Contextual cueing based on specific and categorical properties of the environment

During the analysis of a visual scene, top-down processing is constantly directing the subject's attention to the zones of interest in the scene. The contextual cueing paradigm developed by Chun and Jiang ( 1998) shows how contextual regularities can facilitate the search for a particular element via implicit learning mechanisms. The study presented here reports three experiments that used this paradigm. Experiment 1 showed that regularities in the specific elements of the context can act as cues to the location of the target. Experiments 2 and 3 explored a novel aspect of contextual regularities, namely semantic regularities based on the categorization of contextual elements. Contextual cueing effects were obtained when semantic-category membership of the context predicted the target location. Moreover, in all three experiments, contextual cueing effects were obtained implicitly. The results suggest that in target-detection tasks, implicit learning can be based not only on the specific constituents of the context, but also on the semantic categories of those constituents, depending on their predictive power.