Prediction error is a discrepancy between an expected state (a prediction) and actual sensory data or experience, in which an error signal arises that leads to adjustment of the model and reduction of future discrepancies between expectation and reality.