Simultaneous brightness contrast for flashes of light of different durations. Mathew Alpern

Simultaneous brightness contrast for flashes of light of different durations Mathew Alpern Measurements have been made of the magnitude of simultaneous brightness contrast on two young adult male observers by a binocular brightness matching method. Five different luminances of the inducing pattern were studied and the duration of the exposure was varied between 5 and 150 msec, along an arithmetic scale. For low inducing flash luminances, the longer exposures show the greater contrast effect. However, as the luminance of the inducing pattern was progressively increased, the duration of exposure showing the largest contrast effect systematically decreased. The curves resemble the Broca-Sulzer curves both in the manner, at any given luminance, that the ordinate varies with duration and. in the way any given curve changes as luminance is varied. The data explain a previous contradiction between results from psychophysical and. electrophi/siological experiments and therefore greatly strengthen physiological theories of contrast. s,simultaneous brightness contrast is the familiar phenomenon in which an object appears dimmer when seen against a bright surround than against a dark one. This is illustrated in Fig. 1, which shows two identical gray surfaces on different backgrounds. Johannes Mueller 10 believed that this effect was due to the reciprocal action of adjacent retinal areas and Hering s supported this view, in disagreement with Helmholtz 7 who had attributed the phenomenon to a more complex psychological process which he referred to as an "illusion of judgment." There are a number of excellent reasons for supporting Mueller in this dispute, not From the Departments of Ophthalmology and Physiology, University of Michigan, Ann Arbor, Mich. This investigation was supported in part by a Research Grant NSF G( 10045) from the National Science Foundation. 47 the least of which can be demonstrated with the experimental arrangement illustrated in Fig. 2. In this figure, a is a bright field (the comparison standard) which is maintained at a constant level (about 10 FL. in these experiments) throughout; b is a similar field (the test patch) which is seen as far below the fixation point Z' as a is above it. The luminance of h (B b ) is varied (the dependent variable) in order to make a brightness match with a. c and c' are two similar rectangles (the contrast inducing patches) which are seen by the right eye on either side of b and their luminance can be varied over a wide range (the independent variable). If both b and c-c' are seen by the right eye and a by the left, then as soon as the inducing patches become very bright they make the test patch appear much dimmer and its luminance must consequently be greatly increased in order to re-establish the brightness match (open circles, Fig. 3). On the other hand, if the inducing

48 Alpern In ccsfigativc Ophthalmology February 1963 Fig. 1. The phenomenon of simultaneous brightness contrast. The two gray squares have the identical luminance but the one seen against the black surround appears much brighter..05 pa 2i c b c' JUU 1 Fig. 2. The stimulus pattern used in these experiments, a is seen always by the left eye and c-c' always by the right eye. b is seen by the right eye except in the experiment illustrated in Fig. 3 by the solid circles. Z' is the fixation point viewed binocularly. patches are exposed to one (i.e., the right) eye and the test patch is exposed to the other eye (solid circles, Fig. 3), no such effect occurs at all! On the contrary, when the inducing patch becomes very bright, the test patch actually increases slightly in brightness. Thus, simultaneous brightness contrast requires the test and inducing patches to be exposed to the same eye and this is strong evidence for a retinal interaction effect along the lines proposed by Mueller. It is not too surprising that this is the case, since modern electrophysiological techniques have made it possible to uncover exactly this sort of physiological interaction of adjacent retinal areas by studying the electrical discharge of single retinal ganglion cells. Both Kuffler,0 in the cat, and Barlow,2 in the frog, found that the response of such cells evoked by illumination of the retina with light could be inhibited by illumination of surrounding