1 ns,o,, Rus Vol
16. pp. 1099 10 ,104. Pergrim
Press 1976. Pnntcd I” Great Bntam
ROD AND CONE CONTRIBUTIONS TO PERIPHERAL COLOUR VISION’ BJOKNSTABELLand ULF STABELL Institute of Psychology. Universit! of Oslo. Norway (Rrceioed 21 Nouemher 1975) Abstract-The changes in chromaticities were measured. in a dark-adapted state. when monochromatic lights were moved from the fovea to 2.5’ and 7.5’ temporally. In order to analyze the contributions of the rod and cone mechanisms to these changes. the chromaticities were measured during dark adaptation following high light adaptation. For several min during the cone-plateau period, the colourmixture functions remained invariant and the outlines of the test field appeared distinct and clearcut. Just prior to the cone-rod break of the dark-adaptation curve. however, the chromaticities generally started to change and the shape of the test field became somewhat blurred. The evidence is judged to indicate that the chromaticities measured during the period of invariance are due to pure cone activity. while the shift in chromaticities measured during the further stay in the dark is due 10 the effect of rod intrusion.
Purkinje (1825) reported that the colour sensation may change when a coloured stimulus is viewed with increasing obliquity. Early attempts at describing this phenomenon have been largely qualitative and suffer from lack of information defining the experimental conditions. Nevertheless, most of these early studies agree that as coloured stimuli are moved outward from the fovea. reds and greens tend in general to look yellow, violets and blues to look blue. Furthermore, the colours are said to become desaturated and to appear colourless at the extreme peripheral retina (cf. Hess, 1889). Later on, quantitative methods have been developed in order to describe the colour sensation of the peripheral retina. The first modem investigation of the extrafoveal colour-matching functions was reported by Wright (1946) who determined the spectral coefficient curves at 4’ in the parafovea, using a 2’ bipartite field. In general, the curves were found to be similar to the spectral coei?icient curves obtained from the fovea although small differences were observed for the red and blue coeflicients. The main differences observed were explained on the assumption that the rod activity might desaturate the colour of the extrafoveal field. This study was followed up by Gilbert (1950). She found that at 4’ out and at low intensities, the colour sensation was reduced to a form of dichromatic vision similar to the tritanopic type. As one possible explanation she suggested that the presence of rods in the parafovea adds a colourless quality to the blue and blue-green responses. A new and powerful matching technique was introduced by Moreland and Cruz (1959). A fovea1 comparison field consisting of the Standard Wright primaries was used to match parafoveal stimuli at various eccentricities. Since the normalizing matches
’ The Norwegian Research Council for Science and the Humanities supported this study financially.
were made at the fovea on the wavelengths 494 and 582.5 nm the results could be plotted directly in the fovea1 WDW system. The method may thus be used to obtain quantitative informations on the colour appearance of peripherally presented stimuli as compared with foveal stimuli. Moreland and Cruz (1959) measured extrafoveal spectrum loci from lo” to 50’ and observed a progressive contraction of the peripheral colour gamut with distance from the fovea. The colour matching tended to become dichromatic at 25”30’ and monochromatic at w-50”. On the assumption that rods may feed into “red”, “green” and “blue” cone pathways, the desaturation of the peripheral field-found to be marked over the whole range from violet to green- was explained as an effect of rod intrusion. Furthermore, the change in hue toward green of the shorter wavelengths. observed when the test field was shifted from the fovea to IO’ and 15” out, was explained by a greater proportion of rods combining with “blue” cones than with “green” cones. Alternatively it was suggested that the change in hue could result from the convergence of “blue” cones to “green” pathways. Using the matching technique of Moreland and Cruz. an attempt has been made at isolating the effect of rod intrusion on the colour-matching functions (Stabell and Stabell, 1975). The chromaticities of monochromatic stimuli at 6” in the parafovea were measured twice during dark adaptation: (1) when the sensitivity of the cones, and (2) when the sensitivity of the rods, respectively, were assumed to have reached their absolute dark-adapted thresholds. In addition to a marked reduction in saturation, the rod intrusion appeared to be accompanied with a change in hue: colours of orange and green-yellow were found to change toward yellow, while colours of green-blue and violet changed toward blue. It should be noted that the same kind of changes in hue and saturation are said to occur when coloured stimuli are moved outward from the fovea. Hence, the results appear to indicate that rods may contribute to the
