JOURNAL OF THE OPTICAL SOCIETY OF AMERICA

Rapid Changes in Foveal Sensitivity Resulting from Direct and Indirect Adapting Stimuli*ROBERT M. BOYNTON, WILLIAM R. BUSH, AND JAY M. ENocut

Department of Psychology, University of Rochester, Rochester, New York(Received October 21, 1953)

It is well known that an intense stimulus imaged upon one part of the retina has the capacity to reducethe sensitivity of a spatially remote area. Controversy has long existed regarding the interpretation of thiseffect-whether it is the result of stray light in the eye or of some kind of indirect inhibition. A methodfirst used by Crawford makes it possible to measure sensitivity changes associated with the onset of anadapting stimulus by determining the threshold of a test flash at various times near the onset of the adaptingstimulus. Two adapting stimuli vere used alternately within each experimental session. The "direct" adapt-ing stimulus was concentric with and larger than the foveal test flash; the "indirect" or glare-adaptingstimulus was presented at a glare angle of 18°. When the luminances of these radically different adaptingstimuli were suitably adjusted, the resulting foveal sensitivity changes, as a function of time after onset ofadapting stimulus, were found to be identical. The equivalence held when the luminance of each adaptingstimulus was reduced 1 log unit. The results are interpreted as favoring the stray-light hypothesis.

WHEN a glare stimulus is imaged on the peripheral~v retina and luminance thresholds are determinedfor a test stimulus viewed foveally, a linear relationis found between log intensity of the glare stimulus andlog foveal threshold luminance over a fairly wide range.This same relation holds when an extended field(veiling stimulus) is substituted for the glare stimulus,and the foveal test field is superimposed upon it.'These results have most often been interpreted by thehypothesis that the glare stimulus casts a veil of straylight across the retina which has the same effect as theveiling stimulus. Therefore, whether the adaptingstimulus be direct (veiling) or indirect (glare), thefoveal judgment is considered to be one of intensitydiscrimination.

An alternative viewpoint holds that all, or part, ofthe glare effect results from some kind of neural orelectrical inhibition. By this hypothesis, the peripheralreceptors receiving the glare image instigate activitywhich somehow spreads across the retina to reducefoveal sensitivity or perhaps exerts an inhibitory effectin higher visual centers. The now classic work ofSchouten and Ornstein2 is most often cited as "proving"that the indirect adaptive effects of a glare stimulusare of an entirely different nature from those resultingfrom direct adaptation.

The present experiment is designed to test thesehypotheses by measuring the rapid changes in fovealthreshold which occur just before, during, and im-mediately after the onset of direct and indirect adaptingstimuli. The method used is similar to that devised byCrawford3 and employed more recently by Boynton

* This work was supported by a contract between the U. S.Office of Naval Research and the University of Rochester.

t Jay M. Enoch is now at the School of Optometry, Ohio StateUniversity, Columbus, Ohio.

'W. S. Stiles and B. H. Crawford, Proc. Roy. Soc. (London)B122, 255 (1937).

2 J. F. Schouten and L. S. Ornstein, J. Opt. Soc. Am. 29, 168(1939).

3 B. H. Crawford, Proc. Roy. Soc. (London) B134, 283 (1947).

and Triedman.4 It is hypothesized that if the timecourse of threshold changes differs for the two condi-tions, something other then veiling glare is operatingunder the indirect condition. On the other hand, if thethreshold changes are the same, the stray-light hy-pothesis would seem the most tenable one.

APPARATUS

The apparatus, similar to that previously describedby Boynton and Triedman,4 is schematically shownin Fig. 1. A single light source (50-cp automobilespotlight bulb, operated at 6 v dc) supplied two highlysimilar optical systems by way of M, and M2. Theupper system in Fig. 1 transmitted the test flash;the lower, the adapting stimulus. The shutters (Sh)which controlled the time interval between onset ofadapting stimulus and onset of test flash (adaptinginterval) were located as shown at focal points. Lightwas again collimated, and neutral filters and a neutral

Ml : L eS t1i 4 F | F E Sp 6 \tt 0-- -0; 4 Q-g-E -1 Lid F,;--e~~~~~~~~~~~~---- ..b

DIRECT

18 -)__2-

INERECT55-

------ X ...... S.

