Recurrent connections facilitate occluded object recognition by explaining-away

Despite the ubiquity of recurrent connections in the brain, their role in visual processing is less understood than that of feedforward connections. Occluded object recognition, an ethologically critical cognitive capacity, is thought to rely on recurrent processing of visual information, but it remains unclear whether and how recurrent processing improves recognition of occluded objects. Using convolutional models of the visual system, we demonstrate how a distinct form of computation arises in recurrent-but not feedforward- networks that leverages information about the occluder to "explain-away" the occlusion-- i.e., recognition of the occluder provides an account for missing or altered features, potentially rescuing recognition of occluded objects. This occurs without any constraint placed on the computation and is observed both across a systematic architecture sweep of convolutional models and in a model explicitly constructed to approximate the primate visual system. Building on these computational results, we conduct a behavioral experiment to study explaining-away in humans, and find evidence consistent with explaining-away. Finally, to demonstrate a specific mechanism for explaining-away, we develop an experimentally inspired recurrent model that recovers fine-grained features of occluded stimuli by explaining-away. Recurrent connections capability to explain away may extend to more general cases where undoing context-dependent changes in representations could benefit perception.