Light and Dark Adaptation - Semantic Scholar · Light and dark adaptation W. A. H. Rushton-he nature of visual adaptation is an extremely complex question and our un-derstanding of

Volume 11Number 6

Visual-oculomotor connections 503

The high-frequency cutoff of the semicircularcanals is unknown but probably also lies above5 Hz. The high-frequency cutoff of the extra-ocular muscles is 1 Hz. so that central mechanisms(that introduce phase lead) must compensatefrom 1 to 5 Hz. The pursuit system begins tocut off between 0.5 to 1 Hz. so that it can com-pensate for slow head movements while thevestibular system compensates for rapid headmovements.

RATLIFF: In some cases your integrator seemedto be ideal and in other cases leaky.

ROBINSON: The integrator is not likely to beideal. Its time constant (of the leak) need onlybe larger than a typical slow phase duration ofnystagmus, say 2 to 5 sec. Its time constant maybe revealed by the fact that, on eccentric fixa-tion in the dark, the eyes tend to drift backtoward the primary position at a velocity whichsuggests a leak time constant of 10 sec.

RODIECK: Isn't it curious that in an animal witha fovea one finds less directional selectivity incolliculus and lower visual centers than one findsin afoveate animals?

ROBINSON: Yes. In lower animals it may be thatdirectionally selective units form a special subsetof ganglion cells that are used to stabilize the eyewith respect to the visual world but do not par-ticipate in form discrimination. Lack of such cellsin foveate animals would suggest that this system

has been abandoned in favor of an equivalentsystem which utilizes the more complex process-ing of the visual cortex. This observation is aserious problem in trying to draw an analogy be-tween Collewijn's model for the rabbit's ocularstabilizing system and the smooth pursuit systemof man.

LEVICK: The output of these directional selectiveunits is a signal which corresponds to the velocityof the retinal image. So that if you really wantthe system to stay put you can't use these unitsbecause they don't go down to dc. Couldn't youconsider the concentric units which do giveyou absolute position?

ROBINSON: The directionally selective on-unitsin the rabbit detect velocities of 0.01 degrees persecond and stabilize the eye drift to comparablevalues. If the drift disturbance were unidirectional(it usually isn't) the eye would travel six minutesof arc in ten seconds or about 0.5 degree in aminute. Since the drift disturbance is not unidirec-tional, the long-term drift of mean eye positionis so slow that it would be difficult to distinguishit from a true dc mechanism. All one can sayfrom Collewijn's experiments is that the direc-tionally selective units appear to be capable ofexplaining the observed behavior. This, of course,does not exclude the possibility of an absoluteposition system based on concentric units.

Light and dark adaptation

W. A. H. Rushton

-he nature of visual adaptation is anextremely complex question and our un-derstanding of it received a mighty set-back when Hecht and Wald used theiroutstanding force and eloquence to insistthat everything depended simply upon thelevel of rhodopsin in the rods and thephotopic pigments in the cones. Thus alldesensitization by light was called 'bleach-ing' and all recovery 'regeneration/ I am

From the Institute of Molecular Biophysics, FloridaState University, Tallahassee, Fla.

sorry to note that American psychologistsand others still use this misleading termi-nology and there has been widespreadbelief that it rested upon experimentalproof, though in fact it was a lively specu-lation which not only was unsupported byany measurements of pigment kinetics,but which resolutely ignored the work ofDartnall, Goodeve, and Lythgoe (1938)who had made those kinetic measurementsin the '30s and proved the photochemicaltheory of visual performance to be un-tenable.

In fact the vision of everyday life is

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/932871/ on 06/06/2017

504 Rushton Investigative OphthalmologyJune 1972

BOUGUER (1760) FECHNER (1860)

= K

Fig. 1. Simultaneous contrast of Bouguer whereone half of the bipartite field is AI "brighter"than the other. Also the successive contrast ofFechner where AI is suddenly added at centerof uniform field I.

very little affected by the pigment levelin the rods and cones; it depends in themain upon an adaptive nerve organizationin the retina which we shall now consider.

Parti : Steady backgrounds

If the eye is adapted to a luminance Iand must discriminate a region of lumi-nance (I + AZ), the just noticable differ-ence Al has long been known to be givenapproximately by the relation Al/I = K.This was studied by Bouguer (1760) ina bipartite field (Fig. 1), and by Fechner(1860) a century later using a superposedflash. These conditions are often treatedas though essentially identical, but physi-ologically they are entirely distinct. TheFechner relation can be considered physi-ologically as detection of the increasedoutput of the receptor or the receptor or-ganization when the steady light I is sud-denly increased by A/. In principle thediscrimination could be made by a singlereceptor. In the Bouguer relation, it mustbe the separate outputs from receptors indifferent parts of the retina that are com-pared. To be sure, when we examine therelative brightness of the two halves ofa bipartite field, we often scan across theboundary line and thus discriminate partlyby Fechner's successive contrast and partlyby Bouguer's simultaneous contrast. How-ever, the Bouguer situation may be ob-

tained uncontaminated if we use a stabil-ized image (as described later) or by themethod of Cornsweet and Pinsker (1965)where small fields were briefly presentedsimultaneously on either side of the cen-tral fovea and judgments made as to whichwas brighter. They showed an interestingdifference between Bouguer's Law andFechner's. With Fechner the exact law isclose to

where ID is Fechner's eigengrau ( = "darklight" or receptor 'noise'). With Bouguer thesimple ratio Al/I was found to hold downto such a low illumination level that furthermeasurement was impossible.

