Effects of time intervals and tone durations on auditory stream segregation (original) (raw)

2000, Perception & Psychophysics

Adult listeners rated the difficulty of hearing a single coherent stream in a sequence of high (H) and low (L) tones that alternated in a repetitive galloping pattern (HLH-HLH-HLH ...). They could hear the gallop when the sequence was perceived as a single stream, but when it segregated into two substreams, they heard H-H-... in one stream and L-L-... in the other. The onset-to-onset time of the tones, their duration, the interstimulus interval (lSI) between tones of the same frequency, and the frequency separation between H and L tones were varied. Subjects' ratings on a 7-point scale showed that the well-known effect of speed's increasing stream segregation is primarily due to its effect on the lSI between tones in the same frequency region. This has implications for several theories of streaming. When a sequence of tones, alternating between two frequency ranges, is speeded up, the tendency for the high and low tones to form separate auditory streams is increased. It has been proposed by Bregman (1990) that tones group by their proximity on a frequency-by-time surface. An increase of speed brings the tones closer together in time but does not reduce their frequency separations. This brings the consecutive tones of the same frequency closer together on the frequency-by-time surface, while leaving those of different frequencies almost as far away as they were before. This new proximity favors the grouping of a tone with the next one in the same frequency range even if the two tones are not consecutive, because the alternative grouping (with the tone that comes right after it but is ofa different frequency) requires grouping across a longer distance. So we see that temporal distance is very important. But what is the best way to measure temporal distance? The effect of speed could be due to a change in any of the four types of time intervals shown in Figure 1, which all become shorter when the speed is increased. (Note: SOA means stimulus onset asychrony-i.e., onset-to-onset time, and lSI is the label for interstimulus interval-offset-to-onset time. (1) SOA for consecutive tones in the same frequency range (SOAwithin). Note that in Figure 1 there are two different intervals of this type, one for each frequency, sincethe low tones occur less frequently than the high ones in the galloping pattern. (2) lSI for consecutive tones in the same frequency range (lSI-within). Again, there are two different intervals of this type, since the low tones occur less Support from the Natural Sciences and Engineering Research Council of Canada (Experiment I) and NIMH (Experiment 2) is gratefully acknowledged. We are also grateful for Lisa Weaver's assistance. Correspondence concerning this article should be addressed to A.