From considerations of vagueness, I argue that radical skepticism is false. First, I spell out what the skepticâ??s challenge is by reformulating the traditional skeptical argument in terms of possible worlds. This new look at the old argument brings to the fore a crucial, but unstated premise; i.e., that at some possible worlds many of our claims about contingent facts about the world (e.g., â??I am standingâ?) are true, but at other worlds, worlds at which a skeptical hypothesis is true, these claims are false. I take this crucial premise as my point of attack, arguing (1) that hard, skeptical invariantism about the verb â??knowâ?? is incorrect on the grounds that it disregards the role that the subjectâ??s interests play in determining the truth-conditions of her knowledge claims, and (2) that an Interest-Relative version of moderate invariantism that appeals to practical reasoning and utilizes one of John Hawthorneâ??s notions of practical environment is the best way of accounting for our epistemic world-accessibility.

1. The Skepticâ??s Challenge Spelled Out

Where p is a claim about some contingent fact about the world, the skeptic says we do not have access to the truth-value of p. What we need in our arsenal of cognitive abilities, in order to have access to the truth-value of p, is access to the necessary facts about the actual world. But, according to the skeptic, what we need is what we do not have. We do not have access to the world such that we have access to the truth-value of ordinary claims like â??I am standing,â? â??Her eyes are blue,â? and â??My doctor drives a Porsche 911.â? From the traditional way the skeptical arguments run, it is not always so obvious what the motivations behind the skepticâ??s challenge are. What exactly is the skeptic assuming in her talk of uneliminated possibilities? Of course all skeptical arguments qua skeptical arguments conclude that we do not know that p, where p is a claim about some contingent fact about the world. It may be helpful still to spell out what this means by putting the problem differently. The following is how the traditional argument usually runs:

The Traditional Skeptical Argument (TSA)

P1. I do not know that I am not a brain-in-a-vat (BIV).
P2. If I do not know that I am not a BIV, then I do not know that I am standing.
C. I do not know that I am standing.


(TSA) contains prima facie plausible premises from which the conclusion follows validly. When we first encounter this argument it may be that the biggest question on our minds is how this argument could contain such seemingly plausible premises and yet a conclusion that is anything but plausible. Indeed, the skepticâ??s conclusion is the least persuasive thing about her argument. In order to feel the full force of argument, there are a few crucial assumptions we must make along with the skeptic. The assumption I want to focus on, and bring to the fore, is the implicit claim that even if I were a BIV, I would still believe that I am standing. If we suppose the truth of this claim, what further claims might we deduce from it? For starters, we may say that the limitations of my cognitive abilities are such that the world would be no different to me if I were a BIV. As antiskeptics, our first inclination may be to try to undermine the plausibility of this latter claim. Some antiskeptics may want to do this by defending some version of direct realism or representationalism. But I think better is to uphold the plausibility of the hypothetical claim that the world would not be different to me if I were a BIV and to focus our defense of knowledge instead on the irrelevancy of the claimâ??s truth or falsity.

The approach I set out here does not attend to the problems of perceptual justification or to those of perception generally. The questions I am attempting to answer here are those about the nature of our epistemological language. The big question, then, is what all goes into making our knowledge attributions true or false? In our investigation into the skepticâ??s assumptions, including our spelling out the skepticâ??s crucial, but unstated premise, we may ask ourselves what the skepticâ??s answer is to the big question. Of course it is not altogether clear, to this philosopher at least, how the skeptic defines knowledge (e.g., what the necessary and sufficient conditions for knowledge are), for there is no standard positive definition of knowledge provided by the skeptic, only some implied criteria in the arguments. What is hinted at is that the skeptic might count S as knowing that p if and only if S believes that p and has eliminated every logical possibility that not-p. Our radical skeptic must harp on the idea of uneliminated logical possibilities to make her case. She reminds us that the contingently-true facts about our world may (logically) be false and that the world might (logically) be different from how we conventionally think it is. Our question, then, is how can we reformulate the skepticâ??s argument such that it makes explicit all of this modal-talk?

