Philosophia

, Volume 40, Issue 3, pp 563–575

The Argument from Skepticism for Contextualism

Authors

    • Department of PhilosophyEast Carolina University
Article

DOI: 10.1007/s11406-011-9338-1

Cite this article as:
Newhard, J. Philosophia (2012) 40: 563. doi:10.1007/s11406-011-9338-1
  • 135 Views

Abstract

Epistemic contextualism was originally motivated and supported by the response it provides to skeptical paradox. Although there has been much discussion of the contextualist response to skeptical paradox, not much attention has been paid to the argument from skepticism for contextualism. Contextualists argue that contextualism accounts for the plausibility and apparent inconsistency of a set of paradoxical claims better than any classical invariantist theory. In this paper I focus on and carefully examine the argument from skepticism for contextualism. I argue not only that the prima facie advantage of contextualism is specious, but also that contextualism is in fact at a competitive disadvantage with respect to two classical invariantist views. I also argue that contextualism takes an arbitrary and unsatisfying strategy in its response to skepticism. That contextualism is alone in taking this arbitrary strategy marks a second competitive disadvantage for it. In addition, I argue that the contextualist response to skeptical paradox regenerates a skeptical paradox which contextualism is powerless to solve. Consequently, the argument from skepticism for contextualism fails. Furthermore, this feature of the contextualist response to skeptical paradox completely undermines the motivation and support for contextualism deriving from its treatment of skeptical paradox. I conclude that the argument from skepticism for contextualism fails, and that the contextualist response to skeptical paradox fails to motivate contextualism, pending the success of another argument for the contextualist thesis.

Keywords

ContextualismEpistemic contextualismDeRoseSkeptical paradoxSkepticism

I

The contextualist thesis in epistemology is the claim that the semantic content of a knowledge sentence (i.e., a sentence containing the knowledge predicate) is context-dependent due to the knowledge predicate.1 There are three main arguments for the contextualist thesis: the argument from skepticism, the ordinary language argument, and the argument from the knowledge account of assertability. Originally, contextualism was motivated by the response it provides to skeptical paradox. As Keith DeRose observes: “Contextualist theories of knowledge attributions have almost invariably been developed with an eye toward providing some kind of answer to philosophical skepticism.” (DeRose (1995) p 4) The contextualist argues that the response it provides to skeptical paradox is superior to classical invariantist responses because “our ordinary claims to know can be safeguarded from the apparently powerful attack of the skeptic, while, at the same time, the persuasiveness of the skeptical argument is explained.” (DeRose (1995) p 5) This argument for contextualism is an abductive argument according to which contextualism accounts for the plausibility and apparent inconsistency of a set of epistemic claims better than any classical invariantist theory. The principal advantage of contextualism is that it does not require rejecting outright any of the claims jointly comprising the paradox.

More recently, contextualists have adduced linguistic considerations to support contextualism. Chief among them is DeRose: “The best grounds for accepting contextualism come from how knowledge-attributing (and knowledge-denying) sentences are used in ordinary, non-philosophical talk: What ordinary speakers will count as ‘knowledge’ in some non-philosophical contexts they will deny is such in others.” (DeRose (2009) p 47) 2 According to the ordinary language argument, contextualism offers the best explanation for such pairs of apparently contradictory utterances being true.

The third main argument for contextualism is the argument from the knowledge account of assertability. There are two stages to this argument. The first stage begins with the knowledge account of assertability, that if “the standards for when one is in a position to warrantedly assert that p are the same as those that constitute the truth-condition for ‘I know that p’, then if the former vary with context, so do the latter.” (DeRose (2009) p 106) Together with the claim that warranted assertability is context-sensitive, it is argued that the truth conditions of a knowledge sentence vary with context. In the second stage of the argument, contextualists argue that the context sensitivity of the truth conditions of a knowledge sentence is best explained by the contextualist thesis.

