Decision Device Comparison for Model-based Analysis of ITD Perception in Normal Hearing Listeners

Interaural time differences (ITDs) play an important role in sound localization and speech understanding in noise. Previously (Moncada-Torres et al., 2016), we developed a framework capable of predicting ITD just noticeable differences (JNDs) based on a physiological model of the auditory nerve (AN). However, the decision device employed there did not take into account the variability of the used data in a straightforward manner. In this work, we predicted ITD JNDs using two different decision devices in normal hearing listeners using information at the AN level. AN responses to acoustic stimuli from both ears (in the form of spikes) with and without introduced ITDs were simulated using the phenomenological model proposed by Zilany et al. (2009). Next, we used the shuffled cross-correlogram analysis (SCCs, Joris, et al., 2006) to quantify ITD encoding across the AN of both channels. Assuming that the auditory system is more sensitive to smaller ITDs, we corrected the SCC curves using the weighting function proposed by Stern and Shear (1996). Then, we predicted the imposed ITD by choosing the global maximum of the corrected curves. The distributions of the predicted reference and imposed ITDs were fed to two different decision modules: the receiver operating characteristic (ROC) and the detection index (d'). These allowed us to calculate the ITD JND as the 79.4% and 1.5 point, respectively, of the neurometric curve. Finally, we evaluated the performance of the decision devices' predictions by comparing them against literature behavioural data using pure tones with frequencies from 250 to 1400 Hz. The proposed framework showed similar trends as in psychoacoustical data, with the d' metric being higher correlated with it. Future work will be focused in using the framework's improved pipeline to predict ITD discrimination performance in hearing impaired listeners and well as in optimizing hearing aids/cochlear implants signal processing.