KurisutinaPerspectives on Psychological Science 11(1):3–23, doi 10.1177/1745691615617439; Europe PMC funders' author manuscript, PMC4739500. Read by researcher R4e (innate against learned) for research batch R4, 5 October 2026, in place of Polderman et al. 2015 (see below). Provenance: papers/base/plomin2016_top10_behavioral_genetics.provenance.json.
Why this paper, and what was read
- Swap: the brief's candidate, Polderman et al. 2015 (Nature Genetics, the meta-analysis of 50 years of twin studies), is paywalled at the publisher. Every open author-institution copy found (Amsterdam UMC, UQ eSpace, the CORE mirror) sits behind a Cloudflare or CAPTCHA bot check, which the project's rules forbid getting around. It was not read. This review is the open, cross-domain substitute; it cites Polderman's headline only.
- Read in full: every line of the author manuscript converted from PMC XML: abstract, introduction, all ten findings, "why behavioral genetic results replicate", conclusions, funding, the reference list and the five figure captions.
- Not read: figures as images; the textbook (Plomin et al. 2013) the findings are abstracted from.
- Nature of the numbers: a review. The authors chose the ten findings by their own judgement ("not all … supported by formal meta-analyses"). Each number below is their report of a cited study, named with it.
Cautions the authors state first
- Heritability describes a population, not a person. It "describes 'what is' in a population"; "does not refer to a single individual but rather to individual differences in a particular population at a particular time with its particular mix of genetic and environmental effects"; and "does not imply immutability". If heritability were near 100%, existing environmental differences would not matter for that trait at that age, but new environments still could.
- Samples: most research comes from developed countries; results could differ elsewhere.
- Not yet robust enough to list: differences in heritability between personality traits, sex differences in heritability, and gene–environment interaction.
The ten findings, with the numbers given
| # | Finding | Numbers (cited source) |
|---|---|---|
| 1 | All psychological traits show significant and substantial genetic influence | Intelligence: twin correlations about 0.85 identical against 0.60 fraternal, 10,000 pairs (Bouchard and McGue 1981, as modified by Loehlin 1989); also found in Russia, former East Germany, Japan, and rural and urban India. Heritability of intelligence about 50% (95% CI about 45–55%). Schizophrenia concordance about 50% identical against 15% fraternal (Sullivan 2003). Personality: heritabilities usually 30–50%; 24,000 twin pairs (Loehlin 1992). Well-being: 36% (34–38%), 10 studies, 56,000 people (Bartels 2015). "Political beliefs, religiosity, altruism and food preferences" also show significant genetic influence (textbook; no numbers). Polderman 2015: nearly 18,000 traits from about 3,000 publications and 15 million twin pairs. "We are unable to name an exception." |
| 2 | No traits are 100% heritable | Behavioural traits typically 30–50%, against about 90% for height. Measurement error is part of the non-heritable share. This is "the strongest available evidence for the importance of environmental influence". |
| 3 | Heritability comes from many genes of small effect | Largest GWAS effects: under 1.1-fold odds for schizophrenia; 0.0002 of variance in years of schooling per top SNP |
| 4 | Correlations between traits are largely genetic | Among intelligence, reading, mathematics and language in about 5,000 twelve-year-old twins, genes account for 53–65% (mean 61%) of the correlations (Davis 2009). Anxiety and depression correlate "entirely for genetic reasons" (Middeldorp 2005). Genes account for over 70% of the r ≈ 0.30 between exercise attitudes and exercise behaviour (Huppertz 2014, unreplicated) |
| 5 | Heritability of intelligence rises with age | Abstract: from about 20% in infancy to about 60% in adulthood. 41% at age 9, 55% at 12, 66% at 17, from 11,000 twin pairs (Haworth 2010), with shared environment falling significantly over the same years and non-shared environment unchanged (Figure 3). Later life: possibly up to 80% (Panizzon 2014), about 60% after 80 (Lee 2010), or no change (McGue and Christensen 2013). Personality shows no systematic change in heritability with age (Turkheimer 2014). A meta-analysis of seven other domains found heritability rising in adolescence and young adulthood for externalising and internalising problems and for social attitudes (Bergen, Gardner and Kendler 2007); no domain showed a decrease |