retina. Nevertheless, there is one major difficulty with the view that the effect illustrated in Fig. 1 is due to retinal interaction. Barlow, Fitzhugh, and Kuffler3 found in the cat retina at threshold that the effect was much more pronounced for long flashes (380 msec.) than it was for short ones (7 msec). On the other hand, Chinetti0 found exactly the opposite result by use of psychophysical binocular brightness matching technique above threshold. Because of this obvious discrepancy, measurements of the amount of simultaneous brightness contrast have been carried out for a wide variety of different durations (5 to 150 msec.) and inducing patch luminances (over 3 log,0 units). Method The apparatus (Fig. 4) has already been described in some detail elsewhere.1 The test patch h is formed by light which passes through the field stop E after emerging from aperture N. This aperture is imaged in the plane of the pupil of right eye by Maxwellian view. The comparison standard a is formed by light passing through aperture M and then emerging through a similar field stop at E. The image of M is focused in the

Simultaneous brightness contrast 49 Table I. Mean settings of brightness match (B,, = 6.57 FL. N = 30) Log,, B e.o, Observer Duration (msec.) 5 10 15 20 25 35 45 55 65 75 85 95 105 115 125 150 Steady 4.9 MA. D.C. 93.5 243.0 450.0 468.0 488.0 621.0 404.0 338.0 322.0 263.0 217.0 224.0 195.0 183.0 174.0 188.0 327.0 1,200 1,710 1,780 1,610 1,270 1,210 1,050 975 881 823 811 854 745 760 713 736 457 4.4 MA. D.G. 27.0 39.3 87.2 136.0 140.0 160.0 250.0 204.0 160.0 143.0 139.0 143.0 151.0 161.0 156.0 193.0 137.0 536 674 878 968 1,160 1,190 1,000 941 805 804 733 653 640 681 690 659 240 MA. 14.2 26.7 64.8 86.5 97.8 125.0 128.0 145.0 143.0 126.0 118.0 70.2 76.8 54.2 53.9 67.5 60.5 3.9 D.C. 124 139 188 271 347 371 435 433 390 356 336 341 305 276 293 216 66.6 MA. 12.9 15.4 15.9 22.8 23.9 28.1 30.6 28.1 32.0 29.6 31.6 33.4 28.8 29.5 27.7 24.7 19.8 3.4 D.G. 18.6 18.8 26.7 33.8 42.1 77.5 101.0 114.0 179.0 182.0 189.0 176.0 172.0 155.0 145.0 118.0 23.3 MA. 10.2 11.8 12.6 14.1 16.3 17.5 20.0 28.8 28.8 28.1 31.7 27.4 30.0 25.1 23.9 21.7 14.5 2.9 D.C. 8.81 13.20 15.20 14.60 14.20 14.60 15.00 17.00 20.00 22.40 24.20 25.10 23.30 23.10 21.30 22.80 18.80 K center of the entrance pupil of the left eye. The inducing pattern c-d is obtained by light passing through a field stop at U after emerging from the aperture V. This stop is also focused in the plane of the entrance pupil of the right eye. The image of each of these stops in the pupillary plane is somewhat smaller than the smallest possible natural pupil size, and thus artifacts from variation in light reaching the retina with changes in pupil size are obviated. Rotating and fixed polaroids K and L, and neutral filters at M, N, and V permitted variation of the luminance of the test field continuously and that of all three fields in step fashion. The roof prism and penta prisms were placed one above the other and so arranged that the position of the image of N remained fixed while that for the image of M could be varied by rotating the platform on which the two prisms were mounted. In this way the interocular separation of the two eyes was readily adjusted. For the special experiment in which the data illustrated in Fig. 3 (solid circles) were obtained, it was necessary to put both the test and the comparison patch into the left eye. To do this, the platform containing the roof and penta prism was removed and replaced by a special mounting containing two first surface mirrors (one above the other) which helped to image both M and N in the center of the entrance pupil of the left eye. The amount of light reflected by the upper mirror was slightly greater than that reflected by the roof prism and this accounts for the slight vertical displacement of the two curves in Fig. 3 at the = 100- GJ 30- l- 10-0 1 2 3 4 5 LOG INDUCING FIELD LUMINANCE FT-L Fig. 3. Simultaneous brightness contrast compared when the test was seen by the right eye (open circles) and when the test patch was seen by the left eye (solid circles). In each case the inducing patch was seen by the right eye. The ordinate is the luminance of b (Bb) necessary to establish a match to the standard which was always viewed by the left eye. The value of the luminance of the standard (B,,) for the open circles was 12.6 FL., for the solid circles it was 13.9 FL.

