Crossmodal Predicton in Speech Perception在言语感知的跨模态的预测.pptVIP

* In the present study we hypothesize that if there exists prediction within the same modality, like is the case, and if predictive coding models can account for prediction at a phonological level, then predictive coding can occur across sensory modalities too. If cross-modal prediction occurs, then it would lead to advantages in comprehension and communicative skills, And it would help to explain some of the advantages of cross-modal, over unimodal perception * There are some indirect evidences that make us think that this crossmodal prediction can occurs. Like the study of van Wassenhove, in which she meassured ERPs, while she presented syllables audiovisually or only auditory. The presence of visual speech produces faster but smaller auditory evoked potentials (N1 and P2), compared to the auditory only condition. And also, the latency was smaller when the presented syllable was more visually salient, for ex. /pa/ than with more ambiguous syllables like /ka/. This is an example of prediction because /pa/ is visually a stronger predictor of the auditory syllable, than /ka/. * We wanted to test the existence of the predictive coding from visual-to-auditory modality, and in the opposite way, from audition-to-vision. For both directions we used the same kind of experiment, with the same task, that was an audio-visual matching task. Task : AV matching after an informative or non informative context. … no information * * Previous visual information allow participants to make the task faster, and it can only be due to some kind of visual prediction about what is coming in the sentence. As little the information extracted from the lip-reading is, it seems that is enough to help in the task.