Schedule of Events | Symposia

Linking cognitive domains to static and dynamic models of brain network controllability

Poster Session B - Sunday, March 30, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom

Justin Ng^1,2 (jwk.ng@mail.utoronto.ca), Jamie Feusner^1,2,3, Colin Hawco^1,2; ¹Centre for Addiction and Mental Health, ²University of Toronto, ³Karolinska Institutet

Fundamental cognitive domains essential for academic and professional success, including fluid (gF; novel reasoning) and crystallized intelligence (gC; knowledge application), are hypothesized to relate to characteristics of brain-wide connectivity. Borrowing from network control theory, “modal controllability,” reflecting the capacity to drive the brain into high-energy states, may align with gF - assembling novel representations with high cognitive effort. Conversely, “average controllability,” reflecting the capacity to drive the brain into low-energy states, may align with gC - retrieving established representations with low cognitive effort. gF and gC are comprised of general intelligence (g; shared variance across all tests) and domain-specific components (gF-g and gC-g). Brain connectivity can be modelled using structural connectivity (SC; white matter estimate), static functional connectivity (sFC; time-averaged), and dynamic functional connectivity (dFC; time-varying). Using diffusion and resting-state functional MRI data from the Human Connectome Project (n=945), we applied kernel ridge regression to examine associations between gF, gC, and their components (g, gF-g, gC-g), and controllability measured using SC, sFC, and dFC. We found that modal and average controllability predicted each cognitive domain with similar accuracy, g exhibited the strongest relationship with controllability, and dFC-based controllability demonstrated the strongest predictive power across cognitive domains. These findings reveal that controllability may primarily relate to g, and controllability based on time-varying connectivity models like dFC may offer greater cognitive relevance than static connectivity models like SC and sFC. Importantly however, the hypothesized distinction between modal and average controllability in predicting gF and gC was not strongly supported by our results.

Topic Area: THINKING: Other