In 1979, Thomas Bouchard of the University of Minnesota read a newspaper story about identical twins who had been separated at birth and raised apart. He found James Springer and James Lewis -- both named James by their adoptive families, both married twice (first wife named Linda, second named Betty), both had a son named James Alan, both bit their nails, both had dogs named Toy, both drove the same car model, both chain-smoked Salems, and both vacationed at the same beach in Florida. The coincidences were too many to be coincidences. Bouchard launched the Minnesota Study of Twins Reared Apart -- and the data from that study would overturn what social science had assumed for decades.
The nature-nurture debate has been part of Western intellectual culture since at least Francis Galton, who coined both the phrase (in "English Men of Science," 1874) and the methodology (twin studies) for investigating it. For most of the 20th century, the social sciences operated under a strong environmental assumption: that human character, intelligence, and behavior were primarily products of experience, upbringing, and culture. Behaviorism, the dominant psychological paradigm from the 1920s through the 1960s, held that the organism was a blank slate that environment inscribed. John Watson famously claimed he could take any healthy infant and make them into any type of specialist regardless of the child's "tendencies, abilities, vocations, and race." This assumption shaped everything from child-rearing advice to educational policy to the treatment of mental illness.
Behavioral genetics -- the scientific program of systematically measuring the relative contributions of genetic and environmental variation to human traits -- has largely demolished the blank slate assumption without replacing it with genetic determinism. The emerging picture is more complex and more interesting than either extreme: genes and environments do not add together independently but interact, correlate, and shape each other through mechanisms that make the original question -- is it nature or nurture? -- fundamentally misconceived.
"Behavioral genetics has established that virtually all human psychological traits are heritable; that the effect of being raised in the same family is smaller than the effect of the genes; and that the complexity of human behavior cannot be captured in simple genetic or environmental models." -- Robert Plomin, Behavioral Genetics (2016)
Key Definitions
Heritability: The proportion of variation in a trait within a population that is statistically associated with genetic variation in that population. Not to be confused with "genetic causation" for individuals; a population-level, environment-specific measure.
Twin Studies: The research methodology comparing identical (monozygotic, MZ) twins -- who share essentially 100% of their DNA -- with fraternal (dizygotic, DZ) twins -- who share on average 50% of their segregating DNA, like ordinary siblings. If MZ twins are more similar than DZ twins, the difference is attributed to the additional genetic similarity.
Adoption Studies: Studies of biological and adoptive relatives that allow separation of genetic from environmental influences. Children resemble biological relatives they have never met (genetic influence) and may or may not resemble adoptive relatives they have been raised with (environmental influence).
Shared vs. Non-Shared Environment: The shared environment is the family environment common to siblings raised together -- same parents, same neighborhood, same family income, same household culture. The non-shared environment consists of unique experiences that differ between siblings: different peer groups, different classrooms, different within-family treatment.
Gene-Environment Interaction (GxE): Cases where the effect of an environmental exposure depends on genotype, or where the effect of a genotype depends on environment. Rather than adding independently, genes and environments multiply.
Epigenetics: Changes in gene expression that do not involve changes in DNA sequence -- typically through DNA methylation, histone modification, or non-coding RNA mechanisms. Provides a mechanism through which environment shapes gene activity.
Reaction Norm: The range of phenotypes (observable traits) that a given genotype can produce across different environments. High heritability does not mean a narrow reaction norm -- the same genotype can produce very different outcomes in different environments.
Behavioral Genetics: The scientific discipline studying the relative contributions of genetic and environmental factors to variation in behavioral, psychological, and cognitive traits.
Polygenic Traits: Traits influenced by many genetic variants, each of small effect, across the genome. Intelligence, height, personality, and most common diseases are polygenic. Contrast with monogenic traits (single-gene disorders like Huntington's disease).
GWAS (Genome-Wide Association Study): A research approach scanning hundreds of thousands to millions of genetic variants across the genome to identify those associated with a trait or disease in a large population.
