Representation of spectrum in the auditory system

A problem presented at the UK MMSG Nottingham 2006.

Presented by:
Dr Chris Sumner (MRC Institute of Hearing Research, University of Nottingham)
J Billingham, S Coombes, M Hamadiche, I Pu, C Sumner, C Svensson, R Thul, Y Timofeeva, N Venkov, M Zachariou

Problem Description

The auditory system has to represent sound across 12 orders of magnitude of intensity (120dB). At present we only partly understand how the system is able to function so well across this range. In the brain, problems of representation remain in the representation of broadband spectra. The majority of auditory nerve fibres are highly sensitive and have a dynamic range which only changes in firing rate at very low levels. The dynamic range of the firing rate of these neurons can be as little as 15dB. Thus at normal signal levels (>60dB SPL) almost all auditory nerve fibres are firing at their saturated rate.

The study group was asked how is it that we can recognise sounds and understand speech across the majority of our dynamic range, and to consider a computational model of the auditory periphery coupled to a model of the ventral cochlear nucleus (and in particular so-called chopper cells).

Download the full problem description

Study Group Report

The main achievement of this work is a mathematical exploration of the selective listening hypothesis via an analytical and numerical study of a model dendrite driven by realistic spike train data. Peripheral discussions ranged over mechanical models of the basilar membrane, single neuron coincidence detection, preservation of rate-place codes and network studies of interacting neurons.

Download the full report