Anthropology.net
Evolutionary Insights by Anthropology.net
Race Realism vs. 15,836 Ancient Genomes: What They Show
0:00
-1:04:47

Race Realism vs. 15,836 Ancient Genomes: What They Show

Race science now cites polygenic scores and ancient DNA. Read by the standards of the geneticists who built those tools, the evidence runs the other way.

In April 2026, a team led by Ali Akbari and David Reich at Harvard analyzed1 ancient DNA from 15,836 people who lived across West Eurasia, 10,016 of them with newly generated data, and reported that natural selection had pushed hundreds of genetic variants up or down in frequency over the past ten thousand years. Most of the paper, which appeared in Nature, is about allele frequencies and statistical method. One sentence in the abstract is about something else: combinations of alleles that predict cognitive test performance in living people became more common over that span. When the team posted an earlier version as a preprint in September 2024,2 Emil Kirkegaard wrote in the online magazine Aporia nine days later that “the hunter-gatherers were apparently much duller than the farmers.”

Readers of this site will know the standard reply to race realism, and Jonathan R. Goodman of the University of Cambridge and Rebecca Sear of Brunel University of London made it well in an essay for The Conversation on 13 November, which Ars Technica3 republished this week. Richard Lewontin showed4 in 1972 that most human genetic variation lies within populations rather than between them; ancient DNA shows that every living population is a mixture of older ones; so the idea of genetically distinct races with innate differences in intelligence falls apart. I agree with their verdict. But the two genetic arguments at the center of that reply were built to answer an older kind of race science, the kind that argued from skulls and blood groups. Today’s hereditarians argue from polygenic scores (single numbers that add up a person’s many small genetic associations with a trait) and from selection scans of ancient DNA, which are the newest tools of population genetics itself. One of the companies at the center of the movement renamed itself Polygenic Scores LLC in April 2025.

“We believe it is unequivocal that race science is fundamentally pseudoscience.”

Jonathan R. Goodman and Rebecca Sear, The Conversation, 2026

So this essay picks up where theirs stops. The hereditarian case in 2026 has to be answered inside statistical genetics, on its own terms, and when it is, the answer is stronger than the one Lewontin gave. The controls that geneticists developed over the past decade (family-based studies, tests of how well scores travel between populations, and corrections for hidden population structure) cut directly against the claim that genes explain differences in average test scores between racial groups. Getting there requires conceding something race realists like to point out, which is that human genetic ancestry is real and can be measured.

Is race biological? What Lewontin’s 85 percent does and doesn’t show

Lewontin’s 1972 study took blood-group and protein data, sorted human populations into seven “races” as anthropologists of the period drew them, and asked how total diversity was divided. About 85.4 percent fell within populations, 8.3 percent between populations within the same race, and 6.3 percent between races. Studies with far larger datasets have since put the within-population share anywhere from roughly 80 to 95 percent, and in a 2022 review5 the geneticist John Novembre concluded that the qualitative result has held up for half a century. Lewontin drew a blunt conclusion from his numbers.

“...racial classification is now seen to be of virtually no genetic or taxonomic significance...”

Richard C. Lewontin, “The Apportionment of Human Diversity,” Evolutionary Biology, 1972

In 20036 the Cambridge statistician and geneticist A. W. F. Edwards called the reasoning that people drew from this “Lewontin’s fallacy.” His point was that the information separating populations is “hidden in the correlation structure of the data,” spread across many loci, so that small differences at each one add up. The arithmetic is on his side. When Noah Rosenberg and colleagues genotyped 1,056 people from 52 populations at 377 loci in 2002,7 they found that differences among individuals within populations accounted for 93 to 95 percent of genetic variation, and yet a clustering program given no geographic labels sorted those people into groups corresponding largely to major geographic regions. David Witherspoon and colleagues showed in 20078 why both findings hold at once. With a handful of loci, two people from different continents are more similar to each other than two people from the same population about a third of the time. With many thousands of loci, for populations sampled far apart, that almost never happens.

This is the part of the argument I think science writers handle worst. When an essay tells readers that human variation doesn’t fall into clusters, a race realist can pull up any commercial ancestry report and show a pie chart. Ancestry inference works because populations separated by distance and history drifted apart slightly in allele frequency, and enough slight differences let a computer estimate where someone’s recent ancestors lived. That is a statement about genealogy and geography, and it does not recover the races of the census form or the nineteenth-century textbook, as the 2002 paper itself explains.

Rosenberg’s team warned that because they sampled whole populations rather than people spread evenly across the map, their clusters were probably sharper than random worldwide sampling would produce, and that the differences they detected came mainly from gradual changes in allele frequency across regions. The number of clusters is also a setting the analyst picks. When the program was asked for a sixth group, it did not split Africa or Europe; it pulled out the Kalash, a single isolated community in northwestern Pakistan. Clusters track isolation and genetic drift, which is why a small mountain population can earn its own. A 2020 analysis9 of about 92 million variants by Arjun Biddanda, Daniel Rice, and John Novembre found that a variant common in one region is usually common in every region (63 percent of such variants are widespread, rising to 82 percent outside Africa), and that variants common in one non-African region and absent everywhere else are strikingly rare. When two people differ genetically, the variants responsible are most often ones found worldwide.

