Abstract
Previous work has suggested that genetic confounding is a persistent issue in studies of environmental predictors of executive function (EF). This is largely because controlling for genetic confounding typically requires specialized samples such as twins or adoptees, which are more difficult to recruit. Polygenic scores provide a potential alternative control, scalable to smaller samples and not requiring specialized sample features. The purpose of this study was to determine if polygenic scores of EF could be used to replicate the findings of other genetic confounding studies in a less specialized sample. Confounding models showed evidence for genetic confounding in maternal education, although it was far weaker in magnitude than in other genetically informed studies. However, consistent with previous research, there were no detectable influences of indirect genetic effects on the EF polygenic score, indicating that the detected genetic confounding was likely a true genetic effect. Finally, while environmental factors other than maternal education seemed predictive of EF, confounding models showed that this was best explained by their association with maternal education. Other predictors of EF may thus be confounded environmentally, not just genetically. While polygenic scores are a promising method with a multitude of applications, in their current state they do not replicate the findings of other genetically informed studies of EF. Caution should thus be used when employing them to study genetic confounding in EF.
