27. Adaptive processing of interaural level differences

Johannes C. Dahmen johannes.dahmen@physiol.ox.ac.uk Andreas Schulz andreas.schulz@physiol.ox.ac.uk Peter Keating peter.keating@physiol.ox.ac.uk Andrew J. King andrew.king@physiol.ox.ac.uk

Department of Physiology, Anatomy and Genetics, Sherrington Building, University of Oxford, Oxford, UK

Adaptation phenomena have been reported in all sensory systems and can provide valuable insight into mechanisms of sensory coding. To examine how the mammalian auditory system adapts to the statistics of binaural spatial cues and how adaptation at the neuronal level may translate into perceptual phenomena we recorded from neurons in the inferior colliculus of anaesthetized ferrets and carried out behavioural experiments in human listeners using almost identical stimuli. Stimuli consisted of a combination of dynamic adaptors and static test stimuli. The adaptors consisted of five seconds long sequences of broadband noise with dynamically varying interaural level differences (ILDs). The ILD values of the adaptor were drawn randomly from a Gaussian distribution and changed every 5ms. Each test stimulus consisted of a 100ms long noise burst with a static ILD and was presented immediately after the adaptor.

Employing reverse-correlation we characterized each neuron˘2  019s activity during adaptation in terms of a linear filter and a nonlinear gain function. Responses to the static test ILD were used to obtain more traditional ILD-response functions. During behavioural experiments human listeners were required to lateralize the static test ILDs in a two-alternative forced choice task.

Neuronal adaptation to changes in the mean or variance of the distribution from which ILD values were chosen was closely matched by the perceptual effects observed in human listeners. Filter shape mostly remained constant across different means and ILD-response functions shifted in the direction of the mean of the adaptor’s distribution. Parallel changes were observed in human perception as lateralization was biased away from the mean of the adapter. Decreasing the variance of the adaptor’s distribution was associated with an increase in neuronal gain and steeper ILD-response functions. The psychophysical experiments provided evidence for a perceptual correlate of this increase in neuronal sensitivity as lateralization thresholds improved with decreasing variance of the adapter.