Brit. J. Ophthal. (I975) 59, 545

Use of random-dot stereograms in the clinicalassessment of strabismic patients

J. P. FRISBY, J. MEIN*, A. SAYE, AND A. STANWORTH*From the Department of Psychology, University of Sheffield, and the *Hallamshire Hospital, Sheffield

Random-dot stereograms (Julesz, I97I) have beenused in recent years to study anomalies in stereopsis(Benton and Hecaen, I970; Carmon and Bechtoldt,1969; Richards, 1973). Their special advantage forthis purpose is that they provide a very pure test ofstereopsis, because they present disparity informa-tion in the absence of all other clues to depth andlend themselves to unfakeable test procedures, sofacilitating the collection of reliable measures.Despite this advantage their use in orthoptic andophthalmic clinics has not been widespread,probably because there has been no systematicclinical trial of their usefulness before that ofReinecke and Simons (I 974). They used random-dotstereograms to screen for amblyopia and amblyopia-related visual dysfunctions in young children. Theyfound that, with the exception of some microtropicpatients and one esophoric/esotropic patient, nopatient with a constant horizontal or vertical tropiacould obtain a random-dot stereoscopic perceptwith their stimuli; moreover, those microtropicpatients who passed the test did so only when shownstimuli with a disparity of at least 8', all the patientsappreciating a disparity of less than 4' beingnormal or near normal (most had an intermittentexotropia).We too have examined the abilities of various

categories of strabismic patients to fuse random-dotstereograms. Our approach was similar to that ofReinecke and Simons in that we have comparedresults from random-dot stereograms with standardclinical orthoptic measures. The study we reporthere, however, differs from that of Reinecke andSimons in three major respects: two different typesof random-dot stereograms were used (Figs i and2), the disparity except in pilot studies was always12' (which is greater than either disparity used intheir study), and strabismic patients were selectedfor examination only if their clinical recordsmentioned that some degree of stereopsis waspresent.An initial pilot study revealed that there was no

simple correspondence between the performance ofAddress for reprints: Dr J. P. Frisby, Department of Psychology,University of Sheffield, Sheffield SIO 2TN

strabismic patients on the widely-used Wirt (orTitmus) stereotest and on our random-dot stereo-grams. Some patients could demonstrate a stereo-acuity of better than I3' on the Wirt test and yetprove incapable of fusing random-dot stereogramswith disparities in the range 7' tO 1°24'.We followed-up this pilot finding with experi-

ments which presented two different kinds ofrandom-dot stereogram to patients attending anorthoptic clinic. One kind of random-dot stereo-gram was of the usual type (Fig. I). The secondkind, however, was novel in that it containedprominent uniocularly-identifiable features. Itscentral square-shaped area of disparate elementswas thus enclosed in each field of view by an outlinesquare which possessed the same disparity as theelements it contained (Fig. 2). The rationale under-lying the use of this latter 'contoured' kind ofstereogram was as follows.Random-dot stereograms are complex stimuli

which contain many ambiguities concerning whichelement in the left field is to be fused with whichelement in the right. An enormous number ofpossibilities exists in principle, but in practice thebinocular combination process produces a fusionof the two fields in which relatively dense surfacesare preferred to 'lace-like' patterns in whichindividual elements are scattered in a multitude ofdifferent depth planes (Julesz, 197I). The Wirttest figures, on the other hand, are relatively simplestereograms which incorporate little ambiguity.Each Wirt figure is made up of only four circularelements enclosed in a frame, the observer's taskbeing to identify which of the four is standing outin depth. This distinction between random-dotstereograms and Wirt test figures in terms of theircomplexity and ambiguity might have been thecause of the asymmetry in stereo-ability observedwith them in pilot work. We attempted to investi-gate this possibility by seeing if demarcation of thedisparate area in a random-dot stereogram wouldfacilitate stereopsis for patients who would other-wise fail to obtain it with these stimuli. Demarcationof the disparate area, it can be argued, makes arandom-dot stereogram more like a Wirt test figure

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from

546 Britishsournal of Ophthalmology

0 * -*e* * 0*-*.0. . .0 0*.0*- *-*0 ** 0 . . 0 000- * . .0 0 0

0.00-0. .0 .00 000 000. 0 . 00*

0. 0000 60 o. 0 *.oo.o 00O @0 0 .000 0 .@06 .0.0 0

0 6 ..00 0 0 ** 0 0 0* *

@* 0.S* 0..6 0 00 0000,* *o . 0

00 0 0 0 0 00 s0 0 o0 0

*0 -*00 .0 00.0 0. 0. 000..

