- Email: [email protected]
The influence of cone adaptation upon rod mediated flicker
PART
A
Biochemistry and Molecular Biology Retinal Physiology,
Cell Biology, Neurotransmitters,
Central Nervous System Physiology
Morphology
and Morphology
001’.6989,565340+ O.;H)
VisionRer. Vol. 26. No. 8, pp. I 167-l176.1986
Pergamon Journals Ltd
Printed in Great Britain
THE INFLUENCE OF CONE ADAPTATION UPON ROD MEDIATED FLICKER THOMAS E. FRUMKES, FRAXK NAAREXDORP and STUART H. GOLDBERG Department of Psychology, Queens College of CUNY, Flushing, NY 11367, U.S.A. (Received 7 Notember
1985; in revised form 3 March 1986)
A&tract--The influence of annular fields on sensitivity to sinusoidal flicker was assessed in the dark adapted parafoveai retina. Test stimuli were 2”20’ in diameter; annuli had a 2”20’ inner and 7’30’ outer
diameter. Rod flicker was studied with a “green” stimulus too dim to influence cones. Selective cone flicker was obtained using red and green flicker in counterphase and yoked together in modulation depth and scotopic illuminance. Results showed the following. (1) Annular stimulation of rods slightly facilitated rod-mediated flicker sensitivity to frequencies < 10 Hz. In contrast, annular stimulation of cones greatly facilitated rod-mediated sensitivity, particularly for flicker frequencies >7 Hz. We designate this effect, cone-rod interaclion. (2) Annular stimulation of cones has a negligible influence upon sensitivity to cone-mediated flicker frequencies < 15Hz. In contrast, annular stimulation of rods has a large influence upon sensitivity to cone-mediated flicker, an effect we designate rod-cone infepxrion. (3) Within limits, both rod-cone and cone-rod interaction increase as the annular illuminance increases and as flicker. frequency increases; the limiting frequency and illuminance values, however, are different for the two forms of interaction. Results are compared with prior evidence that rod and cone signals summate to produce an absolute threshold or flicker sensation. We suggest that there are at least three mechanisms for interaction between rod- and cone-related signals. Rods
Cones
Rod-cone interaction
Flicker
Retina
INTRODUCTION
A growing body of evidence indicates that visual signals stemming from the rods and cones of the vertebrate eye interact with one another. Prior psychophysical investigations using flicker have indicated two different underIying mechanisms. First, MacLeod (1972) and van den Berg and Spekreijse (1977) showed that separate flicker signals stemming from rods and cones can summate together, presumably by converging at some common locus in the retina. The retative phase of the rod- and cone-flicker signals, which is influenced both by experimental manipulation as well as the more sluggish response properties of rods, determines whether the separate signals enhance or destructively interfere with each other. This approach and interpretation was also used by Rodieck and Rushton (1976) to interpret flicker responses recorded extraceIlulariIy from cat ganglion cells. Difference in rod vs cone latency also determines whether subliminal rod and cone signals will add together to produce a threshold sensation (Frumkes er al., 1973). More recent investigations have described a “new” type of rod-cone interaction (Alexander
Psychophysics
and Fishman, 1984; Arden, 1985; Coletta and Adams, 1984; Goldberg et af., 1983; but although see also Lythgoe and Tansley, 1929). Cone-mediated flicker sensitivity, especially for frequencies greater than 15 Hz, is depressed when rods are dark adapted and greatly enhanced by selective rod light adap~tion, a phenomenon which we have referred to as suppressive rod-cone interaction (Frumkes and 1986; Frumkes, Naarendorp, Eysteinsson, Eysteinsson, Denny and Goldberg, 1986). This effect can be shown with cornea1 ERGS in humans and other mammals (Arden and Frumkes, 1986; Loew and Arden, 1985) and by means of intracellular recording in distal retinal neurons of amphibians including rhe cones themselves (Frumkes and Eysteinsson, 1986). Suppressive rod-cone interaction is also characteristically abnormal in some individuals with circumscribed retinal pathology (Alexander and Fishman, 1985; Arden and Hogg, 1985) and in some individuals with protanomalous, hereditary color deficiencies (Alexander and Fishman, 1983; Coletta and Adams, 1985; Goldberg and Frumkes, 1983). On the basis of these data, we proposed that the neural circuitry is within the distal most retina and that the phenomenon
1167
THCMAS E. FKLWES
II68
reflects rod modulation of horizontal cell feedback onto con2s (Frumkes el ai.. 1986). This explanation setms rsasonablr sincr feedback of horizontal ~211s upon cones has besn shown in an 2wr growing number of species {e.g. ~22 Stzrling. 1983), and in all species studied by ph.vsioio,oical techniques. horizontal cells receiving cone-input have been shown to additiona!ly receive some rod-input. Since there is no evidence for a horizontat ceil feedback mechanism onto rods. we reason2d that there should be no influence of cone-adaptation upon rod fficker. In spit2 of the apparent absence of relevant neural circuitry, the present results show that the adapted state of cones in neighboring retinal areas exerts a striking influence upon rodmediated flicker. Although this phenomenon bears superficial resemblances to other psychophysical effects, most particularly suppressive rod-con2 interaction, its illuminance and frequency limitations are sufficiently unique to suggest it is a fundamentally distinct type of cone--rod interaction. This paper has been presented previously in abstract form {Goldberg ef rri.. 1985).
METHODS
All data were collected with a 5 channel ~ax~vellian view optical system described previously (Frumkes et al., 1986; Nygaard and Frumkss, 1985). In brief. 4 channels stemmed from ssparate light emitting diod2 (LED) sources. One of these was a “green” LED with a peak navelength at 545 nm (Stanley Electric ESBG5503) and served as the flickering source for the majority of our experiments; the other rhre2 stemmed from “red-orange’. LEDs with peak wavelength at 630nm (General Instruments MV5152) and were used for fixation or control purposes. The temporal waveforms of these sources were controlled by the circuit used by Frumkes and Eysteinsson (1986, Fig. l), and monitored online with phototransistors (see Nygaard and Frumkes, 1982b). At the maximum depth of flicker modulation used, 96%, total distortion was less than 2% and was less with lower modulation depths. Average illuminante lev2l was controlled by means of neutral density wedges and fixed filters. The remaining which provided adaptation stimuli, beam, stemmed from a conventional tungsten source;
:‘i .:.
stimulus wavelength and iliuminance was controlled by means of neutral density wedges or filters and interference filters with half bandwidths of 6-12 nm. The spatial position, size, and shape of all stimuli was determined by placing appropriate targets in collimated beams. The photopic illuminance of all stimuli was calibrated according to the procedure of Nygaard and Frumkes (1982a). Scotopic illuminance was calculated by comparing the relative sensitivity of the CIE photopic observer to various wavelengths (See Wysecki and Stiles, 1967): these calculated scotopic values were occasionally checked according to the procedurrs of Frumkes and Temme (1977).
The final display was presented to the obs2rv2r.s eye in .!vfaxwellian view. This consisted of three stimuli. The first was a central red fixation target. The second was a 2”20’ diamzter. sinusoidally flickering test stimulus presented in the stimulus field 7” to the right of fixation. For the majority of experiments, this was genrrated by a green LED and had a modulation depth of 96%. The third stimulus ~vas an ;tnnular adapting field concentric to the flickering test: the inner diameter was contiguous with the test and the outer diameter was 7 30’. In most experiments, the annular adapting field was 680 nm in wavelength and 0 log phot. td (-2 log Scot. td) in illuminance. This annulus gvas the most effective means we coutd devise for light adapting cones and still preserve sensitivity in nearby rods. Illustrated results are from the left eye nasal field of F.N. (an author) which is emmetropic and has normal color vision. All illustrated experiments were also repeated with at least one of four other observers: these included the other t\vo authors and tsvo other laboratory personnel. For these observers, however, stimuli were prss2nted to the right eye temporal field. All observsrs were well informed regarding the purpos2 and procedure used for all experiments. All observers were dark adapted 25 min prior to data collection.
A sinusoidally flickering stimuius with constant spatial properties can be described by three parameters: its average illuminance, its modulation depth, and its frequency. In any threshold experiment. tivo of these parameters
InHuencc of ionc adaptatton
are fixed by the experimenter and the obserler varies the third to threshold which thus serves as index of flicker sensitivity. In the course of pilot ex~~mentation, we found each of these dependent variables to provide unique advantages and, hence, adopted the use of all three. For the data shown in Figs 1-4, the observer varied the frequency of a flickering stimulus of constant wavelength, illuminance, and modulation depth (96%) until Aicker could just be discerned (i.e. determined the critical flicker frequency or CFF). For the data shown in Fig. 5, the observer increased the modulation depth of a flickering stimulus of constant wavelength, illuminance, and frequency until flicker could just be perceived. Finally, for the data shown in Figs 6 and 7, the observer increased the illuminance of stimuli of fixed modulation depth (96%) and frequency until flicker could be perceived. Hence in all cases, thresholds were obtained by means of an ascending method of adjustment. The data plotted below resulted from observations in a minimum of two experimental sessions and in each session, each threshaid was obtained at least three times. As much as possible, we randomized the order of stimulus presentations within an experimental session. With the exception of very few data points, these psychophysical judgements were easy to obtain and standard errors are usually about the size of a plotted datum. Most of our conclusions were also checked in several sessions involving the use of a staircase or temporal forced choice procedure.
z
E
it69
upon fficksr RESULTS
Figure 1 illustrates typical results when critical flicker frequency (CFF) was determined as a function of the illuminance of a stimulus stemming from a green LED. Results plotted with closed circles were obtained with no background field. CFF increases as illuminance increases and only for the purposes of initial description, are fit by inspection rvith three straight fines. According to classical interpretation (e.g. Hecht and Smith. 1936), the initial segment is attributable to the functioning of rods, the horizontal segment represents the rod plateau, and the upper segment is attributable to cones. We refer to the straight lines with a positive slope as Ferry-Porter functions (e.g. Ferry, 1892; Porter, 1902). The data plotted with open circles are results obtained in the presence of a 680 nm annulus of 1 phot. td. illuminance. In respect to the control data, this annulus shifts the rod Ferry-Porter iunction upward but does not alter its slope, it shifts the “rod-plateau” upward, but has no influence on the cone Ferry-Porter function. Similar rest&s were also obtained with the four other observers used. Annular fields influence the sensitivity of both photopic (e.g. Fry and Bartley, 1936; Keesey, 1970) and scotopic (Nygaard and Frumkes, 1985) flicker. Therefore, before preceding to other stimuius manipulations, we deemed it 5rst necessary to establish that the annular back-
15-
2 u
10 -
SI -2
! .
