A parsimonious, age-invariant, and psychometrically validated neural signature of task-control network functioning

Abstract

Neuroimaging has vast potential for identifying brain features that support individual and developmental differences in cognition, but challenges related to the dimensionality, unreliability, and unknown developmental properties of neuroimaging measures prevent this from being realized. We overcome these challenges by applying latent variable modeling to demand-related responses in the "task-control" frontoparietal and dorsal attention networks (FPN, DAN) within the Adolescent Brain Cognitive Development Study and Human Connectome Project. Task-control activations are well-captured by a bifactor structure with a general factor spanning FPN/DAN and network-specific factors that is remarkably invariant across middle childhood to early adolescence and largely generalizes to adulthood. Trait-state-occasion modeling reveals a general task-control trait that shows robust associations with cognitive abilities, clinical symptoms, and grades. Repeated neuroimaging substantially increases associations. Findings establish a parsimonious neural signature that generalizes across samples, is stable across development, and has broad utility for understanding developmental and individual differences in goal-directed behavior.