Functions 533are effects of objects that do not owe themselves to chance, and which are bestowed
by some external process — and adapts them to the biological realm. Natural
selection takes the role of a designer. If the function of an artefact is what a
designer or user intended it to do, then it is only a short step to say that the
function of an organic trait is whatever it wasselected for. To say that some trait T
was selected for F is to say that T increased its frequency in a population over some
alternative because it had effect F [Sober, 1984, 100]. To say that the dark pigment
of the wings of the peppered moth was selected for camouflage is to say that moths
with dark wings increased their frequency over moths with lighter wings because
the dark wings provided camouflage. So now we can see how the SE account
makes sense of our three connotations of function ascription, and how it does it in
a way that closely follows how we grounded artefact functions. According to the
SE account, function claims explain the presence of the functionally characterised
item because they are, in spite of their present-oriented appearance, historical
claims about the selective history of the trait. Functions are distinguished from
accidents because some of the fitness-enhancing effects of a trait type will be
too new to explain the current frequency of that trait type in the population.
And the reference to natural selection enables us to ground the normativity of
function statements, because a trait can be a token of type whose presence is
explained by selection for some effect, even though the trait in question does not
have that effect. That is how a token heart that is unable to pump blood can
still have pumping blood as its function. Finally, the SE account does not face
the promiscuity objection levelled at CR functions. Since the SE account makes
functions dependent on selection, the SE account restricts functions to systems
that undergo selection, and hence explains why biologists, but not physicists or
chemists, make use of this kind of talk.
One further point of clarification is worth making here. In order to cut off
worries that by looking to a trait’s history of selection, the SE account makes
the loss of function impossible, most theorists these days also stipulate that the
selective regime that determines a trait’s function is that which occurs in the recent
history of a trait. The most widely adopted theory of functions these days is some
variant of the ‘modern history’ account [Griffiths, 1993; Godfrey-Smith, 1994].
To illustrate a last strength of the SE account I want to turn back to Kant.
Kant worries that a purely mechanistic explanation of organic form would always
leave too much of an air of ‘contingency’ to that form:
So where the structure of a bird, for instance, the hollow formation of
its bones, the position of its wings for producing motion and of its tail
for steering, are cited, we are told that all this is in the highest degree
contingent if we simply look to thenexus effectivusin nature, and do
not call in aid a special kind of causality, namely, that of ends (nexus
finalis). [Kant, 1952, 4]Consider a house. One could give a mechanical explanation for how the house
comes to be built by citing the positions of all the bricks and window frames