Africa makes the point sharper. The Human Heredity and Health in Africa (H3Africa) consortium sequenced 426 genomes from 50 ethnolinguistic groups in 13 countries and reported in 2020 more than 3 million variants never seen before, with discovery still not leveling off. The continent has roughly 2,000 ethnolinguistic groups, and only a small fraction have been studied. A racial category like “Black” groups the most genetically diverse people on Earth with one another and with descendants of the Atlantic slave trade, whose ancestry is itself mixed. In a 2015 study of 23andMe customers, African Americans averaged 73.2 percent African and 24.0 percent European ancestry, and most people with less than 28 percent African ancestry identified as European American. About 3.5 percent of European Americans carried at least 1 percent African ancestry. In Brazil, a 2003 study10 by Flávia Parra, Sérgio Pena, and colleagues found that color, as judged by appearance, was “a poor predictor of genomic African ancestry” at the individual level. The All of Us program reported in 2025, from 230,016 genomes, that people within each self-identified racial and ethnic group show gradients of ancestry rather than discrete clusters.

The scientific societies have reached the same conclusion. In 2019 the American Association of Biological Anthropologists stated that race does not accurately represent human biological variation, past or present. In 202311 a committee of the National Academies of Sciences, Engineering, and Medicine went further for researchers in genetics.

“Researchers should not use race as a proxy for human genetic variation.”

National Academies of Sciences, Engineering, and Medicine, Using Population Descriptors in Genetics and Genomics Research, 2023

Governments, meanwhile, write and rewrite the categories. “Mexican” was a race on the US census once, in 1930, and Asian Indians were listed as “Hindus” from 1920 to 1940 regardless of religion. In March 2024 the Office of Management and Budget added a Middle Eastern or North African category, separate from White, and described its standards as sociopolitical constructs that make no attempt to define race biologically. A year later, an executive order criticized a Smithsonian American Art Museum exhibition for presenting race as a social construct, quoting the exhibition’s statement that “Race is a human invention.” On 10 September 2026 the Census Bureau proposed removing race and ethnicity questions from the decennial census altogether, citing “substantial doubts about the meaning, significance, and impact” of sorting people this way; the comment period closed on 2 November. A category that agencies create and retire by rule is the social construction argument in its plainest form. The anthropological position has never been that such categories are empty, though. UNESCO’s 1950 statement12 called race “not so much a biological phenomenon as a social myth,” and that myth has shaped where people lived and what schooling they received for centuries, which is exactly why it contaminates the genetic data that race realists now rely on.

What ancient DNA shows about racial purity

Goodman and Sear’s second genetic argument survives scrutiny better, and the past decade of ancient DNA gives it specifics a short essay can’t carry. Iosif Lazaridis and colleagues showed in 201413 that most present-day Europeans descend from at least three deeply divergent populations: western hunter-gatherers, ancient north Eurasians, and early European farmers. In 201514 Wolfgang Haak’s team found that Corded Ware people in Late Neolithic Germany traced about 75 percent of their ancestry to Yamnaya herders from the steppe, and a 2025 study15 led by Lazaridis found that the Yamnaya were themselves a mixture, drawing about four-fifths of their ancestry from a population in the Caucasus and lower Volga region. Britain’s population changed profoundly twice in under two thousand years. Iñigo Olalde’s team estimated16 that the spread of the Beaker complex, after about 2500 BCE, replaced approximately 90 percent of it within a few hundred years, and Nick Patterson’s team found17 that migrants arriving between roughly 1000 and 875 BCE contributed about half the ancestry of people in England and Wales by the Iron Age.

Gene flow ran into Africa too. A woman buried at Luxmanda in Tanzania about 3,100 years ago carried 38 percent of her ancestry from a population related to the pre-pottery farmers of the Levant, according to a 2017 study18 led by Pontus Skoglund. Even Neanderthal DNA turns up on both sides: Lu Chen and colleagues reported in 202019 that individuals in African populations carry on average about 17 million base pairs of Neanderthal-derived sequence (Europeans carry about 51 million), most of it shared with non-Africans and explained partly by ancient back-migration into Africa.

The traits most often treated as racial markers turn out to have separate histories, and I find that a stronger point than any admixture percentage. The main light-skin variant at the SLC24A5 gene arrived in Europe with farmers from Anatolia, where it was already common, as Iain Mathieson and colleagues showed in 2015.20 Cheddar Man, a hunter-gatherer from Somerset who lived about 10,000 years ago, most likely had dark to black skin and blue or green eyes. A 2025 study21 by Perretti, Barbujani, and colleagues inferred pigmentation from 348 ancient genomes and found dark or intermediate skin in about half the individuals well into the Bronze and Iron Ages, although their Iron Age sample is only 11 people and the authors call their skin-color categories somewhat arbitrary. In Africa, Nicholas Crawford, Sarah Tishkoff, and colleagues studied 1,570 people in 201722 and found that the light SLC24A5 variant entered East Africa through gene flow from outside, while light and dark variants at several other pigmentation genes have coexisted in the human lineage for hundreds of thousands of years. One HERC2 variant associated with lighter skin is most common today in Europeans and in the San of southern Africa, and it dates to roughly 921,000 years ago.

