A New Argument for Skepticism

A NEW ARGUMENT FOR S K E P T IC IS M Baron Reed Northwestern University [email protected] It is commonly thought that the traditional skeptical challenge is fatal for epistemic internalism but merely problematic for epistemic externalism. If knowledge is essentially tied to what is accessible from the “inside,” skeptical scenarios—like vivid dreams and deceiving demons—show that knowledge is impossible. After all, from the inside things can look just the same to a normal person as they do to one who is dreaming vividly. Because the dreamer does not have knowledge, the purportedly normal person cannot, either. On the other hand, if knowledge is grounded in a broader range of facts about the subject, including facts of which she may be unaware, then the traditional skeptical challenge does not show knowledge to be impossible. Externalist theories may still have work to do in response to skepticism—largely because externalism is thought to have too easy an answer for it—but the skeptical challenge is not generally thought to be a direct and immediate danger for externalists. But, even if this correct, it does not mean that externalism is in the clear. After laying out the traditional argument for skepticism and the basic externalist reply to it, I shall construct a new argument for skepticism—one that that is equally damaging to both internalism and externalism. After considering several objections to that argument, I shall close by considering some of the consequences of this new skeptical challenge. I. THE TRADITIONAL SKEPTICAL CHALLENGE Much has been written about the relation between ancient and modern skepticism. Some scholars argue that the ancient skeptics tried to undermine belief while the modern skeptics challenged claims to knowledge. There are also differences in the purposes for which the ancients and the moderns used their skeptical arguments. Skepticism, as it figures in present-day epistemology, is another matter. By and large, it 2 is presented—as it was by Descartes—as a challenge to be overcome, and not as a philosophical position one might reasonably adopt. Despite these differences in the skeptical tradition, it is still possible to discern a common thread running through it. That thread is the basic skeptical challenge posed both to other philosophers and to ordinary subjects. To get clear about the fundamental structure of that challenge, I shall make use of what is perhaps the best-known instance of it: the evil demon hypothesis of Descartes’s First Meditation, which emerges out of the skeptical challenge posed by the possibility of an omnipotent God. How, the meditator asks, “do I know that he has not brought it about that there is no earth, no sky, no extended thing, no shape, no size, no place, while at the same time ensuring that all these things appear to me to exist just as they do now?”1 Although the challenge is presented in the first person, this is not essential to it. Let us suppose A is the victim of an evil demon, while B is a normal human subject. Although they differ in that crucial respect, they are nevertheless exactly alike with respect to how things appear to exist. Now, the skeptic invites us to compare A and B. It is clear that A does not have knowledge of the world around her. Her beliefs are false, which means that the appearances are misleading in her case. Although they make A’s beliefs rational, in the sense of being reasonable or blameless, the appearances do not bring her anywhere near the truth. They are therefore inadequate, from a strictly epistemic point of view. Turning to B, then, notice that her beliefs are grounded in those very same appearances. If they are epistemically inadequate for A, they must be equally inadequate for B. Because there is no relevant difference between the two subjects, B cannot have knowledge of her environment. In short, A and B are alike in all relevant respects. A does not have knowledge, so B cannot have it, either.2 II. THE EXTERNALIST RESPONSE TO TRADITIONAL SKEPTICISM Externalist theories of knowledge and justification are characterized by their grounding of epistemic properties—e.g., knowledge, justification, warrant—in factors that do not have a necessary relation to one’s subjective awareness.3 So, for example, a basic form of reliabilism takes one’s belief to be justified just in case it is the product of a reliable belief-producing mechanism.4 Whether one’s cognitive faculties, like sense perception, are in fact reliable is generally too complex a fact for 3 one to be able to know just by reflection.5 This fact is thus external to one’s subjective awareness. There are, in addition to basic reliabilism, various types of externalism. Some require that the reliable belief-producing mechanisms be virtues of the subject.6 Others focus instead on the modal relations the subject’s belief has to the purported fact known.7 What is common to all of them, though, is the fundamental idea that knowledge and justification are a matter of how well one fits with one’s environment, whether or not one is aware of that fit.8 Given this fundamental conception of knowledge (and other epistemic properties, like justification), the externalist has an easy response to offer the skeptic. The skeptic relies on a comparison between A (the subject in a skeptical scenario) and B (the normal subject), but that comparison is incomplete. There are epistemic differences between A and B—very important ones—and they correctly allow us to attribute knowledge to B even though she cannot tell, simply by reflection, that her situation is different than A’s. Whether or not she is aware