Perception d’attributs des visages parlants par le nouveau-né humain

A visual preference method is used for studying how human newborns can process internal characteristics on speaking faces in a multimodal context. In a first experiment, two mobile faces associated with a speech sound are opposed: one is clear and the other is blurred. A preference for the clear face is observed. A second experiment contrasts two clear faces, showing lip movements respectively congruent and non-congruent with the associated speech sound. No preference is observed in this situation. These results show that newborns process mobile faces by orienting their attention on more information than that corresponding to very low spatial frequencies. It seems however that they cannot detect the correspondences between the lip movements and the associated sounds.