In recent works, several authors have challenged the coherence of epistemic contextualism.Footnote 1 They hold that assuming the factivity of knowledge and epistemic closure, the contextualist cannot coherently assert, or know, the content of her theory. After presenting the objection in section 1, we will examine three responses we deem unsatisfactory in section 2, and then put forward our own solution in sections 3 and 4. In a nutshell, our solution rejects the idea that the contextualist should be construed as asserting that her theory is true. We also argue that the contextualist framework does not entail that high epistemic standards are in place in a context in which contextualism is presented.
1. THE STATABILITY PROBLEM
Let ‘K(S, h)’ stand for ‘S knows that h,’ where S is a contextualist and ‘h’ denotes the proposition that S has hands. According to contextualism, there are ordinary, low-standards, contexts (hereafter CL) in which ‘K(S, h)’ is true. Now, contextualism holds that there is no such thing as knowing that p, simpliciter. On this view, in uttering ‘K(S, h)’ in CL, S asserts that she knows that h relative to low epistemic standards.Footnote 2 In other words, we have
(1) KL(S, h),
where ‘KL’ stands for ‘knows relative to low epistemic standards.’ Contextualism also holds that ‘K(S, h)’ is false in a skeptical, high-standards, context (hereafter CH). Letting ‘KH’ stand for ‘knows relative to high epistemic standards,’ we thus have:
(2) ~KH(S, h).Footnote 3
Finally, many authors contend that contextualists are committed to holding that ‘K(S, (1))’ is true in CH. In other words, contextualists must hold that
(3) KH(S, (1)).
We will examine why contextualists are thought to be committed to (3) shortly. But let us first see why the triad (1)–(3) is said to be inconsistent. According to factivity,
(4) KL(S, h) → h.
And given that contextualists take themselves to know that factivity holds, regardless of the context in which they are located, we have:
(5) KH(S, KL(S, h) → h).Footnote 4
The principle of epistemic closure entails that:
(6) [KH(S, KL(S, h)) & KH(S, KL(S, h) → h)] → KH(S, h).
From (3), (5) and (6), we obtain:
(7) KH(S, h).
But (7) contradicts (2). Hence, assuming epistemic closure and knowledgeH of factivity, the triad (1)–(3) is inconsistent. The contextualist theory can thus not be coherently stated.
But why are contextualists thought to be committed to (3)? First, many commentators have contended, a context in which the contextualist presents her theory, call this a contextualist context, or CC, has the marks of high-standards contexts. As Anthony Brueckner puts it, ‘Whenever I think about the skeptical possibilities, while in a careful, reflective philosophical context like this [i.e., a contextualist context], it strikes me that I do not know that they do not obtain. That's one of the main motivations for attempting to formulate a successful contextualist theory in the first place’ (Reference Brueckner2004: 404).Footnote 5
From here, there are two routes leading to (3). First, one may hold that the contextualist ought to know that the knowledge claim made in CL is true. As Crispin Wright puts it, ‘For if contextualism cannot rationally profess that knowledge, it has no point to make!’ (Reference Wright2005: 243).Footnote 6 Hence, it must be correct for the contextualist in CC = CH to claim to have the relevant second-order knowledge. KH(S, KL(S, h)), or (3), must thus hold.
The second route invokes the knowledge norm of assertion. Many contextualists, including Keith DeRose and Stewart Cohen, endorse the knowledge account of assertion in its meta-linguistic form, according to which, ‘A speaker, S, is well-enough positioned with respect to p to be able to properly assert that p if and only if S knows that p according to the standards for knowledge that are in place as S makes her assertion’ (DeRose Reference DeRose2009: 99).Footnote 7 But surely the contextualist would hold that the statement of her theory in CH, including her assertion that KL(S, h), or (1), is an epistemically proper speech act. If so, the contextualist is committed to holding that she knowsH that (1), i.e., (3).
Opponents of contextualism may contend that the argument generalizes in various ways. First, the putative knower S in CL need not be the contextualist S* herself. Second, they may add, the two contexts in question need not be the radical skeptic's context and the ordinary, low-standards context: the argument only requires that S be in context CX and S* be in context CY, where CY is more demanding than CX. Finally, the proposition that S has hands can be replaced by any empirical proposition p S* takes S to knowX but does not take herself to knowY. The inconsistent triad can thus be represented as:
(1) KX(S, p);
(2) ~ KY(S*, p);
(3) KY(S*, (1)).