The site of Fechner scaling. Thoughthe rise of Fechner threshold might bedue to desensitization of the rods them-selves (as was supposed by the Hecht-Wald theory), in fact it is primarily due todesensitization of the rod community. Theplainest evidence (Rushton 1965a) is thata background lasting 2 sec. which raisesthe rod threshold 3 times is so weak thatless than 1% of the rods catch a singlequantum from it during the 2 sec. of ex-posure. So the 99% of rods that never'saw' the background (since they did notcatch a single quantum from it) had theirthresholds raised 300% by activity of the1% of rods that did see it. Thus the Fech-ner gain organization is proximal to therods. But it is not proximal to the ganglioncells since Donner (1959) has obtainedthe usual Fechner relation by recordingfrom ganglion cells when minimal flashesare superposed on various steady back-grounds. Horizontal cells have always beenthe obvious structures to carry out Fechnerscaling, they are extensive enough to re-ceive many signals even when only 1% ofrods catch one quantum, and their con-nections ramify to an immense numberof receptor-bipolar junctions. Moreover, ascaling mechanism whose object is to dealwith the immense range of light inputthat in fact is handled, ought to start itsscaling as early as possible along the trans-


Volume 11Number 6

Light and dark adaptation 505

X

(b)

(a)Fig. 2. (a) Geometry of contrast flash presentation. Test flash of intensity X falls upon a 2°circle 6° from fixation point (F.P.). The contrast flash of intensity <f> falls on the 8° surroundannulus with 2° black center. These flashes may fall upon steady backgrounds X on # and <Pon 0. (b) Schema of neural interaction. Each region generates from the flash (X or <j>) apulse which is attenuated by the G.-box in proportion to the total steady background (f> + MD),(0 + 0D) (where J«D, #D is Fechner's eigengrau). If X and /* are kept fixed, then the inhibitorysignal N must remain fixed when the X flash is reduced just to threshold. In this way wemay vary <P and 6 for a constant N and establish the relation between them. This interactionpresumably all takes place in the amplitude-coded range of signals before the difficult trans-formation into spike frequency in ganglion cells.

mission line to anticipate overloading. Foreight years I have hoped that the hori-zontal cells are Fechner scalors, and I stillhope so.

Retinal nerve signals. The astonishingrange and accuracy of Fechner scaling canbe shown by the technique of contrast-flash analysis (Alpern, Rushton, and Torii1970a,b,c,d). A surround flash <£ falling up-on an annulus, Fig. 2(a), inhibits the signalfrom a flash A falling upon the dark centerso that its intensity must be increased ifit is still to be visible. Whatever the classof receptor excited by A at threshold (e.g.,rod, red cones, etc.), it is only the quantumcatch from </> by that receptor which raisesthe threshold (Alpern and Rushton 1965).Thus if A lies below cone threshold, the Athreshold will be raised only by the excita-tion of rods in the surround no matter howstrong <£ may be. This allows us to mea-sure rod excitability far above cone thresh-old and to determine the course of roddark adaptation in a normal subject over6 log units range of sensitivity change,

as will be mentioned later (Fig. 5). If thesurround flash does not fall upon the en-tire annulus but upon four sectors of 11^°(like windmill sails) so that only Vs ofthe area of the annulus is illuminated,then the inhibition is reduced to Vs. Thatallows us to measure N, the size of thenerve signal as a function of the intensityof flash. The argument given in Alpern,Rushton, and Torii (1970a) is too long tobe repeated here and we simply state therelation found.

When in 1968 we first determined thisrelation the result surprised us, but it fitswell with recent electrophysiology. If Nis the size of nerve signal and I the in-tensity of flash in the absence of steadybackground we found experimentally that

N =I + <r

(2)

From this formula it is clear that when I =oo, N = 1 hence N is expressed as a frac-tion of its maximum value. When I = a,N = Vz and thus <r is called the "semi-



10

0-8

0-6

04

0-2

A « x ' *

rx-rvK-f — $ * f <

t

- 6 0 -4-8 -2-4 -1-2- 3 6Log/

Fig. 3. Amplitude of turtle cone potentials intracellularly recorded as a function of log lightintensity. Dotted curve plots the relation V/V,nax = 1/(1 + ff), where log a = - 3.8. (FromBaylor and Fuortes 1970.)

saturating constant," i.e., the light flashthat raises N to half its maximum. Forhuman rods a is about 1000 quanta ab-sorbed per rod per flash. This flash bleachesabout 1/10,000 of the rhodopsin.

The absolute threshold for human rodsis about 10"5 of the a value (i.e., 1 quantumcaught per 100 rods) and thus the rod nervesignal, N, increases in direct proportion tothe flash from threshold to 4 log units aboveit: at higher levels it begins to saturate.Cones behave like rods, but thresholds andsemisaturation are both about 2 log unitshigher in light level.

The hyperbolic formula (eqn 2) whichmay be called an H-function, was put for-ward by Naka and Rushton (1966a, b) todescribe the relation between light intensityand S-potentials. It has since been shown tounderlie many of the "amplitude-coded"nerve signals of the retina (e.g., Baylor andFuortes [1970] for turtle cones intracellu-larly recorded, Fig. 3). If we regard eqn(2) as the synaptic transform function forthese amplitude-coded signals, a simple ex-pectation follows from the mathematics.

If the "H curve" output of one synapse

(with some fixed attenuation) is the inputto the next synapse, then its output is easilycomputed and found also to be an H curve.Consequently with a chain of such synapsesthere will be an H relation between the in-put at one end and the output at the other.

