Being an exclusively human ability, art cognition occupies a unique position to inform our knowledge of human cognition. To do so, however, it must be understood in computational terms generalizable to other cognitive processes. Here, such an account of art cognition is sought. Specifically, I argue that Bayesian inference drives art cognition, by showing how it may underlie music and visual art perception. The resulting account provides the sketches of an economical and unified theory of art cognition consistent with contemporary theories of perception.