The effect of nonsimultaneous on-frequency and off-frequency cues on the detection of a tonal signal masked by narrow-band noise.

Listeners' detection thresholds were measured for a 125-ms, 1-kHz tonal signal masked by a similarly gated 50-Hz-wide band of noise. A two-interval, adaptive, forced-choice procedure either with or without temporally surrounding cuing intervals containing 50-Hz-wide bands of noise was employed. When the cues were present, their center frequency was either 1 kHz (on-frequency) or 900 or 700 Hz (off-frequency). In the conditions of principal interest, the envelopes and phase modulations of the bands of noise were "frozen" across the four intervals that defined a trial, but were chosen randomly across trials. Thresholds were lowest with cues centered at 1 kHz and increased substantially when the center frequency of the cues was changed to 900 or 700 Hz. With cues centered at 700 Hz, performance was equivalent to that obtained without cues and with the masking noise "frozen" across the two intervals that defined a trial. A similar pattern of results was obtained with high-frequency stimuli, where sensitivity to fine-structure information is greatly reduced. Roving the level of the stimuli over a 40-dB range generally reduced sensitivity but did not greatly affect the overall pattern of the data. Thresholds obtained in the two-interval task with masking waveforms chosen randomly were compared with thresholds obtained when the masking waveform was "frozen" within, but not across trials. Differences in threshold appeared to be accounted for by the listeners' use of changes in the mean slope of the envelope of the noise produced by adding the tonal signal.