Note on the Interpretation of Experiments Concerned with Perceptual Adaptation (For discussion in Perception Seminar, January 10th)

January 1967

Note on the Interpretation of Experiments Concerned with Perceptual Adaptation
(For discussion in Perception Seminar, January 10th)

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 References to these essays must cite them explicitly as unpublished manuscripts.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 announced publication of Rock’s new book (1966), together with the recent Howard and Templeton book (1966) on spatial orientation, and the earlier book of Taylor (1962) on his strictly behavioral theory of perceptual adaptation points up a very puzzling situation. How can all the experiments that stem (in some sense) from Stratton’s 1896 experiment on vision with inverting spectacles be related to one another, and to psychological theory?

There seem to be three periods of research, first, the earliest experiments, next, the Innsbruck experiments (Kohler, 1951, translated in 1964), and, last, the recent series touched off by Held in 1958 which are mainly concerned with adaptation under conditions of prismatic optical distortion. The difficulty in interpreting all these post-Stratton experiments is to decide what they are about. To what psychological problems are they relevant?

Some investigators have thought they were relevant to phenomenal adaptation, but others to behavioral adaptation. In the 1950’s I saw an analogy between curvature adaptation (with prisms) and certain kinds of sensory adaptation, and showed that the curvature aftereffect obeyed the same laws as a negative afterimage, which is about as pure a phenomenal experience as one can get. But Held believes the adaptation experiments are relevant to voluntary action and a special kind of feedback. Taylor believes they demonstrate a motor theory of perception — that perception can be subsumed under behavior. Harris sees the results in terms of sense modalities — the old puzzle of the relation between vision and kinesthesis (or “touch”). The experiments (other than my 1933 series) certainly involve proprioceptionin some meaning of that term, but what is the meaning of the term? Both the behavioral and the phenomenal adaptation might be said to involve learning but what sort of learning? All kinds of psychologists are now getting into the act. All parties seem to agree that their experiments are concerned with “adaptation” but they do not ask: adaptation to what?

Before I properly understood the Innsbruck experiments, it seemed to me that the most interesting outcome of spectacle-wearing was a shift in the psychophysical correspondence between phenomenal lines and “retinal” lines. The qualities of linear curvature (or bending) and of linear tilt have opposites (as do the quality-pairs blue-yellow and warm-cold) relative to a neutral quality. So also does the quality of motion-one-way and motion-the-other-way (the afterimage of motion). I thought of a line or a motion on a sensory surface as being a sort of basic sensation like a color or a temperature (cp. the modern idea of “edge detectors”). This idea of recalibrating the null point of a phenomenal dimension was also attractive to Ivo Kohler. But this theory of what I called adaptation with negative aftereffect (1937) was lost sight of in two subsequent developments. First, it did not apply to the figural aftereffects of W. Köhler (which are like afterimages but are not dimensionally opposite) and, second, it got confused with theadaptation-level phenomena of Helson (which are “judgmental,” insensitive, and not like afterimages). The result of these developments was theoretical confusion about aftereffects in general, and the muddle was worsened by the hope of the field-theorists that all negative aftereffects (normalization) could be reduced to figural aftereffects (contour repulsion). It is now clear that they cannot (e.g., Bergman and Gibson, 1959). There seem to be at least three general types of aftereffects of attending to the same thing for a long time, or being made to attend to it: the negative aftereffect, the figural, and the judgmental.

To return to the Innsbruck experiments, I finally understood that my theory of the local recalibration of the neutral quality of lines, colors, and retinal motions could not explain the “situational” aftereffects discovered by Kohler. These effects appeared in (what I later called) the phenomenal visual world, not simply in the retinal visual field. They do not, as we say “move with the eyes” as afterimages do; they stick to the edges of things. They are aftereffects of wearing spectacles in the ordinary environment, not of simply continued looking through an optical device or looking at an actually curved line. They were so remarkable and so inconsistent with any existing theory of visual sensations that Kohler repeated the experiments for 10 years before publishing the results (1951). As he understood them, they seemed to prove that the effect of a local retinal stimulus was conditional upon eye-posture. Perhaps every sensation had to be conditioned to every possible eye-posture (see Taylor, 1962). More generally, the fact had to be not merely that a distorted retinal image got phenomenally corrected in the course of time, but that each special retinal image distortion in each possible eye-posture got corrected. There is no limit to the number of points, lines, or forms on the retina that would have to be “relearned”.

Parenthetically, this is the difficulty that, in large part, has led me to deny the relevance of the retinal “image” and its neural projection for visual perception, and to postulate that the stimulus for vision is the structured ambient array of the light —the stimulus, that is, for an ocular system that registers its surroundings.

