Such effects are different from simple after-images produced by staring at a bright, coloured surface. If one looks for half a minute or so at an unpatterned red field, for a few moments all subsequently viewed white objects will appear tinged with green. But in the McCollough effect, although the perceived colours are also the approximate complementaries of the adapting colours, they depend (or are contingent) upon the orientation of the stripes of the adapting and test fields. It is the orientation with respect to the retina that is important. McCollough's subjects were adapted and tested with their heads upright, but if while looking at the test field they tilted their heads sideways, by 45 degrees, the apparent colours disappeared, even though the stripes in the test field still looked horizontal and vertical. If they tilted their heads through 90 degrees the effects reappeared, though now the stripes that had appeared pinkish with the head vertical appeared bluish, and vice versa.
Interest in the McCollough effect, and related effects, has generated a large volume of research. It is worth asking why. The original interpretation of the effects seems to have been right for the Zeitgeist of the late 1960s. It was becoming clear from the discoveries of the neurophysiologists that cells in the visual cortex of the brains of cats and monkeys are specialized to respond to rather specific features of the sensory input. After-effects seemed to suggest a similar organization in man. For example, the movement after-effect (apparent motion of a stationary field in one direction, following prolonged viewing of real motion in the other) is consistent with the idea that there are specific channels sensitive to particular directions of motion, whose outputs are reduced by prolonged stimulation and are compared to give perceived motion. The McCollough effect, as a complicated after-effect, might reveal how the human brain processes more complex patterns. This hope has, however, not been realized. Contingent after-effects have revealed little about how a face is recognized. Like other after-effects they seem to tell us only about the organization of early stages of visual processing.
Many other contingent after-effects (or CAEs) have been discovered since Celeste McCollough's original report. For example, coloured after-effects can be made contingent on the width of stripes, as well as on their orientation. Another effect, reported by H. J. Wyatt, is the size of after-effect contingent upon orientation. After adaptation to coarse horizontal stripes, alternating with fine vertical stripes, medium stripes appear fine when horizontal and coarse when vertical.
There are several ways for finding out where in the brain a particular effect is occurring (see Julesz 1971 for a discussion of them). One trick is to induce a visual effect with one eye, and then see if it can be obtained with the other eye. For example, for the after-effect of movement: if one gazes for a time with one eye at a moving display, which is then stopped, the now stationary display will appear to move in the opposite direction — and this apparent motion can be seen not only with the eye that inspected the moving display but also with the other eye, when the adapted eye is closed. This phenomenon is known as interocular transfer. It implies (provided there is no activity from the closed eye) that the anatomical site of the after-effect lies at a point at or after the combination of the two optic tracts. Some CAEs transfer between the eyes — for example, the movement after-effect contingent on the orientation of a superimposed grating. However, the McCollough effect exhibits little, if any, transfer. It used to be thought that it was a purely monocular effect, whose site lay peripheral to binocular combination, but more recent studies have found evidence for binocular effects. One way to do this is to divide up colour and pattern information between the eyes so that, instead of coloured gratings, the subject sees a plain coloured field with, say, his left eye, and a black and white grating with his right eye. The alternate adapting stimulus used is another colour, seen by his left eye, and a grating at 90 degrees to the first, seen by his right eye. (These adapting conditions are shown diagrammatically in Fig. 1.) McCollough effects are then found, but they are of opposite sign in each eye. So when the subject views the test field with the eye that has seen colour, the perceived colours are the complementary colours to those paired with a particular orientation during adaptation. This is the normal relationship between the colours seen during adaptation and those that appear during testing. But, when the eye that has adapted to black and white stripes is tested, the perceived colours are the same as those with which that orientation was paired during the adaptation. Here the normal relationship is reversed. Perhaps when the left eye is flooded with green light while the right sees a black and white grating, the visual system treats the colour input as 'left eye greener than right'. At any rate, the apparently simple scheme suggested by the early interocular transfer results has now been complicated by such later work.
Although most work on contingent after-effects has been done in vision, they can be found in other sensory modalities such as touch. For example, contingent tactile adaptation can be obtained with a wedge-shaped block of wood, held with the tips of the forefinger and thumb. If your hand moves from left to right as the finger and thumb slide from the thick to the thin end of the wedge, then, after some moments of sliding the hand right to left and left to right, a rectangular block of wood will feel deformed. As the finger and thumb of the adapted hand are moved from left to right the block will feel to be getting thicker, and for the reverse direction of movement it will seem to be getting thinner. Here the tactile adaptation is dependent upon the direction of movement of the hand.
Fig.1. Diagrammatic representation of the dichoptic induction of the McCollough effect, reported by D. M. and V. MacKay. During adaptation, the subject's left eye always sees colour and his right eye always sees black-and-white stripes. The colour and stripe orientations are systematically paired so that, say, red and vertical are always seen together, though by different eyes, and similarly green and horizontal. The pairs of stimuli alternate every 10 seconds or so during adaptation. To test for an after-effect, the subject views a black-and-white grating with one eye. If this is the eye that is adapted to colour, the after-effects are negative, but if it is the eye that is adapted to the gratings the after-effects are positive.
