Metaphysics, Language, and Epistemology Group

Abstract: “When a panel of judges splits neatly along national or factional lines, it feels as though something systematic is going on. It is harder to say what that feeling is tracking. The panel might be divided by ordinary disagreement, some judges might simply be harsher graders, or the judges might be biased. And if they are biased, is it one bias or several? Call this the individuation problem for cognitive bias. I argue that the usual practice of classifying biases by the outputs they produce cannot solve it, because the same pattern of judgment can come from quite different underlying dispositions. This is an extension of the familiar many-to-on mapping problem in cognitive science that I will argue is more pernicious for biases. This matters for a recent debate between Thomas Kelly and Selim Berker on the overgeneration problem for symmetry-theoretic accounts of biases: such accounts seem to count almost everything as biased. Norm-theoretic views answer this by making departure from a norm of correctness part of what it is to be a bias. I agree that something in this pressure is right, but I think it gets the order of explanation backward. We cannot say whether an asymmetry is defective or maladaptive until we know what it is an asymmetry in, and across which circumstances it operates. I propose that individuation comes first and base individuation in the algorithmic learning processes of decision-making under uncertainty. Normativity comes in afterwards, with the question of whether a given bias serves an agent well in the environment it finds itself in.”

Registration: If you do not hold a university card, please contact the seminar convenor or admin@philosophy.ox.ac.uk at least two working days before a seminar to register your attendance.

_{Metaphysics and Epistemology Group Convenors: Nick Jones, Mona Simion, Bernhard Salow and Alex Kaiserman}