What do modern genomic studies say about selection on cognitive ability across populations?
Executive summary
Modern genomic work finds that cognitive traits are highly heritable and polygenic, that many loci and structural variants contribute to cognitive variation, and that some genomic regions linked to brain and language functions show signs of past selection—but the evidence for recent, directional selection on “intelligence” across human populations is tentative and confounded by technical and social factors [1] [2] [3] [4]. Robust conclusions about population differences require much more diverse data and careful controls for stratification, assortative mating, and “genetic nurture” [5] [6] [7].
1. Genetic architecture: highly heritable, strongly polygenic
Twin, family and molecular studies converge on substantial heritability of cognitive abilities—roughly half or more of the variance in modern industrialized samples—while GWAS and sequence studies show cognition is influenced by hundreds to thousands of common variants plus rare variants and copy-number changes with individually small effects [1] [8] [2] [3].
2. Signals compatible with selection, but not a simple story of directional change across groups
Population-genomic scans detect enrichment of selection signals in genes annotated for nervous-system functions and cognition across species, and some human studies infer shifts in allele frequencies over millennia consistent with selection on cognitive-related variants during the Holocene—but these signals reflect polygenic, subtle shifts rather than single “intelligence” genes and are sensitive to the models used [4] [9].
3. Ancient DNA and polygenic score analyses: suggestive temporal trends, strong caveats
Comparisons of ancient and modern genomes have been interpreted as increases in polygenic scores for educational attainment or general cognitive ability over parts of the Holocene—supporting hypotheses of culture–gene coevolution during urbanization—but those studies rely on limited SNP panels, uneven ancient sampling, and modern-derived weights, so inferred selection magnitudes and directions remain debated [9] [4].
4. Functional and domain-specific selection: language as a clearer example
Some rapidly evolved genomic regions tied to transcription-factor binding—including sequences affecting FOXP2 motifs—show associations with language abilities and population patterns consistent with long-term selective pressures; these results point to domain-specific evolutionary changes (language vs. general cognition) and to balancing or complex selection rather than simple directional sweeps [10] [11].
5. Methodological limits and social confounders that muddy cross‑population inferences
Polygenic scores predict cognition best in ancestrally matched groups, and remaining population stratification, assortative mating, and genetic nurture (parents’ genotypes shaping environments) can mimic or inflate signals interpreted as selection or population differences; many large GWAS remain Eurocentric, limiting claims about global population-level selection [5] [6] [7].
6. The balanced bottom line: selection shaped cognitive-related biology, but claims about recent population differences are provisional
Modern genomics shows cognition is shaped by many genetic variants, some with signs of evolutionary pressure, and that selection has plausibly acted on brain- and language-related loci over deep and recent timeframes; however, assertions that contemporary group differences stem from recent directional selection lack conclusive genomic proof given technical confounding, sparse ancient sampling, and limited multi-ancestry discovery efforts—future multi‑ancestry GWAS, whole-genome sequencing, and within‑family designs are required to move from plausible hypotheses to robust, population-specific inferences [8] [12] [7] [6].