Interaural Correlations Between +1 and −1 on a Thurstone Scale: Psychometric Functions and a Two-Parameter Model

Abstract

This contribution investigates how the interaural correlation (IAC/r) is represented in terms of perceptual units, to understand how the binaural auditory system separates meaningful sound sources from diffuse ambient noise. Psychometric functions for the discriminability of narrowband noise signals with different IAC were obtained from pairwise comparisons in a novel 2-pair-AFC-paradigm. The nine reference correlations covered the whole parameter range between -1 and +1.

The shapes and slopes of the psychometric functions critically depended on the reference IAC, and IAC discrimination thresholds were inconsistent with the predictions of signal detection theory (SDT) if the normalized IAC was used as a decision variable. Therefore, a Thurstone model was applied to transform the stimulus IAC into a decision variable T(r) which is, in contrast to the normalized IAC itself, characterized by standard normal distributions in the entire parameter range and thus compatible with SDT. In good approximation, T(r) is proportional to the difference between the dB-scaled energies in the correlated vs. the anticorrelated signal components after noisy peripheral preprocessing. T(r) allows one to predict the percentage of correct discriminations for arbitrary pairs of IAC with a simple analytical two-parameter-formula.