Can machine learning-based predictive modelling improve our understanding of human cognition?

Abstract

AbstractA growing body of research predicts individual cognitive ability from brain characteristics including functional brain connectivity. Most of this research aims for high prediction performances but lacks insight into neurobiological processes underlying the predicted concepts. Here, we encourage designing predictive modelling studies with an emphasis on interpretability to enhance our understanding of human cognition. As an example, we investigated in a preregistered study which functional brain links successfully predict general, crystallized, and fluid intelligence of 806 healthy adults (replication:N=322). The choice of the predicted intelligence component as well as the task during which connectivity was measured proved crucial for better understanding intelligence at the neural level. Further, partially redundant, system-wide functional characteristics better predicted intelligence than connectivity of brain regions proposed by established intelligence theories. In sum, our study showcases how future predictive studies on human cognition can enhance explanatory value by prioritizing comprehensive outcomes over maximizing prediction performance.Significance StatementOur preregistered study “Can machine learning-based predictive modelling improve our understanding of human cognition?” builds on the lack of conceptual insights into the neural underpinnings of human behavior and thought despite the considerable surge in the number of published predictive modelling studies. Exemplarily, we demonstrate how predictive modelling can be applied strategically to enhance our understanding of general intelligence – a hallmark of human behavior. Our study unveils crucial findings about intelligence, e.g., it suggests differences in the neural code of distinct intelligence facets not detectable on a behavioral level and a brain-wide distribution of functional brain characteristics relevant to intelligence that go beyond those proposed by major intelligence theories. In a broader context, it offers a framework for future prediction studies that prioritize meaningful insights into the neural basis of complex human traits over predictive performance.