Toward a Computational Neuropsychology of High-Level Vision

摘要

Visual information processing in humans has recently been studied from three distinct perspectives. This paper develops a way of melding the approaches of Artificial Intelligence, Cognitive Psychology, and Neuropsychology, and explores the advantages of such a hybrid approach. Each of the individual approaches has its strengths and weaknesses, but these are different for the different approaches. It is argued in this paper that by combining the three, we are in a position to take advantage of each one's strengths and may be able to circumvent each one's weaknesses. Additional keyword topics include: mental imagery and computer models.