Milk digestion tells a parallel story. Tishkoff’s team tested 470 people from 43 ethnic groups in Tanzania, Kenya, and Sudan in 200723 and found three different variants allowing adults to digest lactose, none of them the European variant, which was absent, and the most common East African variant spread under strong selection within roughly the last 3,000 to 7,000 years. In Europe, Richard Evershed and colleagues traced24 dairying through about 7,000 animal-fat residues from 554 archaeological sites and found that people drank milk for millennia before the European variant reached appreciable frequency around 2000 BCE, nearly 3,000 years after it first appears in the ancient DNA record. If races were old natural units, the traits used to tell them apart would share a single history, whereas pigmentation and milk digestion each have their own, and neither map lines up with the racial one.

Can polygenic scores reveal racial differences in intelligence?

A genome-wide association study, or GWAS, scans hundreds of thousands or millions of people for genetic variants statistically associated with a trait. A polygenic score adds up a person’s variants, each weighted by its estimated association, into a single number. The largest such study of educational attainment, published by Aysu Okbay and colleagues in 202225 with about 3 million participants, produced a score that explains 12 to 16 percent of the variation in years of schooling among people of European ancestry. In two US samples of people with African ancestry, the same score explained 1.3 and 2.3 percent, roughly 11 and 15 percent of its accuracy in Europeans.

The hereditarian move is to compute average scores for different populations and read a higher average as a genetic advantage. Three findings from inside genetics block it, and the first is that scores lose accuracy continuously with genetic distance from the people they were built on. Yi Ding, Bogdan Pasaniuc, and colleagues showed in 2023,26 across 84 traits, that accuracy falls almost in lockstep with genetic distance from the training sample (a correlation of negative 0.95), and that even among people labelled European, the most distant tenth had 14 percent lower accuracy than the closest tenth. The authors urged the field to “move away from discrete genetic ancestry clusters” and treat ancestry as a continuum. A difference in average scores between two groups partly measures where the GWAS volunteers came from.

The second finding is that a population GWAS captures more than the direct effects of genes. It also picks up anything that travels with genes in a given society, such as parents’ education and wealth, or patterns of who marries whom. Siblings offer a natural control, because each inherits a random half of each parent’s genome while growing up in the same household. In 202227 Laurence Howe and colleagues compared population estimates with within-sibship estimates in 178,086 siblings from 19 cohorts. The estimated genetic effects on educational attainment shrank by 47 percent, those on cognitive ability by 22 percent, and those on height by 10 percent. In Iceland, Augustine Kong’s team found in 201828 that the variants parents carried but did not pass on still predicted their children’s educational attainment, at 29.9 percent of the strength of the variants they did transmit, which means part of the signal works through the family environment parents create. A family-based GWAS preprint29 by Tan, Young, and colleagues, covering 17 cohorts and 34 traits, attributes 48.3 percent of the population signal for educational attainment to factors uncorrelated with direct genetic effects, compared with 10.7 percent for height, and names population stratification as the likely main cause.

I consider this the most damaging evidence against the hereditarian use of polygenic scores, and it comes from behavior geneticists rather than from their critics. A score that is roughly half confounding in the society where it was built cannot be used to compare populations whose access to schooling and wealth has differed for centuries. Kevin Bird ran the test directly in a 2021 paper30 in the American Journal of Physical Anthropology. Using 503 people of European ancestry and 504 of African ancestry from the 1000 Genomes Project, he found that educational attainment scores built from population GWAS weights differed between the groups by more than all but one of 10,000 comparison scores built from randomly chosen variants. When he rebuilt the scores with within-family weights, the squared difference fell by about 95 percent and became indistinguishable from chance, and the variants involved were no more differentiated between continents than control variants were. His conclusion was that the hereditarian claims are “not supported in the least.”

Fairness requires the strongest counterpoint. Cognitive test scores appear less confounded than schooling, as Howe’s 22 percent figure suggests. In April 2026, researchers from the genetics company Herasight reported in the journal Intelligence & Cognitive Abilities that most of their cognitive score’s predictive power survived within families, and that about two-thirds of its accuracy carried over to 1,491 African American children in the US Adolescent Brain Cognitive Development (ABCD) study.31 That is the best current case that a cognitive polygenic score captures real, partly portable genetic effects. The paper makes no claim about group averages. Its authors are employees and equity holders of the company, the journal’s editor appended a note that one of three reviewers recommended rejection, and the journal was launched in 2025 by the former editor of Intelligence after Elsevier appointed new ones.

Even a score that was entirely direct and fully portable would not settle a question about group averages, which is the third problem. Arbel Harpak and Molly Przeworski showed in 202132 that when environments change, the average polygenic score of a population can drift while the trait it predicts stays put, particularly when the trait is under stabilizing selection.

“...their mere existence implies nothing about their relationship to trait differences.”

Arbel Harpak and Molly Przeworski, PLoS Biology, 2021

Geneticists have already watched a polygenic signal of this kind dissolve. Between 2012 and 2018, a series of studies, including an ancient DNA analysis from Reich’s own group, reported that natural selection had made northern Europeans taller than southern Europeans. Most relied on effect sizes from the GIANT consortium, a meta-analysis of about 253,000 people. In 2019 two teams, one led by Jeremy Berg and Graham Coop33 and the other by Mashaal Sohail and Shamil Sunyaev,34 reran the tests with effect sizes from the UK Biobank’s roughly 500,000 participants, a more uniform sample. The signal weakened sharply or vanished, because subtle population structure in the original GWAS had leaked into the scores.