__ It. _ ___ -L|'.IX;KWl

... ___ i ------- I ' 1;- * Jr.. / . _---_ ___--*- *---------- /

FIG. 1. Diagram of apparatus. The upper part of the figureshows the optical system used for the "direct" condition. Changesmade for the "indirect" condition are shown in the lowerright-hand corner. The appearance of the stimuli to the subjectfor the two conditions is shown in the lower left-hand corner.

4 R. M. Boynton and M. H. Triedman, J. Exptl. Psychol. 46,125 (1953).

56

VOLUME 44, NUMBER JANUARY, 1954

CHANGES IN FOVEAL SENSITIVITY

wedge were located as shown. For the direct condition,the two beams were superimposed with a prism andbeam-splitter and then brought into the observer's eyeby L7. The subject saw the stimuli as shown in theupper figure in the lower left-hand corner of Fig. 1(labeled "DIRECT"). In the lower right-hand cornerof Fig. 1 are shown the apparatus changes which wererequired to present the indirect adapting stimulus. Theprism and M3 were mounted on a revolving turn table.With M3 in position, the collimated rays from theadapting system were stopped down and reflected off M4through the last lens, L7. Under this condition, the stim-uli appeared as shown in the extreme lower left-handcorner of Fig. 1 (labeled "INDIRECT"). A bitingboard and fixation light (not shown) were arranged sothat the test stimulus, under both conditions, waspresented foveally. Note that a Maxwellian view wasemployed, thus eliminating the effects of pupillarychanges. All optical elements were rigidly clamped toa 6 in. X6 in. X5 ft steel I-beam. A wall separatedsubject and experimenter, and stray light in the systemwas further eliminated with covers, baffles, etc..

CALIBRATIONS

1. The Shutter System

Two Kodak photographic shutters were used. Shi,in the test-flash system, was set at 1/25 sec. S 2, inthe adapting system, was set on either time or bulb,depending on the length of the adapting interval. Theshutters were mechanically operated by means of adouble-cam system, driven by a 1725 rpm motorgeared so that the cams rotated approximately onceper sec. The adjustable relative position of the twocams was indicated on a dial calibrated directly in0.005-sec intervals. The calibration was carried outby determining the exact time of one revolution fromthe time elapsed when the cams were allowed to rotate50 times. The shorter intervals and the test-flashcharacteristics were checked by observing the responsesof a photocell on a CRO. Prior to each session, the"zero" adapting interval (when test and adaptingstimuli had simultaneous onset) was checked andadjusted, if necessary, by turning the cams slowly byhand until both shutters were simultaneously activated.Throughout the experiment, data at any given adaptinginterval were found to be highly reproducible, and theshutter system gave little trouble.

2. Reference Luminance

A commonly used binocular procedure was employed,with the right eye observing the test stimulus and theleft eye viewing, through a 2-mm diameter artificialpupil, an independently illuminated screen which wassubsequently measured for luminance with a Macbethilluminometer. The adapting stimuli were then com-pared monocularly with the test stimulus, after tem-porarily equating the two for area. The neutral wedge

-. 10

0

.1 0

4.50

J l

I ---+.2 +.4 -. 6 4.8 +1.0

SECONDS

FIG. 2. Stimulus sequence within experimental "trials" forvarious adapting intervals. Adapting interval is indicated on theleft. Small rise is the test flash, larger rise is the adapting stimulus.Note that test flash sometimes is presented before the adaptingstimulus but is more often superimposed upon it. The adaptingstimulus remains on for a full sec for the +.50 condition to avoidthe effects of adapting stimulus extinction.

was calibrated so that identical readings for the twosystems indicated identical luminance values. Withno filters in the system, the luminance of both testand adapting stimuli was 26 000 mL.

EXPERIMENTAL PROCEDURE

Each experimental session began with at least 20-min dark adaptation. The procedure within eachsession was then as follows: (a) After dark adaptation,a measure of absolute threshold was secured. This wasobtained with a descending method of limits, withindividual flashes 15 sec apart in each series. Lightfrom the adapting system was occluded. (b) For agiven interval from onset of adapting stimulus toonset of test flash (adapting interval), a test-flashthreshold was determined. This was accomplished byrepeated presentations of the adapting stimulus-test-flash sequence. Such sequences for four adaptingintervals are illustrated in Fig. 2. For convenience, eachof these sequences will hereafter be called a "trial."At a given adapting interval, the luminance of the testflash was successively reduced, from trial to trial, untiltwo successive negative judgments were recorded; andthe test-flash luminance for the last positive judgmentwas taken as threshold. Depending on where thedescending series happened to start, each thresholddetermination required from two to six minutes (fourto twelve trials at 30-sec intervals). (c) This procedurewas repeated for the other adapting intervals. In halfthe sessions, intervals were sampled from "short tolong" (left to right in Fig. 3); the reverse order wasused in the other sessions. (d) At the end of each session,the absolute threshold was redetermined. A 30-secinterval was scrupulously maintained between trials,and a typical session lasted about two hours.

