Abstract

According to an emerging vision in computational cognitive neuroscience, perception (rich, full-blooded, world-presenting perception of the kind we enjoy) depends heavily on prediction. To visually perceive, if this schema is correct, is to meet incoming visual information with a set of matching top-down predictions that track the evolving visual signal across multiple spatial and temporal scales. This chapter first introduces this general explanatory schema, then displays some recent evidence for the schema, and discusses to what extent it marks a radical departure from previous (feature-detection based) models of perception. It ends by exploring some implications of the schema for questions concerning multimodal and crossmodal effects in sensory processing, and for an understanding of the deep and fundamental relations between perception, imagination, and understanding.