“Therefore, the conclusion of strong polygenic adaptation now lacks support.”

Jeremy J. Berg and colleagues, eLife, 2019

IQ test results themselves keep pointing to the environment. Bernt Bratsberg and Ole Rogeberg35 analyzed IQ scores from 736,808 Norwegian men born between 1962 and 1991 and found that the twentieth-century rise in scores, its peak with men born around 1975, and the decline afterward all appear within families, among brothers, which rules out a shift in who was having children. William Dickens and James Flynn found in 200636 that Black Americans gained 4 to 7 IQ points on non-Hispanic white Americans between 1972 and 2002, a period far too short for genetic change.

Did Europeans evolve to be smarter? Reading the 2026 Nature selection study

That brings us back to Akbari and Reich, the study I read most closely for this piece, because it is the one race realists will cite for the next decade. Their method tests for consistent trends in allele frequency across thousands of dated ancient genomes, which is meant to separate sustained selection from the shifts that migrations and population structure produce. They report 479 independent locations in the genome with a greater than 99 percent probability of directional selection and estimate selection coefficients for 9.7 million variants. To ask about complex traits, they added up the changes in alleles that predict each trait today. On their scale, where 1.0 means the average score moved by one standard deviation of present-day variation over ten thousand years, the score for intelligence test results rose by 0.74, household income by 1.12, and years of schooling by 0.63, mostly before about 2,000 years ago.

The authors hedge these results heavily. They note that the traits matter only in modern societies and could not have been measured in preliterate ones across most of the period when selection acted. They report that the same alleles are strongly correlated with alleles that lower diabetes-related traits, so selection could have been acting on metabolism and carried the schooling predictor along (my own guess at a candidate would be the change in diet that came with farming). They point out that in twentieth-century Iceland, selection on the schooling score ran in the opposite direction. And when they rebuilt the schooling score with direct-effect weights from family-based GWAS, the signal was only nominally significant, while scores built from brain-volume GWAS showed nothing.

“...it remains unclear how these relate to phenotypes that were adaptive in the past.”

Ali Akbari and colleagues, Nature, 2026

Harvard Medical School’s press release was plainer still: “So these results do not mean that Europeans evolved to be smarter or healthier.” I agree, and I would go further. The trait labels are proxies built from twentieth-century data, and the schooling score, which the authors rebuilt from family-based direct effects, measures the trait that the Tan and Young study finds roughly half confounded in ordinary GWAS; once rebuilt, its signal barely clears significance. After the height episode, I would not treat a polygenic selection signal on a trait this confounded as established until it replicates with within-family weights at full statistical power.

The deeper problem for the race realist reading is that the study compares no one. It traces allele frequencies through time within West Eurasia. Claims about racial hierarchy need the same kind of time series from Africa, East Asia, and the Americas, built with scores that work equally well in each. Those series do not exist. When Mark Lipson and colleagues published ancient genomes from sub-Saharan foragers in 2022,37 they reported six new individuals spanning about 18,000 years, and those six doubled the time depth of ancient DNA from the region. Hereditarian authors have tried to fill the gap themselves. Davide Piffer and Emil Kirkegaard published an analysis of European ancient genomes in Twin Research and Human Genetics in 202438 claiming selection for cognitive scores, and they reported finding no correlation between latitude and their intelligence scores, which is the opposite of what the cold winters theory predicts. As far as I can find, no formal reply to the Akbari paper has appeared in print yet.

Who still publishes race science, and who pays for it?

Political demand for these claims is open. At the World Economic Forum in Davos on 21 January 2026,39 President Donald Trump said of Somalis, “I always say: ‘These are low-IQ people.’” Temporary Protected Status for Somalis ended on 14 August 2026, after a federal judge lifted a stay, and for Haitians after the Supreme Court allowed the termination in Mullin v. Doe on 25 June.40

Talk of national IQ draws largely on a dataset assembled by Richard Lynn. Sear’s audit of it, circulated as a preprint in 2022,41 found that of the 25 sub-Saharan African countries with measured data, 40 percent rest on a single sample, and that 72 percent of the estimates come from children; the figure for Angola rests on 19 people, and the one for Somalia on child refugees tested in a camp in Kenya. Jelte Wicherts and colleagues, using explicit inclusion criteria, put the average for sub-Saharan Africans at about 82 against UK norms in 2010,42 where Lynn had put it below 70. Sear has since written to editors at 18 journals and tracks 174 studies built on Lynn’s data. She told Retraction Watch in November 202543 that “there is absolutely no scientific merit whatsoever in the National IQ database.” Proceedings of the Royal Society B retracted a 2010 paper built on it, a decision reported in March 2024, and Cross-Cultural Research retracted another in November 2025. Elsevier has been reviewing Lynn’s papers in its journals since December 2024, and three of his papers now carry expressions of concern.

Money for this work has an older lineage. The Pioneer Fund was incorporated in 1937 by Wickliffe Draper, heir to a textile-machinery fortune, with the eugenicist Harry Laughlin as its first president, and Draper secretly financed the launch of Mankind Quarterly in 1960. An investigation published by The Guardian and HOPE not hate in October 202444 found that Kirkegaard had incorporated the Human Diversity Foundation as a Wyoming company in November 2022, that Andrew Conru, founder of the dating site Adult Friend Finder, had put about $1.3 million into it, and that its leaders described taking over what remained of the Pioneer Fund’s money, about $277,873 in its latest filings. Conru cut ties after publication. The foundation was renamed Polygenic Scores LLC in April 2025.