In the face of the above difficulty with retinal sensations, some theorists have given up trying to explain phenomenal adaptation. It is all a matter, they say, of behavioral adaptation. The assumption seems to be that one can only find out about the environment by behaving in it. Kohler himself was tempted to suggest (1953) that “anyone who wants to see correctly must first be able to manipulate correctly”, although he has also believed that the acts of reaching, grasping, pointing, drawing, etc. during the experiment should not be confused with the fact of change in phenomenal experience. Other theorists, however, (including Kohler) have followed the line of trying to explain how the false visual experiences of space and motion that occur when first wearing spectacles might be corrected. The notion that they are corrected by proprioception (or kinesthesis or touch) that accompanies vision is an old one,

The restoring of rigidity is accomplished insofar as the system can register the new invariants over time that specify the layout of the world. Head-movements would be necessary for isolating these new invariants; perhaps voluntary head movements would help in directing attention to them but passive movements should be sufficient. Manipulation, reaching, and similar behavior might help to produce phenomenal adaptation but, in this formula, they are not necessary. The visual perception of oneself-in-the-world involves the postural system, to be sure, consisting of the eyes in the head on the body on the surface of support, but the adjustments and exploratory orientations of this system do not constitute behavior in the usual meaning of the term. In short, according to this formula there is a way of “finding out” about the environment without necessarily behaving in the sense of performing or executing actions.

A stationary pictorial frozen sample of an optic array does not seem to normalize with time in several respects when it is abnormal in those respects. This fact could be explained by a theory of the averaging of past experiences and of the influences of memory on perception. Many experiments exhibit this fact, mine especially, but it does not tend to become normal with respect to an ideal in the brain or to an average of past experiences; it tends to become vertical with respect to the edges of the environment, however they lie and wherever they are. Likewise, one can learn to draw, write, reach, point, or manipulate correctly when he can only see his hand, a virtual hand, “in” a mirror or “behind” a prism or a lens, or “in” a television picture (Smith and Smith). But this is not the same as spectacle-wearing. The optic array from a device attached to the table-top is not the same as the moving and changing sample of the ambient array when the device is attached to the head. In a spectacle-wearing experiment, the subject has to distinguish anew what is objective (exterospecific) from what is subjective (proprioceptive) in the changing flow of visual stimulation, but he does not have to do so on the basis of a special sense-modality called “proprioception” (Gibson, 1966, ch. 2).

The above formula admittedly requires the substitution of ecological optics for retinal image optics, and demands a new visual physiology of “looking around” instead of the old visual physiology of neural projection from the retina to the brain, but these hoary old doctrines have been losing their force anyway. All the investigators, however, are still thinking of the eyes as a photographic camera. The notion of the eye as a device that samples the available ambient array is found in Gibson, 1966, ch. 12.

If the adaptation is to the wearing of spectacles on the head this will explain why (as Taylor discovered) one can put them on every morning and take them off every afternoon and get used to it. What counts is not the pattern of the head’s field of view but the invariants under transformation of pattern as the head moves. (Note that spectacles have not yet been worn on the shoulders nor, except briefly by Taylor and recently by Festinger, on the eyes). The abnormal form as such (the distortion, or disarrangement or optical transformation, that is, the new form relative to the old form) does not carry information about the world (only about the spectacles) but the non-change with changing form does carry information about the world.

The altered “form” of the sample-field produced by spectacles can be of many sorts: Stratton’s inversion, or Erisman’s reversal, or Kohler’s reversal, or tilt, or displacement (with wedge-prisms), and curvature, and compression, or ordinary telescopic magnification or minification, or single-meridian magnification or minification (Ogle), or for that manner, a non-geometrical bias of the sample-field such as the chromatic aberrations or color. The “invariance theory” should still apply. We ordinarily don’t notice the forms of the visual field anyway, except in pictures: what we detect are invariants.

What else will this formula explain? And what will it not explain? What experiments are suggested by it in contrast to those concerned with vision vs. proprioception, or voluntary vs. involuntary, or behavior vs. experience?


Irvin Rock, The Nature of Perceptual Adaptation, 1966.

I. P. Howard & W. B. Templeton, Human Spatial Orientation, 1966.

J. G. Taylor, The Behavioral Theory of Perception, 1962.

Ivo Kohler, Uber Aufbau und Wandlungen der Wahrnehmungswelt, 1951.

—–, The Formation and Transformation of the Perceptual World, 1964, (Psychol. Issues, #12).

—–, Umgewohnung im Wahrnehmungsbereich. Die Pyramide, 1953 (Translated as a supplement to the above monograph, 1964).

—–, Die Zussamenarbeit der Sinne und das allgemeine Adaptationsproblem. Ch. 16 in W. Metzger (ed.) Handbuch der Psychologie, Vol. 1, 1966.

J. J. Gibson, Adaptation with negative aftereffect, Psychol. Rev., 1937, 44, 222-244.

—–, The Senses Considered as Perceptual Systems, 1966.

R. Bergman & J. J. Gibson, The negative aftereffect of the perception of a surface slanted in the third dimension. Amer. J. Psychol., 1959, 72, 364-374.

Also a series of papers by Held, Hein, Hay & Pick, Harris and others, during the last six or seven years.

Most of these references are given by Rock.