January 1969
“Information” in Visual Theory
J. J. Gibson, Cornell University
The World Wide Web distribution of James Gibson’s “Purple Perils” is for scholarly use with the understanding that Gibson did not intend them for publication. References to these essays must cite them explicitly as unpublished manuscripts. Copies may be circulated if this statement is included on each copy.
There are two radically different meanings of the term information in vision that tend to be confused. One I will call sensory-input information and the other optic-array information.
Existing theories of perception assume that the only information for perception is “sensory” information. But this assumption is vague. If it means that the information for perception must come through thesenses (J. Locke) instead of through extrasensory intuition, most of us would agree. But it can be taken to mean that the information for visual perception consists of impulses in the fibers of the optic nerves (J. Muller) and with this we need not agree. This is what I called sensory-input information. It assumes that afferent “signals” to the brain from the receptors are the basic constituents of perception. To assume that all visual information must come through the visual sense is not to assume that it is transmitted over nerve-fibers in some kind of “sensory code.” For a sense may be considered as an active perceptual system with a capacity for extracting information from obtained stimulation and not merely as a channel from receptors to brain. For vision, this implies that retinal inputs, then ocular adjustments, then altered retinal inputs, and so on are required for the normal process of perception. This exploratory process is distinguished from the brain’s processing of the impoverished or curtailed retinal inputs that are generally studied in the laboratory.
Optic-array information differs radically from sensory-input information. It is information in light, not in nervous impulses. It involves projection to a point of observation, not communication between a sender and receiver. It is outside the observer and available to him, not inside the observer. It has only to be attended to, it does not have to be interpreted.
The concept of an array requires ecological optics for its analysis whereas the notion of an image follows from physical optics. The former implies invariants of structural; the latter implies only correspondence between luminous points in the world and focus-points in the image. The structure of an ambient array can only be registered by an ocular system over time; the structure of a temporary retinal image has been thought to be “transmitted to the brain” point for point over the optic nerve and tract (but this is now so dubious as to be unacceptable). And, indeed, if the only information the nervous system can obtain is from the mosaic of point stimuli at retinal receptors, it is very poor indeed.