Principles such as efficient coding, predictive coding, and temporal invariance have been proposed to explain sensory coding, and have succeeded to some measure in reproducing the receptive field properties of neurons in the visual cortex. Bayesian surprise has been offered as an explanation of attention, and has enjoyed some success in modeling human saccades, while reinforcement learning and intelligent adaptive curiosity have been aimed at explaining how actions are chosen.
In this talk, we will develop a more careful definition of predictive coding and a novel definition of curiosity, and demonstrate characteristics of both. We will then show how these two definitions fit together into a single optimization principle. We propose this principle as a candidate for the hypothesized global optimization principle of the brain, as well as a general definition of "intelligence."