444+6ȸ𰸰ȨȨȨИh8tmH sH h%;mH sH h=LmH sH hCmH sH humH sH h/_mH sH h9gcmH sH h` mH sH hsZmH sH h/SmH sH h JmH sH h Z#mH sH hO0mH sH :Z$[$u$v$''+,4455_9`92;3;;;< <==DDD$0d]0^`a$gd9gc$0d]0^a$gd Z#+6667777^9_9::2;;;n==??@@ @@@]@@@@@@@A.AAAAABCCDD%D,DgDDEEFFFF/GGGGȘ𘨘蘐𘈐h omH sH hdmH sH hmH sH h"1mH sH h` mH sH h<mH sH hO0mH sH hwmH sH hsmH sH hoxmH sH h8tmH sH h9gcmH sH h%;mH sH humH sH h )smH sH 5DDFFwHxHHHKKMMMMNN"N#NRR$0d]0^a$gdo$ & F0d]0a$gdo$0d]0^a$gdd$0d]0^a$gd Z#GGGH(H9H:HvHwHIIKKLLMMNNNOPPPDQIQRRRRRTJTOTPTQTTTTTTNUV!X*XиШzrhhmH sH hhmH sH hhmH sH hhhh5 hhh` mH sH hmH sH h<mH sH hmH sH h%;mH sH h1?mH sH homH sH humH sH he#hmH sH hdmH sH h omH sH h9gcmH sH ,R0R{R|RTTQTTTHZIZ^^``bbbbdd$0d]0^a$gd$$ & F0d]0a$gd$$0d]0^a$gd$ & F0d]0a$gd*XYY!ZEZFZGZHZ\\]]^^^``[btbbbbbb\c_c`ccdd?dAddee eee3f8fYfafgg]g^gȸȸ{sssȸh1?mH sH h {h {5>*mH sH h {mH sH hh5>*mH sH hh$5>*mH sH hh$mH sH hmH sH h$mH sH hmH sH hwmH sH hmH sH h4wmH sH h` mH sH hSmH sH hhmH sH -d?d@d e ekglghhh9iqiiiEj$ & F0d]0a$gdiM@$ & F0d]0a$gdiM@$ & F0d]0a$gdiM@$0d]0^a$gd$0d]0^a$gd$ & F0d]0a$gd^gegfggghgkghhhkkkkkkkkkkll.lZllln#n*n.nynnnnn`oaoooopqqpsssuuvvvGwغآآآآؚؚؒؒ蒺h$ mH sH huImH sH hmH sH h` mH sH h1?mH sH he#hmH sH hO0mH sH h<mH sH hhdmH sH hhiM@mH sH hiM@mH sH h2BmH sH hmH sH h {mH sH hmH sH 1EjFjQjj+kkkll\l]lao^w_wwwxxyy$ & F0d]0a$gdiM@$0d]0^a$gd Z#$ & F0d]0a$gdiM@$0d]0^a$gdiM@Gw]w^wxyy{{{x|y|~~./7<TȁɁ01`bOăCg|·؇vw`SииظȨȨؘؘؐh<mH sH h6mH sH h^dzmH sH h$mH sH hw<mH sH h4wmH sH hNHmH sH hkmH sH h` mH sH hRmH sH huImH sH h@tmH sH h+DmH sH hmH sH hFgmH sH 4yyyo{p{{{y||||}}~~./1Ńwxʊˊ$0d]0^a$gd@t$ & F0d]0a$gd@tӎԎUV$0d]0^a$gd@tSӗ2Sax}۠ PmKL%L]£ģצ &Щ輴̴̴̴̴ĴĴĘ̘̘h3DmH sH hmH sH hh-g5mH sH h-gmH sH h!mH sH hO0mH sH huImH sH hkhioS5mH sH hioSmH sH hkmH sH h6mH sH h,mH sH 9ѩBܭݭ"#²<=¹YZ|$0d]0^a$gd Z#$0d]0^a$gd@tЩѩϪ$(4ij?Onưұ[])/[r۶5U¹ԼԼ𼨼~~~~r~~hShy5mH sH hymH sH hyhymH sH hyh$mH sH he#hCJ(aJ(mH sH h(mH sH h$h$6mH sH