Dissociable neural correlates of multisensory coherence and selective attention.

Previous work has demonstrated that performance in an auditory selective attention task can be enhanced or impaired, depending on whether a task-irrelevant visual stimulus is temporally coherent with a target auditory stream or with a competing distractor. However, it remains unclear how audiovisual (AV) temporal coherence and auditory selective attention interact at the neurophysiological level. Here, we measured neural activity using electroencephalography (EEG) while human participants (men and women) performed an auditory selective attention task, detecting deviants in a target audio stream. The amplitude envelope of the two competing auditory streams changed independently, while the radius of a visual disc was manipulated to control the audiovisual coherence. Analysis of the neural responses to the sound envelope demonstrated that auditory responses were enhanced independently of the attentional condition: both target and masker stream responses were enhanced when temporally coherent with the visual stimulus. In contrast, attention enhanced the event-related response (ERP) evoked by the transient deviants, independently of AV coherence. Finally, in an exploratory analysis, we identified a spatiotemporal component of ERP, in which temporal coherence enhanced the deviant-evoked responses only in the unattended stream. These results provide evidence for dissociable neural signatures of bottom-up (coherence) and top-down (attention) effects in AV object formation. Significance Statement Temporal coherence between auditory stimuli and task-irrelevant visual stimuli can enhance behavioral performance in auditory selective attention tasks. However, how audiovisual temporal coherence and attention interact at the neural level has not been established. Here, we measured EEG during a behavioral task designed to independently manipulate AV coherence and auditory selective attention. While some auditory features (sound envelope) could be coherent with visual stimuli, other features (timbre) were independent of visual stimuli. We find that audiovisual integration can be observed independently of attention for sound envelopes temporally coherent with visual stimuli, while the neural responses to unexpected timbre changes are most strongly modulated by attention. Our results provide evidence for dissociable neural mechanisms of bottom-up (coherence) and top-down (attention) effects on AV object formation.