The Inherent Normativity of Functions in Biology and Technology 107
technology, in contrast, seems much closer to the functional type: pump, knife, and so
forth. It is not commonly recognized in technology that, apart from the functional type,
another type is presupposed, which is identifi ed by the physical and historical features of
existing tokens of a functional type and by the design specifi cations associated with the
type. A defense of the importance of this “narrow” type concept for artifacts is given by
Soavi (this volume).^5
Precisely with reference to the distinction between token and type, it has recently been
argued by Davies (2001) that in fact the etiological theories are just as little capable of
assigning malfunctions, that is, assigning the function to do F to an item that is not capable
of doing F. His argument is specifi cally directed to an SE-type theory, which he presents
as follows (p. 194):
The selected function of type T in organism O in environment E is to do F iff:
(i) ancestral tokens of T in O performed F in E;
(ii) T was heritable;
(iii) ancestral performances of F enabled organisms with T to perform better in E than organisms
lacking T;
(iv) superior reproduction caused organisms with T to out-reproduce those lacking T;
(v) superior reproduction caused organisms with T to persist or proliferate in the population.
This matches the way the SE theory is phrased by Neander, with one crucial difference:
Neander’s SE defi nition assigns functions to tokens, not to types.^6 Davies claims that the
type T in clauses (i) to (iv) is a “success type”: it is the type corresponding to tokens that
actually performed F. And if in clauses (i) to (iv) type T is a success type, then it must
equally be in clause (v). What is explained by the attribution of the function F to T on the
basis of the etiological account of function, therefore, is the persistence and proliferation
in a certain population of organisms of a type T the tokens of which do in fact perform F.
Accordingly the function to do F can be attributed only to tokens of T that have the capac-
ity to perform F, and no token lacking this capacity can have the function. Ergo selected
malfunctions are impossible.
Davies’ argument, however, is based on a too narrow construal of the etiological theory.
Apparently he reads clause (i) in the defi nition extracted here as stating that every token
of T in O performed F in E, clause (iii) to state that ancestral performances of F enabled
every organism with a token of T to perform better in E than organisms lacking T, and
clause (iv) to state that superior reproduction caused every organism with a token of T to
out-reproduce those lacking T, and that only on this reading the conditions stated in clauses
(i) to (iv) explain the presence and proliferation of T in the population.
To invalidate Davies’ reading, take Neander’s version of the SE defi nition (1991:
174):