| 6 | Stability from age to age is mainly genetic; change is mainly environmental | Personality: 80% of phenotypic stability over a decade is genetic (McGue, Bacon and Lykken 1993), confirmed by meta-analyses (Briley and Tucker-Drob 2014; Turkheimer 2014). Intelligence: "genetic amplification" fits better than new genes appearing (Briley and Tucker-Drob 2013; 11,500 pairs). The proposed mechanism is gene–environment correlation: children "select, modify and create environments correlated with their genetic propensities" |
| 7 | Most measures of "the environment" are partly heritable | Average heritability 0.27 across 35 environmental measures in 55 studies (Kendler and Baker 2007). Uncontrollable life events (death of a spouse) show no significant genetic influence; controllable ones do. Parenting in Japan shows more genetic influence than in Sweden (culture moderates heritability) |
| 8 | Links between environment and traits are partly genetic | Home environment and infant mental development: r = 0.44 in biological families against 0.29 in adoptive families. Two-thirds of the correlation between maternal negativity and adolescent antisocial behaviour is genetic (0.40 of 0.61; shared environment 0.16, non-shared 0.05) |
| 9 | Most environmental effects are not shared by children growing up in the same family | For most traits it is genes that make siblings alike; the influential environment is specific to each child. Exceptions: adolescent antisocial behaviour, shared environment about 15% against non-shared about 40% including error (Rhee and Waldman 2002); academic achievement, about 15% (English) and 10% (mathematics) shared, presumably schools (Kovas 2007); intelligence, with shared-environment influence until adolescence that then diminishes (Briley and Tucker-Drob 2013; Haworth 2010). Non-shared effects are probably "many experiences of small effect", possibly chance ("the gloomy prospect") |
| 10 | "Abnormal is normal" | Common disorders are the extremes of continuous traits; exception: severe intellectual disability (IQ < 70) is distinct |
Why the authors think these replicate
Large effects (heritabilities of 30–50%, against under 1% of variance for most sex differences); a focus on variance components rather than single genes or single environmental factors; large samples, consortia and meta-analysis; and a history of controversy that raised the bar.
Limits
- A narrative review of selected findings by four authors of the field's main textbook. Confidence intervals are mostly absent; several numbers come from single studies.
- Twin-design assumptions are acknowledged as criticised but not discussed; the authors rely on twin and adoption designs agreeing.
- Almost nothing on attitudes or values beyond "significant genetic influence" and the rising heritability of social attitudes. No shared-environment figure for attitudes, religion or political views is given.
- Samples are mostly from developed, Western countries; the authors say so.
What it means for Amadeus (inference)
- Heritability does not answer the user's question directly. It partitions differences between people within one population. A trait every human has (a language system, core knowledge) can have low heritability, and a heritable trait says nothing about what everyone shares. What it does say is where person-level variation comes from: about half from genes, little from the family home after childhood, and the rest from each person's own experiences, measurement error included.
- The family environment is weak for dispositions, stronger for content. The shared environment (family, school, neighbourhood: everything siblings share) explains little adult variation in personality, intelligence or most psychopathology. It does explain some school achievement and childhood intelligence, and it fades as people choose their own environments. For the slot: a person's dispositions are mostly not "the culture they grew up in", and mostly not a sum of identifiable life events either.
- Stability is genetic, change is environmental. A person's stable core (personality: 80% of its decade-long stability) traces to their genome; how they change traces to experience. This fits the synthesis's split: a shared rule of change in the base, a stable person-level starting point in the slot.
- "Many small, idiosyncratic experiences" are the non-shared environment's best description. That is why a person's divergences cannot be reconstructed from a list of major life events. It is an argument for dense, personal data (deliberative recall, a lifelog) rather than demographic or biographical summaries.
- Environments are partly chosen. Measured "environments" are 27% heritable on average, because people select and shape them. A replica that carries on should choose its experiences in the person's way, not have them imposed.