Effects of level roving and overall level on correlation change discrimination in naïve and trained listeners

Abstract

Sensitivity to interaural differences is remarkably variable across individuals, even highly trained listeners. In experiment 1, we measured sensitivity to changes in interaural correlation in 28 naïve listeners to determine the expected inter-individual variability for this task. Stimuli were 65-dB-A, 10-Hz narrowband noises with a 500- or 4000-Hz center frequency. Stimuli were tested either without level roving, or with ±5 dB of level roving in an effort to force listeners to attend to binaural width/movement cues rather than binaural loudness cues. At 500 Hz without level roving, the performance for the naïve listeners was much worse than the performance for the highly trained listeners previously reported in the literature. The addition of level roving significantly degraded performance, suggesting a close relationship between detection of decorrelation and loudness. At 4 kHz with and without level roving, none of the naïve listeners could perform the task. In experiment 2, we measured correlation change sensitivity as a function of overall level in trained listeners. Performance improved as overall level increased to about 65 dB-A, then worsened for more intense levels. The data will be discussed in terms of confusions between width and loudness cues, and in terms of Weber’s law.