000 .0060 . 0 o 00000.00 0

0*- ** . * 0- * - .O . * - 0.

0 0.00 .a . .0 0 .0 0 0 0

* .00 0- 0 0. * @ 0. * 0 0 . . - . g

o 000 ** 0 0. o0 0 0... * 0 .

* 0.@-0 @. 00 00. a 0. @ 0 . 0

*0. . 0 0 6 0*0 0 0 05* 6 0 0 0.*O 00 0. 06eee .0 .00 0660@@. -0.O. 0. 0.. o0 .0 .0 00 *'- -*-0 -

00 -0.@ 60s@-.0 000- 00 0.geOS*-*

0 .0 0 * * .0 .0 0. @ . .

* 0 0 .- *.00 . -0 * *0 * 0- 0

* ** 0*00 . 0..0 0 0 0

.000..0.0.0.60 0000 .0 S

0*. .- 0 06 og 0 0e 0

00 *660 00 60 0 0 000 00.0.0 0 0.0* 0. *- . .0 ee0. 0. @6 0* 0.0 0

00* *4000.00-oe e - * * 0

00 0* O 0 * 0 0 00 0 .. . 0* 0 0 *a 0

00 O 0. 0 *-e . 00. e0*0e e- *

e000.0 *0.000 000 0 00060 g000. oOSooeo 0es0 000. 06 00 00O @0 0060 00

*0 Og0 0 0 00 0. 0.0

00 - * . *e 0 .e * .0 o * 0

6 .S 0 0 0. 00 0 0. 0- * @ 0 00 0 -.-..0. * -e0 000.0 000

o *gooo *0

,0 000 0.0000 Os. 00 . . -. 00 0 *. .0 00.

0 . ** 0 0 0 **0 0..0.0 0*0 .0

0 .0000. .*.0 . 0 00 0a S 0 00.

** 0. *00---.0.e.000 0 0 . S. .* * *. . *--* * @@ *g 060@@

0 .0 00000 0 0.0 0 0 000 00

.00 . 6- 0. .0 0. . 0. 0 06.

* *e 0 .0 * 0 .* 0 . *0 0 0 0. *-S Oe * * 0 * . . @ 0 0 0 0. * . * *

FIG. I Non-contoured stereogram used in experiment. For the experiments, stereograms subtended a visual angle of1O0 30'. Brightness of the dots (as measured with an SEI spot photometer) was < about 3.4 cd mr2 while that of theground was about 34 3 cd m 2

* 0* 5 006-o6 6 0. 0.** - . 6* 0 .0 00

.0 60.* *@ *.0.0.00* * - 000 . .0* O**- 0 -. .0 *S0 000 00 0 .@@ 0@

0 * *. .*6 .- 6.0*- 06** 0 0 *

0. 0.0 6.6 .0.6. 00 00 .00.0.@-* @-

6 .- 6 0000 * 0 6 @0 00 *o 0. 0

*.@ 00 @ 0 0 0-.-- * @ . 0. * -

0*06 0 0 6 0666 .- . .00

*0 0 0. . 0. 0 S 0 @* 6 6- 6

*. -0g0S@ *a@000 0.60..0 . -* . **0 *0 * *00 * * 0* 0 6

O 0 o 00 0 a * .0 @0 0.0 0

Og 6.00. 0@0 0000 006 00

000 0. 00 * 0e * * * S

O 060 00. 0. 060 000 0 5

e o 0 * . 0 @ 0e s 0- O 00 0 0

O 6@60 006 O 00 0 00 Oo 6 6

0 0 00o 60660@. 60 *** @-00 '0aO00 0

.06 0 6. - 0@0* .0 0

0*@ 0 . * * 0 .00 -0 0 o60 ..

