One of the largest genetic investigations of cognition ever conducted in young people has found that while boys and girls differ modestly on a range of cognitive tests, those differences are mirrored by real differences in how strongly genes shape performance in each sex. The study, published in the journal Biology of Sex Differences, analyzed data from nearly 4,700 individuals aged 8 to 21 drawn from the Philadelphia Neurodevelopmental Cohort, and applied a statistical genetics framework capable of asking two distinct questions at once: whether the same genes influence cognition in females and males, and whether those genes exert equal force in both sexes. The answers, the researchers report, are “yes” and “no,” respectively, a combination that has rarely been demonstrated at this scale.
The Philadelphia Neurodevelopmental Cohort is a population-based sample from the greater Philadelphia area, originally comprising more than 9,400 young people. For the new analyses, the team restricted the sample to 4,694 white non-Hispanic participants with both genetic and cognitive data, a step taken to control for population stratification, a known source of false positives in genetic studies. The participants had a mean age of 13.8 years and were almost evenly split by sex. Each had completed the Penn Computerized Neurocognitive Battery, a 14-test instrument designed to probe five domains of cognition: executive function, encompassing abstraction and mental flexibility, attention, and working memory; episodic memory, covering verbal, facial, and spatial recall; complex cognition, including verbal reasoning, nonverbal reasoning, and spatial processing; social cognition, measured through tasks of emotion identification, emotion differentiation, and age differentiation; and speed, assessed through motor and sensorimotor reaction-time tasks. Both accuracy and response time were recorded for every test.
To make the data tractable for genetic analysis, the researchers compressed the battery into composite domain scores using confirmatory factor analysis, and derived overarching measures of general accuracy, commonly denoted “g,” and general speed, denoted “gs,” through principal component analysis. Missing cognitive data were handled with multiple imputation by chained equations, a standard statistical technique that fills gaps using age, sex, and available test results. Genotyping was performed on Illumina arrays and imputed against the 1000 Genomes reference panel at the Broad Institute, and the team estimated empirical genetic relatedness among all participants from roughly 50,000 common single-nucleotide polymorphisms. Including related individuals, the authors note, substantially boosts power to detect heritability and, critically, to detect Gene-by-Sex interactions.
The statistical machinery at the heart of the study is an extension of the classical polygenic model implemented in the SOLAR software package. Under a univariate model, the phenotypic variance of a trait is decomposed into additive genetic variance and environmental variance, and narrow-sense heritability is the ratio of the former to the total. The researchers then extended this model in two directions. The first is a quantitative Gene-by-Sex interaction: here, the additive genetic variance is allowed to differ between females and males, and the likelihood of that model is compared with one in which the genetic variances are constrained to be equal. A significant result indicates that genes matter more in one sex than the other, even if the same genes are involved. The second is a qualitative interaction, tested through the genetic correlation between the sexes. If this correlation falls below 1, it implies that different genetic variants influence the trait in females and males; if it equals 1, the genes are the same. All models adjusted for age, age squared, sex, and their interactions, and false discovery rate was controlled at 5 percent.
The behavioral findings, first, are strikingly consistent with decades of prior work. Differences between the sexes were small or negligible on nearly every measure, with standardized mean differences ranging from 0.061 to 0.182, all below the conventional threshold of 0.2 standard deviations that defines a “small” effect. Females showed higher accuracy on memory and social cognition tasks, and faster performance on complex and social cognition speed measures. Males showed higher accuracy on executive and complex cognition, and faster executive and sensorimotor speed. On the omnibus measures, there was no significant difference in general cognitive accuracy, but females were modestly faster on general speed. Notably, the largest single difference, in sensorimotor speed favoring males, remained small at 0.182 standard deviations, underscoring how little overlap-free separation there truly is between the sexes on these dimensions.
The genetic results, however, tell a more nuanced story. All cognitive accuracy measures were moderately heritable, with estimates ranging from 0.417 to 0.718, and all speed measures were heritable in the range of 0.363 to 0.471. But when the polygenic models were refit separately for each sex, the magnitude of genetic influence diverged in domain-specific ways. On executive function accuracy, males showed roughly twice the genetic variance of females, with estimates of 0.598 versus 0.301, and correspondingly lower environmental variance. On complex cognition accuracy the pattern repeated: genetic variance of 0.610 in males against 0.291 in females. Conversely, on complex cognition speed and social cognition speed, females carried the larger genetic burden, with genetic variance of 0.575 and 0.589 respectively, compared with 0.135 and 0.129 in males, whose performance on those speed measures was instead dominated by environmental variance exceeding 0.64. In short, the genes that matter appear to matter differently in different domains, and differently in each sex.
Yet the qualitative picture is one of remarkable unity. Genetic correlations between females and males were not significantly different from 1 on any of the cognitive measures. This suggests complete pleiotropy, the situation in which the same variants influence the trait in both sexes. The finding aligns with a broader literature showing that for most complex human traits, heritability does not differ by sex and genetic correlations between sexes hover at or near unity. It also echoes a recent sex-stratified genome-wide association study of brain anatomy that found largely consistent genetic influences across sexes. In the context of the new results, this means that sex differences in cognition are not a story of different genes at work, but of the same genetic architecture being amplified or muted by sex-specific biological and social contexts.
