It has long been appreciated that learning about the probabilistic structure of events alters our perceptual awareness. However, recent work has demonstrated that this relationship between learning and perception is more complex than previously believed and can theoretically serve a multitude of functions. For example, we must balance demands of representing the perceptual world accurately while effectively updating our models when the world changes. The first part of my talk will present some work that asks how we optimize this balance via predictive mechanisms. The second part will consider how explanations of oscillatory windows of perceptual awareness could link with these ideas. I hope to convince the audience that our models of learning-perception interdependences should move on from some currently popular monolithic accounts (e.g., cancellation; intrinsic fixed sampling rhythms), and stimulate discussion concerning how best to conceptualise these synergistic relationships.