British Journl of Ophthlmology, 1984, 68, 33-39 Rod-cone interction in flicker perimetry KNNTH R. LXNDR ND GRLD. FISHMN From the Deprtment of Ophthlmology, University of Illinois ye nd r Infirmry, Chicgo We hve ssessed the influence of the rod system on cone flicker sensitivity during SUMMRY flicker perimetry. For temporl frequencies bove 18 Hz extrfovel cone-medited flicker thresholds for white test stimulus re s much s 1-5 log units lower when mesured ginst lrge bckground light tht sturtes the rods thn when mesured in drkness. Following Gnzfeld blech extrfovel cone flicker thresholds re t their minimum once the cones hve recovered their sensitivity, but then thresholds rise s the rods begin to recover from the blech. Our results indicte tht the flicker sensitivity of the extrfovel cone system t high temporl frequencies is influenced by the rods surrounding the flickering test stimulus. The rods reduce flicker sensitivity mximlly in the drk dpted stte, nd their suppressive influence is minimised only by strong rod bleches or by lrge bckgrounds tht sturte the rod system. Flicker perimetry hs been used to ssess the sensitivity of the humn cone mechnism both in the norml visul system'-3 n%i in ptients with visul bnormlities.' Cone flicker sensitivity is usully mesured by mintining constnt luminnce nd decresing the flicker frequency from high rte until the flicker cn just be detected (the CFF), or by mintining constnt high frequency (usully t pproximtely 25 Hz) while the minimum luminnce necessry to detect flicker is determined. -It hs generlly been ssumed tht the rod system does not influence cone flicker sensitivity t the high temporl frequencies used in flicker perimetry owing to n inbility of the rod system to follow high rtes of flicker. However, recent reports78 hve indicted tht the rod system cn reduce cone flicker sensitivity t high frequencies if mesurements re mde in the drk dpted eye. These studies of rod-cone interction in flicker sensitivity were performed in or ner the fove by mens of chromtic stimuli. We hve exmined the generlity of this rod-cone interction in flicker perimetry by using white flickering test stimulus presented t loctions cross the visul field. The threshold luminnce for detecting flicker ws mesured both in the drk dpted eye nd with rod sensitivity depressed by bckground light or by blech. If the rod system reduces cone flicker sensitivity in the drk dpted eye, s suggested by the Correspondence to Dr Kenneth R. lcxnder, University of Illinois ye nd r Infirmry, 1855 W Tylor Street, Chicgo, IL 6612, US. previous studies,78 then light dpting or bleching the rods should improve cone flicker sensitivity t ll extrfovel loctions. Mterils nd methods SUBJCTS Five nive observers rnging from 21 to 32 yers of ge prticipted in the experiments. ll hd norml (corrected) cuity nd colour vision. During the experiments the subjects' pupils were dilted nd ccommodtion ws prlysed by the instilltion of 1% phenylephrine nd 1% cyclopentolte drops. xcept where noted no opticl corrections were used, since photopic flicker sensitivity t high frequencies is reltively unffected by opticl blur.' PPRTUS Testing ws performed with Tubinger perimeter. The circulr test stimulus (TS) ws either 14 min or 66 min in dimeter nd ws either white (tungsten), red (Oculus glss filter, bndpss >62 nm), or bluegreen (Oculus interference filter, kmx= 5 nm). The TS ws presented ginst the perimeter diffusingsurfce t number of retinl loctions reltive to 3-min dimeter dim red fixtion trget. For fovel testing the fixtion dot ws replced by dimond shped pttern of 4 red dots 1 min in dimeter nd seprted by 2. Under certin of the experimentl conditions described below the Tubinger fixtion chnnel ws used to present bckground fields rther 33 Br J Ophthlmol: first published s 1.1136/bjo.68.5.33 on 1 My 1984. Downloded from http://bjo.bmj.com/ on 31 ugust 218 by guest. Protected by copyright.
