Personality is often described as something deeply personal: the characteristic ways people tend to think, feel and behave. Yet decades of behavioural genetics have also suggested that some of the differences between people are associated with inherited biology. A large new study now provides one of the most detailed maps yet of the common genetic variation associated with the Big Five personality traits.
Published in Nature, the research brought together 46 cohorts and between 611,037 and 1,136,711 participants per trait. The researchers identified 1,260 approximately independent genetic variants associated with extraversion, agreeableness, conscientiousness, neuroticism or openness to experience. Of these lead variants, 824, or 65%, were located in regions not previously associated with variation in the corresponding personality trait.
The scale of the result is striking, but its interpretation requires care. The study does not show that personality is genetically fixed, nor does it identify a small set of genes that determines who someone becomes. Individual genetic effects were tiny, common variants explained only a minority of measured personality differences, and the researchers explicitly emphasised the substantial role of environment.
A much larger map of personality genetics
The study was conducted by the Revived Genomics of Personality Consortium. Its central analysis examined the Big Five, a widely used framework that organises personality into five broad dimensions. Extraversion captures tendencies such as sociability and assertiveness. Agreeableness includes compassion and trust. Conscientiousness concerns organisation and responsibility. Neuroticism reflects tendencies toward anxiety, depression and emotional volatility, while openness to experience includes curiosity, imagination and aesthetic sensitivity.
Rather than analysing one sample, the researchers meta-analysed genome-wide association studies across dozens of cohorts. Sample sizes differed by trait: 662,617 participants contributed to the extraversion analysis, 611,037 to agreeableness, 641,167 to conscientiousness, 1,136,711 to neuroticism and 611,985 to openness.
Across roughly 10 million single-nucleotide polymorphisms, or SNPs, per trait, the team searched for variants whose statistical association with personality passed the conventional genome-wide significance threshold. This produced 1,260 lead SNPs. The expansion over previous work was substantial. The number of significant loci rose from 3 to 131 for conscientiousness, from 4 to 39 for agreeableness, from 8 to 126 for openness, from 14 to 258 for extraversion and from 224 to 706 for neuroticism.
Most of these signals were trait-specific. Eighty-two percent of genomic regions containing a genome-wide significant variant were associated with only one of the Big Five traits. The average absolute genetic correlation among the five traits was 0.19, supporting the idea that the dimensions overlap to some degree without collapsing into one underlying genetic tendency.
The effects were widespread but individually tiny
A list of 1,260 variants can sound as if researchers have found a genetic blueprint for personality. The effect sizes show why that would be misleading.
For extraversion, for example, the median association among lead variants was only 0.009 standard deviations per effect allele after correction for statistical inflation. The researchers estimated that, on average, more than 16,000 independently associated common SNPs contribute to each trait. Personality genetics therefore appears highly polygenic: many variants each contribute very small statistical effects.
The proportion of personality variation associated with common SNPs was also moderate. In the main analyses, SNP heritability ranged from 4.8% for agreeableness to 9.3% for extraversion. When the researchers accounted for the reliability of typical personality measures, estimates were higher, at roughly 9.3% to 13.3%.
Those estimates remain well below the roughly 40% to 60% heritability often reported by twin and family studies, which can capture genetic influences not represented by common SNPs alone. The gap is scientifically informative. It indicates that common variants measured in genome-wide association studies capture only one part of the biological contribution to personality.
Just as importantly, heritability is not destiny. It describes variation within populations under particular environmental conditions. It does not mean that a given percentage of an individual’s personality is caused by genes, and it does not imply that personality cannot change.
Testing whether family background was creating false genetic signals
One of the study’s strongest features was its attempt to separate direct genetic associations from environmental and family processes that can make genetic results difficult to interpret.
Standard population-level genome-wide studies can be influenced by population structure, assortative mating and what researchers call dynastic effects, where parents’ genetic characteristics shape the environments they provide to children. To probe these possibilities, the consortium assembled within-family genome-wide association data from 12 cohorts, with between 31,544 and 50,725 participants depending on the personality trait.
Within-family comparisons are valuable because relatives share much of their family environment and ancestry. If a population-level genetic association largely disappears within families, that can indicate that the original association was partly produced by confounding rather than a direct biological pathway.
Here, population-level and within-family effects were highly similar. The researchers concluded that personality’s genetic associations were only minimally confounded by shared family environmental factors, population stratification, assortative mating and related processes. That contrasts with findings for traits such as educational attainment, income, cognitive performance and externalising behaviour, where within-family estimates can be noticeably smaller.
The result does not eliminate confounding. The authors explicitly noted that it was not entirely absent and that even small effects can matter. What it does provide is stronger evidence that many of the detected personality associations are not simply artefacts of family background.