Missing Heritability Problem: The observation that GWAS variants identified as statistically significant typically explain much less variance in a trait than twin-study heritability estimates would predict. Multiple mechanisms have been proposed.
Equal Environments Assumption: The assumption in twin studies that the environments of MZ and DZ twins are equally similar. Critics argue MZ twins may be treated more similarly (same bedroom, same activities, more often confused for each other), inflating heritability estimates. Research testing this assumption generally finds it holds for most psychological traits.
Nature of Nurture: Robert Plomin and colleagues' observation that many environmental measures used in psychology research are themselves heritable -- that is, genotype influences the environments people experience, so "environmental" variables contain genetic signal.
The Heritability Concept: What It Means and What It Does Not
Heritability is the central concept in behavioral genetics and the most commonly misunderstood. Getting it right is essential to interpreting the nature-nurture literature correctly.
The technical definition: heritability is the ratio of genetic variance to total phenotypic variance in a population. If heritability of IQ in a population is 0.60, it means that 60% of the variation in IQ scores among people in that population is statistically attributable to genetic variation among those people. The remaining 40% is attributable to environmental variation (including measurement error).
What heritability does not mean is critical. It does not mean that 60% of any individual's IQ "comes from" their genes. There is no meaningful sense in which you can decompose an individual's trait into genetic and environmental portions -- both are necessary conditions for the trait to exist at all. Every phenotype requires both genes and environment; remove either and the organism does not develop. Heritability describes the relative contribution of genetic versus environmental differences to the variation observed across individuals in a specific population.
Heritability is population-specific and environment-specific. This is the crucial mathematical property that makes the concept more subtle than it appears. Consider height. In a population with uniform, adequate nutrition, most of the variation in height will reflect genetic differences -- heritability will be high, perhaps 90%. In a population with highly variable nutrition -- some severely malnourished, some well-fed -- nutritional variation will contribute substantially to height variation and heritability will be lower, perhaps 50-60%. The heritability of the same trait is different in different environments. The implication: high heritability does not preclude large environmental effects. A trait can be highly heritable and still be dramatically affected by changing the environment.
The development of heritability estimates across the lifespan reveals a counterintuitive pattern. Heritability increases with age for most psychological traits. A comprehensive meta-analysis by Haworth et al. (2010), published in Molecular Psychiatry (doi: 10.1038/mp.2009.55), pooled data from twin pairs across four countries and found mean heritability estimates of approximately 41% in childhood (ages 4-10), 55% in adolescence (ages 11-18), and 66% in adulthood (ages 19+) for cognitive ability. For personality traits, similar developmental increases are observed.
Why does heritability increase with age? The most compelling explanation involves gene-environment correlation: as people grow older, they gain increasing autonomy to select environments compatible with their genotypes. A genetically bookish child may be placed in mixed-ability schooling, but a genetically bookish adult selects into universities, libraries, and intellectual careers. This active niche-picking amplifies genetic influences over time. Simultaneously, the shared family environment's influence wanes as children leave the home and its effects dilute.
Twin Study Evidence: The Minnesota Study and Its Legacy
The classical methodology of behavioral genetics compares identical (monozygotic, MZ) twins -- who share effectively 100% of their DNA -- with fraternal (dizygotic, DZ) twins -- who share on average 50% of their segregating DNA, the same as non-twin siblings. If MZ twins are substantially more similar than DZ twins on a trait, genetic factors are implicated. The "equal environments assumption" -- that the environmental similarity of MZ and DZ twins is comparable -- has been tested extensively and generally holds for most psychological traits, though critics maintain it is imperfect.
Bouchard's Minnesota Study of Twins Reared Apart (MISTRA), the most famous behavioral genetics study ever conducted, circumvented the equal environments problem entirely. Reared-apart MZ twins share their genes but not their family environment. Any similarity between them must therefore reflect genetic influence rather than shared upbringing. The study recruited approximately 137 twin pairs (including both MZ and DZ reared-apart pairs) between 1979 and 1999, subjecting each pair to about fifty hours of psychological, physiological, and cognitive testing.