Although there has been much discussion of the contextualist response to skeptical paradox, not much attention has been paid to the argument from skepticism for contextualism. In this paper I focus on and carefully examine this argument. In section II, I present the contextualist response to skeptical paradox, and the contextualist’s abductive argument that the response provided by contextualism is superior to any classical invariantist response. In section III, I argue that the prima facie advantage of contextualism is specious, and in section IV I argue that contextualism is in fact at a competitive disadvantage with respect to two classical invariantist views. In section V, I argue that contextualism takes an arbitrary and unsatisfying strategy in its response to skepticism. That contextualism is alone in taking this arbitrary strategy marks a second competitive disadvantage for it. In section V, I also argue that the contextualist response to skeptical paradox regenerates a skeptical paradox which contextualism is powerless to solve. This feature of the contextualist response to skeptical paradox completely undermines the motivation for contextualism deriving from its treatment of skeptical paradox.

II

Contextualists have argued that the response contextualism provides to skeptical paradox is superior to those provided by any classical invariantist theory. Specifically, contextualism purports to best explain the plausibility and apparent inconsistency of three highly plausible epistemic claims jointly comprising a skeptical paradox.3 For example, consider H1 and H2:
  1. H1

    I know that I have hands.

     
  2. H2

    I do not know that I am not an artificially stimulated brain in a vat.

     
H1 is an ordinary attribution of empirical knowledge, and as such is highly plausible. Difficulties in refuting the brain-in-a-vat hypothesis show that H2 is also highly plausible. Yet H1 and H2 appear to be jointly inconsistent with the highly plausible H3:
  1. H3

    If I do not know that I am not an artificially stimulated brain in a vat, then I do not know that I have hands.

     
The skeptical paradox may be generalized as follows: (Cf. Feldman (1999))
  1. K1

    I know some ordinary empirical propositions.

     
  2. K2

    I do not know that no skeptical scenarios obtain.

     
  3. K3

    If I do not know that no skeptical scenarios obtain, then I do not know any ordinary empirical propositions.

     

As with H1–H3, each of K1–K3 is highly plausible. The contextualist points out that any classical invariantist position, whether skeptical or non-skeptical, must reject one of K1–K3; however, it seems that a classical invariantist will be unable to explain why K1, or K2, or K3, is highly plausible when in fact it is false. By contrast, the contextualist offers an explanation as to why each of K1–K3 is highly plausible.4 K1 is highly plausible because it is true in contexts with ordinary epistemic standards; K2 is highly plausible because it is true in contexts with fairly high epistemic standards; and K3 is highly plausible because it is true “regardless of what epistemic standard it’s evaluated at”.5 Nevertheless, K1 is false in contexts with fairly high epistemic standards, and K2 is false in contexts with ordinary epistemic standards. According to contextualism, K1 and K2 differ in truth value between contexts with ordinary epistemic standards and contexts with fairly high epistemic standards because the semantic content expressed by sentence tokens of these types is sensitive to the epistemic standards of the context. Tokens of K3 express different propositions in contexts with different epistemic standards, though each is true. With respect to a given set of epistemic standards, a contextualist can identify which claim to reject, and can also explain why that claim is plausible by adducing contexts in which that claim is true. Furthermore, there is no single set of epistemic standards with respect to which all of K1–K3 are true, and so there is no single context in which K1–K3 constitute a paradox.6 Whereas classical invariantism is forced to identify one of K1–K3 as false despite being highly plausible, contextualism treats K1–K3 with more parity: for K1 and K2, there are contexts with respect to which it is true (because the proposition it expresses in that context is true) and contexts with respect to which it is false (because the proposition it expresses in that context is false) while K3 is true with respect to any context. The high plausibility of each is accounted for, and none of K1–K3 is rejected outright. Keith DeRose writes that contextualism wins “this battle—that of providing the best resolution to our puzzling conflict of intuitions.”7

Stewart Cohen points out an additional advantage of contextualism: “Because each proposition has independent plausibility, it looks arbitrary and therefore unsatisfying to appeal to any two against the third.” (Cohen (1999) p 63) That is, according to Cohen, not only must a classical invariantist reject one of the three highly plausible propositions, but its strategy in selecting which proposition to reject appears to be arbitrary, and therefore unsatisfying.

III

Contextualists emphasize that these features of contextualism mark significant advantages over classical invariantism, and should appeal to all epistemologists engaged in debate over how best to respond to this skeptical paradox. However, as contextualists now acknowledge, contextualism requires to be forfeited two other intuitions which are both widely shared as well as highly plausible.8 First, and obviously, there is a strong intuition that K1–K3 are paradoxical:
  1. K4

    K1–K3 are paradoxical.