Effect of background on N. This was in-vestigated over a very great range by al-lowing the inhibitory flash <£ to fall upona steady background 9. To inhibit a testflash A of some fixed size, the relation found

was

N = (3)

where 9D is the same as ID eqn (1) and isFechner's eigengrau. The formula heldgood over the whole available useful rangeof 9 intensities (about 5 log units) and fora large range of N-values obtained by usingdifferent intensities of A., the test flash to beinhibited.

Fig. 4 shows the size of N plotted againstlog <£ the inhibiting flash, which in theupper curve fell on a dark background(log 9 = - oo ) and in the others the log


Volume 11Number 6


log <p td sec

Fig. 4. Nerve signal N plotted experimentally as function of inhibitory flash intensity <P, andsteady background intensity 0 upon which it falls. H? curves on left plot log N (scale onleft); Hi are the same curves plotted simply as N (scale on right). All curves are the the-oretical expectations of eqn. (3). (From Alpern, Rushton, and Torii 1970b.)

background was as marked. In the upperleft (H2) curves log N is plotted, in thelower right (HJ , N itself. All the curvesare H curves with o- unaltered by back-ground adaptation. In the H2 plot, thecurve is displaced vertically downward bybackgrounds. In other experiments purebleaching was found to displace the curvenot downward but horizontally to the right.

In many cases it is not easy to tellwhether adaptation is in effect by back-ground or by bleaching (possibly what iscalled 'nervous' or 'chemical'). It looks asthough the direction of shift of the H2

curve of the nerve signal resulting shouldtell us; the vertical shift is the 'nervous'part, the lateral shift is the chemical.

A pleasing result from eqn (3) is that itprovides a quantitative explanation of thefamous "rod saturation" of Aguilar andStiles (1954). For any fixed small value of

N say 1(H (= 100 times the dark threshold)</> will be negligible compared with <r un-less 6 is very large, and eqn (3) will be-

comeNo-

which is identically Fechner's formula (1)with K evaluated as Na/0D. But as 6 con-tinues to increase <f> must also increase untilit no longer is negligible compared with o-.From our fundamental equation

N =e0 + e

(3)

it is clear that however great <f> becomes, Ncan never exceed 0D/(#D + 0) and when,with increasing 6, this falls below the Ncriterion (10~3 in the present case), ob-viously no increase in <f> can restore it. Onthe right of Fig. 5 are displayed log incre-



0 20Minutes in the Dark9 0 % 6 Bleach

-ooAfter Log 6 Scotopic td

Fig. 5. Left, 3 dark adaptation curves, right, 3 increment threshold curves for normal humanrods determined by contrast flash technique. For each curve some fixed supra-threshold flashXi applied to the center (see Fig. 2) was inhibited by the surround flash <t>. Ordinates plotthe least value of log <P which just inhibits Xi when <f> falls on various steady backgrounds 0(curves on the right). The relation between log <P and log 0 corresponds to Aguilar andStiles' saturation, and the curves plot exactly the relation of eqn (3) with log 0o = - 1.9(td) and log <f = 3.3 (td. sec). Curves on the left plot similar results where there is nosteady 0 background, but the region had been 90% bleached at time zero. Contrast flash tech-nique when K lies below cone threshold measures only rod excitation by <£, thus we obtaina pure rod dark adaptation curve over a range of about 6 log units without cone interference(From Alpern, Rushton, and Torii, 1970c.)

merit threshold curves where <j> falls onvarious backgrounds 6 and the criterion for<f> 'threshold* is that the <f> signal after at-tenuation by the 9 background should justextinguish the perception of the fixed Xflash (at the center of the annulus). Thethree curves were obtained with 3 differentlevels of X. What the experimental pointsshow (right half of the figure) is first thefamiliar Fechner log increment thresholdcurve, but at high values the Aguilar andStiles saturation develops. The theoreticalcurves of the figure are simply the plottedexpectation of the fundamental formula (3).All the curves have the same eigengrau-19 log td.) and the same semisaturationo- (3.3 log td. sec), and indeed the whole

family is simply a replot of the results ofFig. 4, which also displays the fundamentalrelation of eqn. (3).

Now equation (3) which implies forrods the saturation which Aguilar andStiles saw was there, equally implies it forcones where they saw that it was not there.Fortunately the explanation is simple. Withcones the a level is some 2 log units abovethe a level for rods, moreover with thesame illumination cones bleach much fasterthan rods do. Consequently, in attemptingto reach the cone a level of quantum catch-ing from a steady background, so muchpigment is bleached away that insufficientremains to achieve the required quantumcatch. Alpern, Rushton, and Torii (1970d)


Volume 11Number 6


not only examined this quantitatively andexplained how in these conditions no steadybackground could saturate the cones, theyalso showed how cone saturation couldnevertheless be displayed. For, if the back-ground I instead of being continuous, shinesonly for a second or so at a time, just whilethe increment Al is flashed upon it, thebleaching produced will be slight and theFechner threshold rise can be observednearly uncontaminated by any impoverish-ment of quantum catching through bleach-ing. When this experiment was done thecone saturation was seen to occur as ex-pected and eqn (3) was followed for cones,as in Fig. 5 it is for rods, but naturally theconstants 0D and a now had values appro-priate to cones not rods (Alpern, Rushton,andTorii, 1970d).

