- Email: [email protected]
Temporal modulation sensitivities of red- and green-sensitive cone systems in dichromats
Huron Rrs. Vol. 24. so. 12. pp. 1995-1999. 1981 Primed m Great Britain. .All rights rrssnrd
0042.6989 8-I 53.00 + 0.00 Copyright (_ iY8-l Pergamon Press Ltd
RESEARCH
NOTE
TEMPORAL MODULATION SENSITIVITIES OF RED- AND GREEN-SENSITIVE CONE SYSTEMS IN DICHROMATS PETER LEBNE
University of Rochester, Center for Visual Science, Rochester. NY 14627. U.S.A. (Received 9 March 1984: in recked form 21 Max 1984) Abstract-The temporal modulation sensitivities of protanopes, deuteranopes and normal observers were measured with sinusoidal grating patterns (of spatial frequency 2.5 c/deg) whose contrast was modulated sinusoidally in time. Contrast sensitivity functions obtained from the three types of obseners are almost identical. If we accept that in each class of dichromat only one cone type (red-sensitive, R. or green-sensitive, G) supports threshold. and that post-receptoral filters in both classes of observers are the same. the results show that R and G cones have the same temporal properties. Flicker
Dichromacy
Cones
INTRODUCTION
The temporal properties of the red (R), green (G) and blue (B) sensitive cone systems of the human eye have been measured by several techniques that attempt to isolate one system at a time (Green, 1969; Kelly, 1974; Estivez and Spekreijse, 1974; Esttvez and Covonius, 1975). There is general agreement that the B cones have substantially lower contrast sensitivity than the R and G cones, with the result that at high temporal frequencies the B system is quite unresponsive to modulations that excite the R and G systems. There is no agreement that the R and G systems have the same temporal properties: Green (1969) isolated the two receptor systems by chromatic adaptation, and found that at low temporal frequencies the R system was the more sensitive while at high temporal frequencies the G was more sensitive. Kelly (1974) who used the same technique (although with less certain isolation), found the G system to be substantially more sensitive at all temporal frequencies. Estevez and Spekreijse (1974) isolated the different cone mechanisms by a substitution technique, and found that at all temporal frequencies the sensitivities of R and G systems were indistinguishable. Cicerone and Green (1978) reached the same conclusion from quite different experiments. One problem, recognized in all this work, is that the limitations to sensitivity may arise not in the receptors themselves, but in post-receptoral pathways. Considerable evidence (e.g. Kelly and van Norren, 1977; King-Smith and Carden, 1976) suggests that the chromatic pathway has poorer temporal resolution than the achromatic one, yet in none of the above work is it clear which pathway determined threshold. A promising way to avoid this problem is to establish the temporal contrast sensitivities of protanopes or deuteranopes, who are believed to lack
one or the other cone type. In these observers the achromatic pathway is presumed to be driven by just one type of cone (R or G), and we can relatively easily isolate it by luminance modulation alone. If postreceptoral filters are linear, then providing signals from cones pass through the same filters in both protanopes and deuteranopes, any difference between the shapes of the contrast sensitivity curves must reflect a difference between the temporal properties of R and G cones. The temporal sensitivities of dichromats have been little studied: Heath (1958) found that at middle and long wavelengths protanopes required greater threshold energy than did normals or deuteranopes at a criterion (high) flicker rate. This suggests that G cones have poorer sensitivity than R cones. Pokorny and Smith (1972) provided further support for this idea when they found that the graph against log luminance was considerably steeper for deuteranopes than for protanopes, with normal observers lying in between. In the light of these observations it is obviously desirable to have the temporal sensitivity of deteranopes and protanopes more fully characterized by the contrast sensitivity function. Kelly (1962) measured the sensitivities of one protanope and one deuteranope, but his methods did not permit precise comparison of observers. METHODS Apparam
Sinusoidal grating patterns were generated on the face of a television display (Joyce Electronics) by techniques described by Schade (1956). The depth of modulation of the gratings could be varied continuously from 0 to 0.95 without change in space-time averaged luminance, which was fixed at I75 cd m-2. The display had a white (P4) phosphor. The modu-
1995
I996
PETER LESSIE