Access to data is the newer front. In January 2026 The New York Times reported that Bryan Pesta, Kirkegaard, John Fuerst, and Jordan Lasker had used NIH-controlled data from the ABCD study, which follows more than 11,000 children, and from the Philadelphia Neurodevelopmental Cohort, about 9,500 children, to produce at least 16 papers claiming genetic links to racial gaps in IQ. The NIH had suspended Pesta’s access in 2021. The ABCD study now says it “condemns this misuse of participant data,” and Bird and Jedidiah Carlson documented similar misuse of the 1000 Genomes and Human Genome Diversity Project resources in 2024.45 Carlson, Brenna Henn, and colleagues showed in Nature in 202246 that the document left by the gunman who attacked a Buffalo supermarket that May reproduced a figure from a population genetics paper, and that its pseudoscientific claims “can be traced to discussions about research papers.” His federal trial opened with jury selection on 13 October 2026.

Journals remain the main route to respectability. Richard Haier, editor of Intelligence from 2016 to 2024, defended publishing research on group differences in a 202047 editorial on academic freedom. Elsevier appointed Dragoș Iliescu and Samuel Greiff as co-editors from January 2025; members of the board resigned, and Haier and Thomas Coyle founded Intelligence & Cognitive Abilities, which published its first issue that June. The psychologist Andrew Winston48 has described scientific racism as a “community project” with its own journals and funding, and in 202449 he, Bird, and John Jackson argued in American Psychologist that psychology should borrow the evidentiary standards of genetics.

“...the current racial hereditarian work would be ineligible for publication.”

Kevin A. Bird, John P. Jackson Jr., and Andrew S. Winston, American Psychologist, 2024

Events around Nathan Cofnas, the philosopher Goodman and Sear mention in their opening, have moved quickly. In February 202450 he published “A Guide for the Hereditarian Revolution” on his Substack, and Emmanuel College, Cambridge, ended his research affiliation that April; Cambridge concluded in October 2025 that his views broke no law or university regulation. Ghent University hired him as a postdoctoral researcher in March 2026, and more than 300 staff and students protested the appointment. On 21 July he published plagiarism allegations against the Cambridge sociologist Jason Arday, who resigned on 5 August and was found dead in London on 14 August, aged 41; Cambridge has commissioned an independent review of how it handled the allegations and supported him. Ghent suspended Cofnas on 20 August and opened a preliminary disciplinary inquiry. The next day Bill White, the US ambassador to Belgium, emailed the rector asking the university to rescind the decision, and he later said on X that the United States would review its relationships with Ghent. Ghent then confined Cofnas to working from home for three months without teaching, and on 30 September Unia, Belgium’s equality body, said it would take legal action under the country’s anti-racism law. The Conversation essay describes the ambassador’s intervention as a response to an inquiry into how Cofnas was hired; the reporting I could find ties it to the suspension and the disciplinary inquiry.

What would it take to settle the question?

My verdict is that the hereditarian claim of innate differences in intelligence between racial groups has no support in current genetics, and the evidence for that verdict is stronger now than it was a decade ago. The newest tools are what make the difference. Family-based studies show that the population signals race realists rely on are heavily confounded for schooling, and portability studies show that score differences between groups track genetic distance from the GWAS volunteers. The one large selection scan that reports change in a cognitive predictor comes with caveats from its own authors that rule out the reading race realists want. Where I would update Goodman and Sear is in emphasis: concede plainly that ancestry is real and measurable, rather than saying only that variation doesn’t fall neatly into clusters, and make the case from within-family genetics, where hereditarians cannot accuse critics of ignoring the data.

A real test of the hereditarian hypothesis would need direct-effect polygenic scores, estimated within families, that predict equally well in every population being compared, plus ancient DNA time series from Africa and Asia on the scale of the West Eurasian one. Even then, Harpak and Przeworski’s argument means a difference in average scores would not by itself show a difference in the trait, so the honest conclusion may be that genetic data collected inside societies organized by race cannot answer the question in the form hereditarians ask it.

The first requirement is the place to watch, because it is the one geneticists can actually meet. The family-based GWAS that supplied the direct-effect weights for the Akbari team’s own check drew on 17 cohorts: 16 of European ancestry, one East Asian, and none from Africa or the African diaspora.

