Friday, June 13, 2008

Determination, Genes, and Explanation, and Eugenics

I commented on an online article by Jim Manzi here. Jim has another recent article, this time from the June 2 edition of the paper version of National Review, which you can get (if you pay for it) here.

The June 2 article has the promising title "Undetermined: There is danger in assuming that genes explain all." The title may lead the reader to suspect that the article will refute genetic determinism. While the article does punch holes in certain grandiose claims made in the name of genetic science, it unfortunately does not really address determinism at all; and this because it conflates the notions of determined and explained.

Something is determined when its nature and destiny are entirely the result of non-rational, physical causes. Given a certain physical state of a system and the universe in which it exists, the future state of the system follows of necessity from the initial state (at least in a statistical sense.) Something is explained (in a scientific sense) when its present or future physical states can be analyzed and predicted in terms of its past states. It is possible for something to be determined yet unexplained. This happens when the problem of analysis becomes intractable.

Such intractability happens all the time. In fact, in physics and engineering, the vast majority of physical problems are intractable. Take a simple coin toss. The trajectory of the coin is entirely determined by the physical forces on the coin. Yet predicting the result of a coin toss based on analysis of the forces is extremely difficult. There are just too many factors involved and even tiny variations in the parameters of the problem can change the result. So, as far as predicting the result of a coin toss, we are at no advantage to ancient Greeks who never heard of Newton. They knew as well as we do that a coin toss is a 50/50 proposition; but we cannot predict the result of an individual coin toss any better than they could. The intractability of many simple physical problems is what keeps Las Vegas in business. Casinos allow players to place roullette bets even after the ball has been sent rolling on the rim and the wheel spun. Given the velocities and locations of the ball and wheel, shouldn't a player be able to predict where it will land? In theory, yes; in practice, no. The problem is way too sensitive to the precise parameters involved and what happens when the ball falls off the rim and bounces around the number slots. The roullette wheel is entirely determined but also entirely unpredictable.

What about determinism with respect to ourselves? Jim Manzi talks particularly about genes, but the philosophical problem is the same whether it is proposed that we are determined by genes, the environment, some combination of the two, or even if we take the ancient view that we are determined by the stars. (Determinism is thought of as a modern philosophical development, but it is really a return to an ancient mode of thought.) If we are determined by non-rational causes (genes, the environment, evolution, the stars, black cats, etc.), then our nature and destiny are entirely functions of those causes. Whether we can explain or predict our destiny by analysis of those causes is entirely another question. A demonstration that such prediction is practically impossible does not answer the claim that we are determined.

The popular belief in genetic determinism, Manzi tells us, comes from the media loosely speaking "of things such as a 'happiness gene', a 'gay gene', or a 'smart gene.'" The inference that there are genes for every aspect of our nature, and that our nature can be engineered through those genes, naturally follows. "Seeing this momentum, it is natural to assume that eventually we will have explained all human behavior, not just diseases caused by one or a small number of interacting genes." But Manzi cites two reasons why such a conclusion is unjustified. One reason is the "correlation vs. causation" problem. The causal relation between two things does not follow directly from their correlation. In Bertrand Russell's famous example, it would be a mistake to conclude from the correlation of umbrellas opening and rain falling that opening umbrellas causes rain to fall. In the genetic case, a gene may be correlated with a certain trait yet not be the cause of it. Chinese people, for example, may possess a certain gene and also be susceptible to a particular disease, but the cause of the disease may be due to peculiarities of Chinese culture rather than genes. It is very difficult to disentangle environmental from genetic causes because of this problem. This point is of little moment, however, because the precise physical basis of determinism is incidental to its philosophical implications. Whether it is environment, genes, or the stars, the philosophical implications of determinism remain the same.

The more interesting reason Jim cites is something called "epistatic interaction", or the fact that many traits are caused by interactions between several genes rather than a single gene. Some traits are caused by the interaction of many genes. It doesn't take many genes interacting to make the problem of prediction intractable, as Jim points out:

"Consider a simplified case in which some personality characteristic - aggressiveness, for example - is regulated by 100 genes, each of which can have two possible states ('on' or 'off'). The combinatorial math is daunting: There are more than a trillion trillion possible combinations of these gene states. Thus we could sequence the DNA of all 6.7 billion human beings and still not know which genes are responsible for aggressiveness."

What Jim has shown, of course, is that genetic explanation may be impossible; what he has not shown is any reason to dismiss genetic determinism. Indeed, the tone of the article seems to concede determinism without a fight. He ends the article with the following:

"Science may someday allow us to predict human behavior comprehensively and reliably, so that we can live in Woodrow Wilson's 'perfected, co-ordinated beehive.' Until then, however, we need to keep stumbling forward in freedom as best we can."