-1
Green flicrte’r illumination (log
I 0
I 1
sot. rd)
Fig. 1. Critical flicker frequency (CFF) as a function of illuminance for sinusoidal Ricker of 96% modulation depth stemming from a green LED. Data were obtained with no annular field (solid circles) or in the presence of a 680 nm annulus of 1 phot. td illuminance (open circles).
1170
Tmms
kk ”
d l
E.
l
/
-;f
Fig. 2. CFF as a function of’ illuminance and of the wavelength of annular fields of the indicated wavelengths but equated for rods at 0.01 Scot. td illuminance.
ground influence which alters the “rod Ferry-Porter” function involves a conemediated annular influence on rod-mediated flicker. We established this with three sorts of control experiments. In the first of these, which is illustrated in Fig. 2, we again determined CFF as a function of ihuminance but in the presence of 512, 620, and 680 nm backgrounds equated for their influence upon rods at - 2 log Scot. td. Because of the difference in spectral sensitivity of rods and cones, the 620 and particularly the 512 nm background fields have a much smaller influence upon cones (i.e. the neurotic illuminance of the 512, 620 and 680 nm fiefds were -2.7. -0.7 and 0 iog td respectively). The open “....
~-.-._----
Wygaard and Frumkes (1985, their Fig. 5) determined the influence of annular fields of 560 and 635 nm of variable scotopic illuminance, upon the sensitivity of green-LED generated flicker of - 1.3 log scat. td (0.05 scat. td) iiluminance; they used spatial parameters quite similar to those in the present experiments. As reported in counterpart photopic experiments (i.e. Fry and Bartiey, 1936; Keesey. 1970). increasing the annulus iiluminance at first increases, then decreases flicker sensitivity. For the two frequencies used, enhancement depended upon scotopic illuminance and not on wavelength. The maximal change in sensitivity observed was about 2% for I Hz flicker and 10% for 7 Hz flicker, which would roughly cotrrespond to a change in CFF of about I Hz. This influence of rod stimLcioring annuli upon rod-derectedjlicker is, hence, quite comparable to that observed in Fig, 2 with a - 1.5 log scat. td test probe and in Fig. 5 with a test probe of 0.03 scat. td and flicker frequencies 6 4 Hz. They never. examined the influence of such annuli with other frequencies or other average illuminance values and, hence, never observed a cone-annular influence upon rod flicker, the chief topic of interest in the present paper.
FRNKES
ef nl.
circles in Fig. 2 again show that the 680 nm background increases the CFF to higher frequencies than observed in the dark. But with the exception of data obtained with flickering stimuli of - I.5 log Scot. td in iIluminance,* (the second set of data from the left), the other two ivavelengrh annuli have a negligible influence on CFF. Clearly, rod matched annuli of different wavelengths have differing influences upon CFF, suggesting that their influence is mediated by cones. In our second control experiment which is illustrated by Fig. 3, we again determined CFF as a function of iliuminance but annuli were equated for their influence upon cones at 1 phot. td. When so equated, the 680, 655, and 620 nm stimuli have different infmences on rods (i.e. they were - 2, - 1.8 and - 1.3 Scot. td respectively) but produced the same influence upon CFF suggesting that the annutar influence is mediated by cones. To be sure, the results obtained with other wavelengths can only indirectly support this conclusion due to light from the annulus straying into the retinal region stimulated by the flickering test probe. For wavelengths ~540 nm, this stray light for the rod system was great enough to directly interfere with the visibitity of the flickering test probe and, hence, depress CFF. The resulting deviation from the control data, however, is in the opposite direction predicted by a rod mediated annular effect. The most parsimonious conclusion from Figs 2 and 3 is that our usual 680 nm annular field mediates its effect by stimulating cones.
15
5
680
0
645 620
A n
540
I i
I
-20
-t 5
Green flicker
illuminatiocl
5’2 / 472
x
d .
-1 0
t -0.5
flog scof. id)
Fig. 3. CFF as a function of illuminance and of the wavelength of annular iieids of the indicated wavelengths but equated for cones at 1 phot. id.
Influence
of cone adaptation
F.N.
20 -
2 ; 15tJ
10 -
,gy 5
.
/’ I -3
I -2
I -1
I 0
I 1
1 2
Flicker illumination (log sco?. td)
Fig. 4. CFF as a function of illuminance for stimuli stemming from a green LED (the circles), a red LED (the triangles), or red/green stimuli flickering in counterphase and yoked together in scotopic illuminance (the squares). Data were either obtained with no annulus (solid symbols) or a 680nm annulus of I phot. td iliuminance (open symbols). For the red/green counterphase condition, the indicated scotopic illuminance on the abscissa applies to the illuminance contribution provided by either red- or greenflicker alone and, hence, is displaced by 0.3 log units. See text for further explanation.
In order to determine whether the “rod Ferry-Porter” limb is in fact mediated by rods, we repeated the experiment shown in Fig. 1 using three different flickering test probes, all with the same spatial properties. The first of these was that used for Figs l-3 and stemmed from a green LED. The second was generated by a red-orange LED and probably stimulated both rods and cones at dim illuminance levels (i.e. see Wald, 1945). The third stimulus stemmed from both LEDs flickering in counterphase and always yoked together so that their modulation depth and scotopic iiluminance was the same. At these equated scotopic illuminance levels, the red LED provided a 1.6 log unit (a factor of 40) greater photopic illuminance. Hence, for cones, this stimulus was virtually identical to that provided by the red LED alone but for rods, the stimulus was nonflickering and twice the illuminance of the green LED alone. Results are presented in Fig. 4. The solid symbols represent CFF with no annular background field while the open symbols show CFF with a 680 nm background field of 1 phot, td. The data plotted with circles were obtained with the green LED and are very similar to those shown in Fig. 1. Notice again that the annulus facilitates CFF for the rod Ferry-Porter and
upon
1171
flicker
rod plateau limbs, but not for the cone Ferry-Porter limb. For stimulus illuminance values dimmer than -0.5 log Scot. td, data obtained with the red LED (the triangles) show the same rod Ferry-Porter function and are almost identical to that obtained with the green LED. But with this stimulus at higher illuminance levels, we obtained no annular influence upon CFF and the data stand above the rod-plateau. Finally, with the red/green counterphase stimulus (squares) which precluded rod flicker, CFF is always uninfluenced by the annular background and in fact, bears a considerably different functional relationship to stimulus illuminance. To emphasize this difference, these data are fit (by inspection) with a straight line of considerably greater slope. We are unaware of previous similar comparisons of rod vs cone sensitivity to flicker but we can discern only one explanation. Flicker is detected by rods alone when the green LED is dimmer than +OS log Scot. td, or the red LED alone dimmer than -0.5 log Scot. td; all other data pertain to cone mediated flicker sensitivity. In other words, rods are always more sensitive than cones to dim flicker. In summary, the results in Figs l-4 show that illumination of nearby cones facilitates CFF mediated by rods. Rod adaptation has a much smaller influence upon rod flicker (the data obtained at - 1.5 log Scot. td in Fig. 2). At the illuminance levels we employed, cone adaptation has no similar influence upon cone mediated flicker. (2) Modulation
thresholds
The above CFF data indicate the range of rod flicker probe parameters which show cone-rod interaction. But CFF is of limited value since by definition, it only examines flicker sensitivity at the limiting high frequency. For this reason, we examined the influence of the annular adapting field used in Fig. 1 upon modulation thresholds over a range of flicker frequencies. In these experiments, the average illuminance of the flickering test probe was fixed at three portions of the rod Ferry-Porter limb shown in Fig. 1: at 0.01 (-2 log), 0.03 (-1.5), and 0.2 (-0.7 log) Scot. td. Results are shown in Fig. 5. The lower coordinates plot modulation threshold amplitude as a function of flicker frequency according to usual convention (i.e. an increase in sensitivity is indicated by an upward shift on the ordinate). The solid symbols again represent data obtained with no annular field, the open
I i-2
THOMAS E. FRLMKES
I
1
III
1
2
3
Flicker
45
I
I
to
15
frequency
Fig. 5. The Intluence of nicker frequency and a 68Onm xmulus upon modulation sensitivity. The lower coordinates shows modulation amplitude as a function of flicker frequrnc~: :rn I~CTCLXon the ordinate indicates a laker rn~~dul~~t~~~n threshold or an ittcr~mc in sensittvity. The solid symbo1.i were obtained with no annulus while the open symbols were obtained with a 680 nm annulus of I phot. td illuminance The upper coordinates plot the ratio of the open to solid symbol value (expressed on the ordinate in %) as a function of dicker frequency. The horizontal dashed lines represent 95% confidence intervals (“2 SE) for the datum in the dark having the most variability. For both upper and Io\srr plots. symbol shape indicates the average illuminancr of the thckcring test probe which was either 0.2 icircles), 0.03 (trianglej). or 0.01 (squares) Scot. td.
symbols represent data obtained in the presence of a 650 nm annular field of 1 log phot. td, while the different shaped symbols represent the different average illuminance levels of the rod flickering test probe as indicated. At these dim levels, notice that each average illuminance value produces a unique function with no common envelope at the high frequency end which is characteristic of corresponding photo@ flicker data and is indicative of a linear system
*The horizontal dashed lines indicated on tire top of Fig. 5 are the YjO% confidence intervals (+Z SE) for the most variable datum obtained with -no annulus present. I Hz flicker of 0.03 scat. td. but ire quite for other data obtained with flicker similar frequencies = ~4 Hz. Variability tended to decrease as Hider incrsasxi further in frequency.
e! d.