When we think of possible alternatives to our ordinary conception of something, X, we are thinking about (logically) possible ways that X can be other than the way we conventionally think it is. Ways we think the world is according to our perceptions (P-worlds, let us call them) are some of the possible ways the world can be. Other ways the world can be, worlds at which a skeptical hypothesis is true (or S-worlds), have it that the world is significantly different from the way we conventionally think it is, such that the truth-value of p at S-worlds is different from what it is at P-worlds. So the problem put in these possible worlds terms looks like this.

The Modal Skeptical Argument (MSA)

P1. I do not know which of the two kinds of worlds I am at--P-worlds or S-worlds.
P2. At P-worlds my claim that I am standing is true and at S-worlds my claim that I am standing is false.
P3. If I do not know which of the two kinds of worlds I am at, then I do not have access to the truth-value of my claim that I am standing.
C. I do not have access to the truth-value of my claim that I am standing.

Everything crucial to (TSA) is here. The difference is that the force of the skepticâ??s challenge has been spelled out. The premises as well as the conclusion have been restated in a way that draws out the important modal notions only implicit in (TSA). According to (MSA), the real problem is that, since at least two possible worlds--one at which p is true and one at which p is false--are indistinguishable in our access to them, we cannot know, properly speaking, which of the two kinds of worlds we are at. The (MSA) skeptic argues that we do not have any way, given the limitations of our cognitive abilities, of distinguishing between P-worlds and S-worlds. That is to say, the two kinds of worlds are subjectively indistinguishable. We may call this the Problem of Indistinguishable Worlds. I suggest that a solution to the problem of indistinguishable worlds is sufficient for a solution to the problem of knowledge. The skeptic has a certain view about epistemic world-accessibility against which the antiskeptic must attack. If the skeptic argues that we do not have access to the necessary facts about the world in order to know whether p is true, the antiskeptic will need to argue that we do, and further explain how it is that we have this access to the necessary facts about the world and therefore access to the truth-value of p. Once this explanation has been given, the antiskeptic is in a position to deny all of the premises of (MSA).

The first important lesson we can learn from the skeptic is that she is presuming the correctness of a hard version of invariantism about knowledge, which has it that the standards for knowing are fixed such that nearly all of our utterances containing the verb â??knowâ?? are false. The antiskeptic will need to make the case against this hard invariantism and further argue for the correctness of either some version of moderate invariantism, which would have it that the standards for knowing are fixed such that many of our utterances containing the verb â??knowâ?? are true, or some version of contextualism, which holds that the standards for knowing are sensitive to the context of discourse such that at least a great many of our â??ordinaryâ?? claims to knowledge are true (by virtue of the â??ordinaryâ?? context of discourse) and at least a great many of our claims to knowledge uttered in contexts in which a skeptical hypothesis has been made salient to the discourse are false (by vice, as it were, of this â??higherâ?? context of discourse).

In the following section, I briefly make such a case against hard invariantism, principally on the grounds that it does not take into account the important role the subjectâ??s interests play in determining the truth-conditions of her utterances, and also that skeptical invariantism destroys at least one intuition we have about the nature of knowledge; namely, its link with practical reasoning.

2. Skeptical Invariantist Treatments of â??Vagueâ?? Claims

Philosophers since Unger (1978) have played a game whereby they make their different points about knowledge by showing a supposed analogy between the verb â??knowâ?? and comparative adjectives like â??flatâ?? and â??tallâ??. The game is to try to show that your way of analyzing these comparative adjectives is the most reasonable, which would show, in turn, that your way of analyzing the verb â??knowâ?? is the most reasonable. Let us see how the game is played. What might we say about the following exchange?

Construction-Speak
House Builder 1 (after using a level on a concrete surface): â??Yep, this floor is flat all
right.�
House Builder 2 (without even looking at it): â??No it ainâ??t.â?
HB1: â??Sure it is. See for yourself.â?
HB2: â??Of course youâ??d think that if all you went by was the level, but if you go by a
somewhat more accurate measure of flatness, say by using a microscope, you
would see millions of bumps, which means that some parts of that floor are
flatter than other parts, which means that that floor ainâ??t flat flat.â?
HB1: â??It is flat flat! In construction, if a floor is level itâ??s flat!â?

