Turned up to eleven: Fair and Balanced

Wednesday, April 24, 2002

Genetics, IQ, and Intelligence (part 1)

This is, as noted below, a response in part to Godless Capitalist and his notion of racial differences in intelligence. The next few posts will be wide-ranging, discussing the topics above and possibly expanding in ways I have not anticipated (I'm going with a stream of consciousness approach here). I will start with a review of the topic of genetics, especially with respect to the nature of heritability studies. I hope that interested readers (if there are any) will challenge any errors, factual or logical, that they see!

In the past 50 years or so, the study of genetics has progressed exponentially, with rapid developments in the 1970's allowing for the isolation of genes in "recombinant" (this means transferring the gene into a different setting, like into a bacterial host for example) systems, and study of single gene function. The development of the "OGOD" paradigm, which stands for "One Gene, One Disease", particularly in the 1980's and 1990's, created a mindset of finding a gene that was responsible for a disease. It was always understood by people in the field that this was an oversimplification, but a useful one in the context of reductionist molecular biologist (I don't use the term "reductionist" pejoratively, incidentally; its what I do!).

At this point it is useful to define some terms. 1)Genotype Generally, genotype refers to the actual set of genes carried by an organism. For example, a bacterium that contains the gene bla, encoding the resistance mechanism to beta-lactam antibiotics (such as penicillin), might be said to have the genotype bla+. More generally, an organisms genotype can be extended based on whatever genes we are interested in. In eukaryotes, which have a diploid genome, contain two alleles at every genetic locus. In case you missed it, click on one of those links to go back and look up a definition for any general genetic term in the rest of this essay. 2) Phenotype The physical attribute of an organism. In the above example of the antibiotic resistant bacterium, growth in the presence of penicillin is the phenotype. The term is extremely general. Obesity is a phenotype; so is brown hair, so is blue eyes, so is big feet. Calling something a phenotype does not mean that there are genes involved. Of course, there are genes "involved" in every aspect of human existence, but the point here is that a phenotype is not the same as a genotype, and "phenotype plasticity", i.e. the amount that environmental conditions affect a given characteristic, is the crux of the issue. 3) Heritability The ability to pass on traits to one's offspring. The goal of phenotype heritability studies is to assess the contribution of genes and environment to the expression of phenotype, and heritability is the statistically assessed value of the genetic contribution.

My major contention with "Godless Capitalist" is that the heritability of intelligence is flawed on both counts both in terms of the measurement of heritability and in terms of defining intelligence. I will address the heritability issue first. The gold standard of heritability studies is the separated identical twins case, often just called "twin studies". The big problem with this and all other attempts to divide phenotype variance into genetic and environmental factors is the assumption of additivity, i.e. phenotype = (genetic influence) + (environmental influence). This ignores the increasingly acknowledged importance of signal transduction in genetic activity. In other words, the activation of a gene to produce its product (whereby the phenotype results) is exquisitely sensitive in many cases to the environment surrounding the cell. In bacteria, this is seen quite clearly in many areas (if you are wondering why I keep bringing bacteria up, it is because I study bacterial genetics for a living), and it is increasingly clear that the development of bacterial communities is very much dependent on the environmental factors which surround them. This logic applies perhaps even more strongly to multi-cellular organisms. The parallels are fairly clear, from a genetic perspective. The developmental pathways that result in the formation of full-grown adult human are complex, and at nearly every step there are interactions between genes and environment. The results of those interactions can affect the nature of future interactions in profound ways (see my previous discussions of feedback loops for details). It is not in any way clear, especially with the small sample sizes of twin studies, that the analysis of variance approach is appropriate, for a number of reasons. I think that the waters are sufficiently muddy with respect to analysis of heritability of any complex phenotype to make me question claims about heritability of intelligence. As a person familiar with the intricacies of molecular genetics, I find it hard to give much weight to claims of "50% heritability" for any trait, let alone one that is as complex and poorly understood as intelligence.

So, is it hopeless to try and identify genotype/phenotype relationships for complex traits? By no means. But it is essential to understand the intricacy of the system you are studying, and to recognize that in a complex, living organism, independent genes are the exception, not the rule. In other words, there are so many networks of connections between organs, tissues, cells, proteins and genes, that it is likely to be very hard to find a gene that acts alone, or even in a relatively small group, to create a phenotype. Most of these genes have been identified, and they usually encode some sort of nutrient deficiency, or single essential enzyme missing or defective (hemophilia, caused by deficiency in Factor 8, for example). Any phenotype that involves multiple genes, with multiple alleles, especially where there is environmental or intercellular communication dependence for gene expression, is going to be much more difficult to understand. The example, from human development, that best shows this (to me) is maturation of the immune system. The production of mature T-cells is influenced by 1) somatic cell recombination events, 2)Selection in the thymus for cells with non-self reactive receptors (negative selection), 3)Selection in the thymus for cells with reactive enough receptors to random display peptides (positive selection), 4) Somatic hypermutation, which specifically acts to increase specificity to a certain stimulus after the cell is circulating, dependent on activation of the T-cell by its target. This is a highly complex developmental pathway, and when its done, a vast array of non-self reactive T-cells, capable of responding to intracellular and extracellular threats, is present. The response of this system is still dependent on receiving its signals in the proper context (bound to MHC on antigen presenting cells with appropriate soluble and insoluble factors). So measuring this activity by "phenotype", i.e. the ability of a person to fight off disease, is a non-starter. I would like to suggest that intelligence is at least as complicated a trait.

Here I will also respond to the notion that "Godless Capitalist" put forth to explain why, though the overall genetic diversity within populations is much larger than between populations, it is still important to study those populations. In his defense, he suggests an orthogonal relationship between the important traits and the unimportant ones. Two immediate questions follow; 1) How do you decide what is important? This introduces the essential subjectivity of nearly all of this type of work, and 2) Where does this notion of independence (graphical right angles are a representation of variable independence) come from? In what way do you establish that the phenotypes you identify as "different" are independent from the ones that are not, in establishing the identity of "races". Quite simply, there is very strong evidence that isolation of populations reproductively has not been a part of human existence for long enough (~40000 years) to cause substantial divergence (modern humans have existed for at least 150,000 years, and some have argued for closer to 1 million years. Contact between peoples that were reproductively isolated has been established at the present time. In fact, here is a prediction for the IQ/race crowd. Increased mixing of previously isolated groups should abrogate the racial/IQ relationship. Start assembling the IQ scores of offspring from interracial marriages, and see how they compare to their parents.

My next post will deal with the subject of intelligence itself, and a semi-mathematical argument for a structure of conscious brain function. Beware!!