- Email: [email protected]
Spatial extent of rod-cone and cone-cone interactions for flicker detection
Virion Res. Vol. 26. No. 6. pp. 917-925, 1986 Printed in Great Britain. All rights reserved
Copyright C
0042-6989~86 $3.00 + 0.00 1986 Pergamon Journals Ltd
SPATIAL EXTENT OF ROD-CONE AND CONE-CONE INTERACTIONS FOR FLICKER DETECTION NANCY J. School
of
optometry.
and ANTHONYJ.
COLETTA*
ADAMS
University of California, Rerkeley, CA 94720, U.S.A
(Received 14 October 1985; in reviscdjiwm 30 December
1985)
Ah&act-Over a large range of tight adaptation levels, sensitivity to 25 Hz flicker improves as the light level of the background increases. Using small background discs and annular surrounds. this c&t was shown to be mediated by the surround and not the average luminance of the test region, in agreement with recent reports. The effect is due to two types of lateral interaction: at mesopic light levels (from 0.1 to 1.0 td), cone-mediated flicker resolution is enhanced by the stimulation of surrounding rods; at photopic light levels (above IO td), ficker sensitivity improves with light stimulation of adjacent cones. The spatial zone, or extent, over which the surround contributes to the flicker threshold was measured. The spatial area over which rods influence the cone 25 Hz Sicker threshold is larger than the analogous spatial area of cone intluence. In the parafovea, at 5 deg azentricity, cone ticker sensitivity for a 20’spot is infiuenced by cones in a 1 deg diameter area centered on the spot; the corresponding area of rod influence is about 3 deg. In the fovea, flicker sensitivity for a 10’ spot is influenced by cone stimulation in an area of about 20’ diameter. Rods which a&t foveal flicker sensitivity appear to occupy an annular zone with about a 2 deg outer diameter and 1 deg inner diameter, centered on the fovea. Flicker
Spatial se&i&on
Rod-cone
interaction
lNTRODUCl’lON
Sensitivity for the detection of a small spot centered on a background depends upon the spatial dimensions of the background disc (Crawford, 1940; Westheimer, 1965, 1967). As background diameter increases, sensitivity first decreases and then increases; this latter effect is known as spatial sensitization. There is some discrepancy in the literature about the specificity of this spatial interaction for single receptor mechanisms. The individual cone types are reported to have independent pathways for sensitization (McKee and Westheimer, 1970), but recent studies find evidence for spatial sensitization between the middle and long wavelength-sensitive cone types (Stromeyer, 1983) and the short wavelength-sensitive cones can be sensitized by other cone types (Haegerstrom-Portnoy and Adams, 1983). Similarly, an earlier study of the sensitization properties of rods and cones found that surrounding rod stimulation is unable to sensitize the cone threshold (Westheimer, 1970). More *Present address, to which correspondence should be sent: Center for Visual Science, University of Rochester, Rochester, NY 14627, U.S.A.
Lateral interaction
Light adaptation
recent studies, however, have demonstrated interactions between rods and cones in the sensitization effect. The cone increment threshold is increased by the presence of small rod-detected backgrounds (Frumkes ef al., 1973; ‘femme and Frumkes, 1977; Martinez er al., 1977; Latch and Lennie, 1977; Buck, 1980; Bauer et al., 1983a,b; Stromeyer and Hill, 1983; Buck, 1985a.b). Surrounding rod stimulation can also sensitize cone thresholds, because the cone threshold elevation in the presence of a small rod field can be reduced by expanding the rod background (Buck, 1980, 1981; Bauer et al., 1983a; Stromeyer and Hill, 1983; Buck, 1985a,b). A similar effect can be produced with annular surrounds. The cone threshold is elevated by the presence of a thin, rod-detected annulus placed close to the stimulus; as the annulus is moved away from the target, the cone threshold falls to the level obtained without an annulus (Buck, 1981; Buck, 1985b). The spatial interactions between rods and cones are complex, because the cone threshold elevation produced by a rod-detected background can also be reduced by the presence of a surrounding cone-detected annulus; also, the cone threshold elevation caused by a cone-detected field can be reduced
917
918
NANCY .I. COLETTAand ANTHONY J. ADAMS
by the presence of a surrounding rod annuius (Stromeyer and Hill, 1983). In the present study we examined the spatial properties for cone-detected sensitization flicker. Flicker sensitivity becomes degraded during the course of dark adaptation, or conversely, is enhanced by increasing the light adaptation level (Lythgoe and Tansley, 1929; Creed and Ruth, 1932; Fry and Bartiey, 1936; Berger, 1953, 1954; Foley, 1956, 1961; Kelly, 1959; Harvey, 1970). The adaptation effect on flicker is due, in part, to rod activity. As rods dark adapt, they suppress cone flicker sensitivity, as recently demonstrated in human psychophysical and ERG studies (Goldberg and Frumkes, 1983; Goldberg et at., 1982, 1983; Alexander and Fishman, 1984; Coietta and Adams, 1984; Arden and Frumkes, 1986) and in animal eiectrophysioiogicai studies (Frumkes er al., 1985). This rod-cone interaction for flicker has been demonstrated for small fovea1 targets: compared to the dark adapted condition, foveai flicker sensitivity is better in the presence of a rod-detected background. Since there are no rods at the fovea, this implies that the suppression of cone flicker sensitivity by dark adapted rods occurs over large retinal distances (Coietta and Adams, 1984; Adams and Coietta, 1984). Over a small range of photopic adaptation levels, there is an additional enhancement of flicker sensitivity as the background light level increases. Unlike the effect observed at mesopic levels, the spectral sensitivity of this effect resembles a cone mechanism (Coietta and Adams, 1984; Coietta er al., 1984). This suggests that light stimulation of cones can also improve cone flicker sensitivity. These adaptation effects on flicker sensitivity are qualitatively similar to the spatial sensitization effect for increment thresholds, because flicker sensitivity is improved as light is added to the region surrounding the target. It is true that, as the background intensity increases under a flickering target, the average luminance of the test target increases; therefore, one might invoke the Ferry-Porter law (i.e. flicker sensitivity improves as the target luminance increases) in order to explain the enhancement of Aicker sensitivity with background intensity. However, this idea does not explain the results obtained with annular surrounds. Flicker sensitivity is enhanced by increasing the intensity of an annuius centered on the target and this effect cannot be accounted for by stray light (Fry and
Bartiey, 1936; Berger, 1953. 1954; Foley, 1956; Harvey, 1970; Goldberg and Frumkes, 1983: Goldberg et al.. 1983; Adams and Coletta. 1984). Furthermore, flicker sensitivity increases as a background disc of fixed intensity is expanded (Foley, 1961: Goldberg and Frumkes, 1983; Alexander and Fishman, 1984; Adams and Coietta, 1984). The surrounding light level is an important factor in the adaptation effects on flicker sensitivity. In this study, the spatial extent of the rod-cone and cone-cone interactions for flicker were investigated by measuring cone flicker thresholds for small targets su~rimposed upon various sized background discs. Annular surrounds with increasing inner diameter were also used to measure the distance over which the lateral effects occur. METHODS
Stimuli The stimulus was a small, 620 nm, circular test target which flickered at 25 Hz. Previous test sensitivity results indicate that this target flicker is detected by cones (Coietta and Adams, 1984). The flickering target was presented for 750 msec every 2.25 set (1.5 set interstimulus interval). In the experiments at mesopic adaptation levels, four dim, red fixation dots arranged in a diamond were placed around the fovea1 target, i/2 deg from the target; for the parafoveai condition, a dim, red fixation spot was placed 5 deg away from the test target. The fixation targets for the photopic experiments were small black dots, arranged in the same fashion. The surrounding illumination was configured either as an annuius or a background disc. For the experiments with annular surrounds, a 500nm, steady, concentric annuius was superimposed on the target: the annuius outer diameter subtended 4 deg and the inner diameter was varied. The intensity of the annulus was also variable. Further details of the stimuli are given in the results section. For the sensitization paradigm, the 500nm, steady background discs were superimposed on the target, and the disc diameter was varied. Two adaptation levels were chosen for each eccentricity: one of the rod-mediated and one on the cone-mediated portion of the flicker threshold vs radiance curve for a 620 nm target on a 500 nm background (Coietta and Adams, 1984). In order to suppress any surrounding rod activ-
919
Size of lateral interactions for flicker