The core finding, published in Bouchard et al.'s landmark 1990 Science paper (doi: 10.1126/science.2218526), was that MZ twins reared apart showed substantial similarity on essentially every psychological measure tested: IQ (heritability estimates around 70%), all five Big Five personality dimensions (40-60% heritability), vocational interests (40-50%), religious attitudes, social conservatism, and specific phobias and quirks. Strikingly, MZ twins reared apart were approximately as similar on most measures as MZ twins reared together -- suggesting that sharing a family environment added relatively little psychological similarity beyond the shared genes.
The James twins were not anomalous in the broader dataset, merely the most vivid illustration of a general pattern. Other pairs showed similarities in mannerisms, speech patterns, sense of humor, and specific habits that seemed implausible as genetic products but were observed repeatedly. Heritability estimates for specific traits from the broader twin literature include: IQ 60-80% (adult); schizophrenia approximately 80%; bipolar disorder approximately 75%; major depression approximately 40-50%; anxiety disorders approximately 30-40%; political orientation (conservatism vs. liberalism) approximately 40-50% (Alford, Funk, and Hibbing 2005, American Political Science Review); sexual orientation in males approximately 18-39% (Langstrom et al. 2010, Archives of Sexual Behavior).
Adoption studies provide convergent evidence. Studies of adopted children consistently show that they resemble their biological parents (whom they have typically never met) more than their adoptive parents (who raised them) on measures of IQ, personality, and temperament by adulthood. The Minnesota Adoption Studies, also run from the University of Minnesota, showed that by early adulthood, adopted children's IQ scores correlated substantially with their biological parents and not significantly with their adoptive parents.
The Non-Shared Environment Puzzle
If heritability accounts for roughly 50% of variance in most traits, and shared family environment accounts for very little (often near zero, as consistently found in twin and adoption studies), then roughly 50% of variance is attributed to non-shared environment -- the unique experiences that make siblings raised in the same home different from each other.
This non-shared environment component represents the main concession of behavioral genetics to the importance of environment. But identifying what it consists of has proven surprisingly difficult. Judith Rich Harris, a developmental psychology textbook author who was rejected by the graduate program that had admitted her and worked largely outside academia, argued in "The Nurture Assumption" (1998) that the answer was peer groups. Children, she argued, adapt their behavior to their peer group, not to their parents. The personality and values of siblings differ because they occupy different social niches -- different peer groups, different classroom social hierarchies -- even while sharing the same parents. Children of immigrants adopt the accent and values of their peers, not their parents, as every immigrant family has observed.
Harris's peer group theory remains controversial. Critics note that parents clearly affect outcomes beyond personality: educational access, physical safety, values explicitly taught, opportunities provided. The meta-analytic literature on parenting effects is complicated by the genetic confound -- studies of parenting and child outcomes rarely separate the genetic from the environmental contribution. Adopted children provide cleaner tests, and adoption studies do show parental effects on education, values, and cognitive outcomes in childhood, even if personality is less affected.
Robert Plomin, the most prolific behavioral geneticist of the modern era, acknowledges the difficulty of identifying the non-shared environment and suggests that part of it may be genuinely random -- stochastic variation in brain development, epigenetic noise, the essentially unpredictable neurobiological events of individual development. If so, some "environmental" variance is neither genetic nor identifiable environmental influence, but developmental noise. This would have the uncomfortable implication that much of what makes individuals different is neither caused by genes nor by identifiable environmental factors, but by processes too fine-grained to be captured in any study.
Gene-Environment Interaction and Correlation
The most important advance beyond the simple variance-partitioning approach of classical twin studies has been the investigation of how genes and environments interact and correlate.
Gene-environment interaction (GxE) describes cases where the effect of an environmental exposure depends on genotype. The classical illustration is Caspi et al.'s 2003 Science paper (doi: 10.1126/science.1083968) examining the MAOA gene and childhood maltreatment. The study followed 442 New Zealand males from birth to adulthood and found that childhood maltreatment predicted antisocial behavior in adulthood primarily in individuals with the low-activity MAOA variant. Those with the high-activity variant showed little elevation in antisocial behavior even when maltreated. A gene previously associated with aggression ("the warrior gene" in popular media) turned out not to have a simple effect but an interaction effect: the gene influenced sensitivity to environment, not behavior directly.