(i.e. see Kelly. 1973). This replicates prior scotupic data from our laboratory (Nygaard and Frumkes. 1985). Yforeover. the influence of the annulus depends both upon the average illuminance and frequency of the flicker test probe. The annulus never has an appreciable influence at the lowest frequency used (I Hz) and always enhances sensitivity at frequencies between 6-10 Hz. Since further generalizations are hard to establish from the lower plot in Fig. 5. we calculated the ratio of flicker sensitivity obtained in the fight to that in the dark. These values are plotted in the upper part of Fig. 5, again on logarithmic coordinates; the same shapes are used to indicate the average illuminance of the rod flickering test probe. The horizontal dashed lines indicates 95% confidence intervals for sensitivity in the dark and hence, data standing outside this space show a sigmficant influence of the annular adapting field.* The data obtained with the 0.2 (circles) and 0.01 (squares) Scot. td illuminance levels are quite consistent. For flicker frequencies between l-4 Hz, the annular field has an insignificant influence on sensitivity. &t as frequency further incrtrases, the background first increases and then again decreases flicker sensitivity. The same tendency is seen for flicker frequencies >4 Hz with the 0.03 td flicker probe, but at lower frequencies, there is still some annular facilitation of flicker. Although we were first surprised by this deviation, it is in fact predictable and represents the slight rod-mediated influence of the annulus upon rod flicker sensitivity seen at the same illuminance value (- 1.5 log Scot. td) in Fig. 2, an effect which is obtained with flicker frequencies of l-7 Hz (Nygaard and Frumkes, 198s). In summary, modulation threshold data show that regardless of the average illuminance level of the flickering rod test probe, the present type of cone-rod interaction is negligible For flicker frequencies $4 Hz and maximal for flicker frequencies between 6 and IO Hz. (31 Iliuminance
thresholds
The above data indicate that cones stimulated by an annulus influence rod flicker in a frequency specific manner. In order to study the influence of annular illuminance at various frequencies, we chose a third index of flicker sensitivity. Accordingly, the observer increased the average illuminance of the flickering test probe fixed in modulation depth (at 96%) and frequency until flicker was just perceived. In
1173
Influence of cone adaptation upon flicker
monotonic influence of adapting illuminance is highly reliable across observers, as is the influence of flicker frequency upon this effect. We address the apparent discrepancy with the modulation threshold data of Fig. 5 in the discussion. Previously, we (Goldberg er al., 1983) obtained functions with superficial similarity to Fig. 6 when examining the influence of rod adapting fields upon cone mediated flicker sensiriviry. Thus, in spite of the control experiments illustrated in Figs 24, the skeptic might argue that all data in Figs l-6 are examining a general influence of an annular field upon flicker sensitivity and are not particular to the receptors influenced by the various stimuli. We, therefore, compared these two separate effects directly always using illuminance thresholds and the same spatial stimulus arrangement. In one situation, we examined the influence of cones upon rod flicker and hence, used the same wavelength stimuli as just described. We refer to this below as cone-rod interaction. The other situation was construed to examine the influence of a rod annulus upon cone flicker, which we refer to below as rod-cone interaction. Hence, the flickering test probe involved the red/green counterphase control procedure used for some experiments illustrated in Fig. 4 above and used an annulus which was 512 nm in wavelength. Figure 7 illustrates results obtained with 12 Hz flicker. The test probe illuminance necessary for flicker detection is again plotted as a
14 HZ 12 HZ 10 HZ 6 HZ 6 Hz I
I
I
I
-1.0
-0.5
0.0
0.5
I -1 680nm
5
annulus
illumination
(log phot. td)
Fig. 6. Minimal iliuminance of a test probe necessary for flicker detection as a function of the illuminance of a 680 nm annulus. The sinusoidally flickering stimulus was modulated at 96% and stemmed from a green LED. Flicker frequency is indicated by the different shaped symbols.
Fig. 6, these illuminance thresholds are plotted as a function of the illuminance of the 680 nm annular field with flicker frequency as a parameter. With this plot, an increase in flicker sensitivity is indicated by a lowering on the ordinate. For flicker frequencies < 5 Hz (not illustrated), increasing annular illuminance to the highest value illustrated had no effect on flicker sensitivity. But for all the illustrated frequencies, flicker sensitivity increases with annular illuminance. This effect is only about 0.45 log units at 6 Hz but increases gradually with frequency to a 1.2 log unit effect at 14 Hz. This
__--* \ /
E ‘Z 0
.c
-2.0
l
S z
-
Rod am.,
P
cone flicker
yz+//
ti
Cone onn.. rod flicker
B G
‘1. \
-2.5
l-Y
F.N. 12 Il.7 I
-Q
I
I
A ”
-3
I -1
-2 Annulus
I
(log
phot. td
0
I
Fig. 7. Minimal illuminance of a I2 Hz test p~r+e necessary for flicker detection as a function of the illuminance of a surrounding annulus. The open triangles show data obtained with a flickering stimulus involving red/green counterphase as in Fig. 4. but surrounded with a 512 nm annulus; the solid symbols show data obtained with green LED surro_und_edby a 680 nm annulus as in Fig. 6. In order to produce maximal correspondence of these functions, the ordinate is scaled in scofopic, the abscissa in phofopic units. For the data plotted by triangles, the ordinate only represents the one component of the red/green counterphase stimulus and, hence, is actually displaced 0.3 log units downward. See text for further explanation.
