Abstract
Human auditory localization ability is generally good for single sources, with errors subtending less than 5 degrees at the midline, and increasing to roughly 12 degrees in the periphery. Errors are generally larger in conditions where task demands are high, such as when multiple simultaneous sources are present. The current study shifts focus from the size of localization errors to the frequency of localization errors, examined under conditions of high and low visual uncertainty. In this study the ability to localize the addition of a new source to a mixture of 4 simultaneous sources was examined under two conditions of visual uncertainty. The task for participants was to listen to two auditory scenes separated by a brief blank interval. The scenes contained sounds edited to 1000ms duration that were representativeof outdoor urban spaces. Each first scene contained four sound sources; the second always contained a change, the addition of a new sound source. At the end of the second scene participants indicated where the change occurred, by pointing to the change location. Change locations were represented as either nine clearly visible and identifiable sound producing targets (speakers) in the free-field (low visual uncertainty condition) or change locations were obscured by occurring within an array of 90 visual targets (positioned 2 degrees apart), creating conditions of high visual spatial uncertainty. In general, performance was poorer when the audio-visual correspondence was reduced. Further, the pattern of results is consistent with other multisource localization studies; performance was most accurate when changes occur directly in front of the listener. Accuracy declined as changes were presented further in the periphery, and the impact of change position was greater under conditions of high visual spatial uncertainty (low audio-visual correspondence).
Meeting abstract presented at VSS 2015