#topographie/colline **articles**: https://www.nature.com/articles/nrn2787 # terms free-energy principle (minimizing) surprise or surprisal (negative log probability of an outcome) internal generative model (of the world) (minimizing) free energy synaptic activity (perceptual inference) perception efficacy (learning and memory) gain (attention and salience) infomax principle minimum-redundancy principle policies (from optimal control theory and reinforcement learning) "The free energy principle is based on the Bayesian idea of the brain as an 'inference engine'. Under the free energy principle, systems pursue paths of least surprise, or equivalently, minimize the difference between predictions based on their internal model of the world and their sense and associated perception. This difference is quantified by variational free energy and is minimized by continuous correction of the world model of the system, or by making the world more like the predictions of the system." # https://arxiv.org/pdf/1705.09156.pdf section 2 # key ideas