Abstract
When exploring an everyday scene, viewers typically exhibit an object-to-object scanpath, presumably to enable them to identify individual objects and develop a coherent percept of the situation. Does this mean that object processing in peripheral vision is limited to an analysis of low-level features, signalling potentially interesting fixation locations? We report a series of experiments showing that peripheral object analysis is automatically influenced by the object's semantic consistency with the surrounding scene. First, object-scene consistency is demonstrated to influence the processing benefit derived from a peripheral object preview. When viewers are given a presaccadic peripheral preview of an object, postsaccadic object fixation times are reduced relative to a control where no preview was available. This preview benefit is stronger for scene-consistent objects. Second, semantic consistency is shown to affect the object's saliency as a potential saccade target. Frequency and latency of voluntary and involuntary saccades towards objects are influenced by the object's semantic consistency with the scene. Consistent objects are more salient targets for involuntary saccades and less salient for voluntary saccades. These two sets of findings are explained in terms of a larger useful field of view for context-supported objects, which is based on rapid access to a global scene theme and activation of object representations linked to that theme. This is assumed to facilitate both the pre-attentive creation of object files for the consistent objects in the scene and the attentive peripheral processing of such an object when it is selected as a potential saccade target.
Supported by Concerted Research Effort Convention GOA 98/01 of the Research Fund K.U.Leuven, and the Belgian Programme on Interuniversity Poles of Attraction Contract P4/19.