ity for the photopic sensitization experiment, a 10 deg diameter blue-green (Wratten No. 64) auxiliary field was used. The auxiliary field was 63 phot td (about 500 scat td) in the fovea and 10 phot td (about 80 scat td) in the parafovea. Apparatus The apparatus for these experiments was a two-channel Maxwellian view optical system, previously described in detail (Coletta and Adams, 1984). Variable background sizes were produced by changing the aperture size in the background channel. For studies with annular surrounds, the annuh were made on photographic slides and placed in the background channel. A third channel was added for these studies in order to provide a dim, red (Wratten No. 70) fixation light for the mesopic experiments, and a blue-green (Wratten No. 64) auxiliary surround to suppress rod activity for the photopic sensitization experiments. The source for this channel was an incandescent lamp of 40 W. The auxiliary field intensity was adjusted with Wratten No. 96 neutral density filters. Subjects Two experienced psychophysical observers with normal visual acuity, the authors, were used in these experiments. Subject N.C. was used for all the stimulus conditions; subject A.J.A., a mild deuteranomal, was used for the mesopic fovea1 condition only. Procedure At the beginning of a test session, the subject dark adapted for 10 min. This was shown to be an adequate dark adaptation period, because the flicker threshold on the dimmest background was the same after 25 min of dark adaptation as after 10 min (Coletta and Adams, 1984). The subject then adapted to the dimmest background or annulus for 2 min. At the end of this adaptation period, the test stimulus was presented and the subject set a flicker threshold by the method-of-adjustment. Flicker threshold is defined as the intensity of the test target needed to just perceive flicker. The flicker threshold is often far above the detection threshold, so the target may still be bright even when the flicker is no longer perceived. To avoid adaptation or habituation to the stimulus, the subject dimmed the target after making each flicker threshold setting, and so began the next setting from below the flicker threshold. At least
three readings were taken for each stimulus condition. In order to examine the effect of surround adaptation level, the intensity of the annuli was varied. After the readings were taken for a given adaptation level, the subject adapted to the next highest surround intensity level for 2 min. The size of the annulus was kept constant during each testing session. The experiments with expanding background discs were done at two intensity levels. The surround intensity level remained constant during each testing session while the diameter of the disc was expanded. Readings were taken with the smallest diameter first; thresholds were measured on successively larger disc diameters. RESULTS
Fovea1 spatial extent: (i) annular surrounds The flicker threshold vs radiance (tvr) curve for a 10’ diameter spot on a 4 deg diameter background disc is shown in Fig. 1 (squares). The curve has two negative sloping portions. At the dim backgrounds, over the 1 log unit range between -0.7 and 0.3 log td (0.2-2 phot td; 1.616 scat td), there is a shallow negative slope. Our previous studies of the cone-detected flicker tvr curves revealed a rod action spectrum at these dim adaptation levels (Coletta and Adams, 1984). The shallow slope indicates that rod adaptation produces a gradual enhancement of flicker sensitivity. Over a range of photopic adaptation levels, between 1.8 and 2.3 log td (63-200 phot td), there is a steep negative slope. We have previously shown that this effect is cone-mediated. The flicker tvr curve measured with an annulus (4 deg outer diameter/ 17’ inner diameter) surrounding the fovea1 target is shown in Fig. I (circles). The data is virtually identical to the result with the background disc. One might attribute this coincidence of the data sets to stray light from the annulus failing on the target area. However, for the 200 phot td condition, the amount of stray light falling on the center of the annulus is not equal to 200 td; thus the stray light “pedestal” under the target is not equivalent to the pedestal provided by the 200 td disc. Yet for these two conditions, the threshold is the same. Additionally, the flicker threshold on a 200 td, 10’ diameter background disc (triangle in Fig. 1) is about 1 log unit higher than the thresholds on either the large disc or annulus. Thus, the light pedestal under the target con-
920
NANCYJ.
COLETTA
and
ANTHONY
J.
ADAMS
2
1
0
Log
retlnol tlluminonce
( phot
td)
Fig. 1. Fovea1 flicker threshold for a 10’ arc, 620 nm, 25 HZ target as a function of background or annulus retinal illuminance. Squares represent results for a 4 deg diameter, 500 nm disc; circles. 4 deg outer/l7’ inner dia 500 nm annulus. Triangle represents flicker threshold on a 200 phot td, IO’diameter 500 nm disc. Subject N.C.
tributes little to the flicker threshold, since the pedestals for the large and small background discs are equivalent. This is consistent with Berger’s result (1954) showing that a small steady light added to a flickering test patch lowered rather than increased the CFF; he concluded that stray light falling on the target could not account for the enhancement of flicker sensitivity by an annulus. Goldberg and Frumkes (1983) and Goldberg et al. (1983) also show that a large disc or contiguous annulus can produce the same flicker threshold. The flicker tvr curves for various annulus inner diameters are shown in Fig. 2. Note there is little change in the rod-mediated effect as the inner annular diameter is increased, up to
Log retlnol
1.5 deg. This would be consistent with retinal anatomy, which shows a rod-free area over approximately the central l-l.5 deg diameter. With larger inner diameters, 2.4 and 3.6 deg, the shallow negative slope is absent, indicating no rod effect. This shows that the rods responsible for the effect on fovea1 cone flicker thresholds are located in an annular zone, just outside the rod-free area, but they do not extend beyond about 2 degrees diameter. With the annular surrounds, a steep negative tvr slope at the bright backgrounds is still evident for inner annular diameters of 17’ arc. As the inner dimension increases, this steep, cone-mediated effect gradually disappears. This shows that the cone effect at photopic levels is
lllumlnonce
(phot Id)
Fig. 2. Fovea1 flicker threshold as a function of 500 nm annulus retinal illuminance. Symbols denote annulus inner diameter; outer diameter was 4 deg. Same test stimulus and subject as in Fig. 1.
921
Size of lateral interactions for flicker
ulation in an annular zone from I to 2 deg, centered on the fovea. The spatial sensitization function for the brighter, cone background is shown in the lower part of the figure. The flicker threshold drops to the minimum at about 20’ diameter. Thus, the threshold-lowering effect is mediated by cones within the central 20‘ diameter. Parafiveal spatial extent: (i) annular surrounds
Diam (min) 40
‘,
-i
’
(20
60
‘,
’
6
240
,’ 1
Log retinal araa I deg z 1
Fig. 3. Fovea1 flicker threshold as a function of background disc area for a 10‘ arc, 25 Hz, 620 nm target. Background wavelength was 500 nm. Vertical bars are the range of three readings. Solid circles depict results on 0.63 phot td background; open circle is dark adapted flicker threshold. Solid squares, results on 200 phot td background surrounded by a 63 phot td auxiliary field; open square, flicker ~re~old on 63 td auxiliary field atone. All data for subject N.C. except triangies, which depict data for subject A.J.A. on 0.63 phot td background.
mediated by stimulation within a small area surrounding the target, within the central 0.5 deg diameter.
In the parafovea, we found separate rod and cone adaptation effects on the flicker threshold, similar to those in the fovea (Coletta and Adams, 1984). In this study, we determined whether the parafoveal adaptation effects on flicker were larger in spatial extent than the fovea1 effects. Like the fovea1 results, the flicker tvr curve in the parsfovea has a shallow sloping portion for dim back~ounds below 2 phot td; this part of the curve has a rod action spectrum (see Fig. 1 of Coletta and Adams, 1984). From about 6 to 63 td, there is a steeper negative slope in the tvr curve, and the action spectrum over this portion of the curve resembles a cone opponent function (see Coletta and Adams, 1984; Coletta et al., 1984; Coletta and Adams, 1986).
I
A
Fovea1spatialextent: (ii) expanding backgrounds We measured the flicker threshold as a function of back~ound disc diameter at two adaptation levels, in order to determine the spatial extent of the separate rod- and cone-mediated effects. These measures confirm the results with annuli, and employ the more familiar spatial sensitization paradigm (Westheimer, 1965). The flicker threshold for the 10’ diameter fovea1 test was measured on a variable diameter 0.63 td background, which is on the rodmediated limb of the flicker tvr curve. The data is shown in the upper portion of Fig. 3. Note that the flicker threshold is unaffected until the background diameter reaches about 1 deg. From this point, up to about 2 deg, the flicker threshold drops by about 0.2 log unit. Although this effect is small, the amount agrees with the tvr data for this background. Since the action spectrum is that of rods, the th~shold-iowe~ng effect of the background is caused by rod stim-
Lag onnulus
innerarea Meg *
)
Fig. 4. Parafoveal Sicker threshold as a function of annulus inner area for a 20’ arc, 620 nm. 25 HE target at 5 deg ccc. Outer diameter was 4 dcg, except for data denoted by diamond symbol, which was for an annulus with outer diam of 7 deg/inner diameter 4.7 deg. Solid circles are results for a 2 td annulu% solid squares, results for 20 td annulus. Open triangle depicts dark adapted flicker threshold; open circle, flicker threshold on a 2 td, 4 deg disc; open square, flicker threshold on a 20 td, 4 deg disc. All data for subject N.C.
