Skip to content

Editor’s Introduction:

Responding to Skepticism
Introduction to Skepticism: A Contemporary Reader (Oxford UP, 1999)

Keith DeRose
Yale University

Note: This is from the disk I sent to Oxford UP.  Since the document on the disk was edited a bit, what appears below is not exactly as the essay appears in Skepticism.  It’s pretty close, though.  In the version that appears in the book, page references to other essays in Skepticism refer to page numbers in the book, while below page references are, for the most part, to the original place of publication of the essays referred to.  Also, I below make one correction (in red) of a factual error contained in the essay as it appears in Skepticism.

1. The Argument by Skeptical Hypothesis

For almost anything you might think you know, there are powerful skeptical arguments that threaten to establish that you know no such thing. Take, for instance, your belief that you have hands. (Those who don’t have hands should change the example.) Surely there is something you not only believe, but also know! What kind of skeptical argument could possibly undermine that solid piece of knowledge?

Well, skeptical arguments come in many varieties, but some of the most powerful of them proceed by means of skeptical hypotheses. Hypotheses explain. What does a skeptical hypothesis explain? How you might be going wrong about the very things you think you know.

Consider, to take an old example, the scenario Descartes describes in First Meditation, in which he is the victim of very powerful and very deceitful “evil genius” who “has directed his entire effort to misleading” Descartes. This hypothesis could explain how Descartes has come to have any number of false beliefs. On this supposition, “the heavens, the air, the earth, colors, shapes, sounds, and all external things” are, Descartes writes, “nothing but the deceptive games of my dreams, by which [the evil genius] lays snares for my credulity.” What becomes of Descartes’s supposed knowledge of the existence of his hands? Descartes makes it clear that his evil genius hypothesis has cast this belief into doubt when, in keeping with his resolution to regard as false anything for which he finds a reason to doubt, he reacts: “I will regard myself as having no hands, no eyes, no flesh, no blood, no senses, but as nevertheless falsely believing that I possess all these things” (Descartes 1980, p. 60).

Much the same effect can be attained by means of the more up-to-date skeptical hypothesis according to which you are a bodiless brain-in-a-vat who has been electrochemically stimulated to have precisely those sensory experiences you have had, perhaps because you are appropriately hooked up to an immensely powerful computer, which, taking into account the “output” of the brain which is you, has seen to it that you receive appropriate sensory “input.”

Other old favorites include the hypothesis that one is dreaming — which Descartes considers in Meditation I before getting to the evil genius hypothesis — and Bertrand Russell’s proposal that the earth came into existence only five minutes ago, but was created complete with all the evidence of great age (including our apparent “memories”) that it actually contains. As these illustrate, some skeptical hypotheses are designed to target more limited ranges of beliefs than the evil genius or brain-in-a-vat hypotheses. Descartes seems to believe the dream hypothesis undermines a narrower range of beliefs than does his evil genius hypothesis, and Russell’s hypothesis seems to target only our supposed knowledge of the past.

By what reasoning can these skeptical hypotheses be used to undermine our supposed knowledge? The skeptic’s argument, at least in its most basic form, is as simple as it is powerful. The skeptic begins by asserting, and asking us to agree, that it is in some way an open question whether or not the situation described in her hypothesis is our actual situation. This may take the form of a premise to the effect that we can’t rule out her hypothesis, that it’s possible that her hypothesis is true, or that we don’t know that her hypothesis is false. (It is of course not necessary to her argument that the skeptic should profess to believe, or ask us to believe, that her hypothesis is true, or even that it is at all probable.) The skeptic then concludes that since we cannot rule out her hypothesis, and must admit that it may be correct, or anyway that we don’t know it isn’t, we don’t after all know the thing we originally supposed we did know.

A skeptical thesis is typically a claim that a certain range of beliefs lack a certain status. In addition, then, to varying in their scope — which specifies the range of beliefs being targeted — skeptical theses, and the arguments used to establish them, also differ in their force — which specifies precisely what lack the skeptic alleges befalls the targeted beliefs. Some skeptics may claim that the beliefs in the targeted range aren’t justified, or that they’re possibly false, or that they aren’t known with complete certainty, etc. But one of the most popular skeptical claims is that the targeted beliefs aren’t known to be true. Keeping our focus on knowledge, the above argument can be rendered as follows, where ‘O’ is a proposition one would ordinarily think one knows, and ‘H’ is a suitably chosen skeptical hypothesis:

The Argument by Skeptical Hypothesis:
1. I don’t know that not-H.
2. If I don’t know that not-H, then I don’t know that O.
So, C. I don’t know that O.

Though there are very different skeptical arguments, we will here focus on responses to this form of skeptical argument. (Many of the responses, or analogues of them, will also apply to other forms of skeptical arguments.) Before describing lines of response that are represented by the essays to follow in this anthology, we will begin by discussing one of the most popular responses this argument, and skeptical arguments generally, meet with in philosophy classes, and a very influential response given by G.E. Moore that is very similar in spirit to the popular response, and which raises important issues about how to approach the topic of skepticism.

2. “Aw, Come On!”

In The Significance of Philosophical Scepticism, Barry Stroud describes one common reaction to arguments by skeptical hypotheses as follows:

I think that when we first encounter the sceptical reasoning outlined in the previous chapter we find it immediately gripping. It appeals to something deep in our nature and seems to raise a real problem about the human condition.(1)

When arguments by skeptical hypotheses are first presented to students in philosophy classes, some do have roughly the reaction that Stroud describes. But many have a very different reaction, finding the arguments farfetched, ridiculously weak, and quite unthreatening; such a reaction is often accompanied by an exclamation somewhat along the lines of, “Aw, come on!” Those inclined to react in this latter way probably grew increasingly impatient of my repeated description, in section 1, above, of the Argument by Skeptical Hypothesis as “powerful,” thinking instead that the argument hasn’t a chance in the world of establishing its absurd conclusion.

Well, the skeptical argument really is powerful — at least fairly powerful — and is certainly not absurdly weak. The argument is clearly valid — its premises imply its conclusion — and each of its premises, considered on its own, enjoys a good deal of intuitive support. For however improbable, farfetched, or even bizarre it seems to suppose that you are a brain-in-a-vat, it also seems that you don’t know that you’re not one. How could you possibly know such a thing? And it also seems that if, for all you know, you are a brain-in-a-vat, then you don’t know that you have hands. How could you know you have hands if, for all you know, you’re bodiless, and therefore handless?

The reaction that the skeptical argument is weak is probably best refined to the claim that, however strong the argument may be, it’s not strong enough to adequately support such a counter-intuitive conclusion. And the reaction that the skeptical argument is absurdly weak is probably best refined to the claim that, however strong the argument may be, it’s nowhere near to being strong enough to support such a counter-intuitive conclusion. This would still make sense of the objector’s sense that the argument is unthreatening and doesn’t have a chance in the world of establishing its conclusion. Such an objector could be feeling that our knowledge that we have hands is, as David Lewis puts it in essay 12 of this collection, “a Moorean fact….It is one of those things that we know better than we know the premises of any philosophical argument to the contrary” (p.*549*). In looking at G.E. Moore’s own reaction to the skeptical argument, we can find strong reasons for even those in the “Aw, come on!” crowd to have ample interest in the skeptical argument. (Those who instead react in the way Stroud describes will of course be well-motivated to study the argument.)

