Kees-Jan Kan recently left a couple of comments in response to some that I made concerning one of his findings. The larger issue under discussion was the nature of g. Here, I’m using a double entendre, referring both to the heritability and the biological structure of g. When it comes to the issue of individual differences and environmental malleability, both issues matter and are related. Kees-Jan is working with a model in which g is a statistical, but not biological reality (See pic below); this model more readily allows for within population differences in g to be a function of (active) gene-environment correlations. (Think of the Dickens and Flynn model.) As a result, environmental influences are less constrained as agents of individual differences. This model can be contrasted with a biological g model (e.g., Jensen, 2011), in which g is conceived of as a biological entity and in which individual differences are largely due to additive genetic effects. Here, I’m contrasting an additive genetic biological g model with a G-E correlational statistical g one. Jensen and Rushton (2010) versus Flynn and Dickens (2001). In terms of environmental malleability, these represent, in my assessment, extremes. In between, there are a number of variants. The point here is that, with regards to malleability, the biological/environment structure of g matters, too. While some of the debate about the heritability of g may be over (1), that about the structure of it is till ongoing.
(1) Apparently, Davies et al (2011) has not ended debate about the breeding value of g. My interpretation (and I was not alone here) was that Davies (2011) established high narrow heritability of g, which necessarily implies the low g-e of g, disconfirming all models of g predicted on g-e correlations, active or otherwise. Apparently, some disagree.