Historical and contemporary philosophical theories of perception fail to account for two basic problems that demand attention in any theory of vision: the inverse optics problem and misrepresentation. I will offer empirical evidence derived from my work on motion perception to illustrate a different approach that succeeds where these other theories have not. Specifically, I will show how motion "illusions" can be explained using a probabilistic model of vision, thereby demonstrating how the visual system can generate biologically useful percepts from underdetermined stimuli. I will also indicate how this approach requires a fundamental shift in how theorists must think about visual (mis)representation. The result is a unique manner to achieve a naturalized philosophy of perception that is amenable to some central concerns in philosophy and science.