A few weeks ago the study about obesity being socially contagious was all the rage. For anyone interested in human behavior this shouldn't be surprising, we are a social creature and our peer group is an essential part of our 'extended phenotype'. The psychologist Judith Rich Harris has famously argued that the 40% of unattributed component of variation of personality is due to our peer groups (10% is parents and 50% is genes). In Harris' model the best thing that parents can do is choose a particular peer group with values which reflect their own priorities. In other words, buying a house in a "good community" with "good schools" is paramount. Your peers are the measure which you take into account and help shape your values, priorities, and goals in life. So with that, I was interested in a new study which tracks male twins and shows the importance of genetic variation in the choices which they later made in terms of assorting into distinct peer groups. From the conclusion:
As male twins mature and create their own social worlds, genetic factors play an increasingly important role in their choice of peers, while shared environment becomes less influential. The individual-specific environment increases in importance when individuals leave home. Individuals who have deviant peers in childhood, as a result of genetic vs shared environmental influences, have distinct developmental trajectories. Understanding the risk factors for peer-group deviance will help clarify the etiology of a range of externalizing psychopathology.
Basically the study is measuring gene-environment correlation, the amplification of initially small innate differences over time due to a series of choices influenced by genetic biases. Here is a clear illustration: imagine two fraternal twins, one who is slightly more bookish and another who is slightly more athletic. Over the years the bookish one finds greater returns in studying in terms of positive feedback from parents and peers, and the athletic one tends to focus on sport because that is the domain in which he excels. By the age of 18 one might predict that the bookish brother will be a much stronger scholar while the athletic brother would be outstanding on the field of play. Can their genes explain this? If the initial difference was so small why are they now so different? Obviously over the years the brothers have invested their finite time differently, you can't be a good basketball player without some practice, nor can one master calculus without the prerequisites. Initially small differences can snowball via gene-environment correlations to make a rather non-trivial difference in phenotypic outcome. A more classic example would be in terms of dairy cows and milk production. Assuming there is a small heritable difference in output across cows, one can imagine that the farmer will divert his or her best feed to the best innate producers. This will of course have the effect of stretching out the distribution as the best producers now have the best raw material to work with while the worst producers will have the least nutritious feed. How do peer groups fit into this? They are environment. There doesn't have to be scientific exposition on this topic, everyone knows from common sense and experience the disparate pressures placed upon you by different peer groups, and the varied expectations. For example, though I was an atheist by the end of high school I generally socialized with the Mormon kids. The main reason was that my earlier peer groups had shifted toward alcohol and marijuana, and that just wasn't my scene. So I "opted out." When you are a small child you have only so much choice in your peer group. You can't leave you neighborhood very easily, your school is chosen by parents and geography, your religious institution is chosen by parents, your free time is often controlled and parceled out at someone else's discretion. Once you enter young adulthood free choice becomes a much more important variable, and the study above indicates that those choices are scaffolded by genetic predispositions. Because of the discrete manner in which social groups exist one can imagine a phenotypic landscape with alternative peaks which different individuals are attracted to in relation to their propensities, some of which have a genotypic component of variation. What does this mean in practical terms? Things are more than the sum of their parts. If you want to change individuals you often have to change their whole social landscape. Environmental pressures can constrain and warp goals and preferences, but those forces need to be maintained. If, for example, a group of individuals have criminal tendencies (e.g., low time preference, lack of empathy, etc.) then it stands to reason that together they will far more anti-social and pathological than they would be apart, where their urges would be buffered by individuals and groups which promote different goals and incentives. Social engineering can't simply be a matter of focusing on individuals, it has to be a culture war. Addendum: In quantitative behavioral genetics it is important not to confuse gene-environment interaction with gene-environment correlation. This article is a good intro to the topic. Basically interactions can be thought of as non-linear effects. For example, a norm of reaction. In contrast, gene-environment correlations tend to amplify the magnitude of extant differences. An interaction is salient if shifting the environment radically changes the proportionate phenotypic effects across different genotypes. For exame, if genotype A and B exhibit the same phenotype in environment 1, but are very different in environment 2, that shows a gene-environment interaction. In contrast, a gene-environment correlation would be genotype A and B developing different alternative extended phenotypes (e.g., attraction toward different ecological niches) which in their turn have significant developmental effects (e.g., the small initial phenotypic difference is exaggerated by the disparate environmental regimes).