of it, B’s beliefs fit well with her environment, whereas A’s do not. This fact, although it is external to B’s subjective awareness, explains why she has knowledge even though A does not.9 It is worth noting that, though this simple answer is available to the externalist, it is rarely defended in quite this way. The explanation for why this is so is not hard to see: if skepticism is obviously false, as it would seem to be on the supposition that externalism is correct, it becomes very hard to see why countless philosophers have been so concerned about it.10 The problem posed by skepticism is thus different for the externalist than for the internalist. The externalist does not face the straightforward challenge of refuting the skeptic’s argument. Although such a refutation is essential to, say, Descartes’s epistemology, it is unnecessary—both for the epistemic subject and for the epistemologist theorizing about knowledge—if externalism is correct. Instead, the externalist faces the rather less worrisome task of explaining why we are prone to mistakenly thinking skepticism is a serious problem.11 III. A NEW SKEPTICAL CHALLENGE The traditional skeptical argument is sometimes presented as depending on the premise that knowledge requires certainty. Because such a 4 conception of knowledge is so vulnerable to skepticism, most philosophers now have abandoned it.12 They are willing to accept that one can have knowledge, even when the epistemic basis for one’s belief is compatible with having a false belief.13 This move, although surely a reasonable one to make, has led to a serious problem in epistemology. Let us suppose that, in order for a subject’s belief to count as knowledge, it is necessary that the epistemic basis for her belief must surpass some minimal threshold of excellence—it must be sufficiently good for her to have knowledge. Under the conception of knowledge we are now working with, this threshold will fall short of what would be needed for certainty. So, it is possible that the subject could hold a belief the basis for which surpasses that threshold, and yet the belief is false. To adapt an example from Bertrand Russell, let us suppose that I walk by a clock everyday on my way to campus.14 The clock has worked perfectly for the past ten years—a fact that I have confirmed countless times by checking what it says against other clocks. Earlier today, however, the clock stopped. When I pass by at 11:30 AM, I do not realize that this is so, and I form the well justified, though false, belief that it is noon. Because the belief is false, it cannot count as knowledge. Now, the problematic twist occurs on the following day. I still do not know that the clock has stopped, but I happen to walk by at noon. The belief I then form is still well justified, given my past experience with the clock’s reliability, but it is also true. Nevertheless, the belief does not count as knowledge. Its truth is accidental—a bit of good luck for me. The problem for epistemology, then, is this: a belief can count as well justified, and also be true, and yet not be knowledge. This, of course, is the so-called Gettier problem.15 Although a huge literature has followed in the wake of Edmund Gettier’s initial paper, the usual reaction has been to regard the cases he presented as counterexamples requiring some sort of addition to standard accounts. No solution to the problem has yet found widespread acceptance, but it has become fairly standard in contemporary epistemology to compartmentalize the Gettier problem—that is, most epistemologists continue working on other issues (e.g., the nature of justification) under the assumption that their views will be unaffected by the requirements of whatever solution ultimately is found. But to see the problem with accidentality in this way, I shall now 5 argue, is to miss the fundamental nature of the difficulty it presents. Cases of the sort that Gettier and others have proposed make possible a new comparative argument for skepticism—but, unlike in the case of the traditional skeptical argument, the nature of the new comparison is such that the move to externalism does nothing to distinguish between the ordinary subject and the subject who does not have knowledge.16 To put the point more plainly, the new skeptical argument affects externalist theories just as much, and in just the same way, as it does internalist accounts. Let us begin with a pair of cases. Car Possession 1: Bartholomew has a lot of very good evidence for the proposition that his friend, Smith, owns a Ford. He has ridden in Smith’s Ford numerous times in the past year, has heard Smith talk about his Ford regularly, and has even seen the title for the car in Smith’s name. Moreover, Smith has never discussed any plans to sell or otherwise get rid of the car. Bartholomew has also recently taken a basic symbolic logic course, and he recognizes that the disjunction introduction rule allows him to derive a true complex proposition by disjoining two propositions where at least one of them is true. So, even though he has no reason to believe that it is now snowing in Albuquerque, he infers (and forms the belief) that Smith owns a Ford or it is now snowing in Albuquerque.17 Car Possession 2: Connor has a lot of very good evidence for the proposition that his friend, Lee, owns a Honda. Connor has ridden in Lee’s Honda numerous times in the past year, has heard Lee talk about his Honda regularly, and has even seen the title for the car in Lee’s name. However, Lee has very recently sold his car. Connor also has taken a basic symbolic logic course and understands how the disjunction introduction rule works. Even though he has no reason to believe that it is now snowing in Albuquerque, he adds this as a disjunct to his belief that Lee owns a Honda. Consequently, Connor forms the belief that Lee owns a Honda or it is now snowing in Albuquerque. Here is another pair of cases. 