3. Moore’s Response

In “Four Forms of Scepticism,” Moore considers a skeptical argument of Bertrand Russell’s to the conclusion that he does not know “that this is a pencil or that you are conscious.”(2) After identifying and numbering four assumptions on which Russell’s argument rests, Moore writes:

And what I can’t help asking myself is this: Is it, in fact, as certain that all four of these assumptions are true, as that I do know that this is a pencil or that you are conscious? I cannot help answering: It seems to me more certain that I do know that this is a pencil and that you are conscious, than that any single one of these four assumptions are true, let alone all four. That is to say, though, as I have said, I agree with Russell that (1), (2), and (3) are true; yet of no one even of these three do I feel as certain as that I do know for certain that this is a pencil. Nay more: I do not think it is rational to be as certain of any one of these four propositions, as of the proposition that I do know that this is a pencil. (Moore 1959, p. 226)

This reaction of Moore’s may be attractive to the “Aw, Come on!” crowd. Rather than having to identify one of the premises of the skeptical argument as positively implausible, one can, like Moore, make the more modest — and more reasonable — claim that however plausible those premises may be, they are not as certain or as plausible as is the thought that we do know the things in question, and thus those premises don’t have enough power to overturn that thought. Indeed, as we see in the above quotation, Moore agrees with Russell’s first three assumptions, so he certainly finds them plausible, though Moore makes it clear that if it came down to it, he’d reject any of those three premises before he’d accept Russell’s skeptical conclusion. And though Moore will ultimately reject it, there’s no indication that Moore finds Russell’s fourth assumption, considered by itself, to be initially implausible.

Indeed, if the premises of a valid skeptical argument are all plausible, then those who judge that the fact that we’re knowers to be a Moorean fact should find great value in the study of skeptical arguments: Such arguments show that some premise that we are tempted to accept must be false, because if it were true, we wouldn’t know what we clearly do know. Such is the fate of Russell’s fourth assumption in Moore’s hands. Skeptical arguments should in that case be seen as a rich source of information about what is, and what is not, necessary in order to know something to be the case.

Though the skeptical argument of Russell’s which Moore is countering above is quite different from our Argument by Skeptical Hypothesis, the basic type of maneuver Moore makes above is applicable to our argument; indeed, Moore himself, in his essay “Certainty,” wrestles in a similar way with a form of the Argument by Skeptical Hypothesis. The skeptical hypothesis which obsessed Moore throughout his career was the dreaming hypothesis. Moore provocatively argues:

I agree, therefore, with that part of the argument which asserts that if I don’t know that I’m not dreaming, it follows that I don’t know that I’m standing up, even if I both actually am and think that I am. But this first part of the argument is a consideration which cuts both ways. For, if it is true, it follows that it is also true that if I do know that I am standing up, it follows that I do know that I’m not dreaming. I can therefore just as well argue: since I do know that I’m standing up, it follows that I do know that I’m not dreaming; as my opponent can argue: since you don’t that you’re not dreaming, it follows that you don’t know that you’re standing up. The one argument is just as good as the other, unless my opponent can give better reasons for asserting that I don’t know that I’m not dreaming, than I can give for asserting that I do know that I’m standing up. (Moore 1959, p. 247)

Here Moore agrees with premise 2 of the Argument by Skeptical Hypothesis (at least in its dream argument version). But rather than accepting the skeptic’s conclusion that he doesn’t know that he’s standing up, Moore instead holds fast to the position that he knows that he’s standing up, and uses that, together with premise 2, to reach the conclusion that he does indeed know that he’s not dreaming. Thus, while the skeptic argues

1; 2; therefore, C,

Moore argues

2; not-C; therefore, not-1,

and claims that his argument is just as good as the skeptic’s (though one cannot help but suspect, especially given the first quotation from Moore above, that Moore considers it more rational to follow his line of argument than it is to draw the skeptic’s conclusion).

What is one to do if a powerful argument is presented toward a conclusion the negation of which one finds very plausible — plausible enough to (at least) rival the plausibility of the premises? In the quotations above, Moore motions towards two suggestions. In the first quotation, it is suggested that we make a choice according to what seems most certain to us. What we face in the situation imagined is a puzzle — a set of statements (the premises of the argument in question together with the negation of the argument’s conclusion) all of which we find plausible, but such that they can’t all be true. If we want to have consistent beliefs,(3) we’ll want to reject, or at least suspend belief in, at least one of the members of that set. Moore apparently finds Russell’s (4) to be the least certain, and so, presumably, it’s that member of the set he denies. If no further progress on the problem can be made, then perhaps making such a “Moorean” choice is the best we can do. Better to reject what seems less certain to us than what seems more certain, after all. To be sure, such a “solution” is not very satisfying. Rejecting something on the grounds that other propositions one finds plausible imply its falsity is not very fulfilling when what one rejects is itself plausible. Still, Moore rightly points out that in the situation in question there are alternatives to accepting the skeptical conclusion — alternatives which many will and perhaps should find preferable.

Can we hope for a better solution to our puzzle? In the second quotation above, Moore suggests that the parties to the dispute look for positive reasons in support of the puzzle members they support.(4) Of course, if that can be done, it would certainly help. But since each of the puzzle members will be very plausible, it may be difficult to find arguments for them whose premises are even more certain than are the puzzle members themselves.

There is an alternative, and perhaps more promising, avenue of possible progress that Moore seems to overlook: One can hope for an explanation of how we fell into the puzzling conflict of intuitions in the first place. Perhaps we can explain how premises that together imply a conclusion we find so incredible can themselves seem so plausible to us. Such an explanation can take the form of explaining, for the member of the set that one seeks to deny, why it seems to us to be true, though it’s in fact false. The game then would not be one of producing more positive support for the aspects of one’s position that are already plausible anyway, so much as it is a game of damage control — of providing a deflationary explanation for why we have the misleading intuition we do have about the plausible statement that one chooses to deny.

4. The Response from Semantic Externalism

We can distinguish between two types of “Moorean” responses to skeptical arguments. One such type of response is any reaction according to which one claims that it’s more certain or more plausible that the skeptic’s conclusion is false than it is that her premises are true, and one therefore takes the argument not to successfully support its conclusion, but to rather show that one of its premises is mistaken. This type of response can be made to any skeptical argument, whether it’s of the form we’re focusing on or not. (In fact, one can make this general kind of response to any argument for a counter-intuitive conclusion, whether or not that conclusion is a skeptical thesis.) And, if it is applied to the Argument by Skeptical Hypothesis, an executor of this type of “Moorean” maneuver needn’t follow Moore in fingering the first premise of the argument as the problem — the second premise can also be denied. But sometimes, when a response to skepticism is described as “Moorean”, what is meant is something quite different: that the responder, like Moore, denies the first premise of the Argument by Skeptical Hypothesis.

A line of response that’s “Moorean” in this latter sense and that has been very influential — largely due to the work of Hilary Putnam (see essay 2 of this collection) — is the Response from Semantic Externalism. According to semantic externalism, the contents of at least some of one’s thoughts are not completely determined by “internal” facts about what is going on inside one’s head, but are at least partially determined by such “external” facts as the nature of the items one has been in contact with. In particular, according to Putnam, you cannot think about, say, trees, if you haven’t been causally connected with trees in the proper way. Thus, a brain-in-a-vat (henceforth, a “BIV”), if it hasn’t been in contact in the proper way with real trees, cannot refer to or think about trees. When such a BIV thinks such thoughts as those she expresses via the sentences “There’s a tree”, or “Here’s a hand”, or “I’m not a BIV”, then, it is not thinking the same thing that those words would express in our mouths/minds (given that we’re not BIV’s), since the BIV is not causally connected with trees, hands, vats, etc., in the way needed to have such thoughts. What does “tree”, as used by a BIV (in “vat-English”), refer to? Putnam lists several different suggestions: “[I]t might refer to trees in the image, or to the electronic impulses that cause tree experiences, or to the features of the program that are responsible for those electronic impulses.” All of these suggestions are in the spirit of semantic externalism, because, as Putnam writes, “there is a close causal connection between the use of the word ‘tree’ in vat-English” and each of these suggested referents (p. *14*). Importantly, on any of these suggestions, the BIV ends up thinking something true when it thinks “There’s a tree”, or “Here’s a hand”, or even “I’m not a BIV”, for, to take the “in the image” reading, the BIV is indeed indicating a tree-in-the-image and a hand-in-the-image, and it indeed is not a BIV-in-the-image (it’s just a BIV).(5)

A semantic externalist, as such, needn’t commit himself to any of these positive suggestions about what “tree”, etc., refers to in vat-English, only to the negative thesis that it doesn’t refer to real trees. Indeed, Putnam not only doesn’t commit himself to any particular positive proposal, he doesn’t even commit himself to the thought that any positive proposal on which the BIVs end up thinking largely true thoughts is correct. In fact, Putnam seems unsure that the BIVs succeed in meaning anything at all; he writes that if we are BIVs, then “what we now mean by ‘we are not brains in a vat’ is that we are not brains in a vat in the image or something of the kind (if we mean anything at all).”(6) What Putnam seems committed to is just this: That if the BIVs succeed in meaning anything true or false by their relevant thoughts involving “tree”, “hand”, “vat”, etc., they mean one of the three above proposals, or some closely related other thing, on which the relevant thoughts turn out to be true.

