Abstract:
Intensity discrimination was measured in a 2AFC task in which the duration of a 1-kHz sinusoid was randomized, independently for each observation interval, between 40 and 400 ms, while keeping the standard intensity constant. Listeners were required to detect a difference in intensity between the two intervals while ignoring the random changes in total energy (and consequently loudness) caused by the duration randomization. Randomization caused a deterioration in performance consistent with listeners being unable to dissociate the intensity change from the random changes in total energy, implying long-duration temporal integration. When notched noise was added to the stimulus, however, performance improved to a level similar to that seen without the randomization. It is possible that listeners can make a relative comparison of the intensity of the tone with that of the notched noise (possibly in a manner similar to profile analysis) that utilizes a much shorter integration period. It is suggested that when making absolute intensity (in other words, loudness) judgments, listeners generally use a long integration time, but that judgments of relative intensity, which are needed for identifying the features that characterize auditory objects, are made over much shorter intervals. [Work supported by The Royal Society.]