Brightness adaptation. The N-signalwhich we have been considering is an in-hibitory signal—one that is just sufficientto reduce to invisibility a fixed strong testflash A. Now Alpern and Rushton (1967)showed that nerve signals were adaptedby steady backgrounds in the same waywhether their destination was to inhibit orto excite. This suggests the question, "Doequal N signals look equally bright?" Sup-pose upon the retina there are two steadybackgrounds of luminances 61} 62, what willbe the energies <f>1} <f>2 flashed upon themwhen those flashes appear equally bright?Will they accord with eqn (3) so thatbrightness is equal when N is equal?

There is a difficulty in making this kindof observation because it is hard to judgethe equality of flashes when seen in con-trast to their unequal backgrounds. Butthis difficulty can be overcome if the back-ground is stabilized on the retina. As hasbeen recognized since the pioneer experi-ments of Ditchburn and Ginsborg (1952)and Riggs, Ratliffe, Cornsweet, and Corn-sweet (1953), images that are stabilizedand thus remain fixed on the retina despiteeye movements, quickly fade away and arenot seen at all. Flashes applied to the tworegions in this condition are therefore pro-jected upon an empty field and thus may

be directly compared in brightness withoutcontamination from background contrast.This experiment has recently been per-formed by Mrs. Cynthia Gosline and me(unpublished) in the University of Cam-bridge. The subject (C. G.) wore a special-ly worked tight contact lens that carried ona stalk the target and a lens to focus it onthe retina.

The principle of the experiment is indi-cated in Fig. 6. The 4° stabilized field con-sists of an upper half of luminance I, anda lower half JV After stabilized fading hasoccurred and the field is empty, the lever ispressed and suddenly Ix is replaced by thebrighter I2 and Jt by the brighter J2. Theaim of the experiment is to find what is theluminance ratio I2/J2 that makes I2 and J2

look equally bright when they suddenlyappear in the empty field.

We shall not digress here to describe theoptics in detail, how the change is madefrom Ix, Ji to I2, J2 and how the ratio I2/J2

is adjusted. In principle, light falls on theglass plate at the polarizing angle thuspressing the level and rotating through 90°the attached sheet polarizer switches thelight from beam 1 to beam 2. Thelower aspect of Fig. 6 shows the eye,with tight contact lens holding on the

stalk a target -| with the lens L4 to focusit onto the retina at U3. The target con-sists of -|, a strip of half-wave cellophanethat covers half of the target circle withhorizontal edge sharply in focus. In con-junction with P, the sheet polarizer withaxis at 45° to that of the cellophane wehave in effect two polarizers with axes atright angles. Horizontally polarized lightfalling on the target illuminates only theupper half I, vertically polarized lightilluminates only the lower half J. Lightpolarized at an angle a to the horizontalilluminates I in proportion to cos2 a, J inproportion to sin2 a. Thus by adjusting a,the plane of polarization of beams 1 and 2we may set the ratio I/J as required.

Fig. 7 displays the results of one experi-ment out of many performed by Mrs.



beamN 'splitter

COMMONBEAM

Fig. 6. Optical equipment to replace the 4° bipartite field I,/Ji by I2/J2. When the lever isdepressed it rotates the sheet polarizer P through 90° so that light falling on the glass plateat the polarizing angle, instead of being reflected as beam 1 is transmitted as beam 2. Thelower part of the figure shows the observer's eye wearing a tight contact lens with a stalkthat holds the target X/2, a half-wave cellophane strip whose horizontal edge is focussedsharply by L4 onto the retina at U3. In conjunction with the sheet polarizer P, X/2 acceptsonly horizontally polarized light in the upper half-field I and only vertically polarized lightin the lower half-field J. Thus in changing from beam 1 to 2 we can alter the total intensity(I + J) by adjusting the neutral filters Fi, F2, and the ratio I/J can be altered by ad-justing the plane in which the beams are polarized, easily set by rotating the half-wave

plates (-^ ) 1 and ( ̂ j 2, which causes the polarization plane falling upon the stabilized target

to rotate through twice that angle.

Cynthia Gosline (subject) and me in theZoological Laboratory of the University ofCambridge. After proper alignment of thetarget on the eye's optic axis and of the eyein the equipment, the subject relaxed andthe two fields Il5 Ji in Fig. 6 faded awayand left emptiness. Then the subjectpressed the lever (which replaced Ix by Land Ji by J2) and quickly released it again.This caused the empty field to be replacedby two contiguous fields, and the subjectsaid which was brighter. The operator thenaltered the polarization plane in beam 2

(by rotating the half-wave plate (4)2) togive a new I2/J2 ratio and the procedure

was repeated. The subject never knew whatchanges were being made and generallywas consistent in log (I2/J2 judgmentscorrect to ± 0.1 in a total log luminancechange of 24 or more. The judgmentswere always made for an increase in bright-ness of the fields. With a decrease, the newfields both looked black and their equalityin blackness was hard to judge.

Fig. 7 shows one critical experiment re-peated on 4 days. A 0.8 neutral densitycovered the J half of the target and inconjunction with suitable polarization ofbeam 1, the adaptation ratio li/]i wasalways maintained so that log (Ii/Ji) —


Volume 11Number 6


1-4. Though the ratio I]/Jx was always thesame (about 24) the actual luminancefor adaptation was varied between experi-ments as shown by abscissae in Fig. 7 fromcomplete darkness (- oo ) to about 1000 td.for Ix. The changeover was always made tofull brightness where I2 was about 10,000td. and J2 was adjusted to look equallybright.

