Abstract
Previous studies showed that contextual cueing expedited visual search. Tseng and Li (2004, Perception & Psychophysics) demonstrated that the decrease in search time occurred as a result of the reduction in the number of saccadic eye movements during an ineffective search phase. Context has no effect in the effective search phase in which successive fixations move closer to the target. In the present study we explored in computational modeling how three alternative factors mediate eye movements during context-guided visual search: attention prioritization, the rate of visual resolution decline in the periphery, and magnitude of inhibition of return (IOR). The selection of successive fixation location is based on the optimal information gain hypothesis: the search model evaluated each possible location and selected one that will maximize information gain from surrounding stimuli. The stimulus expected information gain increases by attention prioritization but declined with increase of eccentricity, and the rate of the decline was contingent on a scaling constant. If a stimulus had been examined, its information decreased by the magnitude of IOR. The model parameters were calibrated by the observed distribution of the amplitude of saccade in no-cued condition and then tested in cued condition. The results showed that both the change of the decline rate of visual resolution and the magnitude of IOR lead to a shortening of the ineffective search phase and a change in the slope of the effective search phase, in disagreement with the empirical data. On the other hand, attention prioritization shortened the duration of the ineffective search phase but left the slope of the effective search phase unaltered. These simulation results suggest that attention prioritization can occur during our visual search paradigm in which the salient task-irrelevant contextual cues serve as a causal venue through which context-guided learning takes place.