Behavioral Genetics

Abstract

Behavioral genetics is the branch of the behavioral sciences that measures how far genetic differences between people explain the psychological differences among them. Using twins, adoptees, and now the whole genome, it partitions the variation in a trait into genetic and environmental components and asks not whether nature or nurture matters but how much each contributes and through what pathways. This article treats the field as a behavioral science: it explains the twin and adoption designs, the concept of heritability and its three laws, the shift from failed candidate genes to genome-wide association studies and polygenic scores, and the interplay of genes and environment. It closes with the postgenomic reframing of a century-old debate. Three interactive demonstrations let the reader estimate heritability, decompose variance, and build a polygenic score.

Keywords: behavioral genetics, heritability, twin study, polygenic score

Behavioral genetics is the study of how inherited differences in DNA contribute to differences in behavior, from cognitive ability and personality to psychiatric risk. It does not ask whether a trait is genetic or environmental, a question the field regards as ill-posed, but how the variation in a trait across a population divides between the two, and how genes and environments act together over development. Its historical root is the recognition, associated with Francis Galton in the nineteenth century, that psychological characteristics run in families and that the resemblance of relatives can be turned into a measurement of inheritance. What distinguishes the field from a vague appeal to heredity is its quantitative method: by comparing people of known genetic relatedness, or by reading the genome directly, it estimates the proportion of trait variation attributable to genetic differences and forces the nature-nurture debate into testable form (Plomin et al., 2016).

What Behavioral Genetics Is

Behavioral genetics is best understood as a science of variation. Its object is not the trait of a single person but the spread of a trait across a population, and its central question is how much of that spread is explained by the genetic differences among the people in it. This framing matters because it dissolves a confusion that dogs popular discussion. To say that height or verbal ability is heritable is not to say that any individual's height or ability is caused by genes rather than by food and schooling; it is to say that, in this population and under these conditions, people who differ in the trait tend also to differ in their genes. The field grew from Galton's insight that the resemblance of relatives is data, and it matured by making that data precise. Two research strategies define it. The quantitative-genetic strategy compares relatives of known genetic relatedness, chiefly twins and adoptees, and infers the genetic contribution from patterns of resemblance without measuring any gene. The molecular-genetic strategy reads the genome itself, testing millions of DNA variants for association with a trait. The two strategies converge on the same conclusion, reached first by the older method and confirmed by the newer: individual differences in behavior are, in part, differences in inheritance (Plomin et al., 2016).

Types of Behavioral Genetics

Genetics, Behavioral is a formal descriptor in the National Library of Medicine's Medical Subject Headings, which files it beneath both the behavioral sciences at tree position F04.096.276 and genetics proper. Beneath the descriptor MeSH hangs the single narrower heading listed in Table 1. Two cautions apply. The list is an indexing classification built to organize the biomedical literature, not a theory that carves the field at its joints, and its one member is a position about heredity rather than a method or subdiscipline of it: genetic determinism is a claim behavioral geneticists overwhelmingly reject, filed here for retrieval, not endorsed as a branch of the science. Only subtypes that are themselves live articles on this site are linked, and at present this descriptor has no page of its own.

Twin and Adoption Designs

The classical tools of the field exploit natural experiments in relatedness. The twin design compares monozygotic twins, who develop from one fertilized egg and share essentially all of their DNA, with dizygotic twins, who develop from two eggs and share on average half of the genetic variants that segregate in a family, the same proportion as ordinary siblings. Because both kinds of twin are typically raised together, both share a family environment, so the extent to which identical twins resemble each other more than fraternal twins do is a signature of genetic influence (Boomsma et al., 2002). This inference rests on the equal environments assumption, the premise that identical and fraternal twins raised together experience equally similar trait-relevant environments, so that the greater similarity of identical pairs reflects their greater genetic similarity rather than more similar treatment; the assumption has been repeatedly tested, for instance in twins mistaken about their own zygosity, and has largely held for the traits behavioral genetics studies (Boomsma et al., 2002). The concordance of a pair, the degree to which they match on a trait, is the raw material: consistently higher monozygotic than dizygotic concordance across a trait indicates heritability. The adoption design attacks the same problem from the other side, separating the two things a biological family confounds. An adopted child shares genes but not environment with the birth parents and environment but not genes with the rearing parents, so resemblance to each set of parents isolates one influence from the other. The most dramatic version studies monozygotic twins reared apart, separated in infancy and raised in different homes; their similarity in adulthood cannot be attributed to a shared upbringing and so estimates heritability directly. The Minnesota Study of Twins Reared Apart found such twins to be substantially alike across a wide range of psychological measures, including general cognitive ability and personality, despite their separate rearing (Bouchard et al., 1990). The demonstration below lets the reader set the concordance of identical and fraternal pairs and watch how their difference translates into an estimate of genetic and environmental contribution.