lating waveform was generated by a PDP 1I ,3-% computer and was delivered to an analog multiplier via a Digital-to-.~nalo~ converter. A second (sinusoidal) signal from the computer was multiplied with the first to produce time-varying changes in contrast. The product was passed through a programmable attenuator before being fed to the Z-axis of the display. A pulse from the computer, synchronized to the modulating waveform, triggered a frame of the display every 5 msec (200 Hz). Obsercers
Normal observers were the author and students from within the department of psychology. Dichromats were recruited through an advertisement in the university newspaper. They were screened initially with ishihara pseudoisochromatic plates, and the City University Colour Test, and were positively identified as dichromats by their Rayleigh matches: all would accept a normal match of green (548 nm) and red (647 nmf to yellow (586 nm) and all could make an exact match between the 548 and 647 nm lights when presented in a 5 deg bipartite field. AH observers were between 20 and 36 years old and had acuity corrected to 6/6 or better. Procediire
Observers sat in a darkened room in a dental chair that provided firm head support, and viewed &he display from 6 m, at which distance the screen subtended 2 x 2.9 deg. A small dark fixation spot was painted on the center of the screen. In all measurements the gratings had a spatial frequency of 25cideg (close to the peak of the spatial contrast sensitivity curve for all observers) and flickered at temporal frequencies ranging from 0 to 45.5 Hz. Thresholds were measured by a two-alternative forced choice technique: each trial (I 760 msec) contained two 705 msec intervals, marked by distinctive tones and separated by 360 msec. During one or the other interval, randomly selected from trial to trial, a flickering grating appeared, with its contrast weighted by a raised cosine of period 705 msec. This appeared as a grating of smoothly increasing then decreasing contrast, and the observer identified the interval in which it had appeared by pressing one of two keys. From trial to trial one of 20 gratings appeared (ten different temporal frequencies each in the first or second interval). If a particular grating was detected correctly on three successive presentations the contrast at which it was presented on its next trial was reduced; a single error at any level of contrast caused contrast to be increased for the next trial. Following the first three correct responses the contrast was reduced by a factor of two; on subsequent downward reversals the amount by which contrast changed was reduced in steps until it reached 25%. The staircase sequence was stopped automatically when there had been five reversals at the smallest contrast step. This procedure permitted
the temporal contrast sensitivity curve to be measured reliably and rapidly. RESL’LTS
Pokorny and Smith (1972) showed that. when 580 nm test fields were used, the relationship between CFF and Logillumination was consistently steeper for deuteranopes than for protanopes. It therefore seemed worthwhile first to measure temporal contrast sensitivity at this wavelength. The observer viewed the display with his natural pupil through a 580 nm interference filter (Balzars, half-width 4 nm) placed 3 cm before the preferred eye. Under these conditions the pupils of all observers had diameters between 5.5 and 7 mm, and for the purposes of calculating mean retinal illumination (206 td) 6.25 mm has been assumed. Spectrophotometric measurements of filter transmittance showed that the greatest possible effect of filter obliquity would be to reduce the wavelength of peak transmittance by about 2 nm, a quite negligible change for these purposes. Figure I shows the results of contrast sensitivity measurements obtained from four normal observers (a), four deuteranopes (b) and five protanopes (c). On the high-frequency descending limb the average contrast sensitivities of protanopes mostly lie below those of normals, whose sensitivities mostly lie below those of deuteranopes, but the differences are all within the standard error of the mean of the normal curve. The different groups of observers could be better compared if their contrast sensitivities could be characterized by some function. One that fits the observations well (except at the lowest temporal frequencies) represents the temporal contrast sensitivity curve as the difference of two exponentials, thus: f‘(w) = Ae-b’,’ _ Cc-d’,’
(I)