We might begin by saying, tendentiously perhaps, that HB1 is a practical guy who at least thinks he is making a sound judgment based on the best (and we might add, most relevant) possible evidence (i.e., what the level reads). For reasons not yet apparent, HB2 is feeling quite skeptical and so challenges the â??flat assertionâ?? made by the practical house builder. What are we to make of this? More importantly, who is right? Are they both right? Are they both wrong? It is not obvious that the skeptical house builder is right and the practical house builder is wrong. It seems flat-out wrong to say that both house builders are wrong. However, in a different regard, it is not so plain to me that both house builders are not speaking the truth. The exchange is about what counts for having the property of flatness. HB1 has what we might call a â??practical viewâ?? of flat (without trying to tip the scale in his favor by using happy terms), according to which a surface has the property of flatness if it has the property of being level. HB2 has a skeptical view of flat, according to which a surface has the property of flatness if and only if it is strictly bump-free. What are the best ways for determining who in this exchange is right? The following are some ways:

Option A: We take â??flatâ?? to mean â??strictly bump-free,â?? and side with the skeptical house builder that the floor is not flat.
Option B: We say â??flatâ??-claims possess an interest-relative property, and so given the interests that are proper to--or if you like, the norm in--construction-speak, being level is sufficient for being flat, and side with the practical house builder that the floor is flat.
Option C: We take â??flatâ?? as an indexical some of whose extensions in different contexts include â??levelâ?? and â??strictly bump-free,â?? and say both house builders are speaking the truth with respect to what counts for the floorâ??s being flat (though both are oblivious to this fact).

Both options A and B have it that one and not the other house builder is speaking the truth, and so both of these options support our intuition (and the house buildersâ?? intuition) that there is a genuine disagreement between the two house builders. According to option C, there is no disagreement per se, only a misunderstanding about the semantics of â??flatâ??. Option C is one way a contextualist might look at this exchange. For the sake of time, I will not directly address the contextualist option in this paper. First, let us have a look at option A.

Before going on, let me call time-out from the game to make the following confession. It is not at all obvious to me that what goes for comparative adjectives like â??flatâ?? goes as well for the verb â??knowâ??. What does seem obvious is that some kinds of words work differently than other kinds and that we should be mindful of this fact if we are going to continue playing our game. Nevertheless, I feel that something valuable can be learned about utterances containing the verb â??knowâ?? by analyzing utterances containing other words about which there can be said conflicting things. The interesting thing that words like â??flatâ?? have in common with the word â??knowâ?? is that both are absolute, but vague terms. Indeed, there is a great deal in common between the problem of knowledge and the problem of vagueness. What is at issue with both problems is the issue of what all is included in an extension of a vague term. With that said, game on!

Certainly option A is the skepticâ??s way of looking at the above discourse, for there is no way that a concrete surface can be strictly bump-free (which explains why HB2 did not have to look at the level to know that the concrete floor was not flat). Of course this option should right away seem implausible. However, if we were to better develop the story behind HBâ??s flat-denial, it may not seem so insane. Suppose HB2 has been attending a philosophy class at night for the past few weeks, and every morning after the class he has been challenging very many of HB1 â??practical assertionsâ?? (much to the annoyance of HB1). Today he is challenging HB1â??s notion of flatness, but just last week he was challenging his view of solidity, arguing that â??since there is more space in that concrete wall--which is constantly gaining and losing atoms--it ainâ??t really solid at all!â?

And, as with this weekâ??s flat-debate, HB1 was steadfast in his assertion that the wall was indeed solid. If HB1 is aware of HB2â??s interests, why would he continue to debate, rather than simply concede something to the effect of â??Well, for my purposes, this floor (or wall) is flat (or solid), but for yours, I suppose itâ??s notâ?? Given that HB1 understands, as well as any of us can understand, the â??interestsâ?? behind HB2â??s skeptical claims, it seems that what is going on here is some issue about what is the norm of assertion.

