Amit Karmon (Reuben College): 'Which Ideal Epistemology?'
Abstract: In her 2022 paper "Why Ideal Epistemology?", Jennifer Carr argued for two claims: (a) ideal epistemology focuses on perfect and computationally unbounded learners, and (b) only ideal epistemology is normatively robust, meaning that it is not conventional or context dependent. While Carr's case for (b) is plausible and some idealisation in epistemology is needed -as in any theoretical field- her argument for (a) is not compelling. I suggest that Carr's reasoning applies equally well to a wide range of ideal models of more realistic learners, who are computationally bounded and therefore inherently limited in what they can learn from their evidence. The model of ideal unbounded learners can be seen as a degenerate limiting case. Consequently, the epistemic norms governing ideal unbounded learners are typically of limited relevance for actual learners, except in some special, non-representative cases. To better understand the epistemic norms that apply to less impressive learners like us, our attention should shift to models of computationally bounded learners, which are better suited for the job.