Both conditions were tested within each session.Direct and indirect adapting luminances were selectedwhich yielded the same test-flash threshold at "zero"adapting interval (simultaneous onset of adapting

January 1954 57

BOYNTON, BUSH, AND ENOCH

+0.5

I

-J

00U

4z=1 -0.5I

4

-I 0I-

, -1. 5

RMB

h A A.A "

* D- 2.6 mLo 0- 24,000 mL

A D .26mLa ' 2,400 mL

JME

-.1 0 +.I +.2 +3 +.4 +.5 -.1 0 +.i t2 t3 +*4 +5

ADAPTING INTERVAL- SEC. ADAPTING INTERVAL-SEC

FIG. 3. Results of main experiment. Log luminance of threshold test flash is shown as a function of adapting interval for the twomain experimental conditions and at two adapting levels for two subjects. The horizontal lines show the luminances of the directadapting stimuli. "D" and "I" in the legend indicate direct and indirect conditions, which are represented in the plot by closed andand open symbols, respectively.

and test stimuli).t An IDDI order was used as thevarious adapting intervals were sampled. Each criticalcomparison between conditions is, therefore, based upondeterminations made within a few minutes of each other.

RESULTS

The results are given in Fig. 3. Each plotted pointfor RMB is the average of three determinations madein separate sessions. Each point for JME is based on asimilar average of two determinations. The followingpoints should be noted in Fig. 3: (1) The direct andindirect adapting stimuli, when their luminances aresuitably adjusted, produce test-flash threshold changesdescribable by a single function. (2) For both adaptingconditions, threshold starts to rise when the test flashis presented before the adapting stimulus onset.§ (3)The curves reach a definite peak and then decline.This means that, following the initial transition fromabsolute to intensity discrimination threshold, Al/Ideclines.

Implicit in the experimental procedure is the assump-tion that the initial dark-adapted state is reestablishedbetween successive trials. In order to determine the

$ The values finally selected resulted from a number of explora-tory sessions. A given "direct" adapting luminance (2.6 mL) wasselected, and the glare luminance was varied. Changes in adaptingluminance of only 0.1-log unit resulted in marked changes in thefunctions. For JME, the desired value fell between the discrete0.1-log unit steps by which we were able to vary adapting lumi-nance. Consequently, determinations were made "on each side,"first with no filters in the adapting system and then with a 0.1-logunit filter. The data reported are the means of those obtained inthe two sessions. For RMB, this difficulty was not encountered.

§ This interesting phenomenon will not be further discussedin this paper. See Crawford (reference 3) and Boynton andTriedman (reference 4) for a discussion of this effect.

minimum "safe" time that should elapse betweentrials, thresholds were determined, for a given adaptinginterval and adapting luminance, with various intervalsbetween trials. These intervals ranged from 2 sec to2 min. Altogether, 6 experimental sessions were devotedto this purpose.

Sample results are given in Fig. 4. Data for the 2-min spacing are not presented, since they differed inno case from the 1-min data. The results show that:(1) 30 sec appears to be the minimum "safe" spacingof trials. (2) For negative adapting intervals, the effectof "massing" trials is, in general, slightly to raise thethreshold from its "true" (30 sec) value. (3) For positiveadapting intervals, the intensity-discrimination thres-hold is spuriously lowered by massing trials. By repeti-tion of the stimulus sequence every 2 sec, it is possibleto drive the threshold down about a half log unit.

DISCUSSION

1. The Stray Light Interpretation

We have shown that the adaptive effects associatedwith the onset of a 7 direct adapting stimulus at 2.6mL can be duplicated by a 5.50 glare stimulus at about25 000 mL, at a glare angle of 180. The resulting func-tions are complex, showing a rise at short negativeadapting intervals, reaching a peak, and declining-inexactly the same way for each condition. Further, thesimilarity of effect holds when the luminance underboth conditions is reduced 1-log unit.