34 thn fixtion trget, nd fixtion ws controlled by n uxiliry dim red cross projected on to the pproprite loction within the perimeter bowl. Under these conditions the subject fixted the lterlly plced fixtion trget while mintining forwrd hed position in the chin rest. Temporl squre wve modultion of the TS ws provided by n episcotister rotting in collimted portion of the opticl pth. The TS ws presented for 3 seconds either in the drk, ginst the perimeter Gnzfeld of *5 log cd m-2 or ginst bckground provided by the fixtion chnnel. The luminnce of the TS ws chnged in -1 log unit steps by series of neutrl density filters. For the bleching experiments the subject's eye ws exposed for 3 minutes to Gnzfeld light of 3-6 log cd m-2 provided by Feldmn dptometer (mericn Opticl Co.). Stimulus luminnces were clibrted by Spectr Spotmeter. Flicker frequencies were checked by photocell nd oscilloscope. The photopic nd scotopic vlues of the chromtic stimuli were clculted from the nominl colour temperture of the tungsten 3 2 N- 1 O cr- -J 2 ' II o Bckground on * Bckground off S: N2 S: N3 8 6 4 2 2 4 6 8C Ns I Retinl Temporl ccentricity (Deg) Fig. I Threshold luminncesforflicker detection for3 observers (N1, N2 N3) using 14-min dimeter25 Hz white test stimulus presented either in drkness (filled circles) or ginst Gnzfeld bckground of-5 log cd m-2 (open circles) t the retinl loctions indicted. Kenneth R. lexnder nd Gerld. Fishmn source, the spectrl trnsmission of the vrious filters (s determined by Crey spectrophotometer), nd the photopic or scotopic luminosity function. The clculted photopic vlues greed with the reltive threshold vlues obtined for these stimuli during the cone plteu portion of drk dpttion. The clculted scotopic vlues were equivlent to those obtined from bsolute threshold mesurements. GNRL PROCDUR Prior to n experimentl session ech subject ws drk dpted for 45 minutes. During the session flicker thresholds were mesured by determining the minimum luminnce necessry to detect flicker in temporlly modulted light of constnt frequency (usully 25 Hz). For ech tril the luminnce ws initilly set below the flicker threshold to void flicker dpttion, nd then the luminnce ws incresed in 1 log unit steps until the subject.responded, by mens of buzzer, tht flicker ws detected (scending method of limits). Threshold ws defined s the medin of 3 repetitions of this procedure. Results LIGHT DPTTION ND FLICKR SNSITIVITY The threshold luminnce for 25 Hz, 14-min white flickering TS ws mesured t selected loctions cross the horizontl meridin of the visul field, first in the drk dpted eye, nd then ginst the stndrd Tubinger Gnzfeld bckground of 5 log cd m-2. The results for 3 observers re shown in Fig. 1. In the fove the flicker threshold is reltively unffected by the presence of the bckground field. However, in the retinl periphery flicker thresholds re s much s 1-5 log units lower in the presence of the bckground light thn in the drk. To determine whether the improved flicker sensitivity of the light dpted eye is chrcteristic only t frequency of 25 Hz we mesured flicker thresholds t 2 in the temporl retin t frequencies from 2-5 to 4 Hz in steps of either 2-5 or 5 Hz. The results for one subject re shown in Fig. 2. t low temporl frequencies the bckground field elevtes flicker thresholds considerbly from the drk dpted level s expected from Weber's lw. However, t temporl frequencies bove 18 Hz flicker thresholds re consistently lower in the presence of the bckground thn in the drk. Therefore, the improved flicker sensitivity in the presence of the bckground is not limited to 25 Hz but extends from 18 to t lest 4 Hz Ṫo determine whether cones or rods medite flicker thresholds under these conditions we mesured flicker thresholds t 5 from the fixtion point in the temporl retin for 14-min TS tht ws Br J Ophthlmol: first published s 1.1136/bjo.68.5.33 on 1 My 1984. Downloded from http://bjo.bmj.com/ on 31 ugust 218 by guest. Protected by copyright.
