Abstract
The “ventriloquist effect” (mislocalization of sound toward a visual stimulus) is consistent with statistically optimal integration of visual and auditory signals (Alais & Burr, Curr. Biol., 2004). Here we report that “temporal ventriloquism”, the tendency for a sequence of sounds to “capture” visual flashes in time (Shams & Shimojo, Nature, 2000), is also consistent with optimal integration. Subjects performed a temporal-bisection task, reporting whether the second (probe) stimulus in a 3-stimulus (800 ms) sequence was closer in time to the first or third. In single-cue sessions, the three stimuli were all either visual flashes or auditory tones. In two-cue sessions, all stimuli consisted of a flash and tone, presented simultaneously for the second stimulus, but at consistently different times for the first and third stimuli. The perceived point of bisection in the two-cue condition was determined more by tone than flash timing, but both cues influenced the bisection. The results were well predicted from optimal combination of the visual and auditory cues with relative weights derived from the single-cue thresholds. Importantly, bisection thresholds in the two-cue condition were significantly better than in either single-cue condition, strong evidence for inter-modal combination. Further discrimination experiments suggested that combination was not mandatory (Hillis et al., Science, 2002); rather subjects retain access to single-cue information as well as the combined information, calling into question the concept of mandatory temporal binding.