« Teachers - They Simply Can't Win | Gene Expression Front Page | The unAdapted mind? »
December 10, 2004

Bad theories & incongruities

A few weeks ago Will Wilkinson posted the following:


Similarly, I think the mind is pretty modular, if not massively so. Thus, I don't think there is anything like a general reasoning capacity...Yet I'm rather impressed by the data on G, and how it predicts quite a number of things. So what is G tracking if there is no general reasoning capacity?

Now, I've wondered about this as well, because I too tend to subscribe to modularity and a belief that g has great predictive power. What's going on? In Prehistory of the Mind Steven Mithen (another Steve & evolution!) suggests that "general intelligence" is a basic facility that is a rough tool that we share with other homonoids. In fact, Mithen believes that chimpanzees "learn" language with their general intelligence, thus, their lack of fluency in terms of syntax. In contrast, humans have a "language instinct," which allows us to naturally grasp the nuances of grammar as well as sponge up vocabularly. There is something to this general sketch, but here is my conception of what's going on. I'm going to start out with two facts:

  • We know that g is normally distributed (roughly).
  • We know that g correlates with many other cognitive capacities, for instance, mathematical aptitude.

For a trait influenced by a large number of genes the result is often a normal distribution in terms of the phenotypic expression. Since g is normally distributed it is plausible that it is a phenotype with a large constellation of genes upstream. And sure enough the work I've seen by behavorial geneticists like Robert Plomin tend to offer very small values for the additive contribution of any one locus (on the order of 1%).

In contrast, capacity for language (though not eloquence), is an on-off feature, that is, we all have the "language instinct," and some would argue that FOXP2 might be the "master" regulatory gene which is central to its function. Language is the archetypical module when we think of higher cognitive abilities.1 People can tell that a sentence sounds "wrong" without being able to elaborate why. Contrast this with mathematics, with its system of proofs. Theoretically mathematics is conceived as reflective rationality incarnate. But if you read books like The Number Sense you will note that mathematics itself seems to emerge from a synthesis of basic analog numerosity as well as linguistic and visuospatial capacities. In cases of individuals with brain injuries there are those who can lose the capacity for geometrical reasoning but may still be able to perform algebraic manipulations, and some individuals who "can not count" who can still make assertions about geometric questions by inspection. The evidence seems to be that mathematics is an emergent property of various cognitive domains. Papers that note the importance of "hemispheric integration" in mathematical aptitude also suggest it is a complex trait with many variables contributing to the full expression of the "phenotype."

And so with g. In Prehistory of the Mind Mithen argues that the interaction between the various modules facilitated by analogical thinking was crucial to the "Great Leap Forward" in human cognition, so he's basically saying the same thing. All the various subroutines, some of which, like language, have very focused primary tasks, nevertheless intersect and contribute to the formation of a "general intelligence" which transcends the limitations of any given module. I know Steven Pinker believes that religion is a byproduct of the interactions of various modules (in concert with environmental cues), so even massive modularity does not imply a hermetic seal between various cognitive processes in the minds of those who espouse that hypothesis. In both The Mating Mind by Geoffrey Miller and The Blank Slate by Pinker you have two books which seem to assume a modular mind and say some positive things about psychometry and the attempt to ferret out g. So the "Thus" part of Will's thought does not seem founded.

Anyway, my thoughts aren't fully fleshed out on this topic, but I simply wanted to elaborate on what I posted on Will's blog.

Addendum: Please note that the site about FOXP2 asserts that mutations to the gene also seem to have a major impact on "general intelligence" (verbal and non-verbal) something that the popular press reports tend to omit. Since language is likely one of the crucial building blocks of general reasoning capacity it stands to reasoning that the removal of this variable would cause havoc downstream.

1 - Steven Pinker has argued that the language module is more naturally decomposed into grammatical & lexical modules. See his work on irregular verbs.

Posted by razib at 01:02 PM