Creativity, Genes, and Racial Differences
Davide Piffer is an Italian evolutionary anthropologist. He received a BA in Anthropology from the University of Bologna and a Master of Science in Evolutionary Anthropology from Durham University in England. His Master’s thesis was on the sexual selection of sleep patterns among humans, and was the first to link mating behavior to chronotype (sleep patterns during a 24-hour period) within an evolutionary framework. His research later shifted to quantitative genetics, and he published one of the first studies of the heritability of creative achievement. In 2013, he switched to molecular genetics, focusing on the polygenic evolution of educational achievement and intelligence, and this remains his main focus. Within this area, his main finding is that ethnic differences in intelligence are largely explained by the thousands of genetic variants that predict cognitive abilities within populations.
Grégoire Canlorbe: As Leonardo da Vinci pointed out in his notebooks, “The black races of Ethiopia are not products of the sun, for if in Scythia a black man makes a child to a black woman, the offspring is black; but if a black man inseminates a white woman, the offspring is gray. Proof that the race of the mother has as much power over the fetus as that of the father.” Besides skin color, how do you sum up what we know — or appear to know — about heritable racial differences in traits such as intelligence, creativity, or even the rate of maturation of male voices?
Davide Piffer: As I said, creativity has been neglected by geneticists and many psychologists, so unfortunately we know next to nothing about race or individual genetic differences in creativity. With regards to intelligence, there is a growing consensus, thanks to the work of myself and colleagues, that racial differences have a genetic basis. This comes from different lines of evidence, using the most recent methods of population genetics: admixture analysis and polygenic scores.
As for voice maturation, Philippe Rusthon proposed the theory that the Mongoloids are the most K-evolved and the Negroids are the least K-evolved, while the Caucasoids fall between the two, although closer to the Mongoloids. This theory is supported by a large amount of data. I made a new contribution to Rushton’s theory by presenting data on race differences in the age at which the voice breaks in boys. The prediction from Rushton’s theory and the hypothesis to be tested is that the voice should break at a younger age in Negroids than in Caucasoids. The hypothesis was successfully corroborated.
Grégoire Canlorbe: Your most conclusive investigations in behavioral genetics deal with the connection between sexual selection and sleep patterns — or that between sexual selection and both sex- and country-level differences in performance on tests of fluid intelligence. Could you tell us more about it?
Davide Piffer: I was the first to investigate the relationship between sleep patterns and sexual behavior among humans as part of my MSc’s dissertation at Durham University. My original work was later replicated by a researcher from Sri Lanka and a group from Germany. Sleep is well known to affect mating behavior in many animal species, and some heritable differences in chronotype also predict mating success among men, both in Western and non-Western societies.
The effect seems to go above and beyond personality (extraversion) and the propensity to engage in social activities at night. Being a night owl is associated with going out at night, thus increasing the chances of meeting a member of the other sex, but also with testosterone levels. We still don’t know if it is also perceived as an attractive feature in males, but it has significant sex differences, with males across different societies being more likely to be night owls than females, so it is a candidate target for sexual selection.
Some years ago, I published a paper in which I proposed an explanation of the paradox that more developed countries, with higher equality of the sexes, had a higher sex inequality in IQ scores. That is, in more developed countries, men are smarter than women, but this sex difference is much smaller or absent in less developed countries. For example, in Muslim countries, women have higher ability than men, so we could call this the Muslim paradox. This is the opposite of what one would expect from a purely environmental perspective, because women get more education in sex-equal countries. What I found was that smarter countries had higher sex differences independent of GDP or equality of the sexes.
A potential mechanism that I did not mention in the paper but that occurred to me later is that in industrialized countries, dysgenic fertility is mainly driven by highly educated females having fewer children and later in life, whereas male intelligence is not generally related to reproductive success. This sex disparity in dysgenic fertility would cause women to decline in intelligence relative to men.
This may not seem obvious. Because both men and women contribute to the genotypes of children, the fact that high-IQ women have few children should depress the IQ of both sexes, not that of women only. The question is more subtle, however, because many complex traits have sex-specific mechanism of expression, so that the same allele has different effects in males and females, even if the genes are not located on the sex chromosomes.