6 Barn Sighting 1: Bridget, who has excellent eyesight, is driving through an ordinary rural area, which has all of the usual features one would normally encounter in the countryside. She sees a barn not too far from the road and accordingly forms the true belief that there is a barn in the field.18 Barn Sighting 2: Cassandra, who also has excellent eyesight, is driving through a rural area when she sees a barn not too far from the road. She also forms the true belief that there is a barn in the field. However, she is in an area where the farmers have built numerous barn façades, which are so cleverly constructed that they cannot be distinguished from real barns by passing motorists. When we consider in isolation the normal (B) subjects in each pair of cases, it seems obvious to almost everyone that they have the knowledge in question. After all, their beliefs are not only true but highly justified as well. By contrast, when we consider in isolation the abnormal (C) subjects in each pair of cases, it seems pretty clear to the great majority of epistemologists that they do not have the knowledge in question.19 Although their beliefs are true and—it is important to note—just as well justified as those of the B subjects, the C subjects have beliefs that appear to be true by accident. Connor’s complex belief was true because the disjunct he added randomly to his prior false belief happened to be true, though he had no reason to think that it was. And Cassandra’s perceptual belief was formed when she just happened to be looking at the one real barn amongst all the barn façades in the area. She easily could have been looking at one of those façades instead and would have formed the same justified belief, though it would then have been false. Knowledge is thought to be incompatible with this kind of luck—it cannot be a mere accident that one’s belief happens to be true rather than false. Although this is the standard way of reading these cases, it has not been appreciated that they permit a comparison much like the one that underlies traditional skepticism. Here is how the new argument for skepticism works: C does not have knowledge. B is just like C in all epistemic respects. Therefore, B does not have knowledge, either. There are only two premises to the argument. Of these, the first 7 seems fairly secure—virtually everyone agrees that knowledge is incompatible with accidental truth. But, one might object, surely the second premise is false. Isn’t it just the case that we are not in a good position to point out what the epistemic difference is between B and C, given that we do not yet have a solution to the problem of accidental truth? If so, the new argument for skepticism is no more troubling than the original Gettier problem. This objection misses the mark, though in an instructive way. Let us suppose that epistemologists have actually found a universally accepted solution to the Gettier problem. Abstracting from the details of the solution, let us say that there is some condition x that B satisfies and C does not; this is what distinguishes all cases of non-accidentally true belief from all cases of accidentally true belief. We can abstract away from the details of the solution in this way because they do not matter, for the purposes of the new argument for skepticism. For notice that, whatever x may be, it has nothing to do with the epistemic performance of B. By hypothesis, the epistemic performance of B is just the same as the epistemic performance of C. Connor’s belief is grounded in exactly the same sort of evidence as Bartholomew’s; Cassandra’s belief is grounded in exactly the same sort of visual experience as Bridget’s.20 It is important to see that the situation does not change when we take into account the properties central to externalist accounts. Cassandra’s faculty of vision, for example, is just as reliable as Bridget’s, and she is using it in an environment that is well-suited for its operation (as is made clear by the fact that she can then form many justified, true beliefs about the color, size, and shape of the structure, the species of nearby trees, etc.). Their epistemic performances are the same, even when we conceive of those performances in the most broadly externalistic way possible. The satisfaction of condition x, then, stands entirely outside of the subject’s performance. It is not merely external to the subject’s subjective awareness, it is also extrinsic to her epistemic performance.21 Thus, the same recourse to externalism that allowed for an answer to traditional skepticism is of no help at all in response to the new argument for skepticism. Even when we conceive of them in the most rigorously externalistic way possible, there is no epistemic difference between B and C. Because one of them does not have knowledge, the other cannot have it, either. 