     

A paradox consists of a set of propositions which are jointly inconsistent and each of which is highly plausible.9 The strength of K4 is indicated by its being shared by epistemologists who nevertheless disagree as to which of K1–K3 should be rejected as well as by the tenacity of the skeptical paradox: if K4 were false, it would be very surprising to discover its falsity only after the tremendous amount of attention which has been paid to solving the skeptical paradox.10

Since classical invariantism maintains that K1–K3 are highly plausible and jointly inconsistent, classical invariantism grants K4, regardless of which of K1–K3 is rejected by a particular response to the skeptical paradox.

Contextualists grant that, with respect to a single context, the semantic contents of K1–K3 are inconsistent. However, depending on the epistemic standards of that context, the proposition expressed in that context either by K1 or K2 is neither highly plausible nor true. Therefore, with respect to a single context, K1–K3 are not paradoxical, according to contextualism. On the other hand, a contextualist holds that the proposition expressed by K1 in a context in which there are low epistemic standards, the proposition expressed by K2 in a context in which there are high epistemic standards, and the proposition expressed by K3 in either of these contexts, are all highly plausible, as well as true. However, these propositions are not jointly inconsistent, and so are not paradoxical. Therefore, according to contextualism, K1–K3 are not paradoxical, and K4 is false.11

Second, there is a strong intuition that the semantic contribution due to the knowledge predicate to a knowledge sentence does not vary with context:
  1. K5

    The semantic contribution to a knowledge sentence due to the knowledge predicate does not vary with context.

     

K5, too, is shared by epistemologists who disagree over which of K1–K3 should be rejected, which indicates its strength. By contextualists’ own lights, the knowledge predicate is not obviously context-sensitive, and arguments must be given to overcome the intuition that it is semantically invariant. DeRose writes, “if the contextualist treatment of skepticism is correct, it is only controversially so, and it is far from being clear to all.”12 Thus, one of the principal contextualists concedes that K5 is highly plausible, and yet, according to contextualism, K5 is false.

Naturally, any forfeited intuitions are germane to the contextualist’s argument from skepticism. In aiming to accept all of K1–K3, the contextualist requires the highly plausible K4 and K5 to be forfeited. Once it is clear that contextualism requires highly plausible intuitions to be forfeited, it is also clear that contextualism has no clear advantage over classical invariantism in the argument from skepticism. The principal alleged advantage of contextualism is that it does not require rejecting outright any of the claims jointly comprising the skeptical paradox. However, since it requires rejecting other highly plausible intuitions, the alleged advantage is specious.

It should be emphasized that the issue here is not that contextualism requires two intuitions to be forfeited rather than just one, but that each response to the skeptical paradox requires rejecting a highly plausible intuition. The importance of K4 and K5 to the argument from skepticism is that contextualism requires these highly plausible intuitions to be forfeited, and therefore does not have a clear advantage over classical invariantism.

IV

A viable response to the skeptical paradox must account for two features of the intuition it recommends rejecting: that it is false, and that it is highly plausible. A contextualist or classical invariantist theory may improve its competitive standing by providing an explanation for one or both features which is superior to the explanations provided by its rivals. For example, a skeptical invariantist may adduce skeptical scenarios to justify her claim that K1 is false, and she may point out that she has sensory experiences which constitute evidence supporting beliefs in ordinary empirical propositions and which make K1 highly plausible, but that this evidence is insufficient for knowledge, so that K1 is nevertheless false.13 A non-skeptical invariantist may defend her claim that K2 is false by adducing the modus tollens inference from K1 and K3, her sensory experiences, and the a priori plausibility of K3; and she may admit that the lack of justification for the negation of K2 may contribute to the plausibility of K2, even though it is false. Both the skeptical and non-skeptical invariantist adduce independent considerations in accounting for the falsity and high plausibility of the intuition she recommends rejecting.14