1099
BJORY STABELL and LLF STABELL
1100
shift in chromaticitv recorded when a test field is moved across the retina. On the other hand. it is generally assumed that cone mechanisms may contribute to this change [cf. Wooten and Wald. 19731. Supposing that both rod and cone mechanisms are involved. the question arises as to their relative contributtons. The aims of the present paper were twofold. Firstly. to measure the shifts in chromaticities. in a darkadapted state. when monochromatic stimuli are moved from the fovea to 2.5 and 7.5 temporally. Secondly. to attempt to analyze the contributions of rod and cone mechanisms to these shifts. For this second purpose, the chromaticities were measured during dark adaptation following high light adaptation. Evidence is presented indicating that the chromaticities measured during the cone-plateau period are due to pure cone activity. while the shift in chromaticities during the further stay in the dark is due to the effect of rod intrusion. XIETHODS (1) The Wright
calorimeter and calibration procedures in detail by Wright (1946). The calorimeter was arranged with three reflecting prisms in the W, spectrum, giving the instrumental primaries 650. 530 and 460nm. The unit matches were made symmetrically at the fovea on wavelengths 494 and 582.5 nm, reflected from the W, spectrum. All the observations were made with the right eye of one of the authors (US). a normal trichromat. In the following experiments the comparison field, consisting of the instrumental primaries, was applied at the fovea, while the test field was applied at the fovea. 2.5’ and 7.5’ temporally to the fovea. The test and comparison fields. each subtending I x 1 , were presented in succession. using 0.5~set flashes. To centre the exit pupil of the calorimeter the disappearance of the coloured fringes. caused by chromatic aberration of the eye. was used as used have been described
470. 480. 490. 500. 520. %(I. iho. JCI.III
The results of Experiment 1 are plotted m the fovea1 WDW system and are shown tn Fig. I, The data points in this and the following figures represent the means of five measurements. The total scuttcr m chromaticity
was about
0.04 m r and p. Figure
I;,
gives the mean chromaticities of monochromattc lights of equal brightness obtained at the fovea and 2.5’ temporally. while Fig. 1b gives the chromaticities at the fovea and 7.5 temporallq. In agreement with previous studies the results show that the saturatton of the different colours is reduced when the monochromatic stimuli are moved from the fovea toward the periphery. Furthermore. reds and greens change toward yellow. while violet changes toward hluc. The changes are somewhat more pronounced at 7.5 than at 2.5’. The chromaticitv shifts observed might be attrtbuted to variation in both rod and cone mechanisms. since rods probably contribute to the peripheral colour sensation under the present conditions of experimentation (cf. Stabell and Stabell. 1975). A new experiment was therefore performed in order to analyze the contributions of the two receptor mechanisms.
a criterion (cf. Wright. 1946). /Z.upcriment 1
(chramaricity
shifts
in a dark-adapted
srate)
The successive phases of the operation were as follows: (L)After 30 min dark adaptation, the eye was illuminated in 0.5~set flashes with a bipartite field subtending 1’ x 2’. located 7.5’ temporally to the fovea. The upper half of the field subtending 1’ x I’ was illuminated by monochromatic light of 650nm at an intensity 2 log units above the absolute dark-adapted threshold for the given stimulus measured at 7.5 . representing an intensity level of about IO photopic td. The lower half (the test field) was illuminated by monochromatic light of 460nm. The intensity of the test field was increased in O.l-log unit steps until a brightness match was obtained between the two fields. To obtain a match 3-5 flashes were needed. (2) 5 min dark adaptation. (3) Stimulation for 0.5 set at 7.5’ temporally with the 1’ x Ii test field of 460 nm at the intensity level recorded in phase I. (4) 1 set dark adaptation. (5) Stimulation for 0.5 set with the fovea) comparison field consisting of the instrumental primaries. The subject tried to establish a match by increasing or reducing the intensity of each of the three primaries. Only small adjustments were required since pre-experimentation had established an approximate match. (6) The match was established after four repetitions of phases 3-5 at 5-set intervals. (7) Phases 1-6 were repeated. except that between different runs the wavelength of the test field was varied: 460.