Putnam himself seems largely uninterested in the potential such results have for the problem of skepticism.(7) But Putnam does claim that semantic externalism furnishes “an argument we can give that shows we are not brains in a vat” (p. *8*), and whether or not he is interested in using this argument for anti-skeptical purposes, understandably enough, others have thought that such an argument may be of use against the skeptic. If we can use semantic externalism to prove that we’re not BIVs, after all, it seems we can, by means of this proof, come to know that we’re not BIVs, and this would block the first premise of the Argument by Skeptical Hypothesis, at least in its BIV form.

But how might such an argument proceed? There are two quite different ways that have been proposed, both as promising anti-skeptical strategies in their own right and as interpretations of Putnam.

The main idea of the first type of argument is this. If we are BIVs, then by, “I am not a BIV,” one means that one is not a BIV-in-the-image (or some closely related true thing), which is in that case true. On the other hand, if we are not BIVs, then by “I am not a BIV,” one means that one is not a BIV, which is in that case true. Thus, whether we are BIVs or whether we are not, our use of “I am not a BIV” is true. Either way, it’s true; so, it’s true.

A problem that quickly emerged for this type of strategy is that it seems to yield only the conclusion that

(a) My utterance of “I am not a BIV” is true,

while what we seem to need, and what Putnam seemed to promise, was an argument to the quite different conclusion that

(b) I am not a BIV.(8)

Essay 4 in this collection by Graeme Forbes is set up as an attempt to take the crucial last step from what this strategy has yielded (a) to what is needed (b).

Another problem is that this first type of strategy seems to require the strong, positive externalist thesis that by her use of such sentences as “I am not a BIV”, a BIV means some true thought such as that she is not a BIV-in-the-image, and it’s unclear whether the thought experiments used to support semantic externalism support such a conclusion. Putnam himself, we saw, only claimed the weaker, negative results about what a BIV could not mean.

The second strategy, employed by both Brueckner and Warfield in their essays in this collection, seems to avoid that problem. The externalist thesis used on this second strategy is only a negative claim about what a BIV cannot mean or think — that by “tree”, “hand”, “vat”, etc., the BIV does not refer to trees, hands, vats, etc., due her lack of causal contact with such items. This negative externalist thesis is then combined with a positive claim to the effect that we do have the thoughts in question — the thoughts that the BIVs cannot have. These together imply that we are not BIVs.(9)

This strategy immediately raises the issue of whether such an anti-skeptic has any right to his second claim — the positive claim that he does have the thoughts in question. Since he does not have such thoughts if he is a BIV, he seems, in claiming to have such thoughts, to be helping himself to a question-begging assumption that he is not a BIV. Ted Warfield seeks to defend his use of this second strategy from this charge in essay 5 of this collection.

There are other potential problems/limitations that inflict both types of response from semantic externalism. Many of these are dealt with in the essays by Brueckner, Forbes, and Warfield in this collection. Each of these essays is, in its own way, quite sensitive to the potential problems of Putnam-style responses to skepticism. (Forbes, while he thinks the strategy does yield a proof that one is not a BIV, thinks that such a proof fails to provide the needed relief from skepticism. Brueckner and Warfield, while sensitive to the apparent problems, express a hope — more confident in the case of Warfield than in the case of Brueckner — that these problems are only apparent, and that the anti-skeptical strategy can succeed.(10)) Here, we’ll bring up only two such problems/limitations.

First, an often-noted potential problem, first noted* in an early paper by Brueckner,(11) that both Forbes and Warfield discuss in their essays in this collection: To solidify his externalist claim that the BIVs he was imagining were not capable of thinking about trees, hands, vats, etc., Putnam imagined a very special scenario in which the BIVs have always been BIVs. In fact, he went further and supposed that all sentient beings had always been BIVs, the universe, by accident, just happening “to consist of automatic machinery tending a vat full of brains” (p. *6). But what of other scenarios? What if I am a brain who has only very recently been envatted, after many years of normal embodiment and causal contact with real trees, hands, vats, etc.? Then, it seems, and it seems consistent with externalism, that I do mean tree by “tree”, vat by “vat”, etc., and so I am falsely thinking, “I am not a BIV.” In short, the response by semantic externalism seems to be effective only against a quite limited number of skeptical hypotheses, while, in order to pack much anti-skeptical power, it must work against them all. (Perhaps it’s precisely this problem that made Putnam himself hesitant in using his argument for anti-skeptical purposes.)

[*My above statement that this problem was first noted by Brueckner is a mistake.  In an earlier 1984 paper (“Could We Be Brains in a Vat?” Canadian Journal of Philosophy 14: pp. 115- 123), Peter Smith, for instance, not only notes the general problem that Putnam’s strategy works only against some forms of the BIV hypothesis, but also suggests the specific possibility of recent envatment as skeptical hypothesis that would avoid the Putnamian counter-attack.  I don’t know if Smith (1984) was the first place these things were done.  My understanding is that these were often noted in the explosion of literature that followed Putnam’s work.  They were ideas “in the air”.]

Second, a less often noted problem, that the authors of the essays in the collection don’t bring up. As opposed to Moore, who settles for the claim that he feels more certain that he does know that he’s standing up than that he doesn’t know that the relevant skeptical hypothesis is false, the Putnam-style response to skepticism attempts — leaving open the question of whether it succeeds — to provide a non-question-begging argument that one is not a BIV. In this way, it is a very aggressive anti-skeptical strategy. (Which explains the placement of the section on this type of response in the present collection: We have sought to order the types of response from the most aggressively anti-skeptical to the most conciliatory to skepticism.) But in another respect, this strategy seems to concede much to the skeptic. By proving that one is not a BIV, one seeks, it seems, in following this strategy, to make it the case that the first premise of the Argument by Skeptical Hypothesis is false, as applied to oneself. In this respect, this strategy is “heroic” in the way Descartes’s response to evil genius argument is: the Putnam-style arguer, like Descartes, seeks by constructing a proof against the obtaining of the relevant hypothesis, to gain knowledge, for himself and all that would follow him, that the hypothesis is false. Externalist semantics has replaced Descartes’s God as the slayer of skeptical hypotheses. Presumably, though, the proof only helps those who follow the hero — who know and understand the argument. But what of people who have never encountered this complicated argument that one is not a BIV? Do such folks fail to know that they have hands? Other strategies, which attempt not to show how to regain knowledge in the face of the skeptical argument, but rather to show how the skeptical argument never worked in the first place, by protecting the knowledge of the unphilosophical, seem in that respect at least to be more aggressively anti-skeptical.* [*I expand on this problem in my paper, “How Can We Know that We’re Not Brains in Vats?”, available on-line by clicking here.]

Forbes’s strategy, by which he argues that there are no relevant alternative situations in which one’s belief that one is not a BIV is false, seems, if it works at all, to secure the knowledge of even the nonphilosophical that they’re not BIVs, and thereby promises to block the skeptic’s first premise, even as it is applied to those who don’t know the proof. In that respect, Forbes’s use of the Putnam-style strategy is unusual; other uses of the strategy at least seem not to share that virtue with Forbes’s.

There may be ways, other than Forbes’s, in which the Putnam-type strategy could be used to show how, in the face of the skeptical argument, we all knew all along.(12) But, to initial appearances, a proof that the BIV hypothesis is false is of help against the skeptic’s first premise only to those who possess the proof, and, at the very least, those engaging in the externalist strategy haven’t done much to tell us how it is of value in securing the knowledge of those who don’t possess the proof against the onslaughts of the skeptical argument.(13)

5. Responses from Epistemic Externalism

As opposed to the semantic externalism discussed in the previous section, epistemic externalism is not a thesis about the content of one’s beliefs. Rather, it concerns the conditions under which a belief is justified or constitutes knowledge.