The contextualist response to skeptical paradox requires rejecting two intuitions, K4 and K5. Contextualism has a ready explanation based on independent considerations for her claim that K4 is highly plausible but false.15 However, this is not so for K5. The contextualist’s main explanation for the falsity of K5 appeals to semantic blindness: “What we fail to realize, according to the contextualist solution, is that the skeptic’s present denials that we know various things are perfectly compatible with our ordinary claims to know those very propositions.” (DeRose (1995) p 5) In other words, the contextualist’s explanation is that we simply do not realize that the semantic content of a knowledge sentence is context-dependent due to the knowledge predicate. While this explanation may be criticized on a number of grounds, what I wish to point out here, in order to contrast it with the explanations given by skeptical and non-skeptical invariantism for the falsity and high plausibility of the intuition each recommends rejecting, is that the contextualist’s explanation presupposes the contextualist thesis. Whereas the skeptical and non-skeptical invariantist adduce independent considerations in accounting for the falsity and high plausibility of the intuition she recommends rejecting, the contextualist’s explanation for rejecting K5 appeals to no independent considerations. Although it is legitimate for contextualism to adduce the contextualist thesis to explain the falsity and high plausibility of K5, the pair of explanations given by contextualism for the falsity and high plausibility of the intuitions it recommends rejecting, sc., K4 and K5, is inferior to the explanations given by skeptical and non-skeptical invariantism, since the explanation for the falsity of K5 appeals to no independent considerations. Consequently, contextualism is at a competitive disadvantage in the argument from skepticism.16

As above, DeRose concedes that K5 is highly plausible.17 It seems that a contextualist could explain the high plausibility of K5 simply by granting that it is a widely held belief among ordinary, competent speakers that the semantic content of a knowledge sentence does not depend on the context; at least, not due to the knowledge predicate. This explanation for the high plausibility of K5 appeals to an independent consideration.18 However, the contextualist’s explanation for the falsity of K5 appeals to no independent consideration.

A contextualist might adduce ordinary linguistic behavior as an independent consideration to explain the falsity of K5. Notice that since K5 is the negation of the contextualist thesis, this amounts to giving the ordinary language argument for contextualism. Even supposing that the ordinary language argument is successful in establishing the contextualist thesis, there is still a problem with this maneuver for the argument from skepticism. Since the success of the argument from skepticism would depend on the ordinary language argument, the argument from skepticism would not establish the contextualist thesis independently, that is, without another, independent argument which is itself sufficient for establishing the contextualist thesis. Thus, the argument from skepticism is superfluous to establishing the contextualist thesis, and contextualism remains inferior to skeptical and non-skeptical invariantism in the argument from skepticism.

Similarly, a contextualist might appeal to the knowledge account of assertability as an independent consideration to explain the falsity of K5, but since K5 is the negation of the contextualist thesis, this amounts to giving the argument from the knowledge account of assertability. As with adducing the ordinary language argument, even supposing that the argument from the knowledge account of assertability is successful in establishing the contextualist thesis, there is still a problem with this maneuver. Since the success of the argument from skepticism would depend on the argument from the knowledge account of assertability, the argument from skepticism would not establish the contextualist thesis independently, that is, without another, independent argument which is itself sufficient for establishing the contextualist thesis. Thus, the argument from skepticism is superfluous to establishing the contextualist thesis, and contextualism remains inferior to skeptical and non-skeptical invariantism in the argument from skepticism.

It is not easy to see what other independent consideration a contextualist might find to explain the falsity of K5. It may be hoped that independent tests may be adduced to determine whether the knowledge predicate is in fact context sensitive. Herman Cappelen and Ernie Lepore have developed three intuitive tests for the context sensitivity of an expression. (Cappelen and Lepore (2005) chapter 7) However, since the knowledge predicate fails all three tests, they do not aid the contextualist’s case in the argument from skepticism.19

The skeptical and non-skeptical invariantists adduce independent considerations in accounting for the falsity and high plausibility of the intuition she recommends rejecting. The contextualist has an explanation based on independent considerations for her claim that K4 is false yet highly plausible. A contextualist may explain the plausibility of K5 simply by granting that it is a widely held belief among ordinary, competent speakers that the semantic content of a knowledge sentence does not depend on the context; at least, not due to the knowledge predicate. While this explanation appeals to an independent consideration, the only explanations available for the falsity of K5 which appeal to independent considerations are independent arguments for the contextualist thesis. Therefore, since the skeptical and non-skeptical invariantists adduce independent considerations in accounting for the falsity and high plausibility of the intuitions she recommends rejecting, while the contextualist does not, contextualism is at a competitive disadvantage with respect to two classical invariantist views in the argument from skepticism.

