Structural and dynamic functional controllability: shared variance between network control metrics limits attribution...

Fluid and crystallized intelligence have recently been hypothesized to reflect the brain network's capacity to facilitate transitions into hard-to-reach, high-energy states, measured via modal controllability, and easy-to-reach, low-energy states, measured via average controllability, respectively. Controllability is typically measured for the structural connectome, but dynamic functional connectivity may better capture temporally varying processes relevant to cognition.

Using data from the Human Connectome Project, we compared the prediction of crystallized and fluid intelligence from controllability metrics measured in both structural and dynamic functional connectivity. We hypothesized dynamic functional connectivity would outperform structural connectivity, and modal and average controllability would differentially predict fluid and crystallized intelligence.

As expected, dynamic functional connectivity outperformed structural connectivity in predicting cognitive performance. However, modal and average controllability displayed similar predictive power for fluid and crystalized intelligence, and performed comparably to regional (or node) strength, a simpler graph metric reflecting the strength of immediate regional connections that strongly covaries with both controllability measures. Furthermore, controllability most strongly predicted general intelligence, the shared factor between crystallized and fluid intelligence. General intelligence was associated with higher average controllability (and greater node strength) in transmodal regions and higher modal controllability (and lower node strength) in unimodal regions in dynamic functional connectivity.

These findings clarify how network control maps onto cognitive ability across the cortical hierarchy, while suggesting that shared variance between controllability metrics hinders specific attribution to crystallized and fluid intelligence, challenging prior hypotheses.