Thesis defense David Richter (Donders series 493)

11 March 2021

Promotor: Prof. dr. F.P. de Lange

Prediction throughout visual cortex
How statistical regularities shape sensory processing

When we look at the world, we use our prior knowledge to predict and make sense of what we see. Usually we only become aware of this process when our predictions are incorrect. Consider for example walking around a corner and almost bumping into someone. The surprise and startle you felt, is the consequence of a prediction error – i.e., you did not predict someone would stand around the corner. In my thesis, I investigated how the brain creates and uses predictions to inform our visual perception. My work shows that we automatically recognize statistical regularities in our environment, and subsequently use these regularities to predict what we see, even without any intention to do so. Moreover, I demonstrate that these predictions influence neural processing throughout the sensory brain. In particular, surprising visual stimuli elicit stronger and clearer neural responses compared to correctly predicted ones, possibly because well-predicted events are less informative than surprising events. Combined my results are well accommodated by casting perception as an inferential process, inferring the most likely sources for sensory input using a combination of sensory information and prior knowledge, derived from statistical regularities in the sensory world.