A THEORY OF MCCOLLOUGH EFFECT AND CONTINGENT AFTER-EFFECT CHUN CHIANG Institute of Physics, Academia Sinica A model is advanced to explain the McCollough effect and the contingent motion after-effect. This model is based on coupling of two units at the inspection stage, each of which is sensitive to a particular feature of the stimuli. In the testing stage, the input of one feature in the stimulus not only excites this particular feature detector, but also induces the excitation of the other feature detector through the coupling to the state compatible to the previous inspection stimulus, and the other newly incomming feature in the stimuli compared with this state produces the after-effect. Comparisons with other theories are discussed. The McCollough effect (McCollough, 1965) is the effect that when viewing alternatively the red/horizontal and green/ vertical grids for some time, the achromatic vertical lined grid appears pinkish coloration and the achromatic horizontal lined grid appears greenish coloration. Even though a lot of experimental data on this effect have been obtained, the theoretical explanation is still in question (Skowbo, Timney, Gentry, & Morant, 1975). Of the many theories, two classes of theories may be classified. The first class of theory, originally proposed by Mc- Collough (1965) and extended by many others, assumes that the effect arises due to chromatic adaptation of color coded edge-detectors. With the inspection of red/horizontal lines, the chromatic component of the horizontal units for red is adapted, thus the achromatic test pattern presented later will appear tinged green by the unadapted horizontal detectors. The same reasoning applies to the appearance of pinkish coloration of achromatic vertical lines. However, this model has been criticized that too many properties have to be assigned to this colorcoded edge-detector (Murch, 1976). For example, if a high degree of retinal area specificity is found (Stromeyer, 1972), this is interpreted to mean that the color-coded edge-detectors have narrow receptive fields. Also, Murch (1969, 1972, 1974, 1975, 1979) has reported some other data which are not compatible with this model. The second class of theories attributes the McCollough effect to " a learning process by which the pairing of a color with a specific grid orientation produces an association in which the grid orientation comes to evoke the response of the visual system to the color " (Murch, 1979). Furthermore, the lined grid in inspection was viewed as the conditioned stimulus and the color functions as the unconditioned stimulus. However, as remarked by Mayhew and Anstis (1972), if it is due to learning, then the effect should be able to be contingent on shape such as triangles or disks. This seems not be the case. Also, Favreau (1979) rejects the classical conditioning explanation on the basis that either member of the contingency can evoke the perception of the negative of the other, and this is not the characteristic of classical conditioning. In the following, a new model is presented which seems to be able to alleviate the above mentioned defects. THEORY Visual system consists of many individual
feature detectors or sensors, which normally detect each in-comming features. However, this is the passive response of our visual system. In order to perceive the in-comming signals through the detector, the visual system also consists of active response, this active response consists of subjective controller and follower (or filter) as shown in Fig. 1. The subjective controller represents the system in which a subjective efforts for the visual system to perceive the signals can be excerted and the subject can direct a mental control over the components of the visual system to match the signals. This subjective efforts and mental control are the meaning of" active response " which represents the autonomous response of the subject. The matched components for the signals are termed the follower or the filter, since the matched components try to follow the incomming signals (for more detail, see Chiang's paper, 1978). The perceived signals in the perception space is the detected signals in the detector filtered through this filter, namely, the color, intensity or the motion in the incomming signals minus the color, intensity or the motion in the matched signals in the filter. Chiang (1978) has used this model to explain the motion after-effect and quantitative equation can be derived to predict the extent of the motion aftereffect. For the present situation, when two different features, feature A and feature B, in the signals are repeatedly presented in the " inspection stage "; the matched filter for feature A, namely the filter A, and the matched filter for feature B, namely the filter B, are coupled together through some physiological processes. Also, the controller for feature A and the controller for feature B are coupled together as well. The coupling is due to the establishment of the correlation of simultaneous firing of neurons for two different features. In the testing stage, with the in-comming signal containing both feature A and a neutral feature, the filter A and controller for feature A are rapidly activated and thus feature A is perceived in the perception space. However, due to the coupling process, the firing of neurons for filter A will induce the firing of neurons for filter B even though feature B is not presented in the signals. Consequently, the perceived " neutral feature" of the in-comruing signals will be the "neutral feature" minus the" feature B" and this gives rise the contingent after-effect. Some experimental evidences can sup- FIG. 1. A block diagram showing the perception and coupling of the units in the visual system sensitive for two different features of the stimuli. The sign -9 indicates the direction of flow and the signindicates the coupling of two units.
