In visual search tasks, a target is detected faster when repeatedly presented within a particular context of distractors. Contextual cueing (Chun & Jiang 1998) shows that our brain keeps trace of previously encountered visual scenes with a high-capacity memory. Formed in a few trials, this memory can be used to guide attention and does not lead to any conscious knowledge. These findings challenge our understanding of the neural mechanisms of memory, attention and consciousness: how can a long-lasting neural memory be acquired in a few stimulus presentations? What are the neural mechanisms of this unconscious guidance of attention?
To investigate these issues, we recorded magneto-encephalographic (MEG) data in normal subjects performing a modified version of the contextual cueing paradigm. In our predictive condition each context is repeatedly associated with a single position of the target. In the non predictive condition, each context is associated on successive presentations with each possible position of the target. Therefore controlling for repetition effects, all contexts were repeated an equal number of times in our experiment. Reaction times were shorter in the predictive condition and this facilitation was implicit. We analyzed the acquisition of contextual associations during the first presentations and how these associations guide attention during the last presentations. Analysis of evoked fields shows that unconscious knowledge about the spatial regularities in vision can influence our perception within 110 ms after stimulus presentation. Further analysis of evoked and induced responses and their localization should help understand the brain mechanisms involved in unconscious influences on visual perception.
MC is supported by a grant from the Délégation Générale pour l'Armement