Large-field trichromacy in protanopes and deuteranopes* Vivianne C. Smith and Joel Pokorny Eye Research Laboratories, The University of Chicago, 950 East 59th Street, Chicago, Illinois 60637 (Received 21 June 1976) Protanopes and deuteranopes do not accept the classical dichromatic matches when field size extends to 8' visual angle. Their unique matches of spectral yellow to a mixture of red and green are then mediated by the photoreceptors of small-field dichromacy interacting with a photoreceptor with the spectral sensitivity of rhodopsin. Our data suggest that large-field trichromacy is a general feature of protanopia and deuteranopia.
Many protanopes and deuteranopes insist that they enjoy a unique percept for red light. Color theorists doubt this claim since protanopes and deuteranopes are dichromats: they can match all spectral colors using suitable mixtures of two primary colors. Protanopes and deuteranopes are thought to see the spectrum as shades of blue and yellow separated by neutral grays in the spectral region corresponding to 500 nm. 1-3 In an anomaloscope with 545 nm (green) and 670 nm (red)
primaries, protanopes and deuteranopes behave as monochromats; they match either the 545 nm primary, or the 670 nm primary, or any mixture of these primaries to spectral yellow. In fact, dichromats will match spectral green to spectral red. The basic dichromatic performance of protanopes and deuteranopes in color matching has been confirmed for foveal fields in many laboratories.
1 4,5 Thus, when a dichromat
claims to see "red, " he is greeted with skepticism and treated to explanations designed to annul his "red" perception. 6 However, occasional studies suggest that by enlarging the field to include parafoveal regions, dichromats may encounter an added dimension of color perception.
6-11
Nagel7 noted that although he was a deuter-
anope with a field of less than 20 extent, he became trichromatic
with a field of 100. Nagel 8 confirmed
these
findings on 30 other dichromats. Based on the appearance of red-green color-mixture fields. Nagel9 concluded that he was a deuteranomalous trichromat with a large field. His suggestion, however, was not based on a study of metameric color matches. Jaeger and Krokerl° found that two of seven deuteranopes
and
six of nine protanopes would not accept a neutral point match for a 220 field. They called these observers incomplete dichromats
and followed Nagel 10 in suggesting
that their color vision was that of the corresponding anomalous trichromat.
1
Boynton and Scheibner
6
and
Scheibner and Boynton" found that some of their dichromats had difficulty setting a neutral point match with a 30field. These data, taken with some color naming data, led them to the hypothesis that such observers had residual red-green discrimination mediated by normal cone photopigments; but they mention rod intrusion as an alternative. Boynton and Scheibner6 suggested that the ability to make red-green discriminations might be a general feature of dichromacy. Thus the studies 6 -11 show little agreement on the universality
of the phenomenon,
nor do they indicate how
such an added dimension of discrimination might be mediated. Further, there have been no studies of the large-field color matches made by dichromats. In this 213
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
paper, we examine dichromatic coefficients and redgreen matches made by protanopes and deuteranopes with field sizes of 10, 20, 40, and 8° to examine the dif-
ferent hypotheses 6 "11 and to see if the matches reveal any basis for the dichromat's claim of a distinct chromatic sensory perception of long-wavelength stimuli.
METHODS
Equipment. A Moreland Universal Anomaloscope'2 was used. Modifications of the instrument have been described previously.13 The left half-field was filled by a single nearly monochromatic beam (test field), the right half-field by an additive mixture of two nearly monochromatic beams, the primaries, P1 and P2 (mixture field). The interference filters used for this study were from various manufacturers. Wavelength of peak transmission and half-height bandwidth were measured in the anomaloscope with a calibrated laboratory-constructed photoelectric spectroradiometer. Half-height bandwidths ranged from 7 to 14 nm. Field stops provided circular
fields subtending 10, 20, 40, and 8°. An
additional field stop provided an 80 circular aperture with the central 40 blackened out. Procedure. The anomaloscope was used in a manner essentially identical to that of the Nagel anomaloscope. 14 The spectral test field was variable in luminance; the primary mixture field had variable proportions of P1 and P 2 . The field was freely viewed; each
observer used his preferred eye. Room illumination was dimmed, and the observers adapted to the field illumination. Observers were cautioned not to stare continuously at the field but rather to use a glance technique in which they looked away every 10-12 s.
When
the 80 annular stop was used, observers were instructed to fixate the center of the black circle. Observers. We advertised locally for color-defective observers. Observers were accepted as dichromats if on the Nagel anomaloscope they matched in hue 670 to 589 nm and 545 to 589 nm, and also if they made
a neutral point match for the 20field between a 7000 K white and a mixture of 450 and 650 nm.
Eighteen of the
individuals responding to our advertisement were classified as dichromats by this method. EXPERIMENT 1: ANOMALOSCOPE SCREENING WITH VARYING FIELD SIZE
Our first attention was directed to the discrepancy between Nagel8 and Jaeger
and Krokerl 0 as to how many
dichromats would refuse a dichromatic large-field Copyright O 1977 by the Optical Society of America
213
TABLE I.
Rayleigh matches accepted
(V) or rejected
Failure to accept dichromatic matches, however, miilghL be explained by postulating increased sensitivity
(x) by
douteranopos and protanopes as a fulnctiollOf field size. 589-670
1°
7
.
mn
589 = 545 nm
Field size 2' 4'
to saturation
Field size
1'
2'
4'
Deuteranope DK MM JK KK DR MB LB JD KG CM TR
450 and 650 nm spectral
x x x x x x x x x
ing a high proportion
size, 15 failures
x x x
x x
of 650 nm primary.
The propor-
to accept a match of 589 to 670 nm
x x x
EXPERIMENT 2: DICHROMATIC COEFFICIENTS AS A FUNCTION OF FIELD SIZE
The purpose of the first experiment was to
of 545 and 670 nm in
the mixture field and a 589 nm light in the test field. The field luminance was 5 candles/M 2 . For each circular stop, 11 deuteranopes and 7 protanopes were asked to match (i) the 545 nm primary, and (ii) the 670 nm primary to the 589 nm light by adjusting the luminance of the 589 nm test field. In addition, two protanopes, SY and CB, and two deuteranopes, DK and MM, were
asked to make the matches using the 80 annular stop. Results and discussion. All observers made both matches with the 1°field (Table I). As the field size increased, more observers reported that they did not have an exact color match. They were more likely to refuse the 589-670 nm condition than the 589-545 nm. With the 80 field, only two 589-545 nm
matches were accepted. The observers commented spontaneously that the fields appeared increasingly different in color as field size increased. Since all eighteen dichromats were unable to make 8° matches, the results show that this is a general phenomenon of dichromatic color vision, as postulated by Nagel. 8 The annular field stop was used to evaluate the possibility that retinal inhomogeneity was the source of the match refusals. With the annular field stop, protanope CB and deuteranope MM accepted the 589-545 match, but not the 589-670 match. Protanope SY and deuteranope DK rejected both matches. These data suggest that retinal inhomogeneity between foveal and parafoveal regions was not the cause of match rejection. 214
show that test wave-
To evaluate this hypothesis, we examined the dichromatic coefficients on those observers who were able to return for further testing,
field size was increased to 8'.
condition.
primaries
x x x
check how many of the 18 dichromats would continue to accept a match of 589 to 670 nm or 589 to 545 nm as the We used primaries
An in-
might be attributed to the dichromat's ability to discriminate a very small saturation difference (less than 1%)between the two lights when large fields are used.
x
Procedure.
lights.
tion is near 0. 98 at 540 nm, approaching unity at 650 nm. These data suggest a very slowly increasing saturation of long-wavelength lights. Since the precision of color matching improves with an increase in field
Protanopes
match.
of the spectral
lengths above 540 nm are matched to mixtures contain-
x
CB SY DP DS JM DV MG
differences
terpretation of dichromatic color-mixture data is that a wavelength which matches a mixture of two spectral primaries is less saturated in appearance than either spectral primary itself. Color matching data4 using
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
The purpose of the second experiment was to determine whether the dichromats would accept a dichromatic match using 450 and 650 nm primaries as the field size was increased to 80. Since the procedure took 3-4 h, only a few of the original 18 observers were able to participate. Procedure.
We used primaries
of 450 nm (blue) and
650 nm (red) in the mixture field and a series of four wavelengths,
589, 545, 523, and 494 nm in the test
field. The field luminance varied between 5 and 15 candles/m
2
, depending upon wavelength and class of ob-
server. Six of the deuteranopes and four of the protanopes were studied. For each test wavelength and field stop, the observer was asked to find an exact color match by adjusting the ratio of blue to red in the mixture field and the luminance of the test field, In addition, protanopes SY and CB and deuteranopes
DK and
MM were asked to match 494 nm using the 8° annular field stop. Results and discussion.
All six deuteranopes could
match 589, 545, 523, and 494 nm to mixtures of 450 and 650 nm when 1° and 20 fields were used (Table II).