Heritability and Its Laws

Heritability is the central and most misunderstood statistic of the field. It is defined as the proportion of the variance in a trait, within a particular population at a particular time, that is attributable to genetic differences among individuals. It is a fraction between zero and one, and it is a property of a population under conditions, not of a trait in the abstract or of any person. A heritability of 0.5 for a trait does not mean that half of any individual's trait is genetic; it means that, in this population, genetic differences account for about half of why people differ. Crucially, heritability depends on how much environments vary: equalize the environment and the remaining variation is more purely genetic, so heritability rises, while a population with wildly unequal environments will show lower heritability for the same biology. The additive genetic variance, the part that adds up predictably across the genes a child inherits, is the component twin studies estimate as narrow-sense heritability. After decades of such studies a set of empirical regularities became robust enough to be stated as laws. Turkheimer's three laws of behavior genetics summarized them: all human behavioral traits are heritable; the effect of being raised in the same family is smaller than the effect of the genes; and a substantial portion of the variation is explained by neither genes nor family environment (Turkheimer, 2000). A meta-analysis of virtually every twin study conducted over fifty years, covering more than seventeen thousand traits, confirmed the first law with a vengeance: the average heritability across all measured human traits was about forty-nine percent, and for most trait domains the pattern of resemblance was consistent with genes of simply additive effect (Polderman et al., 2015). A proposed fourth law added that each trait is influenced by very many variants of very small effect, anticipating the molecular findings described below (Chabris et al., 2015). Table 2 collects the four.

The variance a trait shows can be decomposed into three sources that the twin design is built to estimate. The additive genetic component, conventionally labeled A, captures the summed effect of individual variants. The shared environment, labeled C, is everything that makes members of the same family alike regardless of genes, such as household income or neighborhood. The nonshared environment, labeled E, is everything that makes family members differ despite their shared home and genes, and it absorbs measurement error as well. Turkheimer's second and third laws are, in this notation, the repeated finding that C is surprisingly small for most behavioral traits while E is surprisingly large. Figure 1 shows the three components as a single bar partitioning the whole of a trait's variance, using the values worked out below. The demonstration that follows lets the reader move the three components and see how any allocation of genetic, shared, and nonshared influence sums to that whole.

From Candidate Genes to Genome-Wide Association

Once twin studies had established that behavioral traits are heritable, the natural next step was to find the responsible genes, and the first attempts went badly. The candidate-gene era of the 1990s and 2000s selected a plausible gene, usually one in a neurotransmitter pathway, and tested whether one of its variants was associated with a trait or disorder. A famous example reported that a variant in the serotonin transporter gene predicted depression, but only in people who had also suffered stressful life events, a gene-by-environment interaction that drew enormous attention (Caspi et al., 2003). The trouble was that these findings largely failed to replicate. A systematic reexamination of the historically most-studied candidate genes for depression, using sample sizes orders of magnitude larger than the originals, found no support for any of them, whether acting alone or in interaction with stress (Border et al., 2019). The candidate-gene approach had been defeated by the very structure of the traits it studied: behavioral characteristics are not governed by one or a few genes of large effect but by thousands of variants each contributing almost nothing, exactly the fourth law's claim. The genome-wide association study, or GWAS, was the method that matched this architecture. Rather than guessing a gene, a GWAS scans millions of variants across the entire genome in very large samples, asking which are statistically associated with the trait, and it makes no prior assumption about biology. As sample sizes grew into the hundreds of thousands and beyond, GWAS began to yield robust, replicable associations for cognitive ability, educational attainment, and psychiatric risk, confirming the highly polygenic architecture and identifying specific loci (Visscher et al., 2017). The genetics of intelligence became one of the clearest cases, with heritability rising across development and hundreds of associated variants of tiny effect (Plomin & Deary, 2015). The many small effects can be summed for each person into a single polygenic score that predicts a fraction of the trait, and the demonstration below builds such a score from many loci to show how tiny individual contributions aggregate into a normal distribution with real, if modest, predictive power.