8 IV. OBJECTIONS AND REPLIES Here I shall consider three objections to the new argument for skepticism. According to the first objection, the above argument depends on the supposed fact that the Gettier problem has not been solved. But one of Nozick’s motivations for defending his tracking theory of knowledge was its ability to handle Gettier cases.22 In Barn Sighting 2, for example, Cassandra fails to satisfy Nozick’s sensitivity requirement: if it were not true that p, S would not believe that p. If Cassandra were looking at one of the barn façades rather than at the single genuine barn in that area, she would still believe that she is looking at a barn. Hence, on Nozick’s view, Cassandra would not know that there is a barn in the field. In reply, it can be granted that Nozick’s theory is able to account satisfactorily for the above Gettier cases, Barn Sighting 2 and Car Possession 2. However, there are other cases where the subject’s belief is sensitive, in Nozick’s sense, yet fails (intuitively) to have knowledge: Barn Sighting 3: Claire, who has excellent eyesight, is driving through a rural area when she sees a red barn not too far from the road. She forms the true belief that there is a red barn in the field. However, she is in an area where the farmers have built numerous barn façades, which are so cleverly constructed that they cannot be distinguished from real barns by passing motorists. In order to prevent themselves from becoming confused about which structures are genuine barns, they have rigorously followed a policy of painting the genuine barns red and the barn façades yellow. Notice that if Claire had formed the belief, there is a barn in the field, that belief would have failed the sensitivity requirement: she would continue to hold the belief even if she were looking at a barn façade. As before, Nozick’s account handles that belief properly. However, the belief Claire actually has in this case is sensitive. If there were not a red barn in the field, she would not believe that there was one (she might believe that there is a yellow barn in the field, but of course that’s a different belief). Despite being sensitive, though, Claire’s belief is not knowledge. Although the belief she has could not easily have been false, she still could easily have had another belief in its place which then 9 would have been false. It is a matter of luck that this did not happen. In that sense, then, the belief Claire does have is accidentally true.23 According to a second objection, whatever the anti-accidentality condition may turn out to be, it does not matter that it is extrinsic to the subject’s epistemic performance. Truth, after all, is a condition of knowledge, and it is equally extrinsic to the subject’s epistemic performance. If epistemologists do not think that truth’s being extrinsic is a problem, why should it be so for the condition that rules out cases of accidental truth? By way of reply, I shall argue that truth and the anti-accidentality condition are not really on a par. To see this, notice first that the traditional skeptical challenge could not be effectively answered by simply adding truth to the subject’s epistemic performance when truth is construed as extrinsic to it. That is, it would be an ineffective reply to say, e.g., in response to the evil demon scenario, that B has knowledge while A doesn’t simply in virtue of the fact that B’s beliefs are true. The point of the traditional argument is that B’s justification for her beliefs—her evidence or reasons for them—is not appropriately connected to the truth.24 To take Betty to nevertheless have knowledge is to treat knowledge as consisting in the simple conjunction of justification and truth. But this is not what knowledge is. The widespread intuitive response to Gettier cases shows that knowledge must be more than merely justified belief that is also true. The something more is that there has to be the proper connection between the subject’s justification and the truth of her belief. When that proper connection obtains (assuming this is possible), the truth is the attained end of the subject’s performance. To return to a distinction drawn in section III, the truth may be extrinsic to the subject’s performance, but, when that performance is properly connected to the truth, the truth then is not external to the subject’s awareness. But things are very different for the anti-accidentality condition. The subject’s performance is not aimed at ensuring that it is satisfied. In Barn Sighting 1, for example, Bridget’s performance is directed at determining whether there is a barn in the field in front of her. She is not also trying to determine whether the surrounding fields have genuine barns or mere barn façades. In that sense, Bridget is blind as to whether the antiaccidentality condition has been met. Her epistemic situation of course would be better if she were sensitive to the presence of barn façades in 10 the surrounding fields, but she in fact is not. The simple fact that there aren’t any façades around her makes no epistemic difference to her. What is problematic about the anti-accidentality condition, then, is that it is both extrinsic to the subject’s performance and external to the subject’s awareness. Truth and justification (conceived in an externalistic way) may be one or the other, but they are not both. Finally, according to a third objection, it may be granted that the epistemic performance of a normal subject, B, may be indistinguishable from that of a subject, C, in a Gettier case, but this does not mean that there is no epistemic difference between them. As many virtue epistemologists have argued recently, what is distinctive about knowledge is that, in all and only those cases where the subject does have knowledge, she deserves at least partial credit for the truth of her belief.25 Thus, John Greco says that, in cases of purported knowledge, the subject’s virtues are the most salient part of the explanation for her success.26 According to Ernest Sosa, a performance is apt when its success is “sufficiently” due to the subject’s competence.27 By contrast, the success a subject has in a Gettier case