Socioeconomic status and polygenic propensity to cognition have unique and interactive associations with cognitive performance during middle childhood

Socioeconomic status (SES) and genetic propensity have been reliably associated with cognitive ability. Pseudo-experiments showing that income supplementation enhances educational attainment supports a putative causal effect of SES on cognition. What remains poorly understood is the extent to which genetic propensity and SES have shared, unique, and/or interactive associations with cognition, which has important implications for policy decisions. We estimated whether SES and polygenic propensity to cognition are independently and/or interactively associated with cognitive performance in 5,556 9–10-year-old children of European genomic ancestry who completed the baseline session of the Adolescent Brain and Cognitive Development (ABCD) Study®. Cognition: Principal components reflecting general ability, executive function, and learning/memory were derived from performance on NIH Toolbox cognitive assessments. Fluid intelligence was measured by the Matrix Reasoning subtest on the WISC. SES: Familial and neighborhood socioeconomic status were assessed by parent report of gross annual household income and census tract-associated neighborhood deprivation, respectively. PGS: Discovery GWASs of educational attainment (EDUPGS; n=766,345; Lee et al., 2018), executive function (EFPGS; n=427,037; Hatoum et al., 2022), and IQ (IQPGS; n=269,867; Savage et al., 2018) were used to generate polygenic scores (PGS). Analyses: Mixed effects models nested by site and family were used to estimate whether SES indices and polygenic propensity to cognition have unique and interactive associations with cognitive performance (i.e., general ability, executive function, learning/memory, fluid intelligence) with cross validation. Multiple testing was corrected for using FDR. Covariates included sex, age, and the first 10 ancestral principal components; interaction models included predictor x covariate interactions. All SES (i.e., familial, neighborhood) and polygenic cognition (i.e., EDUPGS, EFPGS, IQPGS) indices showed expected associations with all 4 cognitive performance metrics when examined independently, except for EDUPGS and measured EF as well as neighborhood deprivation and learning/memory, which only displayed a marginal trend (SES: Bs≥0.06, Ps≤0.001; Cognition PGS: Bs≥0.05, Ps≤0.009). In simultaneous models, familial income (Bs≥0.05, p≤0.01) and all Cognition PGS aside from EDUPGS and IQPGS with measured EF (Bs≥0.08, p≤1.5e-06), but not neighborhood deprivation (Bs≤0.04, Ps≥0.05) were associated with unique variance in general cognitive ability, executive function, learning/memory. A neighborhood deprivation x EDUPGS interaction revealed that the association between high EDUPGS and general cognitive ability was most pronounced at higher levels of neighborhood disadvantage. Familial income and genomic propensity to cognition had non-overlapping associations with cognitive performance during middle childhood. In addition to leveraging genomic data to better parse the etiology of differential cognition, this work also emphasizes the continued need to consider public policy solutions combating childhood poverty. While associations between SES and cognition are modest in magnitude and comparable to polygenic associations, SES-related associations with cognition may plausibly reverberate to related variables (e.g., health) and traverse generations for a broader impact.