Gene-Environment Interplay

The clean partition of variance into genetic and environmental boxes is a useful fiction that the field itself has spent decades complicating. Two phenomena blur the boundary. Gene-environment correlation arises because people are not randomly assigned to environments: the same genetic propensities that shape a trait also shape the environments a person encounters, as when a child genetically inclined toward reading is given more books, seeks them out, and is taught by parents who share and cultivate the inclination. Here nature and nurture point the same way and reinforce each other, so that an apparently environmental advantage is partly a genetic effect in disguise. Gene-environment interaction, the second phenomenon, arises when the effect of the genes depends on the environment, or equivalently when the effect of an environment depends on the genes. A striking example concerns the heritability of cognitive ability itself: in young children from impoverished families the heritability of intelligence was found to be near zero, with the shared environment dominating, while in affluent families the same trait was highly heritable, as though deprivation suppresses the expression of genetic potential that adequate resources allow to emerge (Turkheimer et al., 2003). Interactions of this kind mean that a single heritability figure can conceal opposite realities in different subpopulations. Together, correlation and interaction show why behavioral genetics cannot license a genetic determinism: the two inheritances are entangled in the very developmental pathways that build a mind, and separating them statistically does not separate them causally. The connections to learning and to the social environments studied in social psychology are therefore intrinsic to the field rather than external to it.

Worked Example

The logic that turns twin resemblance into a heritability estimate can be made exact with the standard formulas of quantitative genetics. Represent the similarity of a twin pair as a correlation between the two members across many pairs, and write the monozygotic correlation as rMZ and the dizygotic correlation as rDZ. Identical twins share all of their additive genetic variance and their common environment, so their correlation reflects both: rMZ equals A plus C. Fraternal twins share half of their additive genetic variance and their common environment, so rDZ equals one half A plus C. Subtracting the second equation from the first gives rMZ minus rDZ equals one half A, so the additive genetic component is twice the difference between the two correlations: A equals two times the quantity rMZ minus rDZ. This is Falconer's formula, and the doubling is the heart of it. Take a concrete trait for which identical twins correlate at rMZ equals 0.85 and fraternal twins at rDZ equals 0.55. Then A equals two times the quantity 0.85 minus 0.55, which is two times 0.30, or 0.60: sixty percent of the variance is additive genetic, the narrow-sense heritability. The shared environment follows from rMZ equals A plus C, so C equals 0.85 minus 0.60, or 0.25. The nonshared environment is whatever the identical twins do not share, E equals one minus rMZ, or one minus 0.85, which is 0.15. The three components sum to 0.60 plus 0.25 plus 0.15, exactly 1.00, as a complete decomposition of the variance must. This single trait illustrates all three laws at once: the trait is highly heritable at 0.60, the shared family environment is the smallest term at 0.25, and a real nonshared residual of 0.15 remains that neither genes nor family explains. Change the two correlations in the demonstration above and the same arithmetic re-partitions the variance in front of the reader (Boomsma et al., 2002).

Discussion

Behavioral genetics occupies an uncomfortable but indispensable place in the science of mind. It has established, about as firmly as anything in psychology, that genetic differences contribute to nearly every measured difference in behavior, and it has done so twice over, first through the resemblance of relatives and again through the direct reading of the genome. Yet its central statistic is chronically misread, and the field spends much of its energy correcting the misreadings. Heritability is not destiny: a trait can be highly heritable and highly malleable at once, as eyesight is heritable yet corrected by glasses, and a heritable disorder can have an entirely environmental remedy. Heritability is not fixed: it shifts with the range of environments a population contains, which is why the same trait can be more or less heritable in rich and poor settings. And heritability is not a warrant for genetic determinism, because gene-environment correlation and interaction weave the two inheritances together in development. The molecular era has sharpened rather than settled these points. Polygenic scores now predict a portion of variance in cognitive and psychiatric traits, opening real prospects for research and raising equally real ethical questions about prediction, selection, and the reinscription of social inequality as biology. The field's open problems are substantial: much of the heritability estimated by twin studies is still not captured by identified variants, the so-called missing heritability; polygenic scores predict far better within the ancestry groups on which they were trained than across them, a portability problem with clear potential for inequity; and the causal path from a variant to a behavior runs through development and environment in ways a statistical association cannot reveal. What behavioral genetics offers the rest of psychology is not a genetic reduction of the mind but a discipline of variance, a set of methods for asking which differences among people are inherited, under what conditions, and to what degree, and a standing warning that both the hereditarian and the environmentalist extremes are refuted by its data.

