Version 2.3, May 28, 2015
As stated in the introductory module, the project of validation seeks to show that we have knowledge of things of which we ordinarily believe we have it. Throughout the history of Western philosophy, there have been philosophers (“dogmatists”) who have defended claims to knowledge and philosophers (“skeptics”) who have argued against knowledge claims. There seems to be a general, but not universal, presumption among epistemologists that skepticism is something that ought to be resisted in many cases. One way describe the project of resisting skepticism is as a project of “validating” claims to knowledge.
A skeptic might deny for some subject S, circumstances c and proposition p that S knows that p in c.
~Ks,cp.This is a strong form of skepticism that corresponds to the ancient “Academic” form. The Academic skeptics confronted claims of knowledge made by the Stoic philosophers and denied them. A weaker form of skepticism would simply refuse to make attributions of either knowledge or ignorance, on the grounds that this would itself implicitly be a knowledge claim. This kind of strategy is associated with the “Pyrrhonian” school of ancient skepticism. The focus here will be on the stronger form of skepticism.
We can understand skepticism more precisely by recalling our symbolic representation of the most commonly accepted necessary and sufficient conditions for knowing.
Ks,cp ↔ (p ∧ Bs,cp ∧ WKs,cp).A subject S knows that p in circumstances c if and only if p is true, S believes that p in c, and S is sufficiently warranted in believing that p in c. From this analysis, it follows by classical logic that if S does not know that p in c, then either p is not true, S does not believe that p in c, or S is not sufficiently warranted in believing that p in c.
~Ks,cp → (~p ∨ ~Bs,cp ∨ ~WKs,cp).Skepticism in epistemology is primarily directed at the third way in which the conditions for knowledge may fail to be satisfied. Skeptics either claim that S is not sufficiently warranted in believing that p in c, or they claim that the dogmatist has not made the case that S is sufficiently warranted in believing that p in c. We will focus here on the stronger, and more provocative, claims by skeptics that in some ranges of cases, S lacks sufficient warrant.
The question about whether a subject has sufficient warrant for believing a proposition gives rise to further questions. Which account of sufficient warrant is in play? What is the standard of sufficiency? Importantly, the higher the standard adopted by the skeptic, the easier the task of showing insufficiency will be. The most effective skeptical tactic is to adopt, for the sake of argument, the standard of someone making a knowledge claim and then argue that this standard is not met. Alternatively, the skeptic might argue for the adoption of a fixed standard of sufficiency and show that it is not met.
The most extreme form of skepticism would be one that is global or universal. It would hold that for no subject S, no circumstances c, and no proposition p is S sufficiently warranted in believing that p in c.
Global Skepticism. For no s, c, p, WKs,cpGlobal skepticism is almost universally avoided. One obvious problem with it is that it applies to itself. If global skepticism is true, then no one is under any circumstances sufficiently warranted in believing that it is true. So no one could know that global skepticism is true.
Other forms of skepticism are localized, involving one or more of the three parameters, the subject, the circumstances of the subject, and the proposition.
Local Skepticism. For some range of s, c, p, ~WKs,cpThese are typically classified by skeptics into certain kinds or types. For example, internalists may be skeptical about the ability of non-human animals or small children to be sufficiently warranted. One might be skeptical about subjects who are in unusual circumstances, such as when they are in a deceptive environment, as in the case of being in an area where there are many decoys. We shall here assume that the range of subjects is normal human subjects and that the circumstances in which these subjects find themselves are normal. This presents the greatest challenge for skepticism.
Kinds of propositions about which skepticism is quite tempting include those about the future, other minds (what they contain, or even whether they exist at all), theoretical objects of science, the existence of God, what is good and what is evil, what is beautiful and what is ugly. In each case, a skeptic contends that the subject’s epistemic position is not strong enough to allow sufficient warrant. The future has not yet happened, and the course of the future has many times confounded our expectations. We as individuals do not have direct access to the minds of others, and our human faculties of perception do not allow us to observe directly the theoretical objects postulated in science. God is said to be a being beyond nature itself, with properties that defy human comprehension. Good and evil are values that are not objects of physical observation, and beauty and ugliness seem to be, to some extent, in the eye of the beholder.
