1. THE KNOWABILITY PROBLEM AND ITS SOLUTION

The knowability problem is usually derived from considerations of the following sort.Footnote ⁵ Let “K_i(S, P)” (or “S knows_i that P”) stand for “ ‘S knows that P’ is true if uttered in contexts in which knowledge standard i operates”. Assume that there are two candidate standards for knowledge, L (for ‘low’) and H (for ‘high’). Let S be the epistemic contextualist, and let “hands” stand for the sentence “S has hands”. Assume also that knowledge is factive (FACT) and closed under (known) entailment (CLOS).Footnote ⁶ It can be easily shown that propositions (1), (2) and (3) are jointly inconsistent with the principles (FACT) and (CLOS):Footnote ⁷

1. (1) K_L(S, hands),

2. (2) Not: K_H(S, hands),

3. (3) K_H(S, (1)).

The argument demonstrates that, if there is no high-standard knowledge of hands, S cannot know_H that she knows_L this proposition. If (1) and (2) are contextualist theses, epistemic contextualism entails that it itself cannot be known_H, which would put the contextualist in a very uncomfortable position: Given the knowledge norm of assertion (KNA) and given that S finds herself in a high-standard context, S is unable to coherently assert the very theory she wants to defend. Contextualists are under the threat of pragmatic inconsistency.

There has been a number of responses, most of which accept that there is at least a prima facie case for the knowability problem. Brendel (Reference Brendel2003, Reference Brendel2005, Reference Brendel, Rami and Wansing2007, Reference Brendel, Leitgeb and Hieke2009) and Wright (Reference Wright2005) reject EC on the grounds of the knowability problem. Brueckner and Buford (Reference Brueckner and Buford2009) and Williamson (Reference Williamson2001) think that the knowability problem exists, but does not have any further effects: inability to knowingly state a theory undermines the possibility of asserting the position, but does not touch upon its tenability. Baumann (Reference Baumann2008) rejects the closure principle and thus blocks the inference to the effect that (1)–(3) form an inconsistent triad. Ashfield (Reference Ashfield2013) claims that knowingly asserting EC demands only low-standard knowledge of EC, and Montminy and Skolits (Reference Montminy and Skolits2014) arrive at the view that stating a theory is not to assert it.

My own response (Reference Freitag2011, Reference Freitag, Jäger and Löffler2012, Reference Freitag2013b) takes a radically different course. I deny that (1) and (2) are contextualist theses in the first place. Contextualism is not, I argue, committed to any factual statements about the obtaining or non-obtaining of knowledge concerning some specific contingent proposition, such as hands. To illustrate my point: If contextualism were committed to (1), it would be refuted if S did not have any hands. Surely, however, EC is not a position whose truth or falsity depends on such contingencies. Thus the mentioned argument is unable to show that epistemic contextualism has the knowability problem.

However, my aim is not only to dismiss some specific argument for the knowability problem, but to show that no such argument can be successful. To this end, I first prove that the knowability problem arises precisely for those theories that satisfy condition (KP) below, and then argue that contextualism does not satisfy this condition. I defend the following thesis (where 𝔓 is the set of empirical, contingent propositions):Footnote ⁸

A theory Θ entails that it cannot be known_H by a subject S if and only if there is some contingent proposition which Θ entails to be true and to be not known_H by S. Observe that the knowability problem applies only to epistemically self-undermining theories, i.e., theories which logically entail that they are unknowable by their proponents. Theory-external reasons for unknowability are irrelevant here. For example, the necessary falsity of a proposition renders it unknowable, but does not generate the knowability problem (more on this below).

As (KP) states the necessary and sufficient conditions for the knowability problem to arise, there remains only a single question to be answered: Does epistemic contextualism meet condition (KP)? As I have argued in my Reference Freitag2011, Reference Freitag, Jäger and Löffler2012, Reference Freitag2013a and Reference Freitag2013b, (MEC) is close to a faithful rendering of the generic form of epistemic contextualism (assuming it to focus on the two standards H and L; 𝔖 is the set of subjects):Footnote ¹¹

According to (MEC), epistemic contextualism merely claims that it is possible for some (English) sentence of the form “S knows that P” to be true if uttered in a context with knowledge standard L while being false if uttered in a context with standard H. In an object-language gloss, (MEC) says that for some subject S and some proposition P, it is possible that S has low-standard, but not high-standard, knowledge of P. As opposed to the conjunction of (1) and (2), (MEC) violates (KP) and hence is free of the knowability problem: (MEC) does not entail that it cannot be known_H by its proponents.