Some accepted the matches with a 40 field. None would make such matches with the 80 field, Three protanopes could match 589, 545, and 523 nm to mixtures of 450 and 650 nm with 10 and 2° fields
(Table II). Two protanopes rejected the 494 nm match with the 20field, although both had accepted the match on their first visit to the laboratory. One other, SY, was somewhat fussy about the matches even with a 1V field, This protanope also had made a full set of dichromatic matches with the 2° field on an early visit. The field-size data of Table II, collected six months later, reflect many hours of practice. Protanope DP accepted matches to test wavelengths
545 and 523 nm
with the 40field and could match the 523 nm test with V. C. Smith and J. Pokorny
214
TABLE II. Dichromatic matches accepted or rejected by deuteranopes and protanopes as a function of field size. X 1 589 2 545 3 523 4 494
nm=Ai nm=A 2 nm=A 3 nm=A 4
P1 (650 (650 (650 (650
+
nm) + B1 nm)+B2 un) +B3 nm)+B 4
EXPERIMENT 3: RED-GREEN COLOR MATCHES FOR VARIOUS TEST WAVELENGTHS
(450 nm) (450 nm) (450 nm) (450 nm)
Field size 2° 4°
1° 1, 2, 3, 4a
P2
1 ,2
, 3, 4a
8'
1, 2, 3,4
1, 2, 3
x V
xxVTV
x x x x x x
VVVx
xVVx
xxV x
VXXX
X
X
VVVx V
x x
x x
, 4a
Deuteranopes DK MM KK JK DR MB
'I
Protanopes
V
I I
V VT V1
DP SY CB DS
I
V VT?? I/
x
Match No match
x
Unstable match aSingle entry means all four test wavelengths were accepted or rejected. Multiple checks are referred to numeral above. the 8° field.
The four dichromats tested with the 8° annular field refused the 494 nm match. Thus our results cannot be attributed to the Maxwell spot or to other retinal inhomogeneity that normal trichromats observe in certain large-field conditions. 15 The Maxwell spot subtends an ill-defined ellipse with major axis horizontal, extending 10or 20. The spot follows fixation and is best observed by switching rapidly from one colorimetric field to another. For normal trichromats, the phenomenon does not interfere with setting a large-field color match. 1 5 With the 8° field, two dichromats spontaneously reported observing a area of color difference, a "red" spot that moved with fixation. They were instructed, as we instruct normal trichromats, to ignore the spot and were able to do so. Many of the dichromats in Jaeger and Kroker'sl° study accepted a neutral point match using a 220field. We do not confirm this finding and presume the discrepancy reflects procedural differences: we used spectral primaries, while Jaeger and Kroker used colored papers on a rotating color wheel. We do agree with Jaeger and Kroker' 0 in finding, for the intermediate field sizes, that more protanopes than deuteranopes will reject a neutral point match. The failure to obtain a large-field match of 589 to 450 and 650 nm shows that the dichromats were not relying on saturation differences to discriminate 589 and
650 nm. If dichromats refuse large-field dichromatic matches, some visual photopigment must be exerting an independent effect. To determine whether this pho215
topigment is a normal or an anomalous cone photopigment or a rod photopigment, we examined a series of red-green color matches.
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
Rods are known to be active in large-field matches of normal trichromats. Experiment 3 was designed to ask whether large-field dichromatic matches agreed with matches of normal trichromats or anomalous trichromats, or whether they showed evidence of rod mediation. The results may clarify instances6 ' 9 11 where dichromats have shown some trichromatic activity with large test fields. Procedure. We used primaries of 545 nm (green) and 670 nm (red) in the mixture field and a series of six wavelengths, in the test field.
610, 596, 589, 580, 570, and 555 nm The field luminance was about 5 can-
dles/m 2 . The six deuteranopes and four protanopes of experiment 2 were observers. First, using the 20field for each test wavelength, the observers were asked to adjust the test luminance to give the best color match for nine red/green ratios covering the full range of the red-green dial, Then, using the 8° field for each test wavelength, a match range was established with the experimenter setting the red/green mixture. The match range is the set of primary mixtures for which the observer can obtain a color match by adjustment of the test wavelength luminance. Results and discussion. The results for the 8° field are shown in Table III as the proportion of 670 nm light in the mixture at the extremes of the match range, together with the proportion of 670 nm light for the match for a normal trichromat, a protanomal, and a deuteranomal. Except for the 555 nm test wavelength (and a few matches for two deuteranopes), the deuteranopes' matches require more red primary than the deuteranomalous match. The matches for protanopes occur near the protanomalous match, but consistently contain a greater fraction of R primary than protanomalous matches. Previous reports of failure of large-field dichromatic matches did not include a study of metamers. The discrepancy between our data and those of Nagel might be attributed to one of two possibilities: Nagel may have been an extreme anomalous trichromat; alternatively, if indeed a dichromat, he would not be accurate in color naming. We have noted that color names of dichromats with large fields are dependent on intensity and contrast in the field. The dichromats show heightened color contrast similar to that found in small fields 6 "'1or when anomalous trichromats make color judgments. ' Our data suggest that the ability to make unique redgreen matches is a general feature of protanopia and deuteranopia which occurs reliably with fields of 8° or more. Ruddock16 showed that dichromats could make dichromatic matches with two 1045' fields, one the mixture field, fixated foveally and the other the spectral field at 6° parafoveal. The protanopes differed in their parafoveal matches. Ruddock16 speculated that differing rod inputs to luminance and/or chromatic channels V. C. Smith and J. Pokorny
215
TABLE III. Proportion of red in red (670 nm)-green (545 nm) mixtures matched to various test wavelengths. 610
596
589
580
570
555
DR KK JK MM DK MB
0.957-0.965
0.837-0.880 0.822-0.880 0.700-0.907 0.747-0.870 0.867-0.880 0.627-0.747
0.735-0.810 0.700-0.768 0.675-0.768 0.515-0.768 0.400-0.630 0.307-0.400
0.527-0.710 0.350-0.697 0.543-0.655 0.308-0.722 0.180-0.435
0-0.565 0-0.400
0.950-0.957
0.907-0.920 0.850-0.917 0.880-0.907 0.867-0.907 0.880-0.960 0.867-0.907
0-0.425
0.10-0.437 0-0.372
Prediction
0.966
0.919
0.869
0.788
0.650
0.195
CB SY DP DS
0.985 0.985-0.990 0.990 0.972-0.990
0.965-0.972 0.972-0.985
0.925-0.950
0.907-0.960 0.985
0.930-0.965 0.957-0.965 0.930-0.965 0.957-0.972
0.920-0.965 0.957-0.972
0.867-0.917 0.867-0.950 0.100-0.930 0.880-0.950
0-0.880 0-0.747 0-0.880 0-0.880
Prediction
0.994
0.981
0.960
0.936
0.912
0.650
Normal
0.795 0.950 0.400
0.655 0.917 0.255
0.543 0.907 0.180
0.400 0.837 0.140
0.293 0.732 0.085
0.112 0.468 0.045
Deuteranopes
0.950-0.965 0.950-0.985 0.950-0.965
0.957-0.972
0.10-0.352 0-0.425
Protanopes
PA DA
could explain his data. MacLeod and Lennie1 7 also report that dichromats remain dichromatic for fields viewed in the parafovea. Small parafoveal fields may not give reliable trichromatic behavior. Two deuteranopes (MB, DK) differed from each other and the others
in their matches to 570, 580, and 589
nm. We cannot explain these matches but suggest that they reflect procedural rather than biological variance. The fact that these observers exhibit narrow matching ranges in agreement with the other deuteranopes at 596 and 610 nm suggests that the deviant matches are not of visual photopigment origin.
Analysis of red-green matches. For small fields,
match range. Near the primaries, some of the observers did not report a full color match, but only a brightness match. We consider dichromacy as a reduction system of normal trichromacy, assuming that protanopes have the normal MWSand deuteranopes the normal LWSvisual photopigment. When protanopic and deuteranopic luminance matches are superimposed, the intersection represents the unique red-green mixture where both the MWS and LWS photopigments
With a 10 or a 20 field, protanopes
behave as monochromats, mixture
18 matching
and deuteranopes
each red-green
with a suitable amount of test-wavelength
luminance. When the test-wavelength luminance is plotted as a function of red-green mixture, the luminance match to a given mixture represents the total visual photopigment response to the proportion of red primary plus the proportion of green primary in the mixture. These matches are shown for the 2° field (Figs. 1-6) for the six deuteranopes (right panel) and four protanopes (left panel) for test wavelengths 555 to 610 nm. The vertical bars show the interobserver 216
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
have an equivalent
quantal catch for the test and the mixture fields. The 2° match for a group of normal trichromats should occur at the intersection of protanopic and deuteranopic
the two halves of the colorimetric field appear identical when the quantal catches for each half of the field are equal for each active visual photopigment. A match which satisfies this condition is called a quantal match. In the anomaloscope, with both primaries chosen above 540 nm, the short-wavelength-sensitive visual photopigment (SWS)does not participate in the matches. In normal trichromats, red-green matches with a 10or a 20field are mediated by the middle- and long-wavelength sensitive visual photopigments (MWSand LWS). For each test wavelength, there is a unique matching red-green ratio which satisfies the quantal match.
