Scientific Papers ACTA ACUSTICA UNITED WITH ACUSTICA Vol. 91 (2005) 421 – 428 Auditory Spatial Perception with Sources Overlapping in Frequency and Time Virginia Best , Andr´ e van Schaik , Craig Jin , Simon Carlile Department of Physiology ( ) and School of Electrical and Information Engineering ( ), University of Sydney, NSW, Australia Summary The effect of a broadband masker on the localisation of a broadband target stimulus was investigated. Human listeners identified the perceived location of the target stimulus, which was randomly presented with and without the masker. Five stimulus conditions were examined, in which the temporal overlap and total duration of the pair were varied. It was found that, overall, localisation was quite robust in the presence of the masker. However, a small and systematic bias in lateral angle perception that was directed away from the lateral angle of the masker was observed. The effect was most pronounced when the target and masker overlapped in time, and was nearly eliminated when the masker fully preceded the target. Furthermore, the effect was not restricted to stimulus pairs in close proximity. Disruptions to polar angle localisation were far less systematic, varying in both magnitude and direction. In general the results were not consistent with previous models that incorporate spatial channel interactions. When stimuli overlap in both frequency and time, it is likely that interference in the processing of acoustic localisation cues influences perceived location. PACS no. 43.66.Dc, 43.66.Pn, 43.66.Qp 1. Introduction 1.1. Background The auditory system uses acoustic cues available at the two ears to compute the locations of sound sources (for review see [1, 2, 3]). Binaural cues result from the fact that the two ears sample a wavefront from two different positions and include interaural differences in time (ITD) and level (ILD). In addition, spectral cues arise from the spatial asymmetry of the acoustic filtering of the auditory periphery. Location-dependent filtering of sounds by the pinnae, head and torso has been shown to create spectral cues to sound source location. A lack of consistency has characterised previous studies examining sound localisation in the presence of a masker. Many researchers have reported that target stimuli tend to be repelled from the location of a masker when one is present (we will refer to this as a ‘pushing effect’). This has been reported in the horizontal plane for pure tone pairs [4], broadband noise pairs [5], and pure tone and narrowband targets with a noise masker [6, 7]. A push- ing effect has also been reported by one author for con- current noise sources in the median vertical plane [8, 9]. On the other hand, several reports have described quite the opposite, where a target is mislocalised with a shift towards a concurrent masker (‘pulling effect’). This has Received 28 January 2004, Revised 10 August 2004, accepted 30 August 2004. been reported for noise pairs [10], a pure tone in the pres- ence of a rectangular pulse [11], and pairs of sinusoidally- amplitude-modulated tones [12]. Other studies examining the effect of a masking sound on target source localisation have failed to find systematic effects of this nature as long as the target is clearly audible [13, 14, 15, 16]. The large majority of studies of concurrent sound source localisation have adopted stimulus presentation paradigms where the target and masker do not have simultaneous on- sets [7, 6, 4, 8, 9, 5, 13, 14]. In general, the masker on- set precedes the target onset by some amount (from 20 ms up to as long as 10 s) to ensure that it is clearly distin- guished from the masker. It has long been known that com- mon onset enhances the grouping of components into au- ditory objects, with the rationale that sounds from a com- mon object will usually coincide in time (this idea is ex- plored in depth in [17]). This stimulus arrangement has commonly revealed a ‘pushing effect’ of the masker. In contrast, the studies reporting ‘pulling effects’ [10, 11, 12] presented the target and masker with simultaneous on- sets. It seems reasonable then to assume that onset syn- chrony/asynchrony contributes to this complex set of lo- calisation phenomena. Perhaps the pushing effects are an artefact of the auditory system streaming the concurrent pair on the basis of onset differences, and the pulling ef- fects occur when the concurrent objects are grouped. 1.2. Approach The aim of this experiment was to test the hypothesis that simultaneous target and masker onsets result in a pulling c S. Hirzel Verlag EAA 421