532 Tim Lewens
of the circulatory system to bring nutrients to cells, the heart’s contribution is
pumping.
The CR account, at least the entry-level version, has trouble making sense of all
three of the connotations of function ascription listed above (although see [Davies,
2001], for an attempt to defend an augmented CR account). First, CR functions
are explanatory, but they give the wrong kind of explanation. If the function
of a token heart is one of its effects, then while that explains the capacities of
the systems that contain the heart, it does not explain the presence of the token
heart itself.^1 Second, the CR account cannot make sense of the function/accident
distinction. In the context of the containing system of hospital treatment practices,
the heart does have the function of making diagnostic sounds. Third, the CR
account cannot make sense of malfunction; if having a function F demands having
effect F, then malfunctions are impossible, for objects malfunction when they have
function F yet are unable to yield effect F.
We can add a fourth problem to this list, the problem of promiscuity. Remember
that biology seems to be quite distinctive in its use of function talk, so whatever
analysis we give of biological functions should explain why this vocabulary is used
in biology in a way that does not feature in the physical science. But it seems
hard to restrict CR functions to biology: any item — whether organic or inorganic
— that makes a contribution to some capacity of a containing system will have a
CR function. To return to our opening example, in the context of an explanation
of the breakdown of atmospheric ozone, CFCs will be credited with a function.
But chemists do not use heavy function talk. So if heavy function talk is what
we are trying to understand, the CR account does not explain its unique role in
biology. In fact, most current writers on functions are pluralists [Godfrey-Smith,
1993]: they think that some, but not all, biological function talk is heavy function
talk. Sometimes when biologists speak of functions they are doing nothing more
than talking of the role of some item in a containing system. Functional analyses
of these kinds seem to occur in physiology and biomechanics. This is light function
talk. The CR account seems well placed to make sense of light function talk, but
when biological function talk is heavy, the CR account is inadequate.
When we look to artefacts, it seems that we can explain our three connotations
by appeal to the simple view that the function of an object is just whatever its
maker (or perhaps its user) intended it to do. If function claims for artefacts are
claims about their intended effects, then we can see why function claims tell us
why an object is there (they tell us the intention that led to its appearance), we
can see why functions need to be distinguished from accidents (not every beneficial
effect will have been intended), and we can see why functions express norms (an
object that cannot do what it is intended to do is malfunctioning).
The selected effects account takes the bare bones of this approach — functions
(^1) Note that Cummins is well aware of this feature of his view. It does not bother him, because
he denies that the presence of any token heart is genuinely explained by the pumping of blood.
In other words, he doubts that the first connotation of heavy function talk is one a good theory
should seek to justify.