h$mH sH h<mH sH hO0mH sH hSmH sH hh@95mH sH hmH sH h@9mH sH h@tmH sH ,¹ιev{к/UVXZ\eF_qz{| XYgiȾŹh?7mH sH hbmH sH h,nUmH sH h8mH sH h omH sH h oh o5mH sH h oCJ(aJ(mH sH hyhymH sH hfHmH sH hSmH sH hOmH sH h(mH sH hymH sH hmH sH 2Y<X$#9h$0d]0^`a$gdb$0d]0^a$gd|e$0d]0^a$gd Z#ȾоҾ+,;<NVXghiŽ𪠕jh/wiUh/wihH+mH sH hH+hH+mH sH jhH+0JUh|ehbmH sH hmH sH hOmH sH h ohdmH sH hdmH sH h omH sH hhmH sH h?7mH sH hbmH sH h&mH sH %$0d]0^`a$gdb,1h. A!n"n#$n% B@B NormaleCJ_HaJmHsHtHLA@L Car. predefinito paragrafoXiX Tabella normale4 l4a 4k4 Nessun elencoX@X H+Testo nota a pi di paginaCJaJT&@T H+Rimando nota a pi di paginaH*1׻`c׻389^_@AE F Z[uv#$,,--_1`12333334 455<<<<>>w@x@@@CCEEEEFF"F#FJJ0J{J|JLLQLLLHRIRVVXXZZZZ\\?\@\ ] ]k_l_```9aqaaaEbFbQbb+cccdd\d]dag^o_oooppqqqqospsssyttttuuvv.w/w1z{wxʂ˂ӆԆUVѡBܥݥ"#ª<=±YZ|Y<Xָ$#9hȻɻ˻̻λϻѻһԻջػ0000000000000000000000000000000000000000000000000000000000 0000 0000 0 0000 00000000000 0000 000000000 0 0 0 0 000 0 00000 00000000000 0000 0 000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000@000@0X00@0XX00@0X00@0xX00X00X  w+6G*X^gGwSЩ¹ȾbefhjlnpsuvxZ$DRdEjycgikmoqrtwyda?ػ?ػvCvDOv ovtv$=vd=v=v=v$>vd>v>v>v$?vd?v?v?v$@vd@v@v@v$AvdAvAvAv$BvdBvBwBw$CwdCwCwCw$DwdDwDwD w$E wdE wE wE w$FwdFwFwFw$GwdGwGwGw$HwdHwHwHw$IwdIwIwIw$JwdJwJ wJ!w$K"wdK#wK$wK%w$L&wdL'wL(wL)w$M*wdM+wM,wM-w$N.wdN/wN0wN1w$O2wdO3wO4wO5wdP6wP7wP8w$Q9wdQ:wQ;wdRw$S?wdS@wSAwSBw$TCwdTDwTEw$UFwdUGwUHwUIw$VJwdVKwVLwVMw$WNwdWOwWPwWQw$XRwdXSwXTwXUw$YVwdYWwYXwYYw$ZZwdZ[wZ\wZ]w$[^wd[_w[`w[aw$\bwd\cw\dw\ew$]fwd]gw]hw]iw$^jwd^kw^lw^mw$_nwd_ow_pw_qw$`rwd`sw`tw`uw$avwdawwaxwayw$bzwdb{wb|wb}w$c~wdcwcwcw$dwddwdwdw$e  ==ee k#k#D'D'R(R())**`1`13333 5 566v9v9<<J@J@@@GHHHHIILLIRIRTTVVXX{Z{ZR]R]``kbububccccc_o_oppqqqdrdrs7s t tztztuuzxzxyAzAzY|Y|nnʆʆԆԆ++{{TT ءءssPP nn|@@N##1ػ      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~DDll r#r#K'K'Y(Y(%)%)**g1g1:3:35566}9}9$<$<Q@Q@@@HHHHHIILLPRPRTTVVXXZZY]Y]``qbbb