is not due in any significant way to her performance.28 So, even if C’s performance is intrinsically the same as B’s, there is still an important difference in their effects. But there are several reasons for thinking that there are not in fact any significant differences, with respect to creditability, between purported instances of knowledge and accidentally true beliefs. First, salience is typically a matter of context. In some contexts, then, the subject’s virtues might be the most salient part of the explanation for why her belief is true—even when it is true accidentally. To see this, let us return to Car Possession 2. Suppose, now, that Connor has a friend, David, who has all of the same evidence indicating that Lee owns a Honda. Like Connor, David has also taken a symbolic logic course, though he did quite poorly. As a result, David tends to confuse the conjunction introduction rule with the disjunction introduction rule. So, David thinks that he can conjoin a randomly chosen proposition with one that he takes to be true and thereby derive a true complex proposition. So, even though he has no reason to believe that it is snowing in Albuquerque, he (mis)applies the conjunction introduction rule to his original proposition and comes to believe that Lee owns a Honda and it is snowing in Albuquerque. His new belief, unlike Connor’s, is false. What accounts for the success Connor has achieved and David has not? 11 The salient difference between them is simply Connor’s possession of a virtue that David lacks. In that sense, Connor’s success is creditable to him as a product of his virtue. Nevertheless, his belief, though justified, is still accidentally true and not a case of knowledge.29 Second, as Jennifer Lackey has argued, there are purported cases of knowledge where the subject does not seem to deserve any significant amount of credit for the success of her belief.30 For example, suppose that I have just arrived in an unfamiliar city. I randomly choose a passerby and ask for directions to the nearest subway station. As it happens, the directions-giver is reliable and gives me accurate directions. Most philosophers are willing to recognize as knowledge my newly acquired belief that the nearest subway station is six blocks to the west. Nevertheless, it is hard to see why I—rather than the testifier— deserve much credit for the success of that belief.31 Although I may deserve some small measure of credit, so too does the subject in a Gettier case deserve some small measure of credit for the success of his belief.32 As Lackey argues, there does not appear to be any significant difference here. Third, that a subject with a justified, accidentally true belief deserves at least some credit for the success of the belief is clear even in cases like Car Possession 2, which is a paradigmatic Gettier case. But the point is even stronger when we turn to cases like Barn Sighting 2. Cassandra does not deserve less credit than Bridget simply because there are no barn façades in the area around Bridget—especially given that they are equally insensitive to the presence of those façades. Because Cassandra has such a good claim to deserving credit for the success of her belief, Sosa argues that her belief should properly be regarded as knowledge.33 According to his version of virtue epistemology, a belief counts as knowledge when it is apt—i.e., when it is true because it is competent.34 Cassandra’s belief is apt in this sense: it is true because she is exercising a competence—her perceptual faculty—in conditions that are appropriate for its use. This is so despite the fact that Cassandra easily could have exercised her competence in a way that would have been unsuccessful. That is, she easily could have been looking at a barn façade instead of a genuine barn. Still, as Sosa says, “That [the belief] is apt by luck makes it no less apt” (p. 87). Is Cassandra’s belief accidentally true in a way that precludes it from counting as knowledge? Consider the following extension of Barn 12 Sighting 2: Cassandra drives further in the area and sees two more structures. Although she believes that she has seen two more barns, each is in fact a façade. Suppose, then, that Cassandra is told that only one of her three beliefs is true. She ought to abandon (at least) two of her beliefs, but she cannot tell which are the false ones. If she manages to retain the belief that is true, this will be tantamount to a lucky guess. For that reason, it is very hard to see her success as knowledge in any sense. But notice that the epistemic basis for her belief is no worse than it was before she was told of the existence of barn façades in the area.35 Both before and after, that belief is supported by the same perceptual experience. If it is not good enough after learning that two of her similar beliefs are false, it was not good enough before she acquired that information. For the same reason, it was not good enough in the original Barn Sighting 2 case. Whether or not her belief is apt, its success is too accidental for it to count as knowledge. The upshot, then, is that there is no sense in which a subject (B) who purportedly has knowledge deserves credit for her beliefs in a way that a subject (C) with accidentally true beliefs does not. And, so, credit for truth does not serve to distinguish between B and C. They are alike in all relevant respects. C does not have knowledge, so B does not, either. V. PRESUPPOSITIONS AND CONSEQUENCES Every skeptical argument embodies some presuppositions about the nature of knowledge. Without them, it would be impossible to offer an argument of any sort. In some cases, the challenge posed by the skeptical argument can be overcome by abandoning its underlying presupposition. This has happened, for example, with arguments that presuppose knowledge to require certainty. The trick for the skeptic, then, is to find presuppositions that are so fundamental that they cannot be abandoned. Has this happened in the case of the new argument for skepticism? There are only four presuppositions on which it rests. The first holds that subjects who are alike in every epistemic respect are also alike with respect to whether they have knowledge. If this thesis were false, it would mean that knowledge would float freely of everything else that we take to have epistemic relevance. If that were really the case, it is hard to see how any sort of meaningful epistemology would be possible. 