where A and C are scale factors and b and d the time-constants of the two exponentials. The smooth curves in Fig. IA, B and C are the solutions to equation 1 (found by the minimizing routine STEPIT) that best fit the mean contrast sensitivities of each group. These means, and the corresponding solutions to equation 1 are plotted in Fig. ID. The CFF can be calculated by finding the temporal frequency for which J(w) = 1. Since the best-fitting values of A and b are similar for all three groups of observers, it is not surprising that the CFFs are also nearly the same: for normal observers 38 Hz, for deuteranopes 40.5 Hz, and for protanopes 38 Hz. At the corresponding level of mean illumination Pokorny and Smith (1972) found CFFs of 40 Hz for deuteranopes and 31 Hz for protanopes. Since in the earlier work the difference between deuteranopes and protanopes was greater at higher levels of illumination. a second series of measurements was made, this time with the white screen viewed directly, providing mean i~iumination of ciose to 5400 td. The results are shown in Fig. 2, separateiy
Temporal 1ooc
modulation
I997
sensitivities in dichromats
‘r Normal
Cc) Proton
(b) Deutan
td) Means
to)
CC
-/
I
Frequency
LHz)
Fig. I. Temporal contrast sensitivity curves for sinusoidal gratings of 2.5 c/deg. Mean illumination 206td (580 nm). Extreme left-hand set of points indicate thresholds for temporally unmodulated gratings. A: normal observers; B: deuteranopes; C: protanopes; D: mean sensitivities of normals (open circles,
continuous
line), deuteranopes (solid circles, dashed Smooth curves are best-fitting
for normals (2A), deoteranopes (B) and protanopes (2C). The best-fitting solutions to equation I are plotted in each Figure, and together in Fig. 2D. The contrast sensitivities of the three classes of observer are substantially higher than in the measurments of Fig. 1, although the differences between deuteranopes and protanopes are smaller. In fact, a single set of parameters enables a satisfactory fit of equation I to all three sets of means. CFFs calculated from the best-fitting solutions to equation I are 58.5, 58 and 57 Hz for normals, deuteranopes and protanopes, respectively. Extrapolated to the corresponding level of mean illumination, Pokorny and Smith’s results indicate a deuteranopic CFF of 53 Hz and a protanopic CFF of 42 Hz.
line) and solutions
protanopes to equation
(squares, I.
dot-dashed
line).
DISCUSSION
The results show that the temporal contrast sensitivity functions of normal, deuteranopic and protanopic observers, when measured with 580nm or white light, are essentially indistinguishable. To the extent that the results obtained from deuteranopes and protanopes reflect the activity of a single class of cone (which seems very likely in view of the spectral composition of the lights involved) Figs I and 2 confirm the observations of EstCvez and Spekreijse (I 974) that the temporal properties of R and G cones are identical, at least in the range 0 to 50 Hz. Postreceptoral attenuation by the same linear filter in protanopes and deuteranopes could not result in
1999
r
(bl
I 1
Deuran
I
,
IO
100 Frequency
Fig. 2. Temporal
contrast
I HZ I
sensitivity curves for sinusoidal gratings of 2.5 c,‘deg. Mean illumination (white). Other details as for Fig. I.
identical curves were the temporal properties of R and G cones actually different. The present findings conflict with the measurements of Pokorny and Smith (1972), which show deuteranopes to have a clear and progressive advantage in temporal resolution as level of illumination is raised. Where the present findings can be compared with those of Pokorny and Smith, the deuteranopic observers have similar CFFs (58 vs 55 Hz); protanopic and normal observers performed better in the present experiments than in those of Pokorny and Smith. Indeed, for the protanopes’ CFF at 5400 td to be as low as that found by Pokorny and Smith (42 Hz), the contrast sensitivity curve of Fig. 2C would have had to be about 0.9 log unit lower on the ordinate. The principal difference between experiments is
5400 td
that the present work used an extended grating (2.0 x 2.9deg) of optimal spatial frequency (2.5 c deg-‘) while Pokorny and Smith used a uniform disc I .O deg in diameter. Were there a relative dearth of G cones in the central fovea, Pokorny and Smith’s protanopes may have been disadvantaged by comparison with those in the present experiments. However, it is hard to see how a CFF change that corresponds to a 0.9 log unit loss of contrast sensitivity could be produced by this means. A second possibility is that R and G cones have receptive fields of different dimensions, and that G cones are relatively less sensitive to low spatial frequencies. By comparison with R cones, G cones would then be relatively less sensitive to the stimuli used by Smith and Pokorny than to those used in the present experiments.
