Binaural Masking-Level Differences for Pulse Train Signals of Differing Correlation

Abstract

Three trained observers performed a two-interval forced-choice detection task; the noise level in the earphones was controlled by a PEST program. In all conditions, the noise was equal in amplitude but inverted in phase (Nπ) in the earphones. Signal conditions included Sm, S0, and partially correlated (ρ) or uncorrelated signal waveforms in the contralateral ear. The signal in the ipsilateral channel was always a random pulse train composed of 100-μsec pulses of random (+1 or −1) polarity, with an interpulse period of 2 msec. All signals were low-pass filtered at 5 kHz. The signal conditions included two ρ=0 conditions in which the contralateral signal waveform was identical to the ipsilateral except it (a) had (randomly) half of the pulses reversed in polarity, or (b) was a deterministic signal composed of pulses of all negative polarity; and two ρ=0.5 conditions in which the contralateral signal (c) had (randomly) half of the pulses deleted, or (d) was composed of only the negative polarity pulses in the signal. We also ran ρ=0.75 versions of (a) and (c), and reverse polarity versions of (b) and (d). All the partial and zero ρ conditions resulted in psychometric functions which fell in the 6-dB range between those of NπSm and NπS0. The worst performance was obtained for condition (d).