Theory of McCollough Effect port the above coupling induction process. For example, Hebb (1949) found that the firing of neurons in a " cell assembly" can affect the probability that other cells will fire. Also, Leppmann and Allen (1973) have reported that when test patterns are presented for brief periods, the exposure time required to identify the form of the pattern is less than that required for subjects to identify the contingent color after-effect; this fact supports the view that filter A (for viewing the form of the pattern) is activated first, and the coupled filter B (for viewing the color) is activated latter by the coupling and induction process. For other examples in which there exists the evidence of the coupling process, we may cite the classical conditioning and the biofeedback process. In both these cases, repeated exposure of two simultaneous stimuli establishes the coupling process. However, it should be noted that even though the contingent effect, classical conditioning and the biofeedback process all rely on the coupling process, it is erroneous to say that the contingent effect can be explained in terms of conditioning process, because one complicated process cannot be explaining in terms of another complicated process, even though they may both have some thing in common. For a more direct experimental evidence of coupling process, we have to rely on the further research in anatomy and physiology. However, the successes of the above model may be judged by the cases for which this model can explain. Case (a): McCollough effect, the orientation contingent color after-effect. In this effect, the inspection stimuli are the vertically orientated red and black stripes alternating with horizontally orientated green and black stripes. Thus the feature follower for vertical black stripes is coupled with the feature follower for red color and the feature follower for horizontal black stripes is coupled with feature follower for green color in the visual system (see Fig. 1). In test stage, with the viewing of vertical black stripes and white color, the vertical black stripes induce the feature follower for red color through the previous feature coupling mechanism, thus the white color goes through this red color filter (or the white color in reference to this red color) gives the green sensation. Similar explanation applies to the red sensation for viewing the horizontal black stripes and white color. Case (b): Color contingent motion after-effect (rotating spirals). The inspection stimuli are black spirals rotating clockwise in red background alternating with black spirals rotating counter clockwise in yellow background (Mayhew & Anstis, 1972). According to Chiang's theory of motion after-effect (1978), the follower for the motion in the visual system should be a transient. In the inspection stage, the follower follows the rotating stimulus starting from motionless in the beginning to full rotation and to abrupt stopping and then rotates in reverse direction. The coupling features are: Red color (or yellow color) in follower A is coupled to transient clockwise rotation (or transient counterclockwise rotation) in follower B. Furthermore, the switching of color from one color to the other (or from clockwise rotation to counter-clockwise rotation) in stimuli requires great mental effort in the subjective controller to cope with; thus in the present situation, the act of switching to red color (or yellow color) in the subjective controller A is also coupled to the act of initiating the clockwise rotation (or the counterclockwise rotation) in the subjective controller B. In the test stage, with the presentation of stationary spirals in red light, the stimulus of red light sets the follower A to conform to the red light, and since the red color in follower A was coupled to clockwise rotation in follower B, thus the stationary spirals will appear to rotate counterclockwise in reference to this clockwise rotation of follower B. However, the
coupled follower B is transient and will disappear gradually, thus the apparent rotation of the stationary stimulus will also be a transient and disappear gradually. Since the "switching" of color is also coupled to the " initiating " of the rotation in the subjective controller, thus even though the apparent counter-clockwise rotation will gradually disappear in the background of red light, the act of new switching of color to red light in subcequent test can again initiated the clockwise rotation of follower B and the stationary stimulus can appear to rotate counterclockwisely again. With the similar argument, the stationary spirals will appear to rotate clockwisely in yellow light. Thus, these coupling of color with motion and the coupling of switching of color with the initiating of rotation together may explain the observed facts (Mayhew & Anstis, 1972), namely, the color dependence of the direction of the apparent rotation, the gradual disappearance of the apparent rotation and the renewed apparent rotation with the switching of the color. Case (c): Space contingent motion aftereffect (moving stripes). The inspection stimuli are narrow spaced stripes moving downward and medium spaced stripes moving upward (Mayhew & Anstis, 1972). The coupling features are: Narrow stripes are coupled to the downward motion and medium stripes are coupled to the upward motion. Thus in testing, the stationary narrow stripes should appear to move upward according to the above theory and the stationary medium stripes should appear to move downward. This is observed. Mayhew and Anstis (1972) also reported that for some subjects, with the inspection of medium space stripes and broad spaced stripes (instead of narrow space stripes and medium space stripes), the stationary " medium " spaced stripes will appear to move upward (instead of downward) and " broad " spaced stripes (which have not been presented in the inspection) will appear to move downward; and in general, this kind of transposed space or color contingent motion after-effect occurred only if a rest period of about 20 min was given between