Assimilative hue shifts in color gratings depend on bar width

Transcription

1 Perception cl Pgychophysics 1986, 4 (6) Assimilative he shifts in color gratings depend on bar width CLEMENS FACH and LINDSAY T. SHARPE Albert-Ldwigs-Universitiit, Freibrg; West Germany He shifts were measred in isolminant color gratings whose bar width was varied from ' to ' ofvisal angle. Sbjects matchedthe hes in each grating with individal Mnsell swatches. He shifts were largest for bar widths of '; however, they depended on the color combination sed. Green and red shifted toward (i.e., assimilated with) whatever second grating color they were paired with. Ble, on the other hand, assimilated with red and with yellow, bt remained relatively nchanged when combined with green. Yellow shifted only minimally, regardless of the second grating color. He shifts decreased with increasing stripe width and disappeared between4.5' and 7.5'. Comparedwiththe assimilative he shifts, color contrast effects were slight or absent. These reslts cannot be attribted merely to chromatic aberration, maclar pigment, eye movements, or field size. The inflence of two or more adjoining chromatic fields on each other's perceived color and lightness has been known for a long time. In fact, as early as 1835,Chevrel (1969/1835) described the sbtle color changes introdced by weavers into tapestries to compensate for sch effects. After performing extensive experiments to measre these effects on apparent he and satration, he was the first to qantify simltaneos color contrast (e.g., red placed next to ble appears yellow). Later, in 1874, von Bezold reported an opposite phenomenon, the so-called spreading effect. He observed this phenomenon in handpainted ornaments in which areas of different colors were interspersed with white and black patterns. Even thogh the lminance of the colored areas remained constant, they appeared lighter when interspersed with white patterns and darker when interspersed with black patterns. Sch changes in perceived lightness are called lightness assimilation, indicating that the lightness of the achromatic pattern has been indced into, or assimilated by, the adjacent portion of color. A more recent term for assimilation appearing in the psychophysical literatre is similitde (DeValois & DeValois, 1975). In 196, Helson and Joy reported qantitative measrements of assimilation effects in achromatic gratings. According to their reslts, assimilation and contrast depend on stripe width, that is, spatial freqency. In high-spatialfreqency gratings, the lightness of white or black bars This research was spported by the DetscheForschngsgemeinschaft, SFB 7, Teilprojekt A6. We thank L. A. Spillmann for his constant advice. We also thankthe anonymos referee who gave s the idea for Figre 4. K. Knoblach, J. Werner, A. Valberg, and V. Volbrecht read andimproved an earlier draft of this manscript. Parts of the manscript were presented at the spring 1983 meeting of the German Physiological Society, Mainz, FRO, and at the Sixth Eropean Conference on Visal Perception, Lcca. Italy, September The athors' address is Albert-Ldwigs-Universitiit, Nerologische Klinik, Hansastrasse 9, 78 Freibrg, West Germany. drawn onto a gray backgrond is assimilated, so that the gray looks lighter when combined with the white grating bars and darker when combined with the black grating bars. With increasing stripe width, this assimilation becomes weaker and ltimately changes into contrast. The gray then looks darker when combined with the white bars and lighter when combined with the black bars. These observations were confirmed by Walker (1978), who reported a critical stripe width of abot 8' of visal angle for the transition from assimilation to contrast. Assimilative shifts can also be fond in color gratings. This was first described by Schober and Mnker (1967), who sed rectanglar color gratings made from papers of different hes, lightnesses, and satrations. The sbjects' task was to view the gratings at varios distances and to report qalitatively the perceived change in the he of the grating bars. For large observationdistances (i.e., for high spatial freqencies) the sbjects reported a shift of the perceived grating colors toward each other (i.e., assimilation). In a red-green grating, for example, the red appeared yellowish-red and the green appeared yellowishgreen. The assimilation effects disappeared at critical stripe widths between 1' and 3', depending pon the grating color combination. Wider bars, which might have yielded a transition from assimilation to contrast similar to that fond for achromatic gratings, were not stdied. More recently, color contrast was fond in an experiment reportedby Ware and Cowen (198). Using a techniqe adopted from color indction experiments, they measred the he shifts prodced by indcing stripes of 18' on test stripes of 6'. The colors of the indcing stripes were highly satrated, whereas those ofthe test stripes were of varios lower satrations. Ware and Cowen mainly fond colorcontrast effects. Only in a few conditions did they find assimilation. It ths seems that the previos stdies on assimilation and contrast may be smmarized as follows: Both assimi- Copyright 1986 Psychonomic Society, Inc. 41

