Crossmodal semantic congruence can affect visuo-spatial processing and activity of the fronto-parietal attention networks

摘要

Previous studies have shown that multisensory stimuli can contribute to attention control. Here we investigate whether irrelevant audio–visual stimuli can affect the processing of subsequent visual targets, in the absence of any direct bottom–up signals generated by low-level sensory changes and any goal-related associations between the multisensory stimuli and the visual targets. Each trial included two pictures (cat/dog), one in each visual hemifield, and a central sound that was semantically congruent with one of the two pictures (i.e., either “meow” or “woof” sound). These irrelevant audio–visual stimuli were followed by a visual target that appeared either where the congruent or the incongruent picture had been presented (valid/invalid trials). The visual target was a Gabor patch requiring an orientation discrimination judgment, allowing us to uncouple the visual task from the audio–visual stimuli. Behaviourally we found lower performance for invalid than valid trials, but only when the task demands were high (Gabor target presented together with a Gabor distractor vs. Gabor target alone). The fMRI analyses revealed greater activity for invalid than for valid trials in the dorsal and the ventral fronto-parietal attention networks. The dorsal network was recruited irrespective of task demands, while the ventral network was recruited only when task demands were high and target discrimination required additional top–down control. We propose that crossmodal semantic congruence generates a processing bias associated with the location of congruent picture, and that the presentation of the visual target on the opposite side required updating these processing priorities. We relate the activation of the attention networks to these updating operations. We conclude that the fronto-parietal networks mediate the influence of crossmodal semantic congruence on visuo-spatial processing, even in the absence of any low-level sensory cue and any goal-driven task associations.