Donald Glaser

Professor (Molecular & Cell Biology)

Email: glaserda@berkeley.edu

Research areas: Cognitive and Behavioral Neuroscience

Our main goal is to construct computational models of the human visual system based on experimental knowledge of its abilities and architecture. In vision experiments, observers examine visual displays and judge depth, color, velocity, texture, flow, or just the presence or absence of a signal. In a recent study of stereopsis, our subjects judged the relative depth of two adjacent dots in the center of a computer screen. When the "scene" was enclosed in a "picture frame" 50 degrees wide and tilted optically, a systematic bias was discovered in judgments of the depths of the test dots, even though the subjects didn�t know that the frame was being tilted. Thus a powerful subliminal cue was at work. Depth judgments are found in these experiments to depend on items in widely separated parts of an image, not only on the particular objects being examined, proving that the classical theory of stereopsis is incorrect.

The perceived color of a part of an image is known to depend on the colors of adjacent areas. We have also shown that judgments of the speed of moving objects can be greatly affected by blinking lights far from the moving object. Thus local judgments of depth, color, and motion can all depend on items in other parts of an image. Any theory must therefore explain how �global� information affects �local� perceptions. No such theory is known at present, but these experiments demonstrate the need unambiguously.

Since these and many other visual processes are complex and non-linear, conventional mathematical models offer little hope of providing a useful bridge between psychophysics and neurobiology. We therefore use computer simulations of models which mimic the real human visual system. When visual images are presented to these models, they produce responses like those we ask of our subjects.

Selected Publications

Kumar, T., and Glaser, D. A. 1995. Depth discrimination of a crowded line is better when it is more luminant than the lines crowding it Vision Research 35(5): 657–666.

Kumar, T., and Glaser, D. A. 1993. Some temporal aspects of stereoacuity Vision Research 34(7): 913–925.

Kumar, T., and Glaser, D. A. 1993. Initial performance, learning, and observer variability for hyperacuity tasks Vision Research 33(16): 2287–2300.