台大心理系

An important issue facing the empirical study of consciousness concerns how an external piece of information gains access to conscious processing. Previous approaches to this issue have focused mainly on the aware versus unaware dichotomy, concerning how the brain differentially responds to subliminally-presented and supraliminally-presented stimuli. However, according to classic hierarchical models, a stimulus may consist of different levels of representation. In this study, we examined whether conscious access to multiple levels of stimulus representation is mutually independent or shares a single underlying mechanism with the magnetoencephalography (MEG) technique. Participants were asked to view masked target faces that were presented briefly so that the viewer’s perceptual awareness alternated from consciously accessing facial identity in some trials (i.e., conscious face identification) to being able to consciously access facial configuration features but not facial identity in other trials (i.e., conscious face detection). Conscious access to facial identity and detection were correlated with a series of distinct neural events. Even before target presentation, different phase angles became locked to the two conscious processes. This effect was followed by stronger phase clustering in cases of conscious face identification immediately prior to target presentation. Our recent analysis further reveal that those pre-stimulus activities may be partly attributed to the effect of preceding context. First, the behavioral results showed that the performances of both conscious face identification and detection were biased by the preceding context shaped by preceding responses. Specifically, participants were more likely to respond “not seeing the targets” when they made the same responses in preceding trials. This observation is supported by MEG data which showed that preceding contexts differentially modulated the above patterns of phase adjustment. In conclusion, dissociable forms of conscious perception may arise when accessing different levels of stimulus representations and moreover, the context provided by preceding responses may slant current conscious judgement toward assimilative consequences. This view also challenges the classic hierarchical model of face processing in which face detection is a necessary first step for facial identity recognition.