Ambiguous musical figures and auditory streaming

Abstract

Three experiments with musicians and nonmusicians (N=338) explored variations of Deutsch's musical scale illusion. Conditions under which the illusion occurs were elucidated and data obtained which supported Bregman's suggestion that auditory streaming results from a competition among alternative perceptual organizations. In Experiment 1, a series of studies showed that it is more difficult to induce the scale illusion than might be expected if it is accepted that an illusion will be present for most observers despite minor changes in stimuli and experimental conditions. The stimulus sequence seems better described as an ambiguous figure. Having discovered conditions under which the scale illusion could be reliably induced, Experiments 2 and 3 manipulated additional properties of the stimulus (timbre, loudness, and tune) to provide cues to streaming other than pitch and location. The data showed that streaming of this sequence can be altered by these properties, supporting the notion of a general parsing mechanism which follows general gestalt principles and allows streaming by many stimulus dimensions. Finally, suggestions are made as to how this mechanism might operate.listening to music provides an excellent context for exploring the phenomenon of auditory streaming. Bregman and Campbell (1971) described auditory streaming as the perceptual splitting of concurrent auditory events into separate streams or sequences. This must happen, for example, when listeningto an orchestra. The listener needs to decide which melodies are being played by what instruments, where the sounds originated, and how many instruments and melodies there are.Research on auditory streaming has focused on much simpler situations than listening to an orchestra. Bregman and Campbell, for instance, binaurally presented subjects with a rapid repetitive sequence of six different 100·msec pure tones, three high pitched and three low pitched, one and a half octaves apart. Although successive tones alternated from the high-to low-pitch range, subjects organized the high-and low-pitched sounds into separate auditory streams. Some subjects (59%) claimed that these two streams were successive (i.e., three high tones followed by three low tones, or vice versa); the remainder reported that the streams were concurrent. Neither of these percepts are veridical. Deutsch (1975b) found similar effects when she pre-

Structural alterations of an ambiguous musical figure: The scale illusion revisited

“…Experiment 2 included no NS note pattern, and it provided listeners with a closed set of (written) responses (cf. Smith et al, 1982). The closed set is well justified by the finding in Experiment 1 that 90% of all responses fell into two categories ("arc" and "linear") and that only a small portion fell into the "other" category.…”

“…However, there are some considerations that caution against such an interpretation. The presence of a substantial number of veridical responses in a closed set is not an unique finding (see Smith et al, 1982). The scale illusion, while compelling, is not completely convincing.…”

“…This is the essence of an effect called the scale illusion (Butler, 1979;Deutsch, 1975;Smith, Hausfield, Power, & Gorta, 1982). In scaleillusion experiments, listeners are asked to report on the dichoticallypresented note pattern illustrated in Figure lao This pattern is constructed from two diatonic major scales, one ascending and the other descending.…”

The effects of familiarity and previous training on perception of an ambiguous musical figure

“…Seventy percent of her subjects reported hearing the percept shown in Figure 5c; that is, they reported a "wave" pattern percept in which pitch information took precedence over all else, attributing all the high notes to the right ear and all the low notes to the left ear. Smith, Hausfeld, Power, and Gorta (1982) discovered that replication of Deutsch's findings occurred only when subjects' choices of percepts were limited to the possibilities presented in Figure 5. Moreover, they discovered that musicians were more likely than nonmusicians to report streaming by pitch (suggesting that Deutsch's subjects may have had musical training), and that, when additional information (timbre, loudness, and tune) was given, this was used by subjects so that pitch became only one of many possible cues to correct streaming.…”

“…Two ofthe findings of Smith et al (1982) have led to the present experiment: first, they reported that their subjects (particularly musicians) considered the veridical percept ( Figure 5a) to be a most unlikely tune, and second, they suggested that the musicians' tendency to not report the scale percept ( Figure 5b) might have been because they considered this percept a less typical tune than the wave pattern ( Figure 5c) (Smith et aI., 1982, p. 463). In other words, neither the veridical percept nor the scale percept was expected by subjects, which brings us again to the consideration of the baroque technique of presenting tunes separately before interleaving them.…”