If I make it my position in this paper to defend interest-relativity, and not only the practical house builderâ??s flat-claim but also his continuing to debate, it seems incumbent upon me to anticipate and attempt to answer an objection on the grounds that I am placing a normative constraint on these â??vague claimsâ?? when I say that some interests are better than others. Indeed, the practical house builder and I are arguing that the best way to explain why the practical house builderâ??s interests seem right in the environment is precisely because they are right. Why is it that debates with skeptics are an exception to the general rule that with nearly every complicated or controversial matter some kind of method of weighing the costs against the benefits of holding a particular view is crucial to a resolution? My position is that debates with skeptics should not be an exception to this rule. So, with this Normative Constraint on interest-relativity, we can say that of the two house builders in our exchange, since the skeptical house builderâ??s interests can be judged improper to this (and nearly every other) practical environment, and since the practical house builderâ??s interests are proper to this (and nearly every other) practical environment, the practical house builder, in his assertion that the floor (or wall) is flat (or solid), is the one who is right.

Above, I talked about what interests are proper and improper to a practical environment. Because it is so central to my view, I need to give some explication of what I take a â??practical environmentâ?? to be. Before doing this, though, I would like to share a worry (a very big worry) that I have about the merely pragmatic treatments of knowledge attributions that are so common in versions of moderate invariantism. Having done this, I believe the motivations behind the semantics of my own version of moderate invariantism will be made all the more clear.

4. A Worry About Merely Pragmatic Treatments of Knowledge Attributions

Describing his own journey through the treacherous divide between skepticism and fallibilism, Lewis (1996) writes,

If you claim that S knows that P, and yet you grant that S cannot eliminate a certain possibility in which not-P, it certainly seems as if you have granted that S does not after all know that P. To speak of fallible knowledge, of knowledge despite uneliminated possibilities of error, just sounds contradictory. (p. 419)

Rysiew (2001) argues for a pragmatic explanation of this 'sound of contradiction'. Given his pragmatic treatment of knowledge attributions, Rysiew himself could have written this passage, differing principally with respect to the words he would choose to italicize. For Rysiew, to attribute knowledge to S, and to also "grant that S cannot eliminate a certain possibility in which not-P," only seems like you have granted that S does not really have knowledge of p; fallibilist knowledge claims just sound contradictory. Their oddity should be explained pragmatically, rather than semantically. That is to say, nothing semantically encoded in a fallibilist knowledge claim is odd or, for that matter, necessarily untrue; rather, it would be odd that it would ever be appropriate in a conversation to utter an explicitly fallibilist knowledge claim like "I know that the bank will be open on Saturday even though it might not be." Essentially, I agree with Rysiew that the best way to account for this oddity is pragmatic. The worry I have about these pragmatic treatments of knowledge attributions is when they are applied across the board to the analysis of all knowledge claims. Specifically, I worry about the costs of taking our knowledge claims nonliterally.

If it is the case that we are speaking nonliterally when we make most of our claims of knowledge, does that not commit us to conceding too much to the skeptic? Is it not the case that what we are in fact saying is that we do not, literally speaking, know anything? Now it seems perfectly obvious to me that very many of our utterances containing the verb 'know' are really just exaggerations of our epistemic position toward p. For instance, if a schoolyard bully says of another bully, "He won't try to fight me because he knows what I would do to him if he did," has there really been a knowledge attribution made, or would this better be explained as the bully's attempt to convey to the hearer the extent to which he believes that the other bully will not try to fight him? We make exaggerations all the time. This, I feel, is not an important concession. But the concession that we are always (or at least most of the time) speaking nonliterally when we utter our claims of knowledge seems to be too hefty of a concession.

For instance, what does this concession do for epistemic deductive closure? If I say, for example, of a millionaire's pool, that it is a mile long, I am not committed to saying (am I?) that it was longer than two-hundred yards, for it should be obvious to the hearer that I am merely trying to convey that the pool was significantly longer than is typical for a private pool. On the other hand, if I say that I know that there are at least three people in the world who play professional baseball, I should be committed (shouldn't I?) to saying that I know that at least one person in the world plays professional baseball. In the first case, I cannot properly deduce from the obviously nonliteral statement that the pool is a mile long the obviously literal statement that the pool is longer than two-hundred yards. Here the deduction is inappropriate because we usually only make epistemic deductions from propositions, not from 'conveyed information.' In the second case, though, the simple deduction seems more than appropriate. My worry, then, is that if we are speaking nonliterally when we utter our claims of knowledge, then we should not expect these kinds of simple deductions to be made. But we do. It is likely, I feel, that the reason for the appropriateness of the deduction in the second case is that we are speaking literally when we utter most of our claims of knowledge.

