Genetic and environmental continuity in personality development: A meta-analysis

Psychological Bulletin 140(5):1303–1331, doi 10.1037/a0037091; NIH author manuscript, PMC4152379. Read by researcher R4e (innate against learned) for research batch R4, 5 October 2026, as the batch's meta-analysis of the heritability of personality. Provenance: papers/base/briley2014_personality_continuity.provenance.json.

Why this paper

The brief asked for one meta-analysis of the heritability of personality or of social attitudes. This one was chosen because it is open (PMC) and because it answers the developmental half of the innate-against-learned question, not only "how heritable": it follows genetic and environmental influences on personality, and their stability, from age 1 to 86. Alternatives checked in PMC only: Vukasović and Bratko 2015 (personality) is not there; the open attitudes candidate, Hatemi et al. 2014 (PMC4038932), is a pooled twin analysis of five democracies with a genome-wide part, not a meta-analysis. Neither was fetched.

What was read

• Read in full: every line of the author manuscript, converted from PMC XML: abstract, introduction and theory, method, results, discussion, strengths and limitations, conclusion, footnotes, reference list, five figure captions and all eight tables.
• Not read: the online supplement (Table S1 and Figures S1–S3, the self-report-only sensitivity analysis; its results are summarised in the main text); figures as images.

What they did

• Data: 24 longitudinal behaviour-genetic studies from 21 samples: 21,057 sibling pairs (7,787 identical and 12,951 fraternal twin pairs reared together, plus small numbers of twins reared apart, half-siblings and unrelated siblings), more than 40,000 people aged 1 to 86. Samples are from Minnesota, Finland, the Netherlands, Sweden, Australia, Croatia, Germany, Japan (one), Wisconsin, Boston and Southern California. Mostly White; ethnicity was poorly reported.
• Measures: self- and informant-reported temperament and personality scales, coded into the Big Five. The mean interval between waves was 5.4 years.
• Model: a longitudinal model with additive genes (A) and non-shared environment (E). Shared environment (C) was not modelled, because "personality traits are only rarely influenced by this class of effects"; most source studies had already dropped C from their reported models. The environmental component was also corrected for measurement error (Cronbach's α; mean reliability .78), giving lower bounds for true environmental variance.
• Analysis: random-effects meta-regression with age splines (breaks at 3, 15, 30 and 60) and continuous exponential curves, controlling for the time between waves.

Main results (verified)

How much is genetic, by age (proportions of variance):

• Heritability falls steeply in childhood (−.013 a year to age 15) and slowly after. The authors summarise it as roughly 40% genetic throughout adulthood.
• Error-corrected environmental variance rises until about 30 and is flat afterwards.

How stable each part is (correlations over about 5.6 years):

• Genetic influences become perfectly stable by about 30 and stay so.
• True environmental influences keep becoming more stable all through life (corrected .81 at 45, .96 predicted at 90).

Where stability comes from:

• The genetic contribution to personality stability is roughly constant at about .36 across the lifespan (exponential asymptote .38, reached by about age 8). Heritability falls while genetic stability rises, and the two offset each other.
• The environmental contribution rises from near zero in early childhood (.03) to about .17 at 15, .25 at 30 and .31 after 60.
• The authors' conclusion: "the lifespan trend of increasing phenotypic stability … predominantly results from environmental mechanisms." In infancy stability is all genetic; by midlife genes and environment contribute almost equally.

Other findings:

• Differences between the Big Five are "strikingly trivial" (about 1% of variance). Extraversion and conscientiousness are slightly more heritable, agreeableness slightly less.
• Self-report data alone (from age 9) show weaker, non-significant age trends in heritability, but the same conclusion about stability: from age 9 the environmental contribution rises by .31 against .14 for the genetic one.
• Publication bias may affect low-precision studies; dropping effect sizes with SE > .075 left the trends unchanged.

The authors' interpretation

• Both theories are half right. Five Factor Theory (personality as biological maturation) correctly predicted that genetic stability peaks near 30 and that genes contribute throughout. Social-investment theories correctly predicted rising environmental influence and environmental stability. Genetic stability near 1.0 after 30 "suggests that the environmental changes typically experienced in adulthood do not meaningfully influence gene expression relevant to personality."
• Candidate mechanisms:
  • early gene-by-shared-environment interaction (which inflates infant heritability);
  • active gene–environment correlation (people choose environments that fit their genes);
  • accumulating identity-relevant experiences that make even identical twins less alike;
  • stable adult roles that reduce novelty.
• The non-shared environment is real and enduring, not just measurement error and not just passing states. Yet measured environments explain under 5% of it (Turkheimer and Waldron 2000, cited). On the genetic side, common SNPs recover only about a third of twin heritability for personality (Vinkhuyzen 2012, cited).

Limits

• Shared environment fixed at zero by design. This meta-analysis cannot show that it is zero; it assumes so, following earlier literature.
• Correlation matrices were reconstructed from published, often trimmed, models. Correlations, not covariances, so changes in total variance are invisible.
• Early-childhood heritability may be inflated by parent report: activity level at age 2 was .84 by parent report against .55 by observer (Saudino 2012).
• The age breakpoints are partly arbitrary; cohort effects are confounded with age.
• Overwhelmingly Western, mostly White samples.

What it means for Amadeus (inference)

• A person's dispositions are set by genes and by their own accumulated experience, not by the culture or family they share with others. Adult personality is about 40% heritable. The rest is non-shared environment (at least about 30% after removing measurement error), and the family-shared environment was assumed to be about zero. None of this is "base" content: heritable differences are still differences. The genetic share belongs in the slot as a person setting, and the base holds only its population distribution (B8).
• Experience writes the person more as life goes on. In childhood, stability is all genetic; by midlife, stable environmental effects contribute as much. The carrying-on goal concerns adults, so most of what makes this adult this adult, and keeps them so, is consolidated individual experience. That is exactly what the episode store and gated consolidation (B2, B3) are meant to hold.
• A stable core plus slow, environment-driven change. Genetic influences are fixed after 30, and environmental influences keep stabilising. That fits a design where the slot has a fixed part (the person's starting settings) and a part that changes only through consolidated experience, and where change becomes harder with age. The "rule of change" (B6) should probably include age-dependent plasticity.
• Do not expect to recover the slot from a list of life events. Measured environments explain under 5% of the non-shared variance. A person's divergences come from many small, idiosyncratic experiences, which supports dense personal data over biographical summaries.