Abstract
Observers use prior expectations to inform their searching behavior within an environment to facilitate efficient visual search (Aydemir et al., 2013). Search decisions may reflect this bias even when it is contradicted by recent experience (Võ & Wolfe, 2012; 2013). We asked whether searchers can learn to incorporate evidence that contradicts their prior expectations with practice. To estimate prior expectations, 185 participants rated how likely common household objects were to be found in each of 12 locations (6 in the living room and 6 in the kitchen). Then, a separate group of 10 participants searched through two virtual rooms (kitchen and living room) for one of 3 target objects. Objects were hidden from view in the scene and revealed by a mouseclick. Points were awarded for finding a target before the end of each 30s trial, and more points were awarded for searching quickly. Time was expended in clicking locations and traveling between rooms. Three search environments were created in which targets appeared in likely, unlikely, or uniform locations, where location likelihood was determined from the initial ratings. The three environments were tested in separate blocks of trials. We found that there were two types of searchers: those who adapted to the environment quickly when the episodic information did not match prior expectations, and those who required repeated searches in order to learn new sets of expectations. Results were consistent with a Bayesian decision model in which the prior expectations were either consistent with the ratings, inverted, or uniform. Searchers who took more time to adapt to the environment either did not construct a new prior to account for recent experience, or assumed all target locations were random. The findings indicate different individual preferences to form new expectations to fit recent experiences.
Meeting abstract presented at VSS 2016