Twin studies predict that within a given (Western) country, up to 80% of the variation in IQ is heritable. And yet finding genes for intelligence appears to be analogous to finding intelligent life on other planets. We suspect they’re out there given the massive number of places for them to hide, but detecting them is the hard part. So far, out of the billions of genetic variants in the human genome, about 2,411 IQ “genes” have been found (with only about 127 of them being causal) and these explain only about 10% of the IQ variation within a Western country (not 80% like twin studies promised).

But these numbers should greatly increase as we use better measures of IQ (so far the largest studies use education level as a proxy for IQ since administering a million people a hour IQ test is much harder than just asking them their highest degree) and as genetic measures improve (so far the largest studies have only looked at a small fraction of the human genome).

Correlations: Individual vs group level

Currently polygenic scores correlate about +0.3 with IQ in Western countries (about the same as the correlation between IQ and brain size). While this correlation is not large, small correlations among individuals can become huge correlations among groups. For example, while the correlation between IQ and brain size is only about +0.3, the correlation between the average IQ and the average brain size of the 10 “races” studies by Richard Lynn in his 2006 book clocks in at 0.83! The reason for this is that as you move from individual to group level data, all the individual exceptions to a given trend tend to cancel one another out, and the underlying relationship becomes much more stark.

Applying the same concept to IQ “genes”, Davide Piffer found that the average polygenic scores of different races correlate about 0.9 with the average IQ of said “races”. So even though we’re a long way from being able to guess the IQ of an individual from his DNA alone, we can already make very reliable guesses about the average IQ of entire populations, or at least what their IQs would be if they were reared in contemporary America.

I turned to table 5 of one of Piffer’s paper (see appendix below) and noted the PGS GWAS sig scores of Northwest Europeans (IQ 100) and African Americans (IQ 85). Since the IQ gap between these two groups is thought to be 100% genetic (Lynn 2006), I simply equated their polygenic scores to their IQs, and using linear extrapolation, crudely assigned assign IQs to other groups in table 5. For example, since the polygenic scores of East Asians and Ashkenazi Jews were both about a third as far above Northwest Europeans as Africans Americans are below, I assigned them IQs that were as a third as much higher than the White mean defined as 100, as Black Americans are below (105 vs 85 respectively).

But since Black Americans are 15 IQ points below whites but only 75% bantu genetically, I assigned the congoid race an IQ 20 points below the white mean (15/0.75 = 20) which is 80, since they’re virtually 100% bantu. Then using Figures 5 and 11, I assigned Yoruba an IQ of 80 (since they’re Congoid) and the French an IQ of 100 (since they’re white) and by equating the polygenic scores of these groups to their IQs, I could linearly extrapolate to Arabs (Palestinians), Native Americans (indigenous Columbians), pygmies (mbuti), Australoids (Papuans) and Capoids (San). This method often gave wildly disparate IQs for the same race depending on whether I used figure 5 or 11 (perhaps because unlike table 5, these samples are low coverage genomes and thus unreliable) so I increased reliability by averaging. For example Papuans had an IQ of 81 using figure 5 but only 58 using figure 11, so I split the difference and assigned them IQ 70.

The first thing we notice is that the three light skinned races (East Asians, Ashkenazi Jews, and Whites) all have triple digit genetic IQs and the four dark skinned races (Pygmies, Congoids, Australoids and Capoids) all have genetic IQs below 90. The medium coloured races (South Asians, Arabs, Native Americans) are in between.

The next thing we notice is the geographically isolated races (Native Americans, Australoids, and Capoids) score lower than their skin color would predict while races that border a dark Caucasoid race (Arabs or South Asians) do not.

Conclusion

When modern humans first evolved a few hundred thousand years ago, we probably had (genetic) IQs around 70 (capoid level). Then when we finally left Africa about 70,000 years ago, IQ began to increase the further North we went, peaking at around 85 for those who reached Siberia (Native American level). Then sometime after the Native Americans colonized the New World about 15,000 years ago, there was a 10 to 20 point increase in IQ for everyone who bordered the Middle East or South Asia. This included the Congoids of sub-Sahara who jumped from 70 to 80 and the East Asians who jumped from 85 to 105. We don’t know specifically what selection pressures improved IQ by about 1 SD in the last 15,000 years but it looks like Native Americans (isolated in the New World), Australoids (isolated in Oceania) and Capoids (isolated on Africa’s Southern tip) were just too far away from the action, and that action was near the Middle East.

Whether it’s the birth of agriculture, the birth of civilization, or the start of Christianity and Islam, the Middle East has long been the contact place of different races and the land where history was written. Even today, wars in the Arab World, tensions over Iran, and the Israeli-Palestinian conflict dominate the news. It’s thus not surprising that our genetic history was written there too.

Appendix

From Divergent selection on height and cognitive ability: evidence from Fst and polygenic scores