Predicting binaural lateralization and discrimination using the position-variable model

Abstract

The position-variable model was developed as a means to characterize and predict a variety of binaural lateralization, discrimination, and detection phenomena. The model was motivated by a desire for a more complete understanding of the putative mechanisms by which interaural time and intensity differences were combined, as well as the extent to which results in interaural discrimination and binaural detection experiments are mediated by cues based on subjective lateral position. This paper will describe recent efforts to unify and extend the predictions of the model, as well as to develop a publicly accessible version of the model within the framework for comparing evaluating binaural models described by Dietz et al. in this session. Predictions of the model are based on computation of the centroid along the internal-delay axis of the patterns of activity of the display of information proposed earlier by Jeffress and Colburn, derived from the auditory-nerve response to the experimental stimuli. Some of the issues to be discussed include comparisons to other proposed methods of developing lateralization predictions, the impact of internal versus external noise in the model’s predictions, and specific issues involved with modifying the model to render it compatible to the common framework for model comparison.