cccccfofoppqqqkrkrs=sttttuuxxyHzHz`|`|uuццۆۆ22[[ߡߡzzVVxJUU-88ػ  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~=*urn:schemas-microsoft-com:office:smarttags PlaceName=*urn:schemas-microsoft-com:office:smarttags PlaceTypeCR*urn:schemas-microsoft-com:office:smarttagsmetricconverter9*urn:schemas-microsoft-com:office:smarttagsplace8*urn:schemas-microsoft-com:office:smarttagsCity l90 175 cm185 cm6 feet ProductIDRRRRZ_{}mvw|G3T333D5R555,:::C<P<??@@@@CCRJZJJKEKKKLLAMFMXM_MNN1O>OPPCQIQLQOQQQQQ:RERdSlS[ThTTTTT%X-X0X8XYY ]]]"]%]+]ddRe_ekgxg1k9knnnnnnooooppr rttttwwxxyyzz{{!lyq~ MXҜ՜̞ٞ9GKXġΡWdϦβVaHOhq|ŵ&1o|~ٶ!~FI"Hghɻɻ˻˻̻̻λϻѻһԻջػ 3 UgNV'1((*+//Z8f8Y9[9???@D@EE#F(FHHtS~STTUU{WWWWXXZZ[[@\E\`aaa~bbdd#f1fffii kkooq.qqqBsIsuvvvvxxzzc}h}NVݗܘܛ[amz%/7Vl>H)8մִ|2oȶжٶ÷0I͸Ѹhɻɻ˻˻̻̻λϻѻһԻջػ3333333333333333333333333333333333333333333333333333333333333333333333333333333333333EEEEHo]odqqqqpsss "`BXYoٶ;0Klhɻɻ˻˻̻̻λϻѻһԻջػhɻɻ˻˻̻̻λϻѻһԻջػAr¬ J*9%X$ IN#ZHg]s0n^`o(. ^`hH.  L ^ `LhH. ^ ^ ^^ `hH. ..^.`hH. L^`LhH. ^`hH. ^`hH. nLn^n`LhH.^`o(. ^`hH.  L ^ `LhH. ^ ^ ^^ `hH. ..^.`hH. L^`LhH. ^`hH. ^`hH. nLn^n`LhH.^`o(. ^`hH.  L ^ `LhH. ^ ^ ^^ `hH. ..^.`hH. L^`LhH. ^`hH. ^`hH. nLn^n`LhH.^`o(. ^`hH.  L ^ `LhH. ^ ^ ^^ `hH. ..^.`hH. L^`LhH. ^`hH. ^`hH. nLn^n`LhH.^`o(() ^`hH.  L ^ `LhH. ^ ^ ^^ `hH. ..^.`hH. L^`LhH. ^`hH. ^`hH. nLn^n`LhH.]s#ZAJ*9% INqW3v"XV&06:QĴYX $ox$ 3D Z#H+V+O0E5?7[H7%;w<1?iM@2B+DfHuIHJ=LRS/SKSioS,nUsZ9gc|ezg-gFghe#h/wik o )s8t@tu4w^dzi}sNH8"1d~X6wOE9/_S( J w;w$&,b$y@9<C!rokhu {` @uuOuu׻@@UnknownGz Times New Roman5Symbol3& z Arial"1kF0Ć)p_Tp_T#n4 @HX?|e2intelintel      Oh+'0h   $ 0 <HPX`intelNormalintel41Microsoft Office Word@a@;@66@p՜.+,0 hp|  T_ '  Titolo  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz|}~Root Entry F[1Table{YWordDocument.SummaryInformation(DocumentSummaryInformation8CompObju  F#Documento di Microsoft Office Word MSWordDocWord.Document.89q