0 0 0 0a- 0. 00* - - *0 e - - -

,* * - - * *- * . * * .0 la . . .0 . o00. 0 -00- 0 0 0 0s. .*0 00 0.0- - 00. . .o0 0 00* o00a*S* .000 a0

0.0 o 0 00. 0 *..0o 0.00.o0 0.o0** 0-.-0 0.0.0- 0 .0 @0 - 00.0 0-0 0

* 0 *0 0 0 0* 00. @ 0 .0 0

.06g. - .0 0 0. .0 0. 00- 0

*o*O0.*o0 0.000~~~o0. 0. 0 .OD 0 o0 . 0 . 0.00o .0 00. 0. 0.0 0* - 0 0 *. 0000e . . *00 0 S 0 0 0

* 6 * O. - - - - . e e00- - -e . .0 0

00.060000 6 0*-.. 0.000 0 .** 0. 0 0@ *ooo 0 006 00.0.0.0 *- -00 * --** . . @ *o 6 g .-00*...0 .0 0. * 00 00000 0 0 0.

.0 -o 0*- * 00.0 .00 0 06 . 00

6 .6. .0.60 0.00.0. 000 .o 0..

0 00** 0600 0 0 o Oo* 0 0.0 0.06 50.000* 00 .0- . .0o .00 0 e .0 0

0 .0 0* * 0 0 .6 o .0. 0 6 0 .* *.

6* S- 6 0 .0 0 0. 00.0 0 0*

FIG. 2 Contoured stereogram used in the experiment

in that the stereogram is greatly simplified and at least the 'fly' stimulus of the Wirt test), and they were

resolution of the inherent ambiguity is likely, there- expected to be co-operative in a testing situation.fore, to be considerably aided. The stereograms, which always contained a disparity

of I2', were projected through polarizing filters on to avertical aluminized screen situated I67 cm in front of the

Material and methods subject's rigidly-fixed headrest. Polarized filters wereA total of 27 strabismic patients aged between 8 and I 8 mounted on the headrest in such a way that the disparityyears were shown the contoured and non-contoured incorporated in the stereograms was crossed-that is,stereograms described above. All were outpatients the disparate square, if fused correctly-would appear toattending an orthoptic clinic who had been selected lie in front of the screen. The non-contoured and con-for the experiment because they were known to have toured stereograms were shown successively to eachsome clinical record of stereopsis (that is, they could see subject, the non-contoured kind always first. On each

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from

Use of random-dot stereograms 547

presentation, subjects were asked to look carefullyat the centre of the stereogram and report if any part ofit appeared to 'stick out' in front of the rest. An attemptwas made in making this inquiry, and indeed throughoutthe experiment, to put the subject at his ease and toencourage him to describe fully his percepts at anygiven moment. Moreover, any reports of depth ex-periences were followed by further requests for clarifica-tion if these initial reports were either not clear or notfull. For instance, in the case of the contoured stereo-gram, if the subject described a 'square shape stickingout' but did not also state that the elements within thissquare were also 'sticking out', he was asked whetherthis was in fact the case. These full verbal reports werenot taken simply at their face value, but were checked byasking the subject to set a pointer to lie in the apparentdepth plane of the disparate square, if this had beencorrectly reported. This pointer was mounted in a toyrailway truck which could be moved by the subject viaa pulley system along a track lying on a table top betweenthe base of the headrest and the screen. The length ofthe pointer was such that its tip could be positionedimmediately underneath the lower edge of the square-indepth. This toy train arrangement was used to make thetask more interesting for the younger subjects. Averagedepth settings for successfully fused stereograms wereabout 20 cm. The experimental session lasted about 15minutes.

Results

The subjects differed markedly in their ability tofuse the random-dot stereograms and it provedpossible to classify them into the following threegroups:Group i: Normal stereopsis obtained with both

non-contoured and contoured stereo-grams (N= ii).

Group 2: Normal stereopsis obtained only with thecontoured stereogram (N= io).

Group 3: Normnal stereopsis obtained with neitherkind of stereogram (N=6).

Three subjects who failed to fuse the non-contoured stereogram satisfactorily and who in thecontoured stereogram could see only the outlinesquare in depth, and not its enclosed elements, wereassigned to Group 3.The clinical records of the subjects were examined

by an experienced orthoptist (JM) who had no priorknowledge of the subjects' performance on therandom-dot stereograms. The Table presents asummary of items drawn from her assessment whichseemed to discriminate between the subjectgroupings found with the random-dot stereograms.These items were:I. Original and present diagnoses2. Wirt stereoacuity3. Anomalies in ocular movement (as revealed when

tested in the nine cardinal directions of gaze)4. Prism fusion ranges.No clear differences between the groups were found

when size of deviation, visual acuity, or degree ofanisometropia were examined. These details are,therefore, excluded from the Table.