We will devote ourselves here to a skeptical topic originally raised by Descartes: whether one is sufficiently warranted in believing that there exists a physical world external to one’s own mind. This has been called “Cartesian” or “external-world” skepticism. The rest of this module will be devoted to the skeptical claim that human minds cannot in any circumstances know whether an external world exists.
External-World Skepticism. For no c, WKs,cp (where s is my mind and p is the proposition that a physical world exists externally to s)
The classical basis for external-world skepticism, as enunciated by Descartes, is the claim that the only evidence directly available to the mind is its own contents, such as thoughts, perceptions, desires, etc. If the subject is to be warranted in believing that a physical world exists externally to it as a mind, the evidence for such a belief can only be indirect. Then the claim is that this indirect evidence can never provide sufficient warrant.
Arguments for External-World Skepticism
We may construct a general schema for skeptical arguments about what is mind-independent in the following way. Consider any arbitrary subject S, circumstances c in which S exists, and proposition p which presupposes the existence of mind-independent things.
A straightforward rebuttal to this argument is simply to deny the truth of the second premise and hold that S can in fact distinguish on the basis of the contents of S’s mind dreaming from waking (Meditation Six, final paragraph). Perhaps sometimes in our dreams we cannot tell that we are dreaming, but there are many features of our waking life which are quite different from what takes place in dreams, so in circumstances when we are awake, we can distinguish our being awake from circumstances in which we are dreaming. And this is enough to show that in at least these circumstances, our epistemic position is strong enough to undercut the dream argument.
Now the skeptic may try to defend premise 2 by considering situations in which a physical world does not exist, and which by definition, cannot be distinguished on the basis of the contents of S’s mind from situations in which one does. This makes premise 2 of the schema true by definition. Descartes himself described such a situation: one in which a powerful non-physical deceiver supplies S with all the perceptions which S would have if perceiving a physical world, but where no physical world exists (Meditation One). Let us call this the “demon” argument. It states that we cannot be in any circumstances which will allow us to distinguish physical-world experience from demon-induced experience, and so (the argument goes), S cannot be in a strong enough epistemic position with respect to the proposition that a physical world exists externally to S.
Clearly, the demon argument cannot be handled in the same way as the dream argument, because premise 2 cannot be denied. The standard validation strategy is to deny premise 1, which requires that S be able to distinguish on the basis of the appearances to S’s mind between cases where a physical world exists externally to S’s mind and where it does not. Perhaps there is some other basis besides the mere appearances to one’s mind on which the two kinds of situations can be distinguished. Appeal is standardly made to the faculty of reason. It is said, against the skeptic, that S can infer from the appearances to S’s mind to the conclusion that an external physical world exists. Let us call this the “inferential strategy.” In the case of Descartes, and inference is made using pure reason to the existence of a God who would not allow us to be deceived into believing that an external physical world exists, when it does not.
A second way to deny premise 1 would be to restrict the range of situations in which S must be able on the basis of the appearances to S’s mind to distinguish between a physical world and a world of illusion. One might, for example, declare that the demon situation is not relevant to the required strength of S’s epistemic position. If, say, S can distinguish dreams from waking when S is awake, that is enough to put S in a strong enough epistemic position. As will be seen, this approach is taken in one way or another by several contemporary epistemologists.
Contemporary Epistemology and External-World Skepticism
The bulk of the literature in contemporary epistemology is devoted to the linguistic, analytical and normative projects. For the most part, these projects are pursued with little or no reference to the Cartesian skeptical question. Only a relatively few analytic epistemologists take such skepticism seriously enough to discuss it in detail. The most prevalent view is that Cartesian skepticism grows out of long-discarded assumptions made by seventeenth-century epistemologists. The very fact that some epistemogists take external-world skepticism seriously is seen by some to be an embarrassment to epistemology. (Thomas Reid, in the eighteenth century, may have been the first to express this view.)
There are exceptions to the general tendency of contemporary epistemology to dismiss external-world skepticism. Peter Unger and Robert Fogelin have defended forms of skepticism, as did Keith Lehrer in an early paper. Barry Stroud has waged a long, hard, and frustrating campaign advocating that epistemologists should take skepticism seriously. Some internalist epistemologists have grappled directly with skepticism. Among these are John Greco, Laurence BonJour, and Keith Lehrer, whose views we will discuss below. Another is Jonathan Vogel.