Fig. 1. Mean chromaticities within the fovea1 WDW system of monochromatic lights of equal brightness obtained in a dark-adapted state at the fovea and 2.5- (a) and at the fovea and 7.5’ (b). Filled and olain svmbols reoresent the chromaticities obtained at the fovea -and peripherally respectively. The data points in Figs. l-5 give the means of five observations.
Rod and cone contributions METHODS (2) II IS
general]! accepted that cones adapt to darkness at a more rapid rate than the rods. Accordingly. under appropriate experimental conditions. the dark-adaptation curve reveals two distinct branches: a rapid fall of threshold to a first plateau generally ascribed to pure cone function. followed by a second fall of threshold to a final level ascribed to rod function. The absolute threshold level during the cone-plateau period may remain fairly constant for several min. while the concentration of rod pigments changes rapidly. In fact. with a deep-red test light the threshold level may remain constant for more than 20 min. following the initial drop of the dark-adaptation curve (cf. Stabell and Stabell, 1975). Hence. it has been generally accepted that the response of dark-adapted cones may be measured independently of rod activity for a considerable period during dark adaptation. Eventually. however. rods may intrude and change the colour sensation. Evidence indicates that a relatively small participation of rods would cause a large desaturation or even the complete loss of colour (cf. Wald. 1960: 1961; Lie. 1963: Stabell. 1967). Furthermore. the rod intrusion would be expected to be related to the cone-rod break of the dark-adaptation curve. i.e. to occur relatively early for short-wavelength stimuli and successively later for longer-wavelength stimuli (cf. Stabell. 1967). Experiments 2a and 2b measure respectively the absolute threshold and the chromaticity during dark adaptation. Experiment 2a (&solute threshold
during
dark
adaprcrtion)
The successive phases of the operation were as follows. (I) Light adaptation. After 10min dark adaptation. the subject was light adapted for 3 min to a white light of a constant retinal illumination of about 60.000 photopic td. The size of the circular adapting field was 7’, centered 7.5’ temporally to the fovea of the right eye. (2) Threshold measurements. With the subject in complete darkness. the absolute threshold was measured every min during the first step of the dark-adaptation curve. and approximately every 3rd min during the second step. The size of the test field was I’ x 1’ and it was applied 7.5’ temporally to the fovea. using 0.5-set flashes. The intensity was increased in small steps from about 0.5~log unit below the expected threshold. (3) Repetitions of phases 1 and 2. except that the wavelength of the test field was varied between runs: 460, 490. 500. 520. 540. 560. 580. 600. 620 and 640nm were employed. (4) Phases 1-3 were repeated, except that the light adap tation and the test field were centered 2.5’ temporally to the fovea. Erperirnenr 2b (chromaricity shifts during dark cldaprarim) The successive phases of the operation were as follows. (I) Light adaptation. This was a repetition of phase 1 of Experiment 2a. (2) About 4 min dark adaptation. (3) Test-stimulation for 0.5 set with the 1. x I’ test field of 460 nm applied 7.5’ temporally. The intensity was 2 log units above the cone-plateau level as determined in Experiment 2a. representing an intensity level of about 10 photopit td. (4) I set dark adaptation. (5) Stimulation for 0.5sec with the fovea1 comparison field. consisting of the instrumental primaries. The subject tried to establish a match by increasing or reducing the intensity of each of the three primaries. OnIt small adjustments were required, since pre-experimentation had established an approximate match. (6) The match was established after four repetitions of phases 3-5 with 5-set intervals.
