Resolving ambiguous visual scenes: Context, natural statistics and learning
The human visual system generally does a wonderful job in efficiently creating a robust perceptual interpretation of the external world. During the talk, I’ll review several of our recent studies that probe the perceptual processes responsible for selecting one interpretation of a scene over all other possibilities. For example, using a binocular rivalry paradigm, we have shown that the visual system makes use of unambiguous contextual information to bias one perceptual alternative over another. Using a similar paradigm, we’ve demonstrated the visual system’s preference for stimuli that best conform to the statistical properties of the environment. Finally, I’ll discuss recent work that looks at how the perceptual system combines information across and within senses to recalibrate how perceptual stimuli are interpreted.