Temporal Information Processing in the Auditory System

The auditory system is an excellent model to study temporal aspects of information processing be-cause of the pronounced temporal structure of communication sounds. In our analysis of information transmission by neurons in the auditory brainstem we compared two neuron types: octopus neurons and Stellate neurons. Stellate neurons faithfully code spectral information in their rate-place contours, similar to primary auditory nerve fibers. With their long membrane time constants (about 7 ms) they integrate auditory nerve inputs over time. Due to this slow membrane time constant their temporal precision, measured by the information rate, decreases rapidly at characteristic frequencies above 400 Hz. Octopus neurons achieve precise temporal precision by using coincidence detection and are able to code information with a precision in the 20 µs range. In general, these neurons are able to transmit much larger information rates than stellate neurons due to their high firing rates of up to 800 spikes/s, their reliable firing and their extremely precise spike timing. For both neuron types, the major portion of information is coded with a temporal precision ranging from 0.2 to 4 ms. This finding is not only interesting for our understanding of neuronal sound processing but has also important implications for automatic speech recognition, where temporal information is usually processed only with a maximal precision of 10 ms. The electrophysiological experiments within this project focused on the mechanisms underlying precise temporal processing, mainly of interaural time differences (ITD).

Dependence of transmitted information rate on temporal resolution for a stellate neuron and an octopus neuron with characteristic frequencies of 1045 Hz. Note that most information is coded with a precision ranging from 0.2 to 5 ms.