I 17-l
THOM.ASE. I=RUWRESer
function of annular iliuminance. For comparison sake, the ordinate is plotted in scotopic units. the abscissa in photopic units as other combinations would greatly displace the two sets of data from one another. The solid circles show cone-rod interaction. These data appear somewhat different than the I2 Hz function in Fig. 6 because of differences in vertical and horizontal spacing and in the range of iliuminance values on the abscissa, but are in fact in good agreement. Notice that as the “cone annuius increases in illuminance stimulating” from -3.3 to - 1.3 log phot. td, it has little influence upon rod flicker sensitivity. This is expected as over much of this range, this annulus is totally invisible (visibility threshold is inabout _ 1.5 log phot. td). As illuminance crease from - 1.3 to f0.7 log phot. td, however, rod Bicker sensitivity gradually increases and the maximal changes exceeds 0.6 log units. The triangles show rod-cone interaction, Notice that as the annulus increases from a value which is near the threshold for absolute visibility (-3.3 log units) to -0.3 log units, cone sensitivity gradually increases but only by 0.3 log units. At still higher annular ifluminance values (corresponding to photopic illuminan~es for which the 680 nm annulus further increases rod flicker sensitivity), cone flicker sensitivity decreases sharply. We did not directly compare cone--rod and rod-cone interaction at other flicker frequencies. However, scrutiny of Fig. 6 and similar rod-cone interaction data (i.e. Goldberg et al., 1983, Fig. 2) indicate that the range of annular illuminance values over which both rod-cone and cone-rod interaction occur is independent of flicker frequency. On the other hand, the relative magnitude of rod-cone vs cone-rod interaction must vary complexly with flicker frequency. That is, the above experiments show that cone-rod interaction is negligible for frequencies < 6 Hz and increases monotonically as frequency increases from 6 to 14 Hz. In contrast, rod-cone interaction increases gradually with flicker frequency over a range from 1 to 20 Hz (Fig. 2 of Goldberg et al., 1983). DISCUSSION
The foregoing data shows that photic stimulation of cones exerts a marked influence upon rod mediated flicker in a neighboring-area of This effect is specific to stimuli retina. influencing these receptors: stimulation of neighboring rods has a much smaller influence
ai
upon rod mediated flicker, and at the dim illuminan~e levels we employed. stimulation of neighboring cones has a negligible influence upon cone mediated flicker. We. therefore, suggest that this demonstration of cone-rod interaction invotves a unique mechanism. The influence of flicker frequency on this type of cone-rod interaction is not completely clear as CFF, and modulation and iiluminance flicker thresholds suggest somewhat different tendencies. For flicker frequencies <4 Hz, iliuminance and modulation threshold data show a lack of cone-rod interaction; CFF data cannot either support or refute this finding due to the impossibility of data collection (i.e. dimmer stimuli than used in Fig. 1 are not visible). From 6 to 14 Hz, the iliuminance threshold data is reasonably consistent with the CFF data. That is. the CFF data in Fig. 1 show that over the lower range of the rod Ferry-Porter function (between 5 and 9Hz), the cone annular surround increases CFF by a constant amount (about 2.5 Hz): this vertical shift on tbe ordinate is equivalent to shifting the function horizontally by about 0.45 log units. Figure 6 shows that the difference in the il~uminance threshold for 6 Hz stimuli obtained with no background and with the adapting annulus illuminance used for Fig. I (0 log phot. td) is about 0.45 log units as would be expected. However, for iliuminance values greater than -2 log Scot. td, the straight “Ferry-Porter lines” placed amidst the CFF data of Fig. I are misleading. Carefully scrutiny indicates that the relationship between log iiluminance and CFF is curvilinear. and the lzorizontal difference between the set of functions obtained in the dark and with the annulus becomes increasingly greater up to the region where the cone Ferry-Porter function obtains. From this, one would anticipate that the influence of an annular background upon illuminance thresholds would increase with flicker frequency, the tendency in fact observed in Fig. 6. At this time, we do not understand the reason for modulation thresholds showing a smaller annular influence at 12 and 14 than at 8 and 10 Hz. This discrepancy between illuminance and modulation threshold data is also apparent for the other two observers we examined. In combination with prior results, the present data suggests that cones can interact with rods by at least three different mechanisms. First, rod and cone related signals can summate together to produce a threshold signal (Drum, 1982; Frumkes et a/., 1973), to influence increment
1175
Infiuente of cone adaptation upon flicker
thresholds (e.g. Buck, 1985; Bauer et al., 1983; Frumkes and Temme, 1977), and even to produce a perceptible flicker sensation (MacLeod, 1972; van den Berg and Spekreijse, 1977). Although there is uncertainty regarding the underlying spatial parameters. all of these resutts can be modeled by postulating a convergence of like polarity signals at some locus within the visual system (i.e. see Bauer et al., 1983). Physiological and anatomical data in mammals suggest that these effects could be mediated by electrical coupling of photoreceptors and/or a variety of pathways within the retinal inner plexiform layer (Nelson and Kolb, 1984); at any event, this mechanism is clearly evident in extracellular spike records from retinal ganglion cells (Rodieck and Rushton, 1976). Secondfy, dark adapted and photically unstimulated rods exert a suppressive influence on cone-mediated flicker, an effect which is removed by selective rod light adaptation (Alexander and Fishman, 1984; Coletta and Adams, 1984; Goldberg et al., 1983). Results from individuals with circumscribed retinal pathology, with hereditary color defects, and eiectrophysiologicai data suggest that this effect is mediated in the distal neuroretina. If results in amphibians can be extended to mammals, it probably invofves rod modulation of horizontal cell feedback onto cones (Frumkes and Eysteinsson, 1986; Frumkes et al., 1986). The present experiments show that annular stimulation of cones exerts an influence upon rod mediated flicker in neighboring retina. This effect occurs at much dimmer illuminance levels than adaptational influences on flicker involving cone-cone interaction (e.g. see Coletta and Adams, 1984), and has much different parametric properties than rod-rod interaction (Nygaard and Frumkes, 1985) or rod-cone interaction. Therfore, this must involve a third distinct mechanism permitting an interaction between rod and cone signals about which, however, little else is known, We cannot presently state whether the cone annulus removes a tonic suppression or actively facilitates rod flicker, and we are unaware of any evidence suggesting a plausible neural substrate in the retina or cerebrum. We hope to uncover some of the mystery by manipulating spatial parameters and by investigating individuals with circumscribed retinal pathology. Acknowledgements-We thank MS Noreen Denny and Mr Thor Eysteinsson for serving as observers. Supported in part by a Biomedical Grant from NIH to Queens College.
REFERENCES Alexander K. R. and Fishman G. A. (1983) Rod-cone interaction: evidence for a distal retinal locus. J. opt. Sot. rim. 73, 1915. Alexander K. R. and Fishman G. A. (1984) Rod-cone interaction in flicker perimetry. Br. J. Ophtkof. 68, 303-309. Alexander K. R. and Fishman G. A. (1985) Rod-cone interaction in flicker perimetry: evidence for a distal retinal locus. L)ocumen~o Uphth. 60, 3-36. Ardcn G. B. and Hogg C. R. (1985) Rod-cone interaction and analysis of retinal disease. Br. J. Ophtkul. 69, 404-415.
Arden G. B. and Frumkes T. E. (1986) Stimulation of rods can increase cone flicker ERGS in man. Vision Res. 26, 71 l-721. Bauer G. M., Frumkes T. E. and Nygaard R. W. (1983) The signal-to-noise characteristics of rod-cone interaction. 1. Physiol. 337, 101-I 19. Buck S. L. (1985) Cone-rod interaction over time and space. Vision Res. 25, 907-916.
Coletta N. J. and Adams A. J. (1984) Rod-cone interactions in flicker detection. Vision Res. 24, 1333-1340. Coletta N. J. and Adams A. J. (1985) Loss of flicker sensitivity on dim backgrounds in normal and dichromatic observers. Invest. Ophtkal. visual Sci., Suppl. 26, 187. Drum B. (1982) Summation of rod and cone responses at absolute threshold. Vision Res. 22, 823-826. Eysteinsson T. and Frumkes T. E. (1985) Rod suppression of cones and achromatic cone pathways in amphibian retina. Neurasci. Abstr. 11, 476. Ferry E. S. (1892) Persistence of vision. Am. J. Sci. 44, 192-207.
Frumkes T. E. and Eysteinsson T. (1986) Suppressive rod-cone interaction in distal vertebrate retina-Intracellular records from Xenopus and Necfurur. J. Neuropkysiol. To be published. Frumkes T. E., Eysteinsson T. and Arden G. B. (1985) Tonic inhibition of red-cone pathways by dark adapted rods. invest. Opktkai. visual Sci., Suppl. 26, 114. Frumkes T. E. and Temme L. A. (1977) Rod-cone interaction in human seotopic vision-II. Cones influence rod increment thresholds. Vision Res. 17, 673-679. Frumkes T. E., Naarendorp F., Eysteinsson T., Denny N. and Goldberg S. H. (1986) Suppressive rod-cone interactions in distal vertebrate retina-I. Comparison of human psychophysics with intraeelhrlar recording from amphibian retina. J. opt. Sot. Am. To be. published. Frumkes T. E., Sekuler M. D., Barris M. C., Reiss E. H. and Chalupa L. M. (1983) Rod-cone interaction in human scotopic vision-I. Temporal analysis. Ytiion Res. 13, 1269-1282. Fry G. A. and Bartley S. H. (1936) The effect of steady stimulation of one part of the retina upon the critical frequency in another. J. exp. Psychol. 19, 351-356. 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., Naarendorp F. and Frumkes T. E. (1985) Tonic inhibition of rod flicker by dark-adapted cones. Invest. Opktkol. visual Sci., Suppl. 26, 1I. Heeht S. and Smith E. L. (1936) intermittent stimulation by light. VI. Area and the relation between critical frequency and intensity. J. gen. Physiol. 19, 979-991.
II76
THOMAS E. FRU~KES er ai.
Hood D. (1972) Suppression of the frog‘s cone skstem in the dark. l’ision Res. 12. 889-908. Keesey U. T. (1970) Variables determining flicker sensitlrlt> in small fields. J. opr. Sot. Am. 60, 390-398. Kelly D. H (1972) Flicker. In Handbook o/‘Sensor~ Phj,siology IVII,~) (Edited by Jameson D. and Hun;lch L.). pp. 273-302. Springer, Berlin. Loeu E. R. and Arden G. B. (1985) Inhibition of conesby rods in the mammalian eye as demonstrated electrophysiologically using flashing multipoint focal stimuli. Inrest. Ophrhal. tbal. Sri. Suppl. 26, 115. MacLeod D. 1. A. (1972) Rods cancel cones in flicker. h’arrrre 235, I73- 174. Nelson R. and Kolb H. (1984) Amacrine cells in scotopic vision. Ophrhal. Res. 16, 21-26. Nygaard R. W. and Frumkes T. E. (1982a) Calibration of the retinal illuminance provided by Maxwellian view. l%islon Re.r. 22, 433-434. Nygard R. W. and Frumkes T. E. (1982b) LEDs: con-
vement, inexpensive sources for visual cxperlmentatlor. C’ision Res. 22, 135110. Sygaard R. W. and Frumkss T. E. (19S5) Frrquenq dependence in scotopic flicker sensttikitb. I’/.r~orl Rrs. 25. 115-127. Porter T. C. (1902) Contributions to the study of Hicksr. II Proc. R. Sot. 7OA, 3 13-329. Rodieck R. W. and Rushton W. A. H. (1976) Cancsllatlon of rod signals by cones. and cone signals by rods in ths cat retina. J. PhJsiol. 23, 775-755. Smith R. A. (1971) Studies of temporal frequency adaptation in visual contrast sensitivity. J. Ph~siol. 216.53 I-551. Sterling P. (1983) Microcircuitry of the cat retina. .-inn. Rec. !Crrtrosci. 6, l49- 185, van den Berg T. J. T. P. and Spekreijse H. (1977) Interaction between rod and cone signals studied with temporal sine wave stimulation. J. opt. Sot. rim. 67, IZIU-1217. Wysecki G. and Stiles W. S. (1967) Color Scietlce. Wiley. New York.
