Abstract
The gaze cascade effect, in which the gaze leans to the eventually chosen stimulus prior to the decision, was found by Shimojo et al. (2003). They also demonstrated that the manipulation of the eye movement biased the preferential decision toward a face with a longer duration and concluded that eye movements should give a positive feedback to preferential formation. The recent study, however, claimed that the preferential bias can be replicated without eye movements by considering the masking effect, thus the eye movement is not involved in decision making (Bird et al., 2012). To clarify this issue directly, we examined the influence of eye movements by using the gaze-contingent method. Similar to the previous studies, two faces were displayed side by side alternately for 300ms and 900ms respectively, then participants selected a more attractive face. In the central view condition, participants shifted their gaze to the presented face repeatedly. The probability of choosing the longer presented face was 58.3%. In the peripheral view condition, participants viewed two faces in the periphery while fixating the central cross. The bias (51.7%) was not significant. In the gaze-contingent condition, participants were instructed to view the presented face. But the whole stimulus was shifted according to the gaze position, so that the retinal image was identical to that in the peripheral view condition except eye movements. Interestingly, the bias to the longer presented face was significantly decreased to 42.5%. Prior to this condition, participants were trained to inhibit multiple saccades especially toward the longer presented face due to the limited width of the screen. This suppressive gaze shift might affect the decision making in the opposite way. Our findings suggest not only the occurrence of the eye movement but also the intention of the gaze may affect preferential decision.