The statability problem thus has many variations, some of which we will consider below.
2. SOME RESPONSES
Before presenting our own response to the statability challenge, we will examine three other responses that are, in our view, unsatisfactory.
2.1. Freitag's Minimal Epistemic Contextualism
According to Wolfgang Freitag (Reference Freitag2011, Reference Freitag2013), the contextualist doctrine entails neither (1) nor (2). Before we explain why, let us first present what Freitag calls the knowability problem. According to Freitag, an adequate theory T should not satisfy the following condition:
(KP) ∃p [T→(p ∧ ~KH(S, p))].Footnote 8
A theory T faces the knowability problem, writes Freitag, if (and only if) KP holds. In such a case, S cannot knowH that T is true. To avoid the knowability problem, he adds, a contextualist theory should be knowableH.Footnote 9 And as we saw in the previous section, a theory that accepts both (1) and (2), as well as factivity and epistemic closure, satisfies KP.
Freitag holds that contextualism can consistently be knownH to be true, and for this reason it does not face the knowability problem. All contextualism is committed to, according to him, is the thesis he calls minimal epistemic contextualism, or MEC:
(MEC) ∃s ∃p ◊(KL(s, p) ∧ ~KH(s, p)).
According to MEC, there is a subject S and a true empirical proposition p such that it is possible for S to knowL, but to fail to knowH, that p. Freitag takes MEC to be a ‘faithful rendering’ (2011: 281, n. 21) of the contextualist position. Clearly, MEC entails neither (1) nor (2). Furthermore, since it does not satisfy condition KP, it avoids the knowability problem.
MEC presents some interpretive difficulties, though.Footnote 10 Let us first assume, as we have so far in this paper, that ‘KX(S, p)’ denotes the proposition that S knows that p relative to epistemic standards X. If MEC is understood this way, then it is clearly too weak. Contextualism is a semantic thesis about English sentences of the form ‘S knows that p.’Footnote 11 On the current reading, MEC is completely silent about this issue, since it does not connect the English word ‘know’ with either KL or KH. In fact, MEC is compatible with the invariantist thesis that the sentence ‘S knows that p’ always expresses the same proposition, regardless of the context in which it is used. Invariantists do not deny the existence of the propositions that S knowsL that p and that S knowsH that p. They may thus grant that a subject S's epistemic position with respect to proposition p could be such that the proposition that S knowsL p is true, but the proposition that S knowsH that p is false. However, the invariantist would add, the English sentence ‘S knows that p’ always expresses the same proposition, i.e., that S knowsL that p or that S knowsH that p,Footnote 12 as the case may be.
Perhaps MEC should be read differently. Maybe ‘KX(S, p)’ should stand for ‘‘S knows that p’ would be true if uttered in CX,' (or more simply ‘‘S knows that p’ is true in CX) and ‘ ~ KX(S, p)’ should stand for ‘‘S knows that p’ is false in CX.' On this reading, which seems to be the one Freitag intends,Footnote 13 we have:
(MEC*) ∃s ∃p ◊(⌜s knows that p⌝ is true in CL ∧ ⌜s knows that p⌝ is false in CH).
But MEC* still involves an interpretive issue. Surely, the thesis is not that it is possible for the string of letters ‘S knows that p’ to have a true content in CL and a false content in CH. The thesis has to be that there is a subject S and a true empirical proposition p such that the English sentence ‘S knows that p’ can possibly be true in CL and false in CH. MEC* should thus be understood as:
(MEC**) ∃s ∃p [⌜s knows that p⌝ is such that ◊(⌜s knows that p⌝ is true in CL ∧ ⌜s knows that p⌝ is false in CH)].
