Purchase this article with an account.
Kevin Price, Satoru Suzuki, Marcia Grabowecky; Costs of Switching Scene Category in Real-World Visual Search. Journal of Vision 2011;11(11):1330. doi: 10.1167/11.11.1330.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Real-world visual search is heavily dependent on context, and experienced searchers use their knowledge of scenes to facilitate search. While it is known that visual search is facilitated by repetition of a specific scene, scenes in the real world often differ across searches. To compensate for this, people may employ search strategies that vary with contextual categories as well as for a specific scene context. If so, switching from a familiar scene category to a new category may exact a search cost. In other words, if search for birds in urban scenes has become proficient, birds in a forest scene may initially be difficult to find. Several predictions follow from this intuition: (1) Repeated search in a series of novel but categorically similar real-world scenes should facilitate visual search, and (2) A switch to a different real-world scene category should cause a drop in search efficiency. To test these predictions, we had 30 participants search for 20 different examples of birds in 40 different scenes of three different categories: urban, forest, or indoor, with 3 blocks of search trials blocked by category. We observed a significant cost of switching scene categories in search efficiency when search had become proficient in the prior category. Furthermore, search was particularly inefficient when the target bird appeared in the bottom quarter of the scene, consistent with prior results suggesting that knowledge of real-world target probability influences expectations about target locations. The cost of lower scene locations on search was reduced for the indoor scenes, suggesting that knowledge of scene context influences location expectations in visual search. These results suggest that categorical scene context and prior knowledge of target location likelihood interact to affect search strategy and efficiency.
This PDF is available to Subscribers Only