Of Minds and Language (8 page)

And that leads us onward somehow. Peirce argued that if you keep on this track indefinitely, you eventually reach truth. He thought that truth is sort of defined as the limit of scientific experimentation, and he gave a very bad evolutionary argument for this. He said that evolution has adapted us to find the right kinds of solutions to natural problems, but that cannot possibly be true. There is nothing in human evolution that led people to figure out quantum
theory, or classical mechanics, or anything, so that can't be right. That is just one of the worst kinds of pseudo-Darwinism. So maybe what it is leading us to is something totally wrong, and if somebody is looking at this, say some Martian with different cognitive structures, they could see we are just asking the wrong questions. We are not asking the right questions because they are not within our range. We can't ask those right questions; we aren't built for it. And if we can ask them, we can't answer them. So take the questions, this first-person perspective thing, which is a big issue in philosophy: what is it like to be a bat? (a famous article);
³³
what is it like to be me? There are no sensible answers to those questions. I cannot tell you what it is like to be me. If something has an interrogative form and there are no sensible answers to it, it is not a real question; it just has an interrogative form. It is like “How do things happen?” You know, it sounds like a question, but there is no possible answer to it.

So I don't think these are even questions. You can give a naturalist interpretation of such matters, and maybe there is a right question and we just can't formulate it, because we're just not built that way. So if there is one, we may not find it. That is Peirce's concern. Well, to get back to your question, I can't add anything to that, and I don't think anybody has added anything to it in a hundred years. In fact, they haven't even looked at it. The term abduction has been picked up, but it is used for something else. It is used for best-theory construction or something like that, whatever that means, but that is just rephrasing the question.

It seems to me the answer has to come from some kind of study of what this organ is, this science-forming organ. Now Sue Carey, whom I mentioned, has been trying to show that it is just the natural extension of ordinary, common-sense figuring out what the world is like (Carey 1985). But that seems to me to be extremely unlikely. Of course it is interesting stuff, but it seems to me to be going in the wrong direction. Whatever this crazy thing is that scientists do, it seems to me very much disconnected from sort of finding your way in the world. I mean, people talk about it, the search for symmetry – there is a famous book about that
³⁴
– and Galileo talked about how Nature has to be perfect and it is the task of scientists to find it. You do have these guiding intuitions and so everybody follows on, more or less, but they don't seem to have anything to do with sort of getting around in the world. So it is a serious open question, and it could be – it is an empirical question, in principle: what is the nature of Peirce's abductive instinct? Maybe somebody can tell me something. A lot of you know more about this than I do, but I don't know of any work on it, philosophical or
empirical or anything else. It has just kind of been left to the side, and again, as far as I know, Peirce's essay wasn't even discussed for about sixty years.

D
OVER
: You have a long-held view that the human capacity for language is an evolved biological system and, as such, there has to be a genetic basis for it – no different in kind from any other specific feature of human biology. I don't think anyone would want to refute that, but I sense, if I understood you correctly, that you want to go beyond that.

Within the minimalist program, my understanding of which is very shaky indeed, I sense you want to bring forth something beyond the genes. That is, we have what you call principles of natural law. However, I want to point out that the whole thrust of modern-day genetics is going against such ideas of laws of form and principles of natural law, or however you want to phrase it. And indeed, in a very revealing way, Alan Turing was actually wrong with this approach. Just take this one example. He showed mathematically that if you consider a larva of an insect simply in terms of physical/chemical principles of reaction and diffusion amongst free-floating molecules, then the system falls naturally into a series of standing waves of molecular concentrations underlying the appearance of discontinuous bands of bristles along the larval axis (Turing 1952). But we now know from genetic analysis that the positioning of each band is independently determined by a band-specific handful of genes that are networking with each other in a regulatory manner, and if you mutate one or other gene you might knock out, say, bristle band 3 or band 7 and so on. However, knocking out a band does not entail a reorientation of the remaining bands according to physical principles of organization in the remaining larva as a whole – in other words, there are very local molecular interactions making each band independent of the rest, and the ensemble approach of field theory based on physical/chemical principles doesn't seem to come into it. Now we can show this over and over again for almost any aspect of phenotypic form and behavior you'd wish to consider. The evolutionarily constrained yet flexible network, seemingly unique in operation in biology, is very significant, as I shall show in my talk. Biological diversity is a consequence of local differences in the combinatorial usage of modular and versatile genes and their proteins that often stretch back to the origin of life. But nothing seems to be obeying laws of form, out of reach of the genes.

C
HOMSKY
: That can't be. I mean, take, say, the division of cells into spheres, not cubes. Is there a gene for that?

D
OVER
: Yes, of course there is. It could be your worst nightmare (!) for there are tens upon tens, if not hundreds, of genes directly responsible for very wide-ranging
differences in the shapes, sizes, numbers, divisions, life spans, senescence, functions, and behavior of the several hundred types of cells in our species. Cells are not soap bubbles. There are constraints of course but these are a matter largely of history not of physics, over and above the obvious physics/chemistry of molecular contacts.

C
HOMSKY
: No there isn't such a gene. Cells form spheres because that is the least-energy solution. In fact, it has always been obvious that something is channeling evolution and development. It doesn't go any possible way; it goes in the ways that the laws of physics and chemistry allow. Now, Turing's particular proposal about reaction-diffusion, giving discreteness from continuity – first of all, I think it has been partly confirmed, for angelfish stripes and in other instances. But quite apart from that, whether he had exactly the right proposal back in 1952 doesn't really matter. His general formulation just has to be true. And it is presupposed by all the work you are talking about.