MEC** is clearly incompatible with invariantism.Footnote 14 But it is questionable that a contextualist S can knowH that MEC** is true. The problem is that MEC** is arguably an empirical thesis. Granted, when contextualists defend their view, they typically invoke our judgments about hypothetical cases in which agents are stipulated to satisfy certain epistemic standards. But it does not seem that the argument for MEC** comes solely from the armchair. As DeRose (Reference DeRose2009: ch. 2) makes clear, contextualists also appeal to how ordinary speakers do use ‘know’ in various circumstances. But these do not exhaust the considerations in favor of contextualism, for alternative accounts of the data about our context-sensitive use of ‘know,’ such as those proposed by invariantists and other non-contextualists, must also be ruled out. It is thus highly questionable that only a priori considerations are deployed in defending contextualism, and in particular MEC**. Since MEC** seems to be in the domain of propositions our ‘knowledge’ of which a radical skeptic could target, the contextualist cannot plausibly be credited with knowledgeH of MEC**.Footnote 15
Note that our point is not that MEC** satisfies KP. After all, MEC** does not entail that a contextualist S cannot knowH that MEC** is true. Our point is that although S endorses MEC**, she cannot plausibly claim to knowH that MEC** is true. Given how minimal MEC** is, the requirement that a contextualist theory should be knowableH strikes us as unreasonable.
There is a further problem with KP. A contextualist who thinks her theory should not satisfy KP may well be committed to a kind of pragmatic inconsistency, that is, she may not be able to practice what she preaches. This is because our contextualist S, qua ordinary speaker, will inevitably find herself in ordinary conversational contexts in which assertions and knowledge attributions are made. And like any ordinary speaker, S will accept some of these assertions and attributions. She will thus be committed to the truth of some empirical propositions p. However, S should also grant that she does not knowH that p. Again, perhaps her contextualist point of view, narrowly understood as MEC**, does not force her to make this concession. But it would be extremely odd for a contextualist to maintain that she does knowH that p.Footnote 16 This means that in practice, S is committed to both p and ~KH(S, p), for many propositions p. Trying to avoid such a commitment would put her in a very unsavory position. For this reason, the fact that it satisfies KP should not be considered a strike against a contextualist point of view, whether this point of view is understood narrowly (as MEC**) or not.Footnote 17
2.2 Baumann's Modified Closure Principle
Let us now consider Peter Baumann's (Reference Baumann2008) contention that the triad (1)–(3) is actually consistent. His version of the case involves Mary, a contextualist located in CH,Footnote 18 and Frank, an ordinary speaker in CL. He further stipulates that p is the (true) proposition that there is no life on Mars. Baumann holds that it can be the case that Mary knowsH that Frank knowsL that p, even though Mary does not knowH that p. This is because, he writes, the ‘warrant Mary needs in order to count as a knower about Frank's epistemic situation is not (or usually not) the same kind of warrant as the one she needs in order to count as a knower of the things Frank knows’ (Reference Baumann2008: 591). Baumann contends that to knowH that Frank knowsL that there is no life on Mars, Mary needs to knowH about Frank's epistemic position regarding life on Mars, but she only needs to knowL that there is no life on Mars. This view strikes us as arbitrary: we do not see any reason to hold that in order to count as knowing that ‘S knows that p’ is true, one should satisfy higher standards regarding S's position with respect to p than one should regarding the truth of p. Baumann writes that ‘Mary might have very sophisticated (psychological etc.) knowledge about the rudimentary and primitive nature of Frank's knowledge about astronomy; by no means does it follow that Mary has sophisticated knowledge about astronomy’ (2008: 591). But we fail to see how Mary can have sophisticated knowledge that Frank knows that p without having sophisticated knowledge that p. Frank knowsL that p just in case (i) he believes that p, (ii) he satisfies low epistemic standards with respect to p, and (iii) p is true. Baumann's contention that in order to knowH that Frank knowsL that p, Mary has to knowH (i) and (ii) but not (iii) is both mysterious and ad hoc.
In addition to being unmotivated, Baumann's position is counter-intuitive. Baumann rejects the closure principle in the form that is usually accepted by contextualists, and proposes his own version of the principle instead, which we will call Baumann's closure, or BC:
(BC) For all contexts CX there is a context CY (not more demanding than CX) such that: [A knowsX that p and A knowsX that (p → q)] → A knowsY that q.
