September 1973
Further Note on Formless Invariants as Optical Information for Perception
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 References to these essays must cite them explicitly as unpublished manuscripts.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.
The notion of form is so familiar that we do not realize how vague it is. Psychologists keep on studying “form perception” without examining what it is that they are investigating. I have argued that
Object perception does not depend on form-perception but on invariant-detection. And these invariants are “formless,” that is to say, they are not themselves forms. Form perception is thus shorn of its importance (Gibson, 1973, p. 43).
This implies that the figure-ground phenomenon is not basic for object perception, that a flat form is not phenomenally simpler than a solid object, and that a retinal picture of something is not the prerequisite of seeing it. I make two assertions. First, it is possible to perceive an object without having a retinal image of it. A retinal image is defined as a simultaneous set of focus-points projectively corresponding to the set of radial points on the near side of the object, that is, a “mapping” of the facade of the object. Second, it is possible to perceive all sides of an object although the retinal image corresponds only to the near side. I will try to demonstrate these two assertions.
Object perception without a retinal image. By definition all parts of an image, as of a form, are simultaneous and concurrent. A form either exists all at once or does not exist. When an object changes from revealed to concealed and the reverse, with decrementing and incrementing of optical parts, it has no optical image. Nevertheless the object is perceived along with the fact of changing occlusion. The invariants that specify going-out-of and coming-into-sight are not imaged. They are formless. This is the lesson of George Kaplan’s experiment and of the effect that Michotte called screening, as in the “tunnel phenomenon.” He said that such percepts were amodal; I would rather say that they are unimaged.
Even when object and observer are both stationary, as with a frozen optic array, one object is seen behind another at an occluding edge. This is an old paradox in the study of depth perception, called “superposition.” The part of the object that is hidden is obviously not imaged, although it may be specified. (Note that “superposition” is not a cue for depth but a puzzle for depth perception.)
A number of demonstrations are accumulating in all of which a complete object is perceived when parts of it are displayed in succession but never the whole simultaneously. If a slot or window is moved in front of an object or a picture (or if the latter is moved behind the slot) it is perceived without there being a retinal image of the object or picture. The same thing happens if an occluder of some sort is waved in front of a scene or, for that matter, if the hand is waved in front of the eye. Michotte noticed these effects. Note that they cannot be explained by flicker-fusion as one might carelessly assume. I argue further that they cannot be explained by the hypothesis of “gestalt completion,” as Michotte assumed.
The parts or elements of a whole picture can be abruptly exposed (flashed) in succession and the whole will be seen if the intervening delays are not too great. An outline or contour can be perceived as the path of a moving spot even when it leaves no trace on a surface. Rock has described this phenomenon as “form perception without a retinal image.” But that is only a paradox, a contradiction in terms, i.e., a case of form perception without a form. This does not help. We should begin to think about optical information in terms of invariants and cease trying to salvage the concepts of the retinal image.
The perception of all sides of an object at once. Another case of perception without a retinal image is the awareness of the far side of an object as distinguished from its near side, or facade. Actually this is better described as the perception of the continuity of its surfaces, i.e., the connectedness of those facing the point of observation with those not facing it. The occluding edge of the object (either sharp or curved) is the junction of the near with the far side. There are two cases.
1. With the object and the point of observation both stationary the invariants in the frozen optic array that specify the occluding edge (especially if it is curved) will also specify the continuity between the near side and the far side. But this is a limiting case.
2. With either the object or the observer moving the invariants in the optic array emerge under the change, and the continuity of the near and far side is better specified. The junction of near and far changes. The front turns into the back and the back turns into the front; width goes into depth and depth comes into width. The reciprocal going-out-of sight and coming-into-sight at the occluding edges of the object specify the whole surface of the object. (See the MS of my new book for details.)
It should be obvious by now that visual forms are irrelevant to the perception of objects. Forms, including retinal forms, come and go as the perspectives of the object fluctuate. Hence an object cannot be mapped. It has enduring features like convexities and concavities, curves or dihedrals, that do not come and go when the object or the observer moves. They are invariant and hence they have to be specified by invariants in the optic array.
References
Gibson, J. J. The information available in pictures. Leonardo, 1971, 4, 27-35.
Gibson, J. J. What is a form? Psychol. Rev., 1951, 58, 403-412.
Gibson, J. J. On the concept of formless invariants in visual perception, Leonardo, 1973, 6, 43-45.