V

Two further objections may be raised against contextualism’s argument from skepticism. The first objection is that contextualism’s strategy in the argument from skepticism is arbitrary and unsatisfying. That contextualism is alone in taking this arbitrary strategy also marks a competitive disadvantage for it in the argument from skepticism. Compare the support available to a skeptical invariantist, a non-skeptical invariantist, and a contextualist for rejecting the intuition she recommends rejecting in her response to the skeptical paradox. A skeptical invariantist emphasizes the plausibility of K2, and argues that K1 is the least plausible of K1–K3, even if it is highly plausible. Presumably, a skeptical invariantist would regard K1 as less plausible than K4 or K5, given that both intuitions are shared with epistemologists who disagree with her over which of K1–K3 to reject. Similarly, a non-skeptical invariantist emphasizes the plausibility of K1, and argues that K2 is the least plausible of K1–K3, even if it is highly plausible. Presumably, a non-skeptical invariantist would regard K2 as less plausible than K4 or K5, given that both intuitions are shared with epistemologists who disagree with her over which of K1–K3 to reject.

A contextualist emphasizes the plausibility of K1–K3, and argues that K4 and K5 should be rejected. Unlike the skeptical invariantist and non-skeptical invariantist, to whom are available arguments that one of K1–K5 is less plausible than the other four, the contextualist does not argue that K4 and K5 are the least plausible of the five. The argument offered by the contextualist is that K1–K3 should be accepted and K4 and K5 rejected because K1–K3 are all highly plausible. But since K4 and K5 are highly plausible, too, this strategy has the very flaw which Cohen cites in attempting to motivate contextualism: it looks arbitrary and therefore unsatisfying to appeal to the high plausibility of any three as a reason for rejecting the other two. Thus, there is an objection to contextualism that its response to the skeptical paradox is arbitrary and therefore unsatisfying. That contextualism is alone in taking this arbitrary strategy also marks a competitive disadvantage for it in the argument from skepticism.

Discovering the second objection requires a brief discussion about paradoxes and their solutions. A paradox consists of a set of propositions which are jointly inconsistent and each of which is highly plausible.20 A straightforward response to a paradox is to identify one member proposition as the least plausible, and reject it. This is precisely what each invariantist response does, though of course there is disagreement among them over which proposition is the least plausible. As was noticed early in the debate over contextualism, a contextualist solution to skeptical paradox is a completely general solution which may be applied to any philosophical paradox. (Cf. Feldman (1999) section III) Although there has been much discussion of the contextualist response to skeptical paradox, it has not been made explicit which claims are to be included in the contextualist response to skeptical paradox.21 Naturally, it contains the contextualist thesis that the semantic content of a knowledge sentence is context-dependent due to the knowledge predicate. If the contextualist response to skeptical paradox includes only the contextualist thesis, then the solution seems inadequate, because it is too general. To be adequate, a contextualist solution to a philosophical paradox needs a reason why a contextualist solution is called for to solve it, rather than a straightforward, invariantist response. For the skeptical paradox, that reason is that K1–K3 are equally highly plausible.22

Thus, the contextualist’s response to skeptical paradox depends on K1–K3 being not merely highly plausible, but equally highly plausible. Otherwise, if K1–K3 are all highly plausible but not equally highly plausible, it may be argued of one of K1–K3 that it is less plausible than the others, and that the skeptical paradox is best solved by rejecting that proposition. This is precisely what the skeptical invariantist and non-skeptical invariantist do. In fact, the claim that K1–K3 are equally highly plausible would be very surprising and difficult to defend, since all other epistemologists engaged in the debate agree that K1–K3 are not equally highly plausible, despite being highly plausible. Instead, the other epistemologists disagree about which proposition is the least plausible.23

