Publicly available code for "A flexible hub connectivity mechanism for cognitive control". Manuscript in-preparation. Part of C. Cocuzza's dissertation (Aim 3).
Abstract
Neurocognitive processes are embedded in the human brain’s remarkably networked systems. While decades of research have demonstrated that there are functionally-relevant associations between brain activity and brain connectivity, the extent to which information processing is established in network interactions remains unclear. The overarching hypothesis of this dissertation is that network interactions are key to specifying processes that are exhibited locally, as well as specifying flexible processes that are distributed across the cortex. Across all studies, I assessed functional magnetic resonance imaging data from large cohorts of healthy young adult humans. I first tested the specific hypothesis that connectivity patterns in the frontoparietal (FPN) and cingulo-opercular networks (CON) exhibit properties consistent with dissociable yet complementary roles for cognitive control. While prior work demonstrated that FPN regions could flexibly shift their global connectivity patterns across task states, a similar characterization of CON connectivity patterns was not fully established. I found evidence that, in contrast to FPN, CON regions transiently switched their network affiliation in a task-relevant manner. I further demonstrated that only connectivity patterns in the FPN and CON could reliably decode task rules, suggesting that their connectivity-based properties were essential for representing task information. Second, I tested the specific hypothesis that local, functionally-relevant processes could be sufficiently generated by distributed network interactions. I leveraged a technique that models the generation of task-evoked activations in held-out brain regions by explicitly parameterizing activity flowing over an intrinsic connectivity architecture. I applied this approach to visual cortex regions that are well-known to exhibit selectivity for specific visual categories. Evidence that distributed network interactions play a prominent role in the generation of local processes generalized across four models of visual category selectivity. Third, I tested the specific hypothesis that across the cortex, task-set activations occurring with rapidly shifting task demands are coordinated by network interactions exhibiting flexible hub connectivity. This flexible hub connectivity mechanism demonstrated how network interactions are key to implementing flexible, context-dependent processes fundamental to cognitive control. The works in this dissertation provide novel empirical evidence that advances our understanding of how brain network mechanisms underlie local and distributed neurocognitive processes.
Correspondence: Carrisa V. Cocuzza ([email protected]), The Cole Lab (http://www.colelab.org/)