13 The second presupposition is fallibilism. To reject it is to leap from the frying pan into the fire. The traditional skeptical arguments lie in wait should we decide to return to a conception of knowledge that requires certainty. The third presupposition is that fallibilism makes possible instances of justified but accidentally true belief. This has been a matter of some dispute—e.g., Alvin Plantinga has said that the Gettier problem really applies only to internalist theories, whereas Timothy Williamson has argued that it is a problem only for epistemologies that are attempts to analyze knowledge.36 But problems with accidentality have been shown to arise for their favored theories as well.37 At this point, there is no reason to think that any version of fallibilism can escape the need to rule out accidentally true belief. The final presupposition is just this: a belief that is accidentally true—no matter how well justified—cannot be an instance of knowledge. This is one of two bedrock principles in epistemology (the other one being that you cannot know what is false). To say that it should not be abandoned lightly is to grossly understate its importance. Doing so would necessitate a re-conceiving of knowledge so radical that it would represent a concession to the skeptic no less significant than an outright admission of defeat. If the new argument goes through, then, what are the consequences? For example, would it mean that we are rationally compelled to withhold or abandon our beliefs? That does not follow simply from the skeptical argument itself—nor, it should be said, does it follow solely from the traditional skeptical argument, either. The Pyrrhonists thought that the suspension of belief follows naturally from being presented with equally plausible, incompatible arguments. Whether that is so is something that can be left to psychologists to determine for, in any case, it is not the situation that the new argument describes. The new argument for skepticism is not grounded in the presentation of counterbalanced arguments but in reflection on what it means for a belief to be true by accident. Whatever consequences of the argument there may be, they will stem simply from the conclusion that knowledge is not possible. Are we thereby missing anything of value? It seems fairly clear that knowledge would be significantly more valuable than mere justified true belief. Certainly, we would prefer to find ourselves with knowledge rather than 14 in a Gettier case. Why this is so may be difficult to articulate, but I will close by offering two possible explanations for it. First, knowledge would be a more stable possession than mere justified true belief.38 A subject whose justified belief is true by accident would, typically, be easily persuaded to abandon the belief through learning how easily the belief could be false. This would usually not be so in the case of knowledge. Second, and more speculatively, the possession of knowledge seems to be essentially linked with other fundamental values. We do not count accidentally true belief as knowledge because we take it to be important that the subject has acquired knowledge in the right sort of way. It must belong to the person as her doing, much as an action for which a person is responsible must belong to her.39 Although a fuller defense of this claim must wait for another occasion, it would mean that the value of knowledge is perhaps much like the value we accord to free action. In both cases, they are constitutively linked to the value of being a person: knowledge and freedom give depth and substance to our natures as active beings.40 If this is so, then skepticism, like the problems surrounding freedom, is a challenge to our fundamental sense of self. This, no doubt, is why it has proven to be such an enduring part of the philosophical tradition.41 REFERENCES Chisholm, Roderick. 1989. Theory of Knowledge, 3rd. ed. Englewood Cliffs, NJ: Prentice Hall. Cohen, Stewart. 1998. “Contextualist Solutions to Epistemological Problems: Skepticism, Gettier, and the Lottery,” Australasian Journal of Philosophy 76: 289-306. ——. 1988. “How to Be a Fallibilist,” Philosophical Perspectives 2: 91-123. DePaul, Michael and Zagzebski, Linda (eds). 2003. Intellectual Virtue: Perspectives from Ethics and Epistemology. Oxford: Oxford University Press. DeRose, Keith. 1995. “Solving the Skeptical Problem,” Philosophical 15 Review 104: 1-52. Descartes. 1984. The Philosophical Writings of Descartes, Vol. II, ed. by J. Cottingham, R. Stoothoff, and D. Murdoch. Cambridge: Cambridge University Press. Feldman, Richard. 2003. Epistemology. Upper Saddle River, NJ: Prentice Hall. Frankfurt, Harry. 1971. “Freedom of the Will and the Concept of a Person,” The Journal of Philosophy 68: 5-20. Fumerton, Richard. 1995. Metaepistemology and Skepticism. Lanham, MD: Rowman & Littlefield. Gettier, Edmund. 