The most important development in the recent literature on skepticism has been the appeal to contextualism as a way to deal with skeptical arguments. We have dealt with one aspect of contextualism in a previous module, and we shall return to it later in the present one. The rise of externalist accounts of warrant also has implications for skepticism, which will be discussed in turn.
Why External-World Skepticism is Ignored
It appears that there are three reasons that external-world skepticism is largely ignored in contemporary analytic epistemology. One is the widespread adoption of particularism, the contemporary descendant of Reid’s philosophy of common sense. The particularist begins epistemological investigation with the assumption that we have the knowledge that the skeptic denies we have and never looks back.
The second reason for the dismissal of skepticism is the nearly universal adoption of fallibilism. It appears to many epistemologists that the main skeptical problem is posed by extreme skeptical hypotheses such as the Cartesian demon or the brain-in-a-vat. A fallibilist can simply dismiss these hypotheses as being irrelevant to the question of whether we know or have sufficient warrant. (This is what was described above as the second strategy for denying premise 1.)
The third reason is the widespread adoption of externalism. Skepticism is dismissed in a way that is methodist rather than particularist (though it does presuppose fallibilism). Externalist accounts of knowledge seem to be less sensitive to skeptical problems than are internalist accounts. Sufficient warrant for the internalist may require the ability to rule out skeptical hypotheses from “the inside,” but sufficent warrant for the externalist seems to make no such demand at all. For the externalist, whether we know is an empirical question having to do with how reliable, virtuous, properly functioning, etc. our cognitive functions actually are.
Defense of Particularism
Roderick Chisholm is perhaps the most prominent defender of particularism. He claimed that intuitions about what we know are the only plausible starting point for a theory of knowledge. “This may seem like the wrong place to start, but where else could we start?” (Theory of Knowledge, second edition, p. 16). John Greco expands on Chisholm’s comment. “If a theory of knowledge does not explain our considered intuitions about knowledge, what else could it have going for it? What kind of evidence could we bring to bear to show that the theory is correct?” (Putting Skeptics in their Place, p. 23).
What Greco calls “our considered intuitions about knowledge” include not only intuitions about particular cases of knowledge, but also intuitions about what knowledge is. When we engage in theory of knowledge, we start with both kinds of intuitions. Skeptical arguments indicate a tension between them, and the methodists and particularists resolve this tension in different ways.
One objection to particularism is that it begs the question against the skeptic. If our “considered intuitions” about what we know include attribution of knowledge that a skeptic refuses to make, the objection goes, it would not be right to invoke these intuitions in arguing against a skeptic.
Greco’s response is to claim that no question can be begged, because there are no real skeptics with whom to engage in argument. Further, if there were such an argument, it could never be won, and we would never be able to persuade someone who is a skeptic not to be a skeptic. The examination of skeptical arguments in epistemology is useful, on Greco’s view, only insofar as it helps in the shaping of a particularist theory of knowledge.
The “Moorean Argument”
From the point of view of the methodist, particularism is arbitrary. What, the methodist asks, entitles one to begin the investigation of knowledge with the pre-emptive assumption that we know largely what we think we know? There is a line of argument attributed to G.E. Moore that attempts to answer this question.
Consider first the schema of any skeptical argument.
Because the argument is deductively valid, if one concedes premise 2, then one must either accept the skeptical conclusion or reject premise 1. Moore’s claim is that in cases of ordinary beliefs like the present one it is more plausible to reject the alleged condition on knowledge than to reject the attribution of knowledge. This claim can be made into a general anti-skeptical argument.
The methodist has no objections to this argument schema. Its implementation in the present case, however, is another matter. Everything comes down to the question of the relative plausibility of the alleged condition on knowledge and the knowledge attribution. The particularist holds that the knowledge attribution is always more plausible than the condition and for that reason rejects the condition.
A methodist might claim without qualification that the knowledge condition is more plausible than the knowledge attribution, thus flying in the face of “common sense.” This might be the reaction of someone who thinks that knowledge has an essence which is expressed in condition C, and that one has some kind of intuitive access to that essence. Or, one could hold that condition C expresses the meaning of the concept of knowledge and claim that we have a priori access to that meaning. The authority of our source of insight into the nature of knowledge would then said to be greater than that of the deliverances of “common sense.”
