Poster B79, Sunday, March 25, 8:00-10:00 am, Exhibit Hall C
The specificity and robustness of long-distance connections in weighted inter-areal structural brain networks
Richard Betzel1, Danielle Bassett1; 1University of Pennsylvania
The functional repertoire of a brain area depends upon the configuration of its incoming and outgoing connections. The complete set of these connections defines a connectome, which can be represented as a network and analyzed using tools from network science. Among the most salient features of brain networks is their cost efficient spatial embedding, which results in disproportionately many short-range connections. Nonetheless, brain networks exhibit a small number of costly long-distance connections, which are thought to confer functionality to neural systems by enabling efficient signaling and transfer of information by reducing the average number of processing steps separating any two brain areas. However, in real-world networks connection weights are log-normally distributed and weaken monotonically as a function of distance. This observation implies that long-distance connections play a minor role in the network's shortest path structure, suggesting a diminished capacity for promoting efficient inter-areal communication. What then is the functional role of long-distance connections? Using a network analytic framework, we provide evidence suggesting that long-distance connections confer connectional specificity to brain areas and in the process define its functionality. Using four network datasets (human, macaque, drosophila, and mouse), we show that long-distance connections are also highly reinforced and redundant, suggesting that long-distance connections are resilient to most perturbations. However, using dynamical simulations, we show that in their absence the complexity of spontaneous neural activity is dramatically reduced. These findings help clarify the functional roles of costly long-distance architectural features and inform future studies of inter-areal network structure and function.
Topic Area: METHODS: Neuroimaging