According to the externalist’s rival, the epistemic internalist, these matters depend primarily on factors internal to the believer’s point of view and/or factors to which the believer has special access. Most internalists admit that the external matter of whether the belief is true is relevant to the issue of whether it constitutes knowledge, so on the issue of knowledge, internalism is usually the position that only or primarily internal factors are relevant to whether true beliefs constitute knowledge. The epistemic externalist, on the other hand, claims that issues of knowledge and/or justification depend exclusively or primarily on such factors as how the belief was caused or how reliable is the faculty or mechanism by which the subject came to hold the belief — matters which are not in the requisite way “internal” to the subject’s point of view, as can be seen by the fact that you can imagine two subjects whose cognitive lives are identical with respect to what they can tell from their own point of view, but whose beliefs diverge with respect to the matters in question. The internalist about justification will have to hold that the beliefs of such subjects have the same justificatory status, and the internalist about knowledge will have to hold that, where the beliefs of such “twins” don’t diverge in their truth values, they also don’t diverge on the matter of whether they constitute knowledge.

The above characterization of the distinction between epistemic internalism and externalism should strike you as rather murky. How exactly to distinguish “internal” from “external” factors is a very difficult matter, and probably varies greatly from writer to writer.

The paradigm case of an externalist theory, though, is process reliabilism, according to which the justificatory status of a beliefs and/or the issue of whether a true belief constitutes knowledge, hinge on whether the process by which the belief was formed and/or maintained is reliable. The champion of reliabilism is Alvin Goldman, in whose hands process reliabilism has been developed through many stages and has been very ably defended.

Goldman, however, has not done much to apply his reliabilism to the problem of skepticism. And the interaction between process reliabilism or most other forms of externalism on the one hand, and the issue of skepticism on the other can be fascinating. For it’s difficult to see how most skeptical arguments could even ever gain a foothold if reliabilism were correct, since they, for the most part, seem to have no tendency to show that our beliefs are formed by an unreliable process. Consider the BIV argument. Arguably, if we were BIVs, then our belief-forming processes would not be very reliable (though the semantic externalist would contest that judgement.) But the skeptical argument doesn’t endeavor to show merely that we wouldn’t know much about the physical world if we were BIVs, but that we don’t know. But there’s nothing in that argument with any tendency to show that our belief-forming processes are in fact unreliable. The BIV skeptic would probably happily admit that, for all she knows, we are hooked up to the world in a reliable way. But she’ll still insist that we don’t know we’re not BIVs and that if we don’t know that, then we don’t know such things as that we have hands. All this without impugning the reliability of the processes by which our beliefs are formed.

One might conclude from this that externalism promises an antidote to skepticism. If we can establish epistemic externalism, then the skeptic is in trouble, for then her arguments can only work if they establish that our beliefs are unreliable, or false, or that they suffer from some other “external” malady. But most of the weapons in the skeptic’s arsenal seem ill-suited for any such task.

But those more pessimistic about externalism and/or more taken by skepticism will probably draw a different conclusion. The fact that skeptical arguments that don’t even begin to show that the processes by which our beliefs are formed are unreliable can nevertheless be so intuitively powerful (whether or not the skeptical arguments are ultimately sound), they’ll claim, shows that externalism is wrong. How could we find skeptical arguments that don’t address reliability so powerful if our concept of knowledge were that of true, reliably formed belief? The persuasiveness of skepticism can in this way be seen as constituting an objection to reliabilism. (Other forms of epistemic externalism would have similar interactions with the issue of skepticism.)

In essay 7 of this collection, Christopher Hill seeks to meet this objection on behalf of process reliabilism. Hill argues that the process reliabilist can explain the persuasiveness of the skeptical arguments, and thereby disarm this objection. But, according to Hill, while the persuasiveness of the skeptical arguments is explained, it’s not explained in such a way as to vindicate the arguments. Thus, Hill concludes that it’s skepticism that is threatened by externalism, not the other way around.

A common reaction to such externalist responses is to change the subject. Maybe we know various things, for all the skeptic can show, but we might still fall short of epistemic ideals in other ways. Perhaps the lesson of skepticism is that we don’t know that we know the things in question, or that we can’t show that we know them, or some other such thing. This move sets up an interesting dynamic. For the externalist will perhaps be able to employ his tricks on some of the other alleged shortcomings. For instance, it’s not clear that externalism can’t secure knowledge of one’s knowing as well as first-order knowledge. But the externalist may concede some of the alleged shortcomings — particularly the ones that involve epistemic concepts that are clearly internalist in character — but argue that it’s not so interesting or important that we fall short in the ways in question. The search, for the skeptic, becomes one to find a conclusion that is both interesting and established by the skeptical arguments, slightly revised to make them suitable for establishing their new targets. The anti-skeptic responds to each proposal either by showing that it’s no great shortcoming that we fail in the way alleged, or that the skeptical argument cannot establish that particular shortcoming.

Ernest Sosa, in essay 6 of this collection, pursues the anti-skeptical side of this dialectic. Sosa argues that externalism, and in particular reliabilism, can provide an escape from philosophical skepticism, but by philosophical skepticism he does not just mean the claim that we don’t know this or that, but means rather the following thesis: “There is no way to attain full philosophical understanding of our knowledge. A fully general theory of knowledge is impossible” (p. *263*). Sosa argues that there’s no reason to suppose that an externalist theory of knowledge — in particular, a reliabilist theory — would fall short of such full generality. Sosa does admit that externalism will fall short of meeting certain requirements some might want a theory of knowledge to meet, but argues that it’s misguided to want a theory of knowledge to meet such requirements. In particular, Sosa admits that a fully general legitimating account of our knowledge is indeed impossible, where a legitimating account “specifies the sorts of inferences that justify one’s beliefs… without circularity or endless regress” (p. *267*). But when we’re seeking a fully general account — an account of all our knowledge — we of course are not going be able to make it at the same time a legitimating account. We can provide legitimating account of limited stretches of our presumed knowledge, because we can appeal to knowledge from outside of that limited stretch to construct our account. But if the account is to be fully general, it will have to draw its starting point from beliefs that are among those in question, and the account will then suffer from circularity. If reliabilism can provide a fully general account, then it’s giving us everything we can reasonably desire. To ask for an account that is at the same time legitimating is to ask for the obviously impossible.

It should be noted that externalism is quite popular these days, and several of the papers in this collection other than Hill’s and Sosa’s are externalist in character. We’ve collected Hill’s and Sosa’s paper together under the heading of “Epistemic Externalist Responses” because each, while defending a particular version of externalism with respect to its handling of skepticism, does so in such a way as to bring to the fore issues of general importance to the relation of externalism to skepticism.

6. Relevant Alternatives and Denying Closure

According to what is known as the “Closure Principle” for knowledge, if you know some proposition P, and you know that P entails some second proposition Q, then you also know that Q.(14)

This principle looks like it describes how knowledge can be expanded by means of inference: If you know something you can come to know anything it entails by coming to know the entailment. But, in the Argument by Skeptical Hypothesis, the skeptic uses the principle to attack the thesis that we know. For the skeptic seeks to argue that since you don’t know that her hypothesis, H, is false, and since, given closure, you would be able to know that the hypothesis is false if you knew O (the proposition you would ordinarily think you know (e.g., that you have hands)), you must not know that O. Thus, the Closure Principle has come to be seen as underwriting skepticism. In essay 8 of this collection, Fred Dretske goes so far as to write that “Almost all skeptical objections trade on it” (p. *1011*). In particular, this Principle supports premise 2 of our Argument by Skeptical Hypothesis.