0.720-0.950
(a)
40
,
40
(I
b
:E
30
30 0
20
20
0
10
aj
z Z
o
0 D
O
O
0.5
1.0
FRACTION RED
0
-- - --
0.5 FRACTION
1.R RED
FIG. 1. Anomaloscope matches made by protanopes rpanel (a)] and deuteranopes [panel (b)] to a 555 rm test field. The solid
line with vertical bars shows color and luminance matches made by four protanopes and six deuteranopes using the 20 field. The dashed line shows predicted scotopic luminance matches. The solid bar is the unique matching range for protanope CB [panel (a)] and deuteranope DR [panel (b)] using the 8' field. The arrows under the letters N, PA, and DA point to the matches made by a normal, a protanomalous, and a deuterano-
malous trichromat, respectively. V. C. Smith and J. Pokorny
216
l1
l1
(b)
40
(a)
40
(a)
40
7'2
40
-. (b).
a,
30
30
o
30
30 )-
1
0 1
20
20
a. 20
20
0 Iu
.11 10
10
.1
--.
0
1
Z
1o
M-
2 0l
0
0
0.5
1.0
FRACTION
0
DA 4
,
4
0
, I
0.5 FRACTION
RED
01 -,
I
1.0
0
RED
0.5 FRACTION RED
1.0
0
.I
w
0.5
FRACTION
1.0
RED
FIG. 2. As for Fig. 1, with a tes;t wavelength of 570 nm.
FIG. 4.
luminance matches provided all three groups of ob-
dashed line. Matches for the deuteranope DR and the protanope CB occur at or near the intersections of the theoretical rod luminance matches with the 20 luminance matches of the dichromats. The intersection
servers are similar in their inert pigment characteristics. Given the small number of observers, this expectation is well realized. Protanomalous and deuteranomalous trichromats do not accept the normal match: their visual photopigments differ from those of normals. Their unique matches represent intersection points of their underlying photopigments. Mid-match points. The plots in Figs. 1-6 present a graphical representation rather than the usual sets of linear equations. The unique matches made by a deu-
As for Fig. 1, with a test wavelength of 589 nm.
points are included in Table III to allow comparison
of
the predictions with match positions of the other observers. The large-field matches of our dichromats are quantal matches, mediated by the photoreceptors of small-field dichromacy plus a photoreceptor with the spectral sensitivity of rhodopsin.
normals, anomalous trichromats, or dichromats.
Match width. The match widths are strongly correlated with the difference in slope between the rod luminance function and the dichromatic luminance function. The match widths are narrowest at 610 nm where the rod function is steepest, and broadest at 555 nm where rod function is shallowest. In general, the protanopic match widths are narrower than deuteranopic match widths. This trend reverses at 555 nm where the rod and 2° protanopic luminance matches almost coincide over their range. The characteristic match width profile strengthens our evidence that the matches are mediated by the dichromatic cone photopigment and a visual photopigment whose spectral sensitivity is that of rhodopsin.
To evaluate the possibility that the matches were rodmediated, we calculated the luminance matches that
EXPERIMENT 4: MATCHES AS A FUNCTION OF INTENSITY
teranope,
CB, with the 8° field
DR, and a protanope,
are shown as solid bars. The luminance settings indicate that the 80 matches were also matches for the 20 field. For wavelengths 570-610, matches for deuteranope DR contain a higher proportion
of red primary
than those of normal or deuteranomalous trichromats; matches for protanope CB contain a higher proportion of red than those of normal or protanomalous trichromats. These matches are mediated by at least one visual photopigment which does not act in the 20 matches of
would be made if rods alone mediated the response.
These theoretical matches based upon the CIE scotopic luminosity function,
40
Vx are shown in Figs.
(a)
-
40 -
1-6 as a
When rods participate in large-field matches made by normal trichromats, quantal matches are rarely made.
(b)
a,I N
9
30)-
30
1
o 20
20 0
0
z
10
10
4
4 :E X.
A I
j . 0
0
0.5 FRACTION
1.0 RED
DA
O[
4l
)
0
b
!- I. 11.0
N
0.5 FRACTION
FIG. 3. As for Fig. 1, with a test wavelength 217
I
RED
of 580 nm.
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
FRACTION
RED
FRACTION
RED
FIG. 5. As for Fig. 1, with a test wavelength of 596 nm. V. C. Smith and J. Pokorny
217
570 NM (a)
40
40
589
610 NM
~P-OS
I
030
NM
Mb)
0.I
30
g
l*
0.01 0.001
l -20
20
0
10
SY
----
0.I Z
0.0I X
I
l
0.001
-J lO
-
0
,
1.0
0.5
0
0.5
FRACTION RED
FIG. 6.
-
PA -DA-
1.0
-
0.I
FRACTION RED
-J
As for Fig. 1, with a test wavelength of 610 nm.
w
D-MM
0.01 0.001 l
Clarke1 9 uses the term tetrachromacy to allow for the activity of rods in normal color matching. There are three important properties of normal trichromacy which are not fulfilled in tetrachromacy. Normal trichromacy has the properties that (a) three primaries are sufficient for spectral color matching; (b) the relationship between the primaries is invariant, that is, the matches are unique; (c) they hold over a range of 1500 td. These are all properties of quantal matches. With an 80field, three primaries are sufficient and a unique match is made. However, the matches change with field luminance. If four primaries are allowed, the matches are not unique. Trezona20 devised a fourprimary method of systematically iterating between matches made at a scotopic and mesopic luminance until a unique match was achieved. This unique match is a quantal match for each of the four visual photopigments, and is therefore intensity invariant. Experiment 4 was designed to test whether the dichromatic matches
-
0.1
I
0.01 0.001
-
I- .*0
0.5
*
~
-
1.0 0
0.5
WJ
1.0 0
0.5
1.0
FRACTION RED
FIG. 7. Anoma loscope matches as a function of relative flux made by two protanopes (upper rows) and two deuteranopes (lower rows) to test wavelengths of 570 nm (column 1), 589 nm (column 2), and 610 nm (column 3) using the 8' test field. Lu-
minance at unit flux for the various conditions and classes of observer
may be obtained by reference
to Figs.
1, 3, and 5.
Procedure. We used primaries of 545 nm (green) and 670 nm (red) in the mixture field and a series of
each luminance and test wavelength in Fig. 7. The sets of matches show similar trends. Matches are not affected by a tenfold reduction of intensity. The matches tend to widen somewhat with a one-hundred-fold reduction and show greater widening with the one-thousandfold reduction of luminance. There is no shift of the match midpoint to a higher or lower fraction R. The 5 cd/m 2 match is always accepted. These data show an intensity invariance of the quantal matches with a
three wavelengths,
1000 to 1 change in luminance.
would hold with reduction in field luminance.
610, 589, and 570 nm in the test
field. The highest field luminance was about 5 candles/ m2 . The luminance was reduced in one-log-unit steps for a range of 3 log units. Light shielding was provided for the lowest intensities. The observers adapted to the field for up to 30 min. in the low ambient illumination of the darkened room. Two deuteranopes and two protanopes were observers. For each test wavelength and test luminance, a match range was established by the experimenter using the 80 field. Results and discussion.
The matches are plotted for
THEORETICAL IMPLICATIONS
Implicationsfor normal trichromacy. The protanopic and deuteranopic 80 matches may be compared with the 2° and the 8° matches made by normal trichromats (Table IV). The large-field trichromacy we observe in protanopes and deuteranopes shows the properties of a unique match and intensity invariance. The matches therefore show properties more similar to those of normal small-field trichromacy than of normal large-field
TABLE IV. Comparison of trichromacy and tetrachromacy in normal trichromats with large-field trichromacy in protanopes and deuteranopes. Query
Small-field Normal trichromats
Large-field Normal trichromats
Large-field trichromacy in protanopes and deuteranopes
How many primaries?
3
3
4
3
Are unique matches made?
Yes
Yes
No'
Yes
Are matches intensity-invariant?
Yes
No
No'
Yes
`An exception is the technique proposed by Trezona. 20 218
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
V. C. Smith and J. Pokorny
218
tetrachromacy.
field trichromacy.
The comparison points to a distinction between largefield matches of normals and dichromats. Normal trichromacy is both pigment and channel limited. With small fields, normal trichromacy is pigment limited: there are three cone visual photopigments, and at least three processing channels. Intensity-invariant quantal matches are made. With large fields, normal trichromacy is not pigment limited; there are four (three cone and one rod) visual photopigments. However, trichromatic matches can be made: they are not intensity invariant, nor are they quantal. These data indicate that the rods do not have an independent channel; their signals are carried on the channels carrying cone achromatic and chromatic information. In this sense, we say that the data are channel limited. There are only three orthogonal channels for color mixture.
