100 Peter McLaughlin
that fi xes the blade to the handle still has the same function as before and might be thought
to be supposed to do just that independent of the new goals of the agent. A functional
analysis is in fact interesting only in a complex and hierarchically ordered system where
the parts are more or less tailor-made for a particular role. In such a system we may well
be able to identify only one function of a part that fi ts the hierarchical structure—though
this may be due only to lack of imagination of the functional analyst. A cuckoo clock as
a system may be viewed from the perspective of its capacity to act as a counterweight in
a balance, and thus each of its parts may be viewed as contributing to that capacity and
having this contribution as its function. But in such a case each of the parts contributes
only by its individual weight, not by its special structural properties or by its integration
into the organization of the complex hierarchical system. It is indeed hard to imagine what
capacity of the cuckoo clock, other than the uniform motion of the hands, the escapement
mechanism might contribute to for which the complex organization is at all relevant. But
with biological traits the tailor-made character is often much less clear.
These last two aspects of part-whole relations—that organic wholes, as opposed to parts,
tend not to have functions at all and that the functions of parts of artifacts are much more
resistant to arbitrary change than are the functions of the artifacts themselves—however,
may be only obliquely relevant as sources of normativity and rather indicate a source of
confusion in our thinking about functions.
6.3 Types and Tokens
Normativity is also already introduced by the type-token distinction used when character-
izing function bearers. Functions are ascribed to items that instantiate a particular type
that has the function.
The textbook view of the type-token distinction can be misleading. C. S. Peirce uses
the terminology of types and tokens to make certain kinds of distinctions, for instance,
distinctions between the letter A and any particular concrete way of writing it. Here are
fi ve tokens of the letter type A: A A A A A. Often the type-token terminology is introduced
to distinguish between sign types and their concrete instantiations or between token sen-
tences and their content (propositions). There is no obvious connection of this notion of
type and token to normative considerations. But there is also an older philosophical use
of the terminology of types—systematized by William Whewell (1840: 477) with refer-
ence to the traditional distinction between habitus and privation. This more traditional
concept of type, analyzed in detail by Hempel and Oppenheim (1936), sees types as
embodying norms. When we view individuals as tokens of a type, rather than (say) as
elements of a set or members of a class, we have opened up the possibility of introducing
normative considerations. Whereas individuals are either members of a class (elements of
a set) or they aren’t, tokens by their very nature can instantiate a type better or worse. Any