Heritability is a statistical estimate of how much of the variation in a characteristic within a particular population can be associated with genetic differences among individuals. Researchers apply the concept to physical traits such as height and body mass, as well as psychological and behavioral characteristics including cognitive ability, personality, substance use, and vulnerability to certain mental disorders. A heritability estimate does not describe the origins of a trait within one person. It describes why people in a studied population differ from one another under the environmental conditions that existed when the data were collected. Peter Visscher, William Hill, and Naomi Wray emphasized this point in their influential 2008 article “Heritability in the Genomics Era—Concepts and Misconceptions,” describing heritability as a population parameter rather than a measure of genetic determination.
Suppose researchers estimate that a trait has a heritability of 60 percent. This does not mean that 60 percent of an individual’s trait came from genes while the remaining 40 percent came from the environment. A person cannot be separated into genetic and environmental percentages. The estimate means that, within the population and circumstances examined, approximately 60 percent of the observed variation was statistically attributable to genetic variation. Change the population, age range, social conditions, available resources, or measurement method, and the estimate may change. Heritability is therefore specific to a population and historical setting. It is not a permanent property attached to a characteristic.
How Researchers Estimate Heritability
Traditional estimates frequently come from comparisons among relatives. Identical twins share nearly all their inherited DNA sequence, while fraternal twins share, on average, about half of the genetic variants that differ among individuals. If identical twins resemble one another more closely on a trait than fraternal twins do, that additional similarity can be used to estimate genetic influence. Adoption studies compare biological relatives who were raised separately or genetically unrelated individuals who grew up in the same household. Extended-family designs may include siblings, parents, cousins, half-siblings, and spouses, allowing researchers to model several forms of biological and environmental resemblance.
Classical models often divide variation into additive genetic effects, shared environmental effects, and nonshared environmental effects. Additive effects occur when genetic variants make cumulative contributions to a trait. Shared environmental effects include experiences that make people raised together more alike, whereas nonshared environmental effects include experiences that contribute to differences between them. These categories are useful, but they are statistical components rather than simple lists of specific causes. The nonshared category also includes measurement error. Twin models additionally depend on assumptions about mating patterns, environmental similarity, genetic interactions, and the extent to which twins represent the wider population. Heritability estimates should consequently be interpreted as model-based findings with confidence intervals, not exact measurements of nature’s contribution.
What Twin Studies Have Found
One of the broadest examinations of human heritability was published by Tinca Polderman and colleagues in Nature Genetics in 2015. The researchers conducted a meta-analysis covering 17,804 traits reported across 2,748 publications and nearly 14.6 million partly dependent twin pairs. Across all included traits, the reported average heritability was 49 percent. Estimates clustered by functional domain, meaning that different categories of traits showed distinct patterns rather than one universal level of genetic influence. For 69 percent of the traits, the observed twin resemblance was consistent with a model based primarily on additive genetic variation.
This result is sometimes summarized by saying that human characteristics are “about half genetic,” but that wording hides important variation. Some physical traits and neurodevelopmental conditions produce relatively high estimates, while attitudes, behaviors, and health outcomes may show lower or more context-dependent values. Even traits with similar estimates may have completely different biological structures. One may be influenced by thousands of common genetic variants, while another may involve rare variants, developmental processes, indirect family effects, or interactions with environmental exposure. A heritability estimate does not identify the genes involved, explain how they operate, or establish that genetic differences directly cause every observed association.
Narrow-Sense and Broad-Sense Heritability
Geneticists distinguish between narrow-sense and broad-sense heritability. Narrow-sense heritability refers to the proportion of variation associated with additive genetic effects—the cumulative contributions of genetic variants inherited from parents. It is commonly represented as h² and is especially important in breeding, evolutionary theory, and modern genome-wide research because additive effects help predict resemblance across generations. Broad-sense heritability also includes nonadditive influences such as dominance, in which the effect of one allele depends on the other allele at the same location, and epistasis, in which variants at different locations interact.
The distinction matters because two studies may use the word heritability while estimating different components of genetic variation. In a 2017 analysis of SNP-based methods, Jian Yang and colleagues defined narrow-sense heritability as the proportion of phenotypic variance attributable to additive effects generated by all causal variants. They also explained how researchers can estimate the variation captured by measured single-nucleotide polymorphisms, or SNPs, among people who are not closely related. These molecular methods provide evidence based directly on DNA but measure only the variants represented by the genomic data and statistical model.
