The experimental data in Experiments I, II, and III were mainly analyzed with
repeated-measures analysis of variance (ANOVA). The Type-I error rate, α, was set
to 0.05 for all analyses. When multiple comparisons or ANOVAs were performed,
the α in each test was adjusted to maintain a family-wise α of 0.05 using the ˇSid´ak
correction factor. Degrees of freedom were adjusted with the Greenhouse-Geisser’sǫ;
this takes into account any violation to the sphericity assumption in the data. Effect size is reported asη2, the proportion of variance in the data that is accounted for by
Chapter 4
Experiment I: Effect of reverberation context on
sound localization
In Chapter 2, the auditory cues for sound localization and the cue-weighting processes in both free and reverberant fields have been discussed. The role of stimulus context in sound localization, however, has largely been overlooked. In several other audi- tory tasks, for example, frequency discrimination (Demany, 1985; Amitay, Hawkey, & Moore, 2005; Watson, Kelly, & Wroton, 1976) or discrimination of interaural dif- ference (Wright & Fitzgerald, 2001), consistency in stimulus context has been shown to have a positive effect on listeners’ performance. Perhaps when localizing sounds in reverberation, listeners can accumulate information about the reverberation across trials and utilize this information to refine their listening strategies. This chapter presents a study that examines the effect of reverberation context on sound localiza- tion performance.
4.1 Introduction
In a study that examined the effect of listening experience on sound localization performance in a room (Shinn-Cunningham, 2000), it was shown that listening ex- perience can improve the accuracy of localization judgments in azimuthal, vertical, and distance dimensions. The improvement was observed over several days of testing sessions and occurred in the absence of feedback. This indicated that the change in performance was driven by the information inherent in the stimuli. Besides this study, however, there is little research that examines the role of listening experience or the underlying mechanisms that may contribute to improved sound localization perfor- mance with experience. Evidence of listeners’ taking into account stimulus pattern in the context has been demonstrated in several other auditory tasks, such as frequency discrimination of a tonal component within a tonal pattern (Watson et al., 1976), detection of tones in noise (Green, 1961; Creelman, 1973), and word discrimination under different binaural configurations (Nahum, Nelken, & Ahissar, 2010). In the last study, the ability to take advantage of the difference in IPDs of the target and the
masker (referred to hereafter as the binaural advantage) was measured as the differ-
ence in speech reception threshold (SRT) between theSπN0 and S0N0 configurations.
The subscript in S− denotes the IPD of the masker, and the subscript inN− denotes
the IPD of the masker. Binaural advantage was measured under two listening con- ditions differing in the amount of trial-to-trial variability: in the consistent-protocol condition, the SRTs for the S0N0 and SπN0 conditions were obtained with two sepa-
rate adaptive tracks; in the mixed-protocol condition, the SRTs were measured with
two independent but interleaved tracks, the binaural configuration was randomly
interleaved adaptive tracks with an alternating configuration (SπN0-S0N0-SπN0-· · ·).
A significantly higher binaural advantage was obtained by listeners in the consistent- protocol group than that obtained by the listeners in the other groups. Furthermore, following several days of training, listeners in the 1-1-protocol group could achieve an amount of binaural advantage comparable to that achieved by the consistent-protocol group. That is, the listeners could successfully learn to discriminate SπN0 and S0N0
trials within a more complex context, but it took more time for them to do so. The binaural advantage achieved by the mixed-protocol group, however, was significantly lower following the same amount of training.
Additionally, in a study that examined the effect of reverberation context in a distance perception task (Schoolmaster, Kopco, & Shinn-Cunningham, 2003), con- sistency of the listener’s position within a room was shown to yield higher accuracy in distance judgments relative to the condition in which the listener’s position was varied across trials. These studies illustrate that high variability in the context can negatively affect listeners’ performance.
The present study aimed to find further evidence that the ability to localize sounds in the horizontal dimension is affected not only by the acoustic cues present in isolated trials, but also on the cues in the context. The effect of reverberation context on sound localization will be examined; listeners’ ability to determine the direction of a sound source in a listening condition in which the room is consistent across trials (fixed- exposure condition) will be compared to that obtained in a condition in which trials from different rooms are presented within the same listening block (mixed-exposure condition). The simulated rooms (R0, R1, R2, and R3) presented in Chapter 3 are used; recall that the reliability of the spatial cues degraded as the lateral spread of the
reflections increased (i.e., reliability decreased from R0, R1, R2, to R3). Because of a higher degree of trial-to-trial variability in the mixed-exposure condition, we expected listeners to perform better in the fixed-exposure condition.