2 HUE SHIFfS IN COLOR GRATINGS 413 lation and contrast are observed for achromatic gratings, depending pon the stripe width (Helson & Joy, 196; Walker, 1978). Coarse gratings show contrast, whereas fine gratings show assimilation. Only assimilative shifts have been reported for color gratings differing in satration and lightness, provided the dty cycle is eqal (Schober & Mnker, 1967). However, if the dty cycle is neqal, then color contrast, indced by the wider stripes onto the smaller ones, can be obtained (Ware & Cowen, 198). These fmdings prompted the following for qestions. First, do both assimilative and contrast color interactions occr between grating stripes ofeqal dty cycle that have eqal lightness and satration? Second, if so, what are the magnitdes ofthe perceived he shifts? Third, do sch he shifts depend on stripe width? Forth, is the amont of the shift specific to the grating color combination sed? METHOD Apparats and Stimli Sqare-wave gratings of eqal dty cycle were made p of for different Mnsell papers: SR (red), SY (yellow), SG (green), and SB (ble). The papers had dominant wavelengths for cm standard iliminant C of 648 nm, S79 nm, SI4 nm, and 441 nm, respectively, as calclated from the x-y coordinates given in Wyszecki and Stiles (198). They were all approximately eqal in lightness (vale 6) and satration (chroma 6). Chroma 6 was the highest chroma notation that cold be sed while maintaining the grating colors at the same lightness level. All six gratingcombinations were constrcted. The gratings were illminated by a MACBETH daylight lamp (Type BBX-34, Exective) whose central axis of incidence was at an angle of 4So from the normal to the grating srface. The lamp was operated at a color temperatreofabot 7,5 K, approximately corresponding to cm standard iliminant C. Its lminance was S7 cd/m-. By changing the observer's viewing distance while keeping the stimls display fixed in place, we varied the anglar stripe width ofthe grating bars in seven steps from ' (IS cpd) to '(1.S cpd). This procedreentailed thatthenmber ofbars in thegrating stayed constant (in total 4) while the overall width (7.S cm) of the grating varied from I.Iso to I1.So and theoverall height (1 em) varied from I.S3 to IS.3. The gratings were always oriented vertically and were always shown on a gray (Mnsell N6) srrond. Dring constrction of the gratings, great care was taken to paste and align the Mnsell papers exactly. The finished stimli were then careflly inspected to ensre that no peeled-off parts, shadows, or finger marks were conspicosly present. Some tiny imperfections cold not be avoided, bt these were deemed too small to affect the sbjects' jdgments. Sbjects There were S observers, all of whom were emmetropes with normal color vision as indicated by the Farnsworth-Mnsell IOQ-He Test. Procedre The task ofthe observer was to choose from a set of Mnsell color swatches (pls gray) the color that most clearly matched the perceived hes of the grating bars. The observer alternately viewed the grating and the samples. There was no fixation point. The Mn- sell color swatches were 7.S em in width x 1 em in height (i.e., they had the same dimensions as the grating stimli). They were monted directly above the grating stimli and were arranged according to the color circle in steps ofs Mnsell he nits. The entire set of matching samples was always on display. The viewing distances (and sizes) ofthe matching samples and the grating stimli were therefore concomitantly varied throghot the experiments. All ofthe swatches had a Mnsell valelchroma notation of6/6. Ths the effect only of he, and not of satration and lightness, was measred. Before beginning the experiment, the observers were first trained to match the colors of the grating with the Mnsell swatches and to se visal interpolations when necessary. Dring the experiment, the grating color combination and the observation distance were changed in a random order. While sch changes were being made, the stimls display was covered to avoid carryover effects de to persistence or preexposre. Each grating at a given observation distance was presented two separate times. Each sbject participated in two -h sessions. RESULTS Figres 1-3 show the average reslts of the color matches for the six gratings. Each data point is the average of two measrements for each of 5 sbjects. The horizontal bars represent the standard error ofthe means [i.e., the standard deviation!