This course will be built around the contemporary research of vision. First, it will cover the classical approaches of low and high-level vision, visual learning, the neural implementation of perception and learning in the brain, and computational models. Next, it will critically evaluate the state-of-the-art and explore alternative approaches to the same issues. Specifically, it will discuss the probabilistic view on vision, and how it changes the research questions in focus. We will investigate how statistical learning, rule learning, perception and cue-combination as probabilistic inference can expand the range of interpretable phenomena in vision. We will also cover the issue of possible neural embodiment of such computations and review evidence that supports such an interpretation.