Rod-cone interction inflickerperimetry I-J t fi ti ' U ti ei ti ti 2 < s shown in Fig. 3, flicker thresholds for the 2 chromtic stimuli re virtully the sme ginst the 1 / Gnzfeld bckground, demonstrting tht light dpted flicker thresholds re medited by the cone o,/ system. In the drk dpted eye flicker thresholds for the 2 chromtic stimuli re unequl t both low nd -, / high frequencies. t the lowest temporl frequencies the threshold difference corresponds to tht of the rod system. t slightly higher frequencies the,-2 / threshold difference becomes less thn 2-5 log units, S3 N3 which implies tht cones medite flicker detection for X*Bckground Bockground off still the Rhigher TS while frequencies rods detect (bove the15 flickering Hz) there BG remins TS. t -4 consistent difference between the thresholds for the 2 1 photopiclly mtched stimuli in the drk dpted 5 1 15 2 25 3 35 4 45 eye. Since it is unlikely tht rods re detecting the Frequency (Hz) flicker t these high frequencies (see Discussion), the Fig. 2 Threshold luminncesforflicker detection for one difference in thresholds for the photopiclly mtched 1bserver (NJ3) t the temporlfrequencies indicted using test stimuli is likely to be due to the influence of rods 14 tl4-min dimeter white teststimulus presented t n on cone flicker sensitivity. ccentricity of2 (temporl retin) either in drkness (filled The Gnzfeld bckground light-dpts regions of etircles) or ginst Gnzfeld bckground ofo-5 log cd m-2 retin tht re fr removed from the flickering TS s open circles). well s the retin on which the TS flls. To determine the ctul retinl locus tht must be light dpted to either of middle (BG) or. long (R) wvelength. improve flicker sensitivity, flicker thresholds were Fhresholds were mesured t number of temporl mesured t n eccentricity of 5 ginst stedy white requencies both in the drk dpted eye nd ginst bckgrounds of vrious dimeters nd of the sme he Gnzfeld bckground. The results for 2 subjects luminnce s the Gnzfeld bckground. These bckre shown in Fig. 3. Thresholds re plotted in photopic grounds were provided by the Tubinger fixtion nits so tht if cones lone re detecting the flicker chnnel nd were presented in rndom order. The hen the thresholds for the 2 chromtic stimuli will be TS ws reduced in size to 66 min, nd subjects were qul. If rods lone re detecting the flicker then the opticlly corrected for refrctive error nd trget hreshold for the R TS should be 2.5 log units higher distnce using tril lenses. hn the threshold for the BG TS. The effect of bckground dimeter on flicker Fig. 3 Threshold luminncesfor flicker detection for2 observers (NI, --1 _ 1 1 _ N3) t the temporlfrequencies indicted using 14-min dimeter 25 Hz test stimulus tht ws either of 5-2,( - / Bckground long wvelength (squres) or ffi / ffi ffi T ff on middle wvelength (circles) /.opresented 1 red either in drkness (filled >-3L_..._.._Jred symbols) or ginst Gnzfeld i* 1 blue- bckground ofo-5 log cd m-2 green (open symbols) t n eccentricity of (temporl retin). 4 ~~~~~~~~~~~~~~~~~~~~~~5 35 Br J Ophthlmol: first published s 1.1136/bjo.68.5.33 on 1 My 1984. Downloded from http://bjo.bmj.com/ on 31 ugust 218 by guest. Protected by copyright.