Grégoire Canlorbe: You devised a methodology to detect polygenic selection, a mechanism that acts on many genetic variants simultaneously. You have done this in particular with educational attainment. How do you summarize your approach?
Davide Piffer: My approach moves away from classical methods that focused on a single gene, because most traits are polygenic — the result of contributions of many alleles, or gene variants — and the effect of each is so diluted that you need to study many genetic variants in order to detect a pattern of selection. The frequency of a single allele is mainly affected by genetic drift unless that allele has a very strong effect on a trait such as, for example, sickle-cell anemia. Natural selection also affects population frequencies of alleles, but the effect on a single allele is typically so small that it can go unnoticed. However, when you simultaneously look at hundreds or thousands of alleles, you start seeing a pattern because the effects of random drift on each allele cancel each another out, and what is left is the directional effect of natural selection, which acts more strongly on some populations than on others, according to the environmental conditions or the sexual dynamics across the millennia.
Of course, these alleles are not picked at random from the genome. They come from studies called GWAS (“genome-wide association study”) which explore the correlation between millions of genetic variants or SNPs (single-nucleotide polymorphisms) and some phenotype or trait, such as years of education or height, using very large samples (from 100,000 to over a million individuals). These studies then find the SNPs with the strongest phenotypic effect, that is those that can increase an individual’s IQ by half an IQ point or increase height by as much as a centimeter.
The average frequency of alleles weighted by their effect on the phenotype (e.g. IQ) is called the polygenic score. Polygenic scores can be used to predict how well individuals do in school, or their height, or risk for cardiovascular disease. These scores are based on direct biological evaluation of a subject’s genome, so they are unaffected by culture, family differences, or any other factor that is commonly claimed to invalidate traditional IQ testing. I calculated polygenic scores for populations by averaging the scores of samples from different ethnic groups available on public databases. In the case of educational attainment and intelligence, I used between 2,400 and 3,500 of the most significant SNPs to compute polygenic scores.
What I found was that these population-level polygenic scores had a very high correlation of r=0.9 with scores on standardized intelligence tests. Ashkenazi Jews have the highest polygenic scores, followed by East Asians, then Europeans, South Asians, Native Americans and Blacks. A correlation this high would occur at odds of only one in 46,000 if the SNPs had been chosen at random. Furthermore, the SNPs associated with the greatest intelligence differences between populations were also associated with the greatest differences between individuals within a single population. This supports the view that both kinds of difference were the result of selection.
Grégoire Canlorbe: You made the claim that the north-south difference in Italy in fluid intelligence should be understood in terms of genetic differences between the populations. What data corroborate that hypothesis?
Davide Piffer: Historically, literacy levels and economic prosperity have been higher in the north than in the south. Data collected by Richard Lynn have shown that these differences are reflected in scores on tests of intelligence and scholastic aptitude (PISA and INVALSI).
There are strong genetic differences within Italy, recently corroborated by an in-depth study (Raveane at al., 2019). These differences were mostly established in pre-Roman times: Bronze-Age migrations from the Eastern European steppes (Indo-Aryans) in the north and West Asians from the Caucasus in the south. The Latin people who founded ancient Rome belonged to the Indo-European/Aryan group, along with other Italic tribes, such as the Veneti who later founded Venice.
Later migrations strengthened the pre-existing differences. In the first millennium BC, groups of Celtic people (originating from Indo-Aryans, like the native northern Italians such as the Veneti and the Ligurians) settled in the north and mixed with their ethnic cousins (that is why the Roman name for northern Italy was Gallia Cisalpina), and the Greeks heavily colonized the South (Magna Graecia) in the 1st millennium BC. In the Middle Ages, Germanic people invaded Italy and mostly added to the genetic pool of northern Italy (mostly Lombards and Gothic peoples) with some pockets in central and southern Italy (the Lombards in the Ducato di Benevento and the Normans in Palermo), whereas the south was conquered by Arabs.