According to BC, a subject might not validly infer that she knowsX that q from her knowledgeX that p and that (p → q). This is because, as the case of Mary illustrates, for some propositions p and q, this inference is not valid. Hence, a subject cannot confidently extend her knowledgeX by applying deduction to what she knowsX. In some cases, this extension of knowledgeX is not possible. Which cases? Baumann does not provide a general rule. So, as stated, his BC should make anyone reluctant to infer knowledgeX that q from knowledgeX that p and that (p → q). BC thus fails to vindicate knowledgeX extension, which is traditionally the main idea invoked in support of epistemic closure.
Furthermore, BC does not prohibit what Keith DeRose (Reference DeRose1995) calls abominable conjunctions, that is, claims such as ‘While I don't know that I am not a bodiless brain in a vat, I do know that I have hands’ and ‘Even though I don't know that these are not well disguised mules, I know that they are zebras.’ BC sanctions Baumann's contention that someone could knowX that ‘S knows that p’ is true, without knowingX that the truth condition, i.e., the condition that p is true, is satisfied. This means that a contextualist could truthfully say in CH, ‘While I don't know that I have hands, I know that I knowL that I have hands.’ And while in CH, Mary could truthfully say, ‘Even though I don't know that there is no life on Mars, I know that Frank knowsL that there is no life on Mars.’ These conjunctions strike us as very odd.Footnote 19 For these reasons, it would be wise to seek a solution to the statability problem elsewhere.
2.3 Brueckner and Buford's Rejection of (3)
Perhaps the contextualist should simply reject (3). This is the response that Anthony Brueckner and Christopher Buford (Reference Brueckner and Buford2009) recommend. However, they hold that a contextualist can still knowledgeably state in CH the general contextualist doctrine. Like Baumann, Brueckner and Buford stipulate that the contextualist S* in CH and the putative knower S in CL are two different people—and ‘h’ stands for the proposition that S has hands. Hence, they reject their own version of (3), that is, KH(S*, KL(S, h)). However, Brueckner and Buford write, S* can correctly say in CH, ‘Even though I do not know now that ‘K(S, h)’ is true in CL, I do know that the conditions for the truth of that sentence in CL are satisfied up to the ‘truth condition’ [i.e., the condition that h is true]. So if h is indeed the case, then ‘K(S, h)’ is true in CL' (Reference Brueckner and Buford2009: 437, slightly edited to match the current abbreviations). But this response is problematic, for there are many reasons why S* does not knowH that ‘K(S, h)’ is true in CL, other than the fact that she does not knowH that h. Since the proposition that the standards in place in S's context are low, the proposition that S satisfies low standards with respect to h, and the proposition that ‘K(S, h)’ has context-sensitive truth conditions are all empirical,Footnote 20 S* lacks knowledgeH of these propositions. In fact, S* does not even knowH that S exists! So pace Brueckner and Buford, when in CH, a contextualist cannot correctly say, ‘I know that the conditions for the truth of ‘K(S, h)’ in CL are satisfied up to the truth condition.' Her epistemic position with respect to the proposition that ‘K(S, h)’ is true in CL does not grant her such knowledgeH.
Now, to be fair, these criticisms ignore a crucial feature of the scenario discussed by Brueckner and Buford. They assume that the contextualist context is a ‘demanding’ context rather than a skeptical one. And a demanding context, Brueckner and Buford write, is ‘in between’ the ordinary context CL and the radical skeptical context CH we have considered so far. More specifically, a demanding context CD has the following features: (i) some ‘knowledge’-sentences which are truly uttered in CL come out false when uttered in CD; and (ii) some ‘knowledge’-sentences about the existence of and character of contexts other than CD come out true as uttered in CD (Brueckner and Buford Reference Brueckner and Buford2009: 432, n. 4).
But one wonders why the contextualist context CC should be equated with a demanding context CD. As we presented it in section 1, the statability problem is based on the premise that taking an error possibility seriously automatically raises the standards to a level at which a subject counts as knowing only if she can eliminate this error possibility. We will reject this premise in the next section. However, Brueckner and Buford do not question it. Hence, their assumption that CC = CD is admissible only if no radical skeptical scenarios are considered in CC. Their solution can thus be invoked only by a subset of contextualists, namely those who, in defending their view, remain silent about the skeptic's radical error possibilities, and consider only moderate skeptical scenarios.