Temporal
.-lcX-nu~ried.~rmmrs--W. kindly commented at the University from the 4lRC. Rochester with from NIH.
Makous
and
D.
modulation
R.
sensitivities
Williams
on the manuscript. This work was begun of Sussex. with the support of a grant and completed at the University of the support
of a grant
(5 ROI
EYO44-tO)
REFERENCES Cicerone C. M. and Green D. G. (1978) Relative modulation sensitivities of the red and green color mechanisms. L'kion RCS. 18, 1593-1598. Estevez 0. and Cavonius C. R. (1975) Flicker sensitivity of the human red and green color mechanisms. Vision RPS. 15. 879-881. Estivez 0. and Spekreijse H. (1974) A spectral compensation method for determining the Hicker characteristics of the human colour mechanisms. Kcion Res. 14. 823-830. Gwen D. G. (1969) Sinusoidal nicker characteristics of the
in dichromats
I999
color-sensitive mechamsms of the eye. C’iJion Rrs. 9, 591-601. Heath G. G. (1958) Luminosity curves of normal and dichromatic observers. Scienc~e 128, 775-776. Kelly D. H. (1962) yisual responses to time-dependent stimuli: III. Individual variations. J. opr. Sot. Am. 52. 89-95. Kelly D. H. (1974) Spatio-temporal frequency characteristics of color-vision mechanisms. J. opf. Sot. Am. 64, 983-990. Kelly D. H. and van Norren D. (1977) Two-band model of heterochromatic flicker. J. opt. Sot. Am. 67, 1081-1091. Kine-Smith P. E. and Carden D. (1976) Luminance and opponent-color contributions to visual detection and adaptation and to temporal and spatial integration. J. opr. sot. Am. 66, 709-717. Pokorny J. and Smith V. C. (1972) Luminosity and CFF in Deuteranopes and Protanopes. J. opt. Sot. Am 62, 111-117. Schade 0. H. Sr (I 956) Optical and photoelectric analo_g of the eye. J. opt. Sot. hz. 46, 721-739.
0042.6989 8-I 53.00 + 0.00 Copyright (_ iY8-l Pergamon Press Ltd
RESEARCH
NOTE
TEMPORAL MODULATION SENSITIVITIES OF RED- AND GREEN-SENSITIVE CONE SYSTEMS IN DICHROMATS PETER LEBNE
University of Rochester, Center for Visual Science, Rochester. NY 14627. U.S.A. (Received 9 March 1984: in recked form 21 Max 1984) Abstract-The temporal modulation sensitivities of protanopes, deuteranopes and normal observers were measured with sinusoidal grating patterns (of spatial frequency 2.5 c/deg) whose contrast was modulated sinusoidally in time. Contrast sensitivity functions obtained from the three types of obseners are almost identical. If we accept that in each class of dichromat only one cone type (red-sensitive, R. or green-sensitive, G) supports threshold. and that post-receptoral filters in both classes of observers are the same. the results show that R and G cones have the same temporal properties. Flicker
Dichromacy
Cones
INTRODUCTION
The temporal properties of the red (R), green (G) and blue (B) sensitive cone systems of the human eye have been measured by several techniques that attempt to isolate one system at a time (Green, 1969; Kelly, 1974; Estivez and Spekreijse, 1974; Esttvez and Covonius, 1975). There is general agreement that the B cones have substantially lower contrast sensitivity than the R and G cones, with the result that at high temporal frequencies the B system is quite unresponsive to modulations that excite the R and G systems. There is no agreement that the R and G systems have the same temporal properties: Green (1969) isolated the two receptor systems by chromatic adaptation, and found that at low temporal frequencies the R system was the more sensitive while at high temporal frequencies the G was more sensitive. Kelly (1974) who used the same technique (although with less certain isolation), found the G system to be substantially more sensitive at all temporal frequencies. Estevez and Spekreijse (1974) isolated the different cone mechanisms by a substitution technique, and found that at all temporal frequencies the sensitivities of R and G systems were indistinguishable. Cicerone and Green (1978) reached the same conclusion from quite different experiments. One problem, recognized in all this work, is that the limitations to sensitivity may arise not in the receptors themselves, but in post-receptoral pathways. Considerable evidence (e.g. Kelly and van Norren, 1977; King-Smith and Carden, 1976) suggests that the chromatic pathway has poorer temporal resolution than the achromatic one, yet in none of the above work is it clear which pathway determined threshold. A promising way to avoid this problem is to establish the temporal contrast sensitivities of protanopes or deuteranopes, who are believed to lack
