Binocular VEP Summation in Infants and Adults WithAbnormal Binocular Histories

Sandra L. Shea,* Richard N. Aslin,* and Daphne McCullochf

Visual evoked potentials (VEPs) were recorded under binocular and monocular viewing conditions in19 2- to 10-month-old infants, 19 stereodeficient adults, and 12 normal adults. VEPs were elicited bymedium contrast, phase-alternated checkerboards with check sizes ranging from 10-52 min of arc.Binocular VEP summation was defined as the percentage by which the binocular VEP amplitude exceededthe mean of the two monocular VEP amplitudes. In the stereonormal adults, binocular VEP summationwas significantly greater than zero and significantly less than 100%. In the stereodeficient adults, binocularVEP summation was not significantly greater than zero. These findings support the hypothesis that themagnitude of binocular VEP summation reflects the activation of binocular cortical neurons. However,binocular VEP summation in the infants averaged approximately 95%. Moreover, infants under 5 monthsof age, many younger than the reported age of onset for stereopsis, showed binocular VEP summationof nearly 145%. The significantly higher level of binocular VEP summation in the infants was the resultof much larger binocular VEP amplitudes, while monocular VEP amplitudes were equivalent to thoseof stereonormal and stereodeficient adults. These results support the hypothesis that VEP amplitudeis mediated by two independent pools of monocular cortical neurons and that binocular VEP amplitudein stereonormal and stereodeficient adults saturates at a lower level than in infants. Thus, it is hypothesizedthat binocular VEP summation is not representative of the activation of binocular cortical neurons.Invest Ophthalmol Vis Sci 28:356-365, 1987

Performance on a variety of psychophysical tasks isbetter under binocular viewing conditions than undermonocular viewing conditions. If this binocular su-periority is greater than the effect of probability sum-mation, then neural integration of binocular inputs,or binocular summation, can be inferred.1'2 The ab-sence of binocular summation in adults deprived ofnormal binocular inputs because of strabismus providesthe strongest evidence that binocular summation ismediated by binocular cortical neurons.3"6

Despite extensive research on binocular summationin adults, few systematic investigations have been con-ducted on human infants.* There are two reasons why

From the Center for Visual Science,* University of Rochester,Rochester, New York, and the School of Optometry.f Indiana Uni-versity, Indiana.

Supported by NIH Grant EY04392 to RNA. Research conductedat Indiana University—Bloomington, Bloomington, Indiana.

Submitted for publication: October 21, 1985.Reprint requests: Sandra L. Shea, PhD, Center for Visual Science,

University of Rochester, Rochester, NY 14627.* A recent study7 used flash VEPs and anaglyphic stimuli to mea-

sure infants' and adults' interocular suppression. The authors reportedthat while infants as young as 3 months of age showed interocularpattern suppression, they did not show size-specific suppression, im-plying that somewhat different mechanisms may be involved in thesetwo types of interocular suppression. It has also been shown thatadult subjects with abnormal binocular experience may have normalsize-specific interocular suppression4 or pattern suppression8 but lackboth stereopsis and binocular summation. The use of this techniqueto measure interocular suppression, particularly in strabismics, hasbeen criticized by Campos and Chiesi.9

studies of binocular summation in young infants couldprove to be important. First, several recent reports havedemonstrated that stereopsis does not emerge in normalinfants until the fourth postnatal month.10"12 Thesefindings, along with additional evidence,13"16 are con-sistent with the hypothesis that the postnatal emergenceof stereopsis coincides with the onset of functional cor-tical binocular neurons. Alternatively, binocular neu-rons may be present in young infants prior to theemergence of stereopsis, but above-chance performanceon tests of stereopsis may be prevented by attentionaldeficits, grossly inaccurate binocular alignment, orother task demands that are not overcome until laterin infancy. The development of more sensitive assess-ment techniques may reveal the presence of binocularsummation prior to the reported onset age of stere-opsis.17 Thus, if the presence of binocular neuronscould be inferred from the presence of binocular sum-mation, and if binocular summation emerged prior tothe onset age of stereopsis, one could conclude that anecessary neural substrate for stereopsis was presentprior to above-chance performance on tests of stere-opsis.

A second reason for investigating binocular sum-mation in young infants is the need for accurate andreliable indices of binocular functioning in patientswith ocular anomalies, such as strabismus and ambly-opia. Although a variety of clinical measures of bin-ocular functioning are available for use with verbalchildren18"21 and adults,22 recently developed tests of

356

No. 2 BINOCULAR SUMMATION OF VEPs / Shea er ol. 357

stereopsis in infants may be relatively time-consumingand require the absence of gross ocular misalignments.Because the level of binocular functioning is often usedto guide the treatment of ocular anomalies, a rapidassessment that does not require accurate binocularalignment would be advantageous, particularly if itwere possible to determine whether functional corticalbinocular neurons were present in strabismic infantsprior to corrective surgery.