A second problem is that it does not seem that even a contextualist in CD completely avoids the statability problem. If the proposition p is one ordinary folks would take themselves to know in CL, then it is hard to see how the contextualist could fail to knowD p, while knowingD the propositions entailed or presupposed by her doctrine. While our contextualist S* can arguably knowD that S exists and that S is located in another conversational context, it is questionable that S* knowsD (i) what epistemic standards are in place in S's context, (ii) what epistemic position S is in with respect to p, (iii) that the content of ‘K(S, h)’ is context sensitive. In other words, it is doubtful that there are scenarios in which S*'s epistemic position with respect to (i), (ii), and (iii) is superior to her epistemic position regarding p, where p is a proposition ordinary speakers readily take themselves to know in CL.
For example, given that you can tell the difference between a zebra and an ill-disguised mule, you are in a reasonably good epistemic position with respect to the proposition that the animal in the pen is a zebra (call this proposition z). By contrast, you are ignorant about the discriminating abilities of the person standing next to you. Would she note the subterfuge if an ill-disguised mule were to be placed in the pen? Your inability to answer this question indicates that you are not well positioned about the epistemic standards she meets regarding proposition z. The problem is that such a scenario seems typical: unlike the proposition p S is taken to knowL, propositions about other people's epistemic positions do not count among those we readily take ourselves to know, even by low epistemic standards. The same point can be made about (iii): although we are contextualists, it would be unwise for us to hold that we are in a better epistemic position with respect to (iii) than we are regarding, say, the proposition that we have hands! There is thus very little of the contextualist doctrine that S* can knowledgeably state while in CD, if it is assumed that S* fails to knowD p.Footnote 21 Given that cases in which S* knowsD that the conditions for the truth of ‘K(S, p)’ in CL are satisfied ‘up to the truth condition’ are at best exceptional, they cannot be used as a basis for a general solution to the statability problem. It is now time to examine our solution to the statability problem.
3. OUR RESPONSE
A key assumption of the argument generating the statability problem is that while in CC, the contextualist asserts the content of her theory. But one may plausibly hold that the contextualist's utterances have a slightly weaker assertoric force than assertions do. Consider the category of illocutionary acts called weak assertives, which includes conjectures, guesses and hypotheses.Footnote 22 These illocutionary acts aim at truth, but their assertoric force is weaker than that of an assertion. On the current proposal, the force of the contextualist's weak assertives would be somewhere in between the force of a conjecture and that of an assertion. Their illocutionary force would be comparable in strength to that of the weak assertives generated by a parenthetical use of ‘I think.’ Consider the following exchange:
George: Where is Denise?
Sally: I think she's in her office. (Or: She's in her office, I think.)
By prefacing her answer by ‘I think,’ Sally conveys that her level of confidence in the proposition that Denise is in her office is lower than what it would have been if she had straightforwardly asserted that Denise is in her office. In other words, her choice of words places less of her credibility on the line. This strikes us as a plausible description of what typical philosophers do when they defend their views, except that they tend to avoid stylistically frowned upon parentheticals.Footnote 23 In a philosophical context, it is understood that many of the claims made are highly controversial and cannot be established decisively. There is thus an implicit understanding that speakers do not represent themselves as knowing the content of every utterance they make. Utterances expressing controversial philosophical views are thus reasonably interpreted as having weaker assertoric forces than assertions do.
Weak assertives are governed not by a knowledge rule, but by a weaker epistemic requirement: a weak assertive that p is epistemically appropriate only if the speaker has some evidence for p. Suppose that p is the proposition that your ticket will not win the lottery. (Assume that the lottery is fair, the odds of winning are low, etc.) Given the knowledge norm, you cannot properly assert that p – even in CL. However, it would be appropriate for you to make the weak assertive that p, by uttering something like ‘p, I think’ or ‘p, it seems,’ or simply ‘p’ with the right tone of voice. This is because you have reasonably good evidence that p, and an epistemically proper weak assertive that p does not require knowledge that p.