the inspection and test procedures. No explanation was given to this phenomenon. It is postulated here that in the inspection stage, not only the actual physical and physiological space of stripes was coupled to the motion, but the impression of "broad " or "narrow " of the space was also coupled to the motion. Thus in the inspection stage, both the space of the narrow stripes and the impression of being relative narrow are coupled to the downward motion and the space of the medium stripes and the impression of being relative broad are coupled to the upward motion. Consequently, with the stimuli of stationary medium stripes and broad stripes in the test stage, the medium stripes (the impression of being relative narrower) will appear to move upward and the broad stripes (the impression of being relative broader) will appear to move downward. However, this kind of coupling takes time to develop, since the impression of the stripes being broad or narrow requires the brain to interprete, and process of interpretation and its subsequent coupling with motion take time. Whereas the coupling of the actual sensed space with the motion does not require interpretation, thus does not need time to develop. Also, since attention was concentrated in perceiving the motion in the inspection stage and there is not enough time for the brain to interprete the stripes being broad or medium and to make strong coupling in the inspection stage. For these two reasons, the transposed space contingent motion after-effect can not show up right after the inspection stage and have to wait for 20 min to show up. Thus the usual after-effect can appear right away and the motion after-effect with transposition will take about 20 min to develop. The above theory of con-
Theory of McCollough Effect tingent after-effect may not only explain the examples given above, it may also explain many other experimental findings. DISCUSSION While there are many theories advanced in the past (see review by Skowbo et al., 1975), it might be appropriate to compare some of those theories with the present one. For the edge-detection model and the single-unit model (McCollough, 1965; Fide11, 1970; Tell & Clark, 1968; Hepler, 1968), it was postulated that there is a single neural unit sensitive to two stimuli. In the present theory, it is the coupling of two units sensitive to each stimulus that leads to the after-effect, in stead of one unit sensitive to two stimuli. Murch (1972) proposed that " color adaptation in conjunction with a specific line orientation " might be an appropriate way to describe the physiological correlates of the McCollough effect, the fatigued opponent-process color receptors in the lateral geniculate nucleus would " feed into " cortical units having orientation sensitivity; achromatic spatial patterns would appear colored because the lines were processed through fatigued color units on their way to the orientation detectors. Murch (1972) also suggest that the Mc- Collough effect involves only the adaptation and fatiguing of rolor, but probably does not involve a fatiguing of orientationsensitive units. Skowbo et al. (1975) however questioned that why exposure to chromatic gratings could produce no longlasting fatigue of the orientation-detecting mechanism and yet, at the same time, would fatigue color units for a period of weeks. In the present theory, while it also proposes that the orientation elicites the after-effect, similar to Murch's model, however, it does not use the notion of adaptation and fatiguing. The elicitation of after-effect is due to the coupling, which is induced by the simultaneous firing of some neurons by other neurons. This coupling is at a more basic stage and does not necessarily involve the retrieval of the information from the memory. Therefore, this theory is also different from the pure learning or conditioning theory. REFERENCES CHIANG, C.1978 A theory of motion aftereffect. Japanese Psychological Research, 20, 101-104. FAVREAU, O. E. 1979 Persistence of simple and contingent motion after-effects. Perception and Psychophysics, 26 (3), 187-194. FIDELL, L.K.S. 1970 Orientation specificity in chromatic adaptation of human " edge detectors ". Perception and Psychophysics, 8, 235-237. HEBB, D. O. 1949 The organization of behavior. New York : Wiley. HEPLER, N. 1968 Color: A motion contingent after-effect. Science, 162, 376-377. LEPPMANN, P. K., & ALLEN, D. B. 1973 Further Studies of form-specific chromatic aftereffects. Paper presented at the meeting of the Optical Society of America, Rochester, New York, October. MAYHEW, J. E. W., & ANSTIS, S. M. 1972 Movement after-effects contingent on color, intensity, and pattern. Perception and Psychophysics, 12, 77-85. McCoLLOUCH, C. 1965 Color adaptation of edge detectors in the human visual system. Science, 149, 1115-1116. MURCH, G. M. 1969 Size judgments of Mc- Collough afterimages. Journal of Experimental Psychology, 81, 44-48. MURCH, G. M. 1972 Binocular relationships in a size and color orientation specific aftereffect. Journal of Experimental Psychology, 93, 30-34. MURCH, G. M. 1974 Color contingent motion after-effects: Single or multiple levels of processing. Vision Research, 14, 1181-1184. MURCH, G. M. 1975 Orientation specific colored after-effects are classically conditioned responses. Paper presented at association for research in Vision and Ophthalmology, Sarasota, Florida. MURCH, G. M. 1976 Classical conditioning of the McCollough effect: temporal parameters. Vision Research, 16, 615-619. MURCH, G. M. 1979 The role of test pattern background hue in the McCollough effect. Vision Research, 19, 939-942. SKOWBO, D., TIMNEY, B. N., GENTRY, T. A., &
MORANT, R. B. 1975 McCollough effects: Experimental findings and theoretical accounts. Psychological Bulletin, 82, 497-510. STROMEYER, C. F. 1972 Contour contingent color aftereffects: retinal area specificity. American Journal of Psychology, 85, 227-235. TEor, L. W., & CLARK, F. T. 1968 The effects of stimulus density on orientation specific aftereffects of color adaptation. Psychonomic Science, 11, 265-266. (Received May 30, 1980)