The MAOA finding attracted massive attention and generated numerous attempted replications with mixed results. A 2014 meta-analysis by Byrd and Manuck in Biological Psychiatry pooled 27 studies and found support for the interaction in males, though effect sizes were modest and heterogeneity was high. Other GxE interactions in the mental health literature (the 5-HTTLPR serotonin transporter gene and stress, for instance) have been less well replicated. The general lesson is that GxE interactions probably exist, are probably widespread, but identifying specific robust interactions has been methodologically challenging.
Gene-environment correlation (rGE) is conceptually distinct and may be more pervasive than GxE interaction. Three types:
Passive rGE: biological parents provide both genes and environments, which naturally correlate. A child of highly educated parents inherits genes associated with educational attainment and also grows up in a home full of books. The environmental effect of books-in-home is partly a proxy for genetic effects passed through parents. Studies using adopted children, who do not share genes with their environmental providers, provide cleaner estimates.
Evocative rGE: genetically influenced traits evoke different responses from the environment. A highly active, bold child may elicit more opportunities for physical activity and leadership from parents, teachers, and peers. A shy, anxious child may elicit protective and limiting responses. The environment is shaped by genetically influenced traits, creating a correlation between genetic factors and environmental experiences.
Active rGE (niche-picking): individuals actively select environments compatible with their genotype. As people gain more autonomy -- in adolescence and adulthood -- they increasingly select peer groups, activities, careers, and partners that match their genetically influenced propensities. The bookish person gravitates toward libraries; the risk-taker toward extreme sports. This active selection amplifies genetic differences and is a major reason heritability increases with age.
The implication of rGE is that "environmental" measures used in psychological research -- family income, parenting quality, peer group characteristics -- often contain substantial genetic signal. Robert Plomin coined the phrase "the nature of nurture" to describe this: what we think of as environmental influences on children are partly genetic influences routed through genetically influenced environments.
The GWAS Era: Polygenic Scores and Missing Heritability
The revolution in genotyping technology since the mid-2000s has enabled a new form of genetic research: genome-wide association studies (GWAS) that scan hundreds of thousands to millions of genetic variants simultaneously across large samples, identifying those statistically associated with traits of interest.
For educational attainment, Lee et al.'s 2018 Nature Genetics paper (doi: 10.1038/s41588-018-0147-3) studied 1.1 million people of European ancestry and identified 1,271 independent genetic variants significantly associated with years of schooling. Combined into a polygenic score, these variants explain approximately 11-13% of variance in educational attainment -- already more predictive than many single environmental variables, and this from a subset of the genetic variants with effects large enough to reach genome-wide significance.
For schizophrenia, the Schizophrenia Working Group of the Psychiatric Genomics Consortium identified 128 significant genetic loci by 2014 (Nature 2014), with heritability estimates from GWAS-based methods of approximately 20-25% for common genetic variants alone.
The "missing heritability" problem -- why GWAS explains much less variance than twin studies suggest -- has been substantially clarified. The primary explanation is that complex traits are influenced by a very large number of variants each of very small effect, most of which are below the genome-wide significance threshold in any sample. When polygenic scores are built using all variants (not just significant ones) weighted by their estimated effects, prediction improves substantially. A 2019 study by Allegrini et al. in Molecular Psychiatry found that polygenic scores for educational attainment built from millions of variants explained approximately 17% of variance in educational achievement -- still less than twin-study heritability, but substantially more than scores from significant variants only.
Rare variants, gene-environment interactions, and structural genomic variation (copy number variants, insertions, deletions) contribute additional unexplained variance. The complete genetic architecture of complex traits remains a work in progress.
