Abstract
Reinforcement has been shown to play a role for eye movement control in visual search using simple stimuli. For instance, saccades made towards specific locations in a uniform background were followed with a rewarding tone (Chukoskie et al., 2013) or specific sequences of saccades between different items led to target presentations (Paeye & Madelain, 2012). We tested whether visual consequences, displayed after saccades with specific landing positions or movement vectors, can change eye movement behavior in visual search performed in a homogeneously structured background. Participants were instructed to search for a Gabor patch in a 1/f background noise where no target was visible at the beginning of the trial. We used a gaze-contingent display to present the target at saccade offset more often after saccades that landed in a specific quadrant (experiment 1) or after saccades that were moving at a specific angle (experiment 2). Baseline trials did not contain any targets and were cancelled after the execution of ten saccades. Between baselines trials and trials at the end of reinforcement, the proportions of saccades towards the frequently reinforced quadrant or those moving at the frequently reinforced angle nearly tripled. Moreover, changing the reinforcement criteria in favor of other landing positions or movement vectors induced corresponding increases in the proportions of saccades. These results show that seeing the target after specific eye movements determines saccade sequences during visual search in a structured background. They extend the work of Najemnik and Geisler (2005) who elaborated a Bayesian model to account for human visual search. These authors proposed that an ideal searcher chooses fixation locations such that information gain is maximized. By directly manipulating saccadic consequences we showed that an operant learning process can also guide changes in visual search.
Meeting abstract presented at VSS 2014