922
NANCY J.
COLETTA
and
In the parafovea, at 5 deg eccentricity, the effect of annular inner diameter was examined for two surround intensities; one was on the shallow, rod slope of the flicker tvr curve, at 2 phot td (16 scat td), and the other one was at a cone adaptation level, 20 phot td. The results, measured with a 20’ diameter test spot, are plotted as a function of annulus inner diameter in Fig. 4. The flicker threshold on the dim background is about 0.5 log unit lower than the dark adapted flicker threshold. The dim annulus with 17’ inner diameter produces the same threshold as the 4 deg disc, but as the inner diameter is enlarged, the flicker threshold gradually rises, nearly to the dark adapted level. This means that, unlike the fovea1 results, the shallow rod-mediated slope of the curves gradually disappears as the annular inner diameter increases. In the parafovea, rods are present at the target location, so the amount of rod influence could be affected by the small inner diameters. To determine whether the rod effect would still be present for a larger inner diameter if the overall annulus area were increased, the flicker threshold was measured with an annulus of 7 deg outer/4.7 deg inner diameter. The flicker threshold remains elevated for this condition, suggesting the parafoveal rod influence on the cone flicker threshold only extends over a diameter of about 4 deg. At the higher adaptation level (20 td), the cone-mediated effect is present for the 17’ inner diameter and lessened for the 36’ and 43’ inner diameters, but disappears for an inner diameter of I .l deg and larger. This indicates that the photopic effect on flicker sensitivity is mediated by cones within about a 1 deg diameter area, centered on the target.
ANTHONY J. ADAMS
within about a 3.5 deg area centered on the target. The results for the photopic background are shown in the lower portion of Fig. 5. The sensitization curve levels off at about 1 deg diameter; thus the surround influence is mediated by cones within 1 deg of the target. This is consistent with the results with annular surrounds. DISCUSSION
The present results indicate that the stimulation of either surrounding rods or cones can enhance cone-detected flicker sensitivity. Further, the results suggest that at adaptation levels dimmer than 2 phot td, fovea1 cone-detected flicker is influenced by rods located within an annular zone around the fovea. This zone has an outer diameter of about 2 deg and the inner diameter is the size of the rod-free area, about 1 deg in width. Parafoveal cone flicker, at 5 deg eccentricity, is influenced by rods within a 3.5 deg diameter area, centered on the target. Over a certain range of photopic adaptation, there is a further enhancement of flicker sensitivity due to the stimulation of surrounding cones. In the fovea, this sensitization area is about 20’ in diameter. At 5 deg eccentricity, the sensitization area is about 1 deg in diameter.
P c.
Parafoveal spatial extent: (ii) expanding backgrounds
The parafoveal flicker tvr curve is similar to the fovea1 curve over the shallow, rod-mediated slope, but the steep cone slope occurs at a dimmer adaptation level in the parafovea than in the fovea. Again, the spatial sensitization effects were examined for both the rod- and cone-mediated adaptation levels. The spatial sensitization results for the 20’ diameter target on the 0.63 td, rod-mediated background are shown in the upper curves of Fig. 5. Note the curve levels off at 34 deg diameter, suggesting the rod influence on cone-detected flicker in the parafovea is due to the stimulation of rods
\ 20 td
\, Dlam -II
(mln
)
20
10 ’
40
60
120
240
1 0
I -1 Lag
7
retinal
area
I 1 (deg’
I
Fig. 5. Parafoveal flicker threshold as a function of background disc area for 20’ arc, 25 Hz, 620 nm target at 5 deg eccentricity. Upper curve for 0.63 phot td disc; lower curve for 20 phot td disc, surrounded by a 10 td auxiliary field. Open triangle depicts dark adapted flicker threshold; open square depicts Bicker threshold on IO td field alone. Subject N.C.
Size of lateral interactions for flicker
Possible edge efects The influence of the surround on flicker sensitivity cannot be adequately explained as an effect due to edge contours. On a qualitative basis, the threshold is higher when the edge of a disc is close to the target, but for annuli the threshold is lower when the annulus inner diameter is nearer to the target (Figs 1 and 2). The disc condition presents a lower contrast edge because the disc is superimposed on the auxiliary field, yet this lower contrast edge condition produces a higher flicker threshold. Size of the rod and cone surround efects The cone sensitization areas shown in the present results are slightly larger than those previously reported for increment threshold studies. Westheimer (1967) reported cone sensitization areas of 14.2’ diameter at the fovea and 35.7’ at 5 deg eccentricity. Our corresponding values for flicker are 20’ and 60’ diameter, respectively. This discrepancy is probably not due to the use of larger test targets in our study. For increment thresholds, large test probes yield a larger densensitizing zone, and a smaller magnitude of the sensitization effect, but do not alter the overall size of the sensitization area (Alexander, 1974; Matin and Komheiser, 1973; Matin and Komheiser, 1976). Our present results suggest that the sensitization zones for flicker are larger than the zones for increment detection; this implies that channels sensitive to temporal transients involve larger receptive fields or size-tuning. This would be consistent with neurophysiological studies in cat (Movshon et al., 1978) and human psychophysical studies (Keesey, 1972; Kulikowski and Tolhurst, 1973; Tolhurst, 1973) which show that the class of pathways sensitive to higher temporal frequencies tend to be tuned to lower spatial frequencies. The present results also show that the rodmediated effect on cone flicker extends over a larger area than the cone-mediated effect. This would be consistent with the model for rod-cone interaction proposed by Bauer et al. (1983a), in which signals from a rod channel with its own inhibitory surround interact with cone signals at a subsequent detector, which sums inputs from both rod and cone channels. Since rod sensitization areas are larger than those for cones (Westheimer, 1965, 1967), one might expect a large spatial extent for the rod influence on cones. However, caution must be
923
taken in comparing flicker sensitization results with a model designed for increment thresholds. As discussed in the following section, there are many discrepancies between the sensitization properties for flicker and increment thresholds. Comparison tization
with increment
threshold sensi-
The superficial similarity between cone flicker and increment threshold sensitization is that, in each case, an increase in the diameter of a superimposed background produces an increase in sensitivity. A major discrepancy, however, is that, for increment thresholds, the increase in sensitivity is relative, whereas for flicker thresholds, it is absolute. Unlike the spatial sensitization functions for increment thresholds, there is no desensitizing limb for the flicker threshold. At the light levels used in the present experiment, the addition of a background smaller than the target has little influence on the flicker threshold, or even decreases the flicker threshold below the dark adapted level (see Fig. 5). The sensitization properties for flicker also cannot be modelled by the behavior of the flicker tvr curve. In the more familiar case of increment threshold sensitization, adding a surround to a small background disc could reduce the effective stimulation from the center of the background disc; according to the increment threshold vs intensity curve, this decrease in effective light stimulation should lower the increment threshold and produce sensitization. In the case of the flicker, however, the flicker threshold vs intensity curve has a negative slope; i.e. the flicker threshold falls as light stimulation increases. If the addition of a surround reduces the effective stimulation from the central part of the background disc, this should drive the flicker threshold up instead of down. The mechanism for flicker sensitization shown in our results, therefore, is probably not the reduction of a saturating influence from the central part of the background, as proposed for increment thresholds. Sensitization for the increment threshold may involve an indirect effect of the surround upon the detection mechanism, by reducing the effectiveness of the center stimulation in raising the increment threshold. The lateral interaction for flicker, however, could be a direct action of the surround stimulation upon the sensitivity of the flicker detection mechanism. Two other aspects of the rod influence on
NANCY J. COLE~A anId ANTHONYJ. AUAMS
924
cone flicker are inconsistent with rod-cone interaction for cone increment thresholds. First, the cone absolute threshold for detecting small red flashes (presented alone, without a rod background) is relatively unaffected by the state of rod dark adaptation (Bauer et al., 1983b). Rod dark adaptation, however, has a substantial effect on the cone flicker threshold (Goldberg et al., 1983; Coletta and Adams, 1984; Alexander and Fishman, 1984). Secondly, since there are no rods at the fovea, the rod surround effect for fovea1 flicker cannot be modelled with a rod channel having a center-surround organization superimposed on the target region, as proposed for increment thresholds (Bauer et al., 1983a). According to recent el~trophysio~ogica~ and psychophysical reports, the surrounding rod influence on cone flicker occurs at a distal site in the retina, either at the receptor level or outer plexiform layer (Alexander and Fishman, 1984; Arden and Frumkes, 1986; Coletta and Adams, 1984; Frumkes et al., 1985). A direct influence of the surround on the flicker detection mechanism, as discussed above, is consistent with recent evidence that rods influence cone flicker directly through horizontal cell feedback onto cones (Frumkes et al., 1985). The cone-mediated effect at photopic levels may be a similar mechanism, but the presence of an opponent mechanism in the parafoveal photopic sensitization (Coletta and Adams, 1984; Coletta et al., 1984; Coletta and Adams, 1986) suggests that more proximal sites are involved in the cone-cone interaction for flicker.