Epigenetics: How Environment Shapes Gene Expression
Epigenetics refers to heritable (through cell division) changes in gene expression that do not involve changes in DNA sequence. The main mechanisms are DNA methylation (addition of methyl groups to cytosine bases, typically reducing gene expression), histone modification (changes to the proteins around which DNA is wrapped, affecting accessibility), and non-coding RNA regulation.
The foundational animal model comes from Michael Meaney and Moshe Szyf at McGill University. Mother rats differ in how much they lick and groom their pups in the first week of life. High-licking mothers produce offspring with increased glucocorticoid receptor expression in the hippocampus, a larger hippocampus, lower lifetime corticosterone responses, and lower anxiety -- a more resilient stress phenotype. Low-licking produces the opposite. The mechanism is DNA methylation of the glucocorticoid receptor gene promoter: high-licking mothers produce pups with less methylation and more receptor expression. Crucially, cross-fostering experiments showed that the effect was transmitted through maternal behavior, not through genes: pups born to low-licking mothers but fostered by high-licking mothers showed the high-licking epigenetic and behavioral phenotype. An environmental experience in early life was leaving a chemical mark on the genome that altered stress physiology for the life of the animal.
Rachel Yehuda's research on Holocaust survivors and their children claimed to find epigenetic differences in stress-related genes in second-generation survivors, suggesting that the trauma of parents was transmitted to offspring through epigenetic mechanisms. This transgenerational epigenetic inheritance claim has been influential but controversial. Human evidence for true germline epigenetic transmission across generations (as opposed to environmental effects in utero or in shared postnatal environments) is limited and methodologically challenging to establish, because the epigenome is largely reset during the formation of gametes.
Within a lifetime, epigenetic mechanisms are clearly important mediators of GxE interaction: they explain part of how environmental experiences get "under the skin" and alter gene expression in ways that affect health and behavior. The stress physiology of early adversity, the cognitive effects of early enrichment, and the molecular mechanisms of learning and memory all involve epigenetic processes.
The Contemporary View: Why the Dichotomy Is False
The nature-nurture dichotomy was always a false dichotomy, and the science of the past three decades has made this unmistakably clear. Every trait is simultaneously 100% genetic and 100% environmental: both genes and environment are necessary conditions for any phenotype to exist. Remove either and the organism does not develop. The question is never which one -- it is always how they work together.
The more precise framework: genes set reaction norms (the range of possible phenotypes for a given genotype across environments). Environments determine where within that range an individual falls. Gene-environment interactions mean that reaction norms are not the same for all genotypes -- some genotypes are highly sensitive to environmental conditions (high plasticity), others are robust across a wide range of environments.
The practical implications of this framework are important for policy. High heritability does not imply fatalism or the futility of environmental intervention. The phenylketonuria example -- a genetic disease entirely preventable through dietary intervention -- demonstrates that high heritability is compatible with effective environmental treatment. Height has 80-90% heritability but has increased dramatically across populations as nutrition improved. IQ heritability increases with age within well-resourced populations, but Flynn effect data (the secular rise of IQ scores across the 20th century -- approximately 30 points over 100 years) shows that environmental improvements can shift entire population distributions.
What behavioral genetics contributes to practical thinking about human development is not a counsel of despair but a corrective to naive environmentalism. Not all differences in outcome reflect differences in treatment or opportunity; genetic variation contributes substantially to individual differences. This does not mean inequality is natural or inevitable -- it means that eliminating inequality requires more than simply equalizing environments, because individuals bring different genetically influenced starting points to those environments. The goal of good policy is not to overcome genetics but to create environments where all genotypes can flourish.
Robert Plomin, in his 2018 book "Blueprint," argued that the cumulative finding of behavioral genetics -- that genes matter more than shared environment for most psychological traits -- should prompt a fundamental rethinking of education, parenting, and social policy. Others, including behavioral geneticist Kathryn Paige Harden in "The Genetic Lottery" (2021), have argued that acknowledging genetic luck -- the morally arbitrary assignment of advantageous or disadvantageous genotypes -- should motivate greater redistribution and social support, not complacency. The science does not resolve the political debate, but it reframes it.
