Functions and Norms 99
system is viewed not just as the next hierarchical level but as a whole, as the end of the
encasement, as something that displays a certain (perhaps purposive) unity or integrity.
Normally when we pursue the regress beyond the level of the organism, we change our
perspective and say that ecosystems or some other equilibrium systems are what we
happen to be interested in; or we assume that the equilibrium of the system is good for
the organisms in it.
This is our second candidate for a source of normativity in function ascriptions: we
implicitly view the containing system as “more” than a mere aggregate or structure—as
a whole, a hierarchically organized system. The containing system as an organic whole
seems to have a good of its own like the regress stopper of the chain of means and
ends.
In the relations of parts and wholes there also arises a peculiar asymmetry between
considerations of organisms and artifacts. In artifacts both parts and wholes have functions
in the same sense: the governor of the steam engine has a function in relation to the system
of which it is a part, and the steam engine itself serves (or was intended to serve) a purpose
or function for someone or something; the gears of a fl our mill have functions, just as the
mill does. In organisms, on the other hand, parts or organs or traits have functions for the
organism independent of the question of whether the organism itself is thought to have a
function for something or someone else. Something that has a good of its own can of
course also be viewed instrumentally. Just because an entity stops one particular instru-
mental regress doesn’t mean it cannot also be a mere link in the chain or level in the hier-
archy of another regress. The fact that oats are good for horses, not just for the owners of
the horses, does not prevent horses from being good for riding or pulling a plow and thus
being useful to their owners. The two instrumental views may however also come into
confl ict with each other as can be seen in the standard example that eating lamb chops is
good for the sheepdog qua dog but bad for the sheepdog’s owner. That is, although we
may view organisms as artifacts, and by breeding and training even make them artifacts,
they nonetheless retain their ability to stop a functional regress.
Unlike the case of artifacts, which can be good or bad, worse or better, it makes no
sense to ask whether Fred is a good antelope or Linda is a substandard tapir. Now, a good
motor will help to make one car better than another and a good set of teeth will help to
make one sheepdog a better guard than another, but a strong heart or good teeth will not
make one hippopotamus a better hippopotamus than another. A hippo may be better off
with well-functioning organs, but it itself has no function: it has a good. And an organ or
trait can be good for a sheepdog qua dog without being good for the sheepdog qua guard
dog.
Even strongly intentionalistic theories of artifact functions—which allow functions to
come and go with mental events, even without any physical changes in the function
bearers—balk when it comes to the functions of parts within a whole (McLaughlin 2001:
ch. 3). These do not come and go so easily. Even if I turn my ax into a crowbar, the wedge