Opening Up Vision: The Case Against Encapsulation

Abstract

Many have argued that early visual processing is encapsulated from the influence of higher-level goals, expectations, and knowledge of the world. (Early vision is thought to result in perception of three-dimensional shapes and surfaces, prior to object recognition and categorization.) Here we confront the main arguments offered in support of such a view, showing that they are unpersuasive. We also present evidence of top–down influences on early vision, emphasizing data from cognitive neuroscience. Our conclusion is that encapsulation is not a defining feature of visual processing. But we take this conclusion to be quite modest in scope, readily incorporated into mainstream vision science.