« Say CHEEEEESE! | Gene Expression Front Page | Return of the bones »
February 24, 2005

Gene + environment interactions....

When people speak of "Nature vs. Nurture" there is often a tendency to refer to the "proportion of a trait that is genetic vs. environmental" in a rather broad and vague manner. The problem is that the proportionate contribution of genes and environment in an individual human on a particular trait varies from person to person. Additionally, since one can not rewind the developmental arc of an individual and expose them to various environments to get a sense of how powerful genetic and non-genetic inputs might be in determining their particular phenotype it is rather difficult to assert that individual A's phenotype B was determined by quanta of x genes and y environment (and unlike the environment, at least genes are discrete elements).

Rather, we generally speak of, heritability, which is roughly the phenotypic variation attributable to genotypic variation within a population. Heritability is further divided into "broad sense" (additive + dominance effects) and "narrow sense" (additive) flavors.1 But the story doesn't end there as those who are skeptical of claims of heritability would remind you, there are gene-environment interactions (or correlation) as well as epistatic effects (the influence that loci have upon each other, adding a nonadditive factor in the equation). So, the final phenotypic variation in a population can be modeled like so....

Vp = Va + Vd + Vi + Vge + Ve

[Vi is epistatic variance, the others are self-evident]

Update: Please read this correction.

Gene-environment interaction and epistasis are, in my opinion, often used as "get out of jail free" cards by some who come rather close to a "Blank Slate" conception of human nature in implementation if not explicit rhetoric (ironic in the case of epistasis since it is gene-gene interaction). But, I don't think that they get off scott free, especially in the case of gene-environment interaction.

Gene-environment interaction tends to amplify the perception of individual differences. Imagine for example that a farmer notices variation in milk output between his cows. He might maximize his yield by shifting his best feed to his best milk producers. Later, if you decide to analyze milk production you will note a far greater phenotypic variation than you would if they were all given the same feed. The "best" genotype is correlated with the best feed, and the "worst" genotype is correlated with the worst feed. There are also forms of gene-environment interaction where genotype A is "superior" to genotype B in environment A but "inferior" in environment "B" (see norm of reaction), or scenarios where A and B express the same phenotype is environment A but differ in environment B.

In human beings the gene-environment interaction in the behavorial/psychological context can be decomposed into three correlations, passive, evocative and active. A child who is very bright often has very bright parents, and these very bright parents often create a very specific environment for their children.2 A child who has a reputation of being very bright often has expectations built up which teachers are well aware of as they progress through the grades.3 Finally, there is active correlations, where children with certain propensities pursue the activities which they can most easily attain virtuosity because of their inborn talent biases. In other words, two children who differ moderately in brightness may make different choices throughout their childhoods which are important in shaping the final phenotype they manifest in adulthood. By the time they are adults the perceived phenotypic differential (as measured by grades and test scores) might be a yawning chasm in comparison to the mild differences of their youth.

You can, at this point, infer what the obvious solution by some might be to this situation. Proactively encourage the student that is somewhat less bright so that they can make the most of their potential. If you had a population where all students had the same expectations, the same environmental inputs, one could imagine that the phenotypic variance in the next generation might decrease (at this point, the variation due to genotypic variation would actually increase as a proportion of phenotypic variation!). Environment equalization might have an impact on both passive and evocative correlations downstream, which would likely percolate horizontally to active correlations. All is well, and the remediative programs that altered the social dynamics in the first place may be removed, correct?

No. This is where a pure social constructionist and someone who acknowledges the salience of gene-environment interaction must part ways. To a social constructionist once the social system is transformed it can perpetuate itself, just like the previous social system did. Remediation is necessary only to flip the system into another equilibrium (that is, nature does not load the die in any particular direction). On the other hand, if you accept gene-environment interaction, the genetic variation remains, the correlations are simply being dampened by social factors. If those incentives toward equalization and homogenization were removed, then the active correlations at least might start to work their way back into the equation. For example, if "problem" students were not proactively targeted, nurtured, cajoled and guided, they would be more likely to follow a different path than the students for whom schoolwork came naturally. This would likely result in the reemergence of passive and evocative correlations in the next generation. Soon enough all the gene-environment correlations would be salient once more. If you attempt to emphasize gene-environment correlation to explain great phenotypic variation, you must also acknowledge that it is likely that societal remediation or dampening of such correlations must be perpetuated indefintely. In contrast, social constructionists can suggest that once social change is affected culture and socialization can perpetuate the new order.

Social constraints are not necessarily negative. I have referred to the possible "social goods" of monogamous marriage before. Though polygyny is "beneficial" in certain circumstances, I find it plausible that middle class liberal democracies are stabilized by monogamy (though this is only a hypothesis, I don't accept this as a definitive truth). But I also suspect that polygyny is, to some extent, a human male Evolutionarily Stable Strategy (ESS). This means that social restraints and legalistic fiat must come into play to shift the incentives toward monogamy. One can imagine this in fact as a form of dampening gene-environment correlation. In a culture where polygyny is socially acceptable, males may seek out and attain different "marital microenvironments." That is, one may have one wife, or two wives and so forth. In most modern societies this is not possible (at least openly), and serial monogamy is the main variation upon the theme. In a monogamous culture males with "fit" genes may still have more wives over their life, ergo, more children, in a serial fashion, but the variation in the number of children and the lifetime aggregate potential number of wives is far smaller than in polygynous societies (less reproductive skew). The offspring payoff of a "fit" male genotype is far greater in a polygynous society than in a monogamous one, in other words, you are seeing different norms of reaction. Additionally, because of imitation, the number of wives accrued by men with multiple wives may also increase solely because of the signalling impact of his microenvironment (that is, he can afford to have many wives to begin with, which amplifies his desirability, resulting in evocative gene-environment correlation).4