As we saw in section 3, above, the Argument by Skeptical Hypothesis can be profitably viewed as presenting us with a puzzle consisting of three individually plausible but jointly inconsistent theses: the first premise of the argument (I don’t know that not-H); the second premise (If I don’t know that not-H, then I don’t know that O), and the negation of the argument’s conclusion (I do know that O). The skeptic, in accepting the argument’s skeptical conclusion, of course, rejects that conclusion’s negation. The Moorean rejects the first premise of the argument. It was only a matter of time before the puzzle was dealt with by denying the second premise. That time was the 1970’s and 1980’s, and the most prominent of the deniers of closure were Fred Dretske and Robert Nozick. Of course, it’s easy enough to simply deny the second premise and the Closure Principle that underwrites it; what’s more difficult is to give a plausible rationale for taking that approach to the puzzle. Both Dretske and Nozick sought to provide an account of what knowledge is that backed that maneuver.

Nozick, following earlier work by Dretske,(15) advances an account of knowledge that rests heavily on subjunctive conditionals. On this account, what is needed for S to know that P, in addition to the usual requirements that (1) P is true and that (2) S believes that P, is that both of the following subjunctive conditionals hold:

(3) If P weren’t true, S wouldn’t believe that P

(4) If P were true, S would believe that P

The star of Nozick’s show is condition (3),(16) which plays a pivotal role in Nozick’s application of his account of knowledge to the Argument by Skeptical Hypothesis. For your belief that you’re not a BIV fails to meet that condition: If that belief weren’t true (if you were a BIV), you would still believe it was true (you would believe you weren’t a BIV). For, remember, the BIV has had all the sensory experiences you’ve had, and would thus believe everything you believe, including, presumably, that he’s not a BIV. But, Nozick argues, the likes of your belief that you have hands does satisfy this condition for knowledge, as well as the other conditions Nozick posits. Thus, if Nozick’s account of knowledge is correct, you do know you have hands, but you don’t know that you’re not a BIV. The skeptic’s first premise is true, but her second premise, and the Closure Principle that underwrites it, is false.

As I’ve already indicated, Dretske had also proposed that subjunctive conditionals be used in the analysis of knowledge, but Dretske’s “Epistemic Operators” (essay 8 of this collection), especially towards its end, provides an early statement of another approach to knowledge which became quite popular in subsequent years — the “Relevant Alternatives” theory of knowledge. On this theory, the main ingredient that must be added to true belief to make knowledge is that one be in a position to rule out all the relevant alternatives to what one believes. The important implication here is that some alternatives to what one believes are not relevant, and so one can know in the face of some uneliminated possibilities of error. Thus, in one of his examples (pp. *1015-1016), you can know that the animals in a cage are zebras, according to Dretske, without knowing that they are not cleverly painted mules, because the alternative that the animals are merely cleverly painted mules is, in any normal context, an irrelevant alternative to their being zebras. Thus, you can know that the animals are zebras without knowing that they’re not painted mules. Thus, closure does not hold in general. You can know that P without knowing everything that you know that P entails, for P will entail the falsity of all the contraries or alternatives to P, but you need only know the falsity of the relevant alternatives to P in order to know that P. Dretske attempts to support this thesis by means of analogies with “operators” other than “S knows that…” (see especially pp. *1019-1023). Thus, for example, that Brenda didn’t order desert entails that she didn’t order desert and throw it at the waiter; still, some proposition R (e.g., Brenda was on a diet) might explain why Brenda didn’t order desert, while failing to explain why Brenda didn’t order desert and throw it at the waiter. Thus, Dretske concludes, “explains that…” is importantly like “knows that…”: Just as you can know that P without knowing everything that you know P entails, so can a proposition explain another proposition without explaining everything that that second proposition is known to entail. The skeptic, Dretske concludes, is right to claim that we don’t know her skeptical hypothesis to be false, but, since her hypothesis is not a relevant alternative to O, she is wrong to think that we therefore fail to know O.

The primary problem for the strategies of both Nozick and Dretske involves the powerful intuitions most of us have supporting closure.(17) Many, in fact, consider the anti-closure implications of Dretske’s and Nozick’s theories to be reductios of those theories. To their credit, both Dretske and Nozick admit the intuitive power of closure. Dretske admits that producing examples in which closure seems violated will not suffice to support his thesis of non-closure because “the thesis itself is sufficiently counterintuitive to render controversial most of the crucial examples” (p. *1017*). And Nozick goes so far as to compare closure, with respect to its “intuitive appeal,” to a steamroller (p. *206*). Whether either philosopher’s theory of knowledge and/or Dretske’s analogies with other “operators” are sufficiently convincing to warrant accepting the counterintuitive results of their views is a matter we must leave to each reader to decide.

Gail Stine, in essay 9 of this collection, which is largely a response to Dretske, offers a Relevant Alternatives approach to skepticism which does not involve denying closure. Stine agrees that, at least typically, we’re correctly described as knowing various Os despite our inability to rule out various skeptical hypotheses that are alternatives to those Os. And she agrees that this is because those skeptical hypotheses typically fail to be relevant alternatives to the Os we ordinarily think we know. How, then, does Stine avoid denying closure on this Relevant Alternatives approach? By claiming that what the range of relevant alternatives is varies with conversational context, and that this amounts to a change in the meaning of knowledge-attributing sentences. She then claims that in testing closure, we should hold constant the range of relevant alternatives, and thereby avoid changing the meaning of the relevant sentences and committing “some logical sin akin to equivocation” (p. 256*). And, according to Stine, closure does not fail so long as the range of relative alternatives is held constant. In extraordinary contexts, even the hypotheses of philosophical skeptics may be relevant.(18) Stine labels the standards that allow these hypotheses as relevant “extreme” (p. 254*), “very peculiar” (p. 257*), and even “very perverse” (p. 254*), but, apparently, they’re allowable. In those contexts, you don’t count as knowing that the skeptical hypotheses are false, but you also don’t count as knowing such things as that you have hands. In more ordinary, and presumably less “perverse” contexts, you do count as knowing you have hands, but you also count as knowing that the skeptical hypotheses are false. Since in such an ordinary context the skeptical hypotheses are irrelevant, Stine concludes that you “simply know” (p. 259*) that they’re false, with no need for any evidence to that effect: “If the negation of a proposition in not a relevant alternative, then I know it — obviously, without needing to provide evidence — and so obviously that it is odd, misleading even, to give utterance to my knowledge” (p. 258*). Part of the explanation for the oddity of saying you know, say, you’re not a BIV is that there’s a presupposition that not-P is a relevant alternative when it’s knowledge of P that’s in question (p. 255*). But Stine claims that this presupposition can be canceled, and one can truthfully (though perhaps oddly) claim to know such a thing as that one is not a BIV.

7. Contextualist Responses

When students are first presented with skeptical arguments like the Argument from Skeptical Hypothesis, it’s quite common for some of them to react by positing two senses of “know” and claiming that the argument shows only that in some “high” or “philosophical” sense we don’t know we have hands, while, for all the argument shows, we retain knowledge in the “low” or “ordinary” sense of “know”. The skeptic’s argument has induced us to switch over to the “high” sense of “know”, which is why the argument is so persuasive. But the truth of our ordinary claims to know is not threatened by the skeptic’s attack. Some philosophers have developed this type of idea in their treatments of skepticism; see, for example, Norman Malcolm’s distinction between “strong” and “weak” knowledge in his 1952, and Stroud’s distinction between “internal” and “external” varieties of knowledge in his 1984.(19) Such “Two Senses of ‘Knows'” approaches to skepticism are limiting cases of a more general type of approach which have come to be called the “contextualist” approach to knowledge and skepticism.

A problem with such “Two Senses of ‘Know'” theories is that, to avoid making their approach to the problem of skepticism ad hoc, we’d like to find, in ordinary, non-philosophical uses of knowledge-attributing sentences, support for the view that different standards for knowledge govern different contexts. The problem here is not that we find no such variation, but rather that we find more such variation than can be handled by such “Two Senses” theories. A wide variety of different standards for knowledge seem actually to be used in different contexts.

Current contextualists posit such a wide variety of different standards; they look for rules by which what is said in a conversation can change the standards that are in place; and they typically try to (at least partially) explain the intuitive pull of skeptical arguments by claiming that the skeptic, in presenting her argument, exploits one of these rules, raising the standards for knowledge, and thereby making her conclusion that we “don’t know” true. If this is how the skeptic’s argument works, then the truth values of our ordinary claims to know are protected, for the fact that the skeptic can install very high standards which we don’t meet has no tendency to show that we don’t know according to the lower standards that govern our ordinary, non-philosophical conversations.