Acknowledgemenrs-This work was supported by NIH grant EY02271 to A.J.A., and by training grant support to N.J.C.. on EYO7043.
REFERENCES Adams A. J. and Coletta N. J. (1984) Spatial extent of surround influence on flicker sensitivity. Inuesr. Ophfhai. visual Sci., suppl. 24 70.
Alexander K. R. (1974) Sensitization by annular surrounds: the effect of test stimulus size. Vision Rex 14, I 107-I 113. Alexander K. R. and Fishman G. A. (1984) Rod-cone interaction in flicker perimetry. Br. J. Ophthal. 68, 303-309.
Arden G. B. and Frumkes T. E. (1986) Stimulation of rods can increase cone flicker ERG’s in man. Vision Res. 26. In press. Bauer G. M., Frumkes T. E. and Nygaard R. W. (1983a) The signal-to-noise characteristics of rod+one interaction. J. Physiol. 337, 101-I 19.
Bauer G. M.. Frumkes T. E. and Nygaard R. W. (t983b) The influence of rod light and dark adaptation upon rodsone interaction. J. Physiol. 337, 121-135. Berger C. (1953) Area of retinal image and flicker fusion frequency. Acra physiol. stand. 28, 224-233. Berger C. (1954) Iliumination of surrounding field and flicker fusion frequency with fovea1 images of different sizes. Acfa physiol. stand. 30, 161-I 70. Buck S. L. (1980) Cones are desensitized by backgrounds they don’t see. Presented at: Topical Meeting on Recent Advances in Vision, Opt. Sot. Am.; abstr. in meeting Technical Digesi.
Buck S. L. (1981) Spatial patterns of cone-rod interaction and isolated rod and cone systems. Inpest. Op~~~a~.cisuai 5%. Suppi. M, 207.
Buck S. L. (1985a) Cone-rod interaction over time and space. Vision Res. 25, 907-916. Buck S. L. (1985b) Determinants of the spatial properties of cone-rod interaction. Vision Res. 25, 1277-1284. Cofetta N. J. and Adams A. J. (1984) Rodxone interaction in flicker detection. Vision Res. 24, 1333-1340. Coletta N. J., Adams A. J. and Schefrin B. E. (1984) Surround influence on flicker sensitivity: retinal mechanisms. Invest. Ophthal. visual Sci. Suppl. 25, 70. Coletta N. J. and Adams A. J. (1986) Adaptation of a color-opponent mechanism increases parafoveal sensitivity to luminance flicker. Vision Res. 26. In press. Crawford B. H. (1940) The effect of fieId size and pattern on the change of visual sensitivity with time. Proc. R. foe. B129, 94-106. Creed R. C. and Ruth T. C. (1932) Regional variations in sensitivity to flicker. J. Physiol. 74, 407432. Foley P. J. (1956) Can. J. Psychoi. IO, 200-206; as referenced in Foley (1961). Foley P. J. (1961) Inte~lationships of background area, target area and target luminance in their effect on the critical flicker frequency of the human fovea. J. opr. Sot. Am. 51, 737-740. Frumkes T. E., Eysteinsson T. and Arden G. B. (1985) Tonic inhibition of red-cone pathways by dark adapted rods. Imtesz. Op~~ha~.viwaI Sci.. Suppt. 26, 114. Frumkes T. E., Sekuler M. D., Barris M. C., Reiss E. H. and Chalupa L. M. (1973) Rod-cone interaction in human scotopic vision-I. Temporal analysis Vision Res. 13, 3269-1282.
Fry G. A. and Bartley S. H. (1936) The et&t of steady stimulation of one part of the retina upon the critical frequency in another. J. exp. Psychol. 19, 351-356. Goldberg S. H. and Frumkes T. E. (1983) A distal retinal locus for rod-cone interaction? Incest. Ophthai. visual Sci., Suppl. 24, 187.
Goldberg S. H., Frumkes T. E. and Nygaard R. W. (1983) Inhibitory influence of unstimulated rods in the human retina: evidence provided by examining cone flicker. Science 221, 180-182. Goldberg S. H., Frumkes T. E., Nygaard R. W. and Alexander K. R. (1982) Unstimulated rods attenuate cone flicker. J. opt. Sot. Am. 72, 1742. Haegerstrom-Portnoy G. and Adams A. J. (1983) Spatial sensitization properties of the blue-sensitive cone pathways. In Colour Yirion, Physioiegy and Psychophysics (Edited by Motton J. D. and Sharpe L. T.), Academic Press, New York. Harvey L. 0. (1970) Flicker sensitivity and apparent brightness as a function of surround luminance. J. opt. Sot. Am. 60, 860-864.
Size of lateral interactions for flicker Keesey U. T. (1972) Flicker and pattern detection: a comparison of thresholds. J. opt. Sot. Am. 62. 446448.
Kelly D. H.
(1959)
J. opt. Sot.
Effects of sharp edges in a flickering field.
Am. 49. 730-732.
Kulikowski J. J. and Tolhurst D. J. (1973) Psychophysical evidence for sustained and transient detectors in human vision. J. Phvsiol. 232, 149-162. Latch M. and Lennie P. (1977) Rod-cone interaction in light adaptation. J. Physiol. 231, 27-29P. Lythgoe R. J. and Tansley K. (1929) The relation of the critical frequency of flicker to the adaptation of the eye. Proc. R. Sot.. Land. IOSB, 60-92. Martinez J. M.. Sturr J. F. and Nowak L. M. (1977) Spatial interaction between rods and cones determine stcadystate sensitivity in near peripheral retina. Vision Res. 17, 687689.
Matin L. and Komheiser A. (1973) Creation of multiple channels from single channels by an annular surround. Inuesl. Ophthai. vbual Sci., Suppl.. 97. Matin L. and Komheiser A. (1976) Linked changes in spatial integration, size discrimination. and increment threshold with change in background diameter. Vition Res. 16, 847-860. McKee S. and Westheimer Cl. (1970) Specificity of cone
mechanisms
in lateral
925 interaction.
J.
Physiol.
206,
117-128.
Movshon J. A., Thompson 1. D. and Tolhurst D. J. (1978) Spatial and temporal contrast xnsitivity of neurones in areas I7 and I8 of the cat’s visual cortex. J. Physiol. 283. 101-120.
Stromeyer C. F. (1983) Spatial sensitization and desensitization with small adapting fields: interactions of signals from different classes of cones. Vision Res. 23, 621630. Stromeyer C. F. and Hill T. L. (1983) The cone threshold: spatial interactions of rod and cone adapting signals. Vision
Res. 23, 713-722.
Temme L. A. and Frumkes T. E. (1977) Rod-cone interaction in human scotopic vision--III. Vision Res. 17. 681-685.
Tolhurst D. J. (1973) Separate channels for the analysis of the shape and the movement of a moving visual stimulus. J. Physiol. 231, 385-402. Watheimer G. (1965) Spatial interaction in the human retina during scotopic vision. 1. Physiol. 181, 881-894. Westheimer G. (l%7) Spatial interaction in human cone vision. J. Physiol. 190, 139-154. Westheimer G. (1970) Rcdxone independence for sensitizing interaction in the human retina. J. Physiol. 206, 109-l 16.
