Crediting Frans de Waal, Sapolsky writes in Behave.
“culture” is how we do and think about things, transmitted by non-genetic means.
I guess that is close to my preferred definition, which is “socially communicated thought patterns and behavioral tendencies.”
I am about half way through the book. I have two nits to pick.
One nit is that he says that when behavior correlates with a gene in one setting but not another, that proves gene-environment interaction. An example would be that a gene correlates with violence in people who were abused as children, but not in people who were not abused as children. In my view, this might be gene-environment interaction. But it also could be gene-gene interaction. That is, the behavior might be influenced by a gene other than the one on which you are focused, and that gene correlates with whether the person was abused as a child.
Another nit is when he talks about gender and math ability. First, he points out that the very top percentile in math is dominated by males (the fact that Larry Summers was fired for pointing out). Then, he reports on a study showing that male-female math differences are less in egalitarian cultures. However, that is only relevant to the Larry Summers issue if that study refers to the very top percentile. As I read the study, by Guiso, Zingales, and others, it is about averages, not the very top percentile.
The way I see it, a lot of academics are dogmatically insistent that genes matter little and the environment matters a lot. Sapolsky is not one of those, but these examples suggest that he is somewhat biased in the direction of the prevailing dogma.
Why does it matter if the gene-env interaction is mediated through another gene?
I guess by “mediated by another gene” means that the environmental part of the interaction might, itself have a genetic cause. I didn’t think that’s what AK was taking about. Such stories are possible, but stories are also possible where the gene-gene interaction on its won causes correlations in enviornmental conditions.
Simple toy model: Suppose that gene A on its won elevates the chance of environmental outcome X. And suppose A and B together elevate the chance of environmental Y.
Now we expect a correlation between X and Y, and people might come up with theories of how X causes Y. If we know about gene A but not gene B, we might come up with a theory that A and X interact to cause Y.
Even if we know about B, we might imagine that it only infulences Y by causing X. But this would be blown out of the water if the correlation A&B -> Y hold even in those cases where X did not eventuate.
Interestingly, differences between men and women in career choices tend to be *greater* in more egalitarian cultures. To focus on aptitude at the expense of other probably innate psychological difference in this case is to selectively ignore phenomena that contradict conventional wisdom.
Note that the study points out the female reading advantage increases with more equality. It also notes similar results at 95% and 99% but it is unclear whether these are common or group normalized standard deviations.