The results of the repeated experimentsof Fig. 7 (four different symbols) are rea-sonably consistent and run an expectedcourse. When the adapting light is black-ness (- oo) no differential adaptation toIi/Ji could be expected and I2 will appearas bright as J2 when it has the same lumi-nance. Consequently when log Ix = - oothe ordinate log (I2/J2) should be zero,which it is. On the other hand, if the adapt-ing brightness is so high as to be well abovethe eigengrau, we should expect I2 to bescaled down in proportion to Ix, J2 in pro-portion to Jx so that at top right of thecurve the ordinate, log (I2/J2) should ap-proach log (I1/J1) = 14. This also is ful-filled.

Both these results are consistent with thestatement "equal N values are seen asequally bright." In fact the theoreticalcurve (no arbitrary constants) that is drawnthrough the points of Fig. 7 is the mathe-matic expression of that statement.

In the application of the fundamentalformula (3)

1.5,—

- I 0 iogl0 Ilog 11 (trolonds)

Fig. 7. The 4° bipartite field I,/J, was presentedstabilized with log (L/Ji) = 1.4, i.e., L 24 timesthe luminance of Ji. When the stabilized imageshad faded h was increased to I, (about 10,000td.) by pressing the lever, and Ji was increasedto J2 that was adjusted by trial to appear thesame brightness as I2. The experimental points(4 experiments) show for each initial luminancelevel log It (abscissae) what is the ordinatevalue log (I2/J2.) of the flashes that appear equallybright. The curve through the points is that de-rived from eqn (3) on the assumption that equalN signals should look equally bright.

N =<f> + a + 0

(3)

to the half-field h (or the half-field Jx) weput 9 = Il3 0r> = ID, <f> = 1,-1,, < a Thisgives

N, =1,-1, I D

ID + I,(4)

with a similar expression for Nj in terms ofJ. If the condition that flashes appear equal-ly bright is that Nx = Nj we may equatethe expressions (4) and since ID = JD

I» + ID _ J2 + I

I. + ID JI + ID(5)

This is a very simple extension of Fechner'sconcept. The eigengrau ID must always beadded to the real light I or J to which theeye is adapting, and with this concept oftotal light, the N signal is proportional tothe change in total divided by the old total.Equal N signals look equally bright.

In Fig. 7 the theoretical curve was de-rived from eqn. 5, the intensities of I2 andJ2 being well above eigengrau level so thatwe may write

log (IS/JO = log (I, + ID) - log (J, + ID) (6)

In the experiment of Fig. 7 the value of IDwas obtained by ordinary increment thresh-



old measurements where an auxiliary testflash was presented superimposed uponeither field Ii or Jx at the various levelsindicated by abscissae. These incrementthreshold results for backgrounds Ii areshown by black circles at the bottom rightaspect of Fig. 7. The curve shifted 1-4 logunits to the right represents results forbackgrounds Jt. According to Fechner'sformula

log AI = log (I, + ID) + K (upon I,)and

log AI = log (Ji + ID ) + K (upon Jt)

for the two backgrounds. The ordinate dif-ference between the two curves conse-quently is

A y = log (L + ID) - log (J, + ID)

But this from eqn. (6) is precisely log(L/Jo), and the curve through the 4 setsof points in Fig. 7 is in fact simply theordinate difference A y of the lower curvesreplotted. The scale of A y is shown on theright. It is enlarged when replotted tomatch the scale on the left, and with thisalteration, the ordinate of the upper scalemeasured from its zero is equal to theordinate difference of the two lower curves.

The size of amplitude-coded signals. Ourfundamental equation (3) that relates N,the size of signal with <f> the flash energy,and 0 the steady background upon whichit falls appears to be of great importancefor vision.

First it is an experimental relation thatcan be quantitatively checked and has beenfound to hold over the entire useful range—about 5 log units. It may be proved thatat visual threshold N has a fixed size, andthis allows various psycho-physical rela-tions to be predicted from the formula,including the Aguillar and Stiles (1954)saturation, with satisfactory agreement.Two N signals calculated to be equal fromthe formula were found to have theseproperties. If inhibitory, they inhibitedequally so that in the center-surround con-trast flash situation of Fig. 2 they reduceda supra-threshold flash (X) equally just tothreshold. If the N signals were excitatory,

they excited equally so that when imping-ing upon the empty field of a stabilizedimage the two flashes appeared equallybright.

The formula of eqn (3) implies a re-markable organization of signals with goodaccuracy over a great range. It is likely thatmuch of this occurs in the intimate con-nections between receptor terminals, bi-polars, and the processes of the horizontalcells. Expert analysis of this important andcomplex region is proceeding in manylaboratories at present, and we shall soonknow how far the transformation of infor-mation accords with the expectation of eqn.(3), and what remains to be explained byother mechanisms.

Part 2: Bleaching

The slow recovery of visual sensitivity ongoing from bright sunlight into the dark-ness of the tomb was no doubt familiar tothe builders of the Pyramids. Its explana-tion began when Kiihne showed in thefrog's retina that rhodopsin was bleachedby light and in situ regenerated slowly inthe dark. Hecht was correct in his conjec-ture that the dark adaptation curve wascorrelated with the regeneration of rho-dopsin, but he was never able to decidewhat the relation was. His insistence uponthe hopeless attempt to make a singlekinetic system explain two such utterlydivergent adaptations as that from bleach-ings and that from backgrounds has thrownthe whole subject into a lasting confusion.