What do these striking and robust effects tell us about the brain? At present, there is still no generally accepted explanation for CAEs, but current theories are of two main types. One view likens CAEs to other after-effects supposed to result from adaptation in overlapping sensory channels. So, orientation would be coded by a number of neural channels: some tuned preferentially to vertical, some to, say, 10 degrees from vertical, and so on round the clock. Each group of channels tuned to a particular orientation would have members also tuned for red as well as vertical; others for green and vertical, and so on. During McCollough-type adaptation, gazing at a red vertical grating would reduce to below normal the output of channels tuned for both red and vertical. When the test grating is presented, channels tuned for vertical and other colours would give their usual output, but the red/vertical channels would be less active than normal. This imbalance in favour of the other channels would add an apparent green tinge to the vertical stripes. This is sometimes called a 'built-in' theory, since the mechanism producing the CAE is supposed already to exist in the subject's visual system before adaptation. It can be contrasted with a group of 'built-up' theories which propose that some kind of link is forged during the adaptation period, between previously separate mechanisms for processing colour and orientation. Different kinds of link have been proposed: some authors have suggested that an association is made like that in classical (Pavlovian) conditioning (in which, after a bell has been presented together with food on several occasions, the bell alone comes to elicit salivation); others that a neural inhibitory link is formed, so that activity in, say, the orientation system reduces activity in the colour system. However, despite numerous attempts, no experiment has yet been reported which convincingly decides between 'built-in' and 'built-up' theories.
There are several important characteristics of CAEs which a successful theory will have to explain. First, the effects are usually 'negative'. That is, the value taken by a sensory quality in the after-effect is opposite to or shifted further away from the value of that quality during adaptation. So, in the McCollough effect the colour seen on a particular orientation (say, red) in the test field is the complementary one to that viewed on that orientation during adaptation (in this case, green). Second, CAEs can be very long-lived. McCollough effects have been reported six weeks after adaptation, and reports that they can be re-evoked days after adaptation are common in the literature. Third, they do not decay during sleep. Indeed, there is some evidence that to remove the effects one has to look at the usual test field.
It is interesting to ask what the role is in normal perception of the mechanisms that produce the McCollough effect. Presumably, they do not exist simply to provide amusing perceptual demonstrations! One possibility is that, if indeed the initial stages of human perception consist of banks of filters tuned to particular features of the sensory input, then there will be a need to keep the outputs of these filter channels calibrated. For example, the gain of a particular sensory channel (that is, how much output it produces for a given input) might be subject to unwanted drifts, or fluctuations, as, say, a blood vessel narrowed which supplied that area of cortex. The brain could attempt to distinguish between such internal changes and those introduced by stimuli in the external world, by sampling over time the outputs of all channels. On the assumption that red vertical stimuli are about as likely to occur on average as green vertical stimuli, it would turn down the gain of a channel whose output was abnormally high for a long period, or turn up the gain of one that was correspondingly low. Such an automatic gain control system would act to keep the gains of sensory channels roughly equal despite biological drift. It could also remove from the neural image false signals — such as colour fringes introduced by chromatic aberration in the optics of the eye — since these constant errors would be treated in the same way as the unvarying colour/orientation combination in the McCollough effect. But such a system would be misled by prolonged exposure to stimuli exciting only very few channels. It would surely mistake the activity in these channels for internal drift, rather than external stimulation, and turn down their gain. This would give negative after-effects, since these channels would contribute less than their appropriate share to the final percept. The system would also require evidence that the gains of the adapted channels were too low before readjusting them. The best evidence of this would be for the subject to inspect (in the case of the McCollough effect) black and white gratings. Thus presenting the test field should cause CAEs to decay, but withholding it should produce long-lasting after-effects, as is found. CAEs may, therefore, reflect the brain's usually efficient but sometimes erroneous attempts at self-calibration.
(Published 1987)
See also visual adaptation.
— John Harris
- Bibliography
- Dodwell, P. C., and Humphrey, G. K. (1990). 'A functional theory of the McCollough effect'. Psychological Review, 97/1.
- Humphrey, G. K. (1998). 'The McCollough effect: misperception and reality'. In Walsh, V. (ed.), Perceptual Constancy: Why Things Look as They Do.
- Julesz, B. (1971). Foundations of Cyclopean Perception.
- McCollough, C. M. (1965). 'Color adaptation of edge-detectors in the human visual system'. Science, 149.
- Potts, M. J., and Harris, J. P. (1975). 'Movement after-effects contingent on the colour or pattern of a stationary surround'. Vision Research, 15.
- Skowbo, D., Timney, B. N., Gentry, T. A., and Morant, R. B. (1975). 'McCollough effects: experimental findings and theoretical accounts'. Psychological Bulletin, 82.
- Vladusich, T., and Broerse, J. (2002). 'Color constancy and the functional significance of McCollough effects'. Neural Networks, 15/7.