An alternative explanation of our data is that the dichromats have only two processing channels, but reject all but the quantal match because of differing temporal and spatial transients 2 6 generated by cones and rods along the border of the test and matching fields. An unsatisfactory aspect of this explanation is that it implies that only dichromats, and not normal trichromats, are able to use such transient information to make large-field quantal matches under the free-view conditions that we use. We would expect that the use of the 80 annular stop with steady fixation on the black center would reduce information due to temporal border transients. Our observers refused classical dichromatic matches when the annular field was used (experi-
Psychophysical study of rod-cone interactions show a variety of effects, reviewed by Jacobs, 22 including mutual facilitation and mutual inhibition. McCann and Benton2 3 report that rods interacting with LWS cones yield an array of colors similar in color appearance to photopic mixtures
of wavelengths 495 and 656 nm.
Stabell and Stabell24 note that rod color does not occupy a unique locus in the CIE diagram. Jacobs22 observes that the rod interactions reported by Stabell and Stabell suggest that rods interact in the opponent chromatic channels.
Thus, although rods do not have an inde-
pendent activity in color matching, they can modulate opponent color signals. A neurophysiological basis may be noted in data from macaque monkeys. 2 5 When rods enter the chromatic opponents, the rod response is that of the field center. The four receptive field types (red on-center, red offcenter, green on-center, green off-center) would yield two distinctly different types of opponent response. Thus a red- Vx on-center or a red- VAoff-center would give different opponents from a green- Vt on-center or a green-V' off-center. In psychophysics, the four types of receptive field are treated as one chromatic opponent. Both the psychophysics 2 3 ' 2 4 and the Wiesel and Hubel25 data make it clear that any explanation of
rod effects on color appearance requires a more complex approach than is presently available in color theory.
Theoreticalinterpretation of large-field dichromatic
ments 1 and 2). 27 We suspect that while border tran-
sients may enhance color information for dichromats, they are not the cause of the large-field color-matching behavior. A third possibility is that the trichromacy is not mediated by normal rods. Possibly, the dichromats have cones whose photopigment has the spectral response of rhodopsin as postulated for some of the monochromacies. 28-32 If the latter is true, then protanopia and deuteranopia should be regarded as additional forms of anomalous trichromacy. If mediated by normal rods, the term dichromacy may be regarded as a special case of reduced color vision elicited when the small fields selectively allow cone response.
Do dichromats have a uniquesensory perceptfor red? The impetus for our study was the fact that many dichromats insist that they have a unique sensory percept for "red. " Our study is one of color matches which does not tell us anything about the color perceptions of dichromats, However, the fact that protanopes and deuteranopes become trichromatic with large fields means that they do have both the photopigment and possibly the processing channel that would permit an added unique sensory percept. We suspect that their color world is weak compared with that of normal trichromats, However, they do have trichromatic color vision for large fields and it may well be that this trichromacy forms the basis for the perceptual claims made by dichromats. ACKNOWLEDGMENTS
data. The intensity-invariant quantal matches made by our dichromats suggest to us that three cone-rod channels are available to protanopes and deuteranopes. Small-field dichromatic matches are pigment limited; there are two cone visual photopigments and at least two processing channels. With large fields, there are
The major impetus for this research came from a deuteranope, Michael Breton, who insists that he enjoys a unique sensory percept for red. We thank our research subjects for their cooperation.
three (two cone and one rod) visual photopigments
*Supported in part by NIH NEI grants EY00901 and EY70652 to Pokorny, and EY00523 to F. W. Newell. The data from Ex-
and
we suggest that there are at least three channels. The third channel may reflect rod activity in a residual redgreen opponent system. Alternatively, two neural channels may be formed from a blue-yellow channel in a manner similar to that discussed for the red-green channel of normal trichromats. The important point is that the rod signal has independent recognition in dichromatic color opponent channels, allowing the large219
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
periments 1-3 were presented at the annual meeting of A. R. V. 0. in 1976 and those of Experiment 4 at the 1976 Annual Meeting of the Optical Society of America.
'J. H. Parsons, An Introduction to the Study of Colour Vision, 2nd ed. (Cambridge U. P., Cambridge, 1924). D. B. Judd, "Colour perceptions of deuteranopic
2
and protanop-
ic observers, "t J. Res. Natl. Bur. Stand. 41, 247-271 (1948). 3C.
H. Graham,
H. G. Sperling,
Y. Hsia, and A. H. Coulson,
V. C. Smith and J. Pokorny
219
"The determination of some visual functions of a unilateral color-blind subject: methods and results, " J. Psychol. 51, 3-32 (1961). F. H. G. Pitt, "Characteristics of Dichromatic Vision," Med. Res. Council Spec. Rep. No. 200 H.M.S.O. (1935). 5S. Hecht and S. Shlaer, "The color vision of dichromats. I. 4
Wavelength discrimination, brightness distribution, and color mixture, " J. Gen. Physiol. 20, 57-82 (1937). GSee R. M. Boynton and H. Scheibner,
"On the perception
of
red by 'red blind' observers, " Acta Chromatica 1, 205-220 (1967). We used such explanations with dichromatic obser7 8
vers. W. A. Nagel, "Dichromatische
fovea, trichromatische
peri-
2nd ed., (Williams and Wilkins, Baltimore, 1970), pp. 199259. 22 G. H. Jacobs, "Color vision, " Ann. Rev. Psychol. 27, 6389 (1976). 23 J. J. McCann and J. L. Benton, "Interaction of the long-
wave cones and rods to produce color sensations, " J. Opt. Soc. Am. 59, 103-107 (1969).
24U.Stabell and B. Stabell, "Chromatic rod-cone interaction," Vision Res. 14.1389-1392 (1974).
pherie, " Z. Psychol. u. Physiol. Sinnes 39, 93-101 (1905).
25
W. A. Nagel, "Neue Erfahrungen
actions in the lateral geniculate body of the rhesus monkey," J. Neurophysiol. 29, 1115-1156 (1966). 26Thesuggestion that spatial and temporal border transients
jiber das Farbensehen
der
Dichromaten auf grossen Felde, " Z. Sinnesphysiol. 41, 3199
New York, 1973), pp. 319-324.
20p, W. Trezona, "The tetraclhronatic colour match as a colorimetric technique, "1Vision Res. 13, 9-25 (1973). 21 G. S. Brindley, Physiology of the Retina and Visual Pathway,
337 (1907). W. A. Nagel, "Farbenstimmung nesphysiol. 44, 5-17 (1910).
beim Dichromaten,
" Z. Sin-
10
W. Jaeger and K. Kroker, "fber das verhalten der protanopen und deuteranopen bei grossen reizflachen, " Klin MBL. Augen-
heilk. 121, 445-449 (1952). "1H. M. 0. Scheibaer and R. M. Boynton, "Residual red-green discrimination in dichromats, " J. Opt. Soc. Am. 58, 11511158 (1968). 1 2J. D. Moreland and W. B. Young, "A new anomaloscope em-
T. N. Wiesel and D. H. Hubel, "Spatial and chromatic
inter-
could explain our data was first made by D. I. A. MacLeod
at the annual meeting of A.R.V.O. in 1976. Weattempted to evaluate the effects of border transients in
27
two additional ways.
We constructed
an 8° field stop with a
2° black bar. The 2° bar prevents an observer from making direct comparisons across the test-field border. With this test field, discrimination deteriorated. However, protanopes SY and CB and deuteranopes DK and MM continued to reject the neutral point match of 494 rm to a mixture of 450 and
ploying interference filters, " Mod. Prob. Ophthal. 13, 47-55
650 nm. Their red-green matches for wavelengths 570, 589,
(1974). J. Pokorny and V. C. Smith, "Effect of field size on red-
and 610 nm widened, similar
13
green color mixture equations, " J. Opt. Soc. Am. 66, 705708 (1976). 14 A. Linksz, An Essay on1Color Vision (Grune and Stratton, New York, 1964). 15 G. Wyszecki and W. S. Stiles, Color Science (Wiley, New York, 1967).
K. H. Ruddock, "Parafoveal colour vision responses of four dichromats, " Vision Res. 11, 143-156 (1971).
17
D. I. A. MacLeod and P. Lennie, "Red-green blindness fineqj to one eye, " Vision Res. 16, 691-702 (1976).
con-
M. Alpern, "What is it that confines in a world without color? Invest. Ophthal. 13, 648-674 (1974). F. J. J. Clarke, "Needs and Prospects
9
28
G. L. Walls and G. B. Heath, "Typical total colour blindness
reinterpreted, " Acta Ophthal. 32, 253-297 (1954). 29L. Sloan, "The photopic retinal receptors of the typical achromat, " Am. J. Ophthal. 46, 81-86 (1958). 30
M. Alpern,
H. F. Falls,
and G. B. Lee, "The enigma of
typical total monochromacy, " Am. J. Ophthal. 50, 326-341 (1960).
38 J. Pokorny, V. C. Smith, and R. Swartley, "Threshold measurements of spectral sensitivity in a blue monocone
monochromat, " Invest. Ophthal. 9, 807-813 (1970).
for a tetrachromatic
system of large field colorimetry, " in Colour 73, Proceedings of the 2nd Congress of the I. C. A, at York, 1973 (Wiley,
220
to those for one-thousand-fold
reduction in field luminance (Fig. 7). Our second approach was to defocus the field so that the border was invisible. Again, discrimination was reduced, but not to the same degree as with the 2° bar separating the test and mixture fields.