(the nmber ofobservations - 1)]. He shifts were fond for all color combinations. However, the amont of the shift for a given color depended on the color with which it was paired. For instance, in the yellow/ble grating (Figre 1, top) ble appeared greenish; it had therefore changed in appearance in the direction of the other grating color, yellow. This is an example of he assimilation. For the finest grating bars (' ofvisal angle), the shift amonted to more than 5 Mnsell he steps, corresponding to a shift in dominant wavelength of abot 6 om. Unlike the ble bars in the grating, the yellow bars displayed only slight variations in he as a fnction of stripe width; for stripe widths below 6', they even showed a mild he contrast. In the greenlble grating (Figre 1, bottom), the ble bars showed little he shift, except for the narrowest stripes, which shifted in appearance toward ble-green (i.e., they showed he assimilation), and the widest stripes, which showed he contrast. The green bars also appeared ble-green and therefore also displayed he assimilation. However, they did so mch more saliently. Again, the he shift depended on bar width: for a bar width of7.5', no he assimilation was fond. On the other hand, for a bar width of 1', slight contrast was obtained. When combined with red bars, green bars displayed assimilation (Figre, top): theirhe shifted toward yellowgreen. With increasing stripe width, this assimilation was weakened, and at a bar width of4.5' it disappeared. Wider bars showed a slight change in the direction of he contrast. Likewise, the red bars showed assimilation by shifting toward yellow-red. At 6', however, the assimilation disappeared. In the yellow/green grating (Figre, bottom), green again showed assimilation. The effect was largest for bar widths of3' and 4.5', and was still present

3 414 FACH AND SHARPE Yellow-Ble Grating :c e- 1. ;:; >. lo Il) g'.4 -'. 3 Ie- 4.5.S:.5 6 s:; " Q/ - 5Y 1Y 5GY 1GY 5G 1G 5 1 5a Green - Ble Grating 1. 3 < i.8 : c: 5-6.c ILl.t _ " s :c 1.<:.4 Cl..J. 1G SBG 18G 56 Figre 1. Average reslts of color matches for a yeuowlble grating (top) and a green/ble grating (bottom). On the left ordinate, log spatial freqency is plotted (in cpd); on the right ordinate, the corresponding stripe width (in min of arc or visal angle) is given. Mnsen notations are plotted on the abscissa. The thin vertical lines indicate the actal Mnsen notations of the grating colors. The horizontal bars represent the standard errors of the mean matches. II) at 7.5'. At 1', it disappeared. Yellow assimilated for the finest stripes, and displayed contrast for stripes wider than 7.5'. When ble and red were paired (Figre 3, top), both shifted toward each other (i.e., they assimilated); the effect was largest for the narrowest bars, and diminished as the bar width increased. Likewise, when red and yellow were paired, they both displayed assimilation for the fine stripes (Figre 3, bottom). However, for red, the amont of the shift varied as a fnction of stripe width, and the widest red stripes even showed contrast, whereas for yellow the amont of shift stayed constant and was always small in magnitde. Ths, all colors displayed strong assimilation except yellow, which showed little effect in all combinations. In the yellow/ble grating (Figre 1, top), yellow showed no assimilation for bars p to 6', and for wider bars it displayed a small contrast effect. When combined with green (Figre, bottom), yellow showed assimilation only for the narrowest stripes. In combination with red

4 HUE SHIFTS IN COLOR GRATINGS 415 (Figre 3, bottom), yellow showed a somewhat stronger assimilation, bt nlike that ofthe other colors, this shift was independent of stripe width. It shold be noted that in none of the color combinations cold strong contrast be obtained over the range of stripe widths sed. This is made clear in Figre 4, which qalitatively smmarizes the data from Figres 1-3. In this figre, the thin and thick arrows represent he shifts for coarse and fine gratings, respectively. The very slight contrast effects are too small in magnitde to be shown. DISCUSSION On the basis of these reslts the qestions posed at the beginning can now be answered. Strong assimilative color interactions occr between gratings of eqal lminance and satration, bt correspondingly strong color contrast interactions do not. The size of the assimilative effects varies considerably, from 1 to 6 Mnsell he steps, with greater he shifts occrring for the fmer stripe widths. However, the amont ofthe shift depends on the grating color combination sed. It is greatest for ble in combination with yellow and for green in combinationwith ble, and smallest for yellow in all combinations. Three points abot these findings need to be discssed: the small he changes for yellow, the lack of color contrast in coarser gratings, and the strong he assimilation in narrow-striped gratings. For the observation that yellow behaves very differently from other colors when paired in gratings, we can offer Red -Green Grating : Q, f'ii.8!' C7! 14 Ii Q, J..5R 1R SYR 1YR SY 1Y SGY 1GY 56 Yellow- Green Grating 6,J; : 7.5 s 1. <- Vi ;; e- 1. > c:.8.6. ȧ. V1 8'.4.J. 5Y 1'( SGY 1GY 5.li s 1 l!l Figre. Same as Figre 1, for a red/green (top) and a yellow/green (bottom) grating.

