Model-based Analysis of ITD Perception in Normal & Hearing Impaired Listeners

Interaural time differences (ITDs) are a fundamental cue for sound localization. Sensorineural hearing impaired (HI) listeners frequently have worsened localization and lateralization performance compared to normal hearing (NH) listeners, also reflected in deficient detection and discrimination of ITDs. It has been suggested that ITDs are processed by specialized cells that receive input from the auditory nerve (AN) from both ears and function as coincidence detectors. In this work, we developed a physiologically motivated model framework to evaluate temporal coding of ITDs at the periphery level of NH and HI listeners. AN responses to acoustic stimuli from both ears (in the form of spikes) with introduced ITDs (including a reference condition with ITD = 0 μs) were simulated using the phenomenological model proposed by Zilany et al. (2009). Next, we utilized shuffled cross-correlograms (SCCs, Joris, et al., 2006) to quantify the encoded ITD across the AN of both channels. Assuming that the auditory system favors the perception of smaller ITDs, we corrected the SCC curves using the weighting function proposed by Shackleton et al. (1992). Then, we predicted the imposed ITD by choosing the global maximum. As a decision variable, the distributions of the predicted reference and imposed ITDs were processed to obtain a receiver operating characteristic (ROC) in a 2 alternative force choice (AFC) procedure. This allowed us to compute the ITD just noticeable difference (JND) as the 75 % point of the psychometric curve. Finally, we evaluated the model by comparing the simulated ITD JNDs against literature behavioural data using bandpass noise with a center frequency and a bandwidth of 500 and 100 Hz, respectively. For the HI case, we used the same framework, but modified the AN model to account for different degrees of inner/outer haircell (IHC/OHC) impairment based on audiogram information. The proposed framework successfully predicted bandpass noise ITD JNDs of NH listeners. In the case of HI listeners, the model was able to account for trends in the data, although the high variability of participants' performance make the comparison to data difficult. These results provide the basis for a model-based quantification of ITD coding in NH and HI listeners at the periphery level. Future work will be focused in using the current approach to predict ITD discrimination performance in listeners with acoustic and electric hearing to optimize the representation of spatial information in hearing devices signal processing.status: publishe