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
32
M. Alpern,
G. B. Lee, F. Maaseidvaag,
and S. S. Miller,
"Colour vision in blue-cone 'monochromacy,
"
J. Physiol.
(Lond.) 212, 211-233 (1971).
V. C. Smith and J. Pokorny
220
Many protanopes and deuteranopes insist that they enjoy a unique percept for red light. Color theorists doubt this claim since protanopes and deuteranopes are dichromats: they can match all spectral colors using suitable mixtures of two primary colors. Protanopes and deuteranopes are thought to see the spectrum as shades of blue and yellow separated by neutral grays in the spectral region corresponding to 500 nm. 1-3 In an anomaloscope with 545 nm (green) and 670 nm (red)
primaries, protanopes and deuteranopes behave as monochromats; they match either the 545 nm primary, or the 670 nm primary, or any mixture of these primaries to spectral yellow. In fact, dichromats will match spectral green to spectral red. The basic dichromatic performance of protanopes and deuteranopes in color matching has been confirmed for foveal fields in many laboratories.
1 4,5 Thus, when a dichromat
claims to see "red, " he is greeted with skepticism and treated to explanations designed to annul his "red" perception. 6 However, occasional studies suggest that by enlarging the field to include parafoveal regions, dichromats may encounter an added dimension of color perception.
6-11
Nagel7 noted that although he was a deuter-
anope with a field of less than 20 extent, he became trichromatic
with a field of 100. Nagel 8 confirmed
these
findings on 30 other dichromats. Based on the appearance of red-green color-mixture fields. Nagel9 concluded that he was a deuteranomalous trichromat with a large field. His suggestion, however, was not based on a study of metameric color matches. Jaeger and Krokerl° found that two of seven deuteranopes
and
six of nine protanopes would not accept a neutral point match for a 220 field. They called these observers incomplete dichromats
and followed Nagel 10 in suggesting
that their color vision was that of the corresponding anomalous trichromat.
1
Boynton and Scheibner
6
and
Scheibner and Boynton" found that some of their dichromats had difficulty setting a neutral point match with a 30field. These data, taken with some color naming data, led them to the hypothesis that such observers had residual red-green discrimination mediated by normal cone photopigments; but they mention rod intrusion as an alternative. Boynton and Scheibner6 suggested that the ability to make red-green discriminations might be a general feature of dichromacy. Thus the studies 6 -11 show little agreement on the universality
of the phenomenon,
nor do they indicate how
such an added dimension of discrimination might be mediated. Further, there have been no studies of the large-field color matches made by dichromats. In this 213
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
paper, we examine dichromatic coefficients and redgreen matches made by protanopes and deuteranopes with field sizes of 10, 20, 40, and 8° to examine the dif-
ferent hypotheses 6 "11 and to see if the matches reveal any basis for the dichromat's claim of a distinct chromatic sensory perception of long-wavelength stimuli.
METHODS
Equipment. A Moreland Universal Anomaloscope'2 was used. Modifications of the instrument have been described previously.13 The left half-field was filled by a single nearly monochromatic beam (test field), the right half-field by an additive mixture of two nearly monochromatic beams, the primaries, P1 and P2 (mixture field). The interference filters used for this study were from various manufacturers. Wavelength of peak transmission and half-height bandwidth were measured in the anomaloscope with a calibrated laboratory-constructed photoelectric spectroradiometer. Half-height bandwidths ranged from 7 to 14 nm. Field stops provided circular
fields subtending 10, 20, 40, and 8°. An
additional field stop provided an 80 circular aperture with the central 40 blackened out. Procedure. The anomaloscope was used in a manner essentially identical to that of the Nagel anomaloscope. 14 The spectral test field was variable in luminance; the primary mixture field had variable proportions of P1 and P 2 . The field was freely viewed; each
observer used his preferred eye. Room illumination was dimmed, and the observers adapted to the field illumination. Observers were cautioned not to stare continuously at the field but rather to use a glance technique in which they looked away every 10-12 s.
When
the 80 annular stop was used, observers were instructed to fixate the center of the black circle. Observers. We advertised locally for color-defective observers. Observers were accepted as dichromats if on the Nagel anomaloscope they matched in hue 670 to 589 nm and 545 to 589 nm, and also if they made
a neutral point match for the 20field between a 7000 K white and a mixture of 450 and 650 nm.
Eighteen of the
individuals responding to our advertisement were classified as dichromats by this method. EXPERIMENT 1: ANOMALOSCOPE SCREENING WITH VARYING FIELD SIZE
Our first attention was directed to the discrepancy between Nagel8 and Jaeger
and Krokerl 0 as to how many
dichromats would refuse a dichromatic large-field Copyright O 1977 by the Optical Society of America
213
TABLE I.
Rayleigh matches accepted
(V) or rejected
Failure to accept dichromatic matches, however, miilghL be explained by postulating increased sensitivity
(x) by
douteranopos and protanopes as a fulnctiollOf field size. 589-670
1°
7
.
mn
589 = 545 nm
Field size 2' 4'
to saturation
Field size
1'
2'
4'
Deuteranope DK MM JK KK DR MB LB JD KG CM TR
450 and 650 nm spectral
x x x x x x x x x
ing a high proportion
size, 15 failures
x x x
x x
of 650 nm primary.
The propor-
to accept a match of 589 to 670 nm
x x x
EXPERIMENT 2: DICHROMATIC COEFFICIENTS AS A FUNCTION OF FIELD SIZE
The purpose of the first experiment was to
of 545 and 670 nm in
the mixture field and a 589 nm light in the test field. The field luminance was 5 candles/M 2 . For each circular stop, 11 deuteranopes and 7 protanopes were asked to match (i) the 545 nm primary, and (ii) the 670 nm primary to the 589 nm light by adjusting the luminance of the 589 nm test field. In addition, two protanopes, SY and CB, and two deuteranopes, DK and MM, were
asked to make the matches using the 80 annular stop. Results and discussion. All observers made both matches with the 1°field (Table I). As the field size increased, more observers reported that they did not have an exact color match. They were more likely to refuse the 589-670 nm condition than the 589-545 nm. With the 80 field, only two 589-545 nm
matches were accepted. The observers commented spontaneously that the fields appeared increasingly different in color as field size increased. Since all eighteen dichromats were unable to make 8° matches, the results show that this is a general phenomenon of dichromatic color vision, as postulated by Nagel. 8 The annular field stop was used to evaluate the possibility that retinal inhomogeneity was the source of the match refusals. With the annular field stop, protanope CB and deuteranope MM accepted the 589-545 match, but not the 589-670 match. Protanope SY and deuteranope DK rejected both matches. These data suggest that retinal inhomogeneity between foveal and parafoveal regions was not the cause of match rejection. 214
show that test wave-
To evaluate this hypothesis, we examined the dichromatic coefficients on those observers who were able to return for further testing,
field size was increased to 8'.
condition.
primaries
x x x
check how many of the 18 dichromats would continue to accept a match of 589 to 670 nm or 589 to 545 nm as the We used primaries
An in-
might be attributed to the dichromat's ability to discriminate a very small saturation difference (less than 1%)between the two lights when large fields are used.
x
Procedure.
lights.
tion is near 0. 98 at 540 nm, approaching unity at 650 nm. These data suggest a very slowly increasing saturation of long-wavelength lights. Since the precision of color matching improves with an increase in field
Protanopes
match.
of the spectral
lengths above 540 nm are matched to mixtures contain-
x
CB SY DP DS JM DV MG
differences
terpretation of dichromatic color-mixture data is that a wavelength which matches a mixture of two spectral primaries is less saturated in appearance than either spectral primary itself. Color matching data4 using
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
The purpose of the second experiment was to determine whether the dichromats would accept a dichromatic match using 450 and 650 nm primaries as the field size was increased to 80. Since the procedure took 3-4 h, only a few of the original 18 observers were able to participate. Procedure.
We used primaries
of 450 nm (blue) and
650 nm (red) in the mixture field and a series of four wavelengths,
589, 545, 523, and 494 nm in the test
field. The field luminance varied between 5 and 15 candles/m
2
, depending upon wavelength and class of ob-
server. Six of the deuteranopes and four of the protanopes were studied. For each test wavelength and field stop, the observer was asked to find an exact color match by adjusting the ratio of blue to red in the mixture field and the luminance of the test field, In addition, protanopes SY and CB and deuteranopes
DK and
MM were asked to match 494 nm using the 8° annular field stop. Results and discussion.
All six deuteranopes could
match 589, 545, 523, and 494 nm to mixtures of 450 and 650 nm when 1° and 20 fields were used (Table II).
