While our results exhibited a clear relationship between node connectivity and behavior, it is reasonable to speculate that the average level of activation of each node could also contribute to, or even confound, this relationship. For example, nodes that have high levels of activation could potentially influence the correlation of the BOLD parameter estimates, thus potentially explaining their significant functional interactions in the network. To address this idea, we correlated node degree and average parameter estimates (the average of parameter estimates across participants for each ROI) and found no significant correlation (ρ = −0.01, p = 0.9614). In addition, adding the average parameter estimate for each node to the stepwise regression between node degree and behavioral performance contributed no significant variance to the overall model. These analyses suggest that univariate effects alone are unlikely to be driving our results.