Copyright C
0042-6989~86 $3.00 + 0.00 1986 Pergamon Journals Ltd
SPATIAL EXTENT OF ROD-CONE AND CONE-CONE INTERACTIONS FOR FLICKER DETECTION NANCY J. School
of
optometry.
and ANTHONYJ.
COLETTA*
ADAMS
University of California, Rerkeley, CA 94720, U.S.A
(Received 14 October 1985; in reviscdjiwm 30 December
1985)
Ah&act-Over a large range of tight adaptation levels, sensitivity to 25 Hz flicker improves as the light level of the background increases. Using small background discs and annular surrounds. this c&t was shown to be mediated by the surround and not the average luminance of the test region, in agreement with recent reports. The effect is due to two types of lateral interaction: at mesopic light levels (from 0.1 to 1.0 td), cone-mediated flicker resolution is enhanced by the stimulation of surrounding rods; at photopic light levels (above IO td), ficker sensitivity improves with light stimulation of adjacent cones. The spatial zone, or extent, over which the surround contributes to the flicker threshold was measured. The spatial area over which rods influence the cone 25 Hz Sicker threshold is larger than the analogous spatial area of cone intluence. In the parafovea, at 5 deg azentricity, cone ticker sensitivity for a 20’spot is infiuenced by cones in a 1 deg diameter area centered on the spot; the corresponding area of rod influence is about 3 deg. In the fovea, flicker sensitivity for a 10’ spot is influenced by cone stimulation in an area of about 20’ diameter. Rods which a&t foveal flicker sensitivity appear to occupy an annular zone with about a 2 deg outer diameter and 1 deg inner diameter, centered on the fovea. Flicker
Spatial se&i&on
Rod-cone
interaction
lNTRODUCl’lON
Sensitivity for the detection of a small spot centered on a background depends upon the spatial dimensions of the background disc (Crawford, 1940; Westheimer, 1965, 1967). As background diameter increases, sensitivity first decreases and then increases; this latter effect is known as spatial sensitization. There is some discrepancy in the literature about the specificity of this spatial interaction for single receptor mechanisms. The individual cone types are reported to have independent pathways for sensitization (McKee and Westheimer, 1970), but recent studies find evidence for spatial sensitization between the middle and long wavelength-sensitive cone types (Stromeyer, 1983) and the short wavelength-sensitive cones can be sensitized by other cone types (Haegerstrom-Portnoy and Adams, 1983). Similarly, an earlier study of the sensitization properties of rods and cones found that surrounding rod stimulation is unable to sensitize the cone threshold (Westheimer, 1970). More *Present address, to which correspondence should be sent: Center for Visual Science, University of Rochester, Rochester, NY 14627, U.S.A.
Lateral interaction
Light adaptation
recent studies, however, have demonstrated interactions between rods and cones in the sensitization effect. The cone increment threshold is increased by the presence of small rod-detected backgrounds (Frumkes ef al., 1973; ‘femme and Frumkes, 1977; Martinez er al., 1977; Latch and Lennie, 1977; Buck, 1980; Bauer et al., 1983a,b; Stromeyer and Hill, 1983; Buck, 1985a.b). Surrounding rod stimulation can also sensitize cone thresholds, because the cone threshold elevation in the presence of a small rod field can be reduced by expanding the rod background (Buck, 1980, 1981; Bauer et al., 1983a; Stromeyer and Hill, 1983; Buck, 1985a,b). A similar effect can be produced with annular surrounds. The cone threshold is elevated by the presence of a thin, rod-detected annulus placed close to the stimulus; as the annulus is moved away from the target, the cone threshold falls to the level obtained without an annulus (Buck, 1981; Buck, 1985b). The spatial interactions between rods and cones are complex, because the cone threshold elevation produced by a rod-detected background can also be reduced by the presence of a surrounding cone-detected annulus; also, the cone threshold elevation caused by a cone-detected field can be reduced
917
918
NANCY .I. COLETTAand ANTHONY J. ADAMS
by the presence of a surrounding rod annuius (Stromeyer and Hill, 1983). In the present study we examined the spatial properties for cone-detected sensitization flicker. Flicker sensitivity becomes degraded during the course of dark adaptation, or conversely, is enhanced by increasing the light adaptation level (Lythgoe and Tansley, 1929; Creed and Ruth, 1932; Fry and Bartiey, 1936; Berger, 1953, 1954; Foley, 1956, 1961; Kelly, 1959; Harvey, 1970). The adaptation effect on flicker is due, in part, to rod activity. As rods dark adapt, they suppress cone flicker sensitivity, as recently demonstrated in human psychophysical and ERG studies (Goldberg and Frumkes, 1983; Goldberg et at., 1982, 1983; Alexander and Fishman, 1984; Coietta and Adams, 1984; Arden and Frumkes, 1986) and in animal eiectrophysioiogicai studies (Frumkes er al., 1985). This rod-cone interaction for flicker has been demonstrated for small fovea1 targets: compared to the dark adapted condition, foveai flicker sensitivity is better in the presence of a rod-detected background. Since there are no rods at the fovea, this implies that the suppression of cone flicker sensitivity by dark adapted rods occurs over large retinal distances (Coietta and Adams, 1984; Adams and Coietta, 1984). Over a small range of photopic adaptation levels, there is an additional enhancement of flicker sensitivity as the background light level increases. Unlike the effect observed at mesopic levels, the spectral sensitivity of this effect resembles a cone mechanism (Coietta and Adams, 1984; Coietta er al., 1984). This suggests that light stimulation of cones can also improve cone flicker sensitivity. These adaptation effects on flicker sensitivity are qualitatively similar to the spatial sensitization effect for increment thresholds, because flicker sensitivity is improved as light is added to the region surrounding the target. It is true that, as the background intensity increases under a flickering target, the average luminance of the test target increases; therefore, one might invoke the Ferry-Porter law (i.e. flicker sensitivity improves as the target luminance increases) in order to explain the enhancement of Aicker sensitivity with background intensity. However, this idea does not explain the results obtained with annular surrounds. Flicker sensitivity is enhanced by increasing the intensity of an annuius centered on the target and this effect cannot be accounted for by stray light (Fry and
Bartiey, 1936; Berger, 1953. 1954; Foley, 1956; Harvey, 1970; Goldberg and Frumkes, 1983: Goldberg et al.. 1983; Adams and Coletta. 1984). Furthermore, flicker sensitivity increases as a background disc of fixed intensity is expanded (Foley, 1961: Goldberg and Frumkes, 1983; Alexander and Fishman, 1984; Adams and Coietta, 1984). The surrounding light level is an important factor in the adaptation effects on flicker sensitivity. In this study, the spatial extent of the rod-cone and cone-cone interactions for flicker were investigated by measuring cone flicker thresholds for small targets su~rimposed upon various sized background discs. Annular surrounds with increasing inner diameter were also used to measure the distance over which the lateral effects occur. METHODS
Stimuli The stimulus was a small, 620 nm, circular test target which flickered at 25 Hz. Previous test sensitivity results indicate that this target flicker is detected by cones (Coietta and Adams, 1984). The flickering target was presented for 750 msec every 2.25 set (1.5 set interstimulus interval). In the experiments at mesopic adaptation levels, four dim, red fixation dots arranged in a diamond were placed around the fovea1 target, i/2 deg from the target; for the parafoveai condition, a dim, red fixation spot was placed 5 deg away from the test target. The fixation targets for the photopic experiments were small black dots, arranged in the same fashion. The surrounding illumination was configured either as an annuius or a background disc. For the experiments with annular surrounds, a 500nm, steady, concentric annuius was superimposed on the target: the annuius outer diameter subtended 4 deg and the inner diameter was varied. The intensity of the annulus was also variable. Further details of the stimuli are given in the results section. For the sensitization paradigm, the 500nm, steady background discs were superimposed on the target, and the disc diameter was varied. Two adaptation levels were chosen for each eccentricity: one of the rod-mediated and one on the cone-mediated portion of the flicker threshold vs radiance curve for a 620 nm target on a 500 nm background (Coietta and Adams, 1984). In order to suppress any surrounding rod activ-
919
Size of lateral interactions for flicker