Some accepted the matches with a 40 field. None would make such matches with the 80 field, Three protanopes could match 589, 545, and 523 nm to mixtures of 450 and 650 nm with 10 and 2° fields
(Table II). Two protanopes rejected the 494 nm match with the 20field, although both had accepted the match on their first visit to the laboratory. One other, SY, was somewhat fussy about the matches even with a 1V field, This protanope also had made a full set of dichromatic matches with the 2° field on an early visit. The field-size data of Table II, collected six months later, reflect many hours of practice. Protanope DP accepted matches to test wavelengths
545 and 523 nm
with the 40field and could match the 523 nm test with V. C. Smith and J. Pokorny
214
TABLE II. Dichromatic matches accepted or rejected by deuteranopes and protanopes as a function of field size. X 1 589 2 545 3 523 4 494
nm=Ai nm=A 2 nm=A 3 nm=A 4
P1 (650 (650 (650 (650
+
nm) + B1 nm)+B2 un) +B3 nm)+B 4
EXPERIMENT 3: RED-GREEN COLOR MATCHES FOR VARIOUS TEST WAVELENGTHS
(450 nm) (450 nm) (450 nm) (450 nm)
Field size 2° 4°
1° 1, 2, 3, 4a
P2
1 ,2
, 3, 4a
8'
1, 2, 3,4
1, 2, 3
x V
xxVTV
x x x x x x
VVVx
xVVx
xxV x
VXXX
X
X
VVVx V
x x
x x
, 4a
Deuteranopes DK MM KK JK DR MB
'I
Protanopes
V
I I
V VT V1
DP SY CB DS
I
V VT?? I/
x
Match No match
x
Unstable match aSingle entry means all four test wavelengths were accepted or rejected. Multiple checks are referred to numeral above. the 8° field.
The four dichromats tested with the 8° annular field refused the 494 nm match. Thus our results cannot be attributed to the Maxwell spot or to other retinal inhomogeneity that normal trichromats observe in certain large-field conditions. 15 The Maxwell spot subtends an ill-defined ellipse with major axis horizontal, extending 10or 20. The spot follows fixation and is best observed by switching rapidly from one colorimetric field to another. For normal trichromats, the phenomenon does not interfere with setting a large-field color match. 1 5 With the 8° field, two dichromats spontaneously reported observing a area of color difference, a "red" spot that moved with fixation. They were instructed, as we instruct normal trichromats, to ignore the spot and were able to do so. Many of the dichromats in Jaeger and Kroker'sl° study accepted a neutral point match using a 220field. We do not confirm this finding and presume the discrepancy reflects procedural differences: we used spectral primaries, while Jaeger and Kroker used colored papers on a rotating color wheel. We do agree with Jaeger and Kroker' 0 in finding, for the intermediate field sizes, that more protanopes than deuteranopes will reject a neutral point match. The failure to obtain a large-field match of 589 to 450 and 650 nm shows that the dichromats were not relying on saturation differences to discriminate 589 and
650 nm. If dichromats refuse large-field dichromatic matches, some visual photopigment must be exerting an independent effect. To determine whether this pho215
topigment is a normal or an anomalous cone photopigment or a rod photopigment, we examined a series of red-green color matches.
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
Rods are known to be active in large-field matches of normal trichromats. Experiment 3 was designed to ask whether large-field dichromatic matches agreed with matches of normal trichromats or anomalous trichromats, or whether they showed evidence of rod mediation. The results may clarify instances6 ' 9 11 where dichromats have shown some trichromatic activity with large test fields. Procedure. We used primaries of 545 nm (green) and 670 nm (red) in the mixture field and a series of six wavelengths, in the test field.
610, 596, 589, 580, 570, and 555 nm The field luminance was about 5 can-
dles/m 2 . The six deuteranopes and four protanopes of experiment 2 were observers. First, using the 20field for each test wavelength, the observers were asked to adjust the test luminance to give the best color match for nine red/green ratios covering the full range of the red-green dial, Then, using the 8° field for each test wavelength, a match range was established with the experimenter setting the red/green mixture. The match range is the set of primary mixtures for which the observer can obtain a color match by adjustment of the test wavelength luminance. Results and discussion. The results for the 8° field are shown in Table III as the proportion of 670 nm light in the mixture at the extremes of the match range, together with the proportion of 670 nm light for the match for a normal trichromat, a protanomal, and a deuteranomal. Except for the 555 nm test wavelength (and a few matches for two deuteranopes), the deuteranopes' matches require more red primary than the deuteranomalous match. The matches for protanopes occur near the protanomalous match, but consistently contain a greater fraction of R primary than protanomalous matches. Previous reports of failure of large-field dichromatic matches did not include a study of metamers. The discrepancy between our data and those of Nagel might be attributed to one of two possibilities: Nagel may have been an extreme anomalous trichromat; alternatively, if indeed a dichromat, he would not be accurate in color naming. We have noted that color names of dichromats with large fields are dependent on intensity and contrast in the field. The dichromats show heightened color contrast similar to that found in small fields 6 "'1or when anomalous trichromats make color judgments. ' Our data suggest that the ability to make unique redgreen matches is a general feature of protanopia and deuteranopia which occurs reliably with fields of 8° or more. Ruddock16 showed that dichromats could make dichromatic matches with two 1045' fields, one the mixture field, fixated foveally and the other the spectral field at 6° parafoveal. The protanopes differed in their parafoveal matches. Ruddock16 speculated that differing rod inputs to luminance and/or chromatic channels V. C. Smith and J. Pokorny
215
TABLE III. Proportion of red in red (670 nm)-green (545 nm) mixtures matched to various test wavelengths. 610
596
589
580
570
555
DR KK JK MM DK MB
0.957-0.965
0.837-0.880 0.822-0.880 0.700-0.907 0.747-0.870 0.867-0.880 0.627-0.747
0.735-0.810 0.700-0.768 0.675-0.768 0.515-0.768 0.400-0.630 0.307-0.400
0.527-0.710 0.350-0.697 0.543-0.655 0.308-0.722 0.180-0.435
0-0.565 0-0.400
0.950-0.957
0.907-0.920 0.850-0.917 0.880-0.907 0.867-0.907 0.880-0.960 0.867-0.907
0-0.425
0.10-0.437 0-0.372
Prediction
0.966
0.919
0.869
0.788
0.650
0.195
CB SY DP DS
0.985 0.985-0.990 0.990 0.972-0.990
0.965-0.972 0.972-0.985
0.925-0.950
0.907-0.960 0.985
0.930-0.965 0.957-0.965 0.930-0.965 0.957-0.972
0.920-0.965 0.957-0.972
0.867-0.917 0.867-0.950 0.100-0.930 0.880-0.950
0-0.880 0-0.747 0-0.880 0-0.880
Prediction
0.994
0.981
0.960
0.936
0.912
0.650
Normal
0.795 0.950 0.400
0.655 0.917 0.255
0.543 0.907 0.180
0.400 0.837 0.140
0.293 0.732 0.085
0.112 0.468 0.045
Deuteranopes
0.950-0.965 0.950-0.985 0.950-0.965
0.957-0.972
0.10-0.352 0-0.425
Protanopes
PA DA
could explain his data. MacLeod and Lennie1 7 also report that dichromats remain dichromatic for fields viewed in the parafovea. Small parafoveal fields may not give reliable trichromatic behavior. Two deuteranopes (MB, DK) differed from each other and the others
in their matches to 570, 580, and 589
nm. We cannot explain these matches but suggest that they reflect procedural rather than biological variance. The fact that these observers exhibit narrow matching ranges in agreement with the other deuteranopes at 596 and 610 nm suggests that the deviant matches are not of visual photopigment origin.
Analysis of red-green matches. For small fields,
match range. Near the primaries, some of the observers did not report a full color match, but only a brightness match. We consider dichromacy as a reduction system of normal trichromacy, assuming that protanopes have the normal MWSand deuteranopes the normal LWSvisual photopigment. When protanopic and deuteranopic luminance matches are superimposed, the intersection represents the unique red-green mixture where both the MWS and LWS photopigments
With a 10 or a 20 field, protanopes
behave as monochromats, mixture
18 matching
and deuteranopes
each red-green
with a suitable amount of test-wavelength
luminance. When the test-wavelength luminance is plotted as a function of red-green mixture, the luminance match to a given mixture represents the total visual photopigment response to the proportion of red primary plus the proportion of green primary in the mixture. These matches are shown for the 2° field (Figs. 1-6) for the six deuteranopes (right panel) and four protanopes (left panel) for test wavelengths 555 to 610 nm. The vertical bars show the interobserver 216
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
have an equivalent
quantal catch for the test and the mixture fields. The 2° match for a group of normal trichromats should occur at the intersection of protanopic and deuteranopic
the two halves of the colorimetric field appear identical when the quantal catches for each half of the field are equal for each active visual photopigment. A match which satisfies this condition is called a quantal match. In the anomaloscope, with both primaries chosen above 540 nm, the short-wavelength-sensitive visual photopigment (SWS)does not participate in the matches. In normal trichromats, red-green matches with a 10or a 20field are mediated by the middle- and long-wavelength sensitive visual photopigments (MWSand LWS). For each test wavelength, there is a unique matching red-green ratio which satisfies the quantal match.