This insight is generally gained after one has become accustomed to making knowledge attributions sanctioned by “common sense.” Because of this, attribution of knowledge is highly plausible initially. But when one reflects on what it is to know, condition C becomes even more plausible—so plausible that the initial plausibility of the knowledge attribution is diminished if it is seen that the condition is not met. (An example of this process might be Descartes’s initial acceptance of what he was
taught by nature as sufficient for knowledge, which was supplanted by the stricter condition of illumination by the
light of nature. See Meditation Three for this distinction.) The anti-skeptical methodist then sets out to show that the condition is in fact met, while the skeptical methodist argues that it is not met.
How is the dispute between the particularist and the methodist to be solved? It seems that the problem of the criterion poses a threat to any solution. Judging the most plausible approach, accepting the condition that threatens skepticism or accepting the knowledge attribution, seems difficult if not impossible to do from a neutral perspective.
A more specific version of the general skeptical argument involves principle of epistemic logic which in the first module we called Closure Under Detachment.
If S knows in c that p, and S knows in c that if p then q, then S knows in c that q.Most epistemologists think that this condition is too strong, since S might not believe that q, perhaps because S has not worked out the connection between p and q. At any rate, we can formulate a skeptical argument based on closure of knowledge under known detachment.
Some epistemologists have attempted to counter this kind of skeptical argument by denying that Closure Under Detachment is a condition of knowledge. Externalist accounts of knowledge seem forced to give it up, since one might be able to satisfy the external condition for knowing that two human hands exist without satisfying the external condition for knowing that the actual world is not a demon world. On the other hand, some epistemologists think that the principle is too plausible intuitively to give up. (And, of course, a Pyrrhonian skeptic would ask how this dispute could be satisfactorily resolved at all.)
It may be that the plausibility of a closure condition for knowledge, such as Closure Under Detachment, varies with the “tightness” of the connection between what is known and its consequences. One well-known invocation of closure has already been mentioned. Kripke showed that on Nozick’s counterfactual account of knowledge, one would be said to know that he sees a red barn, but not that he sees a barn. This violates (in almost any realistic case) known closure, and it seems to be an egregious violation. As Kripke put it, any intuition that supports knowledge that the subject knowes that he sees a red barn also supports the intuition that he knows that he sees a barn.
The existence of two human hands presupposes that a physical world containing the hands exists, and this presupposes that it is not the case that there is no such world, which in turn presupposes that it is not the case that we are tricked by appearances that exist while a physical world does not. Unlike in the barn case (and Dretske’s zebra case), the proposition that we are not being so tricked contains information extraneous to the proposition at issue, in this case that two human hands exist. Because of this, one might not be inclined to be strict about the application of closure to this consequence of what is known.
The “contextualist” response to the general skeptical argument and the closure version is methodist in character while trying to accommodate particularist thrust of the Moorean argument. The contextualist does not dispute the plausibility of the condition on knowledge, e.g. the elimination of alternative possibilities or closure. He recognizes that despite this we are willing to make knowledge attributions on “common sense” grounds. His aim is to show why we do so in spite of adhering to strong conditions for knowledge.
The key issue is whether the condition C required for sufficient warrant is met in a specific case under consideration. Most of the time, we blithely assume that the condition is met. But when confronted with extreme skeptical hypotheses, we give up this assumption, at least for the time being. We are then unwilling to make the attribution, until our attention is turned elsewhere, and the troublesome hypotheses have faded from our minds. So on the contextualist view, we are temporary skeptics.
Compare what contemporary epistemologists say with what Hume wrote about the matter in the Treatise of Human Nature, Book I, Part IV, Section 7. There, he acknowledged that his confidence was shaken when he was confronted with skeptical arguments, and that he had no way to answer them. On the other hand, they had no force over him when he did not have them in mind. So far, he sounds like a contextualist.
But he also recognized that in the long term, the force of the skeptical arguments had to be respected. In this vein, he counseled that philosophy should always proceed upon skeptical principles. And he cautioned the reader that any claims to knowledge or warrant he might make later should be taken with a grain of salt, as expressive of the strong force of what he was thinking at that time.
