This presentation integrates two theories of visual perception operating on different levels of generality: the speed prior account and predictive processing. The speed prior account is able to account for many disparate motion perception illusions, arguing that perception arises from an aggregation of a prior expectation of a certain motion speed with visual data to arrive at a posterior ‘perception’. For example, the representational momentum effect (an illusion where a moving object is remembered further along in the direction of motion) arises when the expected speed (prior) of a moving object is faster than the actual speed (data), which gives rise to the percept (posterior) that the object was further ahead. While this theory focuses solely on motion perception, another bayes-inspired account is more general. Under predictive processing, perception (and cognition and action) functions by constructing continuously-updated models of the world (posteriors) from which to derive predictions (priors) that filter incoming signals (data). This theory presentation argues that the speed prior account can be viewed as an experimentally-driven, concrete implementation of the more general “grand unifying theory” predictive processing. By integrating the two, it 1) extends the reach of predictive processing into a more specific, falsifiable domain, while 2) the speed prior account is joined to a stronger theoretical framework from which to derive hypotheses. This combined framework operating on conceptual levels from general and abstract to specific and concrete would be a significant step forward in understanding one of the most important aspects of human perception.