Given the inherently noisy nature of sensory signals, the brain must maintain a degree of tolerance for temporal discrepancies when combining multisensory information. While the limits of this temporal window of integration are well established, little is known about how they are affected by sensory experience. Powers et al. (2009) reported that training on an audiovisual simultaneity task narrows the temporal window of integration. However, because subjects were trained on the same task that was used to measure their integration windows, it is unclear whether this effect reflects a genuine change in multisensory integration or a simple improvement in task performance. Here, we examine whether perceptual training affects multisensory integration using independent training and test measures. In pre- and post-training sessions, participants were required to report the position of an auditory noise burst while ignoring a visual stimulus, across a range of stimulus onset asynchronies. The ventriloquist effect induced by the visual stimulus was greatest for simultaneous stimuli and declined with increasing inter-stimulus asynchrony. Training consisted of three 45-minute sessions where participants practiced a 2-IFC audiovisual simultaneity task. Training resulted in narrowing of the time window of integration comparable to that observed by Powers et al., as well as a reduction in the amplitude of the peak ventriloquist effect. Unisensory localisation thresholds were unaffected by training, ruling out changes in auditory and visual sensitivity as the cause of improvements. These findings demonstrate that perceptual training can affect multisensory integration and have implications for the future development of multisensory training protocols.

multisensory integration; perceptual learning; temporal processing