Current Directions

The most consequential recent development is the maturation of the genome-wide approach into a routine engine of discovery, and with it a reframing of the whole field for the postgenomic era. The polygenic architecture that defeated candidate genes is now the working assumption, and effort has shifted from finding the gene for a trait to using the aggregate of thousands of variants as a research tool: polygenic scores are being used to study how genetic propensities unfold across development, how they correlate with the environments people select, and how they interact with social conditions (Harden, 2021). A decade of GWAS has delivered not only associations but a functional-genomic program that asks what the implicated variants actually do, in which tissues and pathways, turning statistical hits into biological hypotheses (Visscher et al., 2017). At the same time the field is reckoning with the limits and hazards of its new power. The missing-heritability gap between twin-study estimates and variant-based estimates is being narrowed by larger samples and better methods but is not closed. The poor cross-ancestry portability of polygenic scores, trained overwhelmingly on people of European descent, is now recognized as a first-order scientific and ethical problem rather than a technical footnote. And a synthesis of the field's most robust findings has been offered as a set of replicated results that any adjacent discipline should take as settled background, from the universality of heritability to the smallness of shared-environment effects (Plomin et al., 2016). The throughline is a discipline that has stopped arguing about whether genes matter and started working out, with new precision and new caution, exactly how. The questions it raises about prediction and individual difference connect directly to the study of cognition and to the broader project of cognitive psychology.

Common Misconceptions

A heritable trait is fixed at birth and cannot be changed.

Heritability describes the sources of variation in a population, not the malleability of a trait. Height is highly heritable yet has risen with nutrition, and a heritable condition can have a wholly environmental treatment. High heritability and high changeability routinely coexist (Turkheimer, 2000).

Heritability indicates how genetic an individual person's trait is.

It does not. Heritability is a property of a population, a statement about why people differ from one another, and it says nothing about the balance of genes and environment within any single person, in whom the two are inseparable (Plomin et al., 2016).

Scientists have found the genes for intelligence, or for any behavioral trait.

There is no such gene. Behavioral traits are influenced by thousands of variants each of nearly negligible effect, which is why single-gene candidate studies failed and why prediction requires summing the whole genome into a polygenic score (Border et al., 2019).

Glossary

Additive genetic variance.

The portion of trait variation due to the summed, independent effects of individual genetic variants; the component estimated as narrow-sense heritability.

Adoption study.

A design that separates genetic from environmental influence by comparing an adopted child's resemblance to birth parents (genes) and rearing parents (environment).

Behavioral genetics.

The study of how genetic differences between individuals contribute to differences in behavior and psychological traits across a population.

Candidate gene.

A specific gene chosen in advance on biological grounds and tested for association with a trait; an approach that largely failed to replicate for behavior.

Concordance.

The degree to which the two members of a twin pair match on a trait or diagnosis; higher in identical than fraternal pairs when a trait is heritable.

Dizygotic twins.

Fraternal twins, developed from two separate fertilized eggs, sharing on average half of the genetic variants that segregate in a family, like ordinary siblings.

Equal environments assumption.

The premise underlying the twin method that identical and fraternal twins raised together share equally similar trait-relevant environments, so their differing resemblance reflects genetic, not environmental, similarity.

Gene-environment correlation.

The nonrandom pairing of genotypes with environments, as when genetic propensities shape the settings and experiences a person encounters or selects.

Gene-environment interaction.

The dependence of a genetic effect on the environment, or of an environmental effect on the genotype, so that influences are not simply additive.

