Sunday, March 2, 2008

probability and public policy VIII: intelligent design

[part one of this series here]

There has been much recent debate about intelligent design theory, whether it presents a significant challenge to evolutionary theory, and whether it should be taught in schools. A recent film criticizing intelligent design (though also the demeanor and attitude of mainstream scientists toward ordinary folk) will soon be followed by one criticizing the treatment of intelligent design by the scientific community, accusing it of violating the basic precepts of free speech (and starring no less an eminent public intellectual than Ben Stein).

Public debate on this question has been confused by several issues. First, the desire by those uncertain in their faith to find some empirical confirmation of their beliefs; yet, as many theologians have argued, belief in a divine entity is fundamentally a matter of faith, and thus not evidential in character. This is the view defended by Kierkegaard under the slogan "leap to faith". Second, the naive scientific realism of many scientists; in the documetary "A Flock of Dodos," for example, evolutionary scientists are repeatedly shown urging that evolutionary theory and not intelligent design should be taught in schools because evolutionary theory is "true." Yet, in the first case, there is much question concerning whether even the most empirically confirmed scientific theories should be judged "true," and in the second, there is no single "theory of evolution," rather there are a number of competing theories within a loosely defined general research program. Third, and in particular, the idea that the structures of living creatures have been selected for optimally, or that all features of such structure are purposeful, or "designed" (even if by a "random process"), has been criticized since the first appearance of Darwin's work, the landmark critique here being, perhaps, Gould and Lewontin's "The Spandrels of San Marco and the Panglossian Paradigm."

The question of whether intelligent design should be taught in school intersects with the concerns of probability theory in two areas. First, the general question of how a theory is supported by data is a probabilistic question. So, both the claim that evolutionary theory is not supported by the data, and the claim that evidence for intelligent design can be found in the data are probabilistic claims. Second, arguments for intelligent design depend (implicitly, if not explicitly) on a characterization of randomness and what kinds of behavior can or cannot be exhibited by a random process. As we saw in our discussion of the clustering illusion, humans exhibit a natural tendency to see patterns and organization in random phenomena. Statistics, thus, and scientific claims based on statistical analysis, must proceed very carefully in order to guard against the inadvertent systematization of fallacious preconceptions about whether or not a particular process exhibits non-random patterns.

Considering the question of how theories relate to data, one is inclined not to dismiss intelligent design theory out of hand. Surely, finding an objective standpoint from which to teach students about scientific methodology so that they can compare for themselves which scientific theory best fits the data would be far more constructive than teaching them the currently most popular scientific theory as dogma. Thus, if comparing intelligent design theory to evolutionary theory can be turned into an objective exercise in scientific methodology, there seems nothing in principle wrong with teaching it in schools. Unfortunately, most parties seem to treat children as future soldiers for their particular cause and are concerned only in indoctrination (i.e. which dogma do we impose upon them?) rather than education proper (which would teach children how to determine for themselves what constitutes dogma and how to pick for themselves amongst alternate views). This, however, is not a point about probability.

Considering the claim that analysis of the structure of organisms can determine whether they were generated by a random process or designed is a much trickier matter. First, the claim of standard evolutionary theory is not that human beings, say, were produced by a random process. Rather, new genetic material is generated at random, but the principles by which this material are selected for or against are rule governed and systematic. Just as relinquishing the claim that an intelligence guides the planets in their motion does not amount to relinquishing the claim that rules can be discovered which govern this motion, relinquishing the claim that intelligence guided the speciation of life on earth does not amount to relinquishing the claim that this speciation was governed systematically by rules.

In fact, although the straw man version of evolutionary theory publicly discussed seems to depend on this fallacious step, there is a similar fallacious step in the publicly available version of intelligent design theory. In particular, the claim that a process was generated by an "intelligence" does not have scientific content. Why not? Because we do not actually have a scientific theory of intelligence. This is the point which Turing emphasizes in his seminal paper "Computing Machinery and Intelligence." There, Turing introduces (what has become known as) the Turing test as a means of determining whether or not a computer is behaving intelligently. Such a test is needed precisely because there is no antecedant notion of intelligence other than by analogy with human behavior. If Turing's analysis is right, than any claim that a system is the result of intelligent behavior amounts to the claim that it looks as if the system were generated by something like human behavior. But then, the question of how to discover the principled rules which govern the system ( the scientific question) just reduces to the question of how to discover the principled rules which govern human behavior. So, intelligent design doesn't look like it could be a scientific theory about how life arose on earth, but rather an argument that no such scientific theory is possible (or, rather, will reduce somehow to an as yet unspecified theory of intelligence in general). Insofar as evolutionary theory is able to generate any systematic rules about the emergence and development of life (other than those directly applicable to the operation of the human brain), it seems to be constantly disproving intelligent design.

A final remark here about the structure of this argument. If intelligent design claims are basically of the form: the systematic rules of evolutionary theor(y/ies) are unable to account for the complex structure observed in living organisms, therefore we must posit some form of intelligent design behind the observed phenomena; then, this argument can only be relevant to the space of possible hypotheses considered, it cannot be an in principle argument against any such theory. This is because (as discussed before here and here) confirmation or disconfirmation of a theory by evidence depends upon the priors placed on the space of possible theories. Even if intelligent design theorists use some objective or principled technique (such as Bayes' postulate, or the Jeffreys prior) to place a prior distribution over evolutionary theories, their argument must be indexed by the space of possible theories. At most, then, such an argument could only amount to pessimism about the prospects for current research, it does not seem (on the basis of the arguments presented in the public sphere) that it could constitute a contentful alternative research program.

next: climate change

No comments: