![]() |
Margaret Sereno Assistant Professor, Department of Psychology,
|
I am building partially prespecified multistage models of the visual system in which
response properties of higher stages develop as the model ³learns from experience.² Like
developing biological systems, such structured learning systems can extract environmental
regularities by combining information from lower levels to represent complex and abstract
properties of the input array and, in so doing, reveal features of the environment
represented at intermediate and higher-level stages of sensory processing.
One of my projects has focused on neural models of motion perception: determining large scale object motion from spatially localized motion signals. These models have made counter intuitive predictions about the perception of the speed and direction of simple patterns and the anatomical basis of position-invariant responses to rotation and dilation in the visual system.
Another project focuses on the effects of two-dimensional motion contrast on the interpretation of three-dimensional depth from motion. My long range goals consist of examiningfrom a psychological and neurobiological perspectivethe interaction of motion, depth, form, and color in the perception of three-dimensional objects and the segregation of figure from ground.