Although stream segregation is typically studied with relatively simplistic stimuli such as pure tones, the parameters that influence basic auditory object formation may be relevant to understanding speech perception in noise. Prelimarary findings (n=9) from this research suggest that cochlear implant (CI) listeners may experience pure-tone stream segregation at slower presentation rates (longer inter-stimulus intervals) than normal hearing (NH) listeners. We speculate that this may reflect increased adaption/habituation in the auditory cortex from CI stimulation. If present, increased patterns of cortical adaption may potentially impair speech-in-noise comprehension, or the ability to follow rapid speech. We are currently measuring speech-in-noise thresholds in our CI group, and will present correlations between stream segregation and speech-in-noise measures.

This research was designed primarily to address aspects of stream segregation in CI users which remain poorly understood. In NH, segregation increases as the frequency separation (ΔF) between alternating tones is increased and/or the inter-stimulus interval (ISI) between tones is decreased. However, while stream segregation in CI listeners appears to be influenced by ΔF, ISI has not been found to affect segregation.

The task required listeners to detect a temporal delay imposed on a single tone. Stimuli were arranged so that any obligatory stream segregation should impair performance, and three ISIs were tested (50, 100, and 200 ms). Preliminary CI results (n=9) show delay thresholds increased with ΔF, and increased further with the addition of a segregation-promoting precursor sequence. Both observations are indicative of stream segregation effects. The CI group showed clear stream segregation effects at ISIs of 100 and 200 ms, but ceiling effects imposed on performance in the 50 ms ISI conditions. In contrast, a NH group (n=9) performed well and showed clear stream segregation effects when the ISI was 50 ms, but showed reduced stream segregation effects for the two longer ISIs. The pattern of results suggests stream segregation may persist at slower presentation rates for CI listeners. We will compare these measures of stream segregation to measures of speech-in-noise performance.