36 2C' 1 2 1 J, S N, o S N4 SN5 ~ oi 2 3 11 Bckground Dimeter ( Deg) Fig. 4 Threshold luminncesforflicker detection for3 observers (N,, N4, N5,) ginst stedy white bckgrounds of vrious dimeters nd luminnce of.5 log cd m-2 using 66-min dimeter, 25 Hz white test stimulus, t n eccentricity of5 (temporl retin). Solid line connects the mens ofthe dt points. thresholds is shown for 3 subjects in Fig. 4. If flicker thresholds re mesured ginst bckgrounds tht re smller thn or equl in size to the TS, flicker sensitivity is equivlent to tht mesured in the drk dpted eye. However, s the bckground is enlrged to light-dpt regions of retin outside the TS, flicker sensitivity improves. Therefore, it is the retin surrounding the TS tht influences cone medited flicker sensitivity. The luminnce of the bckground used to lightdpt the eye ws *5 log cd m-2 (2-3 log scot td on 2. r 1..o - 1. -2 3 I- -5. B--- - - * o S N3 * S: N5 -oo -5. -4. -3. -2. -1.. 1. 2. 3. LOG BCKGROUND LUMINNC (cd/m2) Fig. 5 Threshold luminncesforflickerdetection for2 observers (N3, N5,) ginst stedy white bckground 11 in dimeter nd t the luminnces indicted t n eccentricity of 2 (temporl retin) using 14-min dimeter25 Hz test stimulus. Solid line connects the mens ofthe dt points. Open circles represent detection thresholds for 14-min dimeter, 5 ms durtion, 5 nm test stimulus ginst the sme bckground. Solid line isfit by eye to the dt points. Kenneth R. lexnder nd Gerld. Fishmn the ssumption of pupil dimeter of 8 mm), which is ner the level t which the rod system begins to sturte."' To determine the effect of other bckground luminnces on flicker sensitivity the 25 Hz white TS ws presented in the centre of bckground 11 in dimeter t n eccentricity of 2 in the temporl retin. The bckground ws sytemticlly incresed in luminnce, nd flicker thresholds were mesured t ech bckground luminnce. The results for 2 subjects re shown in Fig. 5 (tringles). s the bckground is incresed in luminnce there is little effect on flicker thresholds until the luminnce is pproximtely -2 log cd m2. The bckground luminnce tht hs the gretest threshold lowering effect corresponds to the Gnzfeld luminnce used in the previous experiments. t bckground luminnces higher thn this vlue flicker thresholds become elevted owing to the incresing light dpttion of the cone system. " lso shown in Fig. 5 is n increment threshold function ginst this sme bckground for 5 nm TS, 14 min in dimeter, nd flshed for 5 ms. ll but the 2 highest dt points re rod determined thresholds s indicted by n bbrevited spectrl sensitivity mesurement t ech bckground luminnce nd by the presence of rod-cone brek in the drk dpttion curves for this TS ginst the vrious bckground luminnces. From Fig. 5 it is pprent tht the bckground begins to elevte rod increment thresholds when it is very dim, yet there is little effect on cone medited flicker thresholds until n J 1 2r -1 _ S N, Bckground on * Bckground off S: N3 fp O 5 1 15 2 Time (Minutes) Fig. 6 Threshold luminncesforflickerdetection for2 observers (N,, N3) following3 min Gnzfeld blech using 14-min dimeter25 Hz white teststimulus presented either in drkness (filled circles) orginst Gnzfeld bckground of.5 log cd m2 (open circles) tn eccentricity of2 (temporl retin). Br J Ophthlmol: first published s 1.1136/bjo.68.5.33 on 1 My 1984. Downloded from http://bjo.bmj.com/ on 31 ugust 218 by guest. Protected by copyright.
