Time course of auditory streaming: do CI users differ from normal-hearing listeners?

Abstract

In a complex acoustical environment, the auditory system decides which stimulus components originate from the same source by forming auditory streams, where temporally non-overlapping stimulus portions are considered to originate from one source if their stimulus characteristics are similar. The mechanisms underlying streaming are commonly studied by alternating sequences of A and B signals which are often tones with different frequencies. For similar frequencies, they are grouped into one stream. Otherwise, they are considered to belong to different streams. The present study investigates streaming in cochlear implant (CI) users, where hearing is restored by electrical stimulation of the auditory nerve. CI users listened to 30-s long sequences of alternating A and B harmonic complexes at four different fundamental frequency separations, ranging from 2 to 14 semitones. They had to indicate as promptly as possible after sequence onset, if they perceived one stream or two streams and, in addition, any changes of the percept throughout the rest of the sequence. The conventional view is that the initial percept is always that of a single stream which may after some time change to a percept of two streams. This general build-up hypothesis has recently been challenged on the basis of a new analysis of data of normal-hearing listeners. Using the same experimental paradigm and analysis, the present study found that the results of CI users agree with those of the normal-hearing listeners: (i) the probability of the first decision to be a one-stream percept decreased and that of a two-stream percept increased as Δf increased, and (ii) a build-up was only found for 6 semitones. Only the time elapsed before the listeners made their first decision of the percept was prolonged as compared to normal-hearing listeners. The similarity in the data of the CI user and the normal-hearing listeners indicates that the quality of stream formation is similar in these groups of listeners.

Keywords: auditory streaming; build-up; cochlear implant; perception; psychoacoustics.

FIGURE 1

Individual durations of answering options “two-streams” (green cross marks) and “one-stream” (blue circles), calculated as proportions of the total sequence duration (30 s), are plotted as a function of the (average) Δf for each listener.

FIGURE 2

Grand means of the proportions of the two percepts from eight listeners for different Δf conditions. Error bars denote the standard error of mean.

FIGURE 3

(A) Time of the first response; individual values from eight listeners. (B) Box plot of the time of first response. Box boundaries mark the lowest and highest quartiles, center lines the median. (C) Type of first decision for different Δf conditions.

FIGURE 4

Time courses of the current probability of the two-stream percept for the different Δf conditions. (A) Conventional display and (B) normalized probability, rescaled by the probability that any response has been given yet (see text).