50 Alpern Investigative Ophthalmology February 1963 Fig. 4. Line drawing of the apparatus as viewed from above. very low luminance levels. The sectored discs I and II exposed the patterns for any desired duration once every four seconds. In the experiments shown in Table I, all the exposures were simultaneous, i.e., the onset and cessation of test, inducing patches and of the comparison standard were always the same. Measurements were made at five different inducing flash luminances and sixteen different durations. In a given experimental session only a single inducing patch luminance level was studied. Six matches were made at each exposure duration, beginning first with the shortest flash and then the next shortest, and so on. At the very beginning and at the very end of the session, sixmatches were made of the value of h required to match a (5 msec, exposures) when the inducing flashes were not exposed at all. Five repetitions of each experiment were carried out on each of the luminance levels of the inducing pattern and for each of the observers. A third observer completed only two experimental sessions under each condition. Nevertheless, his results confirmed in every way those obtained with the other observers. In order to be certain that the matches established between the test patch and the standard did not themselves vary with duration even in the absence of the contrast inducing field, a control experiment was carried out in which the duration of a and b were varied in absence of the inducing pattern. There were no differences in these measurements for different durations within the precision of the measurements (±10 per cent). Results The mean results of the 30 brightness matches for each of the two observers under each condition are summarized in Table I. While differences between the two observers exist, particularly in the magnitude of the effect at the higher inducing patch luminances, similar trends exist in both sets of data, and so the mean results are plotted in Fig. 5. When the luminance of the inducing patterns is very low, the longer flashes have the greater ability to reduce the brightness of neighboring flashes. However, as the luminance of the inducing patterns increases, the flash duration which is associated with the largest ability to reduce the brightness of surrounding patterns becomes shorter and shorter. A second feature of the results in Fig. 5 is the close parallel between flash brightness and flash contrast induction that they suggest. Broca and Sulzer 5 showed that the intensity of a light required to match constant high intensity flashes of different durations was at a maximum for intermediate duration flashes. While this effect has been familiar for a long time,

Simultaneous brightness contrast- 51 no one really understands it.' 1 Perhaps for this reason, attempts to attribute the phenomenon to some measurement artifact appear from time to time. (For the most recent of these, see Raab, Fehrer, and Hershenspn.") In Fig. 6 are plotted two different measurements of the phenomenon made on different observers and with different stimulus conditions. The results nonetheless are quite similar. In the experiments from which the lower set of results were obtained, binocular brightness matches were established between two stimulus patterns (similar to a and b in Fig. 2). a was held at a constant intensity but its duration was varied, while b was held at a constant duration (750 msec.) and its luminance was varied to make the binocular brightness match. The separation of the patterns sufficed to obviate the possibility of any interaction between the two flashes and in this way the possible artifact 100 200 LOG INDUCING FLASH LUMINANCE FLASH DURATION -msec. Fig. C. Two examples of measurements of the Broca-Sulzer curves. A, The test patch was ex 1000 K 800 600 '400 200 0 50 100 150 FLASH DURATION ( MSEC) Fig. 5. Mean data of two observers for the magnitude of B 6 necessary for a match with a when the duration of the entire pattern (illustrated in Fig. 2) was varied. The center of b was separated from the center of c (and c') by % degree. Each point is the mean of 60 measurements. given on the curve and its duration varied as the independent variable, The dependent variable was the energy (luminance x duration) of b needed for the match. The duration of b at any given luminance level of c-c' was held fixed but was varied between luminance levels in the following way: for the top curve 30 msec, for the next curve 15 msec, the one below that 10 msec, and the bottom curve 5 msec. All flashes synchronized at off. Tungsten light had an approximate color temperature 3.0 (10) 3 degrees absolute. B, The test patch was similar to a in Fig. 2, the matching flash similar to b. The test flash luminance was held fixed at the value given on each curve and its duration varied as the independent variable. The matching flash duration was held fixed (0.75 sec.) and its luminance varied to establish the match (dependent variable). The flashes were so synchronized that the test patch always appeared in the middle of the exposure of the matching flash. The light source was the same as that from which the data in A were obtained except that a Wratten No. 74 filter was placed in both beams.