Though I am convinced that (MEC) is sufficient to represent generic contextualism, it is dialectically important to observe that even much stronger positions escape the grip of the knowability problem. Even the following theory (where 𝔖* is the set of all subjects who are not S and 𝔓_t is the set of true empirical propositions) does not have the knowability problem:Footnote ¹²

Even if S were to make the absurdly strong claim that everyone else has low-standard (but not high-standard) knowledge of all true empirical propositions, while she knows_L at least some of them, her claim would not entail that contextualism cannot be known. So even those who assume, falsely in my view, that a denial of scepticism requires that many (or almost all) of our ordinary knowledge claims are actually true, can be satisfied with plenty of versions of contextualism that evade the knowability problem.Footnote ¹³ There is hence no problem of coherently asserting epistemic contextualism, even if the ancillary principles (FACT), (CLOS) and (KNA) are accepted. Any claim to the contrary is based on a misconception of the commitments of EC.

2. MONTMINY'S AND SKOLITS' CRITIQUE

Montminy and Skolits take no issue with my minimalistic conception of epistemic contextualism as such. They also agree that (MEC) does not satisfy (KP),Footnote ¹⁴ object, however, that epistemic contextualism is unknowable_H nevertheless. They propose that (MEC) “is arguably an empirical thesis” and thence conclude that, “since MEC seems to be in the domain of propositions our ‘knowledge’ of which a radical sceptic could target, the contextualist cannot plausibly be credited with knowledge_H of MEC” (Reference Montminy and Skolits2014: 323). If, however, (MEC) cannot plausibly be known_H, Montminy and Skolits claim, the knowability problem is reinstated and my solution obsolete.

I think that this argument is misguided. Suppose that (MEC) is indeed an empirical thesis and that, because high-standard scepticism is true, it cannot be known_H. Nobody, the contextualist S included, would be able to know_H epistemic contextualism to be true. Yet this result, even if correct, is irrelevant to the debate of the knowability problem in its original form. The charge of the knowability problem is not that there are some independent factors – such as a theory-directed form of scepticism – which preclude the knowability of EC, but that contextualism in itself forestalls the possibility of being known by its proponents. The knowability problem arises only if a theory entails that it is unknowable by S. (Montminy and Skolits themselves describe it as the problem “that the contextualist cannot coherently assert, or know, the content of her theory” (Reference Montminy and Skolits2014: 319; my emphasis. See also the title of their paper).Footnote ¹⁵) To illustrate my point, observe that necessary falsehoods cannot, given factivity, be known to be true. So “Montminy = Skolits” is unknowable_H, although it does of course not have the knowability problem: this thesis does not logically entail that it cannot be known.Footnote ¹⁶ Therefore, it is not incoherent to assert this proposition. Similarly, that (MEC) cannot be known_H if high-standard scepticism is true, is irrelevant to the question of whether (MEC) has the knowability problem: epistemic contextualism does not entail high-standard scepticism.Footnote ¹⁷

A critic of my solution would have to deny that (KP) adequately captures the necessary conditions for the knowability problem or that epistemic contextualism violates (KP). Montminy and Skolits do neither. So their arguments have no bearing on my position and I am still able to maintain that epistemic contextualism, properly understood, can be knowingly asserted. Of course this does not mean that Montminy's and Skolits' own suggestion is false that stating epistemic contextualism is a speech act which is not governed by the knowledge norm. It only means that there is at present no need for this manoeuvre.

Let me briefly address the authors' second objection, according to which even a proponent of a minimal form of epistemic contextualism such as (MEC) is pragmatically inconsistent, because “in practice, S is committed to both p and non-K_H(S, p), for many propositions p” (Montminy and Skolits Reference Montminy and Skolits2014: 323–4). My critics reason as follows: “Qua ordinary speaker,” they say, “the contextualist S … will find herself in ordinary conversational contexts in which … S will accept some of [the] assertions and [knowledge] attributions” made in these contexts. “She will thus be committed to the truth of some empirical proposition p” and “it would be extremely odd for a contextualist to maintain that she does know_H that p” (Montminy and Skolits Reference Montminy and Skolits2014: 323–4).

Again, I am puzzled by this reasoning. Observe first that, just like anyone else, a contextualist might violate all kinds of pragmatic norms in everyday discourse. But if that is not a consequence of the theory she holds, no objection to EC can be derived from such possible misdemeanour. So, I take it, Montminy and Skolits aim at demonstrating that contextualists are committed to some form of pragmatic inconsistency. I shall now dispute this thesis.