Further details about the four discriminatoryitems are given below.

I. ORIGINAL AND PRESENT DIAGNOSES

Of the i i patients in Group I, seven (cases I, 2, 3, 6,7, io, and i i in the Table) presented with inter-mittent squints. Of the remaining four only one(no. 4) had a constant manifest squint requiringsurgery, two had microtropias (nos 8 and 9), andthe other one (no. 5) presented with aniso-amblyo-pia. Eight out of I I patients in this group thereforeapparently retained bifoveal binocular single visionfor at least part of the time on initial presentation.Five out of ii patients (nos I, 2, 3, 4, and 5) nowmaintain constant bifoveal binocular single vision,two (nos 6 and 7) show intermittent squints butexercise normal binocular vision alnost constantlyand four (nos 8, 9, IO, and I) show microtropiasbut anomalous binocular function.

In Group 2 there was a much higher proportionof constant manifest squints on presentation (fivepatients out of io: nos I2, I4, i8, I9, and 20).Only one patient (no. 13) presented with an inter-mittent squint and therefore maintained bifovealbinocular single vision for at least some of the time.The remaining four patients (nos 15, i6, 17, and2I) all presented with microtropias with consequentanomalous binocular vision. The present diagnosesare also less good in Group 2. Thus constant bifovealbinocular single vision is present in only one patient(no. 12), another (no. 13) has an intermittent squintbut maintains normal binocular vision nearly all thetime, one patient (no. 2I) has a manifest eso-/exo-deviation and all the remaining six have microtropias(nos 15, i6, 17, I8, I9, and 20).

Summarizing these clinical assessments ofGroups i, and 2, it is evident that the majordifference between them is that there is a muchhigher proportion of subjects in Group i than inGroup 2 who experience bifoveal binocular singlevision for at least part of the time. This difference ishighly significant when original diagnoses areconsidered (Group i-eight subjects out of I I;Group 2-one subject out of IO; Fisher's exact test,P=o-oo6) and it almost achieves significance on thepresent diagnoses (Group I-seven subjects out ofiII; Group 2-two subjects out of Io; Fisher'sexact test, P=o0o56).

In Group 3, five out of the six patients presentedwith constant manifest squints (nos 23, 24, 25, 26,and 27), with the remaining patient (no 22) showingmicrotropia. The results of treatment in this groupare that five patients have small-angle residualdeviations with some degree of anomalous binocular

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from

548 British Journal of Ophthalmology

Table Summary of data taken from clinical recordsOclrmvmn

Originaldiagnosis

Intermittent exotropiaIntermittent exotropiaFully accommodativeManifest esotropiaAniso-amblyopiaFully accommodativeIntermittent exotropiaMicrotropiaMicrotropiaConvergence excess

Fully accommodative

Manifest esotropiaConvergence excess

Manifest esotropiaMicrotropiaMicrotropiaMicrotropiaManifest esotropiaManifest exotropiaManifest esotropiaMicrotropia

MicrotropiaManifest esotropiaManifest esotropiaManifest esotropiaManifest esotropiaManifest esotropia

Presentdiagnosis

ExophoriaEsophoriaEsophoriaAniso-amblyopialFully accommodativeIntermittent exotropiaMicrotropiaMicrotropiaMicrotropiaMicrotropia

Esophoria*Fully accommodativeMicrotropia*MicrotropiaMicrotropiaMicrotropiaMicrotropiaMicrotropia*Microtropia*Manifest eso./exo.