Genome-wide association study.

A method that scans millions of DNA variants across the whole genome in large samples to identify those statistically associated with a trait, without prior biological assumptions.

Heritability.

The proportion of variance in a trait, within a given population and context, attributable to genetic differences among individuals; a population statistic, not an individual one.

Missing heritability.

The gap between the heritability estimated by twin studies and the smaller portion currently explained by identified genetic variants.

Monozygotic twins.

Identical twins, developed from a single fertilized egg, sharing essentially all of their DNA; the more genetically similar of the two twin types.

Nonshared environment.

Environmental influences that make members of the same family differ despite shared genes and home; conventionally labeled E and including measurement error.

Polygenic score.

A single number summing the trait-associated effects of many genetic variants across a person's genome, used to predict a portion of a trait's variation.

Shared environment.

Environmental influences that make members of the same family alike regardless of genes, such as household resources; conventionally labeled C and usually small for behavior.

Twin study.

A design that compares the resemblance of identical and fraternal twins to estimate the genetic and environmental contributions to a trait's variation.

Key Researchers

Thomas J. Bouchard. Psychologist at the University of Minnesota; director of the Minnesota Study of Twins Reared Apart, whose findings on separated identical twins provided direct estimates of heritability. Google Scholar - Wikipedia

Francis Galton (1822-1911). The founder of the study of inherited individual differences at University College London; he introduced the twin method and the framing of nature versus nurture that behavioral genetics turned into a measurement. Wikipedia - Wikidata

Irving I. Gottesman (1930-2016). Behavioral geneticist at the University of Minnesota and the University of Virginia; he developed the polygenic liability-threshold model of schizophrenia and the concept of the endophenotype. ORCID - Google Scholar - Wikipedia - Wikidata

Kathryn Paige Harden. Behavioral geneticist at the University of Texas at Austin; a leading voice on polygenic scores in the postgenomic era and on the social and ethical implications of genetic prediction. Faculty Page - ORCID - Google Scholar - Wikipedia

Robert Plomin. Behavioral geneticist at King's College London; a leading figure in twin and adoption research, the Twins Early Development Study, and the move to polygenic prediction of psychological traits. Faculty Page - ORCID - Google Scholar - Wikipedia

Danielle Posthuma. Statistical geneticist at Vrije Universiteit Amsterdam; she led the fifty-years-of-twin-studies meta-analysis and large genome-wide studies of cognitive and psychiatric traits. Faculty Page - ORCID - Wikipedia

Eric Turkheimer. Psychologist at the University of Virginia; author of the three laws of behavior genetics and of the finding that socioeconomic status moderates the heritability of intelligence. Faculty Page - ORCID - Google Scholar - Wikipedia

Peter M. Visscher. Quantitative geneticist at the University of Queensland; he developed methods for estimating heritability directly from DNA and helped establish the polygenic architecture of complex traits. Faculty Page - ORCID - Google Scholar - Wikipedia

Frequently Asked Questions

What is behavioral genetics?
Behavioral genetics is the study of how genetic differences between people contribute to differences in their behavior and psychological traits. It uses designs such as twin, adoption, and genome-wide studies to estimate how much of the variation in a trait across a population is due to genetic rather than environmental differences (Plomin et al., 2016).

What does heritability actually mean?
Heritability is the proportion of the variation in a trait, within a particular population and set of conditions, that is due to genetic differences among individuals. It is a statistic about why people differ from one another, not a measure of how genetic any single person's trait is, and it changes when the range of environments changes (Turkheimer, 2000).

How do twin studies estimate genetic influence?
Identical twins share nearly all their DNA while fraternal twins share about half, and both types usually share a family environment. When identical twins resemble each other more than fraternal twins do, that extra similarity points to genetic influence, which can be converted into a heritability estimate (Boomsma et al., 2002).

Does high heritability mean a trait cannot be changed?
No. Heritability describes the sources of variation in a population, not the malleability of a trait. A trait can be highly heritable and still highly responsive to environment or treatment, just as heritable poor eyesight is routinely corrected with glasses (Turkheimer, 2000).

