Realism and Kinds

In the recent debate about scientific concepts, pluralists claim that scientists can legitimately use concepts with multiple meanings, while eliminativists argue that scientists should abandon such concepts in favor of more precisely defined subconcepts. While pluralists and eliminativists already share key assumptions about conceptual development, their normative positions still appear to suggest that the process of revising concepts is a dichotomous choice between keeping the concept and abandoning it altogether. To move beyond pluralism and eliminativism, I discuss three options of revising concepts in light of new findings, and when scientists should choose each of them.

One of the core charges against explanationist scientific realism is that is too epistemically optimistic. Taking the charge seriously, alternative forms of scientific realism -- semi-realism and theoretical irrealism -- are designed to be more modest in their epistemic claims. I consider two cases in cosmology and astrophysics that raises novel issues for both views: the cosmic event horizon inverts important tenets in semi-realism ; theoretical irrealism appears incompatible with standard evidential reasoning in the context of the dark matter problem.

Certain philosophers claim the No-Miracles Argument (NMA) for realism commits the base-rate neglect fallacy. I argue that it does not. In general, one commits a base-rate fallacy only when one has access to the relevant base rate. And in the case of scientific realism, we lack access to the relevant base rate. The most natural attempt to save the base-rate objection from this reply leads to unwelcome consequences. However, this dialectic leads to a legitimate concern about the NMA. I conclude by sketching a new type of No-Miracles-style argument, which avoids this concern.

Motivated by a desire to make precise the position of structural realism, David Wallace recently articulated a framework for understanding the relationship between metaphysical and scientific theories. In this paper, I demonstrate two particularly significant consequences of this framework for the metaphysics of science, unrelated to considerations from structural realism: that it underwrites resolutions to both Putnam's (permutation) paradox and the new riddle of induction.