A
Biochemistry and Molecular Biology Retinal Physiology,
Cell Biology, Neurotransmitters,
Central Nervous System Physiology
Morphology
and Morphology
001’.6989,565340+ O.;H)
VisionRer. Vol. 26. No. 8, pp. I 167-l176.1986
Pergamon Journals Ltd
Printed in Great Britain
THE INFLUENCE OF CONE ADAPTATION UPON ROD MEDIATED FLICKER THOMAS E. FRUMKES, FRAXK NAAREXDORP and STUART H. GOLDBERG Department of Psychology, Queens College of CUNY, Flushing, NY 11367, U.S.A. (Received 7 Notember
1985; in revised form 3 March 1986)
A&tract--The influence of annular fields on sensitivity to sinusoidal flicker was assessed in the dark adapted parafoveai retina. Test stimuli were 2”20’ in diameter; annuli had a 2”20’ inner and 7’30’ outer
diameter. Rod flicker was studied with a “green” stimulus too dim to influence cones. Selective cone flicker was obtained using red and green flicker in counterphase and yoked together in modulation depth and scotopic illuminance. Results showed the following. (1) Annular stimulation of rods slightly facilitated rod-mediated flicker sensitivity to frequencies < 10 Hz. In contrast, annular stimulation of cones greatly facilitated rod-mediated sensitivity, particularly for flicker frequencies >7 Hz. We designate this effect, cone-rod interaclion. (2) Annular stimulation of cones has a negligible influence upon sensitivity to cone-mediated flicker frequencies < 15Hz. In contrast, annular stimulation of rods has a large influence upon sensitivity to cone-mediated flicker, an effect we designate rod-cone infepxrion. (3) Within limits, both rod-cone and cone-rod interaction increase as the annular illuminance increases and as flicker. frequency increases; the limiting frequency and illuminance values, however, are different for the two forms of interaction. Results are compared with prior evidence that rod and cone signals summate to produce an absolute threshold or flicker sensation. We suggest that there are at least three mechanisms for interaction between rod- and cone-related signals. Rods
Cones
Rod-cone interaction
Flicker
Retina
INTRODUCTION
A growing body of evidence indicates that visual signals stemming from the rods and cones of the vertebrate eye interact with one another. Prior psychophysical investigations using flicker have indicated two different underIying mechanisms. First, MacLeod (1972) and van den Berg and Spekreijse (1977) showed that separate flicker signals stemming from rods and cones can summate together, presumably by converging at some common locus in the retina. The retative phase of the rod- and cone-flicker signals, which is influenced both by experimental manipulation as well as the more sluggish response properties of rods, determines whether the separate signals enhance or destructively interfere with each other. This approach and interpretation was also used by Rodieck and Rushton (1976) to interpret flicker responses recorded extraceIlulariIy from cat ganglion cells. Difference in rod vs cone latency also determines whether subliminal rod and cone signals will add together to produce a threshold sensation (Frumkes er al., 1973). More recent investigations have described a “new” type of rod-cone interaction (Alexander
Psychophysics
and Fishman, 1984; Arden, 1985; Coletta and Adams, 1984; Goldberg et af., 1983; but although see also Lythgoe and Tansley, 1929). Cone-mediated flicker sensitivity, especially for frequencies greater than 15 Hz, is depressed when rods are dark adapted and greatly enhanced by selective rod light adap~tion, a phenomenon which we have referred to as suppressive rod-cone interaction (Frumkes and 1986; Frumkes, Naarendorp, Eysteinsson, Eysteinsson, Denny and Goldberg, 1986). This effect can be shown with cornea1 ERGS in humans and other mammals (Arden and Frumkes, 1986; Loew and Arden, 1985) and by means of intracellular recording in distal retinal neurons of amphibians including rhe cones themselves (Frumkes and Eysteinsson, 1986). Suppressive rod-cone interaction is also characteristically abnormal in some individuals with circumscribed retinal pathology (Alexander and Fishman, 1985; Arden and Hogg, 1985) and in some individuals with protanomalous, hereditary color deficiencies (Alexander and Fishman, 1983; Coletta and Adams, 1985; Goldberg and Frumkes, 1983). On the basis of these data, we proposed that the neural circuitry is within the distal most retina and that the phenomenon
1167
THCMAS E. FKLWES
II68
reflects rod modulation of horizontal cell feedback onto con2s (Frumkes el ai.. 1986). This explanation setms rsasonablr sincr feedback of horizontal ~211s upon cones has besn shown in an 2wr growing number of species {e.g. ~22 Stzrling. 1983), and in all species studied by ph.vsioio,oical techniques. horizontal cells receiving cone-input have been shown to additiona!ly receive some rod-input. Since there is no evidence for a horizontat ceil feedback mechanism onto rods. we reason2d that there should be no influence of cone-adaptation upon rod fficker. In spit2 of the apparent absence of relevant neural circuitry, the present results show that the adapted state of cones in neighboring retinal areas exerts a striking influence upon rodmediated flicker. Although this phenomenon bears superficial resemblances to other psychophysical effects, most particularly suppressive rod-con2 interaction, its illuminance and frequency limitations are sufficiently unique to suggest it is a fundamentally distinct type of cone--rod interaction. This paper has been presented previously in abstract form {Goldberg ef rri.. 1985).
METHODS
All data were collected with a 5 channel ~ax~vellian view optical system described previously (Frumkes et al., 1986; Nygaard and Frumkss, 1985). In brief. 4 channels stemmed from ssparate light emitting diod2 (LED) sources. One of these was a “green” LED with a peak navelength at 545 nm (Stanley Electric ESBG5503) and served as the flickering source for the majority of our experiments; the other rhre2 stemmed from “red-orange’. LEDs with peak wavelength at 630nm (General Instruments MV5152) and were used for fixation or control purposes. The temporal waveforms of these sources were controlled by the circuit used by Frumkes and Eysteinsson (1986, Fig. l), and monitored online with phototransistors (see Nygaard and Frumkes, 1982b). At the maximum depth of flicker modulation used, 96%, total distortion was less than 2% and was less with lower modulation depths. Average illuminante lev2l was controlled by means of neutral density wedges and fixed filters. The remaining which provided adaptation stimuli, beam, stemmed from a conventional tungsten source;
:‘i .:.
stimulus wavelength and iliuminance was controlled by means of neutral density wedges or filters and interference filters with half bandwidths of 6-12 nm. The spatial position, size, and shape of all stimuli was determined by placing appropriate targets in collimated beams. The photopic illuminance of all stimuli was calibrated according to the procedure of Nygaard and Frumkes (1982a). Scotopic illuminance was calculated by comparing the relative sensitivity of the CIE photopic observer to various wavelengths (See Wysecki and Stiles, 1967): these calculated scotopic values were occasionally checked according to the procedurrs of Frumkes and Temme (1977).
The final display was presented to the obs2rv2r.s eye in .!vfaxwellian view. This consisted of three stimuli. The first was a central red fixation target. The second was a 2”20’ diamzter. sinusoidally flickering test stimulus presented in the stimulus field 7” to the right of fixation. For the majority of experiments, this was genrrated by a green LED and had a modulation depth of 96%. The third stimulus ~vas an ;tnnular adapting field concentric to the flickering test: the inner diameter was contiguous with the test and the outer diameter was 7 30’. In most experiments, the annular adapting field was 680 nm in wavelength and 0 log phot. td (-2 log Scot. td) in illuminance. This annulus gvas the most effective means we coutd devise for light adapting cones and still preserve sensitivity in nearby rods. Illustrated results are from the left eye nasal field of F.N. (an author) which is emmetropic and has normal color vision. All illustrated experiments were also repeated with at least one of four other observers: these included the other t\vo authors and tsvo other laboratory personnel. For these observers, however, stimuli were prss2nted to the right eye temporal field. All observsrs were well informed regarding the purpos2 and procedure used for all experiments. All observers were dark adapted 25 min prior to data collection.
A sinusoidally flickering stimuius with constant spatial properties can be described by three parameters: its average illuminance, its modulation depth, and its frequency. In any threshold experiment. tivo of these parameters
InHuencc of ionc adaptatton
are fixed by the experimenter and the obserler varies the third to threshold which thus serves as index of flicker sensitivity. In the course of pilot ex~~mentation, we found each of these dependent variables to provide unique advantages and, hence, adopted the use of all three. For the data shown in Figs 1-4, the observer varied the frequency of a flickering stimulus of constant wavelength, illuminance, and modulation depth (96%) until Aicker could just be discerned (i.e. determined the critical flicker frequency or CFF). For the data shown in Fig. 5, the observer increased the modulation depth of a flickering stimulus of constant wavelength, illuminance, and frequency until flicker could just be perceived. Finally, for the data shown in Figs 6 and 7, the observer increased the illuminance of stimuli of fixed modulation depth (96%) and frequency until flicker could be perceived. Hence in all cases, thresholds were obtained by means of an ascending method of adjustment. The data plotted below resulted from observations in a minimum of two experimental sessions and in each session, each threshaid was obtained at least three times. As much as possible, we randomized the order of stimulus presentations within an experimental session. With the exception of very few data points, these psychophysical judgements were easy to obtain and standard errors are usually about the size of a plotted datum. Most of our conclusions were also checked in several sessions involving the use of a staircase or temporal forced choice procedure.
z
E
it69
upon fficksr RESULTS
Figure 1 illustrates typical results when critical flicker frequency (CFF) was determined as a function of the illuminance of a stimulus stemming from a green LED. Results plotted with closed circles were obtained with no background field. CFF increases as illuminance increases and only for the purposes of initial description, are fit by inspection rvith three straight fines. According to classical interpretation (e.g. Hecht and Smith. 1936), the initial segment is attributable to the functioning of rods, the horizontal segment represents the rod plateau, and the upper segment is attributable to cones. We refer to the straight lines with a positive slope as Ferry-Porter functions (e.g. Ferry, 1892; Porter, 1902). The data plotted with open circles are results obtained in the presence of a 680 nm annulus of 1 phot. td. illuminance. In respect to the control data, this annulus shifts the rod Ferry-Porter iunction upward but does not alter its slope, it shifts the “rod-plateau” upward, but has no influence on the cone Ferry-Porter function. Similar rest&s were also obtained with the four other observers used. Annular fields influence the sensitivity of both photopic (e.g. Fry and Bartley, 1936; Keesey, 1970) and scotopic (Nygaard and Frumkes, 1985) flicker. Therefore, before preceding to other stimuius manipulations, we deemed it 5rst necessary to establish that the annular back-
15-
2 u
10 -
SI -2
! .