Heritability Can Change Across Development
Heritability estimates are not necessarily stable throughout life. A large international twin study led by Claire Haworth found that the heritability of general cognitive ability increased from childhood into young adulthood. This does not suggest that environments become unimportant as children grow. One interpretation is that people increasingly select, modify, and evoke environments that correspond to their genetically influenced interests and tendencies. A child who enjoys reading may seek books, advanced classes, and similarly interested friends, gradually amplifying an initial disposition. Genetic influence can therefore operate partly through the environments people help create.
Environmental conditions can also change the amount of variation attributed to genetic differences. Eric Turkheimer and colleagues reported in 2003 that socioeconomic status moderated estimates of cognitive ability in a sample of young American twins. Their results suggested stronger shared environmental effects and weaker genetic effects under disadvantaged conditions, although later studies have not found the same pattern consistently across countries and samples. The broader lesson is not that one environment universally increases or decreases heritability. It is that genetic variation may be expressed differently depending on nutrition, education, stress, safety, opportunity, and other conditions.
Heritable Does Not Mean Unchangeable
A characteristic can be highly heritable and still respond strongly to environmental intervention. Height is a familiar example: genetic differences explain much of the variation within many well-nourished populations, yet average height can change across generations as nutrition, disease exposure, and living conditions improve. Wearing corrective lenses also provides a simple analogy. Differences in unaided vision may be substantially influenced by genetics, but eyeglasses can alter the practical outcome. Heritability concerns the sources of variation under current conditions; it does not establish limits on treatment, education, prevention, or social change.
The reverse is also true. Low heritability does not imply that a trait is easy to change. A population may experience a powerful environmental condition that affects nearly everyone similarly, producing little variation for researchers to associate with that condition. Phenylketonuria illustrates the distinction between genetic origin and modifiable outcome. The disorder results from genetic variants that impair phenylalanine metabolism, yet early detection and dietary management can greatly reduce the risk of intellectual disability. Genetic causation and environmental preventability are not opposites. Biology often identifies the pathway through which an intervention can work.
Molecular Genetics and Missing Heritability
Genome-wide association studies have allowed scientists to estimate heritability using measured DNA rather than relying only on family resemblance. These studies often reveal that complex traits are highly polygenic: many variants each make very small statistical contributions. Early genome-wide studies identified associated variants that collectively explained much less variation than twin and family estimates predicted. Researchers called this gap “missing heritability.” Possible explanations included rare variants not captured by standard genotyping, structural changes in DNA, imperfect statistical models, gene interactions, and inflation or confounding in family-based estimates.
Whole-genome sequencing is beginning to close parts of that gap. In a 2025 study of 347,630 unrelated UK Biobank participants of European ancestry, Pierrick Wainschtein and colleagues estimated genetic contributions across 34 complex traits and diseases using common and rare coding and noncoding variants. The measured whole-genome variants accounted for approximately 88 percent of the pedigree-based heritability on average, with noncoding variants contributing most of the estimated genomic heritability. The findings suggest that much apparently missing heritability may be distributed across enormous numbers of variants, including rare variants that earlier technologies measured poorly. The study’s ancestry and sample limitations remain important, however, and its results should not automatically be generalized to every population or trait.
Interpreting Heritability Responsibly
Heritability is valuable because it helps researchers understand variation, design genetic studies, investigate disease risk, and identify biological pathways. It cannot tell us whether a trait is morally desirable, whether social inequality is justified, or what opportunities a person should receive. It also does not reveal an individual’s future. Even when a trait is substantially heritable, genetic information generally produces probabilities rather than certainties. Individual development remains shaped by family relationships, education, culture, health, social institutions, personal decisions, and unpredictable experiences.
Responsible interpretation requires avoiding both genetic determinism and the assumption that genes do not matter. Human differences arise through continuing interaction among genetic variation, biological development, and environmental experience. Heritability quantifies one part of that relationship within a defined population; it does not divide nature from nurture or declare one more important in every circumstance. Properly understood, the concept shows why the familiar nature-versus-nurture debate is misleading. Genes influence how people respond to environments, environments influence how genetic tendencies are expressed, and both operate together across the lifespan.
