Abstract

The aims of the article are related to comparing two versions of structural realism: epistemic structural realism (ESR) and ontic structural realism (OSR). My main goal is to argue for the following: (i) the mature version of ESR admits, despite its declarative agnosticism, a weak metaphysical thesis regarding natural kinds; (ii) this thesis is problematic; (iii) a better metaphysical thesis for the (structural) realist can be found in OSR. In doing so I significantly refer to the notion of semantically interpreted theories. I illustrate my point with a simple example from spacetime physics, where I show that OSR allows for a realist account of curvature facts more compatible with the actual physical theory than the account enabled by ESR.