5 416 FACH AND SHARPE Ble- Red Grating ' a. >. c: ::::I :;: a..4 C'l J c: ụ.. :! 6.s: 7.5 :g B SPa 1PB 5P 1P 5RP 1RP 5R Red-Yellow Grating :c a >- c: c: ::::I - 6..c ' a. C'l..J 1.4 1:. 5 a 5R 1R 5YR 1YR Figre 3. Same as Figre 1, for a ble/red (top) and a red/yellow (bottom) grating. no satisfactory explanation. We can only point ot, first, that the yellow Mnsell paper we chose, 5Y (chroma 6), did not appear strongly yellow at any of the vale levels sed (Nl-N6) (in fact, its appearance ranged from sand to ochre); and second, that the pecliar behavior of yellow is not an isolated phenomenon. It is also observed, for example, in the he constancy of yellow in the Abney effect (see Krtenbach, Sternheim, & Spillmann, 1984). The relative absence of color contrast in the findings presented here is pzzling in view of the fact that sch contrast effects have been reported before for chromatic gratings (Ware & Cowen, 198). Two factors, however, may be responsible for the failre to find strong color contrast in the present stdy: the spatial configration and the satration of the stimli. Ware and Cowen may have fond color contrast becase they sed optimal conditions for indcing it. They sed 'indcing stripes ofthe greatest possible satration and of greater visal angle than the test stripes (l8' vs. 6'). Sttong contrast indction wold therefore be expected (Helson & Joy, 196; V alberg, 1974). In the present stdy, on the other hand, the indcing and test stripes were comparatively lower in satration and had the same width. Conseqently, less contrast wold be expected. Another possible reason for or finding only slight color contrast is that some of or color

6 HUE SHIFTS IN COLOR GRATINGS 417 RP- P-- y : \ I PB BG Figre 4. Schematic diagram smmarizing the reslts of Figres 1-3. Thick andthinarrows represent he shifts for fine and coarse gratings, respectively. The he notationopposite the arrowhead indicates the indcing color. The length of each arrow approximately represents the magnitde of the assimilative he shift. Prple was not sed in or experiments, bt is inclded in the color diagram for completeness. pairs, particlarly 5Y and 5B, were nearly complementary to each other. Ths, any contrast effects were probably manifested as shifts in satration, not as shifts in he. Inasmch as we measred only he changes, we failed to detect sch contrast effects. In the ftre, therefore, experiments shold be ndertaken to investigate systematically the roles ofdty cycle and stripe width in perceived he shifts. Moreover, experiments are reqired in which perceived shifts not only in he, bt also in lightness and satration, are simltaneosly measred, as was done by Ware and Cowen (198). To explain why strong he assimilation is fond for narrow-striped gratings, we mst first consider the effects of varios physical and physiological factors. These inclde chromatic aberration, lminance edges, involntary eye movements, and field size. The first ofthese, the axial chromatic aberration ofthe eye, can be qickly eliminated as a significant factor. It is nlikely to contribte importantly becase the Mnsell papers have complex, continos, spectral reflectance crves. Almost all visible wavelengths are present in each paper, and the dominant wavelength does not necessarily correspond to the peak of the reflectance crve. Moreover, the strongest assimilation effects occr in a direction opposite that predicted by chromatic aberration. For instance, short wavelengths are more strongly affected by chromatic aberration than are middle or long wavelengths. Ths, in the yellowlble grating (Figre I), light from the ble bars wold be expected to overlap and mix with light from the yellow bars, thereby changing the latter's appearance toward bleor, in the case of cancellation, toward gray. Bt in fact, the ble bars display more assimilation than their yellow conterparts. Frthermore, strong assimilation is fond for those color combinations in which the aberration is qite small, for example, red/yellow (Figre 3). The second factor, the lack of lminance edges in the gratings, can also be largely disconted. Withot these edges, accommodation is poor (Wolfe, 1983), and, conseqently, the retinal image may be smeared. However, the color mixing or smearing reslting from the se of eqilminant grating bars cannot explain the present findings, becase he shifts shold be roghly symmetrical, and this was not the case in the present