ity for the photopic sensitization experiment, a 10 deg diameter blue-green (Wratten No. 64) auxiliary field was used. The auxiliary field was 63 phot td (about 500 scat td) in the fovea and 10 phot td (about 80 scat td) in the parafovea. Apparatus The apparatus for these experiments was a two-channel Maxwellian view optical system, previously described in detail (Coletta and Adams, 1984). Variable background sizes were produced by changing the aperture size in the background channel. For studies with annular surrounds, the annuh were made on photographic slides and placed in the background channel. A third channel was added for these studies in order to provide a dim, red (Wratten No. 70) fixation light for the mesopic experiments, and a blue-green (Wratten No. 64) auxiliary surround to suppress rod activity for the photopic sensitization experiments. The source for this channel was an incandescent lamp of 40 W. The auxiliary field intensity was adjusted with Wratten No. 96 neutral density filters. Subjects Two experienced psychophysical observers with normal visual acuity, the authors, were used in these experiments. Subject N.C. was used for all the stimulus conditions; subject A.J.A., a mild deuteranomal, was used for the mesopic fovea1 condition only. Procedure At the beginning of a test session, the subject dark adapted for 10 min. This was shown to be an adequate dark adaptation period, because the flicker threshold on the dimmest background was the same after 25 min of dark adaptation as after 10 min (Coletta and Adams, 1984). The subject then adapted to the dimmest background or annulus for 2 min. At the end of this adaptation period, the test stimulus was presented and the subject set a flicker threshold by the method-of-adjustment. Flicker threshold is defined as the intensity of the test target needed to just perceive flicker. The flicker threshold is often far above the detection threshold, so the target may still be bright even when the flicker is no longer perceived. To avoid adaptation or habituation to the stimulus, the subject dimmed the target after making each flicker threshold setting, and so began the next setting from below the flicker threshold. At least
three readings were taken for each stimulus condition. In order to examine the effect of surround adaptation level, the intensity of the annuli was varied. After the readings were taken for a given adaptation level, the subject adapted to the next highest surround intensity level for 2 min. The size of the annulus was kept constant during each testing session. The experiments with expanding background discs were done at two intensity levels. The surround intensity level remained constant during each testing session while the diameter of the disc was expanded. Readings were taken with the smallest diameter first; thresholds were measured on successively larger disc diameters. RESULTS
Fovea1 spatial extent: (i) annular surrounds The flicker threshold vs radiance (tvr) curve for a 10’ diameter spot on a 4 deg diameter background disc is shown in Fig. 1 (squares). The curve has two negative sloping portions. At the dim backgrounds, over the 1 log unit range between -0.7 and 0.3 log td (0.2-2 phot td; 1.616 scat td), there is a shallow negative slope. Our previous studies of the cone-detected flicker tvr curves revealed a rod action spectrum at these dim adaptation levels (Coletta and Adams, 1984). The shallow slope indicates that rod adaptation produces a gradual enhancement of flicker sensitivity. Over a range of photopic adaptation levels, between 1.8 and 2.3 log td (63-200 phot td), there is a steep negative slope. We have previously shown that this effect is cone-mediated. The flicker tvr curve measured with an annulus (4 deg outer diameter/ 17’ inner diameter) surrounding the fovea1 target is shown in Fig. I (circles). The data is virtually identical to the result with the background disc. One might attribute this coincidence of the data sets to stray light from the annulus failing on the target area. However, for the 200 phot td condition, the amount of stray light falling on the center of the annulus is not equal to 200 td; thus the stray light “pedestal” under the target is not equivalent to the pedestal provided by the 200 td disc. Yet for these two conditions, the threshold is the same. Additionally, the flicker threshold on a 200 td, 10’ diameter background disc (triangle in Fig. 1) is about 1 log unit higher than the thresholds on either the large disc or annulus. Thus, the light pedestal under the target con-
920
NANCYJ.
COLETTA
and
ANTHONY
J.
ADAMS
2
1
0
Log
retlnol tlluminonce
( phot
td)
Fig. 1. Fovea1 flicker threshold for a 10’ arc, 620 nm, 25 HZ target as a function of background or annulus retinal illuminance. Squares represent results for a 4 deg diameter, 500 nm disc; circles. 4 deg outer/l7’ inner dia 500 nm annulus. Triangle represents flicker threshold on a 200 phot td, IO’diameter 500 nm disc. Subject N.C.
tributes little to the flicker threshold, since the pedestals for the large and small background discs are equivalent. This is consistent with Berger’s result (1954) showing that a small steady light added to a flickering test patch lowered rather than increased the CFF; he concluded that stray light falling on the target could not account for the enhancement of flicker sensitivity by an annulus. Goldberg and Frumkes (1983) and Goldberg et al. (1983) also show that a large disc or contiguous annulus can produce the same flicker threshold. The flicker tvr curves for various annulus inner diameters are shown in Fig. 2. Note there is little change in the rod-mediated effect as the inner annular diameter is increased, up to
Log retlnol
1.5 deg. This would be consistent with retinal anatomy, which shows a rod-free area over approximately the central l-l.5 deg diameter. With larger inner diameters, 2.4 and 3.6 deg, the shallow negative slope is absent, indicating no rod effect. This shows that the rods responsible for the effect on fovea1 cone flicker thresholds are located in an annular zone, just outside the rod-free area, but they do not extend beyond about 2 degrees diameter. With the annular surrounds, a steep negative tvr slope at the bright backgrounds is still evident for inner annular diameters of 17’ arc. As the inner dimension increases, this steep, cone-mediated effect gradually disappears. This shows that the cone effect at photopic levels is
lllumlnonce
(phot Id)
Fig. 2. Fovea1 flicker threshold as a function of 500 nm annulus retinal illuminance. Symbols denote annulus inner diameter; outer diameter was 4 deg. Same test stimulus and subject as in Fig. 1.
921
Size of lateral interactions for flicker
ulation in an annular zone from I to 2 deg, centered on the fovea. The spatial sensitization function for the brighter, cone background is shown in the lower part of the figure. The flicker threshold drops to the minimum at about 20’ diameter. Thus, the threshold-lowering effect is mediated by cones within the central 20‘ diameter. Parafiveal spatial extent: (i) annular surrounds
Diam (min) 40
‘,
-i
’
(20
60
‘,
’
6
240
,’ 1
Log retinal araa I deg z 1
Fig. 3. Fovea1 flicker threshold as a function of background disc area for a 10‘ arc, 25 Hz, 620 nm target. Background wavelength was 500 nm. Vertical bars are the range of three readings. Solid circles depict results on 0.63 phot td background; open circle is dark adapted flicker threshold. Solid squares, results on 200 phot td background surrounded by a 63 phot td auxiliary field; open square, flicker ~re~old on 63 td auxiliary field atone. All data for subject N.C. except triangies, which depict data for subject A.J.A. on 0.63 phot td background.
mediated by stimulation within a small area surrounding the target, within the central 0.5 deg diameter.
In the parafovea, we found separate rod and cone adaptation effects on the flicker threshold, similar to those in the fovea (Coletta and Adams, 1984). In this study, we determined whether the parafoveal adaptation effects on flicker were larger in spatial extent than the fovea1 effects. Like the fovea1 results, the flicker tvr curve in the parsfovea has a shallow sloping portion for dim back~ounds below 2 phot td; this part of the curve has a rod action spectrum (see Fig. 1 of Coletta and Adams, 1984). From about 6 to 63 td, there is a steeper negative slope in the tvr curve, and the action spectrum over this portion of the curve resembles a cone opponent function (see Coletta and Adams, 1984; Coletta et al., 1984; Coletta and Adams, 1986).
I
A
Fovea1spatialextent: (ii) expanding backgrounds We measured the flicker threshold as a function of back~ound disc diameter at two adaptation levels, in order to determine the spatial extent of the separate rod- and cone-mediated effects. These measures confirm the results with annuli, and employ the more familiar spatial sensitization paradigm (Westheimer, 1965). The flicker threshold for the 10’ diameter fovea1 test was measured on a variable diameter 0.63 td background, which is on the rodmediated limb of the flicker tvr curve. The data is shown in the upper portion of Fig. 3. Note that the flicker threshold is unaffected until the background diameter reaches about 1 deg. From this point, up to about 2 deg, the flicker threshold drops by about 0.2 log unit. Although this effect is small, the amount agrees with the tvr data for this background. Since the action spectrum is that of rods, the th~shold-iowe~ng effect of the background is caused by rod stim-
Lag onnulus
innerarea Meg *
)
Fig. 4. Parafoveal Sicker threshold as a function of annulus inner area for a 20’ arc, 620 nm. 25 HE target at 5 deg ccc. Outer diameter was 4 dcg, except for data denoted by diamond symbol, which was for an annulus with outer diam of 7 deg/inner diameter 4.7 deg. Solid circles are results for a 2 td annulu% solid squares, results for 20 td annulus. Open triangle depicts dark adapted flicker threshold; open circle, flicker threshold on a 2 td, 4 deg disc; open square, flicker threshold on a 20 td, 4 deg disc. All data for subject N.C.
