Effects of different samples of reproducible noise on detection of a 20-Hz-wide interaural phase shift centered on 500 Hz

Subjects’ thresholds were first determined for detection of an interaural phase shift of a 20-Hz band of equal-amplitude wideband noise. The narrow-band target was centered on 500 Hz, and generated a dichotic, Huggins-type pitch. A diotic band of either 100, 125, 150, or 900 Hz symmetrically surrounded the target band. Interaural phases for frequency components outside of the diotic band were drawn randomly from a rectangular distribution. After thresholds were determined via a tracking procedure, 20 independent samples were generated for each condition at each subject’s threshold. A two-interval forced choice task was used in which each interval was comprised of a 250-ms forward fringe followed by a 250-ms observation period. Sixty trials were presented in a block with each sample presented three times. Order of presentation was randomized across runs and each sample was presented a total of 96 times. P(C) ranged from chance (50%) to over 90% across samples having identical interaural parameters. In conditions where subjects’ thresholds were similar, allowing the same set of samples to be used for two or more subjects, there was a strong inter-subject correspondence with respect to P(C). Possible sources of these sample-specific effects will be discussed. [Work supported by NIH.]