The cross-modal spread of attention is highly automatic

Attention can spread across sensory modalities so that when attention is directed to the visual component of a multisensory object, its task-irrelevant auditory component will also be bestowed with enhanced processing, which operates in either a stimulus-driven or representation-driven manner contingent on whether the visual component is attended based on object category. To date, the visual-to-auditory attentional spreading has been widely hypothesized as an involuntary, automatic process, but this core hypothesis has yet to be tested. In particular, the supraliminal auditory components employed in prior studies might have received some sort of voluntary attention after being consciously detected, which impeded the examination of this hypothesis. The present event-related potential (ERP) study examined it in male and female humans (N = 30) by rendering bimodal objects' auditory components sub-threshold (and hence free from being actively attended) in an object-selective visual recognition task wherein ERP correlates of the representation-driven and stimulus-driven attentional spreading could be dissociated. Our ERP data showed reliable stimulus-driven (200-400 msec post-stimulus) and representation-driven (300-400 msec) attentional spreading to the sub-threshold auditory components, and the estimated neural generators of both included regions of the auditory cortex. Meanwhile, multivariate pattern analysis showed that the object categories could be decoded from ERPs to the inaudible sounds, confirming neural processing of these sounds in the absence of awareness. Collectively, these findings offer strong evidence that the visual-to-auditory attentional spreading is highly automatic, which paves the way for elucidating the psychophysiological mechanisms underlying cross-modal attentional spreading.