Further Reading

Human variation, ancestry, and race categories

  • Bryc, K., Durand, E. Y., Macpherson, J. M., Reich, D., & Mountain, J. L. (2015). The genetic ancestry of African Americans, Latinos, and European Americans across the United States. American Journal of Human Genetics, 96(1), 37–53. https://doi.org/10.1016/j.ajhg.2014.11.010

  • Choudhury, A., Aron, S., Botigué, L. R., et al. (2020). High-depth African genomes inform human migration and health. Nature, 586, 741–748. https://doi.org/10.1038/s41586-020-2859-7

  • Fuentes, A., Ackermann, R. R., Athreya, S., et al. (2019). AAPA statement on race and racism. American Journal of Physical Anthropology, 169(3), 400–402. https://doi.org/10.1002/ajpa.23882

  • Gouveia, M. H., Meeks, K. A. C., Borda, V., et al. (2025). Subcontinental genetic variation in the All of Us Research Program: Implications for biomedical research. American Journal of Human Genetics, 112(6), 1286–1301. https://doi.org/10.1016/j.ajhg.2025.04.012

  • Office of Management and Budget. (2024). Revisions to OMB’s Statistical Policy Directive No. 15. Federal Register, 89, 22182. https://www.federalregister.gov/documents/2024/03/29/2024-06469/

Ancient DNA, admixture, pigmentation, and lactase persistence

  • Brace, S., Diekmann, Y., Booth, T. J., et al. (2019). Ancient genomes indicate population replacement in Early Neolithic Britain. Nature Ecology & Evolution, 3, 765–771. https://doi.org/10.1038/s41559-019-0871-9

Polygenic scores, selection, and test scores

  • Wolfram, T., Moore, S., Li, J. H., Anomaly, J., Davidson, I., & Christensen, M. (2026). Interpreting polygenic prediction of cognitive ability: Evidence for direct, reliable, and portable genetic effects. Intelligence & Cognitive Abilities, 2(1), 1–19. https://doi.org/10.65550/001c.158459

National IQ data, publishing, and the race science network

  • McIntire, M. (2026, January 24). Genetic data from over 20,000 U.S. children misused for race science. The New York Times.

  • Pegg, D., Burgis, T., Devlin, H., & Wilson, J. (2024, October 16). Revealed: International “race science” network secretly funded by US tech boss. The Guardian.

Political and institutional record

  • EUobserver. (2026, August). US ambassador to Belgium claims he got Ghent academic reinstated after Jason Arday scandal. https://euobserver.com/234180/us-ambassador-to-belgium-claims-he-got-ghent-academic-reinstated-after-jason-arday-scandal/

  • Executive Order 14253, Restoring Truth and Sanity to American History (March 27, 2025). https://www.govinfo.gov/content/pkg/DCPD-202500408/html/DCPD-202500408.htm

  • Pew Research Center. (2025, November 3). The changing categories the U.S. has used to measure race. https://www.pewresearch.org/short-reads/2025/11/03/the-changing-categories-the-us-has-used-to-measure-race/

  • Retraction Watch. (2024, March 18). Do some IQ data need a public health warning? A paper based on a controversial psychologist’s data is retracted. https://retractionwatch.com/2024/03/18/do-some-iq-data-need-a-public-health-warning-a-paper-based-on-a-controversial-psychologists-data-is-retracted/

  • Times Higher Education. (2026, August). Former Cambridge professor Jason Arday found dead. https://www.timeshighereducation.com/news/former-cambridge-professor-jason-arday-found-dead

  • Times Higher Education. (2026, September 30). Nathan Cofnas to face legal action in Belgium over alleged racism. https://www.timeshighereducation.com/news/nathan-cofnas-face-legal-action-belgium-alleged-racism

  • U.S. Census Bureau. (2026, September 10). Decennial Census of the Population of Americans; Proposed Residence Criteria and Proposed Regulations for Demographic Questions [Proposed rule].

  • University of Cambridge. (2025, October 2). Statement on disciplinary investigation: Dr Cofnas. https://www.cam.ac.uk/notices/news/statement-on-disciplinary-investigation-dr-cofnas

  • VRT NWS. (2026, August 27). Ghent University imposes 3-month mandatory home-working period on Nathan Cofnas. https://www.vrt.be/vrtnws/en/2026/08/27/ghent-university-imposes-3-month-mandatory-home-working-period-o/

More sources

  • Fuentes, A. (2022). Race, Monogamy, and Other Lies They Told You: Busting Myths about Human Nature (2nd ed.). University of California Press.

  • Goodman, J. R. (2026). How the social brain hypothesis undermines hereditarianism. Evolutionary Human Sciences, 8, e38. https://doi.org/10.1017/ehs.2026.10067

  • Harden, K. P. (2021). The Genetic Lottery: Why DNA Matters for Social Equality. Princeton University Press.

  • Marks, J. (2017). Is Science Racist? Polity.

  • Panofsky, A. (2014). Misbehaving Science: Controversy and the Development of Behavior Genetics. University of Chicago Press.

  • Reich, D. (2018). Who We Are and How We Got Here: Ancient DNA and the New Science of the Human Past. Pantheon.

  • Saini, A. (2019). Superior: The Return of Race Science. Beacon Press.

  • Slobodian, Q. (2025). Hayek’s Bastards: Race, Gold, IQ, and the Capitalism of the Far Right. Zone Books.

1

Akbari, A., Perry, A., Barton, A. R., et al. (2026). Ancient DNA reveals pervasive directional selection across West Eurasia. Nature, 654, 419–428. https://doi.org/10.1038/s41586-026-10358-1

2

Kirkegaard, E. O. W. (2024, September 24). Overwhelming evidence of recent evolution in West Eurasians. Aporia. https://www.aporiamagazine.com/p/overwhelming-evidence-of-recent-evolution

3

Goodman, J. R., & Sear, R. (2026, November 13). “Race realism”: why scientifically flawed approach fails to present a realistic view of race. The Conversation. https://theconversation.com/race-realism-why-scientifically-flawed-approach-fails-to-present-a-realistic-view-of-race-290392. Republished as “’Race realism’ isn’t realistic, and it isn’t good science,” Ars Technica, November 21, 2026.