Bleaching and threshold rise. When thepigments were objectively measured byretinal densitometry in the eyes of livingmen, it was found that rhodopsin regen-erated along an exponential curve with thesame time constant as the rod branch ofthe dark adaptation curve (Campbell andRushton 1955), and that cone pigmentsregenerated similarly along the cone branch(Rushton 1957). Dowling (1960) confirmedthis relation more accurately by measuringin the rat the log ERG threshold and cor-relating it with the extracted rhodopsincontent of the rods, reduced either by


Volume 11Number 6


• A Pigmento Recovery of Interaction 20

10 20 30Minutes in the Dark After Full 8 Bleach

Fig. 8. Rod dark adaptation curve (circles) and the regeneration of rhodopsin in the normalhuman eye (triangles). The pigment was measured by our Florida densitometer (Hood andRushton 1971) from the 90% bleaching level at zero time, till complete recovery. The dark-adaptation log thresholds are a replot from Fig. 5 (left) where the contrast flash techniqueexcludes cones from contaminating the rod threshold curves. Since with suitable scalingboth results fall on the same curve the log threshold rise is proportional to the fraction ofrhodopsin in the bleached state. (From Alpern, Rushton, and Torii 1970c.)

bleaching or by vitamin A starvation—re-sults which coincided, surprisingly enough,despite the two exceedingly differentphysiological conditions in the rods. Thesame relation to bleaching was found in a"rod monochromat" (Rushton 1961) wherethe absence of rapidly adapting cones al-lows the rod threshold to be followed overa range of 6 log units.

All these techniques established the samerelation, namely, that the log threshold israised in proportion to the fraction of rho-dopsin in the bleached state at the time.

log (Al/AIoj = 10 HB (4)

where A I is the threshold during darkadaptation when a fraction B of pigment isin the bleached state, Al0 is the fully darkvalue, and H is a constant being the rise inlog threshold when 10% of the pigment isbleached.

There is little doubt that this holds for

human rods for all times longer than 5minutes in the dark. Fig. 5 shows an un-usual example, for it records the darkadaptation following a full rhodopsinbleach in a normal subject with no conebranch to the curve. The dark adaptationcurve is displayed on the left of Fig. 5where the contrast flash technique elimi-nates the intrusion of cones. Fig. 8 showsthe coincidence of these log rod thresholdsreplotted from Fig. 5 with the rhodopsinlevel in the rods measured on the samesubject by densitometry in a comparablerun. Triangles show two runs of pigmentregeneration over the whole range; circlesshow the log threshold over the later halfof the range. I do not think it possible toelicit a human rod response with more than50% of the pigment bleached. If this isdue to 'saturation' it is not at all theAguilar and Stiles saturation as may be seenfrom comparing the left and right halves of



pigmentstill bleached

TIME (MIN)

90 %recovery

16 20

Fig. 9. Human rod dark adaptation curves following a 40 sec. exposure that ended at zerotime on the figure and left 7%, 2%, 0.5%, and 0.1% of rhodopsin then bleached. Thecontinuous curves show the log threshold rise expected of that amount of bleaching on thebasis of Fig. 8. But early after weak bleaches the threshold is always raised much more thanthis expectation. (From Rushton and Spitzer-Powell 1972b.)

Fig. 5. More likely the sharp threshold risewithin the first 5 minutes is related to therapid adaptation found already by Hechtfollowing weak bleaches. He and especiallyWald supposed that this represented arapid regeneration of bleached rhodopsin.But that misapprehended the problem. Theproper question was not why after thisweak bleach does the threshold return soquickly toward the base line, but why hasit ever left the base line so far?

Early threshold rise. Fig. 9 shows a setof 4 human dark adaptation curves follow-ing 40 second exposures to lights that werefound to bleach fractions of rhodopsin indi-cated. The results in this figure show someof a set (Rushton and Spitzer-Powell 1972a,b) where the pigment regeneration andthe dark adaptation curves were measuredfollowing the whole range of bleachings.The continuous curves of Fig. 9 show thecourse of dark adaptation to be expectedfrom the regeneration of rhodopsin follow-

ing the weak bleaches indicated. Withstronger bleaches than these, the thresh-olds were found to follow these continuouscurves exactly, but in Fig. 9 the experi-mental points start above the curves andonly reach them after five minutes. Theearly rise becomes greater the greater thebleaching light and perhaps drives theearly threshold of Fig. 5 above the picture.

This early threshold rise might be back-ground adaptation from the strong back-ground level of the bleaching light, achange in nerve organization so great thatit takes a few minutes for complete restora-tion. Or it might be the result of photo-products not yet degraded to leave freeopsin (which is what the "rhodopsinbleaching" lines of Fig. 9 represent).

Three lines of evidence support thephotoproduct explanation, (i) Alpern andTorii in preliminary experiments (unpub-lished ) found with contrast flash techniquethat the H2 curve (Fig. 4) relating N and


Volume 11Number 6


</> was displaced not downward as expectedof background adaptation, but laterally asexpected of bleaching adaptation, (ii)Rushton and Spitzer-Powell (1972b) foundthat when the bleaching intensity I and theexposure duration t were varied so that Itthe total energy was constant the adapta-tion was the same from t = V2 second tot = V2 minute. This is what would be ex-pected of a photoproduct, but not of anindependent neural reorganization, (iii)The It = k relation held even when I layfar above N saturation. It is impossible thatincreases in I can produce proportional in-creases in N many times above the Nsaturation level. But It will be proportionalto the amount of photoproduct producedfar above this level. Thus we conclude thatthe early threshold rise is the result of thepresence of a transistoiy photoproduct, per-haps metarhodopsin II as suggested by theevidence of Donner and Reuter (1968).