0.720-0.950
(a)
40
,
40
(I
b
:E
30
30 0
20
20
0
10
aj
z Z
o
0 D
O
O
0.5
1.0
FRACTION RED
0
-- - --
0.5 FRACTION
1.R RED
FIG. 1. Anomaloscope matches made by protanopes rpanel (a)] and deuteranopes [panel (b)] to a 555 rm test field. The solid
line with vertical bars shows color and luminance matches made by four protanopes and six deuteranopes using the 20 field. The dashed line shows predicted scotopic luminance matches. The solid bar is the unique matching range for protanope CB [panel (a)] and deuteranope DR [panel (b)] using the 8' field. The arrows under the letters N, PA, and DA point to the matches made by a normal, a protanomalous, and a deuterano-
malous trichromat, respectively. V. C. Smith and J. Pokorny
216
l1
l1
(b)
40
(a)
40
(a)
40
7'2
40
-. (b).
a,
30
30
o
30
30 )-
1
0 1
20
20
a. 20
20
0 Iu
.11 10
10
.1
--.
0
1
Z
1o
M-
2 0l
0
0
0.5
1.0
FRACTION
0
DA 4
,
4
0
, I
0.5 FRACTION
RED
01 -,
I
1.0
0
RED
0.5 FRACTION RED
1.0
0
.I
w
0.5
FRACTION
1.0
RED
FIG. 2. As for Fig. 1, with a tes;t wavelength of 570 nm.
FIG. 4.
luminance matches provided all three groups of ob-
dashed line. Matches for the deuteranope DR and the protanope CB occur at or near the intersections of the theoretical rod luminance matches with the 20 luminance matches of the dichromats. The intersection
servers are similar in their inert pigment characteristics. Given the small number of observers, this expectation is well realized. Protanomalous and deuteranomalous trichromats do not accept the normal match: their visual photopigments differ from those of normals. Their unique matches represent intersection points of their underlying photopigments. Mid-match points. The plots in Figs. 1-6 present a graphical representation rather than the usual sets of linear equations. The unique matches made by a deu-
As for Fig. 1, with a test wavelength of 589 nm.
points are included in Table III to allow comparison
of
the predictions with match positions of the other observers. The large-field matches of our dichromats are quantal matches, mediated by the photoreceptors of small-field dichromacy plus a photoreceptor with the spectral sensitivity of rhodopsin.
normals, anomalous trichromats, or dichromats.
Match width. The match widths are strongly correlated with the difference in slope between the rod luminance function and the dichromatic luminance function. The match widths are narrowest at 610 nm where the rod function is steepest, and broadest at 555 nm where rod function is shallowest. In general, the protanopic match widths are narrower than deuteranopic match widths. This trend reverses at 555 nm where the rod and 2° protanopic luminance matches almost coincide over their range. The characteristic match width profile strengthens our evidence that the matches are mediated by the dichromatic cone photopigment and a visual photopigment whose spectral sensitivity is that of rhodopsin.
To evaluate the possibility that the matches were rodmediated, we calculated the luminance matches that
EXPERIMENT 4: MATCHES AS A FUNCTION OF INTENSITY
teranope,
CB, with the 8° field
DR, and a protanope,
are shown as solid bars. The luminance settings indicate that the 80 matches were also matches for the 20 field. For wavelengths 570-610, matches for deuteranope DR contain a higher proportion
of red primary
than those of normal or deuteranomalous trichromats; matches for protanope CB contain a higher proportion of red than those of normal or protanomalous trichromats. These matches are mediated by at least one visual photopigment which does not act in the 20 matches of
would be made if rods alone mediated the response.
These theoretical matches based upon the CIE scotopic luminosity function,
40
Vx are shown in Figs.
(a)
-
40 -
1-6 as a
When rods participate in large-field matches made by normal trichromats, quantal matches are rarely made.
(b)
a,I N
9
30)-
30
1
o 20
20 0
0
z
10
10
4
4 :E X.
A I
j . 0
0
0.5 FRACTION
1.0 RED
DA
O[
4l
)
0
b
!- I. 11.0
N
0.5 FRACTION
FIG. 3. As for Fig. 1, with a test wavelength 217
I
RED
of 580 nm.
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
FRACTION
RED
FRACTION
RED
FIG. 5. As for Fig. 1, with a test wavelength of 596 nm. V. C. Smith and J. Pokorny
217
570 NM (a)
40
40
589
610 NM
~P-OS
I
030
NM
Mb)
0.I
30
g
l*
0.01 0.001
l -20
20
0
10
SY
----
0.I Z
0.0I X
I
l
0.001
-J lO
-
0
,
1.0
0.5
0
0.5
FRACTION RED
FIG. 6.
-
PA -DA-
1.0
-
0.I
FRACTION RED
-J
As for Fig. 1, with a test wavelength of 610 nm.
w
D-MM
0.01 0.001 l
Clarke1 9 uses the term tetrachromacy to allow for the activity of rods in normal color matching. There are three important properties of normal trichromacy which are not fulfilled in tetrachromacy. Normal trichromacy has the properties that (a) three primaries are sufficient for spectral color matching; (b) the relationship between the primaries is invariant, that is, the matches are unique; (c) they hold over a range of 1500 td. These are all properties of quantal matches. With an 80field, three primaries are sufficient and a unique match is made. However, the matches change with field luminance. If four primaries are allowed, the matches are not unique. Trezona20 devised a fourprimary method of systematically iterating between matches made at a scotopic and mesopic luminance until a unique match was achieved. This unique match is a quantal match for each of the four visual photopigments, and is therefore intensity invariant. Experiment 4 was designed to test whether the dichromatic matches
-
0.1
I
0.01 0.001
-
I- .*0
0.5
*
~
-
1.0 0
0.5
WJ
1.0 0
0.5
1.0
FRACTION RED
FIG. 7. Anoma loscope matches as a function of relative flux made by two protanopes (upper rows) and two deuteranopes (lower rows) to test wavelengths of 570 nm (column 1), 589 nm (column 2), and 610 nm (column 3) using the 8' test field. Lu-
minance at unit flux for the various conditions and classes of observer
may be obtained by reference
to Figs.
1, 3, and 5.
Procedure. We used primaries of 545 nm (green) and 670 nm (red) in the mixture field and a series of
each luminance and test wavelength in Fig. 7. The sets of matches show similar trends. Matches are not affected by a tenfold reduction of intensity. The matches tend to widen somewhat with a one-hundred-fold reduction and show greater widening with the one-thousandfold reduction of luminance. There is no shift of the match midpoint to a higher or lower fraction R. The 5 cd/m 2 match is always accepted. These data show an intensity invariance of the quantal matches with a
three wavelengths,
1000 to 1 change in luminance.
would hold with reduction in field luminance.
610, 589, and 570 nm in the test
field. The highest field luminance was about 5 candles/ m2 . The luminance was reduced in one-log-unit steps for a range of 3 log units. Light shielding was provided for the lowest intensities. The observers adapted to the field for up to 30 min. in the low ambient illumination of the darkened room. Two deuteranopes and two protanopes were observers. For each test wavelength and test luminance, a match range was established by the experimenter using the 80 field. Results and discussion.
The matches are plotted for
THEORETICAL IMPLICATIONS
Implicationsfor normal trichromacy. The protanopic and deuteranopic 80 matches may be compared with the 2° and the 8° matches made by normal trichromats (Table IV). The large-field trichromacy we observe in protanopes and deuteranopes shows the properties of a unique match and intensity invariance. The matches therefore show properties more similar to those of normal small-field trichromacy than of normal large-field
TABLE IV. Comparison of trichromacy and tetrachromacy in normal trichromats with large-field trichromacy in protanopes and deuteranopes. Query
Small-field Normal trichromats
Large-field Normal trichromats
Large-field trichromacy in protanopes and deuteranopes
How many primaries?
3
3
4
3
Are unique matches made?
Yes
Yes
No'
Yes
Are matches intensity-invariant?
Yes
No
No'
Yes
`An exception is the technique proposed by Trezona. 20 218
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
V. C. Smith and J. Pokorny
218
tetrachromacy.
field trichromacy.
The comparison points to a distinction between largefield matches of normals and dichromats. Normal trichromacy is both pigment and channel limited. With small fields, normal trichromacy is pigment limited: there are three cone visual photopigments, and at least three processing channels. Intensity-invariant quantal matches are made. With large fields, normal trichromacy is not pigment limited; there are four (three cone and one rod) visual photopigments. However, trichromatic matches can be made: they are not intensity invariant, nor are they quantal. These data indicate that the rods do not have an independent channel; their signals are carried on the channels carrying cone achromatic and chromatic information. In this sense, we say that the data are channel limited. There are only three orthogonal channels for color mixture.
An alternative explanation of our data is that the dichromats have only two processing channels, but reject all but the quantal match because of differing temporal and spatial transients 2 6 generated by cones and rods along the border of the test and matching fields. An unsatisfactory aspect of this explanation is that it implies that only dichromats, and not normal trichromats, are able to use such transient information to make large-field quantal matches under the free-view conditions that we use. We would expect that the use of the 80 annular stop with steady fixation on the black center would reduce information due to temporal border transients. Our observers refused classical dichromatic matches when the annular field was used (experi-
Psychophysical study of rod-cone interactions show a variety of effects, reviewed by Jacobs, 22 including mutual facilitation and mutual inhibition. McCann and Benton2 3 report that rods interacting with LWS cones yield an array of colors similar in color appearance to photopic mixtures
of wavelengths 495 and 656 nm.