If particular combinations of proteins and molecules and so on do particular things, that is because of physics and chemistry. The only question is to try to discover in what ways physics and chemistry determine the particular evolution. So again, we are getting into your domain, which you obviously know more about, but take the evolution of the eye. Let's say Gehring's more or less right, okay?
³⁵
What happens is there is a set of molecules, rhodopsin molecules, which happen to have the physical property that they turn light energy into chemical energy? One of them might randomly migrate into a cell. That, according to him, is the monophyletic origin of eyes along with a conserved master control gene, and maybe even everything phototropic.

Everything that happens after that has to do with the intercalation of genes and certain gene sequences, but that all happens in particular ways because of physical law. You cannot intercalate them in any crazy way you dream up. There are certain ways in which it can be done. And he tries to conclude from this that you get the few kinds of eyes that you get. Well, all of that is presupposing massive amounts of maybe unknown physical and chemical principles which are leading things in a certain direction, kind of like cell division into spheres. I mean, there may be a couple of genes involved, but fundamentally it is physical principles.

Now how far does that go? Well, I'm no biologist but I don't agree with your conclusion, or that that is the conclusion of modern genetics. In fact, the whole evo-devo development over the past twenty to thirty years has been moving strongly in the opposite direction, saying that it is all the same genes pretty
much, and that they are conserved; and you get the Hox genes going back to bacteria, and so on, but there are small shifts in the structure of regulatory circuits and the hierarchy and so on; and that through physical principles you get the observed diversity of forms. And it does give you the laws of form. I mean, it is not that the laws of form are like Newton's laws. They emerge from the principles of physics and chemistry, which say that these are the ways in which molecules can work, and not a lot of other ways.

And just conceptually, it has to be like this. I mean, there cannot be anything like selection acting blindly. It is like learning – B. F. Skinner pointed out correctly (in one of the few correct statements he made back in
Verbal Behavior
, in fact) that the logical structure of conditioning, reinforcement theory, is the same as the Darwinian theory of natural selection. He understood Darwinian theory in a very naïve way – random mutation and then natural selection with changes in any possible way; that is all there is. But it can't be true. No biologist ever believed that it was true. It is totally impossible. Something has to channel a mutation in particular ways, not other ways – according to some recent work that I mentioned, in only a few ways. And then selection is just going to have to operate in particular channels and not in others. Skinner took that to be a justification for reinforcement theory, but in fact it is a refutation of reinforcement theory. This naïve Darwinian view is all over the place in evolutionary psychology and fields that touch on the evolution of language, and so on. But it is all just nonsense, as it is often presented. There have to be presupposed physical and chemical laws, and Turing I think was right about that. Maybe reaction-diffusion doesn't explain the stripes of zebras, but the basic principle has got to be right. And it is presupposed in everything that is done. Every time you talk about molecules behaving in a certain way, or genes producing this protein and not another one, and so on, that is all because that is the way physics and chemistry works.

D
OVER
: Well of course. All is chemistry and physics at the level of electrons and protons, and molecular interactions in biology, always based on differences in reduction and oxidation potentials, are not exempt. We don't differ on this point. Nor do we differ on evolved diversity being constrained (life is not a free-for-all). The argument is whether constraints are a reflection of contingent history (given that our single tree-of-life just happens to occupy only a small fraction of phenotypic space), or of the workings of physics/chemistry, or of laws of form above the reach of genes. But I will examine these alternatives in my talk, as well as the other point on which we agree that there is more to the evolution of life than natural selection.

C
HOMSKY
: The point is that if you want to move biology from looking at things as particular cases, if you want to move it from that to a science, then you're going to ask what the guiding principles are that determine what happens – you've got to ask the Why questions: Why did it happen this way and not that way? And that is being done. That is evo-devo work, which is increasingly showing that the course of evolution to a large extent (not always) is more regulatory than structural. I mean, the structures stay and the regulatory mechanisms change, and then you get a lot of diversity. Now they don't have a lot of experimental evidence for it, but that is a leading theme of modern evolutionary developmental biology, and plenty of biologists are staking on its potentially being true, whatever the evidence is.

So I think that Turing is correct in saying that that is the way that biology ought to go as a science. True, you find all sorts of details when you look, but we know that that can't be true generally. In this case, it is very much like the case of language, I think. It looked fifty years ago, and it still sort of looks, like every language is different from every other one and that is all you can say. You study the details. But it is conceptually obvious that that cannot be true, or no-one would ever acquire a language. And it is increasingly understood that it isn't true, and that to some extent you can attribute it to natural law.

About language evolving, yes of course language evolved. We are not angels. But evolution isn't just selection. Now here is an extreme thesis: perhaps language evolved as a result of, say, the explosion of brain size, for whatever reasons that took place maybe 100,000 years ago. It could be. Striedter speculates that a consequence of that is that some neural changes took place. It is not understood well. Even the simplest computation of insects is not understood well.
³⁶
But something is going on, and it could be that explosion of brain size led to some small rewiring which yields unbounded Merge, and everything else that it has come up with, and that yields the semantic interpretations. Then comes the problem of relating two independent systems, this one and the sensorimotor system, whatever it is, and you get complicated solutions to that problem which could be best-possible solutions – a research problem for the future. Well if that is true, then nothing in this particular domain involved selection. I don't really expect that that is going to be true. That is just an extreme speculation. But if that is true, it evolved and nothing was selected. Beyond that, there will be what residue is left in UG after you have extracted all the third-factor principles. And I think the same question arises in the development of organisms. I mean an ant may be developing and you take a look at it and it looks hopelessly complex – this gene did this, and this kind of gene did
something else, and so forth, but there has got to be some physical explanation for that. The problem is to discover it.