Auditory Temporal Processing 1 Running Head: Auditory Temporal Processing the Relation between Auditory Temporal Interval Processing and Sequential Stream Segregation Examined with Stimulus Laterality Differences

This study examines the effects of laterality differences between noise bursts on two objective measures of temporal interval processing gap detection and temporal asymmetry detection, and one subjective measure of temporal organization (stream segregation). Noise bursts were lateralized by presentation to different ears, or dichotic presentation with oppositely-signed interaural level (ILD) or time (ITD) differences. Objective thresholds were strongly affected by ear of entry differences, moderately affected by ILD differences, but unaffected by ITD differences. Subjectively, A and B streams segregated well based on ear of entry or ILD differences, but segregated poorly based on ITD differences. These results suggest that perceptual segregation may be driven more effectively by differential activation of the two ears (peripheral channeling) than by differences in perceived laterality. Auditory Temporal Processing 3 Our sonic environment consists of events from different sources distributed in space and time. Our perceptual system groups acoustic events that presumably emanate from a single source, and segregates those from different sources into separate perceptual streams (after Bregman, 1990). Such processes are likely involved in the cocktail party phenomenon – wherein we selectively attend to a sound source of interest within a multi-source environment. In this study we are interested in how spatial location attributes might be used for segregation. To do so, we exploit an aspect of the perceptual organization of temporally interleaved sounds that temporal judgments are impaired when they must be made between sounds which are perceived to emanate from different distal sources (Warren, Obusek, Farmer, & Warren, 1969; Bregman & Campbell, 1971). Thus, degradation in performance on temporal judgments made between two sounds that vary on some dimension (e.g. location, pitch) might be used as an index by which to determine the relevance of that dimension in stream segregation. Thresholds for the detection or discrimination of auditory temporal gaps (a silent interval between two sounds) may provide a simple measure of auditory temporal acuity for this purpose. In a variant of gap detection, termed ‘between-channel’ gap detection (after Phillips, Taylor, Hall, Carr & Mossop, 1997), the effects on gap thresholds of varying the properties of the sounds preceding and following the silent period have been explored. The two properties which have the most striking effect on gap detection are spectral differences Auditory Temporal Processing 4 between the gap markers (Kinney, 1961; Perrott & Williams, 1971; Fitzgibbons, Pollatsek & Thomas,1974; Williams & Elfner, 1976; Divenyi & Danner, 1977; Penner, 1977; Williams, Elfner & Howse, 1978; Neff, Jesteadt & Brown, 1982; Formby & Forrest, 1991; Hall, Grose & Joy, 1996; Phillips et al., 1997; Formby, Sherlock & Li, 1998; Taylor, Hall, Boehnke & Phillips, 1999; Grose, Hall, Buss & Hatch, 2001; Phillips and Hall, 2000, 2002) and ear-of-entry differences (Divenyi & Danner, 1977; Penner, 1997; Phillips et al., 1997; Taylor et al., 1999). Spectral and ear-of-entry differences between interleaved sounds are also those which have the strongest influence on auditory sequential stream segregation (Van Noorden, 1975; Bregman, 1990; Hartmann & Johnson, 1991). In the gap detection task, when the stimuli bounding the silent period are presented to different ears gap thresholds are often 5 to 10 times higher than those obtained when the same gap markers are presented to a single ear, which can be as low as 2 ms (Penner, 1977; Phillips et al., 1997; Formby et al., 1998). Similarly, gap thresholds were 5 to 10 times longer when markers bounding the silent period originated from free-field sources opposite each ear (i.e., at +/90 degrees azimuth) than when the markers both originated from the same source (Phillips, Hall, Harrington & Taylor, 1998; Boehnke & Phillips, 1999). Two sounds originating from different free-field spatial locations result in different values of the binaural cues which lead to lateralized percepts. These cues include interaural time difference (ITD) and interaural level difference (ILD). The two Auditory Temporal Processing 5 sounds will also differ monaurally in intensity at each ear, and monaural intensity differences have long been known to lead to elevated gap thresholds (Plomp, 1964; Penner, 1977), most likely due to temporal masking processes (see Smiarowski and Carhart, 1975). This raises the question as to whether it is the perceptual change in ‘location’ between the two sounds that leads to elevated thresholds in the free-field case, or the difference with which the sounds activate the frequency channels of each ear. These two possibilities were dissociated in a dichotic gap detection paradigm by Oxenham (2000). Marker differences in ITD alone failed to have an effect on gap thresholds for noise stimuli. ILD differences between leading and trailing markers elevated gap thresholds, but they did so no more than did level differences of the same size as those at a single ear. This finding suggests it is the differential activation of peripheral auditory filters rather than the change in perceived laterality per se that degrades performance on gap detection. While studied separately, no study has compared the absolute values of gap detection thresholds from the same subjects when the markers differ in ear of entry, ILD or ITD alone. What would be still more helpful is knowledge of whether the effects of those spatial attributes generalize to performance in other, arguably more sophisticated, temporal processing tasks which have been directly linked to stream segregation. Another temporal processing paradigm, the temporal asymmetry detection task (Vliegen, Moore, & Oxenham, 1999), has recently been Auditory Temporal Processing 6 introduced as an objective measure of auditory sequential stream segregation (SSS, after Bregman, 1990). A simple demonstration of SSS is accomplished by presenting a sequence of two alternating sounds (‘A’ and ‘B’) differing in some attribute (usually tonal frequency), in the temporal sequence A_B_A___A_B_A... Note that SSS refers to the situation where the two components do not overlap in time. A related phenomenon – concurrent stream segregation – refers to segregation that occurs when the two components are presented simultaneously. In the sequential case, when A and B are similar listeners report perception of a