Having shared this worry about merely pragmatic treatments of knowledge attributions, it should come as no surprise, then, that I feel it necessary to add a semantic component to the pragmatic story.

Consider the following exchange and accompanying story.

Bank

A: I need to deposit this check, but it's after five o'clock, so the bank's closed.
B: Well, you can deposit the check tomorrow.
A: Are you sure the bank will be open on Saturday?
B: I know the bank will be open on Saturday. I've deposited my checks on Saturday at that bank for the past few months.

So the next day, early Saturday morning, A drives to the bank to deposit his check. As he gets closer to the bank he begins to hear the sound of sirens. He then sees the bank, or what was the bank, surrounded by fireman trying to soak the burning building. Suffice it to say, the bank is not open today.

So: did B know what he claimed to know? It seems that we would not have to be completely skeptical about knowledge to answer "no" to this question. There are some things to consider, however, before we commit ourselves to an answer. Someone explaining the above exchange merely in terms of pragmatics might say either (1) that B was harmlessly exaggerating we he uttered his 'knowledge claim,' or (2) that B was conveying more (or less) than what he actually said. Either way we go, we are being skeptics about this case; for both (1) and (2) have it that B was speaking nonliterally when he made his knowledge claim. My answer to the question "Did B know what he claimed?" is yes, that he did know all that he claimed to know.

It is not so reasonable to think that B thought he had, in uttering his knowledge claim, eliminated the possibility of the bank not opening because of a fire. How could he have thought such a thing? In fact, if A had pushed him during their exchange on this matter, B might have qualified his claim in the following way.

B: Well, of course, barring any unforeseen circumstances like the bank being blown up, burned, or robbed-I only meant that I know that the bank managers usually open on Saturday and so probably will continue to do so.

From a pragmatics view, it would be hard to hold in all consistency both (1) that this qualification by B was already conveyed by his earlier utterance of "I know that the bank will be open on Saturday," and (2) that B literally knew that the bank was going to be open on Saturday. I think better is to say that the above qualification was included among his interests at the time he uttered the original knowledge claim and that his interests are semantically encoded in his proposition "The bank will be open on Saturday." After all, B was not claiming knowledge of the proposition "The bank will be open on Saturday" in all possible worlds. B's interests picked out the necessary facts about the world and all and only the relevant not-p worlds-worlds, perhaps, where the bank managers decide for whatever reason not to open the bank on this particular Saturday. The possibility of the bank being burned might not have been included among the relevant alternatives to his specific proposition. If it was not included, then B knew (literally) all that he claimed to know.

In the last section I argue that knowledge propositions (as well as many other propositions containing vague, but absolute terms) possess an interest-relative property. Essentially, I am claiming here that there is an important, though neglected, connection between the psychological and the semantic, which in itself should not be so controversial. I then return to the (MSA) to show how the version of moderate invariantism I favor handles the problem of indistinguishable worlds and accounts for our epistemic world-accessibility.

5. Practical Environment, Interests, & Accessibility to Truth-Values

5.1 Environments

In the Construction-Speak exchange we saw how differing interests and practical reasoning come to play their respective parts in determining who is right in a debate about what counts as satisfying the conditions for a vague, but absolute term. We saw that each house builder had different interests motivating his particular view about what counts as satisfying the conditions for being flat. The practical house builder, whose interests were inspired by the norm of assertion in construction-speak, claimed that the concrete floor is flat because it is level. The skeptical house builder, for his part, had interests inspired by his current philosophical training, and so claimed that the concrete floor is not flat for the reason that no concrete floor can be perfectly bump-free. In determining who was finally right in the debate, we appealed to practical reasoning and practical environments.

I have borrowed the term 'practical environment' from Hawthorne (ms). He offers a couple of ways to develop this constraint. For time considerations, I will only explain here the development I favor. According to one of Hawthorne's ways of understanding this constraint, when one knows p, there is zero epistemic chance that not-p. In that case, glosses of the practical environment idea in terms of epistemic chance will be inevitably circular and rather uninformative. One will likely in that case consider what kinds of practical reasoning we ought and oughtn't to engage in as sufficiently fundamental as not to be tractable in terms of some prior notion of epistemic probability" (ms, p. 62).