1101
to peripheral colour vision
(7) Phases 3-6 were repeated every min during the coneplateau period and approximate]] ever! 3rd min during the second fall of the dark-adaptation curve. (8) Phases l-7 were repeated. except that the wavelength of the extrafoveal test field was varied between runs: 460. 490. 500. 520. 540, 560. 580. 600. 620 and 640nm were employed. (9) Phases l-8 were repeated. except that both the light adaptation and the test field were centered 2.5 temporall!. (10) Phases 1-8 were repeated. except that the light adaptation and the test field were centered at the right eye fovea. while the comparison field was centered at the dark-adapted left eye fovea. During the cone-plateau period the extrafoveal stimuli in the 50%580nm region of the spectrum appeared too saturated to be matched by any fovea1 patch within the colour gamut of the RGB system. In these cases a reasonably good match could be obtained between the test stimulus and a monochromatic wavelength observed at the fovea. RESULTS The results of Experiments 2a and 2b are presented in Fig. Za-f. which gives the changes in the amount
of the instrumental primaries of the fovea1 comparison field needed to match the monochromatic lights of the test field during dark adaptation. together with the dark-adaptation curves. Only the results for some of the wavelengths are presented being the same in all essentials for the intermediate ones. The chromaticities of the monochromatic lights remained invariant and the outlines of the test field appeared distinct and clearcut for several min during the cone-plateau period. Just prior to the cone-rod break of the dark-adaptation curve, however. the chromaticities generally started to change and the shape of the test field became somewhat blurred. The change in chromaticities leveled off after a few min. With deep-red light of 640nm at 2.5. peripherally. the dark-adaptation curve revealed no cone-rod break. However, a shift in chromaticity was recorded at about 22 min dark adaptation. Since the unit matches were made at the fovea on the wavelengths 494 and 582.5 nm, the results can be plotted directly in the fovea1 WDW system. Figure 3 thus shows the spectrum loci in terms of the instrumental primaries obtained at about 5 min (filled symbols) and 30 min (open symbols) of dark adaptation. During the second phase of the dark-adaptation curve. the saturation of the different colours became markedly reduced, reds and green-yellow changed toward yellow and violet changed toward blue. These changes were somewhat more pronounced at 7.5 than at 2.5’. On the other hand no change in chromaticities was observed between 4 and 30min of dark adaptation when the test field and the light adaptation were centered at the right-eye fovea and the comparison field was located at the dark-adapted, left-eye fovea. DISCUSSION
Foilowjng a wide range of bleachings. both the psychophysical dark-adaptation curve of human rods and the regeneration of rhodopsin measured by retinal densitometry. appear to follow an exponential course of recovery (cf. Rushton and Powell. 1972:
2
540
2
6
6
om
460nm
I4
in MHk,
Ttmt
Fig.
14
10
mtn
min
22
22
26
26
30
30
0.5
LO
and
RGB
-3
-2
dark adaptation
curves
f p
measured during
2. Darkxtdaptatron
18
II)
Tune in dork.
K)
Rm-=-
2
nm
6
itt 2.5
(plain
I
rt)
I 22
22
min
min
I6
I
26
26
and 7.5
4 u
L
$
I
-I
-3 I,
0
& 0 b cp -2
z
system of n)o11ocllr(~millic
30
0.5
I 30
(tilled symhois).
the fovr;cI WDW symbols)
within
Time mdwk,
l4’
in dark,
Time
I
I4
f
IO
I
6
G---+-%-V580
I 2
oellicients
I
rim
+-
IIghts
2
(f)
620nm
520
%--
+--+i T,me I” dark.
ml”
-“-e--&”
-.--
30
Rod and cone contributions to peripheral colour vision
840
r
Fig. 3. Chromaticity shifts within the loveal WDW system of monochromatic lights between 5 (filled symbols) and 30 min (open symbols) of dark adaptation obtained at 2.5’ (a) and 7.5’ (b).