4

Lewontin, R. C. (1972). The apportionment of human diversity. Evolutionary Biology, 6, 381–398. https://doi.org/10.1007/978-1-4684-9063-3_14

5

Novembre, J. (2022). The background and legacy of Lewontin’s apportionment of human genetic diversity. Philosophical Transactions of the Royal Society B, 377(1852), 20200406. https://doi.org/10.1098/rstb.2020.0406

6

Edwards, A. W. F. (2003). Human genetic diversity: Lewontin’s fallacy. BioEssays, 25(8), 798–801. https://doi.org/10.1002/bies.10315

7

Rosenberg, N. A., Pritchard, J. K., Weber, J. L., et al. (2002). Genetic structure of human populations. Science, 298(5602), 2381–2385. https://doi.org/10.1126/science.1078311

8

Witherspoon, D. J., Wooding, S., Rogers, A. R., et al. (2007). Genetic similarities within and between human populations. Genetics, 176(1), 351–359. https://doi.org/10.1534/genetics.106.067355

9

Biddanda, A., Rice, D. P., & Novembre, J. (2020). A variant-centric perspective on geographic patterns of human allele frequency variation. eLife, 9, e60107. https://doi.org/10.7554/eLife.60107

10

Parra, F. C., Amado, R. C., Lambertucci, J. R., Rocha, J., Antunes, C. M., & Pena, S. D. J. (2003). Color and genomic ancestry in Brazilians. PNAS, 100(1), 177–182. https://doi.org/10.1073/pnas.0126614100

11

National Academies of Sciences, Engineering, and Medicine. (2023). Using Population Descriptors in Genetics and Genomics Research: A New Framework for an Evolving Field. National Academies Press. https://doi.org/10.17226/26902

12

UNESCO. (1950). The Race Question. https://unesdoc.unesco.org/ark:/48223/pf0000128291

13

Lazaridis, I., Patterson, N., Mittnik, A., et al. (2014). Ancient human genomes suggest three ancestral populations for present-day Europeans. Nature, 513, 409–413. https://doi.org/10.1038/nature13673

14

Haak, W., Lazaridis, I., Patterson, N., et al. (2015). Massive migration from the steppe was a source for Indo-European languages in Europe. Nature, 522, 207–211. https://doi.org/10.1038/nature14317

15

Lazaridis, I., Patterson, N., Anthony, D., et al. (2025). The genetic origin of the Indo-Europeans. Nature, 639, 132–142. https://doi.org/10.1038/s41586-024-08531-5

16

Olalde, I., Brace, S., Allentoft, M. E., et al. (2018). The Beaker phenomenon and the genomic transformation of northwest Europe. Nature, 555, 190–196. https://doi.org/10.1038/nature25738

17

Patterson, N., Isakov, M., Booth, T., et al. (2022). Large-scale migration into Britain during the Middle to Late Bronze Age. Nature, 601, 588–594. https://doi.org/10.1038/s41586-021-04287-4

18

Skoglund, P., Thompson, J. C., Prendergast, M. E., et al. (2017). Reconstructing prehistoric African population structure. Cell, 171(1), 59–71. https://doi.org/10.1016/j.cell.2017.08.049

19

Chen, L., Wolf, A. B., Fu, W., Li, L., & Akey, J. M. (2020). Identifying and interpreting apparent Neanderthal ancestry in African individuals. Cell, 180(4), 677–687. https://doi.org/10.1016/j.cell.2020.01.012

20

Mathieson, I., Lazaridis, I., Rohland, N., et al. (2015). Genome-wide patterns of selection in 230 ancient Eurasians. Nature, 528, 499–503. https://doi.org/10.1038/nature16152

21

Perretti, S., Santos, P., Vizzari, M. T., et al. (2025). Inference of human pigmentation from ancient DNA by genotype likelihoods. PNAS, 122(29), e2502158122. https://doi.org/10.1073/pnas.2502158122

22

Crawford, N. G., Kelly, D. E., Hansen, M. E. B., et al. (2017). Loci associated with skin pigmentation identified in African populations. Science, 358(6365), eaan8433. https://doi.org/10.1126/science.aan8433

23

Tishkoff, S. A., Reed, F. A., Ranciaro, A., et al. (2007). Convergent adaptation of human lactase persistence in Africa and Europe. Nature Genetics, 39(1), 31–40. https://doi.org/10.1038/ng1946

24

Evershed, R. P., Davey Smith, G., Roffet-Salque, M., et al. (2022). Dairying, diseases and the evolution of lactase persistence in Europe. Nature, 608, 336–345. https://doi.org/10.1038/s41586-022-05010-7

25

Okbay, A., Wu, Y., Wang, N., et al. (2022). Polygenic prediction of educational attainment within and between families from genome-wide association analyses in 3 million individuals. Nature Genetics, 54, 437–449. https://doi.org/10.1038/s41588-022-01016-z

26

Ding, Y., Hou, K., Xu, Z., et al. (2023). Polygenic scoring accuracy varies across the genetic ancestry continuum. Nature, 618, 774–781. https://doi.org/10.1038/s41586-023-06079-4

