The effect of spatial configuration in a divided attention task

Abstract

The effect of spatial separation on the ability of listeners to report keywords from two simultaneous talkers was examined. The talkers were presented with equal intensity at a clearly audible level, but were processed to minimize their spectral overlap and reduce energetic interference. The two talkers were presented with various angular separations around references of −45°, 0°, or 45° azimuth. Overall, performance did not vary dramatically with spatial configuration, but depended on spatial separation and reference direction. With the talkers in front or to the left, performance tended to first increase and then decrease with increasing separation. With both talkers to the right, performance tended to improve monotonically with increasing separation. The relative levels of the two talkers at each ear in each configuration partially accounted for results. For each talker a different ear contained a signal-to-noise ratio advantage, and performance was positively correlated with the mean signal-to-noise ratio across the two ‘‘better ears.’’ Thus, when tracking two sources simultaneously, listeners may make use of the information at the two ears independently. Furthermore, the drop in performance for some large talker separations may reflect increased difficulty in following sources that do not fall within a single ‘‘spotlight’’ of spatial attention.