In sensory neuroscience, the neural and perceptual levels of investigation are commonly related through a reductive research strategy based in psycho-neural isomorphisms. Davida Teller's "linking propositions' are a particularly vivid illustration of this epistemology in the context of vision science. For Teller, linking propositions guide the core epistemological practices of vision science by expressing the criteria for acceptable explanations of perceptual phenomena by neural processes and by articulating heuristics for discovering neural properties on grounds of perceptual ones, and vice versa. Furthermore, linking propositions create a uniform discourse in which ontological identity statements can be formulated on the basis of structural terms applicable to both the neurophysiological and perceptual domains.I argue that the epistemic practices of current color vision science are better understood through Machamer-style mechanisms than isomorphism-based linking propositions. For Machamer, activity types of mechanisms are central for both explanation and the discovery of entities and their properties. The activity type of color cancellation is fundamental to the opponent process paradigm of color vision precisely in this manner. The descriptive inadequacy of linking propositions for clarifying vision scientific research is particularly damaging because activity types have a unifying role for the structural facts of color perception. As a form of local reductionism, Teller's proffered epistemology aims to create a uniform domain for neuroscience and perceptual psychology but on the basis of a non-unified, disparate set of structural identity statements. By bringing attention to the unifying role of activity types, Machamer mechanisms allow also for critical epistemological explorations of ontological reductions on grounds of the fit between activity types of higher-level and lower-level sciences. They offer a Machamer-ontological-critical or MOCing framework, for short, against local reductionism.