Talk by Gregory Handy as part of SIAM LS20 virtual minisymposium "Network dynamical systems approaches for understanding neuronal behaviour" Abstract: Inhibitory neurons play a key role in many components of sensory processing, including modulating feature selectivity, mediating response suppression, and maintaining an asynchronous network state. Past mathematical models have successfully replicated and furthered our understanding of some of these processes by considering a recurrent network of excitatory (E) and inhibitory neurons. However, experimental evidence has shown that diversity exists within this inhibitory population, with 80% of neurons falling into one of three major subtypes: parvalbumin-expressing (PV), somatostatin-expressing (SOM), and vasointestinal peptide-expressing (VIP) neurons. The differences in how these subtypes are embedded into the cortical circuit are not minor. For example, in the primary auditory cortex (A1) of mice, SOM neurons project and receive inputs on a broader spatial scale than the other inhibitory subtypes. While this observation suggests that the inhibitory subtypes are primed to play particular roles in deciding network dynamics in A1, these exact roles are unknown. Here, we construct a spatial mathematical model to investigate these individualized roles and how they enable key auditory features to arise. We find that the broad spatial projections of SOM neurons drive network suppression that allows the recurrent network to amplify input signals. This in turn allows neurons to selectively respond to directional auditory sweeps, a critical step for many auditory-guided behaviors.
0 Comments