A Roundup of Population Genetic Research on Selection for Educational Attainment, Intelligence and Brain-Related Traits

Davies et. al (2017) reports that of their neuropsychiatric phenotypes, only bipolar disorder had significant population differentiation (Fₛₜ) while only extraversion (though it did not survive the removal of particular problematic SNPs) and schizophrenia had significant signals of adaption (Qₓ). For their brain phenotypes, only the hippocampus and putamen showed signals of adaptation (Qₓ), while intracranial volume did not. Moreover, the sign of their adaptation showed that there was adaptation for small hippocampal volumesnot larger volumes.

Sanjak et. al (2017) reports slight selection against educational attainment in females and slight (insignificant) selection for educational attainment in males (Sanjak et. al 2018). They also report positive γ (quadratic selection gradient), indicating disruptive selection. However, they also note that all linear selection gradients are quantitatively small and unlikely to have significant phenotypic consequences: “Such weak selection gradients are unlikely to lead to large changes in phenotypic distributions over clinically or socially relevant timescales”. I’m still confused about Table S1 in their results, which seems to report positive selection for educational attainment.

Kong et. al (2017) found that there has been a slight decrease in PGS for educational attainment.

Guo et. al (2018) finds no selection for educational attainment in two samples (Table 1).

Racimo et. al (2018) reports selection for educational attainment in Asians populations, but cautions this could be on a correlated trait (see also Novembre & Barton 2018).

Sanjak et. al (2017) found evidence for slight selection against educational attainment in modern humans, though the estimate differed by sex (Sanjak et. al 2018).

Srinivasan et. al (2018) utilized a post-Neanderthal selective sweep index and found that there was positive selection for intelligence, educational attainment and college completion.

Zeng et. al (2018) reports negative selection for educational attainment in UK BioBank.

Holland et. al (2019) also find that intelligence is under negative selection.

Speidel et. al (2019) reported neither an increase nor a decrease in educational attainment PGS for most of the populations in the 1000 Genomes Project, though Berbers, Iberians, Finnish, British & Gujarati Indians showed marginally significant increases.

Tilot et. al (2019) finds no evidence for local adaptation on cortical surface area.

Urricchio et. al (2019) reports positive selection for educational attainment in UK BioBank.

Zeng et. al (2019) reported selection against both educational attainment and fluid intelligence in UK BioBank.


We should be very wary when it comes to interpreting these tests, they have been known to have significant false positive rates in the presence of population stratification (Berg et. al 2019Sohail et. al 2019), and the complexities of negative selection, gene-environment interactions and population stratification presents simple interpretation of the results of these tests (Coop 2019; Durvasula & Lohmueller 2019; Mostafavi et. al 2019; O’Connor et. al 2019). It is not clear that the PGS for these traits even represents a genetic mechanism (Baverstock 2019; Richardson 2017Richardson & Jones 2019).