Abstract
Past research has shown that when no immediate visual information is available in the periphery, scene context typically dominates search strategies (Castelhano & Henderson, 2007). In contrast, other studies have shown that when peripheral information is available, object content plays a significant role in the selection of potential target locations (corresponding to high spatial frequency information; Parkhurst, Law & Niebur, 2002; van Diepen & Wampers, 1998). The present study examined how search strategies are differentially affected by scene context and the placement of object content. Participants searched for a target using a gaze-contingent moving-window paradigm. The participants saw the search scene foveally, while extra-foveally, the scene was manipulated across five conditions: (1)Full Scene: search scene excluding the target; (2)Fractioned Scene: search scene with clusters of objects removed; (3)Object Array: an array of the scene objects on a grey background; (4)Scene Structure: a structural representation of the scene; and (5)No Scene: a black screen. Thus, the Object Array provided only object content, the Scene Structure provided structural information without gist information, and the Fractioned Scene provided scene context with a smaller number of objects. As expected, search performance was best in the Full Scene condition and worst in the No Scene condition across eye movement measures. Interestingly, there were no differences found in latency to the target between Object Array and Fractioned Scene conditions, but both were slower than Full Scene. Because the Fractioned Scene stimuli do not include clusters of objects around the target, the pattern of results suggests an interaction between scene context and object content. The information provided by objects surrounding the target plays an important role in the selection of potential target locations. Further experiments will report on the role of target information in the periphery and its influence on the selection prioritization of object content based on scene context.
Meeting abstract presented at VSS 2012