-1
Green flicrte’r illumination (log
I 0
I 1
sot. rd)
Fig. 1. Critical flicker frequency (CFF) as a function of illuminance for sinusoidal Ricker of 96% modulation depth stemming from a green LED. Data were obtained with no annular field (solid circles) or in the presence of a 680 nm annulus of 1 phot. td illuminance (open circles).
1170
Tmms
kk ”
d l
E.
l
/
-;f
Fig. 2. CFF as a function of’ illuminance and of the wavelength of annular fields of the indicated wavelengths but equated for rods at 0.01 Scot. td illuminance.
ground influence which alters the “rod Ferry-Porter” function involves a conemediated annular influence on rod-mediated flicker. We established this with three sorts of control experiments. In the first of these, which is illustrated in Fig. 2, we again determined CFF as a function of ihuminance but in the presence of 512, 620, and 680 nm backgrounds equated for their influence upon rods at - 2 log Scot. td. Because of the difference in spectral sensitivity of rods and cones, the 620 and particularly the 512 nm background fields have a much smaller influence upon cones (i.e. the neurotic illuminance of the 512, 620 and 680 nm fiefds were -2.7. -0.7 and 0 iog td respectively). The open “....
~-.-._----
Wygaard and Frumkes (1985, their Fig. 5) determined the influence of annular fields of 560 and 635 nm of variable scotopic illuminance, upon the sensitivity of green-LED generated flicker of - 1.3 log scat. td (0.05 scat. td) iiluminance; they used spatial parameters quite similar to those in the present experiments. As reported in counterpart photopic experiments (i.e. Fry and Bartiey, 1936; Keesey. 1970). increasing the annulus iiluminance at first increases, then decreases flicker sensitivity. For the two frequencies used, enhancement depended upon scotopic illuminance and not on wavelength. The maximal change in sensitivity observed was about 2% for I Hz flicker and 10% for 7 Hz flicker, which would roughly cotrrespond to a change in CFF of about I Hz. This influence of rod stimLcioring annuli upon rod-derectedjlicker is, hence, quite comparable to that observed in Fig, 2 with a - 1.5 log scat. td test probe and in Fig. 5 with a test probe of 0.03 scat. td and flicker frequencies 6 4 Hz. They never. examined the influence of such annuli with other frequencies or other average illuminance values and, hence, never observed a cone-annular influence upon rod flicker, the chief topic of interest in the present paper.
FRNKES
ef nl.
circles in Fig. 2 again show that the 680 nm background increases the CFF to higher frequencies than observed in the dark. But with the exception of data obtained with flickering stimuli of - I.5 log Scot. td in iIluminance,* (the second set of data from the left), the other two ivavelengrh annuli have a negligible influence on CFF. Clearly, rod matched annuli of different wavelengths have differing influences upon CFF, suggesting that their influence is mediated by cones. In our second control experiment which is illustrated by Fig. 3, we again determined CFF as a function of iliuminance but annuli were equated for their influence upon cones at 1 phot. td. When so equated, the 680, 655, and 620 nm stimuli have different infmences on rods (i.e. they were - 2, - 1.8 and - 1.3 Scot. td respectively) but produced the same influence upon CFF suggesting that the annutar influence is mediated by cones. To be sure, the results obtained with other wavelengths can only indirectly support this conclusion due to light from the annulus straying into the retinal region stimulated by the flickering test probe. For wavelengths ~540 nm, this stray light for the rod system was great enough to directly interfere with the visibitity of the flickering test probe and, hence, depress CFF. The resulting deviation from the control data, however, is in the opposite direction predicted by a rod mediated annular effect. The most parsimonious conclusion from Figs 2 and 3 is that our usual 680 nm annular field mediates its effect by stimulating cones.
15
5
680
0
645 620
A n
540
I i
I
-20
-t 5
Green flicker
illuminatiocl
5’2 / 472
x
d .
-1 0
t -0.5
flog scof. id)
Fig. 3. CFF as a function of illuminance and of the wavelength of annular iieids of the indicated wavelengths but equated for cones at 1 phot. id.
Influence
of cone adaptation
F.N.
20 -
2 ; 15tJ
10 -
,gy 5
.
/’ I -3
I -2
I -1
I 0
I 1
1 2
Flicker illumination (log sco?. td)
Fig. 4. CFF as a function of illuminance for stimuli stemming from a green LED (the circles), a red LED (the triangles), or red/green stimuli flickering in counterphase and yoked together in scotopic illuminance (the squares). Data were either obtained with no annulus (solid symbols) or a 680nm annulus of I phot. td iliuminance (open symbols). For the red/green counterphase condition, the indicated scotopic illuminance on the abscissa applies to the illuminance contribution provided by either red- or greenflicker alone and, hence, is displaced by 0.3 log units. See text for further explanation.
In order to determine whether the “rod Ferry-Porter” limb is in fact mediated by rods, we repeated the experiment shown in Fig. 1 using three different flickering test probes, all with the same spatial properties. The first of these was that used for Figs l-3 and stemmed from a green LED. The second was generated by a red-orange LED and probably stimulated both rods and cones at dim illuminance levels (i.e. see Wald, 1945). The third stimulus stemmed from both LEDs flickering in counterphase and always yoked together so that their modulation depth and scotopic iiluminance was the same. At these equated scotopic illuminance levels, the red LED provided a 1.6 log unit (a factor of 40) greater photopic illuminance. Hence, for cones, this stimulus was virtually identical to that provided by the red LED alone but for rods, the stimulus was nonflickering and twice the illuminance of the green LED alone. Results are presented in Fig. 4. The solid symbols represent CFF with no annular background field while the open symbols show CFF with a 680 nm background field of 1 phot, td. The data plotted with circles were obtained with the green LED and are very similar to those shown in Fig. 1. Notice again that the annulus facilitates CFF for the rod Ferry-Porter and
upon
1171
flicker
rod plateau limbs, but not for the cone Ferry-Porter limb. For stimulus illuminance values dimmer than -0.5 log Scot. td, data obtained with the red LED (the triangles) show the same rod Ferry-Porter function and are almost identical to that obtained with the green LED. But with this stimulus at higher illuminance levels, we obtained no annular influence upon CFF and the data stand above the rod-plateau. Finally, with the red/green counterphase stimulus (squares) which precluded rod flicker, CFF is always uninfluenced by the annular background and in fact, bears a considerably different functional relationship to stimulus illuminance. To emphasize this difference, these data are fit (by inspection) with a straight line of considerably greater slope. We are unaware of previous similar comparisons of rod vs cone sensitivity to flicker but we can discern only one explanation. Flicker is detected by rods alone when the green LED is dimmer than +OS log Scot. td, or the red LED alone dimmer than -0.5 log Scot. td; all other data pertain to cone mediated flicker sensitivity. In other words, rods are always more sensitive than cones to dim flicker. In summary, the results in Figs l-4 show that illumination of nearby cones facilitates CFF mediated by rods. Rod adaptation has a much smaller influence upon rod flicker (the data obtained at - 1.5 log Scot. td in Fig. 2). At the illuminance levels we employed, cone adaptation has no similar influence upon cone mediated flicker. (2) Modulation
thresholds
The above CFF data indicate the range of rod flicker probe parameters which show cone-rod interaction. But CFF is of limited value since by definition, it only examines flicker sensitivity at the limiting high frequency. For this reason, we examined the influence of the annular adapting field used in Fig. 1 upon modulation thresholds over a range of flicker frequencies. In these experiments, the average illuminance of the flickering test probe was fixed at three portions of the rod Ferry-Porter limb shown in Fig. 1: at 0.01 (-2 log), 0.03 (-1.5), and 0.2 (-0.7 log) Scot. td. Results are shown in Fig. 5. The lower coordinates plot modulation threshold amplitude as a function of flicker frequency according to usual convention (i.e. an increase in sensitivity is indicated by an upward shift on the ordinate). The solid symbols again represent data obtained with no annular field, the open
I i-2
THOMAS E. FRLMKES
I
1
III
1
2
3
Flicker
45
I
I
to
15
frequency
Fig. 5. The Intluence of nicker frequency and a 68Onm xmulus upon modulation sensitivity. The lower coordinates shows modulation amplitude as a function of flicker frequrnc~: :rn I~CTCLXon the ordinate indicates a laker rn~~dul~~t~~~n threshold or an ittcr~mc in sensittvity. The solid symbo1.i were obtained with no annulus while the open symbols were obtained with a 680 nm annulus of I phot. td illuminance The upper coordinates plot the ratio of the open to solid symbol value (expressed on the ordinate in %) as a function of dicker frequency. The horizontal dashed lines represent 95% confidence intervals (“2 SE) for the datum in the dark having the most variability. For both upper and Io\srr plots. symbol shape indicates the average illuminancr of the thckcring test probe which was either 0.2 icircles), 0.03 (trianglej). or 0.01 (squares) Scot. td.
symbols represent data obtained in the presence of a 650 nm annular field of 1 log phot. td, while the different shaped symbols represent the different average illuminance levels of the rod flickering test probe as indicated. At these dim levels, notice that each average illuminance value produces a unique function with no common envelope at the high frequency end which is characteristic of corresponding photo@ flicker data and is indicative of a linear system
*The horizontal dashed lines indicated on tire top of Fig. 5 are the YjO% confidence intervals (+Z SE) for the most variable datum obtained with -no annulus present. I Hz flicker of 0.03 scat. td. but ire quite for other data obtained with flicker similar frequencies = ~4 Hz. Variability tended to decrease as Hider incrsasxi further in frequency.
e! d.