Stabell and Stabell24 note that rod color does not occupy a unique locus in the CIE diagram. Jacobs22 observes that the rod interactions reported by Stabell and Stabell suggest that rods interact in the opponent chromatic channels.
Thus, although rods do not have an inde-
pendent activity in color matching, they can modulate opponent color signals. A neurophysiological basis may be noted in data from macaque monkeys. 2 5 When rods enter the chromatic opponents, the rod response is that of the field center. The four receptive field types (red on-center, red offcenter, green on-center, green off-center) would yield two distinctly different types of opponent response. Thus a red- Vx on-center or a red- VAoff-center would give different opponents from a green- Vt on-center or a green-V' off-center. In psychophysics, the four types of receptive field are treated as one chromatic opponent. Both the psychophysics 2 3 ' 2 4 and the Wiesel and Hubel25 data make it clear that any explanation of
rod effects on color appearance requires a more complex approach than is presently available in color theory.
Theoreticalinterpretation of large-field dichromatic
ments 1 and 2). 27 We suspect that while border tran-
sients may enhance color information for dichromats, they are not the cause of the large-field color-matching behavior. A third possibility is that the trichromacy is not mediated by normal rods. Possibly, the dichromats have cones whose photopigment has the spectral response of rhodopsin as postulated for some of the monochromacies. 28-32 If the latter is true, then protanopia and deuteranopia should be regarded as additional forms of anomalous trichromacy. If mediated by normal rods, the term dichromacy may be regarded as a special case of reduced color vision elicited when the small fields selectively allow cone response.
Do dichromats have a uniquesensory perceptfor red? The impetus for our study was the fact that many dichromats insist that they have a unique sensory percept for "red. " Our study is one of color matches which does not tell us anything about the color perceptions of dichromats, However, the fact that protanopes and deuteranopes become trichromatic with large fields means that they do have both the photopigment and possibly the processing channel that would permit an added unique sensory percept. We suspect that their color world is weak compared with that of normal trichromats, However, they do have trichromatic color vision for large fields and it may well be that this trichromacy forms the basis for the perceptual claims made by dichromats. ACKNOWLEDGMENTS
data. The intensity-invariant quantal matches made by our dichromats suggest to us that three cone-rod channels are available to protanopes and deuteranopes. Small-field dichromatic matches are pigment limited; there are two cone visual photopigments and at least two processing channels. With large fields, there are
The major impetus for this research came from a deuteranope, Michael Breton, who insists that he enjoys a unique sensory percept for red. We thank our research subjects for their cooperation.
three (two cone and one rod) visual photopigments
*Supported in part by NIH NEI grants EY00901 and EY70652 to Pokorny, and EY00523 to F. W. Newell. The data from Ex-
and
we suggest that there are at least three channels. The third channel may reflect rod activity in a residual redgreen opponent system. Alternatively, two neural channels may be formed from a blue-yellow channel in a manner similar to that discussed for the red-green channel of normal trichromats. The important point is that the rod signal has independent recognition in dichromatic color opponent channels, allowing the large219
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
periments 1-3 were presented at the annual meeting of A. R. V. 0. in 1976 and those of Experiment 4 at the 1976 Annual Meeting of the Optical Society of America.
'J. H. Parsons, An Introduction to the Study of Colour Vision, 2nd ed. (Cambridge U. P., Cambridge, 1924). D. B. Judd, "Colour perceptions of deuteranopic
2
and protanop-
ic observers, "t J. Res. Natl. Bur. Stand. 41, 247-271 (1948). 3C.
H. Graham,
H. G. Sperling,
Y. Hsia, and A. H. Coulson,
V. C. Smith and J. Pokorny
219
"The determination of some visual functions of a unilateral color-blind subject: methods and results, " J. Psychol. 51, 3-32 (1961). F. H. G. Pitt, "Characteristics of Dichromatic Vision," Med. Res. Council Spec. Rep. No. 200 H.M.S.O. (1935). 5S. Hecht and S. Shlaer, "The color vision of dichromats. I. 4
Wavelength discrimination, brightness distribution, and color mixture, " J. Gen. Physiol. 20, 57-82 (1937). GSee R. M. Boynton and H. Scheibner,
"On the perception
of
red by 'red blind' observers, " Acta Chromatica 1, 205-220 (1967). We used such explanations with dichromatic obser7 8
vers. W. A. Nagel, "Dichromatische
fovea, trichromatische
peri-
2nd ed., (Williams and Wilkins, Baltimore, 1970), pp. 199259. 22 G. H. Jacobs, "Color vision, " Ann. Rev. Psychol. 27, 6389 (1976). 23 J. J. McCann and J. L. Benton, "Interaction of the long-
wave cones and rods to produce color sensations, " J. Opt. Soc. Am. 59, 103-107 (1969).
24U.Stabell and B. Stabell, "Chromatic rod-cone interaction," Vision Res. 14.1389-1392 (1974).
pherie, " Z. Psychol. u. Physiol. Sinnes 39, 93-101 (1905).
25
W. A. Nagel, "Neue Erfahrungen
actions in the lateral geniculate body of the rhesus monkey," J. Neurophysiol. 29, 1115-1156 (1966). 26Thesuggestion that spatial and temporal border transients
jiber das Farbensehen
der
Dichromaten auf grossen Felde, " Z. Sinnesphysiol. 41, 3199
New York, 1973), pp. 319-324.
20p, W. Trezona, "The tetraclhronatic colour match as a colorimetric technique, "1Vision Res. 13, 9-25 (1973). 21 G. S. Brindley, Physiology of the Retina and Visual Pathway,
337 (1907). W. A. Nagel, "Farbenstimmung nesphysiol. 44, 5-17 (1910).
beim Dichromaten,
" Z. Sin-
10
W. Jaeger and K. Kroker, "fber das verhalten der protanopen und deuteranopen bei grossen reizflachen, " Klin MBL. Augen-
heilk. 121, 445-449 (1952). "1H. M. 0. Scheibaer and R. M. Boynton, "Residual red-green discrimination in dichromats, " J. Opt. Soc. Am. 58, 11511158 (1968). 1 2J. D. Moreland and W. B. Young, "A new anomaloscope em-
T. N. Wiesel and D. H. Hubel, "Spatial and chromatic
inter-
could explain our data was first made by D. I. A. MacLeod
at the annual meeting of A.R.V.O. in 1976. Weattempted to evaluate the effects of border transients in
27
two additional ways.
We constructed
an 8° field stop with a
2° black bar. The 2° bar prevents an observer from making direct comparisons across the test-field border. With this test field, discrimination deteriorated. However, protanopes SY and CB and deuteranopes DK and MM continued to reject the neutral point match of 494 rm to a mixture of 450 and
ploying interference filters, " Mod. Prob. Ophthal. 13, 47-55
650 nm. Their red-green matches for wavelengths 570, 589,
(1974). J. Pokorny and V. C. Smith, "Effect of field size on red-
and 610 nm widened, similar
13
green color mixture equations, " J. Opt. Soc. Am. 66, 705708 (1976). 14 A. Linksz, An Essay on1Color Vision (Grune and Stratton, New York, 1964). 15 G. Wyszecki and W. S. Stiles, Color Science (Wiley, New York, 1967).
K. H. Ruddock, "Parafoveal colour vision responses of four dichromats, " Vision Res. 11, 143-156 (1971).
17
D. I. A. MacLeod and P. Lennie, "Red-green blindness fineqj to one eye, " Vision Res. 16, 691-702 (1976).
con-
M. Alpern, "What is it that confines in a world without color? Invest. Ophthal. 13, 648-674 (1974). F. J. J. Clarke, "Needs and Prospects
9
28
G. L. Walls and G. B. Heath, "Typical total colour blindness
reinterpreted, " Acta Ophthal. 32, 253-297 (1954). 29L. Sloan, "The photopic retinal receptors of the typical achromat, " Am. J. Ophthal. 46, 81-86 (1958). 30
M. Alpern,
H. F. Falls,
and G. B. Lee, "The enigma of
typical total monochromacy, " Am. J. Ophthal. 50, 326-341 (1960).
38 J. Pokorny, V. C. Smith, and R. Swartley, "Threshold measurements of spectral sensitivity in a blue monocone
monochromat, " Invest. Ophthal. 9, 807-813 (1970).
for a tetrachromatic
system of large field colorimetry, " in Colour 73, Proceedings of the 2nd Congress of the I. C. A, at York, 1973 (Wiley,
220
to those for one-thousand-fold
reduction in field luminance (Fig. 7). Our second approach was to defocus the field so that the border was invisible. Again, discrimination was reduced, but not to the same degree as with the 2° bar separating the test and mixture fields.
J. Opt. Soc. Am., Vol. 67, No. 2, February 1977
32
M. Alpern,
G. B. Lee, F. Maaseidvaag,
and S. S. Miller,
"Colour vision in blue-cone 'monochromacy,
"
J. Physiol.
(Lond.) 212, 211-233 (1971).
V. C. Smith and J. Pokorny
220