These two reasons for studying binocular summationin young infants, one aimed at determining if corticalbinocularity is present prior to the onset age of stere-opsis, and the other aimed at determining if preverbalinfants with ocular anomalies have functional corticalbinocular neurons, both suggest the use of the visualevoked potential (VEP). If the VEP does accuratelymeasure functional binocular vision, then it has enor-mous potential as a diagnostic tool. Conversely, if theVEP does not measure binocularity, then its use inbinocular testing must be qualified. Unfortunately, as-sessments of binocular summation in normal, strabis-mic, and amblyopic adults using the VEP have pro-vided both varying definitions of summation and in-consistent results.9'23"31 As noted earlier, psychophysicalevidence of binocular summation consists of binocularperformance that exceeds monocular performance bya factor greater than 1.4 (probability summation fortwo independent channels reduces the variance by V5).In the VEP literature, however, binocular summationhas been given a variety of names, including summa-tion,24 addition,3233 and facilitation.23 The specific def-initions of these names varies considerably dependingon the experimental condition or calculation used.23'24

Despite these inconsistencies, some general trendshave emerged. Normal adults tested with phase-re-versed patterned transient VEPs yield binocular VEPamplitudes! consistently greater than either of the twomonocular VEP amplitudes, although rarely greaterthan the sum of the monocular amplitudes. Significantlevels of amblyopia (an interocular acuity difference oftwo or more octaves) are characterized by little or noenhancement of the binocular VEP amplitude over thelarger of the two monocular VEP amplitudes.27 Finally,a signal-to-noise analysis of steady-state VEPs in nor-mals24 revealed all possible types of binocular VEP in-teraction, including facilitation (a doubling of themonocular amplitude), summation (greater binocularthan monocular amplitude), and inhibition orsuppression (a lower binocular than monocular am-plitude). Thus, in contrast to the psychophysical evi-

f In the current study, amplitude refers to the difference betweenthe peak that occurs around 100 msec (PI) and the trough (N2) thatfollows PI. While some researchers have used other landmarks (ie,the Nl to PI amplitude), the PI to N2 amplitude was chosen sinceboth PI and N2 occur reliably even in tracings obtained from infants.

dence that amblyopes and strabismics who presumablylack binocular cortical neurons do not show binocularsummation, binocular VEP summation in these pa-tients is extremely variable. These findings suggest ei-ther that binocular summation of the VEP does notrequire binocular cortical neurons, or that other neuralfactors (e.g., cortical inhibition) can modulate the am-plitude of the binocular VEP.

The only study which investigated the phenomenonof binocular patterned VEP summation in human in-fants was reported by Amigo et al.23 They presentedphase-alternated, vertically oriented gratings of differ-ent spatial frequencies and contrasts to normal infantsand children, strabismic children, and normal andstereodeficient adults. Amigo et al23 devised an index,the Incremental Binocular Amplitude (IBA), to mea-sure the difference between monocular and binocularVEP amplitudes. The IBA is the percentage differencebetween the binocular VEP amplitude and the largerof the two monocular VEP amplitudes. Amigo et al23

found widely varying amounts of binocular VEP sum-mation in infants and children between the ages of 2months and 12 years. Moreover, because low levels ofbinocular VEP summation tended to be correlated withabnormal binocular experience, Amigo et al23 con-cluded that this application of VEPs would provide". . .an objective method for screening out defects ofbinocular vision."

There are two problems associated with the Amigoet al23 study (see also Campos and Chiesi9). First, al-though all subjects with normal stereopsis showed sig-nificant binocular VEP summation, some subjectswithout measurable stereopsis, including many infantsunder 4 months of age, showed significant levels ofbinocular VEP summation. Thus, binocular VEPsummation, as assessed by their IBA index, appears tobe a necessary, but not sufficient, correlate of the typicalclinical criterion of normal binocular function (i.e.,stereopsis).

