Some Comments on J. J. Gibson’s Approach to Picture Perception

Leonurdo. Vol. 12. pp. 121-122. Pergamon Press Ltd. 1979. Printed in Great Britain. SOME COMMENTS ON J. J. GIBSON’S APPROACH TO PICTURE PERCEPTION Rudolf Arnheim* In his recent writings James J. Gibson has been quite liberal with descriptions of the confusion, the errors, the ignorance and the irrelevance of what he finds around himself in his field of interest. He has made me, for one, the more eager to receive the final enlightenment, which, according to him, is just around the corner. But if the chapter published in Leonardo [11 is characteristic of his forthcoming book, I am in for another disappointment. Those of us who waited all those years for Gibson to overcome the naive notion that pictures are duplicates of their referents, must be forgiven if they are less dazzled by his recent revelations than he is himself. How a seasoned researcher could be satisfied with undefined and vague concepts such as ‘information’ or ‘invariants’is a mystery. He surely cannot expect many readers to be content with statements such as: ‘The invariants are not abstractions or concepts. They are not knowledge: they are simply invariants’ [11. If, then, I try to extract the meaning of his ‘invariants’ from his examples, I find that it derives from the constancy of shape. He calls his presentation of this phenomenon ‘a very radical hypothesis’, even though the constancy of shape became a common property of psychology textbooks in 1935,at the latest, when Kurt Koffka in his Principles qf Gestalt P.Yychology devoted a chapter to the visual constancies [2].The term ‘constancy of shape’ concerns the observation that in daily life people frequently see the objects of their environment with some of the spatial properties the objects possess in the physical world rather than with those of their projective images. A brick, for example, retains its right-angle shape and objective proportions , no matter from where one looks at it. (The observation applies, however, only to simple, regular shapes. It works for a brick but not, for example, for complex sculptures that change their angles and proportions continuously as one walks around them.) Drawing attention to the constancies was only the beginning of the work that had to be done. The principal task of psychologists was to explain the phenomenon-that is.theyhad to describe itscauses and the mechanisms that made it work. It is generally agreed that the persistence of objective shape is useful for practical identification, but this biological utility cannot account for the complexity of the phenomenon. Therefore, one must ask: Which aspects of the retinal projection enable perception to present the objective shape of things so spontaneously? Under what conditions do these aspects apply? Which underlying principle controls these transformations? (They occur only when the ‘objective shape’ is structurally simpler than the particular projective transformation.) Gibson does not begin to face these problems, because he chooses to deny that visual perception begins physiologically with projective retinal images . I know of no one else who doubts that each shape in perception must be explained as a derivate of a primary retinal input. Gibson avoids this challenge by playing on the ambiguity of the word ‘primary’ [3]. Quite rightly he points out that the optical projections are not the primary content of visual experience. But, in the physiological process that leads from the initial stimulus or input to the final experience, the retinal projection is most certainly primary. The visual gradients, for example , which Gibson accepts as invariants, are at first nothing but sequences of single sizes, shades and densities on the retina. What transforms them into structural crescendos and what makes these crescendos be seen as slants in depth are the essential issues. But they seem to be beyond the ken of Gibson’s thinking. Gibson’s descriptions of the constancy of shape imply that nothing determines perception but the objective ‘physical’shape of things. The variables of the projective image are said to drop out as irrelevant noise. (‘There is no “form” left in a continuous transformation. It has vanished and all that remains is the invariants’ [4].) But this. I am sure, is quite wrong even for ordinary everyday vision. The particular...