Quantifying temporal ventriloquism in audio-visual rhythm perception

We used a rhythm perception paradigm to quantify the effects of small temporal discrepancies between audio-visual stimulus pairs. In this paradigm, observers had to align the onset of a target stimulus (position 3) within a rhythmic sequence of four markers (positions 1, 2, 4, and 5). In Exp 1, the modalities of the markers and targets were crossed in a 2X2 design. In unimodal conditions, the target was placed accurately for both audio (click) and visual (flash) conditions, but in bimodal conditions, there was a consistent 25-30 ms bias in target placement. In Exp 2, the markers were bimodal with various SOAs between the audio and visual components, and the targets were visual flashes. The results demonstrated temporal ventriloquism in which adjustment of the visual target was affected by the timing of the audio components of the bimodal markers, even when observers were told to use the visual components only.