In my handling of the flat debate, I said that the practical house builder's interests were more appropriate than the skeptic's to the environment. How does this normative element of the practical environment idea correspond to my notion of interest-relativity? Certainly, if he were so inclined, the practical house builder could have changed his interests to suit the skeptic's, in which case he might have conceded something like the following: "OK, so it's not flat flat, but it is level by the tool's standards and that's good enough for me." In this case it would seem that the skeptic has the amazing power to change the interests of the would-be knower by changing the nature of what Hawthorne calls the 'deliberative environment' (ibid., p. 62). In that case it would seem that Lewis (1996) is onto something when he argues that the mere doing of epistemology destroys our knowledge. In my opinion, it seems perfectly rational to say that we are capable of losing our knowledge of p, simply on the grounds that we are capable of changing our minds about our belief that p. But I want to reject the idea that the mere introduction of skeptical alternative conceptions of something has the power to necessarily change the deliberative environment and therefore the interests of the would-be knower. The practical house builder could have changed his interests, but he did not, nor should he have, I say. The reason he was right in not changing his interests to suit the skeptic's is that he understood--in some very primitive way--that "in a deliberative environment where one ought to use p as a premise [in practical reasoning], one knows that p" (ibid., p. 62). The practical house builder understands that in construction-speak one ought to use the fact of the floor's being level in reasoning about the floor's being flat or not. So it is for this reason that there need not be much of a leap between our handling of the flat debate and our handling of the knowledge debate: since the practical house builder understands--again, in a very primitive way--that in construction-speak a floor's being level ought to be used in reasoning that it is flat, and since the floor is level and he believes this to be the case, he therefore knows that the floor is flat.

There is still more to be said, however, about how exactly, according to the view of knowledge I endorse, we would handle skeptical arguments like the (MSA). That is to say, how do the interests of a subject of knowledge and practical environment take care of the abovementioned problem of indistinguishable worlds?

5.2 How our interests pick out the necessary facts about the world

I began by explicating the skeptic's challenge that we do not have in our cognitive abilities access to the truth-value of p. It may be important to briefly clarify the distinction (and, indeed, there is one) between having access to a truth-value and knowing a truth-value. I have access to the truth-value of p just in case I have the cognitive ability to know the truth-value of p. Note that having access to the truth-value of p does not entail knowing that p, though knowing that p is sufficient for having access to the truth-value of p; having access to p only means that I am cognitively able to know that p. The skeptic herself should easily agree with this explication of accessibility, for what she is in fact claiming is that knowing that p requires more than what is constituted by our cognitive abilities. Our project, as antiskeptics, is to try to show that we do in fact have this cognitive ability to know that p. I have tried, for my part, to show how our interests coupled with good (but strictly ordinary) reasoning positively affects the truth-conditions of at least many of our 'vague' utterances. I have referred throughout to the vagueness of certain claims containing vague, but absolute terms.

A point of clarification. In Critical Thinking classes we learn that we should dismiss claims in arguments that are 'too vague.' I use 'vague' more broadly to cover any utterance containing a vague, but absolute term, such as 'flat,' 'long,' 'tall,' 'rich,' 'fat,' 'heap,' or 'know'. If a claim is vague in this sense, it certainly need not be 'too vague' to be accepted as a premise in practical reasoning.

Consider the following vague claims:
(1) That woman is rich.
(2) 42nd Street is a flat road.
(3) This is a heap of sand.
(4) He is fat.
(5) I know she likes vanilla ice cream.