galloping rhythm (single-stream percept or auditory fusion). As the difference between A and B is increased, listeners are more likely to report that the sequence segregates into two separate A and B streams (2-stream percept, segregation or fission). For a range of intermediate A and B differences, it is common for listeners to experience a bistable percept (van Noorden, 1975). In the temporal asymmetry paradigm, a repeating sequence of A and B sounds as described above is presented, except that the temporal location of B with respect to A is systematically varied. The task of the subject is to distinguish a standard sequence in which B is centered in time between the two A sounds from a sequence in which B has been temporally shifted towards one of the A sounds. The measured threshold in this objective task represents the smallest detectable asymmetry in the duration of the silent intervals between the A_B and B_A components. The task is easier and produces lower thresholds Auditory Temporal Processing 7 when a galloping rhythm is perceived, than when A and B are perceived to segregate into separate perceptual streams (Vliegen et al., 1999). As such, this task is thought to provide an objective index of sequential stream segregation. Note that in the temporal asymmetry task the listener is effectively biased to hear a single-stream percept (integration) in order to do the task. It is well known that mental set can modulate the emergence of a 1or 2-stream percept (Van Noorden, 1975). Akin to the elevation of gap thresholds which occurs when the markers differ spectrally, temporal asymmetry detection thresholds were elevated as the A and B components are made to differ spectrally (Vliegen et al., 1999). Furthermore, elevation of thresholds as a function of A-B spectral separation mapped onto the amount of phenomenal stream segregation reported by listeners for such A-B differences (Vliegen and Oxenham, 1999), despite differences in task demands (bias to hear 2-streams). While this is suggestive of a link between performance on gap detection and temporal asymmetry detection, no study has ever directly compared thresholds for temporal asymmetry detection with those for gap detection using similar stimuli. The present study seeks to determine if this relation between these tasks holds true for conditions where the components differ in spatial stimulus attributes (after Phillips et al., 1997; Oxenham, 2000). That is, because a disparity between leading and trailing gap markers in ILD, but not ITD, elevates gap thresholds, will A-B disparities in ILD, but not ITD, elevate Auditory Temporal Processing 8 temporal asymmetry thresholds? If so, might A and B differences in ITD lead to less segregation than similar differences in ILD when using a more traditional, subjective measure of auditory stream segregation? We examined these questions in a set of three experiments. In all three experiments we manipulated the same set of spatial cues; these were ear of entry (after Penner, 1977; Phillips et al., 1997), ILD and ITD (after Oxenham, 2000), and level differences between markers at one ear (to control for monaural contributions to the effects of ILD: after Oxenham, 2000, Penner, 1977). In Experiment 1, we re-examined the effects on gap detection thresholds of imposing spatial cue differences on two noise markers bounding a gap. In Experiment 2, we used the auditory temporal asymmetry task (described above) which involves recurrent sequences of two different (A, B) noise markers. Finally, Experiment 3 was a traditional, subjective stream segregation study (after Bregman, 1990), in which listeners tracked the development of perceptual streams prompted by differentiating A and B noises of a repeating ABA_ sequence. In Exp. 2 and 3, it is reasonable to assume that the task demands differently bias listeners towards hearing 1and 2-stream percepts respectively. Experiment 1: Gap Detection The purpose of this experiment was to replicate and extend the findings of Oxenham (2000). Experiment 1 was thus a dichotic study of gap detection thresholds obtained using wideband noise stimuli in which the leading and trailing Auditory Temporal Processing 9 markers differed in their spatial attributes. There were 6 experimental conditions: the leading and trailing noise markers were presented to different ears (‘EAR’ condition), or with different values of ‘ITD’ (+/-0.5 ms) or ‘ILD’ (+/-15 dB), or with different intensities presented monaurally (Loud-Quiet ‘LQ’ or Quiet-Loud ‘QL’, to control for the stimuli presented to each ear in the ILD condition), or both markers were presented diotically (‘Diotic’). The chosen values of ILD and ITD were important, if somewhat arbitrary. The maximum ITDs generated by the human head are in the order of 600-700 μs depending on head size (Middlebrooks and Green, 1991), and the maximum ILDs are in the order of 30 dB (at high frequencies: Middlebrooks, Makous, & Green, 1989). For tonal stimuli, the perceived intracranial lateral displacement of a source saturates after 90 degrees of phase angle irrespective of frequency, and it usually saturates after 15-20 dB ILD, irrespective of frequency (Yost, 1981). The choices of 500 μs ITD and 15 dB ILD thus guaranteed highly lateralized noise stimuli in both dichotic conditions. Whether the ITD and ILD stimuli were equivalently lateralized is another matter. Harris (1960) showed that the timeintensity trading relation for their broadest band of noise (7000 Hz bandwidth) was about 0.04 ms/dB; using this trading ratio, 15dB ILD corresponds to 600μs. Yost, Tanis, Nielsen & Bergert (1975) provided psychophysical evidence suggesting that for noise stimuli, an ITD of 600 μs was judged as roughly equivalent in lateral location to a stimulus with an ILD near 16 dB. The present Auditory Temporal Processing 10 choices of 500 μs ITD and 15 dB ILD differ slightly from those used by Oxenham (2000: 640 μs and 12 dB), but, as will be seen, this was inconsequential to the pattern of results. Most importantly, it was expected that separation of the gap markers by ear of entry would result in significantly larger thresholds than separation by ILD value, which should not differ from monaural level differences. There are two reasons for this. In the EAR (i.e., between-ear) condition, there is no common activation of peripheral channels and thus the task is probably accomplished by a relative timing operation between the events at each ear, while the ILD condition could, in theory, be performed monaurally through within-channel (i.e., withinear) processes. Thus, the operating assumption is that within-ear processing results in lower thresholds than between-ear processing. Method