The Inherent Normativity of Functions in Biology and Technology 109
For that to be the case, heart cannot just mean “any organ that circulates the blood through
an organism’s body,” and pump cannot just mean “any artifact that circulates a liquid
through a system.” As mentioned, biological types can be defi ned and are defi ned in mor-
phological and developmental terms. It would seem that for artifact types we could refer
exclusively to what a particular artifact was intended to be when designed, but structural
criteria are necessary here as well to keep the infl ation of such types by “wishful design-
ing” within bounds; see Thomasson (2007). As a result, structural and historical charac-
teristics also codetermine artifact type defi nitions. A knife is not just anything designed
for cutting. The basic point remains that any theory of function that refers to types must
be able to defi ne these types in a noncircular way.
The CR-type theories can, therefore, account for part of what we mean by attributing
malfunctions. This part, however, contains nothing normative. Malfunction is judged to
be a normative notion only when we take it to include saying, for example of a particular
heart, that although it does not in fact contract and thereby circulate the blood, it is sup-
posed to do so or ought to do so. The PF-type or etiological theories do not account for
this normativity, however, by showing that such statements can be derived from its
theoretical framework. Indeed the theory or defi nition contains exclusively descriptive,
naturalistic terms,^9 so no statement using terms from the intentional vocabulary—like is
supposed to—or more narrowly, from the normative vocabulary—like ought to or should—
can be implied by it. What the theory does is single out, from all the behaviors that a par-
ticular item is capable of, or from all the effects of a particular behavior, one behavior or
effect that is granted a special status. This behavior or effect, then, can be considered the
special thing that the item is supposed to do. For example, the proper function of the nose
is to allow air to reach the lungs, even when the mouth is engaged, and to have this air
pass over the smell receptors at the same time, not to support a pair of glasses. The former
is what the nose should do (and a snotty nose is therefore a temporarily malfunctioning
one), but the latter is not, and there is nothing wrong with a nose that does not. The proper
function of a drinking glass is to hold liquid and allow that liquid to be drunk with the
mouth, not to be smashed against the wall to express rage or anger or joy. The latter is
not what a glass should enable one to do, but the former is, and a leaking glass is a mal-
functioning glass, although it is perfectly good for being smashed against the wall.^10
To bring out the contrast between the PF-type and CR-type theories most clearly, it
should then be noted that the PF-type accounts of normative functions are based on two
elements. First, they interpret “function” as a special, inherently normative notion, and
accordingly narrow down the attribution of functions to items to (usually) one particular
“proper” function per item. On the CR-type theories, in contrast, an item can have many
functions. Second, the PF-type accounts attribute functions to items incapable of the cor-
responding physical behavior. A CR-type theory can try to “approximate” the marking of
particular causal-contribution functions as special by being adaptations, as is suggested
by Krohs in his account of the SD theory (2004: 97). If this is necessary and suffi cient for