How much evidence is required for the epistemic propriety of a weak assertive will of course depend on the strength of the assertive.Footnote 24 One reason to think that in response to George, Sally makes the weak assertive, rather than the assertion, that Denise is in her office, is that it would be odd for George to challenge her utterance by asking ‘How do you know?’ However, George can appropriately ask Sally, ‘What are your reasons for thinking that?’ Similarly, it seems inappropriate to challenge an interlocutor who argues for, say, compatibilism about free will, moral cognitivism or epistemic contextualism, by asking, ‘Do you know that?’ or ‘How do you know?’ This provides further evidence for our claim that the contextualist is not asserting the content of her theory.
Still, philosophers are expected to provide reasons for what they say. Our contextualist should be no different: even though they lack the force of assertions, her assertives should thus be understood as relatively strong. This means that although our contextualist does not represent herself as knowing the content of her theory, she can still make a point (to use Crispin Wright's language), as long as she takes her view to be supported by evidence and arguments.
The statability problem can be avoided by rejecting the assumption that the contextualist is making assertions when she presents her view. We thus reject claim (3), i.e., KH(S, KL(S, h)): S does not assert, but merely makes a weak assertive that ‘K(S, h)’ is true in CL. And that weak assertive is appropriate, provided that S has reasonably good evidence for the proposition that ‘K(S, h)’ is true in CL, with the understanding that reasonably good evidence need not be evidence that would suffice for knowledgeH, or even for knowledgeL.
One might protest that at least some of the contextualist's utterances are assertions. For example, in defending her view, the contextualist appeals to data about speakers' use of ‘know’ in various contexts. Surely, she at least asserts that there are speakers other than her. And arguably, she asserts that speakers are reluctant to claim that they ‘know’ when the stakes are high and error possibilities have been introduced, for example. This may be thought to be enough to generate a tension: given that the presentation of radical skeptical hypotheses suffices to raise the epistemic standards to an unattainable level, the contextualist cannot even begin to defend her view in CC, for, given the knowledge norm of assertion, she cannot properly assert anything in such a context.
We agree that the contextualist should deny the epistemic propriety of asserting any empirical proposition in CH. Like the rest of us, the contextualist does not knowH anything about the external world. Hence, if a defense of her view requires asserting empirical propositions, the contextualist must deny that CC = CH. We think that such a denial is compatible with the contextualist framework.
Consider the objection that in holding that CC ≠ CH, the contextualist fails to take skeptical scenarios seriously, and, contrary to what she officially advocates, the contextualist actually presupposes that skepticism is false.Footnote 25 To respond to this criticism, we need to examine the question of how epistemic standards are contextually fixed according to contextualism. Although he remarks that ‘when the chance of error is salient in a context, the standards tend to rise to a point that falsifies the knowledge ascription’ (1999: 61), Stewart Cohen makes clear that ‘this is not to say that when the chance of error is salient we inevitably shift to a skeptical context’ (Reference Cohen1999: 85, n. 27). In general, Cohen writes, ‘the truth-value of a sentence containing the knowledge predicate can vary depending on things like the purposes, intentions, expectations, presuppositions, etc., of the speakers who utter these sentences’ (Reference Cohen1999, p. 57).Footnote 26
Clearly, the contextualist's purposes, presuppositions, intentions, etc., differ from those of the skeptic. The skeptic's purpose is to question whether we count as knowing anything about the external world, while the contextualist seeks to defend a certain semantical thesis about knowledge sentences. So while the skeptic's discussion takes place within the object language, the contextualist is concerned with meta-linguistic issues regarding ‘knows.’ Furthermore, the skeptic presupposes that one needs to eliminate all error possibilities in order to count as knowing, while the contextualist takes this presupposition to be context sensitive. The skeptic intends to attach high standards to his use of ‘know,’ while the contextualist expresses neutrality between high and low standards. There is thus no reason to hold that high standards are in place in CC.
Opponents of contextualism sometimes invoke David Lewis's Rule of Attention to make their point. According to this rule, ‘if in this context we are not in fact ignoring [a possibility] but attending to it, then for us now it is a relevant alternative’ (Reference Lewis1996: 559). It is not clear exactly how Lewis meant to apply this rule. One difficulty is that, as Lewis himself points out at the end of his article, he should have taken greater care to distinguish between the language we use when we talk about knowledge and the meta-language the contextualist uses to talk about the semantics and pragmatics of the first language. This would have allowed him, among other things, to distinguish between the skeptic's advocating an error possibility and the contextualist's merely reporting on the skeptic's advocating an error possibility. Only the former is an attempt to raise epistemic standards. In other words, contextualists who adopt the Rule of Attention should take into consideration the purpose with which an error possibility is mentioned.
