Robin McKenna

I develop and argue for a version of epistemic contextualism - the view that the truth-values of ‘knowledge’ ascriptions depend upon and vary with the context in which they are uttered - that emphasises the roles played by both the practical interests of those in the context and the epistemic practices of the community of which they are part in determining the truth-values of their ‘knowledge’ ascriptions. My favoured way of putting it is that the truth of a ‘knowledge’ ascription of the form ‘S knows that p’ requires that the subject of the ascription can rule out the relevant alternatives in which not-p, where the relevant alternatives are the ones that those in the ascriber’s context have a reason to consider. What alternatives those in the context have a reason to consider depends on their practical situation and on what alternatives are generally considered relevant within their community. I call this ‘interests contextualism’. The thesis splits into three parts. First, I deal with what I call ‘linguistic objections’, which purport to show that there’s no linguistic evidence that the expression ‘knows’ is context-sensitive, and objections concerning the way ‘knows’ behaves in intra- and inter-contextual disagreement reports. I argue that there are a number of ways in which contextualists can deal with these objections. Consequently, they provide no reason to reject contextualism. Second, there are a number of ways of going beyond the basic contextualist thesis, and I argue that the best way is along the lines indicated above, viz. interests contextualism. In the process I articulate a number of desiderata for a contextualist account of the features of context that are responsible for contextual variation in the truth-values of ‘knowledge’ ascriptions. I argue that, unlike its main rival - which I call ‘conversational contextualism’ - interests contextualism can satisfy all of the desiderata. Consequently, interests contextualism is preferable to conversational contextualism. Third, I argue that there is good reason to prefer interests contextualism to its noncontextualist rivals, strict invariantism, sensitive invariantism and relativism. The objections dealt with in the first part are meant to provide the main reason to prefer a sort of relativism to interests contextualism. Consequently, the upshot of the first part is that relativism is off the table. The considerations that tell in favour of interests contextualism and against sensitive invariantism are of two types. First, I argue that interests contextualism can deal with a wider range of cases than sensitive invariantism. Second, I argue that the influential account of the function of ‘knowledge’ ascriptions developed in Edward Craig tells against sensitive invariantism and in favour of interests contextualism. I also argue that the second consideration tells against strict invariantism as much as sensitive invariantism. Consequently, I conclude that interests contextualism is preferable to all of its rivals.

What is knowledge? What should knowledge be like? Call an epistemological project that sets out to answer the first question ‘descriptive’ and a project that sets out to answer the second question ‘normative’. If the answers to these two questions don’t coincide—if what knowledge should be like differs from what knowledge is like—there is room for a third project we call ‘revisionary’. A revisionary project starts by arguing that what knowledge should be differs from what knowledge is. It then proposes that we revise our account of knowledge accordingly. Our aim in this paper was to develop a methodology for revisionary projects in epistemology. Put roughly, the thought is that we start by looking at the various things that we expect knowledge to do for us. Once we have a list of the various things we expect knowledge to do for us we have a ‘job description’; a list of tasks we need done, and that we expect knowledge to perform. With the job description in hand, we can ask what knowledge would hav..