MicrotropiaManifest esotropiaMicrotropia*Microtropia*Microtropia*Microtropia*

Wirt

402 20

40"f2 20

5o050A

40N6'40'2 20'

I ' 40'

40

2' 20'

6'40'2' 20'

2 20

3'20I '44oS6I40'3'20'6'40o3' 20'

Gross3f2offGross2 20'

GrossGross

Ocular movemnentanomaly

Vertical Horizontal

2 0I 0I 0

I 0o oI 0

I 0

o 0I 0o 0o 0

2 2I 0

0 2o 00 II 20 II 0I 0

0 I

2 I

2 0

I 2

2 0o o2 0

*Surgical treatment givenANo microtropiaMicrotropia-small angle convergent deviation of < S' with evidence of binocular visionOcular movement anomaly: o = no evidence, I = some evidence (at least one report in the subject's case record), 2 = clear evidence

(repeated consistent observations).Prism fusion range: o = Good ( > 30 BO, > 8HBI), I = Fair (20O-2Z9BO, 4A-7A ), 2 = Poor ( < 20"B0, < 4ABI)

vision (nos 22, 24, 25, 26, and 27) while one retainsa constant manifest squint (although of reducedangle). Bifoveal binocular single vision was notpresent in any of the patients of this group at thetime of the original diagnosis nor was it achievedas a result of treatment.

2. STEREOACUITIES

Group i had a median stereoacuity of 50' which wassignificantly better than the 3' 20' of Group 2(Mann-Whitney U = i6'5, N, = I Ii N2 = I0,P < oo I, one-tailed). Group 3 subjects were too fewin number to warrant significance tests but theTable shows that most of them (four out of six)possessed only gross stereoscopic ability on the Wirttest (that is, they were successful with the fly figure

which introduces the Wirt test but not with any ofthe stereoacuity test patterns). Thus the Group 3

Wirt scores were broadly in keeping with the factthat Group 3 subjects were incapable of fusing eitherthe non-contoured or contoured stereograms. (It isperhaps worth noting in this connexion that the twoGroup 3 subjects with better than gross stereoacuityon the Wirt test were at least able to see the outlinesquare in depth in the contoured stereogram,although not the elements enclosed within thissquare.)

3. ANOMALIES IN OCULAR MOVEMENTS

This assessment ascertained whether or not thesubject had a history of horizontal and/or- verticalocular movement anomalies in either or both eyes.

CaseGroup no.

I I

2

3

4

56

78

9I0

II

2 12

'3'4'5i6

I7x8

'920

21

3 22

23

24

25

26

27

Sex

FFFM

M

M

F

F

M

M

M

M

F

M

F

F

M

F

M

M

F

M

M

F

F

F

Age(years)

12

'49II

8

9

9I2

9I0

II

12

9i8

I0

98

8

'312

9

I I

99'3'3'3

Prismfusionrange

0I

0

I

2I

I

I

0I

2

2

I

I

I

2

0

I

2

00

I

2

I

2

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from

Use of random-dot stereograms 549

The evidence from each subject's records concern-ing each direction of movement was rated on athree-point scale: o=no evidence, i = some evidence(at least one report in the subject's case record), and2= clear evidence (repeated observations of a con-sistent kind). The Table shows the results of thisassessment; whereas Groups i and 2 are verysimilar in terms of their histories of vertical ocularmovement anomalies, they are very different whenit comes to horizontal movement anomalies. Thusnone of the i i subjects of Group i had any recordwhatsoever of horizontal anomalies whereas six ofthe io subjects of Group 2 showed at least someevidence of this disorder. This difference betweenGroups i and 2 is significant (Fisher's exact test,P=o0oo3).

4. PRISM FUSION RANGES

The Table also includes a summary of the resultsfrom tests conducted with a prism bar at I/3 m. Thefusion ranges so obtained were rated on a three-point scale as follows: o= Good (>30oABO,> 8ABI); I = Fair (20A-29ABO, 4A-7ABI); 2=Poor (<20ABO, <4ABI). The Table shows thatno patients in Group i had a Poor rating whereasthree in Group 2 had (a non-significant difference).Group 3 patients, however, present a more uniformpattern in that four (out of six) had Poor ratingsand none had a Good one.

DiscussionThe principal finding of the experiment was thatthe 27 subjects fell neatly into three groups on thebasis of their random-dot stereoability. We shallattempt here some interpretation of this classifica-tion.The clinical histories of Groups i and 2 indicate

that Group I subjects in general had never com-pletely lost normal bifoveal binocular single vision,whereas this applied only to one subject in Group 2.The likelihood is therefore that for Group 2 subjectsthe presence of a constant deviation for a period oftime has led to an impairment in their stereopsisor in a failure to develop it in the normal way.The same argument can be applied to Group 3subjects, except that in their case the generally moresevere nature of their diagnoses seems to haveproduced a more serious result. This is reflected inthe fact that four out of six subjects in this lattergroup could obtain only gross stereopsis with theWirt test.The question which arises, therefore, is what is

the nature of the stereopsis deficiency which allowsGroup 2 subjects to see our random-dot stereogramssuccessfully only when a contour is added? In orderto discuss this question and the probable benefitwhich could be conveyed by the contour for these

subjects, a digression into the fusional problemspresented by random-dot stereograms is required.