one or the other cone type. In these observers the achromatic pathway is presumed to be driven by just one type of cone (R or G), and we can relatively easily isolate it by luminance modulation alone. If postreceptoral filters are linear, then providing signals from cones pass through the same filters in both protanopes and deuteranopes, any difference between the shapes of the contrast sensitivity curves must reflect a difference between the temporal properties of R and G cones. The temporal sensitivities of dichromats have been little studied: Heath (1958) found that at middle and long wavelengths protanopes required greater threshold energy than did normals or deuteranopes at a criterion (high) flicker rate. This suggests that G cones have poorer sensitivity than R cones. Pokorny and Smith (1972) provided further support for this idea when they found that the graph against log luminance was considerably steeper for deuteranopes than for protanopes, with normal observers lying in between. In the light of these observations it is obviously desirable to have the temporal sensitivity of deteranopes and protanopes more fully characterized by the contrast sensitivity function. Kelly (1962) measured the sensitivities of one protanope and one deuteranope, but his methods did not permit precise comparison of observers. METHODS Apparam
Sinusoidal grating patterns were generated on the face of a television display (Joyce Electronics) by techniques described by Schade (1956). The depth of modulation of the gratings could be varied continuously from 0 to 0.95 without change in space-time averaged luminance, which was fixed at I75 cd m-2. The display had a white (P4) phosphor. The modu-
1995
I996
PETER LESSIE
lating waveform was generated by a PDP 1I ,3-% computer and was delivered to an analog multiplier via a Digital-to-.~nalo~ converter. A second (sinusoidal) signal from the computer was multiplied with the first to produce time-varying changes in contrast. The product was passed through a programmable attenuator before being fed to the Z-axis of the display. A pulse from the computer, synchronized to the modulating waveform, triggered a frame of the display every 5 msec (200 Hz). Obsercers
Normal observers were the author and students from within the department of psychology. Dichromats were recruited through an advertisement in the university newspaper. They were screened initially with ishihara pseudoisochromatic plates, and the City University Colour Test, and were positively identified as dichromats by their Rayleigh matches: all would accept a normal match of green (548 nm) and red (647 nmf to yellow (586 nm) and all could make an exact match between the 548 and 647 nm lights when presented in a 5 deg bipartite field. AH observers were between 20 and 36 years old and had acuity corrected to 6/6 or better. Procediire
Observers sat in a darkened room in a dental chair that provided firm head support, and viewed &he display from 6 m, at which distance the screen subtended 2 x 2.9 deg. A small dark fixation spot was painted on the center of the screen. In all measurements the gratings had a spatial frequency of 25cideg (close to the peak of the spatial contrast sensitivity curve for all observers) and flickered at temporal frequencies ranging from 0 to 45.5 Hz. Thresholds were measured by a two-alternative forced choice technique: each trial (I 760 msec) contained two 705 msec intervals, marked by distinctive tones and separated by 360 msec. During one or the other interval, randomly selected from trial to trial, a flickering grating appeared, with its contrast weighted by a raised cosine of period 705 msec. This appeared as a grating of smoothly increasing then decreasing contrast, and the observer identified the interval in which it had appeared by pressing one of two keys. From trial to trial one of 20 gratings appeared (ten different temporal frequencies each in the first or second interval). If a particular grating was detected correctly on three successive presentations the contrast at which it was presented on its next trial was reduced; a single error at any level of contrast caused contrast to be increased for the next trial. Following the first three correct responses the contrast was reduced by a factor of two; on subsequent downward reversals the amount by which contrast changed was reduced in steps until it reached 25%. The staircase sequence was stopped automatically when there had been five reversals at the smallest contrast step. This procedure permitted
