110 Maarten Franssen
making such functions special in the sense of allowing a normative reading, a CR-type
theory can argue that if the special function of an item x is to do F by showing behavior
B, then it not only performs F but should perform F as well. A CR-type theory cannot,
however, say of another item x′ of the same type that it is not capable of behavior B, that
this item should still show B or should perform F, since x′ does not have the function to
do F in any CR-type theory. And it may be questioned whether it makes sense to say that
x should be capable of B-ing if it is not at the same time true that x′ should be capable of
B-ing.
7.4 Being Supposed to and Ought to as a Way of Expressing Justifi ed
Expectations
The sketch in section 7.3 of the way the etiological theories handle the supposed
inherent normativity of the functions matches the way Millikan introduces the connection
between normativity and the notion of “proper functions” (1984: 17): “Having a proper
function is a matter of having been ‘designed to’ or of being ‘supposed to’ (impersonal)
perform a certain function. The task of the theory of proper functions is to defi ne this
sense of ‘designed to’ or ‘supposed to’ in naturalistic, nonnormative, and nonmysterious
terms.” So what the PF theory sets out to do is not to give an account of function
from which normative statements containing “is supposed to” and its relatives can subse-
quently be derived; it is to give a direct account of our intuition that some entities can be
supposed to show certain forms of behavior, something that we alternatively express by
attributing a proper function to them. What the PF theory is meant to do is to articulate
this notion of “proper function” in purely naturalistic terms. By giving an account that
singles out, among the many things that a particular item x does, one specifi c behavior B
by which it performs a function F, the theory does not justify normative statements of the
form “x ought to do B,” or state that therefore we can say that x ought to do B or to do F.
The theory merely recovers a function concept that is intuitively associated with these
normative statements, but this recovery does not extend as far as these normative
aspects.^11
There is, however, more than enough reason to ask for a justifi cation of such normative
statements. It may be intuitively all right to say of a particular item existing in the material
world—an organ or a trait or a form of behavior or a device—that it should do something
or is supposed to do something or ought to do something. But can we really make sense
of such statements? Certainly not literally. Only of human beings can it be said that
they should do or ought to do things.^12 One cannot prescribe anything to a mere material
object; to do so would be a classic example of a category mistake. It makes no sense to
say that an electron ought to move in accord with the Maxwell equations. But if such
“ought to” statements are not prescriptive, then what are they? How should we understand