(i.e. see Kelly. 1973). This replicates prior scotupic data from our laboratory (Nygaard and Frumkes. 1985). Yforeover. the influence of the annulus depends both upon the average illuminance and frequency of the flicker test probe. The annulus never has an appreciable influence at the lowest frequency used (I Hz) and always enhances sensitivity at frequencies between 6-10 Hz. Since further generalizations are hard to establish from the lower plot in Fig. 5. we calculated the ratio of flicker sensitivity obtained in the fight to that in the dark. These values are plotted in the upper part of Fig. 5, again on logarithmic coordinates; the same shapes are used to indicate the average illuminance of the rod flickering test probe. The horizontal dashed lines indicates 95% confidence intervals for sensitivity in the dark and hence, data standing outside this space show a sigmficant influence of the annular adapting field.* The data obtained with the 0.2 (circles) and 0.01 (squares) Scot. td illuminance levels are quite consistent. For flicker frequencies between l-4 Hz, the annular field has an insignificant influence on sensitivity. &t as frequency further incrtrases, the background first increases and then again decreases flicker sensitivity. The same tendency is seen for flicker frequencies >4 Hz with the 0.03 td flicker probe, but at lower frequencies, there is still some annular facilitation of flicker. Although we were first surprised by this deviation, it is in fact predictable and represents the slight rod-mediated influence of the annulus upon rod flicker sensitivity seen at the same illuminance value (- 1.5 log Scot. td) in Fig. 2, an effect which is obtained with flicker frequencies of l-7 Hz (Nygaard and Frumkes, 198s). In summary, modulation threshold data show that regardless of the average illuminance level of the flickering rod test probe, the present type of cone-rod interaction is negligible For flicker frequencies $4 Hz and maximal for flicker frequencies between 6 and IO Hz. (31 Iliuminance
thresholds
The above data indicate that cones stimulated by an annulus influence rod flicker in a frequency specific manner. In order to study the influence of annular illuminance at various frequencies, we chose a third index of flicker sensitivity. Accordingly, the observer increased the average illuminance of the flickering test probe fixed in modulation depth (at 96%) and frequency until flicker was just perceived. In
1173
Influence of cone adaptation upon flicker
monotonic influence of adapting illuminance is highly reliable across observers, as is the influence of flicker frequency upon this effect. We address the apparent discrepancy with the modulation threshold data of Fig. 5 in the discussion. Previously, we (Goldberg er al., 1983) obtained functions with superficial similarity to Fig. 6 when examining the influence of rod adapting fields upon cone mediated flicker sensiriviry. Thus, in spite of the control experiments illustrated in Figs 24, the skeptic might argue that all data in Figs l-6 are examining a general influence of an annular field upon flicker sensitivity and are not particular to the receptors influenced by the various stimuli. We, therefore, compared these two separate effects directly always using illuminance thresholds and the same spatial stimulus arrangement. In one situation, we examined the influence of cones upon rod flicker and hence, used the same wavelength stimuli as just described. We refer to this below as cone-rod interaction. The other situation was construed to examine the influence of a rod annulus upon cone flicker, which we refer to below as rod-cone interaction. Hence, the flickering test probe involved the red/green counterphase control procedure used for some experiments illustrated in Fig. 4 above and used an annulus which was 512 nm in wavelength. Figure 7 illustrates results obtained with 12 Hz flicker. The test probe illuminance necessary for flicker detection is again plotted as a
14 HZ 12 HZ 10 HZ 6 HZ 6 Hz I
I
I
I
-1.0
-0.5
0.0
0.5
I -1 680nm
5
annulus
illumination
(log phot. td)
Fig. 6. Minimal iliuminance of a test probe necessary for flicker detection as a function of the illuminance of a 680 nm annulus. The sinusoidally flickering stimulus was modulated at 96% and stemmed from a green LED. Flicker frequency is indicated by the different shaped symbols.
Fig. 6, these illuminance thresholds are plotted as a function of the illuminance of the 680 nm annular field with flicker frequency as a parameter. With this plot, an increase in flicker sensitivity is indicated by a lowering on the ordinate. For flicker frequencies < 5 Hz (not illustrated), increasing annular illuminance to the highest value illustrated had no effect on flicker sensitivity. But for all the illustrated frequencies, flicker sensitivity increases with annular illuminance. This effect is only about 0.45 log units at 6 Hz but increases gradually with frequency to a 1.2 log unit effect at 14 Hz. This
__--* \ /
E ‘Z 0
.c
-2.0
l
S z
-
Rod am.,
P
cone flicker
yz+//
ti
Cone onn.. rod flicker
B G
‘1. \
-2.5
l-Y
F.N. 12 Il.7 I
-Q
I
I
A ”
-3
I -1
-2 Annulus
I
(log
phot. td
0
I
Fig. 7. Minimal illuminance of a I2 Hz test p~r+e necessary for flicker detection as a function of the illuminance of a surrounding annulus. The open triangles show data obtained with a flickering stimulus involving red/green counterphase as in Fig. 4. but surrounded with a 512 nm annulus; the solid symbols show data obtained with green LED surro_und_edby a 680 nm annulus as in Fig. 6. In order to produce maximal correspondence of these functions, the ordinate is scaled in scofopic, the abscissa in phofopic units. For the data plotted by triangles, the ordinate only represents the one component of the red/green counterphase stimulus and, hence, is actually displaced 0.3 log units downward. See text for further explanation.