References
- Bouchard, Thomas J., et al. "Sources of Human Psychological Differences: The Minnesota Study of Twins Reared Apart." Science 250 (1990): 223-228. doi: 10.1126/science.2218526
- Plomin, Robert, John C. DeFries, Valerie S. Knopik, and Jenae M. Neiderhiser. Behavioral Genetics. 7th ed. Worth Publishers, 2016.
- Haworth, Claire M.A., et al. "The Heritability of General Cognitive Ability Increases Linearly from Childhood to Young Adulthood." Molecular Psychiatry 15 (2010): 1112-1120. doi: 10.1038/mp.2009.55
- Caspi, Avshalom, et al. "Role of Genotype in the Cycle of Violence in Maltreated Children." Science 297 (2002): 851-854. doi: 10.1126/science.1083968
- Harris, Judith Rich. The Nurture Assumption: Why Children Turn Out the Way They Do. Free Press, 1998.
- Lee, James J., et al. "Gene Discovery and Polygenic Prediction from a Genome-Wide Association Study of Educational Attainment in 1.1 Million Individuals." Nature Genetics 50 (2018): 1112-1121. doi: 10.1038/s41588-018-0147-3
- Meaney, Michael J. "Maternal Care, Gene Expression, and the Transmission of Individual Differences in Stress Reactivity Across Generations." Annual Review of Neuroscience 24 (2001): 1161-1192.
- Harden, Kathryn Paige. The Genetic Lottery: Why DNA Matters for Social Equality. Princeton University Press, 2021.
See also: How Epigenetics Works, What Is Intelligence, What Is Personality
Frequently Asked Questions
What does heritability actually mean?
Heritability is one of the most frequently misunderstood concepts in science, and getting it right matters for understanding the nature-nurture debate. Heritability is a statistical property of a population: it measures the proportion of variation in a trait within a specific population that is associated with genetic variation in that population. It does not measure how much of any individual's trait is 'due to' genes. A heritability of 0.60 for IQ does not mean that 60% of your intelligence comes from your genes and 40% from your environment. It means that if you take a population of people in similar environments, 60% of the variation in IQ scores among those people is associated with genetic differences. Three crucial caveats follow. First, heritability is population-specific and environment-specific. The heritability of height in a population where everyone eats adequately might be 90%, because most variation comes from genes. In a population with highly variable nutrition, heritability could be much lower because nutrition variation contributes substantially to height variation. Second, high heritability does not mean 'genetically determined' or 'unchangeable by environment.' Phenylketonuria (PKU) is a genetic disease with effectively 100% heritability -- yet it is entirely preventable through dietary modification. Third, heritability is not fixed across development: IQ heritability increases substantially with age. A 2010 meta-analysis by Haworth et al. in Molecular Psychiatry (doi: 10.1038/mp.2009.55), pooling data from 11,000 twin pairs across 4 countries, found heritability estimates of approximately 41% in childhood, 55% in adolescence, and 66% in adulthood -- apparently because as people age they increasingly select environments compatible with their genotype, amplifying genetic influences.
What did the Minnesota Study of Twins Reared Apart find?
The Minnesota Study of Twins Reared Apart (MISTRA), launched by Thomas Bouchard and colleagues at the University of Minnesota in 1979 and continuing through 1999, is the most extensive study of identical twins separated at birth and raised in different families. The study eventually examined approximately 137 reared-apart twin pairs (both identical and fraternal), administering an extensive battery of psychological, physiological, and cognitive assessments. The headline findings were striking. Identical twins reared apart showed substantial similarities in IQ (heritability estimates around 70%), personality (all five of the Big Five personality dimensions showed heritabilities of 40-60%), vocational interests, social attitudes (including political conservatism and religiosity), specific fears and phobias, and even seemingly arbitrary details like favored hobbies, gestures, and speech patterns. Bouchard's 1990 Science paper (doi: 10.1126/science.2218526) reported that on most measures, identical twins reared apart were about as similar as identical twins reared together -- suggesting that sharing a family environment added relatively little to similarity beyond shared genes. The famous 'Jim twins' -- James Springer and James Lewis -- who had both named their sons James Alan, both had dogs named Toy, both married women named Linda then Betty, and both vacationed at the same Florida beach, became symbols of the study's findings, though critics note that such coincidences are more likely to be noticed and remembered than dissimilarities. The study's core finding -- that genetic similarity predicts psychological similarity even across radically different rearing environments -- has been broadly replicated and remains one of the most robust findings in behavioral genetics.
