Virginia M. Richards
G. Bruce Henning
Department of Psychol., Univ. of Pennsylvania, Philadelphia, PA 19104
Observers were required to detect 500-, 110-, or 20-ms bursts of 250-, 500-, or 5000-Hz tones presented in simultaneously gated narrow (30-Hz) noise bands using standard two-interval forced-choice tracking procedures. The noise was in phase at the observers' ears; the signal was in phase in one set of experiments and (pi) radians out of phase in another. Three different ``roving-level'' conditions were used: (1) fixed level the noise-power density, N[sub 0], was fixed (50 dB SPL), (2) inter-interval rove a random gain factor was chosen for every observation interval from a set uniformly distributed in decibels (40-dB range; 0.1-dB steps) to produce N[sub 0] centered on 50 dB SPL. The gain factor affected signal and noise in both ears equally and thus left the signal-to-noise ratio, interaural phase differences and amplitude ratios unaffected, (3) interaural rove the random gain varied from ear-to-ear in every observation interval. Reducing duration reduced BMLD's in all conditions; roving levels reduced BMLD's by small amounts except at 20 ms where interaural rove removes them. [This research was supported by The National Institutes of Health.] [sup a)]On leave from the Dept. of Exp. Psychol., Oxford.