Work in Progress Seminar

I gave a talk at last week’s graduate work in progress (WiP) seminar on Williamson’s account of knowledge and some of the issues I’ve blogged about here recently. In particular, I developed an objection concerning the explanatory value of Williamson’s account, which I then tried to defend the account against on the basis of the burden of proof argument given below.

I think it would be fair to say that my suggestions met with mixed reactions. On the one hand, not everyone felt that I was being fair to Williamson by characterising his account as saying nothing particularly substantive about the nature of knowledge given the fact that he had written an entire book on the subject. On the other hand, some found it difficult to accept that there could be a satisfying explanation of something—in this case whether a given belief either does or does not constitute knowledge—without providing a complete set of necessary and sufficient conditions, which was exactly the position I was arguing against. Conversely, others felt that the constraints upon a theory of knowledge I set up begged the question against the Williamsonian view, even though I was claiming that these constraints could be met by the view provided that they are not interpreted too strongly (i.e. as mandating a set of general principles for knowledge, rather than the requirement to give a principled explanation of each of its instances).

With the benefit of hindsight, it might have been better to structure the talk around just one of the constraints upon knowledge that I discussed—the explanatory constraint being the obvious choice—and to spend more time elaborating my positive proposal, which only appears towards the end of the paper. This would perhaps have made the dialectic a bit clearer, although I think that most of the points I was making still stand. The experience of presenting and discussing these views has, however, made me think about the issue in a slightly different light.

One of the upshots of Williamson’s proposal is that there is (and perhaps can be) no general definition of knowledge. Instead, there are a series of principles, or necessary conditions, such as justification, truth, belief, reliability and so on, that govern our use and application of the concept. Given that we cannot provide a full set of necessary and sufficient conditions for knowledge, we can only provide successive approximations of what knowledge consists in. An account of the necessary conditions for knowledge is therefore only correct to a given degree of approximation. In some cases, the relevant conditions may be limited to particular kinds of knowledge (or ways of knowing), such as perceptual or a priori knowledge. Other conditions may be quite general (e.g. factivity), but in each case may be susceptible to further restriction or qualification in light of the special circumstances that can arise to prevent knowledge, as demonstrated by the numerous counter-examples to justified true belief-based theories.

Although I hadn’t noticed it before, there is an analogy here with Williamson’s views concerning the phenomenon of vagueness. According to his theory of ‘epistemic vagueness’, words have (in some abstract metaphysical sense) particular sharp meanings based upon the linguistic practices of a given community. However, we are unable to track these meanings precisely—in part because they are constantly shifting—thereby giving rise to the phenomenon of linguistic vagueness. Terms such as ‘bald’, ‘heap’ and so on, therefore appear to be vague despite having metaphysically precise meanings. Whilst Williamson isn’t suggesting that the term ‘knowledge’ is vague (although why wouldn’t it be?), the idea seems to be that it has a precise extension (i.e. a particular mental state) whose nature is difficult—perhaps impossible—to codify.

Another parallel can be made with Wittgenstein’s philosophy of psychology. On this interpretation, the concept of knowledge can be likened to a kind of language game, with the various analogies and resemblances between different kinds or ways of knowing forming a series of ‘family resemblances’ (to borrow Wittgenstein’s phrase). However, whilst Wittgenstein can be read as claiming that this is all there is to psychological concepts, which need exhibit no underlying metaphysical similarity, Williamson does claim that there is some definite state of mind corresponding to the possession of knowledge. This opens up the possibility of characterising that state of mind in non-epistemic terms, perhaps as part of a more fully developed theory of mind and mental states (although it is unclear whether such a characterisation would have much in common with the JTB-style conceptual analyses).

As previously mentioned, I am sceptical about Williamson’s assertion that knowledge is a mental state and of his arguments in favour of this view for reasons relating to the nature of such states. To my mind, knowledge, like belief, seems more like a disposition or ability, regardless of whether you consider it as wholly ‘internal’ or partially ‘external’ to the subject. Furthermore, taking the above points into consideration, it seems that we can give a pretty good approximation of what such dispositions or abilities consist in. The crucial difference between this and the reductive view being that such an analysis is never fully precise or complete, but rather always subject to further qualification and elaboration.

The nature of these elaborations will also differ between the two views. On the reductive view, elaborations take the form of further general conditions which may be added to the analysis to give a full account of what constitutes knowledge. On the Williamsonian view, elaborations are principally contrastive, highlighting salient similarities or differences between cases in order to further refine the relevant explanatory principles. Provided that the Williamsonian can always point to some relevant similarity without resorting to ad hoc rules or special pleading on the basis of individuals rather than general features, I argue that the explanatory constraint upon knowledge can be met in the absence of a complete set of necessary and sufficient conditions for knowledge in general.

Another interesting consequence of the view in question is that the various principles and conditions that determine whether something is a case of knowledge or not do not form part of the theory of knowledge per se. They are, rather, additions or refinements to it that map out the various kinds (or ways) of knowing in greater detail. Thus, Williamson’s account has the character of a framework within which explanations can be given rather than itself supplying all the resources upon which such explanations may draw. Again, this may seem too ‘thin’ to constitute a full theory of knowledge. However, if the concept of knowledge is fundamental to the theory of mind, and so not susceptible to further analysis, perhaps this is the best that can be hoped for.