After nearly 2000 years of enforced legal monogamy I do not believe that Western males have changed in their "optimal" ESS greatly. Likely part of this is because serial monogamy, illicit relationships and concubinage have given polygyny preferring males some advantage. But nevertheless, the underlying genotypic variation that might lead to differences in "mate attracting" capacity still remains. If Western society were to legalize and normalize polygyny I woud not be surprised if many high status males began to avail themselves of this option. Already, the sexual revolution has enabled some high status men to maintain harems (Hugh Hefner), even if they are not legally solemnized. For those who have fortunes in the tens of millions I suspect that the constraint of legal monogamy serves as an important dampening influence.

But now I will move to sex differences: even assuming similar aptitudes, if there is an innate preference differential, social engineering must be continued indefinitely if "social goods" are to be perpetuated (or at least until the Baldwin Effect kicks in, but that is on a different timescale). Personally, I do favor some of the social engineering, if children were allowed to choose for themselves it seems plausible that in the early ages boys and girls would want to attend different schools. In a society of universal sufferage, and legal equality, I think that mixed-gender socialization is a crucial dampening factor, and a necessary social lubricant. In some ways single gendered education might be optimal for the academic aspect of a child's life, but the American educational system is also a crucial social engineering tool (just as the American military is). But there is no free lunch, and I have just noted that the socially beneficial aspects of mixed-gender education entail a cost, that is, suboptimal development of academic skillsets.5

Similarly, if society made goals for itself in terms of how many women should be in each profession, there will be a cost that must be paid. If there are many more scholarship opportunities available in the hard sciences for females, the cost might be that women who are more inclined toward humanistic professions where they could excel might settle for a life as a mediocre scientist. The perpetual administrative costs of managing gender equalization programs would be a constant fixed expense.

This does not mean that social engineering is plainly wrong, that depends on the values that you promote. But I think the full implications of acknowledging the importance of gene-environment correlation have not been addressed. Implicitly much of what is basically social engineering seems to assume that its aim is to enable individual free choice. I don't think this is always the case, rather, the engineers in question have an idea of what a "good society" is. This is not necessarily a Left-Right issue, modern Christian conservatives in the United States tend to view a monogamous life-long pairbond as the ideal, but it seems that multiple vectors are pointing to far more of a muddle in this area than the starkness of the values. Nevertheless, even many liberals would assent to the validity of monogamous pairbonds as the ideal for a free democratic society. On the other hand, some on the social Left seem intent on reshaping society in a most unnatural state for reasons that I can not really fathom, or at least I do not believe are warranted. When the Left fights for individual civil liberties and revocation of legal barriers to free choice, I tend to agree with them. On the other hand, there does seem to be a segment that intent on punishing humans for their humanity.6

1 - Narrow sense heritability far more relevant for continuous polygenic traits and has great utility in animal breeding. Given enough time you can theoretically remove all the additive genetic variation from a population by selecting for particular phenotypes, but this is not necessarily so with dominance/recessive variation (homozygous and heterozygous phenotypes might have the same fitness or value in determining selection cutoff for the next generation). Additivity is pretty obvious, on one locus AA = 1, aa = 0, so Aa or aA = 0.5 (more or less). Even in dominant/recessive loci where AA =1 and Aa or aA =1 and aa = 0, the heterozygote is often not exactly 1, but almost 1, or close enough that you can't tell the difference on inspection.

2 - If you believe the Judith Rich Harris line, they select a particular peer group, same difference. I suspect much of the gene-environment correlation is in the peer groups. The implication is that if as a society you have particular social mores to promote, and certain peer groups promote contrary values, you must break up the peer groups.

3 - I once convinced two friends of mine to switch essays for the final paper in my high school AP English course. Their grades were fixed at that point at an A and a B, so it was really not going to make a difference. They wrote up their papers and simply printed it out with the names flipped. The student who normally wrote a B paper received a B grade, even though the student who wrote the B paper was the A student! And the reverse also occurred, the paper written by the B student received an A when the A student's name was on it!

4 - Some have argued that males are important in that reproductive skew is crucial is purifying deleterious mutations from a population. In other words, it is the less healthy and fit males in each generation do not reproduce so take many deleterious mutations with them.

5 - In three dimensions one could imagine that social engineering came favor a particular equilibrium by creating an elevated energetic cordon which isolates the stable region from the plains surrounding it. That is, the favored social state might be conceived of as sitting within a crater. Once the walls of the crater drop, the social state becomes unstable and eventually it slides down the energy peak to a lower state.

6 - To be perfectly, I tend to be on the same page with the individual liberties/rights element of the Left. Though I usually disagree, I can see where the socialist/fiscal liberal Leftists are coming from. On the other hand those who fall under the rubric "identity politics" seem to be incoherent, often irrational and rooted in emotive and interpersonal concerns rather than a "good society." Economic redistributionism in the socialist/Left model seems to be marginal. The individual liberties/rights folk also seem to be in decline (partly because they've won many of the rights and liberties that they fought for!). By default the identity politics folk seem to be looming larger and larger....

Posted by razib at 04:12 PM