Julesz (1971, p. 2I6) has pointed out that fusingrandom-dot stereograms of the kind used here,which possessed fairly large disparities, wouldrequire a vergence shift on the part of the observerduring the fusional process. Consider the stereogramof Fig. i as it appeared in the present experiment.The observer would begin by fixing the screen onwhich the stereogram was projected and, providinghe was capable of experiencing binocular vision,he would immediately secure fusion of the back-ground elements of the stereogram because corre-sponding background elements in each field of viewwould fall on corresponding points on the tworetinae. The elements comprising the centralsquare, however, could not be fused while fixationwas maintained on the plane of the screen. This isbecause corresponding elements in the left and rightimages of the central square possessed a disparity of12', which would cause them to fall outside thedisparity limits for establishing binocular fusion.The evidence for this argument comes from Fenderand Julesz (I967) who studied Panum's fusionalareas for random-dot stereograms using stabilizedimage techniques. They found that left and rightimages have to be aligned within about 6' disparitybefore fusion can take place. A vergence shift willtherefore be required to bring the images of thecentral square to within this disparity. Note that thefusion of the background elements would not bedisturbed by this vergence shift because Julesz andFender have also shown that, once fusion of arandom-dot stereogram has been established, leftand right images can undergo horizontal misalign-ments of about 2° before fusion is lost.The important point made in the above discussion

is that random-dot stereograms with disparitiesgreater than about 6' require a vergence shift beforethey can be fused. The stereograms used in thepresent experiment had disparities of 12' and sothey would require just such a vergence shift. Thesuggestion which immediately presents itself con-cerning the facilitating role of the contour for theGroup 2 subjects is that it helped them producethis necessary shift. Note that the contours of theoutline square were uniocularly identifiable in eachfield, unlike the edges of the 'hidden square' in thenoncontoured stereogram which appear only afterbinocular fusion has been obtained. This propertyof uniocular identifiability could have made theexecution of appropriate vergence movements mucheasier in the contoured stereogram task. Thissuggestion is supported by the work of Westheimer(1971) who showed that the kind of uniocularfeatures possessed by the square would indeed beadequate for initiating and driving vergencemovements.

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from

550 British Journal of Ophthalmology

If this vergence suggestion is valid, the particulardisability characterizing the Group 2 subjects was aweakness in their vergence mechanisms, a weaknesswhich probably had its origin in the period in theirclinical history during which bifoveal binocularsingle vision was absent. The disability might residein one or more parts of the vergence control path-way. It might, for instance, lie in the disparity-detecting cells which are probably used to control,initiate, and guide these movements (Blakemore,I970). Alternatively it may be located in the motormechanisms themselves. In the latter case, itmight be that subjects with the disability find itdifficult to produce integrated vergence shiftsvoluntarily or spontaneously (which is exactly whatan uncontoured random-dot stereogram with a largedisparity requires), but that they can do so understimulus control (that is, when, as in the contouredrandom-dot stereogram, there is an adequatestimulus to trigger the vergence movementsdirectly).Some qualitative observations made by various

Group 2 subjects confirm this vergence interpreta-tion of their random-dot performance. They re-ported while viewing the non-contoured stereo-gram that, although they could not see a square-shaped area of elements standing out in a singledepth plane, none the less some elements in thecentral area did appear to stand out on their own,each in a different depth plane. This suggests thatthese subjects were fusing the two halves of thestereogram in an unusual manner, perhaps becausethey had become locked into an imperfect fusion ofthe two fields from which they could not 'escape'because of the inability to control their vergencemovements appropriately. Such imperfect fusionswould, of course, be possible because of the in-herent ambiguity of random-dot stereograms (seepage 545).That the Group 2 subjects could do as well as