My critics start out with the claim that our contextualist S is ‘in practice’ committed to the truth of some proposition p, but I think that this is false. A contextualist might well be a Pyrrhonian Sceptic with respect to ordinary empirical propositions and hence abstain from any judgment about p (for arbitrary p∈𝔓). Nor is she committed to the view that she does not have high-standard knowledge of some particular proposition p she believes, as this does not follow from (MEC).Footnote ¹⁸ Montminy and Skolits themselves acknowledge this point when they say that “perhaps her contextualist point of view, narrowly understood as MEC, does not force her to make this concession [that she does not know_H that p; W. F.]” (Reference Montminy and Skolits2014: 323). Yet they attempt to justify their thesis by adding that it would be “extremely odd” for a contextualist to claim possession of high-standard knowledge (presumably because outright denial of high-standard scepticism would make EC much less interesting for epistemology).Footnote ¹⁹ Yet this observation, even if true, does not support their thesis. First, I don't see that contextualists must avoid this oddity. If contextualists denied high-standard scepticism, that would seriously diminish the interest of their thesis, but the thesis could still be true. Second, even if epistemic contextualists could for some reason or other not claim high-standard knowledge, the desired conclusion would not follow. We cannot infer from the fact that S does not maintain that she does know_H p that she maintains that she does not know_H p! So the epistemic contextualist does not even in practice have the allegedly problematic commitments.

Suppose, however, that we waive these (rather serious) concerns, and assume that in ordinary contexts S is indeed committed to both p and non-K_H(S, p) for many propositions p. A pragmatic inconsistency does not arise unless, firstly, S not only believes, but also asserts p, and, secondly, the knowledge norm of assertion demands high-standard knowledge of the asserted proposition. Neither of these conditions need be satisfied. A contextualist might simply refrain from asserting what she believes and hence avoid pragmatic inconsistency. But even if she asserted p, she would not necessarily violate the norms of assertion either, because S's practical context as described by Montminy and Skolits is ‘ordinary’ and hence presumably a context in which low-standard knowledge norms operate. There is hence no reason to assume that a contextualist must not assert that p, even if she does not possess high-standard knowledge thereof. Contextualism, in the form of (MEC), is free of any pragmatic inconsistency. Hence there is still no argument against my solution to the knowability problem.

3. CONCLUSION: A NEW KNOWABILITY PROBLEM?

I maintain that the knowability problem in its original form is solved. There is no problem of coherently knowing and asserting epistemic contextualism properly understood. It might, however, be suggested that Montminy and Skolits identify another problem, according to which, even though contextualism is coherently knowable_H, it is not knowable_Hgiven the way things are. I will conclude with some remarks on this ‘new’ knowability problem.

The new problem differs from the original one in that it arises only given some additional condition. More specifically, Montminy and Skolits claim that epistemic contextualism is unknowable_H due to the fact that (MEC) and all stronger versions of epistemic contextualism fall into the scope of high-standard scepticism. According to their view, (MEC) is an “empirical thesis” and “since MEC seems to be in the domain of propositions our ‘knowledge’ of which a radical sceptic could target, the contextualist cannot plausibly be credited with knowledge_H of MEC” (Montminy and Skolits Reference Montminy and Skolits2014: 323). Let us therefore assume that epistemic contextualism is indeed a proposition that falls into the scope of high-standard epistemic scepticism. As a consequence, EC cannot be known_H – if high-standard scepticism is true. However, Montminy and Skolits have not shown that high-standard scepticism with respect to EC is true. Nor may it be presupposed in the present context; as I have already mentioned, contextualists are not committed to any form of scepticism. Montminy's and Skolits' crucial premise is therefore unsupported and the new knowability problem as yet unmotivated.

Yet even if we assume high-standard scepticism of the form required, there does not seem to be any problem worth the name. If theories about the semantic meaning of the English word ‘knows’ are themselves not knowable_H, it would indeed follow that epistemic contextualism is not knowingly statable.Footnote ²⁰ But this cannot be plausibly held against epistemic contextualism, since its only competitor, epistemic invariantism, is, like contextualism, a theory about the semantics of ‘knows’ and would hence be as little knowable_H as contextualism itself.Footnote ²¹ If we can't know_H semantic facts about ordinary language, we can't know_H that the meaning of ‘knows’ is context-independent either. I therefore seriously doubt that the new knowability problem needs our attention. If it does, however, Montminy's and Skolits' view that proponents of EC don't have to assert their theory might well prove to be the right response.