Our common-sense epistemological thinking represents the various species of knowledge (self-knowledge, perceptual knowledge, inferential knowledge, knowledge of other minds, and so on... ) as displaying important differences in structure. Some knowledge seems to rest essentially on evidence or justification, some doesn't. Some knowledge requires making a discovery, some doesn't. Some knowledge seems to concern a logically independent object, whereas some seems at least partly constitutive of its object. We have first person authority in relation to some of the things that we know, but not others - and so on. The challenge for any account of knowledge is to accommodate and explain such differences, and at the same time to explain what it is that different kinds of knowledge have in common: why these phenomena which are at one level rather different from one another, are at another level instances of a single phenomenon, being all, equally, knowledge.

My project introduces this challenge, and responds to it by developing what I call Epistemological Pluralism, a two-tier epistemological framework which takes seriously the idea that there are significant structural differences between various kinds of knowledge. Epistemological Pluralism seeks to accommodate and explain these differences, whilst at the same time explaining how the various species of knowledge form a unified genus.

My current working hypothesis views knowledge at the level of genus as a certain kind of rational ability (an idea we can find in Ryle, but which has been developed more recently in detail by John Hyman and Stephen Hetherington). Thinking of knowledge in this way allows us to explain the differences between the different species of knowledge in terms of the different ways in which the ability which is constitutive of knowing is explanatorily underwritten. Having this ability in all cases requires being (as I put it) "mentally engaged" with a fact. But, I argue, this mental engagement can take different forms, and can relate a person to a fact for different reasons. It is these differences - in the ways in which and reasons for which one can be mentally engaged with a fact - which gives rise to the structural differences between different species of knowledge.

I have developed this idea in relation to practical and psychological self-knowledge, and more recently in relation to perceptual knowledge. In practical knowledge, I have argued (see here and here), a person is mentally engaged with the fact which she knows - and so in possession of the rational abilties constitutive of knowing it - in virtue of executing her intention. I am mentally engaged with the fact that I am now typing, for instance, in virtue of executing my intention to type and thereby precipitating or constituting the fact that I am typing. In psychological self-knowledge, This idea develops Elizabeth Anscombe's conception of practical knowledge as 'the cause of what it understands' and 'undermined by a mistake in perfrmance', and indeed the whole project is strongly influenced by some ideas I find in Anscombe - although I don't suggest that she would agree with everything I want to say about knowledge.

Moving on to psychological self-knowledge, the idea is that I have the rational abilities constitutive of knowing that I am in some state of mind M, simply in virtue of being in M. Why? Because it is essential to our states of mind that they are expressible, and because we can express our mental states in ways which co-opt our broadly rational capacities. The rational ability constitutive of knowing that one is in some state M is thus contained within the ordinary expressive potential of being in that state. (I sketch this form of Constitutivism about self-knowledge here, and develop it by appealing to the idea of rational expression in a paper forthcoming in PPR). In more recent work-in-progress, I argue that in perceptual knowledge, the rational ability which is constitutive of knowing that p is explained by a state of epistemic perception. This is not the explanation of perceptual knowledge in terms of an evidential or other epistemic basis, but a version of the idea that perceiving that p can itself be knowing that p. Perceiving that p can be knowing that p, I argue, because perceiving that p puts one in possession of the rational abilities which are constitutive of knowing that p. Putting this together with my account of self-knowledge gives the result that perceiving that p can directly explanatorily ground both knowledge that p, and self-knowledge that one is perceiving that p.

The next step is to consider additional forms of knowledge, (e.g.) testimonial knowledge, inferential knowledge, and knowledge of other minds. In particular, we need to understand a) how the rational abilities constitutive of knowing are explained in relation to these species of knowledge, and b) how these different explanations might make sense of formal differences between these knowledge-species, as represented in our ordinary thinking. These understandings will then feed back into a clearer understanding of the nature of knowledge in general. In particular, they are needed to get clear on how best to understand the nature of the rational ability or capacity (or abilities, or capacities) which constitute propositional knowledge.

Underwriting Epistemological Pluralism is a critique and reversal of standard methodology in analytic epistemology. The standard methodology tends to abstract a general account of knowledge by considering a restricted class of knowledge-types, and then tries to force additional species of knowledge into this mould. Some kinds of knowledge do seem to require an evidential basis, or justification, for example. But when this structure is taken as a model for all knowledge, we get into trouble. My knowledge that I have a headache isn't obviously based on evidence for thinking that I have a headache. Nor, as Austin pointed out, is my knowledge that there is a pig in front of me when I am simply visually confronted with it. The traditional approach develops an account which is supposed to fit all forms of propositional knowledge, but does so without paying sufficient attention to the variety of forms which propositional knowledge can actually take.

The result is a literature which is full of general accounts of knowledge which clash with our common-sense understanding of various particular forms of knowledge. Rather than responding to this result by questioning the general account of knowledge at issue, proponents of the standard approach tend to question these common-sense conceptions. Psychological self-knowledge, for example, is often represented as resting on some kind of epistemic justification or evidence, but of a kind which is 'hidden' from the common-sense perspective, which itself is dismissed as mistaken.

This methodology is highly suspect. In the abstract, it builds a general account of knowledge out of our common sense conception of certain knowledge-species, and then uses this to reject our common-sense understanding of other knowledge-species, but offers no non-question-begging justification for taking the common-sense data seriously when they support the theory, and dismissing them as 'mistakes' when they look like counterexamples.

Epistemological Pluralism moves away from this problematic methodology, by seeking first to describe the structures of various species of knowledge without making hard assumptions about the structure of knowledge in general. It treats these descriptions as equally theoretically relevant to the task of describing the general form of knowledge, and abstracts from these an account of propositional knowledge at the level of genus. As noted, my current working hypothesis has it that knowing that p is, at its most general, having a certain kind of rational ability in relation to the fact that p. So far this assumption has proved very fruitful. But this general account of knowledge is only as defensible as is its capacity to accommodate the various forms of knowledge. So how it is best developed, and ultimately whether or not we should accept it in any form, are questions which are not yet answered.