Second, the IBA index is based on the superiorityof the binocular VEP amplitude over the larger of thetwo monocular VEP amplitudes. Amigo et al23 usedonly the larger of the two monocular VEP amplitudes,because their normal subjects' monocular VEP am-plitudes did not differ by more than 10%. However,other researchers27 have reported much larger inter-ocular differences in the VEP amplitudes of amblyopes,with the amblyopic eye typically showing the loweramplitude. Therefore, if the difference in monocularVEP amplitudes is consistently more than 10%, thelogic of the IBA must be reexamined. Using the largerof the two monocular VEP amplitudes always yieldsthe smallest, most conservative estimate of binocularVEP summation, because it ignores the contributionof the amblyopic eye. Conversely, using the smaller ofthe two monocular VEP amplitudes yields the largest,

358 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / February 1987 Vol. 28

most inflated estimate of binocular VEP summation,because it ignores the contribution of the non-am-blyopic eye. A measure of binocular function shouldrelate the combined information obtained from botheyes to the information obtained individually from thetwo eyes. Apkarian et al24 devised an alternate indexof binocular VEP summation that expresses the bin-ocular VEP amplitude relative to the mean of themonocular amplitudes.^ Other researchers have usedthis measure28 and similar indices929"31 to evaluatebinocular VEP summation in normal and strabismicadults. As in the Amigo et al23 study, these reports havedemonstrated that binocular VEP summation is ex-tremely variable and is not consistently present inadults with strabismus, amblyopia, and/or suppression.Thus, it appears that the results on binocular VEPsummation, regardless of the index used, do not parallelthe psychophysical evidence that supports the absenceof binocular cortical neurons in strabismic adults.

These variable findings and interpretations in ste-reodeficient adults, together with the limited data onbinocular VEP summation from young infants, makeit important to determine the relation between bin-ocular VEP summation and stereopsis in these twopopulations who presumably lack binocular corticalneurons. If binocular VEP summation is always presentin stereonormal adults and never present in stereode-ficient adults, then stereopsis and binocular VEP sum-mation would appear to share the same underlyingmechanism. If binocular VEP summation is alwayspresent in stereonormal and stereodeficient adults, thenstereopsis and binocular VEP summation would notappear to share the same mechanism. If binocular VEPsummation is always present in stereonormal adults,but present in only some stereodeficient adults, thenagain there would appear to be at least two underlyingmechanisms. Finally, if infants younger than the onsetage for stereopsis show binocular VEP summation, andif the neural mechanism for stereopsis is absent priorto this age, then separate mechanisms must underliebinocular VEP summation and stereopsis.

These possible relations between binocular VEPsummation and stereopsis were assessed by measuringVEP amplitudes in infants younger than the onset agefor stereopsis, infants older than the onset age for ste-reopsis, stereonormal adults, and stereodeficient adults.It is important to note that, despite differences in theVEP waveform between infants and adults,34"41 large

t To illustrate the difference between this index of binocular VEPsummation and the IBA used by Amigo et al,23 consider the followinghypothetical VEP amplitudes: OU = 12 ^V, OD = 8 (iV and OS= 10/zV. The IBA index would be(12-10)/10 X 100 = 20%, whereasthe Apkarian et al24 index would be 1.33, and the Nuzzi and Franchi28

modified index would be {12 - [(10 + 8)/2]}/[(10 + 8)/2] X 100= 33%.

and reliable VEPs can be recorded from young infantsif the eliciting stimuli are of sufficient contrast and size.

Materials and Methods

Subjects

Nineteen infants between 73 and 305 days of agewere recruited by letter and by phone from the Bloom-ington, Indiana, area. An additional 11 infants wereexcluded from the final sample because of failure tocomplete the testing sessions due to crying, restlessness,or an intolerance of the eye patch. Infants were dividedinto a young (< or = 140 days, n = 9) and an old (> or= 160 days, n = 10) age group.

Adult subjects were recruited from newspaper ads(19 adults with a history of strabismus) and existinglaboratory records (12 normal adults). An additionaltwo strabismic adults were excluded from the samplewhen it was determined that their strabismus was con-founded by other factors. Specifically, one had retro-lental fibroplasia and a detached retina, and the otherhad optic nerve atrophy. The remaining adults with ahistory of strabismus were classified as small anglestrabismics (<10 prism diopters lateral deviation withor without amblyopia, n = 7), esotropic (ETs) (>10prism diopters convergent or vertical deviation withor without amblyopia and with and without alternatingsuppression, n = 8), and exotropic (XTs) (>10 prismdiopters divergent deviation with alternating suppres-sion, n = 4). 1 HCSC ciasMiicauons were based on patienthistories and on diagnoses performed on the day oftesting. Details of these subjects' histories, as well astheir deviations and acuities at the time of testing, arepresented with their summation scores in Table 1.Subjects were paid $5 for each visit. Informed consentwas obtained from each subject, or from a parent ofthe subject, after the procedure had been explained indetail and before testing began.

