When a never-seen but less-occluded image is better recognized: evidence from same-different matching experiments and a model.

Empirical studies of human object recognition have found that a query image different from that previously seen almost invariably gives rise to inferior recognition performance than a query image identical to that previously seen. In the present study of same-different matching, we demonstrate that a query image of a face or a Chinese character that was less occluded than that previously seen can yield more accurate positive identification than a query image identical to that previously seen. However, when occlusion of the second image was further reduced, or when the faces were inverted, this effect disappeared. These findings indicate that the representation of a partially occluded object is effectively less occluded, and that the ability of the visual system to overcome occlusion is limited and dependent on its familiarity with the perceived object. A model with limited capacity to overcome occlusion was proposed to qualitatively account for the results.