How likely is it that race differences in IQ are 100% environmental?
Literally impossible
As of now, the mainstream consensus is that the latest wave of out of Africa migration happened somewhere between 50,000 and 70,000 years ago, which is the wave that makes up most of the non-African population’s admixture. For the sake of argument, I’ll assume that the wave happened 50,000 years ago and that the non-African population descends completely from the OOA wave.
To calculate shifts in a population’s mean IQ based on natural selection, the following formula is used:
Z is the phenotypic change, h^2 is the narrow sense heritability, and S is the selection differential (average IQ of the population, weighed by the number of surviving children). If the Reich Lab’s recent results are to be believed, then Western Hunter Gatherers had a polygenic score for intelligence that was three standard deviations below the population mean — which corresponds to an IQ of 55 (assuming correspondence between PGS and phenotype). For now, that should serve as a conservative estimate of the average levels of intelligence of 50,000 year old hominids.
Traditionally, the narrow sense heritability of intelligence is estimated to be 60%, based on midparent~child regression studies, though this value could be potentially upwardly biased by environmental or cultural transmission. Early molecular evidence did not corroborate this, but better methods that use whole-genome GCTA or adjust for the effects of rare variants find heritability estimates closer to the ones that were found by the family studies (48% for the former, 54-76% for the latter).
Assuming that humans increased in IQ by 3 standard deviations in 50,000 years, and that the average generational time was 23.2 years, then the average IQ of each generation climbed 0.0014 standard deviations each generation, implying a mean selection differential of 0.0023.
Let me try to visualize how small that is. Let us assume that the true correlation between fertility and surviving children is perfectly zero. In the year 8000 BCE, the world population was estimated to be 5 million. If 5 million people randomly mate in a population, then just by random chance, shifts of about 25-50% of this magnitude would be expected to happen regularly.
For races to be equal in their predispositions to intelligence, the average selection differential would have to be equal for all 50,000 years. Which is simply not possible. Note that the rate at which intelligence is selected for within the Western Eurasian population has been historically nonlinear.
There was a huge explosion in intelligence about 8,000 years ago, then a plateau, followed by a fall, followed by a rise, and then a fairly constant plateau. If these patterns can differ within the same region, then it would be extremely unreasonable to think that they do not differ between regions.
Unsurprisingly, there has been no news coverage of Reich’s paper. I’m not surprised.
Where did the 55 IQ estimate for 50 kya come from? It didn't jump out at me when I looked at the paper (which, admittedly, I didn't read as scrupulously as I should have).
John Fuerst also made these drift-related arguments. By drift alone, one expects some genetic differences in phenotypic values between separated groups in random directions. Strict equality is not a thing unless there's the exact same stabilizing selection across time and place. Basically impossible. The question is not really whether there are some gaps, there must be for any phenotype, but the size of these.
By the way, you cannot just assume that PGS gaps of 1 SD correspond to 1 SD IQ gap, or in any other phenotype. The scaling functions are not known.