Abstract
Imagine that you are searching for a red pen, and you know it could be on either of two desks. The top of one desk is clean, while the other desk is cluttered with papers, books and coffee cups. Common sense suggests an efficient visual system would not waste any time on the clean desk. Here we test whether optimal fixation strategies can be employed using an analogous experimental setup: stimuli consisting of arrays of line segments, homogenous on one half, and heterogeneous on the other. Using peripheral vision only, observers could accurately detect whether there target was present or absent in the homogeneous but not the heterogeneous array. If search is optimal, the proportion of fixations directed to the heterogeneous side on any given trial should be approximately one. One out of the 14 observers in this experiment followed this strategy, but the rest did not. Individual differences in the relative proportion of fixations on the homogenous vs. heterogeneous side during the first five fixations on target absent trials were highly correlated with how quickly observers detected the target when it was present, both for the heterogeneous (r=0.91, p< 0.001) and homogeneous (r=0.79, p=0.001) side of the display. This correlation clearly demonstrates that observers who have a tendency to search the homogenous side were slower to find the target when it was present. In a follow-up experiment, we directly compared search of uniform homo- and heterogeneous arrays to search of half homo- half heterogeneous. Reaction time should be the average of RT on the two uniform displays, but observers actually take 20% longer than predicted (t(13)=30.95, p< 0.001). Together the results suggest superfluous fixations are directed to the homogeneous side, inconsistent with optimal search.
Meeting abstract presented at VSS 2016