98 Peter McLaughlin
out of the cistern; he needs the key to open the shed, to get the tools, to fi x the bicycle,
to ride to town, to buy fl our, to bake a birthday cake for his cat Findus. In this example,
the chain of relative ends defi nitely stops some place with a different kind of purposive-
ness or means-end relation; it comes to an end with a benefi ciary. The birthday cake baked
by old Pettersson was good for the cat Findus, and that stops the regress: the cat need not
be good for anything. This second kind of purposiveness cannot be iterated. It denotes a
relation to something whose good is not merely relative to its contribution to something
else. The cat Findus may also be good for catching mice or keeping company, but he has
a good of his own independent of whatever usefulness he might also have for the farmer.
When we view a causal chain as a series of means and ends, we presuppose something
that stops the regress, something that has a good. And this applies whether it is an inten-
tional agent, an organism, or simply anything that can be said to have interests—whether
or not it consciously takes interest in them. We presuppose an entity somewhere down the
line which has some kind of interests that (ceteris paribus) ought to be served.
Although every effective means to an end is also the cause of an effect, causes only
have effects. It’s not that they ought to have their effects—they just do, or they aren’t
causes at all. But when we view causes of effects as means to ends we presume that the
means to the ends are supposed to facilitate these ends.
This is our fi rst candidate for a source of normativity in function ascriptions.
6.2 Parts and Wholes
A series of cause-effect relations and means-ends relations can be viewed not only as a
process (chain) but also as a system (hierarchical structure). Just as a system can be seen
to be causally dependent on its parts, so too the parts can be seen as means to the end of
the whole. In hierarchical systems, we can also to a certain extent iterate the means-ends
relation: part A of system B contributes to some performance of B, which contributes to
some performance of larger system C (which contains B), and C in turn is part of D and
makes a contribution to what D performs, and so on. But we rather quickly run out of
containing systems: the valves in the veins contribute to blood circulation, which contrib-
utes to the metabolism of the organism, which may be taken to have a role in the ecology
of a particular region, but after that we have trouble fi nding an appropriate larger contain-
ing system, the performance of which is supported by the ecosystem. Nonetheless, the
primary use of functional ascriptions is actually in hierarchical systems where parts are
ascribed functions for the whole. In some cases the regress is stopped arbitrarily: a per-
formance of the system is good for some external agent. For instance, the governor of the
steam engine contributes to the engine’s ability to deliver regular power and this is good
for the factory owner. A second kind of case is where the benefi ciary that stops the func-
tional regress is actually the function bearer’s containing system itself. Here the containing