For the contemporary contextualist, the recognition of the plausibility of the condition on knowledge and its failure to be satisfied does not matter when it is out of mind. It is as if the skeptical hypothesis had never been encountered. Now it may be that this is an accurate description of human behavior in general. But it seems that the contextualist epistemologist ought, like Hume, to be aware of his own epistemic situation. And this seems to indicate that he, like Hume, should be committed to skepticism while recognizing that he often lapses into ordinary modes of knowledge attribution.
Externalism and Skepticism
Externalist accounts of knowledge allow that S can know that p on the basis of considerations that which lie outside S’s
logical space of reasons and thus may be inaccessible to S. This has consequences for skeptical arguments based on internal indistinguishability. These are the standard arguments from Descartes, noted above, that depend on claims that one cannot distinguish a waking state from a vivid, coherent dream, or a perception of a physical object from a state induced by an evil demon or a computer. Let us call the former states “veridical” and the latter “non-veridical.”
If the mechanism that produces the belief that p is real psycho-physical sense-perception, and real psycho-physical sense-perception is reliable, virtuous, or functioning properly, then S knows that p when p is true. Beliefs that are formed on the basis of other processes, such as the imagination in dreaming, or by computer stimulation, presumably are not reliable, but that does not matter when they are not used in the formation of the belief that p.
This way of undermining skeptical arguments can be criticized on the grounds that it leaves S in the dark as to which process is actually the one involved in producing the belief. S is left wondering whether the belief was produced by a real psycho-physical perceptual process or by a process whose misleading output is indistinguishable from that of a real psycho-physical perceptual process. (Barry Stroud has advanced this objection a number of times.)
Another objection is that the externalist condition on knowledge just described is flawed. According to this objection, sense-perception (at least in many cases) does not induce belief on its own. Instead, it produces a mental representation. Belief is produced when the mental representation is taken to be about a real physical object. That there is such an extra step in the process of producing perceptual belief can be seen from our failure to believe when our mental representations are irregular in some way, or when we have information about some abnormality in our perceptual process.
If this psychologically naïve description of the formation of perceptual belief is correct, then there is a suspicion that the process as described is not reliable because of the indistinguishability of veridical from non-veridical mental representations. If the process is not reliable, then S lacks knowledge of what he believes as the end-result of the process.
One response is to relativize reliability to an environment. It could be held that our process of forming perceptual beliefs is reliable in the environment of a physical world. If our circumstances find us in this environment, we do a good job of avoiding beliefs based on illusion, hallucination, dreaming, etc., and this is all the reliability that is required for knowledge. If we abstract from that environment to a more general space of possibilities including a physical world and a non-physical “demon world,” we might be said to be unreliable. The other possibilities in the “demon world” are not “relevant” to whether one knows in a physical world. This is the “relevant alternatives” response to skepticism.
There are at least two problems with this response. The first problem was discussed in an earlier module. The issue is how to specify what makes an alternative relevant. Suppose that I am unable to distinguish twin brothers Hassan and Ali. If they are both in the same room with me, the process by which I form the belief that I am talking to Hassan is unreliable. But what about Ali’s presence in another room? In another building? In another city? In another country? On another continent? The farther away Ali is, the more reliable becomes my belief that I am talking to is Hassan, but Where is the line to be drawn? Similarly, we can ask why a demon world is too “far out” to be relevant to the reliability of our process for forming perceptual beliefs.
The second problem is the same as that posed earlier in the present module. Given that the environment in which we form our perceptual beliefs is the actual world, then the mechanism forming them is reliable. But then the question arises as to whether the actual environment is a physical-world environment. From the first-person point of view, we are once again left wondering whether we have perceptual knowledge.
We have seen how contemporary epistemologists have ignored, dismissed, contextualized, or evaded skepticism. Very few have taken on Descartes’s challenge of meeting it head on. It is generally conceded that there is no conclusive proof that skepticism is false. So those who do try to take skepticism head-on try to show at least that there are good, though inconclusive, reasons to think that we are fallibly warranted in believing propositions about the external world, and, given fallibilism, that skepticism is false.
The most common approach is to argue that the existence of a physical world is the best explanation for the pattern of our perceptions. (For a classical example, see Chapter II of Bertrand Russell’s The Problems of Philosophy.) As we have seen in a previous module, one can hold that being the best explanation confers warrant on our beliefs.