A contextualist is likely to defend the claim that K1–K3 are equally highly plausible by pointing out that K1, K2, and K3 seem equally highly plausible when considered individually. That is, a contextualist may insist that 〈K1〉LOW—the proposition expressed by K1 in a context with ordinary (LOW) epistemic standards—has the same high plausibility as 〈K2〉HIGH and 〈K3〉LOW (or 〈K3〉HIGH). However, these three propositions can be evaluated as equally highly plausible only by evaluating them according to the same standards. In evaluating the plausibility of these three propositions, the germane standards are epistemic standards, since the propositions expressed by K1–K3 are epistemic claims. So, these three propositions can be assessed as equally highly plausible only by appealing to the same set of epistemic standards. Evaluating 〈K1〉LOW, 〈K2〉HIGH, and 〈K3〉LOW according to ordinary epistemic standards (those of LOW) does not yield that these three propositions are equally highly plausible. Nor does evaluating these three propositions according to the epistemic standards of HIGH. In fact, even according to contextualism there is no single set of epistemic standards on which all three propositions are equally highly plausible. Consequently, according to contextualism, there is no epistemic standard on which the propositions expressed by K1–K3 are equally highly plausible.24

Worse, if there were a single epistemic standard on which the propositions expressed by K1, K2, and K3 in a given context (not necessarily the same for all three) were equally highly plausible, it would be because the epistemic standard of the contexts for all three propositions is the same. But then the semantic contribution made by the context due to the knowledge predicate would be the same for all three propositions. Consequently, the three propositions would be inconsistent, and would comprise a skeptical paradox. Further, since contextualism would grant that the three propositions are inconsistent and equally highly plausible, contextualism would be powerless to solve that paradox. Thus, if there is a single epistemic standard on which the propositions expressed by K1–K3 are equally highly plausible, it follows that those propositions are jointly inconsistent and paradoxical, and that contextualism is powerless to resolve the inconsistency, and is therefore powerless to solve the skeptical paradox.

As above, to be adequate, a contextualist solution to a philosophical paradox needs a reason why a contextualist solution is called for to solve it, rather than a straightforward, invariantist response. For the skeptical paradox, the reason is that the propositions expressed by K1–K3 are equally highly plausible. Since epistemic contextualism can maintain that the propositions expressed by K1–K3 are equally highly plausible only on pain of inconsistency and paradox, the argument from skepticism for contextualism fails.

VI

I have argued that the prima facie advantage of contextualism is specious, because, like skeptical and non-skeptical invariantism, its response to skeptical paradox requires highly plausible intuitions to be forfeited. I also argued that contextualism faces two competitive disadvantages. The first competitive disadvantage is that, unlike skeptical and non-skeptical invariantism, there seem to be no independent considerations available for contextualism to explain the falsity and high plausibility of the intuitions it recommends rejecting. The second competitive disadvantage is that contextualism is alone in taking an arbitrary and unsatisfying strategy in rejecting the intuitions it recommends rejecting, sc., K4 and K5. Finally, I argued that, to be adequate, a contextualist solution to a philosophical paradox needs a reason why a contextualist solution is called for to solve it, rather than a straightforward, invariantist response. For the skeptical paradox, that reason is that the propositions expressed by K1–K3 are equally highly plausible. I argued further that maintaining that the propositions expressed by K1–K3 are equally highly plausible regenerates a skeptical paradox which contextualism is powerless to solve. Consequently, the argument from skepticism for contextualism fails.

Nevertheless, if another argument succeeds in establishing the contextualist thesis, then the contextualist response to skepticism is available on that basis, and no additional reason is called for. Hence, the success of the contextualist response to skepticism depends on the success of the ordinary language argument, the argument from the knowledge account of assertability, or some other argument for the contextualist thesis. Although the contextualist response to the skeptical paradox is available pending the success of one of these other arguments, the argument from skepticism for contextualism fails.

The response to skeptical paradox greatly motivates contextualism in addition to being the explanation contextualism appeals to in the argument from skepticism. Consequently, the failure of the contextualist response to the skeptical paradox, pending the success of another argument for the contextualist thesis, not only results in the failure of one of the three main arguments for contextualism, but it also completely undermines the motivation for contextualism deriving from its response to skeptical paradox.