Rod-cone interction inflickerperimetry o -J Q -1-2 1-3 I 1 " O L L ~ S * e m * - * ODo @ U * -4 - o q -5 2r 5 1 15 2 25 Time (Minutes) * * 3 35 4 45 Fig. 7 Tringles represent threshold luminncesforflicker detection mesured in drknessfollowing the Gnzfeld blech for 2observers (N,, N,), replottedfrom Fig. 6. Circlesrepresentluminncethresholdsfor2observers (N2, N3) underthesme conditions s in Fig. 6 except tht the test stimulus wsflshedfor5 ms rther thnflickered t 25 Hz. the bckground luminnce ners the level of rod sturtion. BLCHING ND FLICKR SNSITIVITY fter 3-minute Gnzfeld blech, flicker thresholds for 25 Hz 14 min white TS were mesured t 2 in the temporl retin either in drkness or ginst the stndrd Gnzfeld bckground. The results for 2 subjects re presented in Fig. 6. For both subjects flicker thresholds initilly following the blech offset re the sme whether mesured in the drk or ginst the bckground. Thresholds ginst the bckground then remin constnt over the next 2 minutes. In the drk, however, flicker thresholds begin to rise fter 2-3 minutes nd rech n symptote t pproximtely 15 minutes. comprison of flicker drk dpttion with more typicl drk dpttion curve is shown in Fig. 7. o S: Ni. S: N2 O S N3 37 U I pl I I I I I I j The circles represent luminnce thresholds for the 14-min white TS flshed for 5 ms. The thresholds were mesured by n scending method of limits, nd ech dt point represents the threshold vlue for one scending series. The tringles re the flicker drk dpttion functions replotted from Fig. 6. It is pprent from Fig. 7 tht flicker thresholds first begin to rise only during the cone plteu period of drk dpttion. Since the cones hve fully recovered their sensitivity during this period, the elevtion of the flicker thresholds is pprently due to the rpidly recovering rods. There is n symptote in the flicker thresholds t pproximtely 15 minutes following the blech offset before the rods hve fully recovered their drk dpted sensitivity. This implies tht the rods do not hve to be fully drk dpted to elevte cone flicker thresholds mximlly. Furthermore, s shown in Fig. Br J Ophthlmol: first published s 1.1136/bjo.68.5.33 on 1 My 1984. Downloded from http://bjo.bmj.com/ on 31 ugust 218 by guest. Protected by copyright.
38 7, the effect of the rods on cone flicker thresholds begins to lessen only when the blech hs reduced rod sensitivity by 2 log units or more. This is consistent with the finding presented erlier (Fig. 5) tht rod sensitivity needs to be reduced by pproximtely 2 log units by bckground light before cone flicker sensitivity improves. Discussion For temporl frequencies bove 18 Hz cone flicker thresholds in the prfovel nd peripherl retin re lower in the presence of bckground light thn in the drk dpted stte. This result is consistent with previous reports tht the CFF is higher when the eye is light dpted.'2 IS We hve shown (1) tht the reduction in flicker thresholds with light dpttion is miniml in the fove but cn be s lrge s 1-5 log units t greter eccentricities, (2) tht the improved flicker sensitivity with light dpttion does not begin until rod thresholds hve been elevted by t lest 2 log units, (3) tht the effect of the bckground light is mximl t bckground luminnces tht sturte the rod system, nd (4) tht the bckground light must be lrger thn the flickering test stimulus to improve flicker sensitivity. Furthermore, we hve lso found tht, following Gnzfeld blech, flicker thresholds re t minimum once the cones hve regined their sensitivity. Then flicker thresholds become systemticlly rised s the rods recover from the blech. These results re consistent with recent reports of rod-cone interctions in the. detection of high frequencies of flicker.7 In those studies bleching or light dpting the rods ner the fove improved cone flicker sensitivity. Our results suggest tht rods hve significnt effect on cone