Were the effects of our glare stimulus the result ofneural inhibition, the direct and indirect conditionscould produce the same effects, as a function of time,only by the most unlikely coincidence. If the glare

58 Vol. 44

CHANGES IN FOVEAL SENSITIVITY

stimulus raises the foveal threshold as a result of aprocess instigated in the retinal area receiving the glareimage, which in turn must be transmitted to the fovea,or a projection of the fovea in a higher visual center,such a process would be expected to introduce sometime delay. (On the basis of present physiologicalknowledge, and the speculative nature of the locus ofinteraction, the extent of this delay would be hard tocalculate.) Our data do not indicate such a temporaldifference between conditions. Also, if some sort ofneural interaction were involved, there would be noreason to suppose that the direct and indirect deter-minations would vary in the same way as a functionof adapting luminance-but they do, at least over a1-log unit range. On the other hand, this result isexactly what the stray-light hypothesis would predict.

These same objections hold for the Schouten andOrnstein electrical inhibition hypothesis. In addition,the notion of electrically mediated inhibition is toospeculative and untestable to be very useful, and thedata which led to the hypothesis are now suspect(see following).

We, therefore, conclude that the adaptive effect ofour indirect stimulus is the result of stray light. Thisconclusion is supported by a growing body of outsideevidence. First, a number of investigators'2 5-7 haveimaged a glare stimulus on the optic disk, and nodiminution of the various glare effects under study hasever been noted. Second, there appears to be enoughstray light in the eye to account for the b-wave com-ponent of the electroretinogram to small area stimuli.8 -'0

6

2

I

I

u0w

0.0 -.03

atI3

1.6 _

1.4

1.2 _

1.0 _

0.8 _

0.6 _

0.4 -

0.2 _

0.0

II , I r . I

-. 1 0 +.I 4.2 +.3ADAPTING INTERVAL-SEC

+.4 +.5

FIG. 5. Intensity discrimination as a function of adaptinginterval. Data are for two subjects (denoted by open and closedcircles), under the direct condition, with an adapting luminanceof 2.6 mL.

Third, the present authors" have obtained directmeasures of stray light in excised eyes which show morethan enough stray light to account for the glare effectsobserved by us and others.

On the 'other hand, very few studies have given anyevidence for the electrical or neural inhibition hy-potheses. The most widely quoted of these, that ofSchouten and Ornstein,2 has recently been re-examinedby Fry and Alpern,12 who have failed to reproduce theresults of one of Schouten and Ornstein's most criticalcontrol experiments and who conclude that theSchouten-Ornstein effects can be explained in terms ofstray light.

-1.0 -

-1. I .08

I ,1 I , Ii 1 I , 1 I10 20 30 40 50 60

TIME BETWEEN AAPT.-TF SEQUENCES SECS.

FIG. 4. Sample results of time-control experiments showingtest-flash thresholds as a function of time between trials. Logluminance of adapting stimulus is 2.6 mL (direct). Parameter isadapting interval in sec.

6 P. W. Cobb, Am. J. Physiol. 29, 76 (1911).6 S. H. Bartley and G. A. Fry, J. Opt. Soc. Am. 24, 342 (1934).7 Y. LeGrand, Rev. d'Optique 16, 201, 241 (1937).8 G. A. Fry and S. H. Bartley, Am. J. Physiol. 111, 335 (1935).9 R. M. Boynton and L. A. Riggs, J. Exptl. Psychol. 42, 217

(1951).10 R. M. Boynton, J. Opt. Soc. Am. 43, 442 (1953).

2. Relation to Other Studies

The data just presented are in accord with those ofCrawford,3 Baker,3 and Boynton and Triedman,4 inshowing that after an initial threshold rise, the intensity-discrimination threshold starts to fall. Our data for thedirect 2.7 mL condition are replotted in Fig. 5, withAl/I on the, ordinate instead of log test-flash luminance.At the peak of the curve, which occurs about 0.04 secafter the onset of the adapting stimulus, intensitydiscrimination is extraordinarily poor, as the test flashmust cause a momentary luminance increase of theadapting field of more than 100 percent in order to beperceived. Within a half sec, the threshold dropsgreatly. Apparently, this same kind of threshold dropmay occur when the adapting stimulus is intermittent,

11 Boynton, Enoch, and Bush (to be published).12 G. A. Fry and M. Alpern, J. Opt. Soc. Am. 43, 187, 189

(1953).13 H.~ D. Baker, J. Opt. Soc. Am. 39, 172 (1949).

