Phase effects for a sine wave masked by reproducible noise

Abstract

A tone-in-noise detection task was used to assess the filtering properties of the auditory system. In the first experiment, ten 150-ms samples of reproducible noise were used as maskers. The signal was a 500-Hz tone either 20 or 100 ms in duration. The 100-ms signal was centered temporally in the noise; the 20-ms signal occurred either at the beginning, center, or end of the noise. The starting phase of the signal was varied from 0°–315° in steps of 45°. Signal thresholds, collected by the method of adjustment, were a cyclical function of starting phase and could be described by an energy model [Green and Swets, Signal Detection Theory and Psychophysics (Krieger, Huntington, NY, 1966/1974)]. A vector description of these data revealed an invariant property of each sample of noise, which we call the ‘‘noise vector.’’ The relationship among the parameters of the noise vectors over the various signal conditions suggest the presence of temporal interactions due to narrow-band filtering. These relationships are consistent with the bandwidth estimates from the energy model which were 36, 36, and 52 Hz for the three subjects. In the second experiment, a 20-ms signal was used and 32-ms maskers were derived from the original ten maskers by deleting portions of the noise that had not occurred simultaneously with the signal. The noise vectors measured for these stimuli differed from those obtained in the first experiment. Moreover, the vectors were better predicted by the energy model than by the vectors of the first experiment, supporting such a model and adding further evidence of temporal interactions due to narrow-band filtering by the auditory system.