Such contextualist strategies also seem to provide a tighter tie between the skeptic’s and the ordinary use of ‘know’, than do “Two Senses of ‘Know'” theories, which may help them to explain why skeptical arguments can seem to threaten our knowledge ordinarily so-called. For if these current theories are correct, the standards for knowledge are variable even in ordinary, non-philosophical settings. It’s not as if we had always used a single set of standards and the skeptic is now introducing some new and different standards in an unprecedented way. Rather, the skeptic is making use of a feature of “know” that shows itself in non-philosophical contexts, and is employing rules we’re already accustomed to.

Both of our contextualist authors, Keith DeRose and David Lewis, follow Stine (see section 6, above) in upholding closure, relative to any set standard for knowledge. Both follow Stine in seeing the skeptic as employing higher standards than are usual, though neither follows Stine in describing the skeptic’s standards as “perverse”.(20) And both seek to view the skeptic as employing a rule for the raising of standards which is common in non-philosophical discussion.

Lewis, maintaining closer ties with the relevant alternatives tradition than does DeRose, gives the following account of knowledge:

S knows proposition P iff S’s evidence eliminates every possibility in which not-P — Psst! — except for those possibilities that we are properly ignoring.(21)

Here, the “possibilities that we are properly ignoring” amount to what, on the relevant alternatives approach, are the irrelevant alternatives to P. On Lewis’s account, the relevant alternatives — the alternatives which our evidence must eliminate if we’re to be knowers — are the possibilities in which not-P that we’re either paying attention to or else are ignoring, but improperly so. Much of Lewis’s paper consists in spelling out the rules that govern what possibilities are properly ignored. The skeptic, on Lewis’s account, exploits the “Rule of Attention”, according to which those alternatives to which we’re paying attention are relevant. By calling attention to various possibilities which we typically (and properly) ignore, the skeptic makes those alternatives relevant. Thus, if our evidence fails to eliminate them — and the skeptic’s hypotheses are always carefully chosen to be such that our evidence doesn’t eliminate them — the skeptic succeeds in creating a context in which we don’t count as knowers. However, that doesn’t mean we speak falsely when we claim to know those very same O’s in contexts in which no skeptic’s are calling our attention to those hypotheses. Even if our evidence doesn’t eliminate those possibilities, we can still be correctly attributed with knowledge, on Lewis’s account, if those who are calling us knowers are properly ignoring those uneliminated possibilities.

DeRose, seeking to hold on to what he finds correct in Nozick’s subjunctive conditionals account while avoiding its pitfalls, claims that the conversational rule that the skeptic exploits in the Argument by Skeptical Hypothesis is the “Rule of Sensitivity”. Where a belief that P is called “sensitive” if it meets Nozick’s third condition for knowledge — where it’s true that you would not have believed that P if P had been false — this rule states:

When it asserted that some subject S knows (or does not know) some proposition P, the standards for knowledge tend to be raised, if need be, to such a level as to require S’s belief in that particular P to be sensitive for it to count as knowledge. (p. *36*)

Since, as DeRose argues, you must be in a very strong epistemic position before any belief you might have that a skeptical hypothesis is false can be sensitive, the skeptic’s assertion of her first premise (“I don’t know that not-H”) will, by the Rule of Sensitivity, raise the standards for knowledge to a level at which you count as knowing neither that not-H, nor that O. The Rule of Sensitivity, DeRose argues, explains why we won’t call someone a knower if we think their belief is not sensitive, and thus explains why Nozick’s account of knowledge can seem so attractive. But, DeRose argues, his contextualist account accomplishes this without licensing the counterintuitive violations of the Closure Principle that Nozick’s non-contextualist account is committed to.

A major problem for contextualist solutions, at least in the eyes of many, is that they concede much to the skeptic, since these strategies allow that we don’t know, according to the high standards that are put in place by the presentation of the skeptical argument, the O’s we usually take ourselves to know. How much of a concession is this? Much will depend on how important one thinks it is to have such “high” knowledge. At one extreme, if one, like Stine, thinks of the skeptic’s standards as “perverse”, one probably won’t mind our failing to meet them. On the other extreme, it will seem to some that it’s the issue of whether we have the “high” knowledge put into play by the skeptic that’s been the important issue all along. To these folks, the contextualist is conceding everything of value to the skeptic, and the fact that the contextualist protects our “low” (“vulgar”?) knowledge is of little importance. Middle positions, according to which there is interest both in what the skeptic is granted and in what she is denied by the contextualist approach are of course possible, and will seem to many to be quite sensible.

The above potential problem involves how significant the contextualist response would be if it were successful. Another problem for these approaches, which will be developed in section 8 below, threatens to show that they won’t be successful in the first place. It involves the presence of a rival theory, which threatens to handle all the phenomena contextualism seeks to explain, while holding the standards for knowledge constant in all contexts.(22)
8. Concessive Responses

In important work on skepticism in the early and mid 1970’s, which culminated in his book, Ignorance: A Case for Scepticism,(23) Peter Unger argued that, in order to really know something, one must be in a very strong epistemic position with respect to that proposition — so strong, in fact, that it would be impossible for anyone ever to be better positioned with respect to any matter than you are now with respect to the matter in question. Unger admitted that varying standards for knowledge govern our use of sentences of the form, “S knows that P”, but did not endorse contextualism, because Unger claimed that these varying standards were standards for whether it was appropriate to say that S knows; the truth conditions for the sentence — the conditions under which what the sentence expresses would be true — were, according to Unger, constant, and very high. Thus, the skeptic is right when she says we don’t know, and we are saying something false (though perhaps appropriate) when, even in ordinary, non-philosophical discussions, we claim to know this or that. This is the rival to contextualism mentioned in the above paragraph. The “rival” came first, however: It was largely in response to this “invariantist” theory of Unger’s that the early contextualist views of the late 1970’s and early 1980’s — like that expressed by David Lewis in a short section of his 1979 and in contextualist versions of the Relevant Alternatives approach(24) — were developed.

Unger’s 1984 book Philosophical Relativity, the guts of which constitutes essay 13 of the present collection, contained what was at that time — and for some time to come, for that matter — easily the most complete exposition of the contextualist view. But while this book represented a change of mind for Unger from his skeptical writings of his Ignorance period, he was not advocating contextualism in Philosophical Relativity. Instead, he defended the “relativist” conclusion that contextualism and his earlier invariantist views which led to skepticism were equally good theories, and that there simply is no fact of the matter as to which view is correct. Unger’s relativism, defended, as it is, by parity considerations, according to which the advantages and disadvantages of contextualism and invariantism balance each other out in such a way that there is no winner, is a precarious view to defend: Any contextualist who succeeds in defeating invariantism will conquer Unger’s relativism as an automatic corollary, and the same will happen for any invariantist who produces a successful argument against contextualism. But here Unger laid out very carefully the rival to contextualism, together with an argument that it was, while not superior to contextualism, at least an equal of it. In sections 15 and especially 16 of essay 11, below, DeRose attempts to show why, in the face of Unger-like considerations, contextualism is superior to invariantism.

Contextualism, as we saw in the previous section, can already be viewed as quite concessive to the skeptic. Unger’s relativism, which straddles contextualism and out-and-out skepticism, is even more so. Concessive enough to make the “Concessive Responses” section of this book.

Though the general drift of recent discussions of skepticism has been fairly anti-skeptical, there has also been an undercurrent of writing, other than Unger’s, more friendly to the skeptic, a couple of important examples of which we’ve included in this collection.

