The functioning of the brain is still an enigma and many models have been advanced to try to explain how we perceive the outside world. Predictive coding, proposed about twenty years ago, is based on the idea that the brain is capable of generating a mental representation of the outside world that is then compared to actual sensory inputs. The differences between prediction and reality will lead to neural correction activity; if the sensory inputs match the model perfectly, no activity is generated. In their paper to be published in PLOS Biology, Andrea Alamia and Rufin VanRullen demonstrated that a predictive computational model explained the development and characteristics of a fundamental property of cortical processing of visual information: alpha brain waves at 10 Hz. A 2-layer predictive model with only neural transmission times as time constraints shows the appearance of alpha rhythms similar to those observed experimentally on electroencephalograms. The same multi-level model explains the appearance of waves propagating through these different levels with a direction depending on the cognitive state: upward direction during the processing of visual information and downward direction during rest. The predictions obtained are perfectly verified by analyzing two sets of experimental EEG data recorded on the scalp of subjects in response to visual stimuli and during rest. These innovative results suggest that alpha rhythms, omnipresent in the brain and whose functional roles are still being debated, could reflect the neural calculation involved in predictive coding.