The authors offer several candidate mechanisms for why genetic variance might be larger in one sex than the other. One prominent hypothesis involves gene-environment correlation: individuals may select or evoke experiences that align with their genetic predispositions, and the degree to which this happens may itself be sex-dependent. Girls, for instance, may be more likely to elicit or seek out prosocial experiences that reinforce genetic tendencies in social cognition, amplifying genetic differences on that domain. Boys may do the same for activities that exercise spatial and executive skills. Hormonal factors are also on the table. Puberty brings substantial testosterone exposure, and prior work has linked pubertal testosterone to mental rotation performance in males, suggesting that sex hormones may magnify the expression of existing genetic influences during adolescence. Epigenetic mechanisms, in which sex-differential methylation or chromatin modification alters gene expression, provide a further layer of plausibility.
The study connects its findings to psychiatric epidemiology in a way that gives the results added clinical weight. Social cognitive deficits are documented features of schizophrenia, depression, and bipolar disorder, and the prevalence of these conditions differs markedly between females and males. If the genetic architecture of social cognition is more strongly expressed in females, as the new data suggest, then models linking social cognition to psychiatric risk may need to be parameterized by sex. Similarly, understanding why genetic influences on executive and complex cognition are more strongly expressed in males could illuminate why neurodevelopmental conditions with executive dysfunction show male-skewed prevalence. The authors caution, however, that their findings are correlational and population-level, and should not be read as deterministic predictions about any individual.
Limitations temper the conclusions. The sample was restricted to individuals of European ancestry, and the authors explicitly call for replication in African, Asian, and Hispanic populations, noting that their cohort included too few participants of other ancestries for stratified analysis. The cohort spans ages 8 to 21 but the models did not account for puberty staging, which given the hormonal hypotheses above is a meaningful gap. And while the sample is large by the standards of behavior genetics, the statistical power to detect genetic correlations below 1 is famously demanding: the authors calculate that even with 80 percent power, detecting a genetic correlation of 0.7 given a heritability of 70 percent requires a sample of this approximate size, while detecting a correlation of 0.9 would require more than 30,000 twins in a classical twin design. Smaller-than-perfect genetic correlations may therefore exist but evade detection.
What the study makes unambiguous is that the social-versus-biological framing that has long dominated discussion of cognitive sex differences is inadequate. The effect sizes for phenotypic differences are small, and no finding here should be mistaken for evidence that one sex is cognitively superior to the other. But the genetic variance findings, the authors argue, demonstrate that genetic factors are not a negligible or secondary consideration. Instead, genes, environment, hormones, and social context appear to operate jointly, with genetic influences being channeled, buffered, or amplified by sex-specific developmental pathways. Untangling that interplay, the researchers conclude, is the necessary next step for a field whose subjects, cognition and sex, sit at the intersection of psychology, neuroscience, genetics, and social science.
Subject of Research: Gene-by-Sex interactions in the genetic architecture of cognitive abilities in children and adolescents
Subject of Research: Medicine
Article Title: Gene × sex interactions on cognition in the Philadelphia neurodevelopmental cohort
Article References: Mollon, J., Knowles, E. E. M., Mathias, S. R., Rodrigue, A., Gur, R. C., Peralta, J. M., Weiner, D. J., Robinson, E. B., Raznahan, A., Gur, R. E., Blangero, J., Almasy, L., & Glahn, D. C. (2026). Gene × sex interactions on cognition in the Philadelphia neurodevelopmental cohort. Biology of Sex Differences, 17(1), Article 116. https://doi.org/10.1186/s13293-026-00929-2
Image Credits: AI Generated
DOI: 10.1186/s13293-026-00929-2
Keywords: cognitive sex differences, heritability, Gene-by-Sex interaction, Philadelphia Neurodevelopmental Cohort, executive function, social cognition, genetic correlation, polygenic model
Cite Scienmag News
APA
MLA
Chicago
Juliet Wilcox. (September 4, 2026). How genes and sex jointly shape cognition in Philadelphia youth cohort. Scienmag. https://scienmag.com/how-genes-and-sex-jointly-shape-cognition-in-philadelphia-youth-cohort/
Juliet Wilcox. “How genes and sex jointly shape cognition in Philadelphia youth cohort.” Scienmag, 4 September 2026, https://scienmag.com/how-genes-and-sex-jointly-shape-cognition-in-philadelphia-youth-cohort/. Accessed 4 September 2026.
Juliet Wilcox. “How genes and sex jointly shape cognition in Philadelphia youth cohort.” Scienmag. September 4, 2026. https://scienmag.com/how-genes-and-sex-jointly-shape-cognition-in-philadelphia-youth-cohort/
Copy citation
Download RIS
Tags: childhood and adolescent cognitive testingcontrolling for population stratification in genetic studiesdevelopmental genetics in neuropsychologygender differences in genetic contribution to brain functiongene-by-sex interactions in neurodevelopmentgene-sex interaction in neurodevelopmentGenetic influence on cognition in youthGenetic influences on cognition in youthimpact of genetics on cognitive test scoresimpact of genetics on youth cognitionlarge-scale genetic analysis in youthlarge-scale youth cognition geneticslong-term genetic influences on youth brain developmentneurodevelopmental geneticsPhiladelphia Neurodevelopmental Cohort studypopulation stratification in genetic studiespopulation-based genetic research in adolescentspopulation-based genetics researchrole of sex in gene expression and cognitionsex differences in cognitive performancesex-specific genetic effects on cognitionsex-specific genetic effects on intelligence
