Why do some toddlers bound through a room full of strangers while others cling to a parent or become easily upset? A large international study published in Nature Human Behaviour provides evidence that common variations in DNA are linked to these early behavioral tendencies during the first three years of life. The research also indicates that the genetic variants linked to toddler temperament overlap with those linked to adult personality, cognition, attention-deficit/hyperactivity disorder (ADHD), and autism.
During the first three years after birth, the human brain grows more rapidly than at any other stage of life as children learn to move, express emotions, and build relationships. Young children also display distinct behavioral styles known as temperament, defined as early-developing personality traits governing how a child reacts to their surroundings.
One widely used psychological model divides early temperament into four traits: emotionality, activity, shyness, and sociability. Emotionality reflects how easily and intensely a child experiences negative feelings like anger or fear, while activity captures physical energy and movement tempo. Shyness describes feeling tense or inhibited around strangers, whereas sociability reflects the desire to interact with others.
Observational research shows these early traits often foreshadow later behavioral patterns. For instance, a 2018 study of infants co-authored by Mark H. Johnson, one of the new study’s authors, followed children, many with an older sibling with autism, into mid-childhood. High activity and poor self-control in toddlerhood were linked specifically to later ADHD symptoms.
Don’t miss the next big psychology finding
Every morning, PsyPost sends a short briefing on the most interesting new peer-reviewed research in psychology and neuroscience.
Early shyness and fearfulness were linked to later anxiety, but also to autism symptoms, and the shyness link faded once family history of autism was accounted for. Similarly, a 2021 meta-analysis of 136 studies found that differences in early development and behavior in the first five years were linked to a higher likelihood of later ADHD.
Twin and family studies estimate that genetics account for 30% to 60% of the variation in these four temperament traits among toddlers, but cannot pinpoint where in the DNA those differences lie. As two of the new study’s authors, Angelica Ronald and Anna Gui, noted in a 2024 review, very few studies have scanned the genome for specific variants linked to infant behavior. Lead author Anja Hollowell of Birkbeck, University of London, senior author Angelica Ronald of the University of Surrey, and their colleagues designed the new study to identify common DNA variants associated with toddler temperament and test whether they overlap with later outcomes.
The researchers preregistered their study design on the Open Science Framework and posted their protocols and summary statistics publicly, though the analyses averaging across the two ages, which produced several of the main results, were added beyond the preregistered plan. They combined five European cohorts: the Norwegian Mother, Father and Child Cohort Study, the Avon Longitudinal Study of Parents and Children in England, the Twins Early Development Study in England and Wales, the Dutch Lifelines cohort, and the Netherlands Twin Register.
Parents or caregivers rated their children’s emotionality, activity, shyness, and sociability during their second year (18 to 24 months) and third year (36 to 38 months). The researchers analyzed each age separately and as a cross-age average, across 43,963 to 72,663 European-ancestry children. The Norwegian cohort supplied roughly 39,000 to 55,000 of these children. Some cohorts used the standard temperament questionnaire, while others built matching scales from items in other questionnaires, and Lifelines parents rated their children retrospectively, years later.
Using genome-wide association analyses adjusted for factors such as age, sex, and family relatedness, the team tested millions of common single-letter DNA variations for links to each trait. Estimating how much of the variation in each trait tracked with common DNA differences across the genome, the researchers found that common genetic variation accounted for about 15% of the differences in cross-age shyness, 10% in activity, 7% in emotionality, and 3% in sociability. Genetic patterns were stable between the second and third years, and analyses run separately for boys and girls indicated that mostly the same variants are linked to temperament in both.
Across all analyses, the team identified 10 independent genomic regions, or stretches of DNA, that met the strict threshold for a genome-wide association with emotionality, activity, or shyness, while none reached the threshold for sociability. Aggregating variants within whole genes identified about 20 distinct temperament-associated genes, nearly all linked to activity. Comparing these against a Genomics England clinical database indicated that temperament-linked genes were enriched for established autism genes at more than ten times the database’s base rate, though this rests on just two genes.
To examine how the 10 genomic regions might operate in the brain, the scientists used colocalization analysis. This tested whether the same DNA variant is linked to both toddler temperament and gene expression, meaning how actively a nearby gene is being used, in post-mortem tissue from 1,433 adult cerebral cortex samples. Two regions showed a high probability of sharing an underlying variant with brain gene expression.
The genetic signal for third-year activity on chromosome 12 had an estimated 93% probability of sharing an underlying variant with a signal for RHEBL1 expression in the cortex. The version of the variant tied to lower toddler activity was tied to higher RHEBL1 expression, and that same variant has previously been linked to more years of schooling and higher cognitive test scores. The protein produced by RHEBL1 helps regulate a cellular signaling pathway involved in brain cell growth and connections.
In the second match, a chromosome 1 signal linked to lower cross-age emotionality had a 99% probability of sharing an underlying variant with increased cortical expression of MR1, a gene not previously linked to behavior. It produces a protein that presents microbial byproducts to specialized immune cells, which the authors note fits with evidence linking the immune system to brain development and behavior. The brain tissue came from adults rather than toddlers, and they add that laboratory follow-up is needed to test how both genes work.
