Abstract
Despite an ever-changing and noisy visual world, our visual experience appears remarkably continuous and stable over time. Recently, orientation and face perception were shown to be biased towards previous percepts of orientation and faces, respectively (Fisher & Whitney, 2014; Liberman et al., 2014). This serial dependence effect was proposed as a mechanism to facilitate perceptual stability, compensating for variability in visual input. A key unanswered question is whether serial dependence operates in complex environments. Here, we tested whether serial dependence occurs in context, examining two main cases of contextual modulation. First, crowding: the deleterious influence of nearby objects on object perception. We presented a sequence of oriented Gabors and asked observers to adjust the orientation of a bar to match each Gabor's orientation. In the crowding condition, the Gabor was flanked by four additional Gabors. In the non-crowded condition, a single Gabor was displayed. We observed a classical serial dependence effect in the non-crowded condition: perceived orientation was biased toward the orientation of the previously presented Gabor. However, when the previous Gabor was crowded, the serial dependence effect on the following single Gabor ceased. Second, summary statistics: the beneficial ensemble perception of multiple objects. We presented a 3x3 array of nine Gabors with random local orientations, and asked observers to adjust a bar's orientation to match the ensemble orientation. We found strong evidence for serial dependence: the reported ensemble orientation was pulled toward the orientation of the previous Gabor array. Further controls showed that serial dependence occurred at the ensemble level, and that observers averaged 5/6 Gabors per trial. Our results show that serial dependence does not occur in crowded viewing conditions. However, serial dependence can still occur between summary statistical representations. Hence, we provide a mechanism through which serial dependence can maintain perceptual stability in complex environments.
Meeting abstract presented at VSS 2016