stdy. The third factor, involntary eye movements, similarly fails to explain or findings. One cold arge that sch eye movements transfer afterimages onto adjacent stripes and thereby inflence the perceived color ofthose stripes. Again, however, the reslting he shifts shold be symmetrical for both grating colors. As evidenced above, this was not fond to be the case. The forth factor, the field size, cannotbe so easily disconted. This is becase we varied spatial freqency by changing the observer's viewing distance. This procedre had the advantage of keeping the total nmber of bars constant. Ths it controlled for the effect of stimls redndancy on the potency of assimilation and contrast. However, the procedre allowed the field area to vary over a range of 1: 1. Therefore, it might reasonably be arged that some ofthe observed phenomena reslted not only from the effect ofvarying spatial freqency, bt also from the effect ofconcomitantly varying field size. This is plasible for three reasons. First, it is well known that field size has an effect on the precision of color matching: Observers can repeat color matches more precisely when they are obtained in large fields (1 ) than when they are obtained in small fields ( ), nder otherwise identical observing conditions (Wyszecki & Stiles, 198). However, sch differences have been fond for small, concentric, color-matching stimli and not for more complex grating stimli in which the field size is mch larger than the stripe width. Ths it is nlikely that the difference between the largest (11.5 x 15.3 ) and the smallest (1.15 x 1.53 ) field sizes, per se, prodced or phenomena. Second, field size cold have confonded the reslts de to the inhomogeneos distribtion ofthe maclar pigment across the retina. This pigment is most dense in the foveal region and trails off in the periphery. Ths, for the highest spatial freqencies (viz., the smallest field sizes), the grating was almost entirely screened by the maclar pigment, whereas for the low spatial freqencies (viz., the largest field sizes), only the central portion of the grating was screened by the maclar pigment. Since the pigment absorbs maximally at a wavelength of abot 46 om (Wyszecki & Stiles, 198), it acts like a filter to redce the effect ofshort-wave stray light. Ths, for or low-spatial-freqency gratings more short-wavelength light reached the peripheral photoreceptors. This cold

7 418 FACH AND SHARPE considerably alter the phenomena, especially those involving the 5B Mnsell grating combinations. Indeed, it is tre that the lower freqency gratings with the 5B Mnsell paper (see Figre 1, top and bottom, and Figre 3, top) shift in the direction of greater bleness. However, we think it nlikely that the differing extents to which the peripheral retina was stimlated cold have significantly contribted to the phenomena, not only becase the sbjects were instrcted to always base their responses on the central portion of the gratings, bt also becase they were not reqired to se strict foveal fixation. Discrimination in the ble-green region ofthe spectrm becomes poor only when very small fields (1'-' in total extent) are sed and when strict foveal fixation is maintained (Bedford & Wyszecki, 1958; Wyszecki & Stiles, 198). Third, field size cold have affected the reslts de to the changes in distribtion of the ble or short-wavesensitive cones with retinal eccentricity. The ble cones have a minimm in the central fovea and a maximm abot 1 to in the parafovea (Williams, MacLeod, & Hayhoe, 1981a, 1981b). Moreover, their distribtion varies systematically throghot the entire retina, with discrete sensitive peaks being separated by large insensitive gaps (Williams et al., 1981a). However, it is nlikely that sch spatial inhomogeneities cold have significantly affected or reslts, becase most ofor grating stimli had bars well above the acity limit ofthe ble cones, which is abot 1-16 cpd for lminance gratings (Stromeyer, Kranda, & Sternheim, 1978). At any rate, the fndamental limit on resoltion is imposed not by the spacing of the receptors, bt by other physical factors (Williams, Collier, & Thompson, 1983). Only for the very smallest grating spatial freqencies sed (15 cpd) is it conceivable that the resoltion limit of the ble cones cold have significantly contribted to assimilation. For the reasons given above, we feel that the assimilation phenomena reported here are not explained