1963. “Is Justified True Belief Knowledge?” Analysis 23: 121-3. Goldman, Alvin. 1979. “What Is Justified Belief?” in G. Pappas (ed.), Justification and Knowledge (Dordrecht: Reidel): 1-23. ——. 1976. “Discrimination and Perceptual Knowledge,” Journal of Philosophy 73: 771-791. Greco, John. 2007. “The Nature of Ability and the Purpose of Knowledge.” Philosophical Issues 17: 57-69. ——. 2003. “Knowledge as Credit for True Belief,” in DePaul and Zagzebski (2003): 111-134. ——. 2000. Putting Skeptics in Their Place. Cambridge: Cambridge University Press. Greene, Richard and Balmert, N.A. 1997. “Two Notions of Warrant and Plantinga’s Solution to the Gettier Problem,” Analysis 57: 132-9. Hetherington, Stephen. 1999. “Knowing Failably,” The Journal of Philosophy 96(11): 565-587. 16 Lackey, Jennifer. Forthcoming-a. “Knowledge and Credit,” Philosophical Studies. ——. Forthcoming-b. “What Luck Is Not,” Australasian Journal of Philosophy. ——. 2007. “Why We Don’t Deserve Credit for Everything We Know.” Synthese 156: 345-61. Nozick, Robert. 1981. Philosophical Explanations. Cambridge, MA: Belknap Press. Plantinga, Alvin. 1993. Warrant and Proper Function. New York: Oxford University Press. Reed, Baron. 2007. “The Long Road to Skepticism,” The Journal of Philosophy 104: 236-62. ——. 2005. “Accidentally Factive Mental States,” Philosophy and Phenomenological Research 71: 134-42. ——. 2002. “How to Think about Fallibilism,” Philosophical Studies 107: 143-57. Riggs, Wayne. Unpublished. “Two Problems of Easy Credit.” ——. 2002. “Reliability and the Value of Knowledge.” Philosophy and Phenomenological Research 64: 79-96. Russell, Bertrand. 1948. Human Knowledge: Its Scope and Limits. New York: Simon and Schuster. Sosa, Ernest. 2007. Virtue Epistemology: Apt Belief and Reflective Knowledge, Vol. I. New York: Oxford University Press. ——. (2003) “The Place of Truth in Epistemology,” in DePaul and Zagzebski (2003): 155-179. 17 ——. 1999. “How Must Knowledge Be Related Modally to What Is Known?” Philosophical Topics 26: 373-84. Van Cleve, James. 1979. “Foundationalism, Epistemic Principles, and the Cartesian Circle,” Philosophical Review 88: 55-91. Williams, Michael. 1999. “Skepticism,” in J. Greco and E. Sosa (eds.), The Blackwell Guide to Epistemology (Oxford: Blackwell): 35-69. Williamson, Timothy. 2000. Knowledge and Its Limits. New York: Oxford University Press. Wolf, Susan. 1990. Freedom Within Reason. New York: Oxford University Press. Zagzebski, Linda. 1996. Virtues of the Mind. Cambridge: Cambridge University Press. 1 First Meditation; Descartes (1984), p. 14. 2 Though it falls outside the scope of this paper, I believe that a similar structure can be shown to be at work in the other main arguments in the skeptical tradition, including Descartes’s dreaming argument, the Modes of Pyrrhonism, and the problem of the criterion. 3 That is, externalists reject epistemic internalism. There are various ways in which internalism has been characterized; for a helpful discussion, see Fumerton (1995), pp. 60-9. One of the most widely shared is Chisholm’s characterization of epistemic justification as “internal and immediate in that one can find out directly, by reflection, what one is justified in believing at any time” (1989, p. 7). 4 See Goldman (1979) for a statement of basic reliabilism. 5 Moreover, even if a particular subject does have some awareness of the reliability of her faculties or belief-forming processes, the justification for her beliefs is independent of that awareness. 6 See Sosa (1991) and Greco (2000) for two of the central defenses of virtue epistemology. See Zagzebski (1996) for a different type of virtue 18 epistemology, grounded in the Aristotelian tradition. 7 See Nozick (1981). Sosa (1999) defends the requirement of a modal connection between belief and purportedly known fact in addition to the other elements of his virtue epistemology, though he appears to abandon this requirement in his (2007). See also Williamson (2000). Nozick argues on behalf of sensitivity (if p were not true, one would not believe p), where Sosa and Williamson defend safety (if one were to believe that p, it would be true that p). 8 This is compatible with a view such as Sosa’s, which takes the awareness of one’s fit in one’s environment to yield a better epistemic status for one’s belief than it would have had in the absence of that awareness. See his distinction between animal and reflective knowledge (2007, p. 24). 9 For one example of this strategy, see Van Cleve (1979) on epistemic circularity. 10 This has been one of the primary motivations for contextualism; see Cohen (1988 and 1998) and DeRose (1995). 11 For externalist responses to this problem, see, e.g., Nozick (1981) and Sosa (2007). 12 See, e.g., Williams (1999) and Feldman (2003). Although Feldman thinks that fallibilism is an adequate response to some forms of skepticism, he does acknowledge that there are skeptical arguments which do not rely on the impossibility of certainty; see p. 128. 13 See my (2002) for an account of fallibilism. 14 Russell (1948), p. 154. 15 See Edmund Gettier’s classic paper (1963). Though Russell’s brief example was earlier, proper focus on the problematic nature of cases of that sort begins with Gettier. 16 I should emphasize, however, that neither Gettier nor any of the other philosophers who have proposed similar cases involving accidentally true justified beliefs have suggested a comparison of the sort that I will be making. See my (2007) for a different sort of presentation of the new argument for skepticism; there, I also defend the new argument against some popular anti-skeptical strategies, including contextualism and Moorean common sense. 