52 Alpern Inveslij-nth : Ophthalmology February 1963 suggested as an explanation for the effect by Raab and associates 11 has been avoided. The similarity of these curves to these in Fig. 5, both in the way they vary with flash duration at any given luminance level and the way any given curve changes as luminance level is changed, is very remarkable indeed. Discussion The results in Fig, 5 make it possible to bring the psychophysical experiments on contrast into agreement with the electrophysiological data obtained on the cat retina. In the latter case Barlow and coworkers 3 found, at threshold, that a very long flash was a much more powerful contrast inducer than a very short one. Chinetti 11 found the opposite result psychophysically but he used flashes which were very much above threshold. Fig. 5 shows how a short exposure has a greater ability to induce contrast than a longer one at a high intensity and a smaller ability to induce contrast than a long flash at, or near, threshold. Thus, the present experiment by bringing the psychophysical and electrophysiological data into accord greatly strengthens theories of contrast based on retinal interaction effects such as the one Mueller proposed many years ago. It is not possible, however, to be nearly as certain when the question is raised as to whether or not the similarity of the curves in Figs. 5 and 6 is anything but fortuitous. The results illustrated in these figures are from different observers and quite different stimulus conditions so that from the available data it is not yet possible to say that those stimulus flashes which have the largest ordinate on the Broca and Sulzer plot also have the largest ability to induce a contrast effect upon neighboring flashes. Apparatus limitations, among other technical difficulties, have so far prevented an exhaustive analysis of this important empirical question, but a preliminary attempt to do so is illustrated in Fig. 7. For the Broca-Sulzer Fig. 7. The relationship between the ability of a flash to induce a contrast effect and its ordinate in the Broca and Sulzer experiment. Broca- Sulzer data from N. B., B«= 146 FL. Contrast data from M. A., B,, = 79 FL. Each point represents a flash of a different duration. B\ is the value of Bt, obtained in each experiment under steady state conditions. curve, rectangle a was exposed to the left eye for a variety of durations but held at a fixed intensity (146 FL.), while rectangle b (exposed to the right eye) was held at a fixed duration (750 msec.) while its luminance was varied so that a brightness match was obtained. The contrast experiment was exactly like those from which the data in Fig. 5 were obtained. The inducing flash luminance was 79 FL.; the test flash luminance was 6.57 FL. Taking into account the difference in the area of the inducing pattern in the contrast experiment as compared to that of the test flash in the Broca-Sulzer experiment, the stimulus conditions were approximately comparable, although different observers were used in the two cases. The results from each experiment revealed that the intermediate flashes had the greatest effect, the maxima in the two experiments being approximately the same (85 msec, in the contrast experiment, 95 msec, in the other). In each case, what is plotted in the figure is the ratio of B\, needed to establish a brightness match with a for each flash duration and that required to

Volume 2 Number 1 Simultaneous brightness contrast 53 1 do so when a and b (and in the contrast experiment, c and c') had reached a steady state (B' b ). While the data do seem to show some sort of linear relationship between the two very different kinds of measurements, the extent to which the measured points fall on the empirical line is not nearly as satisfactory as one would wish, particularly at the upper right-hand part of the graph. The relationship between ordinate on the Broca and Sulzer curve and ability to induce contrast is clearly not 1:1, since the slope of the line in the figure is only 0.26. This means that unit increase B h /B\ in the contrast experiment is associated with an increase of 3.84 in B b /B\ in the Broca and Sulzer settings. The relation between the ordinate of the Broca and Sulzer curves and the ability to induce contrast in neighboring flashes illustrated in Fig. 7, as well as the relations illustrated in Fig. 3, makes it very tempting to devise a neurophysiological model relating the brightness of a flash and its ability as a contrast inducer in a more specific way. However, with the dearth of data presently available it does not seem worth while to attempt any detailed theoretical analysis of possible relationships between these two different results of retinal excitation. Further measurements need to be made with identical flashes in each of the two experiments on the same observers before more specific theoretical treatment can prove profitable. REFERENCES 1. Alpern, M., and David, H.: The additivity of contrast in the human eye, J. Gen. Physiol. 