If parents do not matter much for personality, what does?
This is the puzzle at the heart of modern behavioral genetics. Twin studies consistently show that heritability accounts for roughly 50% of variance in most personality and psychological traits, shared family environment (the environment shared by siblings raised in the same home -- same parents, same neighborhood, same income) accounts for very little (often near zero), and non-shared environment accounts for the remaining 40-50%. Non-shared environment includes all the unique experiences that make siblings different from each other despite sharing parents and a home: different peer groups, different classrooms, different friendships, different experiences within the family (being the eldest versus youngest), chance events, and random developmental variation. Judith Rich Harris, in 'The Nurture Assumption' (1998), argued controversially that the correct conclusion is that peer groups -- not parents -- are the primary environmental influence on personality development. Children's personalities are shaped by their social group outside the home: they learn to behave, dress, speak, and value things in ways appropriate to their peer group, not their parents. This explains the surprising finding that children of immigrants typically adopt the values and accents of their peers, not their parents. Harris's thesis remains controversial. Critics note that parents clearly matter for education, safety, values, and many outcomes beyond personality narrowly defined. Robert Plomin, one of the leading behavioral geneticists, acknowledges the non-shared environment finding but notes that identifying what specifically constitutes the non-shared environment has been empirically difficult. It may include largely random developmental noise -- stochastic variation in brain development that is neither genetic nor obviously environmental.
What is gene-environment interaction and why does it matter?
Gene-environment interaction (GxE) describes cases where the effect of a specific environmental exposure depends on an individual's genotype, or equivalently, where the effect of a genotype depends on environmental conditions. Rather than genes and environment adding together independently, they multiply: the same environment has different effects on people with different genotypes. The most famous GxE study is Caspi et al.'s 2003 Science paper (doi: 10.1126/science.1083968), which examined whether childhood maltreatment predicted antisocial behavior in adulthood differentially depending on a functional variant in the MAOA gene (which codes for an enzyme that metabolizes monoamine neurotransmitters). They found that maltreatment predicted antisocial behavior only in individuals with the low-activity MAOA variant, not in those with the high-activity variant -- a classic interaction. The finding attracted enormous attention and generated over 100 attempted replications. Results have been mixed: some meta-analyses support the interaction, others do not. The MAOA study illustrates both the promise and the methodological challenges of GxE research. Gene-environment correlation (GxE correlation) is distinct: rather than genotype changing the response to environment, genotype influences the environments people experience. Passive GxE correlation occurs when biological parents provide both genes and environment (a child of musical parents inherits musical genes and grows up in a music-rich home). Evocative GxE correlation occurs when genetically influenced traits evoke different responses from others (a highly active child may elicit more physical activity opportunities from coaches and parents). Active GxE correlation (sometimes called niche-picking) occurs when individuals with different genotypes actively select compatible environments -- genetically bookish people gravitate toward libraries. Active GxE correlation increases with age as individuals gain more control over their environments, which is part of why heritability increases through development.
What have genome-wide association studies found about complex traits?
