TOWARDS A THEORY OF COMMONSENSE VISUAL REASONING

In this paper we propose a cognitive architecture underlying visual perception and mental imagery that explains analog mental imagery as well as symbolic visual representations. This architecture has an inherent duality: it can represent and manipulate visual aspects of objects in both symbolic and imaginal forms. It is argued that this architecture resolves many issues raised by the analogic-propositional debate in cognitive psychology. Then we discuss a computational theory, inspired by this architecture, for solving a class of commonsense spatial problems. This involves a novel approach to spatial reasoning that advocates the use of representations with both symbolic and imaginal aspects and visual cases.