Inter-Layer Correlation in a Feed-Forward Network with Intra-Layer Common Noise

Abstract

Neural networks generate correlated neural activities. In a multi-layer network, experimental studies have shown that spike correlations appear within a layer and between different layers. It is input common among neurons in each layer that realizes such correlated activities. Theoretical studies have demonstrated that common input given to neurons within a layer, which we call “intra-layer common noise”, generates spike correlation within the layer, which is “intra-layer correlation”, in a feed-forward network. However, it has not been studied whether the common noise can generate spike correlation between different layers, which is “inter-layer correlation”. In this study, we constructed a theory of inter-layer correlation and calculated the theoretical values of the inter-layer correlation in a multi-layer feed-forward network with intra-layer common noise. Our theory revealed that the common noise generates the inter-layer correlation, which coincided with results of simulation.