Abstract
When searching in real world scenes, our visual system needs to efficiently sift through complex visual information to quickly find the target. One prevailing question is how the visual system prioritizes and processes information to perform tasks effectively. Researchers have postulated that scene representations consist of both background and foreground elements, where background elements provide a scaffold for more detailed foreground elements (Davenport & Potter, 2004; Henderson & Hollingworth, 1999; Munneke, Brentari & Peelen, 2013). Here, we defined background information as boundary defining elements such as walls, floors, ceilings (Oliva & Torralba, 2001) and foreground information as moveable objects nested inside the background (Henderson & Hollingworth, 1999). In the current study, we were interested in how scene information from different planes impacts search for targets (foreground versus background). We introduced a new stimulus set: chimera scenes, which have the foreground set of objects belonging to one scene category, and the surrounding background structure belonging to another. We posit that differences in how the background and foreground are processed will result in different search strategies across scene planes. Participants performed search in scenes that had either consistent (Normal) or inconsistent (Chimera) foreground and background contexts. Target objects could appear in either the foreground or background of the scene. Results showed an interesting effect of target placement. Although participants had a longer target latency for foreground than background targets, they had a shorter response times, had fewer fixations to the target, and had a more direct scan path to the target when the target was in the foreground. Moreover, participants were able to discern foreground targets from farther in the periphery than background targets. These results suggest that there are processing differences between searches for target in different planes of the scene. Implications for scene representations and search mechanisms will be discussed.
Meeting abstract presented at VSS 2018