The effect of local inter-inhibitory connectivity on the dynamics of an activity-dependent neuronal network growth model

Abstract

The balance between excitation and inhibition in a neuronal network is considered to be an important predictor of neural excitability. Various processes are thought to maintain this balance across a range of stimuli/conditions. However, the developmental formation of this balance remains an open question, especially regarding the interplay between network blue-print (the spatial arrangement of excitatory and inhibitory nodes) and homeostatic processes. In this paper, we use a published model of activity-dependent growth to show that the E/I ratio alone cannot accurately predict system behaviour but rather it is the combination of this ratio and the underlying spatial arrangement of neurones that predict both activity in, and structure of, the resulting network. In particular, we highlight the particular role of clustered inter-inhibitory connectivity. We develop a measure that allows us to determine the relationship between inter-inhibitory connectivity clustering and system behaviour in an exhaustive list of spatial arrangements with a given fixed number of excitatory and inhibitory neurones. Our results reveal that, for a given E/I ratio, networks with high levels of inter-inhibitory clustering are more likely to experience oscillatory behaviour than networks with low levels, and we investigate the network attributes which characterise each global behaviour type produced by the model. We identify possible approaches for extensions of the current work, and discuss the implications these results may have on future modelling studies in this field.