Professor, Cognitive Sciences
School of Social Sciences
Donald Bren School of Information and Computer Sciences
School of Humanities
PH.D., Massachusetts Institute of Technology, 1983
Phone: (949) 824-6795
Fax: (949) 824-2307
University of California, Irvine
2169 Social Sciences Plaza A
Mail Code: 5100
Irvine, CA 92697
Machine and human vision; visual recognition; artificial intelligence; virtual reality; consciousness and cognition; shape from motion.
Distinguished Scientific Award of the American Psychological Association, 1989.
Troland Research Award of the National Academy of Sciences, 1994.
Who's Who in America.
The image formed by your eye has two dimensions: up/down and left/right. Yet you see the world in three dimensions. You add a dimension of depth to the flat image at your eye. But this means that you construct all the depth you see around you. You use motion, and shading, and stereo, and other forms of information to construct a world with depth.
I study how you do this, using several research methods. I build computational theories---mathematical models which can do the same kind of depth construction that you do. I prove theorems about their performance, their strengths and weaknesses. I implement them in computer programs that, in effect, endow the computer with the ability to "see" depth. I then compare the performance of these programs and the predictions of the mathematical theorems against the performance of human subjects in controlled visual experiments. I also compare them with what is known of the neurophysiology and neuroanatomy of the human visual system. The programs and mathematical models motivate new experiments. The experiments, in turn, suggest new mathematical models, in that interplay of theory and experiment that is the hallmark of science.
You construct all the depth you see. But then you construct everything that you experience visually: colors, motions, lines, objects, and scenes. I use the same interplay of theory and experiment to explore how you construct all you see, and how you recognize visual objects.
The results of this research are useful in several ways.
First, they tell us something about us---how we see. We want to understand who we are and how we tick, and this research addresses that natural curiosity.
Second, they lead to robot vision. The mathematical models and programs that emerge from this research can give computers the ability to "see" through the eyes of video cameras. The potential benefits of this are enormous, from automated assembly lines for manufacturing, to prosthetic devices for the blind, to the home robot that vacuums the house without vacuuming the cat.
Third, they lead to better displays of virtual reality. You construct all you see, and this research finds out how. This knowledge is critical to designers of virtual worlds, who want to "trick" you into creating the visual worlds they want you to see. You are the real creator of virtual reality, and designers of virtual reality displays must understand how you do it if they want compelling displays.
Fourth, they give new insights into the age-old problem of consciousness. Since before Plato we have wondered about the relationship of our conscious experience to the outside world. Which is primary, mind or matter? How are they related? By showing in detail how you construct your visual experience, these results suggest new ways of thinking about consciousness and its relation to the world.
Facial attention and spacetime fragments. Axiomathes, 2003, 13, 303-327. T. Davies, D. Hoffman
Flank transparency: The effects of gaps, line spacing, and apparent motion. Perception, 2002, 31, 1073-1092. D Wollschlaeger, A. Rodriguez, D. Hoffman.
Attention to faces: A change-blindness study. Perception, 2002, 31, 9, 1123-1146. T. Davies, D. Hoffman
Visual Intelligence: How We Create What We See. W.W. Norton, 1998.
Observer Mechanics. New York: Academic Press, 1989 (with B. Bennett and C. Prakash).
The interpretation of visual illusions. Scientific American, 1983, 249, 6, 154-162.
The interpretation of biological motion. Biological Cybernetics, 1982, 42, 3, 197-204 (MIT AI Memo) (with B. Flinchbaugh).
Structure from two orthographic views of rigid motion. Journal of the Optical Society of America, A, 1989, 6, 1052-1069 (with B. Bennett, J. Nicola, and C. Prakash).
Unity of perception. Cognition, 1991, 38, 295-334 (with B. Bennett and C. Prakash).
Institute for Mathematical Behavioral Science