Is there a gene for intelligence or personality?
No single gene accounts for a behavioral trait. Such traits are highly polygenic, influenced by thousands of genetic variants that each contribute a tiny amount, which is why studies of single candidate genes failed to replicate and genome-wide methods were needed (Border et al., 2019).

What is a polygenic score?
A polygenic score sums the small trait-associated effects of many genetic variants across a person's genome into a single number that predicts a portion of a trait. Built from genome-wide association data, such scores capture the highly polygenic architecture of behavior and predict modestly but reliably (Visscher et al., 2017).

Why did the candidate-gene approach fail?
Candidate-gene studies tested one plausible gene at a time and reported associations, such as a serotonin-transporter variant predicting depression, that did not hold up in larger samples. Behavioral traits are shaped by thousands of variants of minute effect, so testing a single gene had almost no chance of detecting the true, distributed architecture (Border et al., 2019).

Can genes and environment be cleanly separated?
Only statistically, not causally. Genes and environments are correlated, because genetic propensities shape the settings people encounter, and they interact, because a gene's effect can depend on the environment. The heritability of intelligence in early childhood, for instance, is much lower in impoverished than in affluent families (Turkheimer et al., 2003).

References

Boomsma, D., Busjahn, A., & Peltonen, L. (2002). Classical twin studies and beyond. Nature Reviews Genetics, 3(11), 872-882. https://doi.org/10.1038/nrg932

Border, R., Johnson, E. C., Evans, L. M., Smolen, A., Berley, N., Sullivan, P. F., & Keller, M. C. (2019). No support for historical candidate gene or candidate gene-by-interaction hypotheses for major depression across multiple large samples. The American Journal of Psychiatry, 176(5), 376-387. https://doi.org/10.1176/appi.ajp.2018.18070881

Bouchard, T. J., Lykken, D. T., McGue, M., Segal, N. L., & Tellegen, A. (1990). Sources of human psychological differences: The Minnesota Study of Twins Reared Apart. Science, 250(4978), 223-228. https://doi.org/10.1126/science.2218526

Caspi, A., Sugden, K., Moffitt, T. E., Taylor, A., Craig, I. W., Harrington, H., McClay, J., Mill, J., Martin, J., Braithwaite, A., & Poulton, R. (2003). Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT gene. Science, 301(5631), 386-389. https://doi.org/10.1126/science.1083968

Chabris, C. F., Lee, J. J., Cesarini, D., Benjamin, D. J., & Laibson, D. I. (2015). The fourth law of behavior genetics. Current Directions in Psychological Science, 24(4), 304-312. https://doi.org/10.1177/0963721415580430

Harden, K. P. (2021). "Reports of my death were greatly exaggerated": Behavior genetics in the postgenomic era. Annual Review of Psychology, 72(1), 37-60. https://doi.org/10.1146/annurev-psych-052220-103822

Plomin, R., & Deary, I. J. (2015). Genetics and intelligence differences: Five special findings. Molecular Psychiatry, 20(1), 98-108. https://doi.org/10.1038/mp.2014.105

Plomin, R., DeFries, J. C., Knopik, V. S., & Neiderhiser, J. M. (2016). Top 10 replicated findings from behavioral genetics. Perspectives on Psychological Science, 11(1), 3-23. https://doi.org/10.1177/1745691615617439

Polderman, T. J. C., Benyamin, B., de Leeuw, C. A., Sullivan, P. F., van Bochoven, A., Visscher, P. M., & Posthuma, D. (2015). Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nature Genetics, 47(7), 702-709. https://doi.org/10.1038/ng.3285

Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current Directions in Psychological Science, 9(5), 160-164. https://doi.org/10.1111/1467-8721.00084

Turkheimer, E., Haley, A., Waldron, M., D'Onofrio, B., & Gottesman, I. I. (2003). Socioeconomic status modifies heritability of IQ in young children. Psychological Science, 14(6), 623-628. https://doi.org/10.1046/j.0956-7976.2003.psci_1475.x

Visscher, P. M., Wray, N. R., Zhang, Q., Sklar, P., McCarthy, M. I., Brown, M. A., & Yang, J. (2017). 10 years of GWAS discovery: Biology, function, and translation. The American Journal of Human Genetics, 101(1), 5-22. https://doi.org/10.1016/j.ajhg.2017.06.005