Over the past 15 years there have been considerable advances in understanding visual processing. These advances have been fuelled by such developments as, among others, the use of powerful computing systems that enable processing algorithms to be carried out in real time, the application of experimental paradigms developed with normal observers to provide more accurate analyses of patients with impaired vision, and the development of new learning algorithms that can be applied in artificial neural networks. The study of vision is a study in which researchers from different disciplinary backgrounds - computer science, neuropsychology, cognitive science, artificial intelligence - can usefully cooperate and share the research results and developments from the individual fields. Indeed, it is likely that future advances will depend on just such interdisciplinary collaborations. "Understanding Vision" brings together leading researchers from the different disciplines. The contributors attempt to provide a state-of-the-art review of work being carried out in their field, together with a discussion of how this work fits in to the broader perspective of vision research.