flicker sensitivity throughout the extrfovel retin under conditions used in flicker perimetry. The influence of rods is minimised only by strong rod bleches or by bckgrounds tht sturte the rod system. There re severl possible wys in which the rod system could interfere with flicker detection in the drk dpted eye. One possibility is tht the rods tht re stimulted by the high frequency flicker my generte temporlly modulted signl tht is time shifted with respect to the cone signl nd impirs temporl sensitivity through destructive interference, s hs been reported t lower temporl frequencies. 14 '" Two considertions suggest tht this is not the cse. First, we mesured flicker thresholds in subject with complete chromtopsi to determine whether the rod system is cpble of detecting high frequencies of flicker under the conditions used in these experiments. lthough this subject hd norml rod flicker response t low temporl frequencies, he ws unble to detect flicker t frequency Kenneth R. lexnder nd Gerld. Fishmn of 25 Hz nywhere in the visul field whether the eye ws light dpted or drk dpted. This result is in greement with previous reports of reduced flicker sensitivity in complete chromtopsi'6 '7 nd is consistent with the finding tht the rod electroretinogrm (RG) in the norml eye does not respond to temporl frequencies bove 18 Hz.'8 Second, it hs recently been demonstrted tht the reduced flicker sensitivity of the drk dpted eye occurs whether the light input to the rods is temporlly modulted or stedy.7 Therefore, even though the rod system of the norml eye cn pprently respond to high frequencies of flicker under specilised conditions,'9 it seems unlikely tht it is temporlly modulted signl from rods tht reduces the flicker sensitivity of the cone system under the conditions studied here. second possibility is tht stedy rod signl generted by the flickering TS reduces the mplitude of the cone signl nd thereby reduces flicker sensitivity. However, if such were the cse, middle wvelength TS should elevte drk dpted flicker thresholds more thn long wvelength TS, since the middle wvelength TS produces lrger rod signl. Yet, s shown in Fig. 3, drk dpted flicker thresholds re higher for the long wvelength TS which is in greement with previous report.21 Second, it is difficult to remove such stedy rod signl, but it is esy to dd to it by superimposing the flickering TS on stedy bckground of the sme dimeter s the TS. s shown in Fig. 4, flicker thresholds ginst bckground of the sme size s the TS re equivlent to thresholds in the drk dpted eye. It is only when the bckground is lrger thn the TS tht flicker sensitivity is ffected. Therefore it ppers unlikely tht stedy signl from the rods underlying the TS is responsible for the reduction of flicker sensitivity in the drk dpted eye t the temporl frequencies used in these experiments. third possible explntion is tht the rods surrounding the TS in the drk dpted eye lterlly inhibit the cones, reducing their flicker sensitivity. Such n explntion ws originlly proposed by Hood2' to ccount for n nlogous finding in the photopic flicker RG of the frog retin, nd hs recently been suggested s n explntion for humn psychophysicl results s well.7 This lterl inhibition is different from tht typiclly encountered in receptive field mesurements22 but is consistent with the finding tht the mximum relese of neurotrnsmitter by the photoreceptors occurs in the drk dpted stte, s discussed by Hood2' nd Goldberg et l.7 In conclusion, when cone flicker thresholds re mesured in the extrfovel visul field of the drk dpted eye, the rods surrounding the test stimulus pper to reduce pprecibly the flicker sensitivity of the cone system. The inhibitory effect of the rods Br J Ophthlmol: first published s 1.1136/bjo.68.5.33 on 1 My 1984. Downloded from http://bjo.bmj.com/ on 31 ugust 218 by guest. Protected by copyright.
