Cognitive-Behavioral Predictors of Individual Variability of Functional Connectivity in Healthy Young Adults

While stable patterns of fMRI task-evoked brain activity and functional connectivity (FC) exist at the population level, a growing body of research emphasizes that variability exists across individuals. These differences define the critical idiosyncrasies in cognition and behavior across individuals that make individuals unique. Resting-state fMRI data (60 minutes) were examined from 1012 participants from the HCP dataset of healthy adults between the ages of 22 and 37. Functional connectivity was estimated between 360 regions, and variability was defined by each individual's mean correlational distance (MCD) to all other participants. High MCD indicated a more 'idiosyncratic' connectivity pattern deviating from the group pattern. Hierarchical regression was used to determine predictors of variability in FC. The base model (demographics, sleep, sex, brain volume) explained 9.22% of the variance in heterogeneity in functional connectivity. Increased variance was explained by cognition, squared cognition, and NEO personality scores, while emotional scores and fitness explained no additional variance. The final model explained 11.9% of the variance in MCD. Low MCD (i.e., being closer to average) was associated with higher BMI, greater crystalized cognitive scores, more positive emotional valence, and NEO Agreeableness. Greater variability was associated with age, brain volume (potentially a sex difference), and NEO Extroversion. The model underestimated variability in the highest MCD participants, suggesting unexplained factors in highly variable individuals. Differences were observed between males and females, and monozygotic twins showed similar variability, suggesting a genetic component. These results suggest benefits for a connectivity pattern being more similar to the group average.