The claim that should standout among these is (5). Of course we understand that not only comparative adjectives qualify as vague, but also nouns like 'heap' and 'dollop'. But is 'know' really a vague, but absolute term? Empirical linguists are wholly undecided about the meaning of knowledge in our standard usage. My idea is that the vagueness of our interests makes for the vagueness of words like 'know' and 'flat'. Consider that from (4), we could very easily generate a sorites sentence by suggesting that for any n, if weighing n grams is sufficient for his being fat, then so is n -1. Or from (1), for any n, if having n cents is sufficient for her being rich, then so is n -1. Following Graff (2000), I want to deny these sorites sentences on the grounds that the source of their generation is a misunderstanding about the vagueness of our interests in uttering our fat- and rich-claims. Of course not all vague, but absolute terms (e.g., 'flat' and 'know') generate sorites paradoxes. In any case, my intuition is that the vagueness of our purposes or desires motivating our flat- and knowledge claims best explains why these kinds of claims can generate such conflicting (and in the case of knowledge claims, paradox-like) evaluations. "Our purposes and desires are vague" is no compelling thesis, I know. I only want to put it out there so that any worry about whether this vagueness translates into ignorance may be calmed. Of purposes, Graff writes that "just as sentences have truth conditions and desires have satisfaction conditions, purposes have achievement conditions: those propositions, or states of affairs, which are such that if and only if one of them obtains will the purpose have been achieved" (2000, p. 46). The suggestion, then, is that the reason for the apparent 'boundarylessness' of our fat-claims is that our purpose for making them is vague such that there seems to be no 'cut-off point' or minimum number of grams a person would have to weigh to still be considered fat, though there really is a cut-off point or minimum number of grams. That our purposes are vague does not mean that we are ignorant; for we need not know exactly what the minimum number of grams is for us to know what counts as 'fat.' So it may be that weighing n grams and weighing n -1 are the same for our purposes. In my own view, this is where practical reasoning joins the party and adds a Similarity Constraint.

Consider Graff's 'coffee' example:

When I make my coffee in the morning and spill a few grains from the scoop, there is a cost to discriminating between the two amounts at hand. That is, [in this case] there is a cost to taking the time to count and replace the grains I've spilled. The cost outweighs the benefits, so the two amounts are the same for present purposes (ibid., p. 70).

This is an obviously good piece of reasoning. I want to suggest that the simple strength of this line of reasoning persists when we use it to address our problem of indistinguishable worlds, for the problem rests on the skeptic's assumption that the differences between the two kinds of worlds are relevant to the truth-value of our knowledge claims. I want to deny this.

Consider again (MSA):

P1. I do not know which of the two kinds of worlds I am at--P-worlds or S-worlds.
P2. At P-worlds my claim that I am standing is true and at S-worlds my claim that I am standing is false.
P3. If I do not know which of the two kinds of worlds I am at, then I do not have access to the truth-value of my claim that I am standing.
C. I do not have access to the truth-value of my claim that I am standing.

In my view, then, given any practical environment, P-worlds and S-worlds are the same for our purposes. And in any deliberative environment ('practical' or not), the cost of discriminating between the two kinds of worlds outweighs the benefits. So I reject the conclusion of (MSA) by denying P2 and P3 on the grounds that the truth-value of my claim that I am standing is determined at least in part by my interests at the time the claim is made, and so my interests may be (and, I would add, should be) such that the differences between the two kinds of worlds are wholly irrelevant to both the truth-value of my claim that I am standing and to my ability to know its truth-value.



References

Cohen, Stewart. (1992). "Contextualism and Knowledge Attributions," in Philosophy and Phenomenological Research 52: 913-929.

DeRose, Keith. (1995). "Solving the Skeptical Problem," in The Philosophical Review 104.1: 1-52.

Graff, Delia. (2000). "Shifting Sands: An Interest-Relative Theory of Vagueness," in Philosophical Topics 28.1: 45-81.

Hawthorne, John. (ms). "Knowledge and Lotteries".

Lewis, David. (1996). "Elusive Knowledge," in Australasian Journal of Philosophy 74.4: 549-567. Reprinted in Lewis, Papers in Metaphysics and Epistemology (1999), Cambridge: Cambridge University Press.

Rysiew, Patrick. (2001). "The Context-Sensitivity of Knowledge Attributions," in Noûs 35.4: 477-514.

Stanley, Jason. (ms). "On the Case for Contextualism".

Unger, Peter. (1978). Ignorance: A Case for Skepticism, New York: Oxford University Press.

van Fraassen, Bas. (1980). The Scientific Image, Oxford: Oxford University Press.

Williamson, Timothy. (1994). Vagueness, London: Routledge.