The brain is a prediction machine
In predictive processing (PP), a multi-area, multi-level system perceives the world by trying to predict its own sensory states. Each level tries to predict the unfolding signal at the level below. Differences between the predictions and the current level are calculated and only residual prediction errors are sent forward (and sideways) to make better predictions (and drive learning). What we perceive is a combination of top down knowledge based prediction and bottom up incoming sensory evidence (signals from sensory receptors). To see a complex world is to deploy world knowledge (in the form of a generative model) to try to construct the incoming signal for itself using stored probabilistic knowledge about the world.
This brand new paper provides stunning experimental support for the predictive processing framework.
The hollow face illusion provides a striking exammple of how the brain overrides the incoming visual data with so called 'priors' that are stored in the brain as a result of prior exposure to faces under different lighting conditions.
We present this paper in a video lecture that you can also find on our website www.behavioralhealth2000.com.