In essay 14, Thomas Nagel argues that skeptical doubt is the natural result of a realist picture of the world and our place in it — a picture according to which “there is a real world in which we are contained, and…appearances result from our interaction with the rest of it” (p. *68*). This picture naturally leads us to wonder which aspect of the appearances reflect the way the world really is, and which are misleading results of our interaction with the world. As we pursue greater objectivity — a view which relies “less and less on certain individual aspects of our point of view” (p. *67*) — we will not be able to leave skeptical doubts behind. For as we develop a more objective account of our place in the world, an account which explains why the world appears to us as it does, this new picture will also be the result of our interaction with the world, though a more complicated interaction. Nagel argues: “If the initial appearances cannot be relied upon because they depend on our constitution in ways that we do not fully understand, this more complex idea should be open to the same doubts, for whatever we use to understand certain interactions between ourselves and the world is not itself the object of that understanding. However often we may try to step outside of ourselves, something will have to stay behind the lens, something in us will determine the resulting picture, and this will give grounds for doubt that we are really getting any closer to reality” (p. *68*).

Here Nagel raises some of the problems that arise from attempting to develop what Sosa, in essay 6, called “a fully general philosophical understanding of our knowledge.” The question naturally arises: Which epistemic achievements are blocked by our inability to get completely outside of ourselves in the way Nagel outlines? Does that prevent us from having knowledge of what the world is like? Some of Nagel’s remarks appear to indicate that knowledge is thereby undermined. For instance, he writes: “Skepticism is really a way of recognizing our situation, though it will not prevent us from continuing to pursue something like knowledge” (p. *74*, emphasis added), hinting that knowledge itself is not in the cards. But Nagel also writes that “definitions of knowledge cannot help us” with our problem, for, “The central problem of epistemology is the first-person problem of what to believe and how to justify one’s beliefs — not the impersonal problem of whether, given my beliefs together with some assumptions about their relation to what is actually the case, I can be said to have knowledge” (p. *69*). Here Nagel indicates that knowledge isn’t the real issue. But what if knowledge itself, correctly defined, and not just “something like” knowledge, could be attained without our getting fully outside of ourselves in the way Nagel says we cannot? What if even loftier epistemic states, like, say, knowing for certain, properly defined, could survive the lack Nagel alleges? After all, it’s in no way obvious that getting fully outside of oneself in the way Nagel claims we cannot should be construed as a necessary condition for knowing for certain that one has (real, objective) hands.

We’d then be left with the question of how significant in its own right is the result that we cannot get fully outside of ourselves if that result doesn’t lead to any of the more familiar shortcomings that we might have feared are shown by skeptical arguments. Some (perhaps Sosa?) would urge us to see the attempt to meet such an impossible demand as misguided: “Of course we can’t attain a fully objective view of ourselves if that’s what you mean by a fully objective view. Who would have ever thought we could have?”

But some otherwise inclined will urge that, though it does follow fairly automatically from a realist picture, the lack Nagel points to is an important result, for it is the inevitable outcome of pursuit of objectivity, a pursuit deeply embedded within us. Even if it is somehow inevitable that this will lead to the disappointment, doesn’t that only intensify the worry? That the realist picture and the resulting search for greater objectivity lead inevitably to our inability to complete the task we set out on is a deep problem, however inevitable it was that this would be the result.

One may well wonder whether, not too far down this dialectical path, there will be much of a real disagreement between skeptics and anti-skeptics. Perhaps there will be agreement about which epistemic states are undermined by skepticism and which aren’t, the only differences being over the significance of the lacks established by skepticism.

In essay 15, Barry Stroud seems to reach such a point in his debate with Sosa, whose essay 6 is Stroud’s primary target. Stroud writes:

Here, perhaps, we approach something that Sosa and I can agree about. What I have tried to identify as a dissatisfaction that the epistemological project will always leave us with is for him something that simply has to be accepted if we are going to have a fully general theory of knowledge at all. He appears to think, as I do, that it is endemic to the epistemological project itself. We differ in what moral we draw from that thought. (p. *306*)

The lack in question is roughly that, in any fully general account of our knowledge, we’ll have to rely on a starting point which we simply must accept to be true without a non-circular reason for thinking it true. In Sosa’s terminology (see section 5, above), no fully general account of our knowledge could succeed in also being a legitimating account. But though Stroud does draw a different moral from this than does Sosa, he doesn’t think the lack necessarily leads to skepticism. At the close of his essay, he indicates that skepticism will be the winner if we pursue epistemological theories in the traditional way. The moral Stroud draws is that if skepticism is to be successfully resisted, “the resistance has to start farther back” (pp. *306-307*) — with a re-examination of the traditional epistemological project itself.

Sosa, however, could claim to have already proposed the needed revision to that project. The problem with the traditional project is that it seeks a fully general, legitimating account of our knowledge. Sosa’s advise would be: You can seek a fully general account or you can try for a legitimating account. You can in fact pursue both types of account separately. What you can’t have, and shouldn’t seek, is an account of our knowledge that is, at the same time, both fully general and legitimating. Stroud, apparently, would reject this as the revision he was seeking, but it’s difficult to say exactly what’s wrong with it as such a revision.

But we shouldn’t assume that we are destined to reach a point where there’s agreement about what skeptical results can be established and disagreement only over the significance of the skepticism that holds. Skeptics and anti-skeptics should strive to make precise exactly what are the lacks shown by skeptical arguments, and to investigate exactly what consequences such lacks may have for familiar issues of knowledge, certain knowledge, justification, etc. We’ll then be better positioned to evaluate the shortcomings that philosophical skepticism really establishes.(25)


Bouwsma, O.K. 1949. “Descartes’ Evil Genius,” Philosophical Review 58: pp. 141-151.

Brueckner, Anthony L. 1985. “Skepticism and Epistemic Closure,” Philosophical Topics 13: pp. 89-117.

_____. 1986. “Brains in a Vat,” Journal of Philosophy 83: pp. 148-167.

_____. 1996. “Trying To Get Outside Your Own Skin,” Philosophical Topics 23: pp. 79-111.

DeRose, Keith. 1992. “Contextualism and Knowledge Attributions,” Philosophy and Phenomenological Research 52: pp. 913-929.

Descartes, Rene. 1980. Donald A. Cress, tr., Discourse on Method and Meditations on First Philosophy. Indianapolis: Hackett Publishing.

Dretske, Fred. 1971. “Conclusive Reasons,” Australasian Journal of Philosophy 49: pp. 1-22

Foley, Richard. 1993. Working Without a Net: A Study of Egocentric Epistemology. New York: Oxford University Press.

Forbes, Graeme. 1984. “Nozick on Scepticism,” Philosophical Quarterly 34: pp. 43-52.

Klein, Peter. 1985. “The Virtue of Inconsistency,” Monist 68: pp. 105-135.

Lewis, David. 1979. “Scorekeeping in a Language Game,” Journal of Philosophical Logic 8: pp. 339-359.

Malcolm, Norman. 1952. “Knowledge and Belief,” Mind 51: pp. 178-189.

Moore, G.E. 1959. Philosophical Papers. London: George Allen & Unwin LTD.

Schiffer, Stephen. 1996. “Contextualist Solutions to Scepticism,” Proceedings of the Aristotelian Society 96: pp. 317-333.

Stroud, Barry. 1984. The Significance of Philosophical Scepticism. Oxford: Clarendon Press.

Unger, Peter. 1975. Ignorance: A Case for Scepticism. Oxford: Oxford University Press.


1. Stroud 1984, p. 39. The “sceptical reasoning” to which Stroud refers, and which was presented in his first chapter, is Stroud’s rendition of Descartes’s dream argument.

2. As the quotations in the text are about to show, Moore vacillates freely between knowing and knowing for certain, sometimes presenting the skeptical arguments as attempts to reach the conclusion that we don’t know the things in question for certain, and sometimes as urging the conclusion that we don’t know them. This vacillation is explained by Moore’s belief that knowing and knowing for certain are the same thing — that there is no knowing that isn’t knowing for certain. Throughout, we treat Moore as addressing the issue of knowledge, but the reader should be aware that Moore was also writing about certain knowledge, which he thought was the same thing.

3. It’s not to be automatically assumed that we should maintain consistent beliefs in such a situation. Especially where the members of the set all seem to have about the same, high degree of plausibility, the option of continuing to believe all of them, while, of course, realizing that they can’t all be true, and so tempering the degree of one’s belief in each, seems to some an attractive possibility. It’s where one member of the set is, while somewhat plausible, not nearly as plausible as the other members that a rejection of a belief seems the best choice. For arguments that one can rationally hold beliefs one knows to be inconsistent, see Klein 1985 and Chapter 4 of Foley 1993.

