Browse Books

Reviewer: Robin Jeffries

.abstract The study of visual cognition has seen enormous progress in the past decade, bringing important advances in our understanding of shape perception, visual imagery, and mental maps. Many of these discoveries are the result of converging investigations in different areas, such as cognitive and perceptual psychology, artificial intelligence, and neuropsychology. This volume is intended to highlight a sample of work at the cutting edge of this research area for the benefit of students and researchers in a variety of disciplines. The tutorial introduction that begins the volume is designed to help the nonspecialist reader bridge the gap between the contemporary research reported here and earlier textbook introductions or literature reviews. — From the Preface This book is actually a collection of papers previously published as a special edition of the journal Cognition. As an introduction to a new and exciting research area for the technically oriented reader, this book is to be highly recommended. The six chapters, actually independent papers, consist of a literature review and five research reports that span many important issues in the field—from models of shape recognition to the physiological basis for visual mental imagery. All of the authors take a computational approach to understanding visual cognition, and several of the papers contain specific computational models. The introductory chapter, Visual Cognition: An Introduction, by S. Pinker, is billed as a tutorial introduction to the field. I would categorize it more as a literature review. This is a highly volatile area, where knowledge and theories are rapidly changing, and where there are many more questions than answers. Pinker lays out the issues quite well, but he has packed a large amount of information into the paper. It serves very well as a review for someone generally familiar with the area; pointers to the original literature are numerous. The serious novice will find this challenging reading, but it is reasonably complete, self-contained, and accessible for those willing to make the effort. It is not a suitable introduction for the casually interested layperson. The other chapters are a varied sampling of current research in the field. Parts of Recognitions, by D. D. Hoffman and W. A. Richards, describes a theory of object recognition by decomposing shapes into parts based on concave discontinuities and points of maximum negative curvature. Visual Routines, by S. Ullman, attempts to identify some of the primitive operations of the visual recognition system from computational, psychophysical, and physiological evidence. The author then considers how visual routines driven by particular computational goals could be built from this core set of primitive operations. Upward Direction, Mental Rotation, and Discrimination of Left and Right Turns in Maps, by R. N. Shepard and S. Hurwitz, looks at the sorts of reference frames people use to determine whether a turn shown on a map is a right or left turn. The authors demonstrate that the upward direction plays a privileged role in such reference frames. The last two papers deal with mental imagery. Individual Differences in Mental Imagery Ability: A Computational Analysis, by S. M. Kosslyn, J. Brunn, K. R. Cave, and R. W. Wallach, applies the Kosslyn and Schwartz theory of mental imagery (see [1]) to the understanding of how people differ in their ability to use mental images. The authors demonstrate that imagery is a skill with several distinct components. They show that the subskills posited by the Kosslyn and Schwartz theory do an excellent job of accounting for the variability in imagery ability among individuals. The Neurological Basis of Mental Imagery: A Componential Analysis, by M. J. Farah, reanalyzes previous reports of mental imagery deficits in cases of brain damage. Again using the Kosslyn and Schwartz theory of imagery, the author is able to show reasonable statistical connections between the loss of particular imagery subskills and damage to specific portions of the brain. This book would be highly useful for an advanced graduate seminar in computational models of visual cognition or a more general course that sampled computational models in various domains. It could also serve as an overview of the field for someone working in closely related areas (for example, someone building a computational model that included or interacted with a visual processing component). I would not recommend it as an introduction to the field unless the prospective reader was highly conversant in one or more of the following domains: neurophysiology of vision, cognitive psychology, or artificial intelligence.