Effects of Background Noise on Auditory Detection of Noise Bursts

Abstract

The ability of human observers to detect signals that are bursts of white noise is measured by two experimental procedures. In both procedures, the observer's task is to select the interval containing the signal from the two temporal intervals, marked by lights, which constitute a trial. In the first, the signal is added to a continuous background of white noise; in the second, the signal is added to one or the other of two equal noise bursts, that are added to the continuous background noise during the observation intervals. The psychometric functions obtained in the two experimental conditions are different. In the first experimental condition the psychometric functions are consistent with the assumption that the observer is uncertain about either the exact time when the signal occurs or its exact duration, or about both. In the second procedure, in which two noise bursts mark the observation intervals, the psychometric functions are consistent with the prediction of a statistical-decision model that assumes exact knowledge of the temporal occurrence of the signal. In this second procedure the signal to noise ratio for some constant level of detectability is considerably affected by the relation between the continuous background noise and the level of the two noise bursts. The minimum value of signal-to-noise ratio occurs when the continuous noise is ∼5–10 dB more intense than the equal noise bursts; the signal to noise ratio increases from this minimum as the background level is changed in either direction.