Temporal factors in referential intensity coding.

Three experiments investigated the finding [Plack et al., J. Acoust. Soc. Am. 97, 1141-1149 (1995)] that intensity discrimination under backward masking can be improved by presenting an additional, "proximal," tone burst shortly before or after the pedestal. All the stimuli used in the experiments were 30-ms, 1-kHz sinusoids. In the first experiment, intensity discrimination was measured for a 50-dB SPL pedestal presented 100 ms before an 80 dB SPL masker. A proximal tone burst was presented either before or after the pedestal, separated from the pedestal by a brief silent gap. For the conditions in which the proximal burst was before the pedestal, adding a proximal burst with a higher level than the pedestal produced an improvement in intensity discrimination. The most effective level of the proximal burst increased as the gap was increased. For the conditions in which the proximal burst was after the pedestal, two listeners showed an improvement when the proximal burst was lower in level than the pedestal, and one listener showed an improvement when the proximal burst was higher in level than the pedestal. In a second experiment, detection threshold measurements showed that good performance was not dependent on the proximal burst making the pedestal in one of the two observation intervals. The final experiment used a selective training procedure to demonstrate that listeners were basing performance on two conflicting strategies, namely, to pick the interval that sounded as if it had three tone bursts in it when the proximal level was higher than the pedestal level, and to pick the interval that sounded as if it had two tone bursts in it when the proximal level was lower than the pedestal level. A model of temporal resolution is presented that can explain certain aspects of the results in terms of the detection of "bumps" in the temporal excitation patterns produced by the stimuli. In conditions of backward masking, these relative features seem to provide a superior cue for intensity discrimination than absolute intensity, which is actively rejected as a cue.