Persistence of a pitch-segregating echoic memory.

Auditory stimuli were computer generated in order to measure the persistence of echoic memory. The stimuli consisted of 18 bursts (lasting 307 msec) of six equal-amplitued dichotic tones (frequencies: 392, 440, 494, 523, 587, and 659 Hz), each having a different interaural time disparity. For each stimulus a canonical distribution of interaural time disparities was defined. Five of the interaural time disparities in each burst were equal to canonical disparities for that stimulus; the sixth was not. The deviant tones in successive bursts constituted a musical scale. These deviant tones were perceptually segregated when the interburst interval was short, even though individual bursts sounded like noise when played separately. The interburst intervals for which five subjects could identify with 71% accuracy whether the scale was ascending or descending (obtained by an adaptive psychophysical procedure) averaged about 1 sec. This figure represents a lower bound on the average half-life of echoic memory. A sixth subject performed perfectly even with an inter burst interval of 9.7 sec. Two further experiments were carried out with this subject to support the claim that his performance was due to echoic memory.