Scope of this review

This paper presents a review of the processing of complex stimuli in the auditory system with an emphasis on work carried out by the present authors. The paper will focus on frequency discrimination of complex stimuli on the one hand and on the coding of auditory stimuli in single fibers of the auditory nerve on the other hand. To a certain extent this review reflects the development in time of our ideas and insights on these topics.

Historically there have been two competing views on how complex tones are coded in the auditory system, the spectral and the temporal coding. In one view complex tones are spectrally analyzed (1-3) and tonotopically projected onto the cochlea and the central nervous system. In the other view precise temporal information on the stimuli is coded by means of synchronized occurrences of action potentials in the auditory nerve. These views can be merged into concepts in which both play a role (4,5). In this paper we shall show how partially analyzed stimuli can be precisely coded in various temporal aspects of trains of spikes in single nerve fibers. The discrimination of complex stimuli is often described in spectral models, also called place models (3,6). In such models the stimuli are analyzed spectrally by the auditory system and the system is often described as quasi-linear. Then, the result can be described as the convolution of the stimulus spectrum and (a measure of) the filter shape of the analyzer. Due to the limited analyzing properties, the stimulus components are only  partially resolved. Thus the internal spectrum reflects characteristics of the stimulus spectrum and of the analyzer. More refined models (7-9) also take into account how the filter outputs are coded as a function of time in the trains of action potentials in single auditory nerve fibers. A major part of this paper is devoted to the study of such temporal responses.