Next, the researchers estimated genome-wide genetic correlations, which measure how much two traits share the same genetic variants across the genome. Among the four toddler traits, shyness and sociability showed a strong negative genetic relationship, while moderate negative overlap emerged between emotionality and sociability and between activity and shyness. Emotionality and shyness shared a modest positive link, whereas correlations between activity and emotionality or sociability were not statistically significant.
Comparing early temperament against 17 other traits, from age at first walking to adult personality, the team found widespread genetic overlap with neurodevelopmental, psychiatric, cognitive, and personality outcomes. Variants linked to higher toddler activity showed a moderately strong positive association with ADHD, a strong link to walking at an earlier age, and a moderate negative association with adult educational attainment and cognitive test performance.
Variants linked to higher toddler sociability were moderately associated with a lower likelihood of autism and higher adult extraversion, while variants linked to higher shyness were associated with lower extraversion, a lower likelihood of ADHD, and walking slightly later. Variants tied to higher emotionality showed modest positive overlap with autism, adult neuroticism, depression, and schizophrenia. Genetic links with lifetime anxiety disorder, bipolar disorder, conscientiousness, and agreeableness were not statistically significant. That includes toddler shyness, which showed no statistically significant genetic overlap with anxiety disorder, even though observational studies have linked early shyness to later anxiety.
Follow-up modeling estimated that 86% to 97% of the common variants linked to toddler shyness overlap with those linked to ADHD, extraversion, and age at walking. However, many of these shared variants were associated with the two traits in opposite directions, so overlap does not mean the variants relate to both traits in the same way.
Polygenic scores add up many DNA variants into a single score per child, with each variant weighted by how strongly it was associated with the trait in the main analysis. These scores predicted parent-rated temperament in cohorts left out of the discovery step, accounting for 0.06% to 1.9% of behavioral differences. In fraternal twin comparisons, differences between within-family and between-family genetic predictions were not statistically significant, suggesting the signals were not heavily distorted by shared family environments, though the modest predictive power of the scores limits firm conclusions.
To test whether the findings extended beyond European populations, the team ran meta-analyses adding two cohorts assessed with the Child Behavior Checklist: 5,618 to 5,653 Japanese-ancestry children from Japan’s Tohoku Medical Megabank Birth and Three-Generation Cohort Study, and 595 to 843 Black African and mixed-ancestry children from South Africa’s Drakenstein Child Health Study. In this combined sample of 56,083 to 78,894 children, the chromosome 12 variant near RHEBL1 remained associated with third-year activity, and multi-ancestry polygenic scores performed similarly to European-trained scores when tested in European children.
These findings align with an earlier genetic study of seven-month-old infants, which found that babies carrying specific genotypes of the CD38 gene displayed higher cuddliness and touch-seeking behavior than infants with a different genotype. Both studies connect genetic variation to early social-emotional temperament, although the earlier work focused on a single gene related to physical affection at seven months rather than genome-wide associations in toddlers.
As with all research, there are a few caveats to keep in mind. Common DNA variants accounted for 3% to 15% of the variation in temperament, and polygenic scores explained less than 2% of the differences in temperament across children in the samples tested, meaning that genetic links are probabilistic and do not lock in a child’s personality or guarantee a future diagnosis. A child’s score also depends on the population and statistical model used to build it, and could change if either were different. In addition, the 10 genomic regions were not replicated in an independent sample, the cross-age analyses were not preregistered, and the gene-expression analyses relied on post-mortem brain tissue from adults rather than toddlers.
The common-variant heritability estimates were also lower than the 30% to 60% estimates from twin studies. For sociability in particular, the authors suggest the low estimate could reflect differences in the size of a child’s social world, such as how many siblings they have, or measurement error in the sociability questions. Temperament was measured by parent report, which the authors note can be colored by parents’ own views. Parent ratings of twins can also exaggerate differences between fraternal twins, which can push twin-based estimates upward.
Finally, more than 90% of the participants had European ancestry, and fewer than 6,500 children came from the Japanese and South African cohorts, which used a different questionnaire. The authors call for larger multi-ancestry samples, laboratory studies of genes like RHEBL1 and MR1, and research on how early environments interact with these genetic associations.
The study, “Genome-wide association studies of infant and toddler temperament in European and multi-ancestry populations,” was authored by Anja Hollowell, Anna Gui, Emilie Wigdor, Morgan J. Morgan, Laurie J. Hannigan, Elizabeth C. Corfield, René Pool, Susanne Bruins, Helga Ask, Christel M. Middeldorp, Beate St Pourcain, Meike Bartels, Dorret I. Boomsma, Catharina A. Hartman, Aoi Noda, Ippei Takahashi, Mami Ishikuro, Taku Obara, Shinichi Kuriyama, Mary S. Mufford, Marilyn T. Lake, Dan J. Stein, Heather J. Zar, Nadia Hoffman, Elise B. Robinson, Anders D. Børglum, Xinhe Zhang, Varun Warrier, the Autism Spectrum Disorder Working Group of the Psychiatric Genomics Consortium, Tomoki Arichi, Mark H. Johnson, Frank Dudbridge, Stephan J. Sanders, Alexandra Havdahl, and Angelica Ronald.