- . . I

January 1954 59

. ____ 0)

I I II

I

BOYNTON, BUSH, AND ENOCH

since this is one way to consider our time controlexperiment. For example, a 1-sec exposure of theadapting stimulus every 2 sec lowered the test-flashthreshold (in the presence of and 0.1 sec after the onsetof each adapting stimulus) about a half-log unit belowthat obtained with the "safe" 30-sec spacing betweentrials. The data of Crawford3 need to be considered inthe light of this new evidence, since he repeated hisstimulus sequence every 7 sec. His values for positiveadapting intervals are, therefore, in all probability lower(and some of his other values higher) than had he notmassed trials. At any rate, his threshold values reflecta very complex state of the visual system, where notonly the effect of a given adapting stimulus is measured,but also the cumulative effect of those preceding it.

The striking similarity of our data to those of Brocaand Sulzer4 (on the subjective growth of the visualresponse) and to those of Adrian and Matthews"(on the impulse frequency in the optic nerve resultingfrom stimulus onset) should be noted. It is apparent

14 A. Broca and D. Sulzer, J. de Physiol. et Path. Gen. 4, 632(1902).

16 E. D. Adrian and R. Matthews, J. Physiol. 63, 378 (1927).

that the onset of a visual stimulus has associated withit a rapidly increasing and then subsiding activity ofthe visual system which in turn limits the effectivenessof an added increment.

3. Light Adaptation

Baker 3 took as a criterion of light adaptation theintensity-discrimination threshold and was surprised tonote that this threshold dropped during the firstminute or so following the onset of his adapting stimulusbefore starting to rise toward a steady state value.Baker, however, made no measurements during thefirst few seconds following the onset of the adaptinglight. The present data help bridge the gap between thedark-adapted threshold, on the one hand, and theintensity-discrimination data of Baker. To the extentthat one wishes to accept the intensity-discriminationthreshold as a valid index of the state of light adapta-tion (see Baker'3 and Boynton and Triedman4 fordiscussions of this point), the present data describe,for both experimental conditions, the initial timecourse of light adaptation.

JOURNAL OF THE OPTICAL SOCIETY OF AMERICA VOLUME 44, NUMBER JANUARY, 1954

A Study in Binocular Flicker

FRED H. PERRIN*Institute of Applied Optics, University of Rochester, Rochester, New York

(Received June 22, 1953)

If both eyes are subjected to flickering stimuli of equal luminance and frequency, the critical fusionfrequency is usually somewhat higher when the stimuli are in the same phase than when they are in theopposite phase. The difference in fusion frequency for the two cases is herein termed the Sherrington effectafter its discoverer. Apparatus for measuring this effect is described in the present paper. It was found that,for a 2° visual field and dark surround, the Sherrington effect was proportional to the mean critical frequencyfor the two eyes as the field luminance was varied, being about 8 percent of it for the present observer. Atconstant field luminance, the effect approached zero as the field angle approached zero but was approxi-mately constant for angles greater than 20. The mean monocular fusion frequency for seven observerswas below the simultaneous (in-phase) binocular frequency, and for a 0.250 field it was even below thealternate (out-of-phase) frequency; for three observers, it was below the alternate frequency for a 20 field.The present data appear to be inadequate to serve as a basis for deciding between alternative explanationsof the phenomena.

I. INTRODUCTION

T HE modern study of flicker seems to have startedT in 1740 with Segner's' experimental determinationthat a visual sensation persists about one-tenth of asecond. In the two centuries that have elapsed sincethen, countless experiments have been made in monocu-lar flicker, but very little has been done on the inter-action between the two eyes when both are subjected

' Now at Research Laboratories, Eastman Kodak Company,Rochester, New York.

I Segner, De Raritate Luininis (Gottingen, 1740), quoted withlater workers by Ferry, reference 15.

to flickering stimuli. To be sure, Allen2 has studiedextensively the effect of subjecting the unused eye toa steady stimulus, and Schaternikoff' compared thecritical frequencies for the two eyes of a single observerwhen one was photopic and the other was scotopic;but only two studies of true binocular flicker had beendiscovered by the present author in a more than

'F. Allen, J. Opt. Soc. Am. and Rev. Sci. Instr. 7, 583 (1923).This is but one of a series in this journal and in Phil. Mag. from1911 to 1926.

3 H. von Helmholtz, Physiological Optics (Optical Society ofAmerica, New York, 1924), Vol. II, pp. 373-374. An extensiveearly bibliography of flicker sensitivity is on pp. 224-226.

60 Vol. 44