merely by chromatic aberration, eye movements, maclar pigment, or field size. Rather, we feel that they are better explained in terms of the receptive field organization of the visal system. This idea also occrred to Hrvich and Jameson (1974), who proposeda model based on the physiological observations that receptive field sizes vary from the periphery to the fovea and show considerable variation in any given retinal region. Hrvich and Jameson (1974) arged that when a grating pattern is imaged on the retina, nerons that have large receptive fields will integrate over a nmber of stripes and will respond as if the light were actally mixed on the retina, whereas nerons that have small receptive fields, at the same patch of the retina, will resolve the individal stripes and maintain the spatial resoltion of the linear pattern. Ths, the net reslt for fine grating stripes, bt not for coarse ones, will be a blending ofhes and brightnesses (assimilation), bt a preserving of spatial pattern resoltion. This general model of assimilation, which does not inclde contrast, may be made more specific by the attribtion of doble-opponent properties to the responsible nerons (see Michael, 1978a, 1978b, 1978c). Sch an expanded version of the model can be sed to explain why he shifts are obtained for nonopponent colors, sch as green and ble (see Figre 1), and why hes remain constant when the bar width exceeds the receptive field center's size (see Hrvich, 1977, 1981, for more details). REFERENCES BEDFORD, R E., & WYSZECKI, G. (1958). Wavelength discrimination for point sorces. Jornalofthe Optical Society ofamerica, 47, 564. CHEVREUL, M. E. (1969). De la loi de contraste simltane des colers. Paris: Leon Laget. (Original work pblished 1835) DEVALOIS, R., & DEVALOIS, K. (1975). Neral coding ofcolor. In E. C. Carterette & M. P. Friedman (Eds.), Handbook ofperception (Vol. 5, pp ). New York: Academic Press. HELSON, H., &: JOY, V. L. (196). Domains of lightness assimilation and contrast. Psychologische Beitrage, 6, HURCH, L. M. (1977). Two decades of opponent processes. In G. Wyszecki & B. F. Billmeyer, Jr. (Eds.), Color 77, International Color Association (pp ) New York: Adam Hilger. HURCH, L. M. (1981). Color vision. Snderland, MA: Sinaer Associates. HURVlCH, L. M., & JAMESON, D. (1974). Opponent processes as a model of neral organization. American Psychologist, 9, KURTENBACH, W., STERNHEIM, C. E., &:SPILLMANN, L. (1984). Change in he of spectral colors by diltion with white light (Abney effect). Jornal ofthe Optical Society ofamerica, A,I, MICHAEL, C. R. (l978a). Color vision mechanisms in monkey striate cortex: Dal opponent cells with concentric receptive fields. Jornal ofnerophysiology, 41, MICHAEL, C. R. (l978b). Color vision mechanisms in monkey striate cortex: Simple cells with dal opponent-colorreceptive fields. Jornal ofnerophysiology, 41, MICHAEL, C. R. (1978c). Color sensitive complex cells in monkey striate cortex. Jornal ofnerophysiology, 41, ScHOBER, H., &: MUNKER, H. (1967). Unterschngen z den Ubertragngseigenschaften des Gesichtsinns fiir die Farbinformation. Vision Research, 7, STROMEYER, C. F., KUNDA, K., &: STERNHEIM, C. E. (1978). Selective chromatic adaptation at different spatial freqencies. Vision Research, 18, VALBERG, A. (1974). Color indction: Dependence on lminance, prity, and dominant or complementary wavelength of indcing stimli. Jornal ofthe Optical Society ofamerica, 64, VON BEZOLD, W. (1874). Die Farbenlehre, Branschweig, West Germany: Westermann. WALKER, J. T. (1978). Brightness enhancement and the Talbot level in stationary gratings. Perception & Psychophysics, 3, WARE, C., &: COWEN, W. B. (198). Changes in perceived color de to chromatic interaction. Vision Research,, WILUAMS, D. R., COLUER, R. J., & THOMPSON, B. J. (1983). Spatial resoltion of the short-wavelength mechanism. In J. D. MoHon & L. T. Sharpe (Eds.), Color vision, physiology and psychophysics (pp ). London: Academic Press. WILUAMS, D. R., MACLEOD, D. I. A., &: HAYHOE, M. (198Ia). Foveal tritanopia. Vision Research, 1, WILUAMS, D. R., MACLEOD, D: I. A., &:HAYHOE, M. (l98ib). Pnctate sensitivity of the ble-sensitive mechanism. Vision Research, 1, WOLFE, J. (1983). Hidden visal processes. Scientific American, 48,, WYSZECKI, G., &: STILES, W.S. (198). Color science (nd ed.). New York: Wiley. (Manscript received May 8, 1985; revision accepted for pblication September 17, 1986.)