106 Maarten Franssen
theory of function applies only to biological items—at least, as long as the general skepti-
cism regarding the operation of natural selection in the development of social systems
lasts. All remaining theories, however, among those introduced here claim to be general
theories that apply to the entire spectrum of function attributions in (scientifi c) practice.
I disregard here all variants of the SE theory that trade past selection for current fi tness
or remote-past selection for recent-past selection, since these differences are, I think,
immaterial to the arguments I develop in this chapter.
7.3 How Function Theories Account for the Normativity of Functions
The twofold division of function theories sketched in section 7.2 coincides with the sharp
divide concerning the possibility of malfunction that is mentioned in the introductory
section. Only in the etiological theories—PF, SE, and variants—can it occur that it is the
function of an item x to do F while x is incapable of showing the behavior that counts as
performing F. This is so because on these theories the function of an item is defi ned by
reference not to the item’s causal role but to the causal contribution of related items,
its—technically defi ned—ancestors. All other theories, defi ning the function of an item in
terms of the item’s causal role, cannot grant an item that fails to show the required physical
behavior a function, regardless of whether other similar items are capable of showing this
behavior and are therefore attributed the corresponding function.
As far as the functions of biological items are concerned, the PF and SE theories attri-
bute functions, and therefore also malfunctions, to items that are identifi ed as tokens of a
type. By malfunction, then, the following situation is meant: 1) x is a token of the type X,
2) x is attributed the function to do F at least partly on the basis of its being a token of
the type X, and 3) x is not capable of showing the behavior by which tokens of X normally
or usually perform F.^4 Any theory that refers to causal interactions in defi ning func-
tions—and both the CR and the PF theories do so—must attribute functions to tokens,
since only tokens are causally effi cacious. Types can be assigned functions in a derived
sense, in terms of the functions of their tokens.
Care must be taken to distinguish between the type that an item is presumed to be a
token of, in receiving a function on account of a particular theory of function, and the
functional type itself, that is, the type that is defi ned as consisting of all tokens that have
this particular function. The type to which the theory of function refers cannot be the
functional type, on pains of circularity: in order to know whether x is a token of the func-
tional type XF, it must be known whether x has the function F, but in order to know whether
x has the function F, it must be known whether x belongs to the functional type XF. The
dominant type concept in biology is indeed the one that theories of function presuppose,
identifi ed by a combination of structural (morphological and physiological), historical
(developmental), and comparative (homology) features. The predominant type concept in