the temporal contrast sensitivity curve to be measured reliably and rapidly. RESL’LTS
Pokorny and Smith (1972) showed that. when 580 nm test fields were used, the relationship between CFF and Logillumination was consistently steeper for deuteranopes than for protanopes. It therefore seemed worthwhile first to measure temporal contrast sensitivity at this wavelength. The observer viewed the display with his natural pupil through a 580 nm interference filter (Balzars, half-width 4 nm) placed 3 cm before the preferred eye. Under these conditions the pupils of all observers had diameters between 5.5 and 7 mm, and for the purposes of calculating mean retinal illumination (206 td) 6.25 mm has been assumed. Spectrophotometric measurements of filter transmittance showed that the greatest possible effect of filter obliquity would be to reduce the wavelength of peak transmittance by about 2 nm, a quite negligible change for these purposes. Figure I shows the results of contrast sensitivity measurements obtained from four normal observers (a), four deuteranopes (b) and five protanopes (c). On the high-frequency descending limb the average contrast sensitivities of protanopes mostly lie below those of normals, whose sensitivities mostly lie below those of deuteranopes, but the differences are all within the standard error of the mean of the normal curve. The different groups of observers could be better compared if their contrast sensitivities could be characterized by some function. One that fits the observations well (except at the lowest temporal frequencies) represents the temporal contrast sensitivity curve as the difference of two exponentials, thus: f‘(w) = Ae-b’,’ _ Cc-d’,’
(I)
where A and C are scale factors and b and d the time-constants of the two exponentials. The smooth curves in Fig. IA, B and C are the solutions to equation 1 (found by the minimizing routine STEPIT) that best fit the mean contrast sensitivities of each group. These means, and the corresponding solutions to equation 1 are plotted in Fig. ID. The CFF can be calculated by finding the temporal frequency for which J(w) = 1. Since the best-fitting values of A and b are similar for all three groups of observers, it is not surprising that the CFFs are also nearly the same: for normal observers 38 Hz, for deuteranopes 40.5 Hz, and for protanopes 38 Hz. At the corresponding level of mean illumination Pokorny and Smith (1972) found CFFs of 40 Hz for deuteranopes and 31 Hz for protanopes. Since in the earlier work the difference between deuteranopes and protanopes was greater at higher levels of illumination. a second series of measurements was made, this time with the white screen viewed directly, providing mean i~iumination of ciose to 5400 td. The results are shown in Fig. 2, separateiy
Temporal 1ooc
modulation
I997
sensitivities in dichromats
‘r Normal
Cc) Proton
(b) Deutan
td) Means
to)
CC
-/
I
Frequency
LHz)
Fig. I. Temporal contrast sensitivity curves for sinusoidal gratings of 2.5 c/deg. Mean illumination 206td (580 nm). Extreme left-hand set of points indicate thresholds for temporally unmodulated gratings. A: normal observers; B: deuteranopes; C: protanopes; D: mean sensitivities of normals (open circles,
continuous
line), deuteranopes (solid circles, dashed Smooth curves are best-fitting
for normals (2A), deoteranopes (B) and protanopes (2C). The best-fitting solutions to equation I are plotted in each Figure, and together in Fig. 2D. The contrast sensitivities of the three classes of observer are substantially higher than in the measurments of Fig. 1, although the differences between deuteranopes and protanopes are smaller. In fact, a single set of parameters enables a satisfactory fit of equation I to all three sets of means. CFFs calculated from the best-fitting solutions to equation I are 58.5, 58 and 57 Hz for normals, deuteranopes and protanopes, respectively. Extrapolated to the corresponding level of mean illumination, Pokorny and Smith’s results indicate a deuteranopic CFF of 53 Hz and a protanopic CFF of 42 Hz.
line) and solutions
protanopes to equation
(squares, I.
dot-dashed
line).