27

Howe, L. J., Nivard, M. G., Morris, T. T., et al. (2022). Within-sibship genome-wide association analyses decrease bias in estimates of direct genetic effects. Nature Genetics, 54, 581–592. https://doi.org/10.1038/s41588-022-01062-7

28

Kong, A., Thorleifsson, G., Frigge, M. L., et al. (2018). The nature of nurture: Effects of parental genotypes. Science, 359(6374), 424–428. https://doi.org/10.1126/science.aan6877

29

Tan, T., Jayashankar, H., Guan, J., et al. (2024, revised 2025). Family-GWAS reveals effects of environment and mating on genetic associations. medRxiv (preprint). https://doi.org/10.1101/2024.10.01.24314703

30

Bird, K. A. (2021). No support for the hereditarian hypothesis of the Black–White achievement gap using polygenic scores and tests for divergent selection. American Journal of Physical Anthropology, 175(2), 465–476. https://doi.org/10.1002/ajpa.24216

31

ABCD Study. Frequently asked questions [statement on misuse of participant data]. https://abcdstudy.org/faq

32

Harpak, A., & Przeworski, M. (2021). The evolution of group differences in changing environments. PLoS Biology, 19(1), e3001072. https://doi.org/10.1371/journal.pbio.3001072

33

Berg, J. J., Harpak, A., Sinnott-Armstrong, N., et al. (2019). Reduced signal for polygenic adaptation of height in UK Biobank. eLife, 8, e39725. https://doi.org/10.7554/eLife.39725

34

Sohail, M., Maier, R. M., Ganna, A., et al. (2019). Polygenic adaptation on height is overestimated due to uncorrected stratification in genome-wide association studies. eLife, 8, e39702. https://doi.org/10.7554/eLife.39702

35

Bratsberg, B., & Rogeberg, O. (2018). Flynn effect and its reversal are both environmentally caused. PNAS, 115(26), 6674–6678. https://doi.org/10.1073/pnas.1718793115

36

Dickens, W. T., & Flynn, J. R. (2006). Black Americans reduce the racial IQ gap: Evidence from standardization samples. Psychological Science, 17(10), 913–920. https://doi.org/10.1111/j.1467-9280.2006.01802.x

37

Lipson, M., Sawchuk, E. A., Thompson, J. C., et al. (2022). Ancient DNA and deep population structure in sub-Saharan African foragers. Nature, 603, 290–296. https://doi.org/10.1038/s41586-022-04430-9

38

Piffer, D., & Kirkegaard, E. O. W. (2024). Evolutionary trends of polygenic scores in European populations from the Paleolithic to modern times. Twin Research and Human Genetics, 27(1), 30–49. https://doi.org/10.1017/thg.2024.8

39

Trump, D. J. (2026, January 21). Remarks and a question-and-answer session at the World Economic Forum in Davos, Switzerland. Daily Compilation of Presidential Documents. https://www.govinfo.gov/content/pkg/DCPD-202600046/html/DCPD-202600046.htm

40

Mullin v. Doe (U.S. Supreme Court, June 25, 2026), allowing termination of Temporary Protected Status for Haiti and Syria. Reported by SCOTUSblog.

41

Sear, R. (2022). “National IQ” datasets do not provide accurate, unbiased or comparable measures of cognitive ability worldwide. PsyArXiv (preprint). https://doi.org/10.31234/osf.io/26vfb

42

Wicherts, J. M., Dolan, C. V., & van der Maas, H. L. J. (2010). A systematic literature review of the average IQ of sub-Saharan Africans. Intelligence, 38(1), 1–20. https://doi.org/10.1016/j.intell.2009.05.002

43

Retraction Watch. (2025, November 25). Meet the researcher aiming to halt use of “fundamentally flawed” database linking IQ and nationality. https://retractionwatch.com/2025/11/25/meet-the-researcher-aiming-to-halt-use-of-fundamentally-flawed-database-linking-iq-and-nationality

44

HOPE not hate. (2024). Race Science Inc. https://investigations.hopenothate.org.uk/race-science-inc/

45

Bird, K. A., & Carlson, J. (2024). Typological thinking in human genomics research contributes to the production and prominence of scientific racism. Frontiers in Genetics, 15, 1345631. https://doi.org/10.3389/fgene.2024.1345631

46

Carlson, J., Henn, B. M., Al-Hindi, D. R., & Ramachandran, S. (2022). Counter the weaponization of genetics research by extremists. Nature, 610, 444–447. https://doi.org/10.1038/d41586-022-03252-z

47

Haier, R. J. (2020). Academic freedom and social responsibility: Finding a balance. Intelligence, 82, 101482. https://doi.org/10.1016/j.intell.2020.101482

48

Winston, A. S. (2020). Why mainstream research will not end scientific racism in psychology. Theory & Psychology, 30(3), 425–430. https://doi.org/10.1177/0959354320925176

49

Bird, K. A., Jackson, J. P., Jr., & Winston, A. S. (2024). Confronting scientific racism in psychology: Lessons from evolutionary biology and genetics. American Psychologist, 79(4), 497–508. https://doi.org/10.1037/amp0001228

50

Cofnas, N. (2024, February 5). A guide for the hereditarian revolution. Nathan Cofnas’s Newsletter (Substack). https://ncofnas.com/p/a-guide-for-the-hereditarian-revolution

Discussion about this episode

User's avatar

Ready for more?