Disentangling prenatal and inherited influences in humans with an experimental design:

Exposure to adversity in utero at a sensitive period of development can bring about physiological, structural, and metabolic changes in the fetus that affect later development and behavior. However, the link between prenatal environment and offspring outcomes could also arise and confound because of the relation between maternal and offspring genomes. As human studies cannot randomly assign offspring to prenatal conditions, it is difficult to test whether in utero events have true causal effects on offspring outcomes. We used an unusual approach to overcome this difficulty whereby pregnant mothers are either biologically unrelated or related to their child as a result of in vitro fertilization (IVF). In this sample, prenatal smoking reduces offspring birth weight in both unrelated and related offspring, consistent with effects arising through prenatal mechanisms independent of the relation between the maternal and offspring genomes. In contrast, the association between prenatal smoking and offspring antisocial behavior depended on inherited factors because association was only present in related mothers and offspring. The results demonstrate that this unusual prenatal cross-fostering design is feasible and informative for disentangling inherited and prenatal effects on human health and behavior. Disentangling these different effects is invaluable for pinpointing markers of prenatal adversity that have a causal effect on offspring outcomes....

I guess that's what economists would call a "natural experiment." In any case, these results are in light with the sort of arguments Judith Rich Harris lays out in her books. Much of the correlation we see behaviorally between parents and children actually being genetic, even though we have prior assumptions as to the importance of socialization in the behaviors in question. Here's the primary figure:

From the results:

These results therefore point to the importance of inherited factors in the association between prenatal smoking and offspring antisocial behavior and suggest that gene-environment correlation is important in explaining this association

The data show that mothers who smoked in the sample tended to have anti-social offspring. There was a correlation. But when broken down by relatedness the data showed that the effect existed among related mothers and offspring, but not unrelated. The inference then is that the smoking-anti-sociality likely had more to do with personality traits in women who would smoke while pregnant, and who would pass those personality traits to their offspring.

Addendum: What kind of maniac would go through in vitro and then smoke! Just sayin'.

Labels: Behavior Genetics