Does the McGurk effect always fool the brain?

Lebib, R.1,2, Papo, D.3, Douiri, A.4, de Bode, S.5, and Baudonnière, P.-M.2
1Cognitive Neurosciences and Psycholinguistics Team, University of La Laguna, Tenerife, Spain; 2Cognitive Neurosciences and Brain Imaging Lab, CNRS UPR 640, Paris, France; 3Cognitive Psychology Laboratory, CNRS FRE 2071, Marseille, France; 4Department of Computer Science, UCL, London, UK; 5Neurology Department, UCLA, CA
E-mail: Riadh.Lebib@chups.jussieu.fr

The key role played by visual speech information in our subjective perception of auditory speech input is clearly illustrated by the audiovisual speech illusion known as the McGurk effect. Typically, subjects do not notice the lack of congruence between the acoustic and visual input and this illusion is known to be quasi-irrepressible. However, the question of a possible non-conscious detection of this existing incongruence is still open. Here we show that while subjects mainly identified the McGurk stimuli according to the illusory percept they are supposed to yield, the event-related brain potential responses recorded for these stimuli resulted in larger amplitudes for a late positive component, namely the P600, as compared to those elicited by congruent audiovisual speech stimuli with the same auditory information. When the task consisted in explicitly detecting the incongruence between the bimodal speech inputs, however, this modulation in the P600 amplitude no longer occurred, although again the McGurk stimuli were mainly categorized as a coherent percept. These results are in accordance with recent findings showing that the P600 reflects cerebral repair and reanalysis processes. Moreover, the results indicate that this brain response, which may be reflecting a non-conscious detection of the non-coherence between the bimodal speech inputs for McGurk stimuli, seems to depend on the focus of attentional resources.