922
NANCY J.
COLETTA
and
In the parafovea, at 5 deg eccentricity, the effect of annular inner diameter was examined for two surround intensities; one was on the shallow, rod slope of the flicker tvr curve, at 2 phot td (16 scat td), and the other one was at a cone adaptation level, 20 phot td. The results, measured with a 20’ diameter test spot, are plotted as a function of annulus inner diameter in Fig. 4. The flicker threshold on the dim background is about 0.5 log unit lower than the dark adapted flicker threshold. The dim annulus with 17’ inner diameter produces the same threshold as the 4 deg disc, but as the inner diameter is enlarged, the flicker threshold gradually rises, nearly to the dark adapted level. This means that, unlike the fovea1 results, the shallow rod-mediated slope of the curves gradually disappears as the annular inner diameter increases. In the parafovea, rods are present at the target location, so the amount of rod influence could be affected by the small inner diameters. To determine whether the rod effect would still be present for a larger inner diameter if the overall annulus area were increased, the flicker threshold was measured with an annulus of 7 deg outer/4.7 deg inner diameter. The flicker threshold remains elevated for this condition, suggesting the parafoveal rod influence on the cone flicker threshold only extends over a diameter of about 4 deg. At the higher adaptation level (20 td), the cone-mediated effect is present for the 17’ inner diameter and lessened for the 36’ and 43’ inner diameters, but disappears for an inner diameter of I .l deg and larger. This indicates that the photopic effect on flicker sensitivity is mediated by cones within about a 1 deg diameter area, centered on the target.
ANTHONY J. ADAMS
within about a 3.5 deg area centered on the target. The results for the photopic background are shown in the lower portion of Fig. 5. The sensitization curve levels off at about 1 deg diameter; thus the surround influence is mediated by cones within 1 deg of the target. This is consistent with the results with annular surrounds. DISCUSSION
The present results indicate that the stimulation of either surrounding rods or cones can enhance cone-detected flicker sensitivity. Further, the results suggest that at adaptation levels dimmer than 2 phot td, fovea1 cone-detected flicker is influenced by rods located within an annular zone around the fovea. This zone has an outer diameter of about 2 deg and the inner diameter is the size of the rod-free area, about 1 deg in width. Parafoveal cone flicker, at 5 deg eccentricity, is influenced by rods within a 3.5 deg diameter area, centered on the target. Over a certain range of photopic adaptation, there is a further enhancement of flicker sensitivity due to the stimulation of surrounding cones. In the fovea, this sensitization area is about 20’ in diameter. At 5 deg eccentricity, the sensitization area is about 1 deg in diameter.
P c.
Parafoveal spatial extent: (ii) expanding backgrounds
The parafoveal flicker tvr curve is similar to the fovea1 curve over the shallow, rod-mediated slope, but the steep cone slope occurs at a dimmer adaptation level in the parafovea than in the fovea. Again, the spatial sensitization effects were examined for both the rod- and cone-mediated adaptation levels. The spatial sensitization results for the 20’ diameter target on the 0.63 td, rod-mediated background are shown in the upper curves of Fig. 5. Note the curve levels off at 34 deg diameter, suggesting the rod influence on cone-detected flicker in the parafovea is due to the stimulation of rods
\ 20 td
\, Dlam -II
(mln
)
20
10 ’
40
60
120
240
1 0
I -1 Lag
7
retinal
area
I 1 (deg’
I
Fig. 5. Parafoveal flicker threshold as a function of background disc area for 20’ arc, 25 Hz, 620 nm target at 5 deg eccentricity. Upper curve for 0.63 phot td disc; lower curve for 20 phot td disc, surrounded by a 10 td auxiliary field. Open triangle depicts dark adapted flicker threshold; open square depicts Bicker threshold on IO td field alone. Subject N.C.
Size of lateral interactions for flicker
Possible edge efects The influence of the surround on flicker sensitivity cannot be adequately explained as an effect due to edge contours. On a qualitative basis, the threshold is higher when the edge of a disc is close to the target, but for annuli the threshold is lower when the annulus inner diameter is nearer to the target (Figs 1 and 2). The disc condition presents a lower contrast edge because the disc is superimposed on the auxiliary field, yet this lower contrast edge condition produces a higher flicker threshold. Size of the rod and cone surround efects The cone sensitization areas shown in the present results are slightly larger than those previously reported for increment threshold studies. Westheimer (1967) reported cone sensitization areas of 14.2’ diameter at the fovea and 35.7’ at 5 deg eccentricity. Our corresponding values for flicker are 20’ and 60’ diameter, respectively. This discrepancy is probably not due to the use of larger test targets in our study. For increment thresholds, large test probes yield a larger densensitizing zone, and a smaller magnitude of the sensitization effect, but do not alter the overall size of the sensitization area (Alexander, 1974; Matin and Komheiser, 1973; Matin and Komheiser, 1976). Our present results suggest that the sensitization zones for flicker are larger than the zones for increment detection; this implies that channels sensitive to temporal transients involve larger receptive fields or size-tuning. This would be consistent with neurophysiological studies in cat (Movshon et al., 1978) and human psychophysical studies (Keesey, 1972; Kulikowski and Tolhurst, 1973; Tolhurst, 1973) which show that the class of pathways sensitive to higher temporal frequencies tend to be tuned to lower spatial frequencies. The present results also show that the rodmediated effect on cone flicker extends over a larger area than the cone-mediated effect. This would be consistent with the model for rod-cone interaction proposed by Bauer et al. (1983a), in which signals from a rod channel with its own inhibitory surround interact with cone signals at a subsequent detector, which sums inputs from both rod and cone channels. Since rod sensitization areas are larger than those for cones (Westheimer, 1965, 1967), one might expect a large spatial extent for the rod influence on cones. However, caution must be
923
taken in comparing flicker sensitization results with a model designed for increment thresholds. As discussed in the following section, there are many discrepancies between the sensitization properties for flicker and increment thresholds. Comparison tization
with increment
threshold sensi-
The superficial similarity between cone flicker and increment threshold sensitization is that, in each case, an increase in the diameter of a superimposed background produces an increase in sensitivity. A major discrepancy, however, is that, for increment thresholds, the increase in sensitivity is relative, whereas for flicker thresholds, it is absolute. Unlike the spatial sensitization functions for increment thresholds, there is no desensitizing limb for the flicker threshold. At the light levels used in the present experiment, the addition of a background smaller than the target has little influence on the flicker threshold, or even decreases the flicker threshold below the dark adapted level (see Fig. 5). The sensitization properties for flicker also cannot be modelled by the behavior of the flicker tvr curve. In the more familiar case of increment threshold sensitization, adding a surround to a small background disc could reduce the effective stimulation from the center of the background disc; according to the increment threshold vs intensity curve, this decrease in effective light stimulation should lower the increment threshold and produce sensitization. In the case of the flicker, however, the flicker threshold vs intensity curve has a negative slope; i.e. the flicker threshold falls as light stimulation increases. If the addition of a surround reduces the effective stimulation from the central part of the background disc, this should drive the flicker threshold up instead of down. The mechanism for flicker sensitization shown in our results, therefore, is probably not the reduction of a saturating influence from the central part of the background, as proposed for increment thresholds. Sensitization for the increment threshold may involve an indirect effect of the surround upon the detection mechanism, by reducing the effectiveness of the center stimulation in raising the increment threshold. The lateral interaction for flicker, however, could be a direct action of the surround stimulation upon the sensitivity of the flicker detection mechanism. Two other aspects of the rod influence on
NANCY J. COLE~A anId ANTHONYJ. AUAMS
924
cone flicker are inconsistent with rod-cone interaction for cone increment thresholds. First, the cone absolute threshold for detecting small red flashes (presented alone, without a rod background) is relatively unaffected by the state of rod dark adaptation (Bauer et al., 1983b). Rod dark adaptation, however, has a substantial effect on the cone flicker threshold (Goldberg et al., 1983; Coletta and Adams, 1984; Alexander and Fishman, 1984). Secondly, since there are no rods at the fovea, the rod surround effect for fovea1 flicker cannot be modelled with a rod channel having a center-surround organization superimposed on the target region, as proposed for increment thresholds (Bauer et al., 1983a). According to recent el~trophysio~ogica~ and psychophysical reports, the surrounding rod influence on cone flicker occurs at a distal site in the retina, either at the receptor level or outer plexiform layer (Alexander and Fishman, 1984; Arden and Frumkes, 1986; Coletta and Adams, 1984; Frumkes et al., 1985). A direct influence of the surround on the flicker detection mechanism, as discussed above, is consistent with recent evidence that rods influence cone flicker directly through horizontal cell feedback onto cones (Frumkes et al., 1985). The cone-mediated effect at photopic levels may be a similar mechanism, but the presence of an opponent mechanism in the parafoveal photopic sensitization (Coletta and Adams, 1984; Coletta et al., 1984; Coletta and Adams, 1986) suggests that more proximal sites are involved in the cone-cone interaction for flicker.