they did on the Wirt test while failing to fuse thenon-contoured stereogram is, of course, com-patible with regarding them as subjects withvergence difficulties. The Wirt test contains manyprominent contours which could guide vergencemovements in just the same manner as the outlinesquare might have done in the contoured stereo-gram. (Caution is necessary, however, when makingdetailed comparisons between performance on theWirt and performance on our stereograms since animportant difference exists between the two testingsituations: the Wirt is usually placed in the hands ofthe patient and viewed at reading distance, whereasour stereograms were projected at some distancefrom the patient.)Nornal subjects also benefit from the inclusion

of prominent contours in random-dot stereogramsif the disparities are very large and therefore require

large vergence shifts. Thus Saye and Frisby (I975)have shown that contours shorten the stereopsisperception time for stereograms with a disparity of> I° but that they do not do so for similar stereo-grams with a disparity of only 5'. This result iseasily explicable in terms of vergence, with thecontour facilitating the large vergence shift requiredfor the high disparity stimulus but with no benefitconveyed for the low disparity stimulus becauseno such shift is needed.On initial consideration, it might appear that our

vergence explanation of the subject groupings wouldpredict less good prism fusional ranges in Group 2than in Group i as this range measure provides anindex of vergence capability. In fact, the two groupswere not significantly different in this respect,although the presence of three subjects in Group 2with poor fusional ranges (nos i6, I9, and 2I)contrasts with the absence of subjects in Group iwith this assessment and is therefore suggestive.However, two considerations must be borne inmind when assessing the prism fusional rangescores in connexion with the present subjectgroupings:i. The disparity incorporated in the random-dot

stereograms was only I2'. This represents avergence shift equivalent to that required for aprism of considerably less than IAk It is generallyrecognized that it is impossible in the clinicaltesting situation to detect reliably fusional move-ments to prisms of less than about 2A\. Ac-cordingly, even if Group 2 had in fact possessedsignificantly poorer prism fusion range scores,they might still have had sufficient fusionalreserves to deal with the very slight vergenceshift required by the present stereograms. Inshort, the prism fusional ranges scores arerather too crude to be directly relevant to thepresent stimuli.

2. Inability to produce appropriate vergence move-ments to the uncontoured stereograms, thedefining characteristic of Group 2 subjects asfar as our vergence explanation is concerned,need not necessarily be associated with a poorfusional range as measured with a prism bar,regardless of the question of sensitivity. This isbecause the uncontoured stereogram presentsquite a different fusional task from that providedby the prism bar. In the latter case, there aremany prominent uniocularly visible contours inthe field of view which can guide an appropriatevergence movement (indeed, the prism task issimilar to the contoured stereogram situation inthis respect). But in the former case, there are nostimulus features which can trigger the requiredvergence movements. Consequently, even a goodfusional range score as measured with a prism barneed not imply the capacity to fuse the uncon-

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from

Use of random-dot stereograms 55x

toured type of random-dot stereogram. If we

consider the possibility discussed earlier, that theGroup 2 subjects possessed a limitation inproducing integrated vergence shifts voluntarilyor spontaneously in the absence of directstimulus control, then the prism fusional taskcould be unhelpful in understanding the natureof their difficulties.A very marked difference between Groups i and

2 concerned ocular movement anomalies. Group i

contained no subjects with horizontal ocular move-ment anomaly but there were six out of io subjectswith this movement in Group 2. In discussing thisresult, it must be emphasized that the limitations or

overactions in ocular movement recorded were

minimal and there was no evidence to suggest thatany of them were related to paralytic squint. In no

case were they symptom-producing, nor did theygive rise to a compensatory head posture. Two of thethree patients in Group 2 who had clear evidence ofhorizontal anomaly had acquired this limitationafter surgery (nos 12 and 14) but even with thisfinding they should have had no difficulty in main-taining binocular single vision of a centrally-placedstimulus once they had obtained fusion.The findings of this marked difference between

