The effects of attention on the temporal integration of multisensory stimuli

Abstract

In unisensory contexts, spatially-focused attention tends to enhance perceptual processing. How attention influences the processing of multisensory stimuli, however, has been of much debate. In some cases, attention has been shown to be important for processes related to the integration of audio-visual stimuli, but in other cases such processes have been reported to occur independently of attention. To address these conflicting results, we performed three experiments to examine how attention interacts with a key facet of multisensory processing: the temporal window of integration (TWI). The first two experiments used a novel cued-spatial-attention version of the bounce/stream illusion, wherein two moving visual stimuli with intersecting paths tend to be perceived as bouncing off rather than streaming through each other when a brief sound occurs near in time. When the task was to report whether the visual stimuli appeared to bounce or stream, attention served to narrow this measure of the TWI and bias perception toward "streaming". When the participants' task was to explicitly judge the simultaneity of the sound with the intersection of the moving visual stimuli, however, the results were quite different. Specifically, attention served to mainly widen the TWI, increasing the likelihood of simultaneity perception, while also substantially increasing the simultaneity judgment accuracy when the stimuli were actually physically simultaneous. Finally, in Experiment 3, where the task was to judge the simultaneity of a simple, temporally discrete, flashed visual stimulus and the same brief tone pip, attention had no effect on the measured TWI. These results highlight the flexibility of attention in enhancing multisensory perception and show that the effects of attention on multisensory processing are highly dependent on the task demands and observer goals.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 59 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.