Rod-cone interction inflickerperimetry begins to lessen only when rod sensitivity is depressed by t lest 2 log units either by sufficiently intense lrge bckground field or by blech. The influence of the rods is lest nd thus flicker sensitivity is mximl during the first few minutes following blech or ginst lrge bckground field tht sturtes the rod system. In studies of ptients with retinl dysfunction to be reported in subsequent rticle we hve found tht this interction between rods nd cones in flicker perimetry cn be bnorml, suggesting tht comprison of drk dpted nd light dpted flicker thresholds my be useful test of visul function for the extrfovel retin. Supported in prt by NIH Trining Grnt Y738, NIH Core Grnt Yt)17792, nd grnt from the Ntionl Retinitis Pigmentos Foundtion. We thnk Drs T. Frumkes, R. Mssof, nd M. F. Goldberg for helpful comments on the cxperimcnts nd mnuscript, nd Ms M. Gcrc for editoril ssistncc. References I Wolf, Gct M, Gcer S. Criticl flicker frequencics in flicker perimetry: stndrds ind confidence limits. rch Ophthltnol 1968; 8: 347-5 1. 2 King-Smith P, Crden D. Luminnce nd opponent-color contributions to visul dctection nd dpttion nd to temporl nd sptil integrtion. J Opt Soc m 1976; 66: 79-17. 3 Fitzkc FW, Mssof RW. bsolute cone thresholds derived from the Fcrry-Portcr lw. Invest Ophthlnol Visul Sci 198; 19 (RVO suppi): 212. 4 Cmpbell CJ, Rittlcr M. The dignostic vluc of flickcr perimetry in chronic simpie glucomt. Trts in cd Ophthlmol Otolrvnigol 1959; 63: 89-98. 5 Mssof RW, Fleischmn J, Finc SL, Yoder F. Flickcr fusion thrcsholds in Bcst mcultr dystrophy. rch Ophthlmnol 1977; 95: 991-4. 39 6 Mison RJ, Sneiger RS, Foster DHI, Heron JR, Jones R. bnormlitics of chromtic nd luminnce criticl flicker frcquency in multiple scierosis. Invest Ophthlhnol Visul Sci 1982; 23: 246-52. 7 Goldberg SH, Frumkes T, Nygrd RW. Inhibitory influencc of unstimultcd rods in the humn rctin: vidence provided by exmining cone flicker. Sc-ietice 1983s; 221: 18)-2. 8 Colett NJ, dms J. Rod-cone interiction in flickcr detection. Invest Ophthltnol Visul S(ci 1983; 24 (RVO suppl): 187. 9 Kelly DH. Flickering pitterns nd lterl inhibition. J Opt Soc in 1969; 59: 1361-7. 1i) guilr M, Stiles WS. Sturtion of the rod mcchnism of the retin t high levels of stimultion. Optic ct 1954; 1: 59-65. 11 Keily DH. dipttion effects on sptio-temporl sine-wivc thresholds. Vision Res 1972; 12: 89-1)1. 12 Lythgoe RJ, Tinslcy K. The reltion of the criticll frcqucncy of flicker to the dptttion of the cyc. Proc R Soc Lotid 1929; 15: 6-92. 13 Hylkem BS. Fusion frcquency with intcrmittcnt light under vrious circumstinccs. ct Ophthhlnol 1942; 2: 159-1811. 14 McLeod DI. Rods cnccl concs in flicker. Nture 1972; 235: 173-4. 15 Vn dcn Bcrg TJTP. Spekrcijsc H. Intcrction betwecn rod ind conc signls studied with tcmporll sinc wve stimultion. J Opt Soc n 1977; 67: 1211)-17. 16 Skottun BC, Nordby K, Mgnussen S. Photopic ind scotopic flicker scnsitivity of t rod monochromit. Invest Ophtlm1Zol Visul Sci 1981; 21: 877-9. 17 vn der Twccl LHI, Spekrcijsc HI. Psychophysics ind clcctrophysiology of rod monochromtt. Doc OphthIlhnol Proc Ser 1973; 2: 163-73. 18 Sokol S, Riggs L. lcctricl ind psychophysicl rcsponscs of the humn visul system to periodic vrition of luminnce. Invest Ophthlmnol Visul Sci 1971; 1: 171-81). 19 Conncr JD. The temporl propertics of rod vision. J Phsysiol (Lotud) 1982; 332: 139-55. 2( Mrks L, Bornstcin MH. Spcctrll sensitivity by constnt CFF: effcct of chromitic dpttion. J Opt Soc in 1973; 63: 221-6. 21 Hood DC. Supprcssion of the frog's conc systcm in the drk. Vision Res 1972; 12: 889-91)7. 22 Brlow HB, FitzHugh R, Kufflcr SW. Chingcs of orgnizition in the rcccptivc ficids of the cit's retin during dirk idpttion. J Physiol (Lond) 1957; 137: 338-54. Br J Ophthlmol: first published s 1.1136/bjo.68.5.33 on 1 My 1984. Downloded from http://bjo.bmj.com/ on 31 ugust 218 by guest. Protected by copyright.