4. Though Moore writes, “The one argument is just as good as the other, unless my opponent can give better reasons for asserting that I don’t know that I’m not dreaming, than I can give for asserting that I do know that I’m standing up,” indicating that both he and the skeptic will be presenting such a positive case, in what follows in “Certainty,” he merely looks at what case the skeptic might give, arguing that the skeptic’s argument won’t be convincing, and never gives any reasons for his position that he does know that he’s standing up.

5. This aspect of these responses to skepticism was anticipated in Bouwsma 1949. Bouwsma argues that a victim of Descartes’ Evil Genius would not be fooled into holding false beliefs, but would in fact be thinking thoughts that were largely true.

6. P. 15, emphasis added. Along the same lines, Putnam writes, “So, if we are brains in a vat, then the sentence ‘We are brains in a vat’ says something false (if it says anything)” (p. 15, emphasis added).

7. Of the BIV scenario, Putnam writes early in his essay: “When this sort of possibility is mentioned in a lecture on the Theory of Knowledge, the purpose, of course, is to raise the classical problem of scepticism with respect to the external world in a modern way. (How do you know you aren’t in this predicament?) But this predicament is also a useful device for raising issues about the mind/world relationship” (p. 6*). And in what follows, Putnam’s own interest seems confined to the “mind/world relationship”, for the “classical problem of scepticism” is hardly ever again mentioned.

8. See especially Brueckner 1986.

9. See footnote 11 of Essay 3 in this collection, where Brueckner relates some of the advantages of this second strategy.

10. In his more recent 1996 (see especially section 9), Brueckner reverses his judgment, deciding the strategy won’t work after all.

11. See again Brueckner 1986. In Essay 3 of this collection, Brueckner rehearses this objection to the first type of strategy in footnote 10.

12. For instance, as Brueckner construes the skeptical argument, it depends on a premise that the BIV hypothesis is logically possible (see p. *200* of this volume). And at places Robert Nozick seems to construe the skeptic as relying upon the claim that her hypotheses are logically possible or coherent (see pp. *167-169*). The externalist strategy could then be seen as undermining that premise. But that premise at least seems unnecessary. As I’ve been formulating the argument here, there is no such premise, and Nozick sometimes ignores this premise, construing the skeptic as proceeding merely from her true (according to Nozick) insight that we don’t know the hypothesis to be false, together with her “short step” (short, but mistaken!) from that insight to her conclusion that we don’t know such things as that we have hands. But maybe, for reasons I can’t see, that premise really is needed by the skeptic. Or maybe there is some other way the Putnam-style approach can derail the skeptical argument even as it applies to the non-philosophical.

13. For further critique of this type of response to skepticism, see section 2 (pp. *71*-*74*) of Thomas Nagel’s essay 14 of this collection.

14. The Principle, so formulated, is not exactly correct: One could know that P, and know that P entails Q, without ever having put these two pieces of knowledge together in order to infer that Q. In such a case, you might not even believe that Q, much less know it. The details may be difficult to get exactly right, but the following formulation of the principle may at least come close: If you know that P, and you know that P entails Q, and you believe that Q based on an inference from your beliefs that P and that P entails Q, then you know that Q. Such a fancier principle seems equally useful to the skeptic, who can argue that if you really knew that O, you could, given that principle, come to know that not-H, provided that you knew that O entails that not-H. Given that more complicated version of the second premise of the Argument by Skeptical Hypothesis, the first premise would also have to be doctored up a bit, but not, it seems, in such a way as to diminish its plausibility. The skeptic can claim that you don’t know that not-H, and can’t come to know it by means of an inference from your “knowledge” that O. Thus, you must not really know that O, after all. Unfortunately, I suspect even the fancier version of the Principle is not correct. Just as I think one can just barely know that P and just barely know that Q, yet just barely fall short of knowing the conjunction of P and Q — even when holding constant the standards for knowledge — because of the accumulation of doubt, so also can one know that P and know that P entails Q, while falling short of knowing that Q, even if one’s belief that Q is based on one’s knowledge of P and knowledge of the fact that P entails Q.

15. Nozick was apparently unaware of Dretske’s much earlier work until after writing a draft of the relevant portion of his book; see note *53* of Nozick’s essay in this collection. For an example of the precursors of Nozick’s treatment, see especially Dretske 1971, where Dretske advances an account of knowledge in which subjunctive conditionals loom large. See also note 4 of Dretske’s essay in this collection, where Dretske explains how any account of knowledge in subjunctive conditionals play such a role will result in failures of closure.

16. Condition (4) is problematic. Many — perhaps most — students of subjunctive conditionals believe that where the antecedent and the consequent of such a conditional are both true, then so is the conditional. Thus, where Nozick’s first two conditions for knowledge hold, condition (4) will always be met. Thus, (4) can do no work. Anyone who meets conditions (1) and (2) will automatically meet (4) as well. Others, noticing how odd is to use the subjunctive (“If P were true”) in (4) if P is true, may conclude that if P is true, then (4) cannot be met. This would make knowledge impossible, since in that case nobody could ever meet both (1) and (4). What Nozick needs is an account of “true-true subjunctives” — subjunctive conditionals whose antecedent and consequent are both true — which makes them neither always true nor always false. Such an account is controversial at best. Alternatively, Nozick could perhaps give up on formulating his fourth condition in terms of subjunctive conditionals, and instead put it directly in terms of possible worlds, as follows: “In all of the possible worlds very close to the actual world in which P is true, S believes that P.” This could give rise to problems for Nozick, however. First, how close is close enough to count as “very” close? Our understanding of subjunctive conditionals cannot in any obvious way answer this question if we’ve given up on expressing the fourth condition in terms of subjunctive conditionals. Second, such a strategy would involve Nozick in a more serious use of the apparatus of possible worlds than he seems to want (see note *8* of Nozick’s essay 10 in this collection).

17. For more on this, especially as it applies to Nozick, see section 9 (pp. **-**) of DeRose’s essay 11 of this collection and Forbes 1984. For a good general discussion of the anti-skeptical approach of denying closure, see Brueckner 1985.

18. Stine writes, “But the skeptic has an entering wedge, and rightly so. It is an essential characteristic of our concept of knowledge that tighter criteria are appropriate in different contexts. It is one thing in a street encounter, another in a classroom, another in a court of law — and who is to say it cannot be another in a philosophical discussion? And this is directly mirrored by the fact we have different standards for judging that there is some reason to think an alternative is true, i.e., relevant. We can point out that some philosophers are very perverse in their standards (by some extreme standard, there is some reason to think there is an evil genius, after all) — but we cannot legitimately go so far as to say that their perversity has stretched the concept of knowledge out of all recognition — in fact they have played on an essential feature of the concept” (p. *254*). I assume Stine would say the same thing about the BIV hypothesis that she says above of the evil genius possibility.

19. Warning: Stroud treats different historical figures in each chapter of his book, and the labels “internal” and “external” seem to mean something different — though perhaps tied together by some hard-to-describe common thread — in each of these chapters. But in some of those chapters, the distinction seems to amount to a difference between two senses of “know”.

20. In his earlier 1979, Lewis warns against the opposite mistake of thinking that the skeptic’s standards are better or more legitimate than the lower, ordinary standards, or that what’s true relative to the skeptic’s standards is somehow more true than what is true relative to ordinary standards (p. 355, see also p. 353).

21. P. *554*; “iff” is an abbreviation for “if and only if”.

22. For further critique of the contextualist approach to skepticism, see Schiffer 1996.

23. Unger 1975. This book incorporates, with some improvements, Unger’s important journal articles from the early 1970’s, while adding new material as well.

24. For a discussion of the relation between the Relevant Alternatives approach and contextualism, and, in particular, of when a Relevant Alternatives theory is a contextualist view, see section II of DeRose 1992.

25. Thanks to Anthony Brueckner, Graeme Forbes, and Ted Warfield for helpful comments.

Skip to toolbar