The purpose of the present experiment was to investigate the integration of sequences of visual and tactile events. Subjects were presented with sequences of visual flashes and tactile taps simultaneously and instructed to count either the flashes (Session 1) or the taps (Session 2). The number of flashes could differ from the number of taps by ±1. For both sessions, the perceived number of events was significantly influenced by the number of events presented in the task-irrelevant modality. Touch had a stronger influence on vision than vision on touch. Interestingly, touch was the more reliable of the two modalities—less variable estimates when presented alone. For both sessions, the perceptual estimates were less variable when stimuli were presented in both modalities than when the task-relevant modality was presented alone. These results indicate that even when one signal is explicitly task irrelevant, sensory information tends to be automatically integrated across modalities. They also suggest that the relative weight of each sensory channel in the integration process depends on its relative reliability. The results are described using a Bayesian probabilistic model for multimodal integration that accounts for the coupling between the sensory estimates.

Sequences of auditory beeps and tactile taps were simultaneously presented and participants were instructed to focus on one of these modalities and to ignore the other. We tested whether (i) the two sensory channels bias one another and (ii) the interaction depends on the relative reliability of the channels. Audition biased tactile perception and touch biased auditory perception. Lowering the reliability of the auditory channel (i.e. the intensity of the beeps) decreased the effect of audition on touch and increased the effect of touch on audition. These results show that simultaneous auditory and tactile stimuli tend to be automatically integrated in a reliability-dependent manner.