BonJour’s Inference to the Best Explanation
Laurence BonJour has given an argument for why the existence of a physical world is the best explanation for the pattern of our perceptions (BonJour and Sosa, Epistemic Justification). Here we will sketch BonJour’s argument. It begins with a description of the structure of sense-experience. (BonJour adapts the description given by H.H. Price in his 1950 book Perception.) The key feature of the description has to do with three-dimensionality. Next he argues that this feature of sense-experience would be explained if it were the product of a three-dimensional world with a structure very much like what we find in our sense-experience.
It remains to show that this is the best explanation. It is taken to be obvious that this explanation is better than one in which the relevant features of sense-experience occur by brute chance. Moreover, it is hard to see how there could be a better explanation that involves a world with another number of dimensions, or a three-dimensional world with a structure unlike that which we find in our sense-experience.
Any other type of explanation (such as one involving God or an evil demon) would work only by describing a means for simulating the appearance of a three-dimensional world resembling the structure of our sense-experience. BonJour argues that it would be arbitrary to prefer the explanation by a simulating process. Moreover, we can only understand the simulation itself in terms of how a three-dimensional world would produce the pattern of our perception.
Lehrer’s Argument from Coherence
Keith Lehrer mobilizes his coherence account of warrant to reject skepticism. His claim, more specifically, is that subject S can know that skeptical hypotheses are false. For S to be warranted in accepting that p (for any p), it is sufficient that p be more reasonable for S to accept than it is to accept any objection to p. That one lives in a demon world constitutes an objection to the claim that one lives in a physical world, since the two propositions cannot both be true. So to be warranted in accepting that one lives in a physical world, it is enough for it to be more reasonable for S to accept that proposition than to accept the skeptical objection.
Relative reasonableness, for Lehrer, depends on the system of what S accepts. So the question is whether the rest of S’s acceptance system makes S’s acceptance that he lives in a physical world more reasonable than acceptance of the skeptical objection. It is pretty obvious that this is the case if we can appeal to the vast bulk of our acceptance system that implies that a physical world exists.
But this approach seems too easy. Appealing to acceptances about physical objects, which imply that physical objects exist, seems to be the worst kind of circular reasoning. So we should see what happens if appeal to those of our acceptances that do not imply that physical objects exist. This would mostly consist in what we accept about the way things appear to us.
Yet if the relevant part of the acceptance system is reduced in this way, we would have to find something in the system that connects appearances to a physical world. The most plausible candidate is the acceptance that the pattern of our appearances is best explained by being caused by physical objects. If this is right, then Lehrer’s account makes no real advance over other appeals to best explanation.
It can be objected that even if our acceptance system contains what is needed to make our acceptance of the existence of physical objects more reasonable than any objection, this does not by itself confer warrant. For the supporting acceptances might not themselves be warranted.
Lehrer tries to cover this objection by requiring, for the warrant of anything one accepts, an acceptance of one’s own trustworthiness with regard to whatever one accepts. He claims that this “principle of trustworthiness” is itself warranted because it applies to itself. It can be argued on the contrary that it is better to think of this acceptance as warranted because it is supported by other things a person accepts.
Even if we grant that one can be warranted in BonJour’s or Lehrer’s way in accepting that a physical world exists, there is a further question of whether one’s warrant is sufficient and one knows that it is. Lehrer concludes that we at least can know it, given that the other conditions for knowledge are satisfied. But he concedes that we are unable to determine in some strong sense whether we know that a physical world resembling our sense-experience exists.
It appears that the prospects for a wholly satisfying response to skepticism about the external world are dim. There are many approaches that have their appeal as well as their shortcomings. Each of these approaches reflects a different perspective on the attribution of knowledge. Perhaps many approaches are satisfying enough from one perspective but useless from another. And perhaps the shortcomings of all the approaches from a comprehensive perspective is just another reason to abandon the search for a comprehensive account of knowledge.
[ Previous Module | Assignment Page | Course Web Page ]
[ G. J. Mattey’s Home Page | UC Davis Philosophy Department Home Page | UC Davis Home Page ]
This page and all subordinate pages copyright © G. J. Mattey, 2004, 2010, 2011, 2014, 2015. All rights reserved.