Leslie R. Bernstein
Constantine Trahiotis
Ctr. for Neurological Sci. and Surgical Res. Ctr., Dept. of Surgery, Univ. of Connecticut Health Ctr., Farmington, CT 06030
This study principally evaluated whether the normalized cross covariance (Pearson product-moment correlation) or the normalized cross correlation describes discriminability of changes in interaural disparities conveyed by the stimulus enveloped. In a four-interval, two alternative task, listeners detected which interval contained a 4-kHz tone added antiphasically to diotic, 200-Hz-wide, noise (NoS(pi)). The ``nonsignal'' intervals contained the tone added homophasically (NoSo). Discriminability (d') was measured as a function of S/N for values between -30 and +30 dB (really!). For all S/N's, overall level was 70 dB SPL. Listeners' performance was very well accounted for by the normalized cross correlation but not the normalized cross covariance. Additionally, listeners were tested in a ``direct'' discrimination task where changes in envelope correlations ((Delta)(rho)) were produced by ``mixing'' two independent Gaussian noises. Although (Delta)(rho)'s at threshold (d'=1.0) obtained from the two tasks were similar, the psychometric functions obtained with the direct discrimination task were more steep. Discussion will include how the normalized cross correlation of the envelope accounts for classic data concerning discriminability of interaural time differences at high frequencies as a function of depth of modulation. [Work supported by NIH DC02103.]