We live in a largely predictable world: e.g. objects usually don’t suddenly change shape, and light usually comes from above. Prediction is a hallmark of understanding, and our ability to capitalize on predictable structure in the environment lies at the heart of intelligence. A central hypothesis of my lab is that the brain is a prediction machine, comparing incoming signals with prior expectations. We seek to elucidate how this process is implemented in neural hardware. In my talk, I will discuss recent behavioral and neural findings from our lab that shed light on the the nature of predictive processing. Drawing on behavioral data, computational modeling, invasive electrophysiology in mice and neuroimaging in humans, I will show how various forms of prior knowledge shape the neural computations that give rise to perceptual experience.