Object continuity enhances selective auditory attention

Abstract

In complex scenes, the identity of an auditory object can build up across seconds. Given that attention operates on perceptual objects, this perceptual buildup may alter the efficacy of selective auditory attention over time. Here, we measured identification of a sequence of spoken target digits presented with distracter digits from other directions to investigate the dynamics of selective attention. Performance was better when the target location was fixed rather than changing between digits, even when listeners were cued as much as 1 s in advance about the position of each subsequent digit. Spatial continuity not only avoided well known costs associated with switching the focus of spatial attention, but also produced refinements in the spatial selectivity of attention across time. Continuity of target voice further enhanced this buildup of selective attention. Results suggest that when attention is sustained on one auditory object within a complex scene, attentional selectivity improves over time. Similar effects may come into play when attention is sustained on an object in a complex visual scene, especially in cases where visual object formation requires sustained attention.

Fig. 1.

Overall performance is best when spatial location is fixed between digits; moreover, even up to 1 s of advance knowledge of where to direct spatial attention does not overcome the cost of switching spatial attention. Across-subject mean scores (±SEM) for Exp. 1, where the target voice switches between digits (Upper), and Exp. 2, where the target voice is fixed across digits (Lower). Data are plotted as a function of interdigit delay for conditions F (squares and solid lines), SS (circles and dotted lines), and SL (triangles and dashed lines).

Fig. 2.

The cost of switching spatial attention decreases with interdigit delay but is always positive. Moreover, the cost of switching tends to be greater when voice quality is fixed between digits (Exp. 2) (Lower) than when the voice changes between digits (Exp. 1) (Upper), especially at short interdigit delays. Each plot shows the across-subject mean difference in performance (±SEM) between condition F and each of the conditions SS (circles and dotted lines) and SL (triangles and dashed lines).

Fig. 3.

When the target sequence is continuous in spatial location, performance improves from digit to digit, an effect that is enhanced when the target voice quality is continuous between digits. Across-subject mean scores (±SEM) as a function of temporal position for Exp. 1 (with random voice) (Upper) and Exp. 2 (with fixed voice) (Lower). The four plots within each row show data for the four different interdigit delays. Data are plotted as a function of temporal position within the target sequence for F (squares and solid lines), SS (circles and dotted lines), and SL (triangles and dashed lines).

Fig. 4.

Spatially directed attention filters out sources from the wrong direction, and this filtering becomes more refined over time when target location is fixed across digits. (Upper) Percentage of responses that corresponded to a digit presented from a nontarget loudspeaker are shown as a function of the distance between the target loudspeaker and the loudspeaker presenting the reported digit. Responses that did not correspond to any of the presented digits are shown at the far right (rand). Responses are pooled across all subjects and all delays for F (squares and solid lines), SS (circles and dotted lines), and SL (triangles and dashed lines). (Lower) Incorrect responses in the F condition as a function of distance between the target loudspeaker and the loudspeaker presenting the reported digit for each temporal position within the sequence (light to dark gray showing results for target digits 1–4). Responses are pooled across all subjects and all delays.

Fig. 5.

Schematic of the auditory and visual stimuli for the fixed and switching conditions. Five different digits were presented simultaneously from the five loudspeakers (circles) in each of four temporal positions of the stimulus. During each of the four temporal positions, the LED on one loudspeaker was illuminated (filled circle) to indicate the target digit. (Upper) In the fixed condition, the target digit came from the same loudspeaker in each of the temporal positions. (Lower) In the switching conditions, the target came from a different random loudspeaker in each temporal position. The visual cue from the target LED came on simultaneously with the auditory stimuli in the F and SS conditions but preceded the auditory stimuli in the SL condition (diagram not shown).