Does this entail that the contextualist fails to take skeptical possibilities seriously? Not at all. Recall that according to contextualism, there is no such thing as knowing that p, simpliciter. Our subject S knowsL, but fails to knowH, that h. When in CC, the contextualist can coherently accept both the ordinary speaker's and the skeptic's knowledge claims. She can agree with the ordinary speaker that S knowsL that h, and agree with the skeptic that S does not knowH that h. Hence, the fact that CC ≠ CH does not entail that the contextualist ignores the seriousness of skeptical scenarios. She explicitly grants that the fact that these possibilities cannot be eliminated implies that we lack knowledgeH. Hence, the contextualist can coherently deny that the mention of skeptical scenarios raises the epistemic standards in CC, while agreeing with the skeptic's contention that our inability to rule out these scenarios entails that we lack knowledgeH.
4. WHAT EPISTEMIC STANDARDS ARE IN PLACE IN THE CONTEXTUALIST CONTEXT?
One may wonder what epistemic standards, if not high ones, are in place in CC. We think there are at least two answers available to the contextualist. We will not try to adjudicate between these here. Which position is to be preferred is an issue that contextualists should resolve among themselves. But we will argue that each position is compatible with the general contextualist framework.
Consider first the position that CC = CL. As we saw in the previous section, according to contextualism, epistemic standards are fixed by the interests of the conversational participants, including their presuppositions. While arguing for her view, the contextualist quite plausibly presupposes that her assertions – about our context-sensitive use of ‘knows,’ the existence of high- and low-standards contexts, the existence of other speakers, etc. – are epistemically proper. This means that if the knowledge norm of assertion is correct, the epistemic standards in CC must be low, since by assumption the contextualist only knowsL. The contextualist's interests, more specifically her presuppositions, thus arguably entail that low standards are in place in CC. Again, we do not see that this in any way begs the question against the skeptic, for the contextualist concedes that our inability to rule out skeptical alternatives entails that we lack knowledgeH.
It is worth noting that these low standards need not be the lowest admissible standards that can be associated with ‘know.’ The current proposal is that the standards have to be such that the contextualist satisfies them with respect to the propositions she asserts, bearing in mind, as we mentioned in the previous section, that she may not assert every proposition that she defends in CC.
Another option for the contextualist is to hold that epistemic standards are unsettled in CC.Footnote 27 In the previous section, we pointed out that standards are fixed by a host of features such as the purposes, intentions, expectations, presuppositions, etc., of the speakers, as well as the conversational maneuvers they perform. We also remarked that the contextualist's purposes, presuppositions, intentions, etc., do not favor high standards. They do not promote low standards either. When defending her point of view, the contextualist actually expresses neutrality between high and low standards, and this is reason to hold that standards are unsettled in such a context.
Consider by analogy a context in which flatness ascriptions, that is, utterances of the form ‘X is flat,’ are discussed. The contextualist about flatness ascriptions points out that ordinary speakers tend to associate different standards with ‘flat’ in different contexts. Suppose our contextualist uses the table in front of her to make her point. While ordinary speakers would describe the table as ‘flat,’ she points out, in some ‘high-standards’ context in which, say, a physicist remarks on ordinary objects' microscopic irregularities, the same table would not be deemed ‘flat.’ On that basis, the contextualist concludes that utterances of ‘X is flat’ express different contents in different contexts. What standards of flatness are in place in the context in which this semantic account is presented? Suppose one of the contextualist's interlocutors asks, ‘But is the table flat or not?’ It seems inappropriate for the contextualist to simply answer either ‘yes’ or ‘no.’ This suggests that as such, her presentation of her view did not put in place any particular standards of flatness. This should not come as a surprise: the contextualist professes neutrality between low and high standards of flatness, and presenting her view merely requires her to mention the word ‘flat’ rather than use it. It is thus plausible that flatness standards are simply unsettled in a context in which contextualism is defended.Footnote 28 This indeterminacy does not preclude talk about flatness. Conversational participants who wish to address the question of whether the table is ‘flat’ should be explicit about the standards they attach to ‘flat.’ In other words, when they start using the term ‘flat,’ they need to say, for instance, that the table is flat relative to everyday, ordinary standards, but not flat relative to the physicist's demanding standards.