Alpem. 1971). The results suggest that the sensitivity of the rods and the amount of rhodopsin increase rapidly during the first branch of the long-term darkadaptation curve. In Experiment 2. however, the chromaticities remained invariant for several min during the cone-plateau period. With deep-red test light, for instance. the chromaticities were found to remain invariant for more than 15 min. The present results, therefore, suggest that the chromaticities may be measured independently of rod activity for a considerable part of the cone-plateau period. This suggestion is strongly confirmed by the results of Stabell and Stabell (1976). which show that the spectral luminosity function determined during the cone-plateau period at a number of retinal illuminations above the threshold level may coincide with the fovea1 luminosity curve in the 520-700 nm region of the spectrum. Thus. there appears to be no evidence of any contribution of rod activity to the extrafoveal luminosity function in this region. The differences between the fovea1 and extrafoveal luminosity functions obtained at shorter wavelengths could very well be ascribed to the absorption spectrum of the macular pigment. However. this latter factor does not provide an adequate basis for interpreting the chromaticity shift observed at the short-wave region. since any screening effect of pre-receptor pigmentation would be rendered ineffective through the use of monochromatic lights (cf. Wright. 1946). in conclusion then. the results of Experiments 2a and 2b suggest that the differences presented in Fig. 4 between the chromaticities measured at the fovea and at 2.5 and 7.5 during the cone-plateau period. reflect differences between fovea) and extrafoveal cone mechanisms
1103
The results of Fig. 4 show that both reds (620 and 640 nm) and greens (500-540 nm) change toward yellow. while violet (460 nm) changes toward blue when monochromatic stimuli are moved from the fovea toward 2.5’ or 7.5” temporally. Furthermore. the saturation of the short wavelengths becomes marked11 reduced. On the three-response theory these shifts in chromaticities with eccentricity may be interpreted in a number of ways. In general. three kinds of mechanism are in question (cf. Wooten and Wald. 1973). We might assume that (a) the spectral sensitivit,y.functions of the photopigments broaden or shift posltlon in the spectrum toward the periphery so as to overlap more widely, (b) the relative contributions of the three types of cone change with eccentricity, or (c) the response paths of the three colour systems tend to become fused toward the periphery. Presupposing that assumptions (a) or (b) are valid and that the shape of the photopic luminosity function reflects the spectral sensitivites and the weighted contributions of the three cone photopigments. the photopic luminosity function should change with retinal location. Contrary to this prediction. the spectral luminosity curve measured during the cone-plateau period was found to coincide with the fovea1 curve in the 520-700nm region (cf. Stabell and Stabell. 1976). suggesting that within this region there is no change in the spectral sensitivity and the weighted contributions of the cone photopigments. The chromaticity shift observed with monochromatic lights from the red and green regions of the spectrum thus probably involves other mechanisms than alternatives (a) and (b) and might more adequately be explained on alternative (c). The suggestion of Le
Fig. 4. Chromaticity shifts within the fovea1 WDW system of monochromatic lights between fovea and 2.5 la) and between fovea and 7.5 (b). The chromaticities of the peripheral test field (plain symbols) were obtained during the cone-plateau period.
1104
BJOR~ STABELLand
Gros Clark and Chacko t 1947) that the impulses from the “red” and “green” receptors in the periphery are conveyed to a common mass of geniculate cells and give rtse to a yellow sensation. might provide such a basis. In accordance with this suggestion. both reds and greens changed toward yellow when the stimulus field was moved from the fovea to 3.5 or 7.5 temporally. The well-known relatively higher short-wave sensitivity of the extrafoveal luminosity function. as compared with the fovea1 function. has previously partly been accounted for by a relatively greater contribution of “blue” receptors in the peripheral retina (cf. Weale. 1953: Wooten, Fuld and Spillmann. 1975). However. this factor would not easily account for the marked desaturation observed peripherally in the short-wave region. A more adequate explanation of the reduced saturation might be based on neural mechanisms. suggesting that relatively more of the peripheral cone impulses are conveyed to nonopponent cells (cf. De Valois. Abramov and Jacobs. 