Snake eyes of the die..."it lurks in your your genes!"

Dave and Jonah have both commented on this piece in The New York Times which is something of a mismash of recent studies coming out of the field of behavior genetics. The best thing about the piece, from my selfish angle, is that it references Contingency Table, now absorbed into my other weblog, who was riffing off one of my older posts here at Science Blogs.

The biggest problem with pieces like this isn't the genetics or psychology, it is the utter lack of focus on the importantce of probability distributions and the concept of expectation. There are several types of genetic traits. Some of them are the simple Mendelian ones we are all familiar with, many autosomal recessive diseases are controlled by this dynamic, you have a loss of function on one locus, and boom! it all breaks down. This one-gene-one-trait model is not generally what we are talking about in behavior, here we are looking at quantitative traits which are polygenic, as well as exhibiting a great deal of environmental input. These quantitative traits, often modelable to a rough normal distribution (i.e., Bell Curve), are not "simple" by their nature. When people say that something is "50% genetic," they usually mean this: 50% of the variation within a population can be attributable to genotypic variation, and specifically, additive genetic variation. There is obviously an environmental component to the variation, but there are complexities like gene-environment correlation (e.g., mildly bookish children are more likely to seek out books, and so amplify their bookishness over time while mildy book averse may go in the reverse direction, so a modest 'innate' difference ends up be a rather large realized difference). There is also the all important issue of gene-environment interaction (see norm of reaction), which can result in unexpected and divergent phenotypic outcomes in alternative environments (these "percent genetic" numbers usually control for wild environmental variations because of biases in the studies). I won't even get into more esoteric issues like gene-gene interactions (epistatic effects). But, the "big picture" is that when someone says a trait is "50% genetic," they don't mean that in you & I half of the character of that trait is attributable to genetics and half to non-genetics, like mixing two solutions to a 1:1 ratio in a beaker in chemistry. Rather, the quantity they are referring to is a particular statistical construct that emerges, at the root, from the molecular level and the various interactions which lead up to the phenotype. In other words, because these are population level generalizations you only get rough expectations in terms of individuals, so you should be cautious about using any genetic test as a "blueprint" for your life.

That being said, I am not one to discount behavior genetics because of its manifold problems. It is a messy and difficult science, but many people who are highly skeptical of behavior genetics accept with much less objection "evidence" derived from less rigorous and even messier fields in the social sciences. To me, much of the critique of behavior genetics does seem to be politically motivated, and a response to the excesses of the past century (though I would argue that if such is the measure of disciplines, then the social "science " of Marx has quite a bit of blood on its hand, and by association much of the scientific analysis of human sociality). Behavior genetics is important because it tells people the cards they were dealt, and though those cards don't determine the outcome, they allow you to generate a personal calculus of costs, opportunities and risks. In terms of the reality of how human behavior works, it seems clear were a supremely facultative creature, not an obligate one. That is, many of our behaviors are conditional upon a host of inputs, starting with genetics, mediated by development and our social milieu. Different groups of individuals will on average follow alternative paths given the same set of choices because of initial differences of genetics. Consider the work that has emerged out of studying the MAOA locus. It seems that there are two forms of this gene, alleles, within the population. To grossly simplify, one form seems relatively immune to environmental input in regards to early childhood abuse (i.e., the abused do not become abusers), while another form follows the classic generalization where early childhood treatment matters a great deal in regards to how they behave as adults (i.e., the abused become abusers, the non-abusers do not). If I was a social engineer, say a director of an adoption agency, it seems clear to me that this genetic information is relevant in terms of allocating finite resources in screening and selecting parents. The reality is that the children with the loss of function form of MAOA are much more vulnerable to bad parenting, so in a circumstance where resources are rationed these children should be placed with greater surety in "low risk" households vs. those children who have more innate resiliency. I am suggesting here that we acknowledge the facultative nature of human behavior, and take advantage that genes bias, but do not determine.

Finally, several minor points. I agree with the commenter on Dave & Greta's site in regards to "working" on the environmental component of variation in a given trait (though as I pointed out, eliminating the environmental component of variation may simply render the trait "fully genetic" in population variance!). It is about cost vs. benefit, and society has at its disposal scarce resources. Also, there is a tendency for many to assume that focusing on the environmental component of variation is taking a "glass half full" attitude, but far too often I believe neglecting that genes have a strong role in shaping average outcomes in aggregates can cause serious public policy misteps. To get from A to Z you need the various parameters which will shape the decisions you make on the table. Given enough social priming & conditioning many things are possible, but how much, and what, is contingent in part on the character and nature of the genetic component, so just focusing on the environmental aspect because of its perceived malleability is short-sighted. Consider for example gene-environment correlation. One can take steps to break up these correlations, e.g., if cognitive "take off" and "mediocrity" is in part due to the segregation of children into academic and non-academic cliques early in elementary school and the subsequent feedback effects over the next decade leading into adulthood, there are clearly steps we can take. But, just saying that steps can be taken does not mean those steps will be easy, and the nature of those steps are contingent upon the nature of socialization and our confidence that once the children are no longer under the control of social engineers (e.g., at age 18) they won't simply revert back to their most "natural" state (i.e., the non-academic types won't over the years drift away from intellectual engagement without cajoling).

Lastly, the article does not mention one important point which is gaining prominence over the last decade, and that is Judith Richard Harris' argument elucidated in her two books, The Nurture Assumption and No Two Alike, and that is that most of the environmental (social) variation among humans is due to peer groups, milieu, with only about 10% being attributable to parental influence. Again, remember the facultative and conditional nature of human development, but these conclusions drawn from behavior genetics are important in how we view the issue. As it is, human intuition about the importance of parents rules the day. After all, we all think we have a Ph.D. in the university of life, one of whose major requirements is a natural psychological intuition.