This is my doctoral dissertation (link below).
The predominant view in contemporary philosophy regarding natural kinds is broadly realist. More specifically, putative natural kinds are not just considered to be those kinds that we happen to find useful, but rather are taken to show us something about the mind-independent structure of nature. There are at least two important problems with how such a position is motivated in the literature. The first concerns the realists‘ use of claims regarding our inductive knowledge, while the latter involves their use of explanatory power as a tool in confirmation. These problems motivate the alternative, anti-realist view developed in this dissertation. Drawing from the ideas of Hume (1740, 1777), Quine (1969, 1973, 1995), and Stemmer (1981, 2007), I argue that natural kinds are just those classes about which subjects have inductive expectations; using a phrase from psychology, natural kinds are generalization classes. Far from being a radical view, this account accords well with key philosophical insights and approaches from Locke (1975), Hume [1777(2007)] and Mill (1970). This anti-realist view of natural kinds is preferable to natural kind realism because it adequately describes how we talk about and use natural kinds, while at the same time requiring fewer metaphysical and epistemological assumptions. The account of natural kinds offered in this dissertation provides an alternative for those who do not wish to embrace natural kind realism and its associated requirements.