1966). It would be mere speculation, however. to press the analysis in any detail. To summarize. the present results support the suggestion of Wooten and Wald (1973) that neural rather than photochemical cone mechanisms may contribute to the changes in chromaticities observed when a test colour is moved from the fovea toward the periphery. In addition. the present results strongly suggest that also rods may contribute to this change. Thus. the chromaticities remained invariant during the cone plateau of the dark-adaptation curve and generally started to change just prior to the cone-rod break. Hence. it appears that the changes in saturation and hue observed during the second step of the dark-adaptation curve may be ascribed to the effect of rod intrusion (cf. StabelI and Stabell. 1975). In addition to a marked reduction in saturation, the added rod activity seems to be accompanied with a change in hue toward blue in the short wavelength region and toward yellow in the long and medium regions (cf. Fig. 3). The desaturation effect of rods is well-known and has been explained on the basis of the Duplicity Theory of vision, which assumes that rods contribute an achromatic quality to the visual response. The shift in hue associated with the rod intrusion appears more difficult to explain but might be accounted for on electrophysiological evidence, which indicates that both rods and cones feed into spectrally opponent cells (Wiesel and Hubel. 1966). On this basis it has been suggested that rods and cones may interact to produce sensations of hue (cf. Stabell and Stabell. 1973). Thus, in opposition to the assumption that the rod activity contributes an achromatic colour quality to the visual sensation. it has been suggested that the
LLF
STABELL
chromatic response of the rods may chance as d l‘utiction of selective chromatic stimulation ;I’ the cones Accordingly. the present results Indicate that the added rod activity may contribute a blue component in the short-wavelength region of the spectrum and a yellow component m the long and medium regrons. REFERE\CES Alpern M. (1971) Rhodopsm kmettcs m the human e\c. J. PhJsiol.. Lomf. 217. 447-471 De Valois R. L.. Abramo\ I. and Jacobs G. H 11Qbh1 Analysis of response patterns of LG1 cells. J opr. Sot Am. 56. Q66Q77. Gilbert M. (1950) Colour petceptlon m parafoveal btsmn. Proc. phys. See. B.
63. d&89.’
Hess C. (1889) Uber den Farbensinn beI indnectem Sehen. Arch. bphthal. 35. l-62. Le Gros Clark W. E. and Chacko L. 119471 A possible central mechanism for colour vision. A’ururu. Loud. 160. 123-124. Lie 1. (1963) Dark adaptation and the photochromatlc mterval. Documetlta ophrh. 17. 41 I-510. Moreland J. D. and Cruz A. 11959) Colour perception with the peripheral retina. Optica Acra 6. 1IT-15I. Purkinje J. (1825) Beohaclmmgen und krsuche xr P/I ~‘.swlogic der Sinne. Neue Beirriigr :ur Kennrniss drs Srhnzs itr Su&cricer Hinsicht. 6d. 2. Reimer, Berlin. Rushton W. A. H. and Powell D. S. (1972) The rhodopsm content and the visual threshold of human rods. t’ision Res. 12. 1073-1081. Stabell B. and Stabell U. (1973) Chromatic rod vision--IX: a theoretical survey. Visrort Rea. 13. 449-455. Stabell U. (1967) Rods as color receptors in photopic vision Scund. J. Psyzhol. 8. 139- 144. Stabell U. and Stabell B. (1975) The effect of rod activity on colour matching functions. Vision Res. IS. I I 19-l 123. Stabell U. and Stabell B. (1976) Absence of rod activity from peripheral vision. Msiorr Rrs. 16 (in press). Wald G. (1960) Analysis of retinal function by a two-filter method. J. opr. Sot. Am. 50. 633-641. Wald C. (1961) Participation of rods and cones in the visual responses. (Reply to the comments of C. S. Bridgman). J. opr. Sot. Am. 51. Z41-343. Weale R. A. (1953) Spectral sensitivity and wavelength discrimination of the peripheral retina. J. Physiol.. Loud. 119. 170-190. Wiesel T. N. and Hubel D. H. (1966) Spatial and chromatic interactions in the lateral geniculate body of the rhesus monkey. J. Nruroph.rsio/. 29. 1I 15-I 156. Wooten B. R.. Fuld K. and Spillmann L. (1975) Photopic spectral sensitivity of the peripheral retina. J. opt. Sot. Am. 65. 334342. Wooten B. R. and Wald G. (1973) Color-vision mechanisms in the peripheral retinas of normal and dichromatic observers. J. gen. Physiol. 61, 15-145. Wright W. D. (1946) Researches on .Vormnl and D@rice Co/our
Vision.
Henry Kimpton. London.
Wyszecki G. and Stiles W. S. (1967) Color Science. Concepts
and
Merhods,
Wiley, New York.
Quanrirariw
Dnra
artd
Formulas.