This is an animated but polite discussion between Jerry Fodor and Elliott Sober -- very interesting, and I think I understand Fodor a little better now (i.e., motivates what I believe to be his error --- other than the fact that he's worried about problematic teleological notions like function being necessary to natural selection as a theory). And I do believe I've come up with an argument that addresses Fodor's central argument more directly than Block's review (superb as it is). I think it may involve a confusion of the concepts of explanation and cause. Here goes.
Fodor demands of a theory that it provide general laws that cover all phenomena within its domain. Newton's laws, for instance, cover all bodies in the universe and explain the trajectories of their movements, regardless of their rich phenotypic (so to speak) diversity -- ellipses, parabolas, and so on. According to Fodor, natural selection is not a theory because it doesn't provide a predictive law for all phenotypes. It doesn't tell you that trait x (such as speed) will increase fitness (defined as greater reproductivity relative to alternative traits in a given environment) for all organisms. And when speed helps zebras yet slowness helps sloths, there is no overarching law that explains each case in the same way that Newton's laws explain both a parabola and an ellipse. Where Newton can give us something like f=ma, natural selection can only say that trait x will increase fitness when in fact it increases fitness.
As Sober points out, biologists can and do create models that predict phenotypic expression in given circumstances; if some organisms in a given ecology produce only males and others produce only females, and there are many more males than females, we can predict that there will be selection for female-producing organisms. So we have a local law here, in which change in a population is explained by selection for female-producing organisms.
Why does Fodor believe natural selection must involve a more general law to be explanatory? I'm not sure. On the one hand the motions of bodes are explained by Newton's laws. On the other evolution is explained by natural selection. The former gives us general laws predictive for all bodies; the latter gives us no general predictive laws. Is Fodor's point a semantic quibble about what we call a theory? Or is he saying that natural selection has no explanatory power? If the former, then it seems to matter very little--we could call it a meta-theory or anything we like. And the latter seems demonstrably not to be the case. After all, we are to take local models concerning selection as instances of the general theory. And clearly natural selection, because it is an explanation, rules out alternative explanations such as Lamarckianism (apparently with some exceptions) and intelligent design. Biologists are not busy developing models explaining changes of traits of a population based on God's aesthetic preferences.
Newton's initial insight, incidentally, was that all bodies are covered by a single principle, gravity, which explains both the dropping of an apple and the movements of the planets. Suppose Newton's subsequent fleshing out this principle led to mathematical laws that were not universal: that there were different and unrelated equations to explain the movements of different sets of bodies, and a very specific line dividing two domains of applicability -- let's say a spherical plane 50 miles from the surface of the earth (so that if I rocketed an apple into space it would go from behaving according to the first set of laws to the second). Incidentally, Newton himself entertains all sorts of interesting alternative theories in his Principia Mathematica to show that only an inverse square law creates a nice stable solar system and that other laws can be ruled out almost a priori; an inverse cube law, for instance, describes a death spiral; what's required is to find a Goldilocks force that in some cases, if we want to explain the revolution of planets, can just balance the centrifugal force caused by their motion without always simply sucking them in or letting them go on their merry way.
Would this mean that there is no theory of "gravitation"? Against this view, we might note that there are still some common concepts and central principles whether we're talking about inverse cube laws or inverse square laws: mass, inertia, etc. These things by themselves do not give us laws of motion (as the inverse cube example points out). Rather, they imply that in the cases of both apples near the earth and planets far away from it, we take their motions to be a function of mass even if those functions are different. If this were not the case -- and if close to the earth the paths for the movement of bodies were a function of mass and gravitation and outside the fifty-mile mark it were a function of the will of God, then I'd say we don't have a single theory. But in the case I've described, we can have two laws not related by a more general predictive law but involving the same terms and concepts--and so say that they belong to the same theory. That's because these terms and concepts point to a single causal mechanism, mass (avoiding all our thorny Humean epistemological problems, which apply equally to any scientific theory). And likewise with evolution, it seems clear that we take natural selection to state its general mechanism (by implicating reproductivity rather than than Lamarckianism) -- even if in both cases we suppose there are no predictive universal laws but only a universal mechanistic concept. And in fact, I think it's the case that the applicability of a single mechanism to a domain is a better criterion for something's being a theory of that domain than the universality of laws employing that mechanism.
In the case of natural selection, that mechanism is reproductivity. Fodor doesn't like the fact that it leads only to local laws and is triggered by differing circumstances (lions in the case of slow Zebras, soot in the case of white moths, etc.). But that doesn't make it, as he claims, "empty"--whether or not he wants to use the word "theory" to describe it. And so he makes the seemingly absurd claim that natural selection can't tell the difference between heart pumping and heart thumping (or zebra running and zebra thumping) as the survival-relevant trait. I think I finally understand this claim, and it is true insofar as we think that there must be a universal mechanism (lions, soot, etc.) acting on a particular trait in every case, when those mechanisms are obviously very local. But the relevant universal mechanism is reproductivity and the other mechanisms are varying triggers; just as we might think of mass as a universal mechanism and the throwing of an apple as a trigger. I can't give a theory in which the movement of apples is always explained by people throwing them, or in which a given trait is always explained by the presence of lions in the environment. But I still have a theory as to why it is that the apple moved in the way it did after it was thrown and why it is that zebras became fast (and why it is that throwing and lions are even relevant to what happened next). This theory is based on a universal mechanism in each case (mass and reproductivity). In the former case, we get predictive laws covering the entire domain of explanation and in the latter we do not.
There's a reason, by the way, that I chose mass as the relevant mechanism and not gravity, and I think this is relevant to Fodor's confusion. Mass is a mechanism in the sense that the presence of mass x causes movement in mass y (and vice versa). The presences or absence of gravity itself is not the relevant counterfactual here, because it's the law we're using counterfactuals to describe. To test whether body 1 caused body 2 to move, I add or take away body 1; not gravitation itself. Likewise, I'm not asking precisely whether natural selection caused evolution. I'm asking whether or not it explains evolution insofar as reproductivity causes changes in traits. Fodor seems to be conflating explanation and cause, or demands that a theory or general law be the latter rather than the former. But a general law describes something for which there is a mechanism; natural selection "itself" need not be able to "distinguish" relevant traits (heart pumping) and free riders (heart thumping) in order to be a theory. Rather, it must be the case that reproductivity is causally related to the one and not the other.
Seth Paskin says
The March episode of Julian Baggini’s podcast has an interview with Fodor (amongst other things)
Wes Alwan says
Thanks Seth — I’ll check it out.
Jim Lippard says
The updated version of Fodor and Piatelli-Palmarini’s book addresses your response (well, Sober’s version) on pp. 183-186. They express skepticism on the grounds that the number of “place-holders” in such law schemata would be enormously high (where they are not in the case of the physics example).
Wes Alwan says
Thanks — will check it out.
Jim Lippard says
Actually, although pp. 183-186 is relevant, there’s also a specific response to the theory schema suggestion on pp. 131-132.