Darwin has taught us how to translate future directed, teleological biological ‘just so’ stories into conventional past to future causal explanation. Contrast:
Why do giraffes have long necks? To eat higher, less competed for, leaves.
Why do humans have noses? To balance their spectacles on.
In the former, we can tell a story which runs, roughly: Chance mutation led to a long necked giraffe-ancestor which was thus able to eat higher, less competed for, leaves and thus able to be fitter for both predator survival and reproduction, thus leading to more long necks in its descendents who are thus able to eat higher, less competed for, leaves and thus...
In the latter, we can tell no such general story of why noses have helped via spectacle balancing and thus that isn’t the biological function of the nose.
In general, this approach can be described thus:
A natural function of a biological mechanism is an effect of the mechanism that explains the existence, maintenance or nature of the mechanism via the same essential process (whatever it is) by which prototypical nonaccidental beneficial effects... explain the mechanism which cause them...
It turns out that the process that explains the prototypical non-accidental benefits is natural selection acting to increase inclusive fitness of the organism. [Wakefield 1999: 471-2]
There is some debate amongst philosophers about how precisely to define a biological function but Wakefield’s account captures the central idea. Roughly speaking, a biological function which a particular trait of an organism exemplifies is a function that explains the evolutionary success and survival value of that trait [cf Millikan 1995].
Crucially, biological functions are distinct from dispositions (ie what biological systems are disposed to do). Engineering limitations might cause the actual behavioural dispositions of a trait to diverge from the biological function it thus only partially exemplifies. The divergences might themselves be life threatening and play no positive part in explaining the value of the trait. The best explanation of the survival of that organism and those like it cites the function which helped propagation or predator evasion, for example, and not those aspects of its behavioural dispositions which diverged unhelpfully from it.
One example of this is the blindspot of the human eye. The eye is disposed to give awareness of objects in front of it except in its blindspot. But that is not its function. That is, its function is not to give awareness of objects in front of it except in its blindspot but rather to give awareness of objects in front of it. It is just that it does this a bit badly.
This point is sometimes put by saying that what matters is not what traits or dispositions are selected, but what function they are selected for. The distinction between ‘selection of’ and ‘selection for’ can be illustrated by the example of a child’s toy [Millikan 1995, Sober 1984]. A box allows objects of different shapes to be posted into it through differently shaped slots in the lid. The round slot thus allows the insertion of balls, for example. It may be that the actual balls allowed through or ‘selected’ in one case are all green. But they are selected for their round cross section and not their green colour.
Millikan stresses the fact that the biological function of a trait may be displayed in only a minority of actual cases. It is the function of sperm to fertilise an egg but the great majority of sperm fails in this regard.
But whilst biological functions and thus biological functional explanation need not presuppose any teleological future to past causal influence, it is not clear that it can reduce to causal event-explanation. To see this, think of this argument:
According to Wakefield, on this model, “the heart exists for the purpose of pumping the blood in the sense that past hearts having this effect causally explains how hearts came to exist and be maintained in the species and the genesis of the heart’s detailed structure” [Wakefield MS: 28]. The most basic difficulty here is that past hearts’ pumping the blood figure in all sorts of causal stories, stories in which agents die young, or agents do not reproduce, or agents reproduce but defectively, and so on. It is not the case that hearts pumping the blood have simply caused hearts to exist. So hearts pumping the blood does not causally explain how hearts came to exist. This cannot therefore be the sense in which the pumping of the blood is a functionally explicable activity of the heart. [Megone MS: 20]
This is not right because Megone illicitly runs together causal explanation and ‘simple’ causal relations. By contrast, Wakefield, like Millikan, relies on specifying the function of systems through what best explains their continued existence. Thus Wakefield can reply that the divergent causal relations do not explain the heart’s existence. Recall again the distinction between the dispositions of a system and its natural function. Its dispositions are also genuine causal consequences of the system’s nature but do not explain its existence.
But this response suggests a problem if one aims to show not just that biological functional explanation does not presuppose future to past causal influence or the presence of a designer but rather to show that biological functional explanation is simply past to future causal event explanation.
Recall, that what singles out a proper function is what best explains the continued existence of biological trait. It is this which provides the determinacy which underpins the distinction between behaviour which accords with proper function and that which does not. But this notion of best explanation turns on seeing an explanatory pattern in the myriad causal relations which goes beyond the resources of simple causal description of biological happenings, beyond the data of evolutionary success and failure. That austere description lacks the resources to select between different dispositions as being ‘correct’ or ‘incorrect’ realisations of a function.
One might try to put this point by saying that whilst the function that the behavioural dispositions of a biological system partially exemplifies has to be consistent with that behaviour, it is not determined by it. In fact however, the function does not even have to be consistent with the behaviour, because the behaviour only partially exemplifies the function. But given that the causal transactions of evolutionary history provide some constraint in the ascription of function, there are still an infinite number of different functions that meet that constraint. These include all the functions which only diverge at a still future date. Systems with those functions would have been just as successful in the past. Thus the data themselves do not determine a particular function even if we find it natural to interpret the data in particular ways.