Abstract
Introduction: Where do people look when searching for multiple targets under time pressure? Previous studies showed that when there are multiple targets, fixating the most uncertain location is most efficient, as it maximizes information gained. However, observers tend to use the suboptimal strategy of fixating likely target locations (Verghese, 2012). We test the generality of this finding using a simpler task that predicts a stronger contrast in performance by the two alternative strategies. Method: Targets (vertical Gabors) occurred independently at each of 2 locations - 5 degrees left and right of center. Each location also had noise of low or high contrast. Two observers (one author; one participant with practice in other eye movement studies) reported the presence or absence of a target at each location. Brief presentation time (400 msec) typically allowed a single saccade. After stimulus presentation, observers manually selected target locations and received performance feedback for each location. Different strategies make different behavior predictions. The strategy favoring uncertain locations inspects high noise locations - independent of target presence. The strategy favoring likely target locations inspects target-present locations, particularly when targets are highly visible (i.e., in low noise). Results: One observer did not show a clear strategy preference. The other, practiced, observer fixated uncertain locations more frequently. In part this might be because eccentric targets were less visible for this observer, making saccades more informative. Nonetheless, it is important to note her clear preference for uncertain locations. Conclusion: These results contrast with findings that observers choose likely (not uncertain) locations when planning sequences of saccades. Rather, in our study one observer tended to fixate uncertain locations. This suggests that eye-movement strategies are not completely outside of cognitive control and that observers can be efficient in a simple task where they plan a single saccade.
Meeting abstract presented at VSS 2013