I 17-l
THOM.ASE. I=RUWRESer
function of annular iliuminance. For comparison sake, the ordinate is plotted in scotopic units. the abscissa in photopic units as other combinations would greatly displace the two sets of data from one another. The solid circles show cone-rod interaction. These data appear somewhat different than the I2 Hz function in Fig. 6 because of differences in vertical and horizontal spacing and in the range of iliuminance values on the abscissa, but are in fact in good agreement. Notice that as the “cone annuius increases in illuminance stimulating” from -3.3 to - 1.3 log phot. td, it has little influence upon rod flicker sensitivity. This is expected as over much of this range, this annulus is totally invisible (visibility threshold is inabout _ 1.5 log phot. td). As illuminance crease from - 1.3 to f0.7 log phot. td, however, rod Bicker sensitivity gradually increases and the maximal changes exceeds 0.6 log units. The triangles show rod-cone interaction, Notice that as the annulus increases from a value which is near the threshold for absolute visibility (-3.3 log units) to -0.3 log units, cone sensitivity gradually increases but only by 0.3 log units. At still higher annular ifluminance values (corresponding to photopic illuminan~es for which the 680 nm annulus further increases rod flicker sensitivity), cone flicker sensitivity decreases sharply. We did not directly compare cone--rod and rod-cone interaction at other flicker frequencies. However, scrutiny of Fig. 6 and similar rod-cone interaction data (i.e. Goldberg et al., 1983, Fig. 2) indicate that the range of annular illuminance values over which both rod-cone and cone-rod interaction occur is independent of flicker frequency. On the other hand, the relative magnitude of rod-cone vs cone-rod interaction must vary complexly with flicker frequency. That is, the above experiments show that cone-rod interaction is negligible for frequencies < 6 Hz and increases monotonically as frequency increases from 6 to 14 Hz. In contrast, rod-cone interaction increases gradually with flicker frequency over a range from 1 to 20 Hz (Fig. 2 of Goldberg et al., 1983). DISCUSSION
The foregoing data shows that photic stimulation of cones exerts a marked influence upon rod mediated flicker in a neighboring-area of This effect is specific to stimuli retina. influencing these receptors: stimulation of neighboring rods has a much smaller influence
ai
upon rod mediated flicker, and at the dim illuminan~e levels we employed. stimulation of neighboring cones has a negligible influence upon cone mediated flicker. We. therefore, suggest that this demonstration of cone-rod interaction invotves a unique mechanism. The influence of flicker frequency on this type of cone-rod interaction is not completely clear as CFF, and modulation and iiluminance flicker thresholds suggest somewhat different tendencies. For flicker frequencies <4 Hz, iliuminance and modulation threshold data show a lack of cone-rod interaction; CFF data cannot either support or refute this finding due to the impossibility of data collection (i.e. dimmer stimuli than used in Fig. 1 are not visible). From 6 to 14 Hz, the iliuminance threshold data is reasonably consistent with the CFF data. That is. the CFF data in Fig. 1 show that over the lower range of the rod Ferry-Porter function (between 5 and 9Hz), the cone annular surround increases CFF by a constant amount (about 2.5 Hz): this vertical shift on tbe ordinate is equivalent to shifting the function horizontally by about 0.45 log units. Figure 6 shows that the difference in the il~uminance threshold for 6 Hz stimuli obtained with no background and with the adapting annulus illuminance used for Fig. I (0 log phot. td) is about 0.45 log units as would be expected. However, for iliuminance values greater than -2 log Scot. td, the straight “Ferry-Porter lines” placed amidst the CFF data of Fig. I are misleading. Carefully scrutiny indicates that the relationship between log iiluminance and CFF is curvilinear. and the lzorizontal difference between the set of functions obtained in the dark and with the annulus becomes increasingly greater up to the region where the cone Ferry-Porter function obtains. From this, one would anticipate that the influence of an annular background upon illuminance thresholds would increase with flicker frequency, the tendency in fact observed in Fig. 6. At this time, we do not understand the reason for modulation thresholds showing a smaller annular influence at 12 and 14 than at 8 and 10 Hz. This discrepancy between illuminance and modulation threshold data is also apparent for the other two observers we examined. In combination with prior results, the present data suggests that cones can interact with rods by at least three different mechanisms. First, rod and cone related signals can summate together to produce a threshold signal (Drum, 1982; Frumkes et a/., 1973), to influence increment
1175
Infiuente of cone adaptation upon flicker
thresholds (e.g. Buck, 1985; Bauer et al., 1983; Frumkes and Temme, 1977), and even to produce a perceptible flicker sensation (MacLeod, 1972; van den Berg and Spekreijse, 1977). Although there is uncertainty regarding the underlying spatial parameters. all of these resutts can be modeled by postulating a convergence of like polarity signals at some locus within the visual system (i.e. see Bauer et al., 1983). Physiological and anatomical data in mammals suggest that these effects could be mediated by electrical coupling of photoreceptors and/or a variety of pathways within the retinal inner plexiform layer (Nelson and Kolb, 1984); at any event, this mechanism is clearly evident in extracellular spike records from retinal ganglion cells (Rodieck and Rushton, 1976). Secondfy, dark adapted and photically unstimulated rods exert a suppressive influence on cone-mediated flicker, an effect which is removed by selective rod light adaptation (Alexander and Fishman, 1984; Coletta and Adams, 1984; Goldberg et al., 1983). Results from individuals with circumscribed retinal pathology, with hereditary color defects, and eiectrophysiologicai data suggest that this effect is mediated in the distal neuroretina. If results in amphibians can be extended to mammals, it probably invofves rod modulation of horizontal cell feedback onto cones (Frumkes and Eysteinsson, 1986; Frumkes et al., 1986). The present experiments show that annular stimulation of cones exerts an influence upon rod mediated flicker in neighboring retina. This effect occurs at much dimmer illuminance levels than adaptational influences on flicker involving cone-cone interaction (e.g. see Coletta and Adams, 1984), and has much different parametric properties than rod-rod interaction (Nygaard and Frumkes, 1985) or rod-cone interaction. Therfore, this must involve a third distinct mechanism permitting an interaction between rod and cone signals about which, however, little else is known, We cannot presently state whether the cone annulus removes a tonic suppression or actively facilitates rod flicker, and we are unaware of any evidence suggesting a plausible neural substrate in the retina or cerebrum. We hope to uncover some of the mystery by manipulating spatial parameters and by investigating individuals with circumscribed retinal pathology. Acknowledgements-We thank MS Noreen Denny and Mr Thor Eysteinsson for serving as observers. Supported in part by a Biomedical Grant from NIH to Queens College.
REFERENCES Alexander K. R. and Fishman G. A. (1983) Rod-cone interaction: evidence for a distal retinal locus. J. opt. Sot. rim. 73, 1915. Alexander K. R. and Fishman G. A. (1984) Rod-cone interaction in flicker perimetry. Br. J. Ophtkof. 68, 303-309. Alexander K. R. and Fishman G. A. (1985) Rod-cone interaction in flicker perimetry: evidence for a distal retinal locus. L)ocumen~o Uphth. 60, 3-36. Ardcn G. B. and Hogg C. R. (1985) Rod-cone interaction and analysis of retinal disease. Br. J. Ophtkul. 69, 404-415.
Arden G. B. and Frumkes T. E. (1986) Stimulation of rods can increase cone flicker ERGS in man. Vision Res. 26, 71 l-721. Bauer G. M., Frumkes T. E. and Nygaard R. W. (1983) The signal-to-noise characteristics of rod-cone interaction. 1. Physiol. 337, 101-I 19. Buck S. L. (1985) Cone-rod interaction over time and space. Vision Res. 25, 907-916.
Coletta N. J. and Adams A. J. (1984) Rod-cone interactions in flicker detection. Vision Res. 24, 1333-1340. Coletta N. J. and Adams A. J. (1985) Loss of flicker sensitivity on dim backgrounds in normal and dichromatic observers. Invest. Ophtkal. visual Sci., Suppl. 26, 187. Drum B. (1982) Summation of rod and cone responses at absolute threshold. Vision Res. 22, 823-826. Eysteinsson T. and Frumkes T. E. (1985) Rod suppression of cones and achromatic cone pathways in amphibian retina. Neurasci. Abstr. 11, 476. Ferry E. S. (1892) Persistence of vision. Am. J. Sci. 44, 192-207.
Frumkes T. E. and Eysteinsson T. (1986) Suppressive rod-cone interaction in distal vertebrate retina-Intracellular records from Xenopus and Necfurur. J. Neuropkysiol. To be published. Frumkes T. E., Eysteinsson T. and Arden G. B. (1985) Tonic inhibition of red-cone pathways by dark adapted rods. invest. Opktkai. visual Sci., Suppl. 26, 114. Frumkes T. E. and Temme L. A. (1977) Rod-cone interaction in human seotopic vision-II. Cones influence rod increment thresholds. Vision Res. 17, 673-679. Frumkes T. E., Naarendorp F., Eysteinsson T., Denny N. and Goldberg S. H. (1986) Suppressive rod-cone interactions in distal vertebrate retina-I. Comparison of human psychophysics with intraeelhrlar recording from amphibian retina. J. opt. Sot. Am. To be. published. Frumkes T. E., Sekuler M. D., Barris M. C., Reiss E. H. and Chalupa L. M. (1983) Rod-cone interaction in human scotopic vision-I. Temporal analysis. Ytiion Res. 13, 1269-1282. Fry G. A. and Bartley S. H. (1936) The effect of steady stimulation of one part of the retina upon the critical frequency in another. J. exp. Psychol. 19, 351-356. 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., Naarendorp F. and Frumkes T. E. (1985) Tonic inhibition of rod flicker by dark-adapted cones. Invest. Opktkol. visual Sci., Suppl. 26, 1I. Heeht S. and Smith E. L. (1936) intermittent stimulation by light. VI. Area and the relation between critical frequency and intensity. J. gen. Physiol. 19, 979-991.
II76
THOMAS E. FRU~KES er ai.
Hood D. (1972) Suppression of the frog‘s cone skstem in the dark. l’ision Res. 12. 889-908. Keesey U. T. (1970) Variables determining flicker sensitlrlt> in small fields. J. opr. Sot. Am. 60, 390-398. Kelly D. H (1972) Flicker. In Handbook o/‘Sensor~ Phj,siology IVII,~) (Edited by Jameson D. and Hun;lch L.). pp. 273-302. Springer, Berlin. Loeu E. R. and Arden G. B. (1985) Inhibition of conesby rods in the mammalian eye as demonstrated electrophysiologically using flashing multipoint focal stimuli. Inrest. Ophrhal. tbal. Sri. Suppl. 26, 115. MacLeod D. 1. A. (1972) Rods cancel cones in flicker. h’arrrre 235, I73- 174. Nelson R. and Kolb H. (1984) Amacrine cells in scotopic vision. Ophrhal. Res. 16, 21-26. Nygaard R. W. and Frumkes T. E. (1982a) Calibration of the retinal illuminance provided by Maxwellian view. l%islon Re.r. 22, 433-434. Nygard R. W. and Frumkes T. E. (1982b) LEDs: con-
vement, inexpensive sources for visual cxperlmentatlor. C’ision Res. 22, 135110. Sygaard R. W. and Frumkss T. E. (19S5) Frrquenq dependence in scotopic flicker sensttikitb. I’/.r~orl Rrs. 25. 115-127. Porter T. C. (1902) Contributions to the study of Hicksr. II Proc. R. Sot. 7OA, 3 13-329. Rodieck R. W. and Rushton W. A. H. (1976) Cancsllatlon of rod signals by cones. and cone signals by rods in ths cat retina. J. PhJsiol. 23, 775-755. Smith R. A. (1971) Studies of temporal frequency adaptation in visual contrast sensitivity. J. Ph~siol. 216.53 I-551. Sterling P. (1983) Microcircuitry of the cat retina. .-inn. Rec. !Crrtrosci. 6, l49- 185, van den Berg T. J. T. P. and Spekreijse H. (1977) Interaction between rod and cone signals studied with temporal sine wave stimulation. J. opt. Sot. rim. 67, IZIU-1217. Wysecki G. and Stiles W. S. (1967) Color Scietlce. Wiley. New York.