According to this idea it is the produc-tion of metarhodopsin II which alwayscauses visual excitation. When this is pro-duced in such quantity that it remains un-decomposed for some time, a strong after-image persists and is the cause of the rise ofthreshold. Certainly the early threshold riseis associated with enhancement in thebrightness of the after image. We mayunify the concept of threshold rise due tometarhodopsin II with that due to "bleach-ing" (i.e., unregenerated rhodopsin) if wesuppose something like the followingkinetics

Meta II ?± opsin + a\\-trans retinal

This reaches equilibrium in about 5 min-utes and thereafter the amount of meta IIwill be proportional to the amount of freeopsin, and as that is removed from thearena by a stable combination with 11-cisretinal (regeneration) so will the meta IIbe removed in proportion, and hence therise in log threshold that is proportional tometa II will also be proportional to the un-regenerated rhodopsin (after 5 minutes),as observed.

The site of bleaching adaptation. An im-portant unification in the effects of bleach-

ings and backgrounds was put forward byStiles and Crawford (1932) and supportedby good evidence. It was that adaptationwas only of one kind in its operation de-spite the many ways in which it could beinduced. However brought about, a givenstate of adaptation would operate in thesame way for all input—output transforma-tions. "The equivalent background" ofbleaching is an example: if at some stageof bleaching the threshold is raised asmuch as it is by some particular back-ground, then those two states of adaptationwill be found equal as judged by everyother test. An example is seen in Fig. 5 forvarious N-values, where it is seen to breakdown near saturation.

The universal truth of the equivalentbackground concept would be very strongsupport for the view that bleaching andbackgrounds raised thresholds by the samemechanism. However, it now appears thatthe truth is not so universal, and thoughpart of the adaptation mechanism is prob-ably thrown into play by both backgroundsand bleachings, parts of the mechanismsare certainly distinct. Fig. 4 shows thatbackgrounds scale down the N signal with-out affecting a, the input sensitivity to light.Bleachings on the other hand may act byincreasing a (the light requirement) with-out affecting the size of saturated signal.

This difference in the mechanism ofadaptation by backgrounds and by bleach-ings was found also by electrophysiology.Naka and Rushton (1968) showed thatbackgrounds produced a steady hyper-polarization in the S-potential whereasbleachings raised by several log units thethreshold for S-potentials without produc-ing any steady change at all. This desensi-tization therefore appears to occur distal tothe horizontal cells, confirmed also byDowling and Ripps (1970, 1971).

One fact above all we need to know.How far is adaptation found in the rodsand cones themselves, intracellularly re-corded? Since the pioneer work of Tomita,Kaneko, Murakami, and Pautler (1967)many workers have suddenly developed theskill to do what no one could do before,



C.1 0'

0.8

0.6XOE

s.> 04

0.2

-BEFORE /

BLEACH r

o7 AFTER

1I/ •/ /

/ #'

° ft

BLEACH37 min.

^ •

22 min-

r ! J/ t.t ^—•

o/ If / 17 min.

/ / / /

9min-

- 6 -4 -2LOG I

Fig. 10. Potential from axolotl rods intracellu-larly recorded, in response to a 200 msec, flashof log I value shown. White circles before bleach-ing; results fit an H curve with log <r of -4.5.The retina was then 45% bleached and did notregenerate any rhodopsin subsequently. At firstthe response was abolished but it gradually re-covered, the increase in log sensitivity going handin hand with decay of a photo product absorbingat 400 nm. During this recovery the log a re-mains fixed at about -3.5 (where it shifted onbleaching). (From Grabowski, Pinto, and Pak.)

and we shall soon have clear answers toadaptation in the receptors. At present myopinion is that most of background adapta-tion is signal scaling organized proximal tothe receptors, probably by the horizontalcells and their connections. It does notseem as though 1 quantum caught by 1%of the rods will reduce the response of allthe other rods by 70% as it does the N-signals in man. On the other hand, muchof bleaching adaptation is a change ofsensitivity in the receptors themselves. ForFig. 10 I am indebted to Drs. Grabowski,Pinto, and Pak for allowing me to use someof their yet unpublished work. The pointsplot the height of intracellular potentialsfrom axolotl rods as a function of log in-tensity of a 200 msec, flash. Before bleach-ing, the expected H curve was obtained

(white circles). After bleaching the re-ceptors at first were unresponsive and theblack circles show the improvement withtime in the dark. The retina was excisedand densitometry measurements showedthat the exposure bleached 45% and thatno rhodopsin regenerated thereafter. Therecovery of threshold however was ratherwell correlated with the decay of metarho-dopsin II (or whatever was measured at400 nm.) over a range of 3 log units. Fig.10 shows that the bleaching of rhodopsinshifts a to the right by 1 log unit, and thatthe decay of meta II at 37 min. half re-stores the initial Vmax without change of a.Thus within the rod itself there seem to betwo mechanisms in action.

(a) Removal of half the rhodopsin fromthe rods will naturally require twice asmuch incident light to give the same quan-tum catch. This would displace log a 0.3to the right. In fact it is displaced 1.0, thusthe rods are not twice but 10 times reducedin sensitivity by removal of rhodopsin, andsignals in addition are attenuated to 60%(perhaps by residual meta II). (b) Themain adaptive effect is due not to the ab-sence of rhodopsin but to the presence ofmeta II. The log threshold falls over 3 logunits in proportion to meta II concentra-tion and this is entirely due to attenuationof signals like the "background adaptation"of Fig. 4.

