Natural kinds

4. Conventionalism

Kripke and Putnam assume that what natural kinds there are, and which objects are members of those kinds, are matters which obtain independently of our attempts at classification. Conventionalism about natural kinds denies this assumption. It says that our classification of objects into natural kinds does not reflect any preexisting divisions in nature. Instead, it is based on human convenience, and its goal is the prediction and control of observable events, not the discovery of underlying essences.

Locke ultimately took this view. Frequently, he is interpreted as taking it for an epistemological reason, namely that real essences, if there are any, would be unobservable and so unknowable. More recently, Bas van Fraassen (1980) has taken a similar view, but for a different reason, believing that science need be concerned only with observable entities, and so not with real essences. Hence the latter are redundant: they are ‘metaphysical baggage’. The dispute between realists and conventionalists about natural kinds is then located within the more general dispute between scientific realism (which claims that science describes an unobservable world) and constructive empiricism (which is agnostic about whether science does this). A further reason for scepticism about essentialism about natural kinds, and one which Locke and van Fraassen share, is a suspicion of what de re or metaphysical necessity is supposed to be. The suspicion is that the thesis that there are necessities in nature is obscure, and that instead the only source of necessity lies in human conventions. Accordingly, the thesis that objects have real essences in virtue of which they are members of certain natural kinds is held to be infected with this obscurity.