DISCUSSION
The results show that the temporal contrast sensitivity functions of normal, deuteranopic and protanopic observers, when measured with 580nm or white light, are essentially indistinguishable. To the extent that the results obtained from deuteranopes and protanopes reflect the activity of a single class of cone (which seems very likely in view of the spectral composition of the lights involved) Figs I and 2 confirm the observations of EstCvez and Spekreijse (I 974) that the temporal properties of R and G cones are identical, at least in the range 0 to 50 Hz. Postreceptoral attenuation by the same linear filter in protanopes and deuteranopes could not result in
1999
r
(bl
I 1
Deuran
I
,
IO
100 Frequency
Fig. 2. Temporal
contrast
I HZ I
sensitivity curves for sinusoidal gratings of 2.5 c,‘deg. Mean illumination (white). Other details as for Fig. I.
identical curves were the temporal properties of R and G cones actually different. The present findings conflict with the measurements of Pokorny and Smith (1972), which show deuteranopes to have a clear and progressive advantage in temporal resolution as level of illumination is raised. Where the present findings can be compared with those of Pokorny and Smith, the deuteranopic observers have similar CFFs (58 vs 55 Hz); protanopic and normal observers performed better in the present experiments than in those of Pokorny and Smith. Indeed, for the protanopes’ CFF at 5400 td to be as low as that found by Pokorny and Smith (42 Hz), the contrast sensitivity curve of Fig. 2C would have had to be about 0.9 log unit lower on the ordinate. The principal difference between experiments is
5400 td
that the present work used an extended grating (2.0 x 2.9deg) of optimal spatial frequency (2.5 c deg-‘) while Pokorny and Smith used a uniform disc I .O deg in diameter. Were there a relative dearth of G cones in the central fovea, Pokorny and Smith’s protanopes may have been disadvantaged by comparison with those in the present experiments. However, it is hard to see how a CFF change that corresponds to a 0.9 log unit loss of contrast sensitivity could be produced by this means. A second possibility is that R and G cones have receptive fields of different dimensions, and that G cones are relatively less sensitive to low spatial frequencies. By comparison with R cones, G cones would then be relatively less sensitive to the stimuli used by Smith and Pokorny than to those used in the present experiments.
Temporal
.-lcX-nu~ried.~rmmrs--W. kindly commented at the University from the 4lRC. Rochester with from NIH.
Makous
and
D.
modulation
R.
sensitivities
Williams
on the manuscript. This work was begun of Sussex. with the support of a grant and completed at the University of the support
of a grant
(5 ROI
EYO44-tO)
REFERENCES Cicerone C. M. and Green D. G. (1978) Relative modulation sensitivities of the red and green color mechanisms. L'kion RCS. 18, 1593-1598. Estevez 0. and Cavonius C. R. (1975) Flicker sensitivity of the human red and green color mechanisms. Vision RPS. 15. 879-881. Estivez 0. and Spekreijse H. (1974) A spectral compensation method for determining the Hicker characteristics of the human colour mechanisms. Kcion Res. 14. 823-830. Gwen D. G. (1969) Sinusoidal nicker characteristics of the
in dichromats
I999
color-sensitive mechamsms of the eye. C’iJion Rrs. 9, 591-601. Heath G. G. (1958) Luminosity curves of normal and dichromatic observers. Scienc~e 128, 775-776. Kelly D. H. (1962) yisual responses to time-dependent stimuli: III. Individual variations. J. opr. Sot. Am. 52. 89-95. Kelly D. H. (1974) Spatio-temporal frequency characteristics of color-vision mechanisms. J. opf. Sot. Am. 64, 983-990. Kelly D. H. and van Norren D. (1977) Two-band model of heterochromatic flicker. J. opt. Sot. Am. 67, 1081-1091. Kine-Smith P. E. and Carden D. (1976) Luminance and opponent-color contributions to visual detection and adaptation and to temporal and spatial integration. J. opr. sot. Am. 66, 709-717. Pokorny J. and Smith V. C. (1972) Luminosity and CFF in Deuteranopes and Protanopes. J. opt. Sot. Am 62, 111-117. Schade 0. H. Sr (I 956) Optical and photoelectric analo_g of the eye. J. opt. Sot. hz. 46, 721-739.