Similarly, in presenting her view, the epistemic contextualist does not favor one set of epistemic standards over another. She professes neutrality among epistemic standards, and rather than using ‘know,’ she merely mentions that word. Again, suppose that right at the conclusion of her presentation, the contextualist is asked, ‘But does S know that h or not?’ It would be odd for her to answer this question by a simple, unqualified ‘yes’ or ‘no.’ Hence, one can reasonably hold that, while she is presenting her contextualist doctrine, epistemic standards are unsettled. Given this, the contextualist who wishes to make knowledge claims should be fully explicit, since she may take herself to meet different epistemic standards with respect to different propositions. She may say, ‘I know that this is a zebra relative to low epistemic standards, but not relative to moderately higher epistemic standards, since I can't tell zebras from well-disguised mules.’ And, she may say, ‘I know that I have hands, and by this I mean that I meet fairly high epistemic standards relative to that proposition, but I do not meet the skeptic's unattainable standards.’
We have relied on the analogy between ‘know’ and ‘flat.’ But one might object that there is a crucial difference between these two words: if epistemic standards are unsettled in CC, then so is the epistemic position a speaker must satisfy to assert properly. It is thus indeterminate whether a given assertion made in that context is proper or not. (By contrast, an indeterminacy of flatness standards has no similar impact.) Things may be thought to be even worse: as we saw in discussing the first position, given that the contextualist takes herself to be properly asserting many things in CC, it seems that the norm of proper assertion should be set at knowledgeL. This would mean that epistemic standards cannot be unsettled in CC: they have to be low, if we take the contextualist's assertions to be proper.
But this is too quick. One need not construe the meta-linguistic knowledge norm of assertion in this way. A proponent of that norm can coherently hold that epistemic standards may be unsettled in some contexts. When this happens, speakers ought to be clear about their own epistemic positions, that is, what standards they meet, in order to assert properly. For instance, to assert properly that p, a speaker should also specify what epistemic standards she meets with respect to p. This is especially important, given that the contextualist's epistemic position is not the same with respect to every proposition being considered in her context. Now, as we remarked earlier, the controversial philosophical claims she defends will be merely weakly asserted. And regarding more mundane propositions, she can make assertions of the form, ‘P, and I know that p relative to ordinary standards, but I don't know that p relative to the skeptic's high standards.’ Such assertions can be perfectly appropriate, even for advocates of the knowledge norm. As one such advocate notes, one may properly say, ‘P and by Descartes's standards I cannot be absolutely certain that p.’Footnote 29 This is the type of language a careful contextualist will employ. She will indicate, either spontaneously or in response to challenge, the standards relative to which she knows the propositions she asserts. In doing so, the contextualist represents herself as satisfying a corresponding norm of assertion. And her assertion is epistemically proper just in case her representation is accurate. None of that requires that there be unique, constant standards governing the whole context in which the conversation takes place.
5. CONCLUSION
We have argued that contrary to several authors' contention, contextualists who endorse factivity, epistemic closure, and the knowledge norm of assertion can coherently defend their theory without abandoning their response to skepticism. First, we have questioned the assumption that the contextualist needs to be taken to assert the content of her theory. Given that she and her conversational partners are aware of the controversial nature of this theory and what it entails, we can reasonably hold that her endorsement of that theory should be viewed as a weak assertive rather than an assertion.
However, we have conceded, the contextualist does assert some propositions in arguing for her view. These assertions can be appropriate, we have argued. First, given the interests of the contextualist, specifically her presuppositions that her assertions are proper, it is plausible to hold that the epistemic standards in place in her context are low. Alternatively, given that in presenting her view, the contextualist is neutral about what epistemic standards are the ‘right’ ones, one can plausibly hold that the epistemic standards in the contextualist context are unsettled. On both options, contextualism avoids the statability problem.
