14.01.2013

Colloquium in Cognitive Sciences by Felix Wichmann

Department of Psychology, Schleichstr. 4, Level 2, Room 4231 on Thursday, January 17th, 2013, 17-19h

Understanding vision: Inferring Critical Features from Behavior

Understanding visual perception and visual cognition in terms of the algorithms underlying overt behaviour requires a solution to the feature identification problem: Which are the features on which sensory systems base their computations? Thus one of the central challenges in vision science is system identification: We need to infer the critical features, or cues, human observers make use of when they see or hear. What techniques can we use to identify them? What aspect of the visual or auditory stimulus actually influences behaviour if faced with real-world, complex stimuli? In my laboratory we develop data-driven system identification techniques based on modern machine learning methods to infer the critical features from human behavioural judgments. I will present these methods and show what their benefits are over the traditional “reverse-correlation” approach and the “bubbles technique”.