Do you like that image or swipe it away? Where do you want to live? Numerous decisions partly depend on options’ sensory appeal, their aesthetic value. Yet, we have a poor understanding of how objects and experiences gain such value. We propose the theory that aesthetic value is a signal for maintaining and adapting the states of the cognitive-sensory system to process stimuli effectively now and in the future. Two interlinked components generate an object’s aesthetic value: 1) processing fluency 2) learning. We show that a computational model that realizes the theory can replicate past experimental results on the effects of exposure, symmetry, and complexity as well as predict people’s liking judgments on a trial-by-trial basis. Our theory along with its computational implementation has several important implications. One, because it postulates an essential link between sensory processing and valuation, it provides a theoretically grounded reason for why the features of standard convolutional neural networks can be used to predict aesthetic value judgments. Two, its dynamic nature calls to attention sequence effects that have so far often been ignored. Third, said dynamics link aesthetic valuation to other phenomena that empirical aesthetics has so far rarely investigated. Specifically, we propose that our theory naturally offers an account of exploration, exploitation, and boredom in the sensory domain.