Eye gaze during route learning in a virtual task

Looking behavior during route learning is uniquely attuned to information relevant fornavigation, which unfolds continuously over time. To better understand how visual informationis used to learn routes, we tracked eye-position in participants on a guided walk through asimulated urban environment. The route was presented twice to participants in either a naturalsequential order or a randomly scrambled order. After viewing the route, participants werepresented with images of the intersections and asked to report the correct direction to continuetravel along the route. Predictably, the sequential group performed much better on memory forthe route than the scrambled group. Analyses of the eye-data revealed reliable differencesbetween groups during route learning but no differences during scene viewing in the memory testtrials. Specifically, while both groups looked at objects such as people and cars during routelearning, looking at the path ahead was highly predictive of being in the sequential group,whereas looking at buildings ahead was predictive of being in the scrambled group. The resultssuggest that route learning involves encoding, and anticipating, information about the path to-betraveled, perhaps reflecting formation of a temporally continuous representation of the route. Incontrast, those in the scrambled group appeared to rely on building landmarks to attempt tocreate a map of the route, albeit unsuccessfully.