Robustness of hierarchical network organization in models of functional connectivity


Ali Safari

28. November 2018, 17.00
WW8, Raum 2.018, Dr.-Mack-Str. 77, Fürth

In this work, we addressed the question, as to which extent we can consider functional (coactivation) networks to be hierarchical as their structural counterparts. We have shown that this is the case as long as the dynamic process generating the functional patterns is not supercritical. Below the critical point, functional networks indeed inherit essential properties of the underlying structural HMNs, namely the small spectral gaps and the exponential degree distributions. This correspondence is lost for supercritical types of dynamics, where we can safely conclude that the resulting functional networks are not hierarchical. Our work, of course, relies on the essential simplifications of the original neuroscience problem. Namely, our simulated dynamics is a simplification of neural activation processes, and our definition of the functional network, constructed here for convenience from a coactivation matrix, differs slightly from the methods used in experiments. Nevertheless, we believe that our results provide useful insights regarding the connection between structure and function in brain networks, and a viable technique to assess the persistence of the hierarchical organization in functional networks.