Comment by tptacek
21 hours ago
Some of the stuff on Gusev's substack is pretty startling, and I highly recommend it.
Thank you for taking the time to comment here!
21 hours ago
Some of the stuff on Gusev's substack is pretty startling, and I highly recommend it.
Thank you for taking the time to comment here!
> > twin studies and they have taken in serious beating in the past decade, especially in light of GWAS.
Here's a twin study from 2015, newer than the books (Clean Slate etc) and papers you (David) mentioned.
"Thinking positively: The genetics of high intelligence" https://pmc.ncbi.nlm.nih.gov/articles/PMC4286575/
Figure 3 indicates that intelligence is pretty strongly inherited, and they arrive at 0.44.
Now you're saying that that doesn't matter because of GWAS? Sounds a bit hand-wavy to me.
> > Sasha Gusev's Substack as an entry point: https://theinfinitesimal.substack.com
Blog post looks biased. So there's a controversy: https://en.wikipedia.org/wiki/Missing_heritability_problem
And there's two camps:
https://www.clearerthinking.org/post/the-missing-heritabilit...
(I like that article!)
And the two of you (Davind and Thomas) seem to be in the "The DNA Proponents" camp. The other is "The Twin Study Advocates" camp.
I guess now I'm in "The middle ground" camp, no longer in the "Twin Study Advocates".
Thanks for that. Maybe I'll check back in 10 years later and see what has happened.
I don't think there are really warring camps here††. There are inferences drawn from twin studies, and they have not held up in GWAS studies, especially modern GWAS studies that actually try to "deconfound" indirect genetic effects. Basically: you start out with twin study inferences that give relatively high heritability numbers for complex behavioral traits; the first round of GWAS work cuts those numbers maybe in ~half; direct contribution studies (like intrafamily GWAS) cut them basically down to zero.
(For an intuition on things like intrafamily GWAS and IGE vs. DGE, consider assortive mating: tall people tend to marry tall people [that's an empirically verifiable statement]; that means they're more likely to pass down "real tall-people genes", but also that they're likely to pass down "fake tall-people genes" that they just happen to have but have nothing to do with height. If you don't correct for this --- and early GWAS EA studies didn't! --- you end up inflating genetic contribution to height by factoring in the "fake" genes --- also by increasing the likelihood of inheriting shared environmental factors. Those effects† explains something like 10% of height's h2, but most of educational attainment.)
You might get something out of Gusev's review of Turkheimer's book, also on his substack. Turkheimer is also a very interesting source for this stuff, particularly if you want to go a step past whether the h2 numbers we have are holding up (they aren't), stipulating them, and then seeing how little even that means.
"Blog post looks biased" is not a good way to address this unless you actually practice in the space, like the author does, and are in conversation with other practitioners in the space, like the author is.
† Really, effects _like_ this --- I'm terribly oversimplifying, in part because I'm several steps past the frontier of my understanding of this research.
†† Greenberg thinks there are, but who are the twin-studies researchers rejecting GWAS results?