Analog very large-scale integrated (VLSI) implementation of a model of amplitude-modulation sensitivity in the auditory brainstem

An analog very large-scale integrated (VLSI) implementation of a model of signal processing in the auditory brainstem is presented and evaluated. The implementation is based on a model of amplitude-modulation sensitivity in the central nucleus of the inferior colliculus (CNIC) previously described by Hewitt and Meddis [J. Acoust. Soc. Am. 95, 2145-:2159 (1994)]. A single chip is used to implement the three processing stages of the model; the inner- hair cell (IHC), cochlear nucleus sustained-chopper, and CNIC coincidence- detection stages. The chip incorporates two new circuit: an IHC circuit and a neuron circuit. The input to the chip is taken from a 'silicon cochlea' consisting of a cascade of filters that simulate basilar membrane mechanical frequency selectivity. The chip which contains 142 neurons was evaluated using amplitude-modulated pure tones. Individual cells in the CNIC stage demonstrate bandpass rate-modulation responses using these stimuli. The frequency of modulation is represented spatially in an array of these cells as the location of the cell generating the highest rate of action potentials. The chip processes acoustic signals in real time and demonstrates the feasibility of using analog VLSI to build and test auditory models that use large numbers of component neurons.