Recent research in awake, behaving organisms revealed the strong modulatory effects of movement on sensory coding. Surprisingly, these effects are not consistent across species. For example, in rodents and insects locomotion increases the magnitude of visual responses [1-3], while in primates, locomotion has a weak suppressive influence [4]. These differences raise intriguing questions about the computational purpose of such modulations and the generality of the underlying principles of sensory processing. Here, we address these questions from a theoretical perspective. Our approach is grounded in the efficient coding hypothesis, which proposes that sensory systems adapt to the statistics of sensory signals in order to maximize information transmission subject to diverse constraints [5]. Our key insight is that similar to changes caused by external factors, the statistics of sensory inputs are also strongly affected by an animal's movement. Therefore to maintain the accuracy and efficiency of sensory representations, sensory neurons should adapt to the locomotion state. To confirm these intuitions, we first simulated an agent moving in spatially diverse environments. In spatially heterogeneous environments, a fast-moving agent experiences stimuli with much larger dynamic ranges and more rapid fluctuations than a static agent. These effects were diminished in homogeneous environments. To verify the relevance of the proposed principles for the processing of natural stimuli, we analyzed the statistics of videos recorded in various natural scenes while stationary and moving. We processed the videos with a set of filters reminiscent of receptive fields in the visual system. We found that the outputs of filters whose properties matched the distribution of receptive fields in mice and flies had higher variance during movement. In contrast, the output variance of filters matched to the properties of receptive fields in the primate visual cortex decreased weakly during locomotion. We determined that a model neuron optimized to maintain efficient information transmission consistently across these fluctuations, qualitatively reproduces a range of modulatory phenomena observed in mice, flies, and primates [1-4, 6, 7]. Our results demonstrate that locomotion-induced modulations observed across the animal kingdom, while diverse, may be a manifestation of the same principle: maintaining a constant accuracy of sensory codes across environments and behavioral states.