Groups i and 2 was unexpected and it is difficultto relate it satisfactorily to their different per-

formances with the random-dot stereograms.From one point of view it could be taken as evidencein favour of our vergence explanation of theGroup i/Group 2 separation in that it suggests thatthe two groups did indeed differ in terms of eye

movement control. But it must be emphasized thatthe gross conjugate eye movements required in theocular movement test are very different from theslight vergence movements required for fusion of therandom-dot stimuli. It may be that the importantcommon property shared by these two test situa-tions is that they both involve a certain degree ofvoluntary control over eye movements. In the ocularmovement test the patient has to direct his gaze

deliberately to the point indicated in the field ofview; and while attempting to fuse an uncontouredrandom-dot stereogram he has to scan the stimulusand create a vergence shift in the absence of any

prominent feature. The contoured random-dotstereogram presents a contrasting task. There thepresence of a contour in each field of view producesthe required vergence shift directly, without any

deliberate effort on the part of the subject. Thisspeculation about voluntary control can thus makesense of the data but it would of course be unwiseto draw any firm conclusions at this stage. Furtherevidence from a larger sample is clearly requiredand we are currently engaged on this task.Groups i and 2 also differed in terms of their Wirt

stereoacuity scores and so the question arises

whether this factor alone can provide any basis forthese groupings. First, it is important that con-siderable overlap was shown by the two groups (seethe Table) and it would be incorrect to concludethat the two groups were clearly separated by theWirt test. Secondly, it is difficult to see why in-clusion of a contour in a random-dot stereogramshould have helped subjects with relatively less goodstereoacuity as measured on the Wirt test. In thisconnexion it was noted that the worst Wirt stereo-acuity found in either Group i or Group 2 was6' 40. This, despite the differences existing betweenthe Wirt and random-dot stereogram test situations,suggests that all these subjects would probably havehad no difficulty in coping with the I2' disparityincorporated in the random-dot stereograms ifstereoacuity was the critical factor. Thus it seemsunlikely that the important underlying factor de-termining the Group i/Group 2 separation wasstereoacuity per se and it is more probable that thediffering stereoacuities of the two groups wasassociated with the generally less severe clinicaldiagnoses for the Group i subjects as comparedwith the Group 2 subjects.

Finally the relevance of our findings for clinicalpractice needs to be assessed. The prime conclusionis that random-dot stereograms can discriminatebetween patient categories. In particular, the simpleuncontoured stereogram is unlikely to be fusedsuccessfully by any patient who has lost bifovealbinocular single vision for any length of time.This conclusion confirms that drawn by Reineckeand Simons. The present investigation, however,goes further in showing that random-dot stereo-grams can be fused by many patients with thishistory if they are helped by the inclusion of acontour. We suggest that this fact may be veryrevealing in assessing the consequences for stereop-sis in a patient with a history of disturbed binocularsingle vision.

SummaryRandom-dot stereograms were shown to a sampleof strabismic patients for whom there was clinicalevidence of stereopsis. Two kinds of stereogramswere used, one of the usual sort and the other havinga contour surrounding the disparate area in eachfield of view. The patients tested could be clearlyclassified as belonging to one of three responsegroups. The first group could fuse both kinds ofstereogram, the second could fuse only the con-toured kind, and the third could not fuse eitherkind. This grouping was found to relate to thedegree to which bifoveal binocular single vision hadbeen absent in the clinical histories of these patients.The result is discussed in terms of the conse-quences for vergence and stereopsis of a period ofabsence of normal binocular function.

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from

552 British Journal of Ophthalmology

References

BENTON, A. L., and HECAEN, H. (1970) Neurology (Minneap.), 20, I089BLAKEMORE, C. (1970) J. Physiol. (Lond.), 209, 155CARMON, A., and BECHTOLDT, H. P. (I969) Neuropsychologia, 7, 29FENDER, D., and JULESZ, B. (I967) J. opt. Soc. Amer., 57, 8i9JULESZ, B. (1971) 'Foundations of Cyclopean Perception'. University of Chicago Press, ChicagoREINECKE, R. D., and SIMONS, K. (I974) Amer. J. Ophthal., 78, 714RICHARDS, W. (1973) US Air Force Office of Science Research Report No. 73-0439, 1400 Wilson Boulevard,

Washington, DCSAYE, A., and FRISBY, J. P. (1975) (In press)WESTHEIMER, G. (I97I) In 'The Control of Eye Movements', eds P. Bach-y-rita, C. C. Collins, and J. E. Hyde,

pp. 473-482. Academic Press, New York and London

on 18 May 2018 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.59.10.545 on 1 O

ctober 1975. Dow

nloaded from