Modelling the spectral-edge sensitivity of primary auditory cortex neurons

The spectral edge of a sound serves as an important cue in human sound perception. A recent physiological study revealed that there are four types of neurons in the primary auditory cortex (Al) based on the edge sensitivities of the cells. In this study, in order to explain the behavior of Al neurons, we have developed an auditory model that consists of peripheral auditory functions and weighted summation of the peripheral responses. The peripheral auditory system was modeled using a cochlear filter bank as well as compression characteristics. Neural weighting was represented with a combination of excitatory and inhibitory weightings in which the weighting function was approximated with a normal distribution. The weighed summation of the outputs from the model of the auditory periphery was assumed to be the discharge rate of an A I cell. To achieve the best fit to the physiological results, a simulation to find appropriate weighting values was conducted. As a result, the physiological data were successfully replicated using the proposed model. Because edge sensitivity is also important for vowel discrimination, the proposed model can be used as a preprocessor for an automatic speech recognition system.