Figure/ground segregation enhanced by spatially directed attention

Abstract

We recently reported that detecting a “figure” of repeated chords amidst “background” tones was worse when the figure was spatially separated from the background, a counterintuitive result given the published studies of spatial unmasking. Because of the way we blocked trials and provided instructions/feedback, listeners may have directed attention to the front where most figures appeared. Here, we investigated whether spatial cues would improve figure detection when listeners were instructed that spatial cues could aid performance. Figure detection was recorded for a condition where both the background and figure were presented diotically and for two binaural conditions. In one binaural condition, the figure occurred as a brief token at a lateral position separated from the background. In the other binaural condition, the background was composed of both a diotic stream and a lateralized stream, which contained the target (if present). Figure detection accuracy was better for both binaural conditions than for the diotic condition. Accuracy was best for the binaural condition in which the background was diotic and the figure was a brief lateralized token. These results reinforce the idea that binaural cues can be used to direct attentional focus, enhancing detectability of a target object amidst competing sounds.