Abstract
In dynamic environments, gaze patterns are often driven by competing task goals (Hayhoe and Ballard, TICS, 2005). However, it is not clear how observers determine gaze priorities, especially in uncertain environments. We studied whether experience with events in the environment influences distribution of gaze in a real environment. Subjects walked along a circular path with pedestrians who sometimes veered onto a collision course with the subject for a period of 1 sec (approximately). Subjects were given experience with 2 types of pedestrians: one who always veered on a collision course on every circuit around the path (Rogue pedestrian) and one who never veered (Safe pedestrian). In a subsequent trial the roles of Safe and Rogue pedestrian were interchanged. Experience with the Rogue pedestrian elevated fixation probabilities of the Safe pedestrian to 70% (instead of 50% without experience). Conversely, experience with the Safe pedestrian lead to 80 % fixation probability of the Rogue pedestrian (instead of 89% without experience).When subjects had an added task of following a lead pedestrian, overall fixation rates on pedestrians were reduced by about 17%, but the effect of experience (with Rogues and Safe pedestrians) was maintained. These results confirm previous observations in a virtual environment (Jovancevic et al, JOV, in publication), although overall rates of fixating veering pedestrians were higher in the real environment. Thus subjects are sensitive to the probabilistic structure of environmental events and use this information to adjust gaze priorities in situations when there are multiple task goals, consistent with predictions of computational models of attentional allocation while walking (Rothkopf and Ballard, VSS, 2005). This anticipatory allocation of gaze reduces the need for reactive gaze control to unexpected events.
This research was supported by National Institutes of Health Grants EY-05729 and RR-09283