Colloquium: Functional kinds, structural kinds, and the irreducibility of the special sciences.

First, I argue that a unified conception of the notion of function across technology, biology and society can be achieved. This unified conception is grounded in the idea that the notion of function cannot be explicated by a single predicate, as all existing theories of function presuppose. Instead, an item can either have a function, in the precise sense of having been invested with it, or can perform this function, and that these two ways a item can be ‘related’ to a function only jointly exhaust the notion of function. Only in this way can room be made for one of the defining characteristics of function, i.e. malfunction. Making this room, however, entails that functions can only be had by items of a particular kind, a kind that must be definable independently of the function, in non-functional, structural-developmental terms. The biological-organ kind ‘heart’, for example, can be introduced as a functional kind, but only derivatively on the basis of an underlying structural kind ‘heart+’, or a disjunction of such kinds.
Second, as a special application of this view, apart from its relevance (insofar as it will be able to stand up to criticism) to a unified understanding of functional discourse, I argue that it is of relevance to the cogency of currently standard argument against the reducibility of concepts and laws from the so-called special sciences to concepts and laws from underlying sciences, in particular, or ultimately, concepts and laws from physics. Central to this argument is the claim that concepts in the special sciences are functionally defined, and therefore multiply realizable out of more basic, ultimately physical, constituents, thereby making the derivability of generalizations governing these kinds from laws in terms of underlying kinds a hopeless task. However, if functional kinds can only be introduced through underlying structural kinds, to which we must have sufficient empirical and nomological access to ground them as kinds in the first place, the anti-reductionist argument loses most of its force.