Based on noisy sensory evidence, the brain ?decides? what we see. Current models implicitly assume that this decision process is bottom-up and passively driven by input from the sensory areas. This project tests if this view of ?pure vision? is adequate, or whether perception is an active process. First, we use non-invasive brain imaging methods to clarify the neural mechanisms underlying perceptual decision making. In doing so, we aim to resolve several controversies and contradictions in the literature. Having identified these mechanisms, we then investigate how expectations modulate the decision process, and whether such top-down effects may change sensory representations.