Footnotes
1

It is potentially misleading to formulate the contextualist thesis as the claim that the knowledge predicate is a context-sensitive expression, since a contextualist may hold that it is an unarticulated constituent semantically associated with the knowledge predicate, but not the semantic content of the knowledge predicate itself, which is context-dependent. Thus, contrastivism is a species of contextualism, even though according to contrastivism the knowledge predicate is not a context-sensitive expression. For various versions of the contextualist thesis, see Schaffer (2004) p 73; Stanley (2005) p 16; Fantl and McGrath (2009) p 30; and Rysiew (2009).

 
2

Cf. DeRose (2005) p 172; DeRose (2006) Introduction; and Cohen (2000) p 96 f.

 
3

Similar arguments can be run by appealing to epistemic claims comprising other epistemic paradoxes, such as the lottery paradox. For discussion, see DeRose (1996); Cohen (1998); Nelkin (2000); and Hawthorne (2004).

 
4

What follows is the explanation given by contextualists who hold that knowledge sentences are sensitive to the epistemic standards of the context. Contrastivism offers a different explanation of the plausibility of K1–K3, but it, too, is one according to which there is no single context in which all three of K1–K3 are true. According to contrastivism, there are contexts in which each of K1–K3 is true, and the latter feature accounts for their plausibility. See Schaffer (2004). Other versions of contextualism have been proposed according to which the epistemic standards are determined at least in part by the speaker’s intentions (Montminy (2007) and Montminy, “The Role Of Context In Contextualism” (forthcoming)), pragmatic presuppositions (Blome-Tillman (2009)), and ascriber interests and purposes (McKenna (2011)). These three views gives the same explanation as in the main text.

 
5

DeRose (1995) p 39. DeRose is commenting on his premise “2. If I don’t know that not-H, then I don’t know that O.” where H is a skeptical hypothesis, and O is an ordinary proposition about the external world. The full sentence is: “Our verdict regarding (2) is that it’s true regardless of what epistemic standard it’s evaluated at, so its plausibility is easily accounted for.”

 
6

A case where two speakers disagree as to whether a subject has knowledge may be regarded as taking place within a single context. It may seem that a skeptical paradox could arise in such a context due to the presence of both sets of epistemic standards in this context. However, there are various conversational mechanisms available to a contextualist for handling cases of disagreement among conversational participants which prevent such a conversation leading to paradox. For a discussion of these options, see DeRose (2009) chapter 4.

 
7

DeRose (1995) pp 49–50. Stewart Cohen also makes this point: “One strength of the contextualist approach is that it can account for the truth of our everyday knowledge ascriptions while still explaining the force of skeptical arguments.” (Cohen (1998) p 291)

 
8

The error theory objection has been raised by quite a few philosophers, beginning with Schiffer (1996). The objection is addressed by Cohen (1999); Cohen (2001); DeRose (1999); and DeRose (2004).

 
9

According to a standard definition of a paradox, “a paradox arises when a set of individually plausible propositions is collectively inconsistent.” (Rescher (2001) p 6; his italics) Strictly speaking, the definition of ‘paradox’ must restrict membership in the set of paradoxical propositions to all and only those propositions which are required to generate the inconsistency.

 
10

It should be emphasized that K4 is not the weaker proposition that K1–K3 seem paradoxical, but that they are paradoxical. The weaker proposition, that K1–K3 seem paradoxical, is compatible with contextualism, and need not be rejected by a contextualist. Observing that K4 is strong and highly plausible does not beg the question against contextualism, since a contextualist may argue that K4 should be rejected, or may attempt to explain it away.

 
11

Although contextualism is committed to rejecting K4, contextualism may account for the high plausibility of K4 as follows. Regarding K1 and K2, there are contexts with respect to which the propositions they express in those contexts are true. This accounts for their plausibility. K3 is true with respect to any context, which accounts for its plausibility. The propositions expressed by K1–K3 in a single context are jointly inconsistent. Thus, contextualism can account for the plausibility of K4, but must reject it.

