Realism and Artifact Kinds 193
Usually what we mean by “o is used for F” is that o is normally used for F, but what
does “normally” mean? In the debates over the notion of “normal,” typically quantitative
and qualitative interpretations are distinguished. In this case the most likely quantitative
interpretation seems to be the following:
a. Most of the times that o is used, it is used for F.
A qualitative interpretation seems to be more diffi cult to give. A plausible possibility is
the following semiqualitative criterion:
b. Most of the times that o is used by competent people, it is used for F.
Unfortunately both (a) and (b) are false or, more to the point, do not meet the common
classifi cation of artifacts into functional kinds.
We seem to be perfectly comfortable with the thought that something is still a chair
(whatever “being a chair” means) even if it is used most of the time as a small stepladder.
There is a certain intuitive resistance to considering such an object to be a stepladder rather
than a chair. Analogously in the case of (b), it is perfectly conceivable to consider cases
where we—supposedly competent users—will use the very same chair we are sitting on
as a stepladder for years without forming the opinion that it is or has become a
stepladder.
I take it to be symptomatic that we have linguistic instruments for distinguishing
between instances of something being a chair and instances of something being used as a
chair. There are circumstances in which we are willing both to assert that something is a
chair and to deny that it is normally used as a chair, and yet other circumstances in which
we deny that something is a chair even if it is commonly used as such. Indeed these lin-
guistic facts can be viewed merely as cues for the thesis that we do not normally classify
artifacts according to the function they perform or are used for.
Another possibility is to take o in (2) to range over types instead of tokens. This implies
that we need some sort of paraphrase: a type is an abstract entity, so it does not seem to
make much sense to say that O is literally used for F or that O is normally used for F. We
can consider the following paraphrases:
c. Most of the times that tokens of O are used, they are used for F.
d. Most of the times that tokens of O are used by competent people, they are used
for F.
These two solutions do endure. The critics moved to (a) and (b) because according to (c)
and (d) a certain token is an artifact of a certain kind if it is a token of a type whose tokens
are most frequently used for F. So it is perfectly possible for a single chair to be normally
used as a stepladder but to nonetheless not be considered as a stepladder as long as other
tokens of the same type—or at least most of them—are used as chairs. The problem is to