Huanping Dai
Beverly A. Wright
Psychoacoust. Lab., Psychology Dept., Univ. of Florida, Gainesville, FL 32611
The results of two experiments demonstrated that the detectability of an auditory signal depends on the subject's expectation about its duration. The signal (a 250-, 1000-, or 4000-Hz tone, or a broadband noise) had one of six durations between 5 and 295 ms. The masker was a continuous, broadband noise (N[sub 0]=25 dB). Each 2IFC interval was always marked by a fixed-duration visual display whose onset coincided with the onset of all potential signals. When the signal duration was fixed, each signal was correctly detected on about 85% of the trials. in experiment 1, the six subjects were encouraged to expect a particular signal duration (5 or 295 ms) by presenting the signal most often at that duration, and only occasionally at each of the five remaining unexpected durations. Performance for the unexpected signals decreased from 85% to chance as their durations departed from the expected duration. In experiment 2, the signal was randomly presented at each of the six durations with equal probability, and the subjects were so informed. The mean percent correct value was about 79%, with the longest signals being affected the most by the duration uncertainty. The results are analyzed using signal-detection models incorporating the idea of an adjustable temporal-integration window. [Work supported by NIH.]