43: 109-126, 1959. 2. Barlow, H. B.: Summation and inhibition in the frog's retina, ]. Physiol. 119: 69-88, 1953. 3. Barlow, H. B., Fitzhugh, R., and Kuffler, S. W.: Change of organization in the receptive fields of the cat's retina during dark adaptation, J. Physiol. 137: 338-354, 1957. 4. Brindley, C. S.: The physiology of the retina and the visual pathways, London, 1960, Edward Arnold & Co., p. 155. 5. Broca, A., and Sulzer, D.: La sensation lv.minen.se en fonction du temps, J. de Physiol. 4: 632-640, 1902. 6. Chinetti, P. J.: The effects of reduced exposure duration on simultaneous brightness contrast, Ph.D. thesis, University of Wisconsin, 1957. 7. Helmholtz, H. L. F.: Handbook of physiological optics, 11 (Trans, from the 3rd German edition by J. P. C. Southall). Optical Society of America, 1924, pp. 264-296. 8. Hering, E.: Zur Lehre vom Lichtsinne, Sitzungsb. d. Wiener Akad. G8 (a$fe): 186-201, 229-244, 1873. "*" 9. Kuffler, S. W.: Discharge patterns and functional organization of mammalian retina, ]. Neurophysiol. 16: 37-68, 1953. 10. Mueller, J.: Elements of physiology (Trans, from the German by W. Baly), London, 1842, Taylor & Walton, vol. 2, p. 1187. 11. Raab, D., Fehrer, E., and Hershenson, M.: Visual reaction time and the Broca-Sulzer phenomenon, J. Exper. Psychol. 61: 193-199, 1981. Discussion Dr. Hermann M. Burian, Iowa City, Iowa. Dr. Alpern's paper is a difficult one to discuss, if by discussion is meant the raising of critical questions. His work is always so careful and meticulous that one is hard put to find fault with his methodology or his results. This also applies to the paper which he has just presented to us, which as anyone familiar with Dr. Alpern's work realizes, is only a tile in the mosaic of his work on contrast phenomena. In this particular study he has given us first of all an elegant demonstration that simultaneous contrast is indeed due to the interaction of adjacent retinal areas, as was taught by Johannes von Miiller and Hering. Nevertheless, I am somewhat uneasy about the fact that he makes binocular brightness matches, even though he has assured me yesterday when I discussed this with him that this is a well-accepted technique. Dr. Alpern has furthermore confirmed with refined methods the observations of Broca and Sulzer. This confirmation is of value since it can no longer be maintained that the effects noted by Broca and Sulzer might be measurement artifacts. Lastly, a question about Dr. Alpern's premise that there is an obvious contradiction between the neurophysiological results of Kuffler and Barlow and the psychophysical findings of Chinetti. Since the former worked at threshold, the latter with flashes very much above threshold, there was no reason to assume a priori that the results of these investigations should give comparable results. In fact one might have rather expected a discrepancy. Dr. Alpern has confirmed the difference in response to light flashes at threshold intensity and at supraliminal intensities and thus

54 Alpern Investigative Ophthalmology February 1963 removed any question about a contradiction between the p.sychophysical and neurophysiological data. I wish to thank Dr. Alpern for his courtesy in supplying me with his manuscript in advance of the meeting. Dr. Alpern (closing). Binocular brightness matching was introduced in 1934 by W. D. Wright (Proc. Roy. Soc. s.b., 115: 49-87) and since then it has proved to be an extremely powerful tool in our efforts to comprehend a variety of contrast and adaptation phenomena. However, there are limitations to the kinds of questions one can ask with these methods, and I have little doubt that as our knowledge increases we will be required to develop better techniques in order to answer the most basic remaining questions. While I agree with Dr. Burian that we do not know enough about the physiology of brightness perception to predict a priori that suprathreshold and threshold light flashes would behave similarly in a lateral inhibition paradigm, this was clearly the most parsimonius assumption. The fact, as these experiments demonstrate, that this assumption was invalid makes the ultimate understanding of brightness perception more, not less, difficult.