17 This and the following case are modifications of one of the two examples in Gettier (1963). As before, I shall use “B” names to refer to 19 subjects in normal situations; I shall now use “C” names to refer to those whose beliefs are accidentally true. 18 This and the following case are modifications of an example that appears in Goldman (1976), to whom it was suggested by Carl Ginet. 19 Hetherington (1999) is an exception. He regards Gettier cases as borderline instances of knowledge. Sosa (2007) agrees that Connor does not have knowledge in Car Possession 2, but he thinks that Cassandra’s belief should count as knowledge in Barn Sighting 2. I shall return to this claim below. 20 If we like, we can even make the epistemic performance of the C subjects better than that of the B subjects. For example, we could allow Connor’s belief to be grounded in a more extensive range of evidence (including, perhaps, a look at an affidavit signed by Lee stating that he will never sell his car) and Cassandra’s belief to be grounded in a better perceptual experience (e.g., she sees the barn from a closer vantage point and in better light). Still, the beliefs of the C subjects would be accidentally true. 21 For more on this distinction, see my (2007). 22 Nozick (1981), pp. 173-5. 23 For cases like Barn Sighting 3, see Lackey (forthcoming-b). 24 Given this way of thinking of the traditional argument for skepticism, we can see why externalism provides a plausible reply to it. If externalists are correct about what is required for knowledge (e.g., justification or warrant), Betty’s justification is appropriately connected to the truth. She could not have the justification she does unless her beliefs are probably true. 25 See Riggs (2002) and (unpublished), Greco (2003) and (2007), Sosa (2003) and (2007), and Zagzebski (2003). 26 See Greco (2003). 27 See Sosa (2007), pp. 79 and 97. 28 Sosa says that, in a Gettier case, the subject’s competence may be the explanation for why she has the belief in question, but it does not explain why it is true—in other words, the competence accounts for the belief’s existence but not for its correctness (2007, pp. 95-6). 29 I am grateful to Jennifer Lackey for discussion of this point. 30 See Lackey (2007) for this argument. See Greco (2007), Sosa (2007), and Riggs (unpublished) for responses to Lackey and Lackey 20 (forthcoming-a) for her further defense of the argument. 31 This case is drawn from Lackey (2007). She also there presents cases in which a subject apparently has knowledge but without deserving much credit for it, where the purported knowledge in question is not testimonial. 32 For example, in the case above, Connor is clearly performing better intellectually than David is. Connor’s disjunctive belief is not only justified, it would count as knowledge if the original disjunct were true. By contrast, David’s belief is not justified, and it would not count as knowledge even if both conjuncts happened to be true. 33 See Sosa’s discussion of the kaleidoscope believer—a case which is structurally similar to the barn façade case (2007, pp. 31-4, 96 n. 1, 99101, and 104-9). 34 Sosa (2007), pp. 23-4. To be precise, Sosa would say that Cassandra’s belief is animal knowledge but not reflective knowledge (pp. 36-7 and 100-9), where reflective knowledge is apt belief aptly noted (p. 32). In what follows, I shall largely ignore the distinction between animal and reflective knowledge, as I shall be objecting to the claim that Cassandra has any sort of knowledge. 35 One might object here that Cassandra has been given a defeater (counterevidence) for her belief, so her epistemic situation is in fact worse than it was before. But the point can be put in the third-person just as well. An observer who learns that Cassandra has seen only one genuine barn and two barn façades would say that the epistemic basis for her true belief is inadequate. It is not that the belief was wellsupported and has since been outweighed by stronger evidence to the contrary. Rather, the epistemic basis for the belief was never good in the first place. 36 See Plantinga (1993), p. 36, and Williamson (2000). 37 See Greene and Balmert (1997) and my (2005), respectively. 38 In the Meno, Plato says that when a subject has knowledge, there is a “tether” for her belief so that it cannot run away. 39 This is not to say that the subject must deserve credit for the truth of her belief. Rather, the point is merely that the knowledge must be attributable to her, even if someone else is largely responsible for putting the subject in a position to have a true belief. Similarly, an action may be attributable to an agent even in cases where someone else 21 deserves most of the credit for making the action possible—e.g., a ballet instructor may deserve the credit for enabling her pupil to perform a pas de chat, though of course it is the pupil who is actually performing it. For more on attributability, see my (2007). 40 For this claim as it relates to freedom of the will, see Frankfurt (1971) and Wolf (1990). 41 I am grateful to audience members at the Midwest Epistemology Workshop—especially E.J. Coffman, Sandy Goldberg, Matt McGrath, Andrew Moon, Ernest Sosa, and John Wynne—for helpful comments on an earlier version of this paper. Most of all, I am very grateful to Jennifer Lackey for her helpful comments on this paper and, even more, for the many, many insights I have gained through our years of discussing the philosophical problems with which it is concerned. I am also thankful that she is not too skeptical of my skepticism.