Jerry Fodor complains that the criticisms of the objections he makes to the theory of natural selection in his paper Against Darwinism don't address his argument. If he is wrong then someone should say exactly where he has gone wrong. Fair enough - I will address his argument directly and show where I think that he has made a mistake. In summary, my disagreement with Fodor is that he only takes seriously two ways of making the distinction between being selected and being selected for and that, despite his claims to the contrary, there is a third and satisfactory way of making the required distinction.

The crucial question for Fodor is how we can distinguish between selection of and selection for. Darwin claimed that giraffes have long necks so that in times of drought they can feed on leaves that other, less fortunate, giraffes cannot reach. This increased their chances of surviving to breed, and so on. As necks grow longer, so the overall weight of giraffes increases. Both weight and neck length have been selected, but if we accept the Darwin's explanation (about which more later), then only length of neck has been selected for, and the concomitant increase in weight, although selected, is simply a side effect.

Fodor points out that the distinction between selection for and selection of is intensional: had neck length not been accompanied by weight gain (by decreasing bone density, say) neck length would still have been selected. If, though, weight increase was selected for and not neck length, then had weight gain not been accompanied by increase in neck length the heavier giraffes would still have been selected. The distinction between selection for and selection of is a modal distinction; to say that there has been selection for a property P is to say that the counterfactual 'those entities not having P would not have been selected' is true and to say that there has been selection of but not for P is to say that such counterfactuals are false.

The task then is to find a way of making the distinction between selection of and selection for that respects the intensional, modal nature of the distinction. Here is an example, deliberately non-biological, but non-intentional, that will show how the distinction can be made. Suppose that you are walking along the beach with a friend on a volcanic island the day after a violent storm. You come across a cave and you notice that there are unusually many small pieces of volcanic rock on the cave floor. You ask yourself: why are there so many small rocks on the cave floor? You look up and you see a crack in the ceiling of the cave, sufficiently large to let small rocks fall through. You therefore form the hypothesis that the reason why there are so many small rocks is because over a period of time they fell through the gap. Your friend, when told of your hypothesis, responds by saying that although that is a possible explanation, there are others. For example, during the storm the wind may have been blowing in from the sea, and the small pieces of rock were light enough to have been blown through the mouth of the cave. These hypotheses don't exhaust the options, nor are they mutually exclusive. Children might have been playing in the area the previous day and they might have collected small rocks and dumped them in the cave at the end of the day. Some of the stones could have been blown there, and some might have fallen through the crack in the ceiling. If we were really keen, we could test the hypotheses. We could ask the children whether they put any rocks in the cave, we could build a model of the environment and see whether either the blown-in-the-wind or fall-through-the-gap hypothesis is plausible.

But regardless of how we might determine which, if either, of the two initial hypotheses are correct, each provides a different explanation of why there are small rocks on the cave floor. If after having examined in detail all the various alternatives we decided in favour of one hypothesis over another, we would not feel that the explanation was in some sense incomplete - we might think that we have explained what really happened.

This kind of explanation could reasonably be called a sieve explanation. To sieve flour there are two things that we need to do: first, we need a sieve, and second we need to pour the flour through the sieve. Once we have done so, we have separated the sieved from the lumpy flour and we can then explain why there is no lumpy flour in the bowl - it has been sieved. In general, when giving a sieve explanation we claim (i) that there exists a condition that objects must satisfy that separates those objects in a larger set that satisfy the condition from those that don't and (ii) that the conditions must have been applied to the objects in the larger set. We can now say that those objects that have been sieved have been selected for the condition. The point of the earlier example was simply to illustrate that not all sieves are artefacts.

Sieve explanations are intensional (but not necessarily intentional) and they support the relevant counterfactuals. In the cave example, the blown-in-the-wind hypothesis it is true that had there been other objects suitably placed in the environment that were too big to fall through the gap in the cave but light enough to be blown around then some of these would have been blown into the cave; on the fall-through-the-gap hypothesis, it is true that had there been other heavier rocks around that were small enough to fall through the hole, then they would have dropped onto the floor of the cave. But as volcanic islands tend to be geologically pretty dreary, with lots of similar rocks around and not much else, examination of the rocks on the cave floor would not help in deciding amongst the two hypotheses.

The two explanations support different counterfactuals and select for different properties (weight against shape). However, although counterfactuals are supported, they do not directly invoke laws of nature - being sievable is not a causal property we would normally ascribe to flour. But in itself this is not a problem. Laws of nature support counterfactuals, but not all counterfactuals support laws of nature, and the original requirement was to find a form of explanation that supports counterfactuals.

Fodor is not unaware of this kind of argument, which he ascribes to Sober (Dawkins also talks of filters and sieves), but he only discusses this proposal in a footnote and gives it pretty short shrift. He writes (about Sober's pebble sorting machine):

...what grounds the counterfactuals in Sober's example is the structure of the mechanism; given how it works, it lets the round pebbles through but no others; one's intuitions about which trait is  selected for follow not from what laws of selection per from mechanics. Notice, for example, that whereas competition plays a central role in the explanation of every bona fide Darwinian selection, it plays no role at all in explaining how Sober's machine sorts for round marbles.

Note first that, as I pointed out earlier, selection for traits does not have to follow from laws of selection; they just have to support the appropriate counterfactuals. Second, although Fodor is right in claiming that competition plays a crucial role in Darwinian selection, this does not mean that it has to play a crucial role in the explication of the distinction between selection for and selection of. Sieve explanations are not local to the theory of natural selection or any other theory that involves competition.

Although sieve explanations have a temporal aspect, they are ahistorical. But it is easy to see, in outline anyway, the role that sieves play in the theory of natural selection. Assuming once again Darwin's explanation of why giraffes have long necks, there is a sieve for neck length in giraffes that filters out those giraffes that cannot reach the leaves at the top of the trees. Without competition for resources amongst giraffes the existence of such a sieve is irrelevant to natural selection, for in such a case how well-fed a giraffe is would have no differential impact on the number of offspring produced. But if there is such competition, and if further the sieve is in operation over a number of generations, and neck length is heritable, then there would be upward pressure on neck length leading to the fixation of the phenotype that the theory of natural selection is supposed to explain.

Adaptationist explanations are historical explanations that do not depend on laws of selection in Fodor's sense, but do depend on sieve explanations that support counterfactuals and are thus able to make the distinction between selection simpliciter and selection for. They are also highly context-dependent and post hoc, but as far as I can tell no-one is disputing this - for example, the familiar claim that if we were to rewind natural history then we would get a different story, is an expression of a general recognition of the contingency of the actual historical account. The context-dependency and post hoc nature of adaptationist explanations are only problematic if they have to be based on laws of selection construed as causal laws connecting traits to fitness, and laws of selection in this sense are unnecessary - sieve explanations provide an alternative.

The discovery of suitable sieves is a complex matter - possible explanations are not necessarily true explanations..Darwin's theory of giraffe neck length could well be wrong. Craig Holdrege ('The Giraffe's Short Neck' , http://www.natureinstitute.org/pub/ic/ic10/giraffe.htm) writes: "The giraffe's neck carries out a variety of functions - it allows feeding from high branches, serves as a weapon in males, brings the head to elevated heights that give the giraffe a large field of view, is used as a pendulum while galloping, and so on". On Holdrege's view, Darwin got the explanation wrong, but the difficulty of finding the right explanation does not mean that there is no explanation to be had, and indeed Holdrege's alternatives are themselves based on empirical observation.

Those of us who share Fodor's dislike of evolutionary psychology might share my concern that if evolutionary psychology is to stand or fall with the theory of natural selection in general, then it will stand.