Procedure

VEPs were recorded from each subject using threegold Grass electrodes placed (a) along the midline 1-2 cm anterior to the inion, (b) on one earlobe as ref-erence, and (c) on the other earlobe as ground. Imped-ances, checked before testing and between each trial,remained below 15 kOhms (x stereonormal adult= 7.59, sd = 3.82; x infant = 5.40, sd = 2.73; x ster-eodeficient adult = 5.97, sd = 3.76). VEPs were ac-cumulated by, and stored in, a Nicolet CA-1000 signalaverager (Nicolet, Madison, WI) containing an artifactrejection mode which eliminated any sweeps contam-inated by blinks, head movements, or other transientartifacts. The signal averager defined an artifact as anysignal that deviated more than 95% from the amplitudeof the cumulative mean. Pilot data and other work42

No. 2 BINOCULAR SUMMATION OF VEPs / Sheo er ol. 359

Table 1. Characteristics of subjects with abnormal binocular histories

Group

Small Angle

ET

XT

Sub.

5*8*

10*11*15*18*19

123*7

1214*1617*

4

f 9*13

Deviation

noneETETnone/amblETETnone/ambl

accom ETETETET/amblET/amblEThypotropeET/ambl

alter XTalter XTalter XTalter XT

OD error

+.25-J5 X 180X

+ 1.75-1.5 X 180+6.0-.75 X 120-.50- .50X 180+4.0-4.0 X 100+4.0-.50 X 90

+4.5-.75 X 20+.75-1.0X20-1.0- .50X 160-4.0- .50X 170-.50-.50 X 90+.75-.5 X 180-.50-.25 X 150+2.25-1.0 X 180

-1.5-2.0 X 180- 2 . 0 - I . 0 X 7 5-1.75-1.5 X 100plano-.75 X 180

OS error

+.50-.50X 180X

+3.25-1.75 X 180+6.0-1.25 X 135+.50-.50X 180+4.0-2.0 X 105~-1.5-1.0 X20 + +

+4.25-.25 X 180plano-.50 X 135-.50-1.0 X 180~-7.0-.75 X 130r

-1.75-.75 X 120r

+2.0-1.75 X 180-.75-.5OX 180+.75-.75 X 180

-3.0-1.75 X 180-2.O-.5O X 90r

-2.0-.50 X 55r

+.75-.75 X 180~

TNOscore

>1980">1980"

60">1980">1980">1980"

30"

480">1980">1980"

X

>1980"X

>1980">1980"

>1980">1980">1980">1980"

Summation

26'

16.17-32.32-18.21

52.594.33

23.2119.72

34.59X

298.8220.64

131.8357.7519.8250.00

-30.86-39.30

53.52-9.76

W

60.2952.88

-2.5035.2431.3916.67

-17.07

-50.0015.53

121.0512.5413.029.8230.86

-16.36

-59.77-32.61

63.93-33.90

x unavailable data* Strabismus surgery' refractive prescription

indicated that 50 sweeps, in our case, of 250 msec each,were sufficient to obtain a reliable VEP signal.

Infants were held on a parent's lap in a chair posi-tioned 1 m from a 25-inch black and white televisionmonitor (visual field size of 27° X 23°). An experiencedobserver utilized a hand-operated interrupt switchwhich was hard-wired to the signal averager to halt theaveraging process when the infant looked away fromthe screen or started to cry. This observer also manip-ulated small, noisy toys in front of the television screento maintain the infant's attention. The observer stoodto one side of the monitor and could easily judge thedirection of the infant's gaze. Adult subjects sat in achair at the same distance from the screen.

At the end of each viewing condition, the observermonitoring the infant's gaze during data collectionprovided a judgment of the infant's overall attentive-ness (1 = highly attentive, 2 = cooperative with oc-casional lapses of attention, 3 = very uncooperative).Most viewing conditions yielded acceptable scores(44%, 51%, and 5% of l's, 2's, and 3's, respectively).Records with an attention level of 3 were not includedin the data analysis. When possible, data dropped be-cause of inattention were replaced by data collectedon a subsequent session.