Determinism is a necessary but not sufficient condition for physical explanation. That is, we "may someday" be able to predict human behavior comprehensively only if human behavior is in fact determined, whether we can explain it now or not. And if human behavior is determined, then we are not free, whether anyone can predict our behavior now or not. If scientists someday come up with that comprehensive and reliable explanation of human behavior, they will not kill freedom at the moment their theory is complete; they will have shown that freedom was an illusion all along. But it isn't even necessary for scientists to come up with that comprehensive theory for freedom to die. The acknowledgment of the possibility of such an explanation is already to concede the philosophical battle to determinism, for such explanations are only possible if human behavior is in fact determined. The truth of determinism is the ground of the possibility of a comprehensive scientific explanation of man.

Jim's argument amounts to a "freedom of the gaps." The reader may recall the famous "god of the gaps" - the god who is conceded to exist as an explanation for those things science has yet to explain. As science advances, the god of the gaps recedes, for there is less and less for him to do. Jim grounds our freedom on the present ignorance of science, and hopes that science will not advance sufficiently to wipe out freedom completely. But the freedom of the gaps, like the god of gaps, is a poor imitation of the original.

Freedom is philosophical, and the reasons for sustaining true freedom in the face of determinism are philosophical rather than scientific. Human nature and behavior cannot possibly be determined by any set of non-rational causes (the only causes of which science is cognizant), for the simple reason that we are rational, or knowing, beings. Through knowledge, we transcend what we know, and break free of its determinations. Science, for instance, is itself an example of human behavior. Any scientific theory that attempted to comprehensively and reliably explain human behavior would have to include scientific behavior itself in the account. It would have to predict the behavior of scientists in their laboratories and in their theorizing. But such prediction assumes that the scientists can't discover anything radically new that is not already accounted for in the deterministic theory. In other words, the only way a comprehensive and reliable prediction of scientific behavior is possible is if science has come to an end, for only if it is at an end can we be sure that a scientist won't discover something that upsets our comprehensive deterministic theory of man. The comprehensive deterministic theory of man, then, must be the Final Theory of Everything, leaving nothing in the universe out of account that might possibly be a discovery of the future.

But even if someone managed such a Final Theory of Everything, he would still find one thing left out of his theory: Himself. His theory cannot account for his own understanding of the theory, for his understanding of the theory doesn't exist until he has finished it. And once he has finished it, he has created a novel human behavior that is not yet accounted for by the theory; the discovery of this particular Final Theory of Everything. So there cannot possibly be a comprehensive and reliable deterministic explanation of man, for the explainers themselves will escape the deterministic explanation.

The argument can be put another way: Any claim of determination must cite the non-rational causes that are the ground of determination. We are determined by genes, evolution, the stars, whatever. But the very argument for non-rational determination is itself an example of rational causation that transcends the set of non-rational causes. The geneticist may propose a "happiness gene", but he never proposes a "genetic theory gene"; in other words, a gene that causes someone to propose the very genetic theory he is proposing. But unless there is such a gene (or epistatic interaction of genes), then his deterministic theory cannot be a comprehensive explanation of man.

In the end, I agree with Jim's title. We are indeed "undetermined." But the reasons we are undetermined are philosophical, not scientific, and the best Jim's scientific argument can show is that we are (currently) "unexplained", not that we are "undetermined."

The title of this post mentions eugenics, but I will save that for a coming post...

4 comments:

Anonymous said...

Daivd:

Thanks for another great post. I didn't pick the title of the article. You are correct that I was making (or at least trying to make the point) that science does not currently have the ability to meaningfully predict individual or group behavior in reliable, non-trivial and non-intutitve ways. I reamined silent in the article about whether this was possible in principle, becuase that further claim was not central to the larger argument that I was trying to make.

I'm looking forward to the continuation of this post.

Best,
Jim

David said...

Jim,

I guess I don't see the question of the possibility of a deterministic theory of man as a further claim, but as a prior one. For me, when you write that science "does not currently have the ability to...", it implies the concession that science could, in principle, have that ability. We wouldn't write "science does not currently have the ability to produce perpetual motion machines" because the "currently" implies a merely temporary failure, when in fact perpetual machines are impossible in principle. I believe that is why the patent office summarily rejects them.

The article as written, I think, successfully demonstrates that scientists do not have in fact a deterministic theory of man, but implies the concession that such explanations are possible in principle. Do you agree that deterministic theories of human behavior are not possible in principle? Perhaps we disagree on this point.

Cheers,
David

Anonymous said...

David:

If I were writing an article to say that all of these science popularizers who are arguing that we are figuing out how to a build a perpetual motion machine, I could make the point they were wrong just by showing that there is no such scientific finding, and not have to make the more difficult case that there will never be such a scientific finding.

I doubt very much that there will ever be a determinstic science to predict human behavior, but would obviously be prepared to change my mind in the face of contrary evidence.

Best,
Jim

David said...

Jim,

I appreciate your response to my question, and I think I understand where our basic difference lies. I think you hold that a comprehensive deterministic explanation of man is at least theoretically possible, if not practically achievable. I hold that such a comprehensive explanation is not even theoretically possible, for reasons I give in the original post. My reasons are based on the nature of scientific investigation itself, not any particular state of the evidence, which is why I don't think any accumulation of evidence can ever change the conclusion.

I've enjoyed this exchange. Please stop by any time.

Cheers,
David