Conclusion

The magnificent achievements of micro-electrophysiology and microspectrophotom-etry are rapidly and expertly taking overthe whole experimental field in conjunc-tion with the findings of microanatomy. Thepsycho-physicist must now gracefully re-tire from his attempts to explain how theretina works. But perhaps he may still men-tion what it is that has to be explained, forhe alone is professionally concerned withthe possibility that the nerves studied bythe others may have something to do withvisual perception! So let me propose foryour solution, O, colleagues of incompar-able expertise, the two key problems in


Volume 11Number 6


visual adaptation: (i) What organizationunderlies Fechner's Law (Fig. 1) and whatBouguer's? (ii) And when you can explainwhy a real image raises the threshold inproportion to the light that falls, tell mewhy an after-image raises it in proportionto the antilogarithm of the light that fell,for the explanation in my Ferrier Lecture(Rushton 1965b) is now rather outdated.

REFERENCESAguilar, M., and Stiles, W. S.: (1954) Saturation

of the rod mechanism of the retina at high levelsof stimulation, Optica Acta 1: 59-65.

Alpern, M., and Rushton, W. A. H.: (1965) Thespecificity of the cone interaction in the after-flash effect, J. Physiol. 176: 473-482.

Alpern, M., and Rushton, W. A. H.: (1967) Thenature of rise in threshold produced by con-trast flashes, J. Physiol. 189: 519-534.

Alpern M., Rushton, W. A. H., and Torii, S.:(1970a) The size of rod signals, J. Physiol.206: 193-208.

Alpern, M., Rushton, W. A. H., and Torii, S.:(1970b) The attenuation of rod signals bybackgrounds, J. Physiol. 206: 209-227.

Alpern, M., Rushton, W. A. H., and Torii, S.:(1970c) The attenuation of rod signals bybleachings, J. Physiol. 207: 449-461.

Alpern, M., Rushton, W. A. H., and Torii, S.:(1970d) Signals from cones, J. Physiol. 207:463-475.

Baylor, D. A., and Fuortes, M. G. F.: (1970)Electrical responses of single cones in theretina of the turtle, J. Physiol. 207: 77-92.

Campbell, F. W., and Rushton, W. A. H.: (1955)Measurement of the scotopic pigment in theliving human eye, J. Physiol. 130: 131-147.

Cornsweet, T. N., and Pinsker, H. M.: (1965)Luminance discrimination of brief flashes undervarious conditions of adaptation, J. Physiol.176: 294-310.

Dartnall, A. J. A., Goodeve, C. F., and LythgoeR. J.: (1938) The effect of temperature onthe photochemical bleaching of visual purplesolutions, Proc. R. Soc. A. 164: 216-230.

Ditchburn, R. W., and Ginsborg, B. L.: (1952)Vision with a stabilized retinal image, Nature170: 36-37.

Donner, K. O.: (1959) The effect of a colouredadapting field on the spectral sensitivity offrog retinal elements, J. Physiol. 149: 318-326.

Donner, K. O., and Reuter, T.: (1968) Visualadaptation of the rhodopsin rods in the frog'sretina, J. Physiol. (Lond.) 199: 59-87.

Dowling, J. E.: (1960) The chemistry of visualadaptation in the rat, Nature 188: 114-118.

Dowling, J. E., and Ripps, H.: (1970) Visualadaptation in the retina of the skate, J. Gen.Physiol. 56: 491-520.

Dowling, J. E., and Ripps, H.: (1971) S-poten-tials in the skate retina, J. Gen. Physiol. 58:163-189.

Hood, C, and Rushton W. A. H.: (1971) TheFlorida Retinal Densitometer, J. Physiol. 217:213-230.

Naka, K. I., and Rushton, W. A. H.: (1966a)S-potentials from colour units in the retina offish (Cyprinidae), J. Physiol. 185: 536-555.

Naka, K. I., and Rushton, W. A. H.: (1966b)An attempt to analyze colour reception by elec-trophysiology, J. Physiol. 185: 556-586.

Naka, K. I., and Rushton, W. A. H.: (1968)S-potential and dark adaptation in fish, J.Physiol. 194: 249-269.

Riggs, L. A., Ratliffe, F., Cornsweet, J. C , andCornsweet, T. N.: (1953) The disappearanceof steadily fixated visual test objects, J. Opt.Soc. Am. 43: 495-501.

Rushton, W. A. H.: (1957) Physical measure-ment of cone pigment in the living human eye,Nature 179: 571-573.

Rushton, W. A. H.: (1961b) Rhodopsin measure-ment and dark-adaptation in a subject deficientin cone vision, J. Physiol. 156: 193-205.

Rushton, W. A. H.: (1965a) The sensitivity ofrods under illumination, J. Physiol. 178: 141-160.

Rushton, W. A. H.: (1965b) The Ferrier Lectureto the Royal Society. Visual Adaptation, Proc.Roy. Soc. B 162: 20-46.

Rushton, W. A. H., and Spitzer-Powell, D.(1972a) The rhodopsin content and the visualthreshold of human rods, Vision Res. 12: 1073-1081, 1972.

Rushton, W. A. H., and Spitzer-Powell, D.:(1972b) The early phase of dark adaptation,Vision Res. 12: 1083-1093, 1972.Equivalent adaptation levels in localized retinalareas, Report of Discussion on Vision, pub. byPhysical Soc, London, pp. 194-211.

Tomita, T., Kaneko, A., Murakami, M., andPautler, E. L.: (1967) Spectral response curvesof single cones in the carp, Vision Res. 7: 519-531.


Documents

Light and Dark Adaptation - Semantic Scholar · Light and dark adaptation W. A. H. Rushton-he nature of visual adaptation is an extremely complex question and our un-derstanding of