The dynamics of auditory stream segregation for tone sequences with gradually and abruptly varying stimulus properties

Abstract

The nine experiments presented within this thesis explored the dynamics of stream segregation in repeating ABA tone sequences with gradual or abrupt changes in their acoustic properties. Experiments 1-6 used a continuous monitoring method to investigate the effect of these changes on the number of streams perceived (1 or 2). Experiments 1 and 2 demonstrated that abrupt and gradual changes in sequence base frequency had a much stronger effect on the build-up of streaming over time than those in interaural time difference (ITD), an outcome consistent with either functional or neural accounts of the build-up of segregation. Experiments 3 and 4 demonstrated that abrupt changes either in timbre (using pure tones and narrowly spaced tone dyads) or level could produce resetting (partial loss of build-up) but that the direction of the transition was important. Notably, an overshoot in stream segregation followed the tone-to-dyad transition, despite no significant change in the pattern of peripheral excitation. Experiments 6 and 7 demonstrated that resetting was not a result of correlated changes in A and B tone subsets. In both experiments, anti-correlated level changes tended to produce resetting (B"A#) and overshoot (B#A"), respectively. This outcome favours a neural mechanism of build-up based on subtractive adaptation. Experiments 7-9 investigated the influence of an induction sequence on the perception of a subsequent test sequence. Experiments 7 and 8 achieved capture of a tone subset in the test sequence by adjusting the difference in frequency or level between inducer tone subsets, such that only one subset matched its test-sequence counterpart. This resulted in greater stream segregation. Experiment 9 attempted capture using a harmonic complex synchronous with the lower subset. However, the fusion of the synchronous complex with the corresponding tone subset failed to disrupt capture, presumably because it did not change the rhythm of the sequence. Overall, these experiments demonstrate that abrupt changes in stimulus properties can cause resetting of build-up or overshoot, depending on the nature of the transitions, and stream capture can be achieved by manipulating the difference between tone subsets in an inducer.