Acknowledgemenrs-This work was supported by NIH grant EY02271 to A.J.A., and by training grant support to N.J.C.. on EYO7043.
REFERENCES Adams A. J. and Coletta N. J. (1984) Spatial extent of surround influence on flicker sensitivity. Inuesr. Ophfhai. visual Sci., suppl. 24 70.
Alexander K. R. (1974) Sensitization by annular surrounds: the effect of test stimulus size. Vision Rex 14, I 107-I 113. Alexander K. R. and Fishman G. A. (1984) Rod-cone interaction in flicker perimetry. Br. J. Ophthal. 68, 303-309.
Arden G. B. and Frumkes T. E. (1986) Stimulation of rods can increase cone flicker ERG’s in man. Vision Res. 26. In press. Bauer G. M., Frumkes T. E. and Nygaard R. W. (1983a) The signal-to-noise characteristics of rod+one interaction. J. Physiol. 337, 101-I 19.
Bauer G. M.. Frumkes T. E. and Nygaard R. W. (t983b) The influence of rod light and dark adaptation upon rodsone interaction. J. Physiol. 337, 121-135. Berger C. (1953) Area of retinal image and flicker fusion frequency. Acra physiol. stand. 28, 224-233. Berger C. (1954) Iliumination of surrounding field and flicker fusion frequency with fovea1 images of different sizes. Acfa physiol. stand. 30, 161-I 70. Buck S. L. (1980) Cones are desensitized by backgrounds they don’t see. Presented at: Topical Meeting on Recent Advances in Vision, Opt. Sot. Am.; abstr. in meeting Technical Digesi.
Buck S. L. (1981) Spatial patterns of cone-rod interaction and isolated rod and cone systems. Inpest. Op~~~a~.cisuai 5%. Suppi. M, 207.
Buck S. L. (1985a) Cone-rod interaction over time and space. Vision Res. 25, 907-916. Buck S. L. (1985b) Determinants of the spatial properties of cone-rod interaction. Vision Res. 25, 1277-1284. Cofetta N. J. and Adams A. J. (1984) Rodxone interaction in flicker detection. Vision Res. 24, 1333-1340. Coletta N. J., Adams A. J. and Schefrin B. E. (1984) Surround influence on flicker sensitivity: retinal mechanisms. Invest. Ophthal. visual Sci. Suppl. 25, 70. Coletta N. J. and Adams A. J. (1986) Adaptation of a color-opponent mechanism increases parafoveal sensitivity to luminance flicker. Vision Res. 26. In press. Crawford B. H. (1940) The effect of fieId size and pattern on the change of visual sensitivity with time. Proc. R. foe. B129, 94-106. Creed R. C. and Ruth T. C. (1932) Regional variations in sensitivity to flicker. J. Physiol. 74, 407432. Foley P. J. (1956) Can. J. Psychoi. IO, 200-206; as referenced in Foley (1961). Foley P. J. (1961) Inte~lationships of background area, target area and target luminance in their effect on the critical flicker frequency of the human fovea. J. opr. Sot. Am. 51, 737-740. Frumkes T. E., Eysteinsson T. and Arden G. B. (1985) Tonic inhibition of red-cone pathways by dark adapted rods. Imtesz. Op~~ha~.viwaI Sci.. Suppt. 26, 114. Frumkes T. E., Sekuler M. D., Barris M. C., Reiss E. H. and Chalupa L. M. (1973) Rod-cone interaction in human scotopic vision-I. Temporal analysis Vision Res. 13, 3269-1282.
Fry G. A. and Bartley S. H. (1936) The et&t of steady stimulation of one part of the retina upon the critical frequency in another. J. exp. Psychol. 19, 351-356. Goldberg S. H. and Frumkes T. E. (1983) A distal retinal locus for rod-cone interaction? Incest. Ophthai. visual Sci., Suppl. 24, 187.
Goldberg S. H., Frumkes T. E. and Nygaard R. W. (1983) Inhibitory influence of unstimulated rods in the human retina: evidence provided by examining cone flicker. Science 221, 180-182. Goldberg S. H., Frumkes T. E., Nygaard R. W. and Alexander K. R. (1982) Unstimulated rods attenuate cone flicker. J. opt. Sot. Am. 72, 1742. Haegerstrom-Portnoy G. and Adams A. J. (1983) Spatial sensitization properties of the blue-sensitive cone pathways. In Colour Yirion, Physioiegy and Psychophysics (Edited by Motton J. D. and Sharpe L. T.), Academic Press, New York. Harvey L. 0. (1970) Flicker sensitivity and apparent brightness as a function of surround luminance. J. opt. Sot. Am. 60, 860-864.
Size of lateral interactions for flicker Keesey U. T. (1972) Flicker and pattern detection: a comparison of thresholds. J. opt. Sot. Am. 62. 446448.
Kelly D. H.
(1959)
J. opt. Sot.
Effects of sharp edges in a flickering field.
Am. 49. 730-732.
Kulikowski J. J. and Tolhurst D. J. (1973) Psychophysical evidence for sustained and transient detectors in human vision. J. Phvsiol. 232, 149-162. Latch M. and Lennie P. (1977) Rod-cone interaction in light adaptation. J. Physiol. 231, 27-29P. Lythgoe R. J. and Tansley K. (1929) The relation of the critical frequency of flicker to the adaptation of the eye. Proc. R. Sot.. Land. IOSB, 60-92. Martinez J. M.. Sturr J. F. and Nowak L. M. (1977) Spatial interaction between rods and cones determine stcadystate sensitivity in near peripheral retina. Vision Res. 17, 687689.
Matin L. and Komheiser A. (1973) Creation of multiple channels from single channels by an annular surround. Inuesl. Ophthai. vbual Sci., Suppl.. 97. Matin L. and Komheiser A. (1976) Linked changes in spatial integration, size discrimination. and increment threshold with change in background diameter. Vition Res. 16, 847-860. McKee S. and Westheimer Cl. (1970) Specificity of cone
mechanisms
in lateral
925 interaction.
J.
Physiol.
206,
117-128.
Movshon J. A., Thompson 1. D. and Tolhurst D. J. (1978) Spatial and temporal contrast xnsitivity of neurones in areas I7 and I8 of the cat’s visual cortex. J. Physiol. 283. 101-120.
Stromeyer C. F. (1983) Spatial sensitization and desensitization with small adapting fields: interactions of signals from different classes of cones. Vision Res. 23, 621630. Stromeyer C. F. and Hill T. L. (1983) The cone threshold: spatial interactions of rod and cone adapting signals. Vision
Res. 23, 713-722.
Temme L. A. and Frumkes T. E. (1977) Rod-cone interaction in human scotopic vision--III. Vision Res. 17. 681-685.
Tolhurst D. J. (1973) Separate channels for the analysis of the shape and the movement of a moving visual stimulus. J. Physiol. 231, 385-402. Watheimer G. (1965) Spatial interaction in the human retina during scotopic vision. 1. Physiol. 181, 881-894. Westheimer G. (l%7) Spatial interaction in human cone vision. J. Physiol. 190, 139-154. Westheimer G. (1970) Rcdxone independence for sensitizing interaction in the human retina. J. Physiol. 206, 109-l 16.