Self-organized neuronal subpopulations and network morphology underlying superbursts

Neural bursts are an important phenomenon that needs to be understood for their relevance to many different neurological diseases as well as neural computations. While there are different types of neuronal bursts, in this study we investigate the nature of population (as opposed to intrinsic cell-level) bursts, in particular, superbursts (SBs) that are a small (∼100 ms) packet of several population bursts (PBs). It has been suggested that neuronal PBs occur when there exists a delicate balance of system-wide excitation and inhibition and when recurrent excitation loops exist in the network. However, there has been no rigorous investigation on the relation between network morphology and (super)burst dynamics. Here we investigate the important issue based on a well-established Izhikevich network model of integrate-fire neurons. We have employed the overall conduction delay as our control parameter for tuning network morphology as well as its matching burst dynamics. Interestingly, we found that initially identical neurons self-organize to develop several distinct neuronal subpopulations, which are characterized by different spike firing patterns as well as local network properties. Moreover, a few different motifs of SB emerge according to a distinct mixture of neuronal subpopulations that, on average, fire at slightly different phases. Our analyses suggest that recurring motifs of different SBs reflect complex yet organized modular structures of different subpopulations.