Although it feels to us as if we perceive the external world objectively, perception is strongly influenced by prior expectations and beliefs (i.e. "priors"). How these priors are neurally implemented and combined with sensory input is still largely unclear. In the current proposal we focus on the role of spontaneous fluctuations of neural activity in perception. We hypothesize that these fluctuations constitute the ?priors?, which maintain an internal model of the environment. By studying their dynamics, relevance and implementation in the visual cortex, we hope to uncover a crucial computational role of spontaneous activity in embodying a prior world model.