 
12

DeRose (2004) p 37. Elsewhere, DeRose writes: “Insofar as we do have this belief, that the conditions for truly saying that someone knows do not depend on the sorts of contextual factors we have been discussing, then contextualism goes against at least one of our beliefs.” (DeRose (1992) p 927; his italics)

 
13

In what follows, I abbreviate “skeptical classical invariantist” to “skeptical invariantist” since the invariantist views being discussed are restricted to varieties of classical invariantism. Similarly, I abbreviate “non-skeptical classical invariantist” to “non-skeptical invariantist.” I give the name “non-skeptical invariantist” to a philosopher who rejects K2, without wishing to imply that the only way for an invariantist to resist skepticism is to deny K2. I also do not wish to suggest that a non-skeptical invariantist must admit that the only justification available for her belief that no skeptical scenarios obtain is that conferred by inference from K1 and K3.

 
14

A classical invariantist inclined to deny the closure principle might marshal her argument for rejecting closure in order to argue that K3 should be rejected. Some, though not all, of the standard arguments against the closure principle are independent of skeptical paradox. (Cf. Luper (2010)) If these arguments are given to support rejecting K3, then the points made regarding skeptical and non-skeptical invariantism apply to closure-denying invariantism as well. Since the competitive advantages of classical invariantism are not augmented by considering closure-denying invariantism, I do not discuss it in the text.

 
15

Cf. footnote 11: there are contexts with respect to which the propositions expressed by K1 and K2 are true, which accounts for their plausibility. K3 is true with respect to any context, which accounts for its plausibility. The propositions expressed by K1–K3 in a single context are jointly inconsistent. Although there are no three contexts (not necessarily different) with respect to which K1–K3 are highly plausible and jointly inconsistent, these features may be adduced to explain why K4 is highly plausible yet false.

 
16

The non-skeptical invariantist adduces one independent consideration and one uncontroversial consideration to meet her burden of proof. Should the uncontroversial consideration be viewed as controversial, non-skeptical invariantism would be at a competitive disadvantage with respect to skeptical invariantism to the extent of the controversy.

 
17

Cf. footnote 12 and the quoted sentence it is appended to.

 
18

I am grateful to an anonymous referee for pressing this point.

 
19

The knowledge predicate’s failure of all three tests indicates the intuitive strength of K5, and, consequently, that contextualism incurs a burden to argue for the contextualist thesis. Nevertheless, it should be noted that there are grounds for disputing the legitimacy of the tests, including that they beg the question. For a critical discussion of the three tests, see Montminy (2006) especially §§ 5–7; Hawthorne (2006); and Stanley (2005) pp 49–51.

 
20

According to a standard definition of a paradox, “a paradox arises when a set of individually plausible propositions is collectively inconsistent.” (Rescher (2001) p 6; his italics) Strictly speaking, the definition of ‘paradox’ must restrict membership in the set of paradoxical propositions to all and only those propositions which are required to generate the inconsistency.

 
21

Some other claims which might be included concern how epistemic standards are determined for a given context, and how contexts and standards change during discourse. These issues may be set to the side for the purpose of evaluating the argument from skepticism for contextualism.

 
22

Cohen comes close to making this explicit when he argues that “Because each proposition has independent plausibility, it looks arbitrary and therefore unsatisfying to appeal to any two against the third.” (Cohen (1999) p 63) Although he writes that each proposition has “independent plausibility”, it is only arbitrary and unsatisfying if the three propositions are equally plausible.

 
23

It may also be difficult for a contextualist to defend the claim that K1–K3 are equally highly plausible if K3 is taken to be more plausible than K1 or K2, since it is highly plausible in every context.

 
24

A contextualist might try to maintain that K4 and K5 are less plausible than any of K1–K3, and should be rejected, whether or not K1–K3 are equally highly plausible. There are two problems with this response. First, it is simply not very plausible, even by the contextualist’s own lights, to claim that K4 and K5 are less plausible than any of K1–K3; otherwise, this solution to the skeptical paradox would have gained currency long ago. Second, to claim that K5, which is the negation of the contextualist thesis, is less plausible than any of K1–K3, begs the question in the argument from skepticism.

 

Acknowledgements

My thanks to Ray Elugardo, Steve Ellis, Harry Gensler, Nicholas Georgalis, Steven Hales, Robert Howell, Matthew McGrath, Martin Montminy, Wayne Riggs, Chris Swoyer, and Mike Veber for helpful comments, discussion, and suggestions.

Conflict of Interest

None.

Copyright information

© Springer Science+Business Media B.V. 2011