Prefrontal cortical morphometric covariance and cognitive task performance in healthy young adults: an exploratory analysis of Human Connectome Project data

Abstract

The prefrontal cortex (PFC) is implicated in many cognitive tasks, including working memory, reward-based decision-making, executive control, and metacognition. This exploratory analysis investigated whether individual expression of PFC cortical thickness and surface area covariance components is associated with cognitive task performance. The sample comprised Human Connectome Project (HCP) MRI data from 743 healthy adults (age 22–36+). Oblimin-rotated principal component analysis (PCA) was applied to residualized thickness and surface area for 22 PFC regions, controlling for age, sex, and ICV. Component scores were entered as predictors in separate regression models for the Flanker Age-Adjusted Score (inhibitory control), the Gambling RT Difference (RT for the same “larger” guess in reward minus punishment blocks), and 2-Back median RT (working memory speed). Parallel analysis favored two thickness components; a prespecified three-component solution (Global/Lateral PFC, RC1; Orbitofrontal, RC2; Anterior Cingulate, RC3; 53.7% of variance) was retained by decision, and RC1 and RC2 were only moderately stable across bootstrap resamples. RC1 was nominally associated with lower Flanker scores (β = −0.103, p  = 0.027), opposite to the predicted direction, and RC2 with a smaller Gambling RT Difference (β = −0.103, p  = 0.025); neither survived false discovery rate (FDR) correction ( q  = 0.054). In uncorrected models added at review, whole-cortex mean thickness alone robustly predicted Flanker scores but not the Gambling RT Difference; neither component association survived regressing mean thickness out before PCA. In surface area analyses, greater orbitofrontal surface area (AC2) was associated with faster 2-Back RT (β = −0.125, p  = 0.003; q  = 0.032 across 12 surface area tests and 0.064 across all 24 component tests), not explained by a speed-accuracy tradeoff. This unpredicted association persisted in most sensitivity analyses but was attenuated and non-significant when total surface area was regressed out before PCA (β = −0.065, p  = 0.078). In conclusion, no prefrontal-specific association was robust. Global cortical thickness accounts for the Flanker association more parsimoniously, the Gambling RT Difference association was not robust under either account, and the orbitofrontal surface area association could not be separated from global surface area; all associations require replication.