A network approach to fMRI condition-dependent cognitive activation studies as applied to understanding sex differences

Abstract

Network approaches to analysis of functional neuroimaging data provide a powerful means with which to understand the complex functioning of the brain in health and disease. To illustrate how such approaches can be used to investigate sex differences in neurocognition, we applied the multivariate technique of Principal Components Analysis (PCA) to an fMRI dataset obtained during performance of mental rotation – a classic visuospatial task known to give rise to sex differences in performance. In agreement with prior results obtained using univariate methods, PCA identified a core mental rotation network (principal component [PC]1, accounting for 53.1% of total variance) that included activation of bilateral frontal, parietal, occipital and occipitotemporal regions. Expression of PC1 was similar in men and women, and was positively correlated with level of education. PC2, which accounted for 5.7% of total variance, was differentially expressed by men and women, and indicated greater mental rotation-associated neural activity in women in such high-order cortical regions such as prefrontal cortex and superior parietal lobule, in accord with prior findings, and with the idea that women may take a more “top-down” approach to mental rotation. By quantifying, in a data-driven fashion, the contribution of factors such as sex and education to patterns of brain activity, these findings put the magnitude of neural sex differences during mental rotation into perspective, confirming the commonsense notion that, as humans, men and women are more alike than they are different, with between-individual variability (such as level of education, which, importantly, is modifiable) generally outweighing between-sex variability. This work exemplifies the role that multivariate analysis can play in identifying brain functional networks, and in quantifying their involvement under specific conditions and in different populations.