# the predictive mind chapter 3 **Opening (not a real abstract):** Prediction error, context, and precision Perceptual inference is a matter of hierarchical prediction error minimization. Top-down predictions are compared to the actual sensory signal, or with the inference of the level below, and the difference, the prediction error, is used as a feedback signal for the internal models generating the predictions. This sort of scheme sounds very much like linear predictive coding that was invented in the 1950s to compress data. However, these linear schemes would only work if the world was populated only with very simple, linearly interacting causes. Of course, the world is replete with causes that interact non-linearly—this is part of what makes us worry about the relation between causes and effects not being one–one. Examples range from simple perceptual interactions like a cat being partially occluded by the fence, to deep and **Authors:** [[hohwy]]