Genetic patterns travelled across measures and groups, but not perfectly
The researchers also asked whether genetic associations remained similar when personality was measured in different ways or among different groups. Across European-like cohorts, splitting the data into independent halves produced a mean genetic correlation of 0.94, indicating highly consistent genetic effects.
Associations were also broadly similar across geography, age, personality instruments and whether people rated themselves or were rated by someone close to them. This matters because personality studies use many questionnaires and sampling strategies, creating a risk that genetic findings could be tied to a particular instrument rather than the broader trait.
The study also included participants with African-like genomes. Polygenic indices derived primarily from the much larger European-like discovery data significantly predicted extraversion, neuroticism and openness in African-like samples, with meta-analytic standardised coefficients of 0.099, 0.071 and 0.047 respectively. The researchers described this as the first significant polygenic prediction of personality beyond participants with European-like genomes.
However, prediction generally weakens when discovery and target populations differ in genetic ancestry. The imbalance in available genomic data therefore remains an important limitation rather than a solved problem.
Personality genetics overlapped with health and behaviour
The consortium went beyond locating variants. It examined genetic correlations between personality and 67 behaviours and outcomes across nine domains, including physical and mental health, daily behaviour and social outcomes.
Some of the most intriguing associations involved research participation itself. Genetic variants associated with openness and agreeableness were also associated with completing optional questionnaires, with average genetic correlations of 0.22 and 0.11. Neuroticism showed a genetic correlation of minus 0.23 with the number of questionnaire responses, while it correlated positively, at an average of 0.32, with selecting responses such as “I don’t know” or “prefer no response” after beginning a questionnaire.
This raises a methodological issue that extends beyond personality research. If traits partly associated with participation also influence who responds to surveys, large observational datasets can become systematically selective. The genetics of personality may therefore contribute to who enters, stays in and responds fully to research studies.
The researchers also applied Mendelian randomisation to a selected set of outcomes where the method’s assumptions were considered more plausible. Thirty-three exposure-outcome pairings produced directionally consistent evidence across estimators, with 24 analyses pointing from personality toward later outcomes. Examples included associations suggesting that greater conscientiousness reduced body mass index and smoking initiation, while neuroticism was associated with poorer scores on a healthy ageing and longevity factor.
These analyses are more informative about possible direction than ordinary correlations, but they are not equivalent to randomised experiments. Mendelian randomisation depends on assumptions about how genetic instruments affect outcomes, and violations can produce misleading causal interpretations. The findings should therefore be treated as evidence consistent with possible causal pathways rather than proof that changing a personality trait would necessarily change a health outcome.
What this study changes
The study’s main contribution is not a claim that genes explain personality. It is a much sharper estimate of how common genetic variation relates to personality, supported by unusually large samples, cross-cohort replication and within-family analyses.
Three conclusions stand out. First, the genetic architecture is highly distributed: there is no single “extraversion gene” or “conscientiousness gene”. Second, common variants account for a meaningful but limited share of differences between people. Third, many associations survive stronger tests for family-level confounding, making them more robust than might be expected for complex behavioural traits.
For psychology, that strengthens the case for treating personality as a genuinely biopsychosocial phenomenon. Biological differences matter, but they operate probabilistically and alongside development, relationships, culture, opportunity and experience. For genetics, the work also illustrates why enormous samples are needed to detect effects that are individually minuscule.
Important limitations remain
The sheer sample size should not obscure the study’s boundaries. Most statistical power still came from participants with European-like genomes. Although African-like cohorts were included and provided an important test of portability, genomic research remains globally unequal. Findings and prediction tools cannot automatically be assumed to transfer with equal accuracy to all populations.
Personality measurement also introduces noise. Different cohorts used different instruments, and self-report measures can be influenced by response style, context and interpretation. The researchers found that more reliable measures yielded higher heritability estimates, showing that measurement quality directly affects what genomic analyses can detect.
Finally, genome-wide association identifies statistical relationships, not simple biological mechanisms. A significant SNP can sit near multiple genes, affect regulation rather than protein structure, or tag a broader region through linkage disequilibrium. Translating these associations into specific cellular pathways will require further functional research.
The broader message is therefore more nuanced than genetic determinism. Personality differences have detectable genetic architecture, but that architecture is diffuse, probabilistic and incomplete. Even in a study exceeding one million participants, the environment remains central to explaining why people differ.
Source Information
Study: “Robust inference and correlates from genetic associations with personality”
Authors: Ted Schwaba, M. L. Clapp Sullivan, W. A. Akingbuwa and colleagues
Journal: Nature
Published: 2 September 2026
DOI: 10.1038/s41586-026-10992-9
Study type: Genome-wide association meta-analysis across 46 cohorts, supplemented by within-family analyses, polygenic prediction, genetic correlation and Mendelian randomisation
Sample: 611,037 to 1,136,711 participants per Big Five trait in the primary analyses, with up to 50,725 participants in within-family GWAS
