Feb. 17, 11:45, Wei Ji Ma. Optimality and Probabilistic Computation in Visual Categorization

Barbara Anne Dosher

Cognitive Sciences

University of California, Irving

Visual perceptual learning: Changing the state of the observer

Perceptual learning improves perceptual task performance in distinctive ways. External noise tests and noisy ideal observer models can characterize the effects of perceptual learning (or attention) on an observer’s performance.  Across a range of tasks, two independent mechanisms can be seen: tuning of the task relevant perceptual template (external noise exclusion) and enhancing the stimulus (reducing absolute threshold). In the context of perceptual learning, both mechanisms may reflect re-weighting of information from early sensory codes. Principles of reweighting provide an account of a range of phenomena in perceptual learning, including the roles of feedback in performance and principles of transfer and specificity. An augmented Hebbian reweighting model (and extensions) provides one approach to understanding how we learn and how it generalizes.