Neural activity sequences arising from spatially correlated asymmetry in EI-networks

Abstract

Sequential neuronal activity is associated with tasks like spatial navigation and encoding memory and time. Recently, it has been shown that correlated spatial asymmetry in recurrent excitatory connections of neurons can generate neuronal activity sequences. In present work, we show that correlated spatial asymmetry in any type of connections (i.e. excitatory (E) to inhibitory (I), I to E, and I to I) is sufficient to generate neuronal activity sequences. We found that when correlated spatial asymmetry is present in inhibitory connections, sequences are slower. In these networks, the ratio of symmetric and asymmetric connections is an important variable that determines whether the network will exhibit sequences or not and determines the stability of the ongoing activity state. On a structural level correlated asymmetric connectivity introduces effective feedforward pathways in the network, and here we demonstrate that such feedforward paths predict and are the causal mechanism behind the emergence of activity sequences. We ask if this network can show a spatially random background activity state, parallel to a biological network.