Date: Tue, March 13, 14:00-16:00 Place: Room Dw601, D Block, IIS, The University of Tokyo Invited Speaker: Dr. Kosuke HAMAGUCHI (Université-René Descartes (Paris V)) Title: Inhibition-dominated network can explain multistability and highly irregular states in prefrontal cortex activity Abstract: The spike patterns of neurons in vitro in response to current injection is typically very regular, however in vivo, from the sensory to prefrontal cortex, the degree of spiking irregularity is known to be high. In the prefrontal cortex, neurons seem to switch from a background state to an active firing state during delayed response tasks and their spikes trains are highly irregular in both types of states. Although it is easy to obtain self-sustained states (bistability) with a connectivity dominated by excitation, CV of spike trains typically decreases when the firing rate increases. One robust mechanism to obtain high spike irregularity is to balance the excitation and inhibition in recurrent connections. In balanced or inhibition dominated networks, firing patterns can be irregular depending on the parameters of the network, however it is usually difficult to obtain bistability because the mean input to the neuron is not depolarizing. Here it remains unclear how to achieve simultaneously high firing rate and high spike train irregularity. Here we focus on the possible firing states in a spatial working memory model. More specifically, we investigate the dynamics of sparsely interconnected spiking neurons embedded in a ring structure. We construct a self-consistent theory of the macroscopic variables of the system, and calculated the stable states of the system. The activity of recurrent network of spiking neurons have been studied by the mean-field approach. We found that the network shows various patterns of activity, e.g., traveling wave, oscillating wave, standing wave, uniform activity states, and quiescent state. In the seminar, I report that spatially localized activity (bump) states can exist in inhibition dominated regions with sub-threshold external input. In these region, multistability can be achieved thanks to fluctuations of the recurrent input which provides localized positive feedback. Similar fluctuation driven multistability might be relevant to the spontaneous activity in the primary visual cortex.