Churchland, Ramachandran and Sejnowski (1994: 23-25) argue that the traditional paradigm of “pure vision” is a “fiction.” They caricature "pure vision" as creating “a detailed model of the world in front of the eyes in the brain” They are convinced that this view actually “impedes progress.” In similar vein,, they declare that we can no longer assume that vision is a one way, hierarchical process, proceeding from the firing of retinal cells to the higher recognition functions of the brain cortex. Their “interactive” model for vision questions, from an evolutionary and a computational perspective, the necessity for “a perfect internal recreation of the three-dimensional World.” They adopt an embodied approach to vision that includes interactions “with other sensory systems as partners in helping to guide actions”:

…Although organisms can of course see when motionless or paralyzed, the visual system of the brain has the organization, computational profile, and architecture it has in order to facilitate the organism's thriving at the four Fs: feeding fleeing, fighting, and reproduction.

Churchland, Ramachandran and Sejnowski (1994: 25) concede the “undeniable feeling of having whole scene visual representation,” but they dismiss this as a subjective illusion. Incessant, rapid-fire saccades provide “a partially elaborated representation of the visual scene.” “At a given moment “only immediately relevant information is explicitly represented.”

How much of the visual field, and within that, how much of the foveated area, is represented in detail depends on many factors, including the animal's interests (food, a mate, novelty, etc.), its long-and short-term goals, whether the stimulus is refoveated, whether the stimulus is simple or complex, familiar or unfamiliar, expected or unexpected, and so on. The recognition can be faster and more accurate if the animal can make exploratory movements, particularly of its perceptual apparatus, such as whiskers, ears, and eyes. There is some sort of integration across time as the eyes travel and retravel a scan path foveating again and again the significant and salient features. One result of this integration is the strong but false introspective impression that at any given moment one sees, crisply and with good definition, the whole scene in front of one.

Patricia Churchland, V. S. Ramachandran, Terrance Sejnowski (1994). A Critique of Pure Vision, in Large-Scale Neuronal Theories of the Brain, edited by Christof Koch and Joel Davis, MIT Press, 1994.