Genome-wide association studies (GWAS) scan hundreds of thousands to millions of genetic variants across the genome simultaneously in large samples, testing each variant for association with a trait. They have transformed our understanding of the genetic architecture of complex traits -- and revealed that the architecture is far more complex than classical genetics suggested. For educational attainment -- one of the most heavily studied traits -- Lee et al.'s 2018 Nature Genetics GWAS (doi: 10.1038/s41588-018-0147-3) studied 1.1 million people and identified 1,271 independent genetic variants associated with years of education, together explaining about 11-13% of variance in educational attainment. This sounds modest but it is remarkable: these common genetic variants alone -- which can be combined into a polygenic score -- have more predictive power for educational attainment than any single environmental factor studied. For schizophrenia, the Schizophrenia Working Group of the Psychiatric Genomics Consortium identified 128 genetic loci in their 2014 Nature paper. For BMI, height, and coronary artery disease, hundreds of significant loci have been identified. But a persistent puzzle has been the 'missing heritability' problem: twin studies suggest that height, for example, is about 80% heritable, but GWAS variants explain only about 20-25% of variance. Multiple mechanisms have been proposed: (1) a very large number of variants each of tiny effect that GWAS has insufficient power to detect; (2) gene-environment interactions that inflate twin study heritability estimates; (3) rare variants not well captured by standard GWAS; (4) structural variation (copy number variants) not well captured in SNP arrays. Current evidence suggests explanation (1) is primary -- polygenic scores built from all variants, not just genome-wide significant ones, explain substantially more variance.
What does epigenetics add to the nature-nurture picture?
Epigenetics -- changes in gene expression that do not involve changes in DNA sequence -- has become one of the most exciting and overhyped areas of biology in relation to the nature-nurture question. The core finding is that environmental experiences can alter the chemical marks on DNA and the proteins around which DNA is wrapped (histones), changing which genes are expressed and how strongly, without changing the DNA sequence itself. Michael Meaney and Moshe Szyf's rat licking studies at McGill University are among the most cited. They showed that the amount of licking and grooming a mother rat provided to her pups in early life altered the methylation of the glucocorticoid receptor gene in the pups' hippocampus, changing the stress response system in lasting ways. High-licking mothers produced pups with more glucocorticoid receptors, better able to dampen their stress response. Low-licking produced more reactive offspring. Critically, these effects could be reversed by cross-fostering: pups born to low-licking mothers but raised by high-licking mothers showed the high-licking epigenetic profile. The mechanism of environmental influence on gene expression was identified. Rachel Yehuda's research on Holocaust survivors and their offspring reported that children of Holocaust survivors showed lower cortisol levels and altered stress responses, with epigenetic differences in genes regulating stress response -- a claim of transgenerational epigenetic inheritance in humans. This finding has been influential but controversial: replication has been partial and the mechanisms for human transgenerational epigenetic inheritance are unclear, as epigenetic marks are largely reset during the formation of sperm and eggs. Epigenetics is a genuine and important mechanism of GxE interaction within a lifetime, but claims of transgenerational inheritance in humans require stronger evidence than currently available.
Does high heritability mean we cannot change outcomes?
No -- and this is arguably the most important practical point in the entire nature-nurture debate. High heritability does not imply that traits are fixed, unchangeable, or unresponsive to environmental intervention. The proof-of-concept example is phenylketonuria (PKU). PKU is a genetic metabolic disorder caused by mutations in the gene encoding the enzyme phenylalanine hydroxylase. Without the enzyme, phenylalanine accumulates to toxic levels, causing severe intellectual disability. The disorder is essentially 100% heritable -- essentially all variation in who gets PKU is explained by genetic variation. And yet PKU is entirely preventable through a simple dietary intervention: eliminating phenylalanine from the diet prevents the toxic accumulation and allows normal development. All newborns in developed countries are screened for PKU at birth. High heritability and effective environmental intervention are entirely compatible. Similarly, the heritability of height is approximately 80%, but populations have grown substantially taller over the 20th century as nutrition improved -- a pure environmental change operating on a highly heritable trait. The heritability of educational attainment is around 50-60%, but educational interventions -- high-quality early childhood programs, effective schooling, tutoring -- improve educational attainment. For policy purposes, heritability tells us about sources of individual differences in a population at a given time; it tells us nothing about whether changing the environment will improve average outcomes. The dichotomy between 'genetic' and 'changeable' is simply false. Every trait is simultaneously genetic and malleable.