4 Linguistic Arguments for Intellectualism
4.3. The Semantic Implementability Constraint
In this chapter I have set up the context for our discussion of knowledge-how in the rest of the thesis, setting out the main contours of the debate about the nature of
knowledge-how with a focus on the importance of linguistic arguments. I have offered an interpretation of Ryle’s discussion of knowledge-how that shows him to be principally interested in non-linguistic issues, and argued that he doesn’t hold that knowing how to do something is just any kind of ability. I have suggested that – somewhat surprisingly – linguistic evidence entered into the knowledge-how debate on the side of Anti-
Intellectualism via linguistic motivations for the standard view, and argued that the linguistic motivation for this view is weak.
In §4, I set out what I take to be the linguistic arguments for Intellectualism and worked through some of the responses to this argument to get clearer on the legitimate uses of linguistic evidence in this debate. The overall assessment of these criticisms is partly concessionary to the Intellectualist. I have conceded that truth-conditional semantics are relevant to our account of the nature of knowledge, meaning that one can legitimately use linguistics as a source of evidence about the nature of knowledge-how. However, I have argued that we shouldn’t overstate the significance of linguistic evidence. For one thing, we should not expect linguistic uniformities to follow through to
61 Stanley accepts this this point (2011b, pp. 148–9), and it is natural to read chapters 7 and 8 of
Know How as a defence of the idea that knowledge-how involves theoretical knowledge, rather than some other relation.
metaphysical uniformities. Furthermore the standard semantics offered by contemporary linguistics are not the final word in our account of the truth-conditions of knowledge-how ascriptions, and might be over-ridden by philosophical considerations. I have also
suggested that the linguistic evidence — both concerning the correct semantics for English ‘knows how’ ascriptions, and the cross-linguistic evidence — is more complex than Stanley and Williamson claim. Even if Stanley and Williamson’s semantics for English is correct, I have argued that it does not resolve some crucial issues about the nature of knowledge- how: leaving open the question of how we should characterise the knowledge-how relation.
Going forward, it would be helpful to have a positive picture of what the legitimate role of linguistic evidence in this debate that accepts the relevance of linguistic
considerations to the nature of knowledge-how, without endorsing the general priority of linguistic over philosophical considerations. I think that one helpful way to think about legitimate role of linguistic evidence is in this debate is as providing a semantic
implementability constraint on accounts of knowledge-how. Supposing that knowledge- how is the state which is picked out by our ordinary ascriptions of the form ’S knows how to V’, it is a condition on an acceptable account of the nature of knowledge-how that it is compatible with a linguistically plausible account of the truth-conditions of ’S knows how to V’. All we need to get this connection is the truth schema; we don’t need to be
committed to the controversial linguistics-first methodology. Stanley hints at this kind of implementability constraint:
Suppose one produces an analysis of knowing how to do something. Surely, it would be a worry with such an analysis if there is no correct compositional semantics of English according to which ascriptions of knowing how to do
something express that analysis. If there were no plausible compositional semantics for English ascriptions of knowing how that assigned to them one's favored
analysis, then that would show that one's analysis could not possibly be what English speakers mean when they use such ascriptions. (Stanley, 2011 quoted in Bengson & Moffett, 2011b, p. 36).
I think that this constraint provides a plausible picture of the role of linguistic evidence in this debate. We can understand the linguistic criticisms of the Standard picture as applications of this constraint. Supporters of the standard view claimed that a non- propositional account of knowledge-how was the only semantically implementable account of knowledge-how, and their Intellectualist opponents quite reasonably pointed out that their view was not semantically implementable, given standard views about ‘knows how’ ascriptions. This constraint also gives some prima facie support to Propositionalism. Since an Propositionalist account can be semantically implemented, via Stanley and Williamson’s semantics, it has one up over other accounts of knowledge-how, for which there is not an obvious semantic implementation.
One place where I want to depart from Stanley concerns what happens if an account fails the implementability constant. If it turns out that an account of knowledge- how is not compatible with any off-the-shelf semantics, then this is a strike against that account, because this provides evidence that our ordinary ‘knows how’ ascriptions are not picking out the state posited by that account, leaving the proponent of that account open to the change that they are changing the subject away from our ordinary notion of
knowledge-how (Bengson & Moffett, 2011b, pp. 36–7; Stanley, 2011b, p. 144). However, there are a number of ways in which the supporter of an unimplementable account might respond to this challenge. One would be to offer a novel account of the semantics of ordinary ‘knows how’ ascriptions that is amenable to their metaphysics (perhaps along the lines of (Santorio, 2016)). Taking this option allows the supporter of this account to claim that their analysis connects to ordinary meaning; the point is just we needed to do some substantial philosophical work to get a grip on that meaning.
An alternative response is to disconnect their account of knowledge-how from the semantics of ‘knows how ascriptions entirely. One might opt for an error theory of our ordinary ‘knows how’ ascriptions, claiming that all sentences of the form ’S knows how to V’ are false, meaning that we need to employ other constructions to pick out knowledge- how. Alternatively one might pursue a revisionary philosophical project, opting for an explication of knowledge-how, which aims to tidy up our ordinary notion of knowledge- how in ways that deliberately depart from the ordinary concept (Bengson & Moffett, 2011b, p. 44). If our ordinary knowledge-how ascriptions are simply false, or the philosophical project is to develop a new notion of knowledge-how, then the charge of
changing the subject loses its sting, and theories of knowledge-how can be developed independently of linguistic considerations.
In closing, I want to connect the criticisms of Intellectualism to the chapters to come. In §4.2.2. I argued that there might be philosophical reasons to think that knowledge-how is not a species of propositional knowledge. In chapter 3 I give an example of such an argument, contending that Intellectualism faces a significant problem in isolating a kind of propositional knowledge that is plausibly identical with knowledge-how. In §4.2.4. I pointed out that there are a number of alternatives to the interrogative semantics offered by Stanley and Williamson. In chapter 2 I will consider the Free Relative semantics suggested by Abbott and Bach, argue that it provides the only way for Objectualists to semantically implement their account of knowledge-how, and show that it is a linguistically implausible account of ‘knows how’ ascriptions. In §4.2.5. I considered Glick’s argument that linguistic evidence fails to establish the interesting kind of Intellectualism. In chapter 7, I will build on his discussion, arguing that the linguistic evidence is compatible with knowledge-how being a species of ability — the ability to answer a question.