Neural bases of monkey call perception

Abstract

We propose a three-layered neural network model for the perception of complex FM signals, such as monkey calls. The first layer is a two-dimensional cochleotopic map with a frequency axis and a propagation axis. Stimulation with a monkey call that is characterized by a complex FM signal activates in a temporal manner the peripheral neurons that are arranged along the frequency axis. The activities of the peripheral neurons then propagate to auditory centers along the propagation axis, whereby a spatiotemporal neuronal firing pattern specific to the monkey call emerges. The second layer decomposes the signal into a series of simple FM components that are then integrated by the third layer network. By stimulating the model with various monkey call signals, we show that the monkey call signals are processed in a hierarchal manner, which is quite similar to visual processing in that information about the monkey calls is decomposed into auditory features (simple FMs) by the lower cortical network, and then integrated into a perceptual image by the higher cortical network.