Abstract
Emotion theory is beset by category disputes. Examining the nature and function of scientific classification can make some of these more tractable. The aim of classification is to group particulars into <<natural>> classes - classes whose members share a rich cluster of properties in addition to those used to place them in the class. Classification is inextricably linked to theories of the causal processes that explain why certain particulars resemble one another and so are usefully regarded as <<of the same kind>>. The need to base categories on underlying causal processes explains why mere careful definition (including operational definition) need not produce categories that are productive objects of scientific study. Because different causal processes produce different patterns of similarity there is unlikely to be a single classification that is optimal for addressing all scientific questions. Cultural categories should not be contrasted to natural categories, but should be treated as natural classes generated by underlying social processes. Our capacity to introduce epistemically optimal categories is often restricted because categories play a role in social and political, as well as epistemic, projects. This account of classification has many implications for emotion theory.