Effects of topology on the controllability of brain connectomes through sparsity promoting control

The fundamental mechanisms underlying the brain's ability to switch between dynamic (or physiological) states in response to cognitive demands are elusive, and have not been systematically correlated with the topology of neural circuits, particularly in underdeveloped brains. We used a sparsity promoting closed-loop control framework, large datasets of resting-state connectomes from early adolescents and synthetic graphs, to investigate the role of graph topology on regional (node) controllability and control action on the connectome. Feedback costs were examined in ranges corresponding to nodes becoming self-controlled, losing their control action, or remaining self-controlled. Their associations with node connectedness and strength, and network modularity, fragility and resilience were assessed. Highly connected nodes that were central to the network became self-controlled and maintained their control action on the network under high feedback cost, suggesting that brain regions with such properties may play critical roles in the connectome's controllability. In addition, nodes in more modular, fragile and less resilient networks were self-controlled under overall higher feedback costs.