The McGurk effect demonstrates the existence of a fusion process in audiovisual speech perception. We consider here the assumption that fusion is controlled by an upstream conditional binding process, which can block fusion in the case of strong audiovisual inconsistencies, as in the case of a dubbed film. To test this hypothesis, we designed two experiments in which a consistent or inconsistent audiovisual context is placed before McGurk stimuli, and we show that inconsistent contexts can remove the McGurk effect.

Keywords: McGurk effect, binding, multisensory fusion, AV speech perception and scene analysis.