The stimulus display was a black and white check-erboard§ with an average luminance of 2.5 ft lamberts

§ A checkerboard pattern was used instead of a horizontal or verticalgrating because it provided more contours to attract the infants' at-tention, it has been utilized in other infant studies,42-43 and pilot work

++ bifocals OU of +2.0"" error over and above prescription

(8.58 cd/m2) and a contrast of 0.50. These values weresimilar to earlier work.2328 All samples were collectedwith the room lights off. The checkerboard pattern wasphase-reversed at a rate of 1.88 alternations per second(0.94 Hz). Check size was 26 min and 52 min for theinfants and 10 min and 26 min for the adults. Thesecheck sizes were chosen to present one pattern that wasclearly above each subject's threshold for producing aclear VEP signal, and one pattern that was near orbelow threshold (for infant acuity thresholds see Dob-son and Teller;45 Teller46). The smaller of the two checksizes used with adults (10 min) was the smallest checksize the display system could generate at the 1 m view-ing distance, and was clearly above threshold for mostadults. Adults were tested while wearing their refractivecorrection.

Each subject was first tested under binocular con-ditions (OU) while viewing the larger of the two pat-terns. If a significant VEP amplitude was not obtained,impedances were checked and, if necessary, the checksize was increased by one octave. If a consistent butsmall VEP amplitude was obtained, the check size wasnot changed. If a consistent and large VEP amplitudewas obtained, the check size was decreased by one oc-tave.

After collecting this initial OU VEP and selectinga check size, a minimum of three additional VEPs

with normal adults indicated that with our protocol, stimulus displayand electrode placement, a checkerboard pattern elicited a larger andmore consistent response than did gratings (for a similar finding usingflash VEPs, see MacKay & Jeffreys44).

360 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / Februory 1987 Vol. 28

20CM

O 1 6 0 -

oo5

120 -

80 -

4 0 -

I I 52 min

26 min

10 min

1

StereonormalAdults

Infants StereodeficientAdults

Fig. 1. The percentage of binocular VEP summation shown bynormal adults, infants, and stereodeficient adults as a function ofcheck size. Bars = 1 SE.

were obtained in each complete session. These threeVEPs consisted of (a) binocular viewing, (b) monocu-lar viewing by the left eye (OS), and (c) monocularviewing by the right eye (OD). The order of these threeconditions was randomized across subjects. In themonocular conditions, the nonviewing eye was coveredwith a black patch. If the infant remained cooperativeafter this set of four VEPs, a second set of four wascollected using the next smaller check size. Alterna-tively, the second set was collected on a subsequentsession. Both sets of VEPs were collected from alladults on the same day.

In addition to the VEPs, a dynamic near retinoscopywas performed on all subjects. A 12-inch black andwhite television monitor, located 1 m from the subject,presented a portion of a movie or cartoon from anRCA videodisc to attract and maintain the infant's at-tention. The brightness and contrast levels of this moviematched the checkerboard pattern, and the retinoscopywas performed at the same distance (1 m) as the check-erboard. The purpose of this retinoscopy was to deter-mine whether (a) the subjects were accommodatingaccurately to the stimulus distance (±0.50 D), (b) therewas a gross difference in the refractive error betweenthe two eyes (± 1.0 D), and (c) any spectacle correctionworn by the adults was adequate. All subjects met thesecriteria.

Adults were also tested for their near Snellen acuityand their stereoacuity using a Rosenbaum pocket visionscreener and the random element TNO test21 (LamerisInstrumenten b.v., Utrecht), respectively.

Results

The performance of the two adult groups on theTNO test was compared to determine if the adult sub-jects with a history of strabismus were, in fact, stereo-deficient. As seen in Table 1, it was not possible tocompare mean thresholds on this task because 14 ofthe adults with abnormal binocular histories could notdetect the depth in any of the TNO targets (maximumdisparity = 1980 sec). In contrast, 10 of the 12 adultswith normal histories had stereoacuities of 60 sec orbetter. When the scores on the TNO were divided intowhat would constitute normal (60 sec or better) orworse than normal (480 sec or worse) scores, there wasa significant deficit in stereoacuity in the adults withabnormal binocular histories compared to the normaladults (X2 = 17.11, P < .001). The group of adultswith abnormal histories will subsequently be referredto as stereodeficient adults. Although we did not obtainstereoacuity estimates from the infants, summary datacompiled by Teller46 suggest that, based on the ages ofthe infants, only 63% would have shown evidence ofstereopsis.

The binocular summation index devised by Apkar-ian et al,24 and modified by Nuzzi and Franchi,28 com-pares the binocular VEP amplitude to the mean mon-ocular VEP amplitude, and was used throughout thepresent report. Figure 1 shows these mean binocularVEP summation scores for the stereonormal adults,the infants, and the stereodeficient adults at the threecheck sizes. Multiple comparisons using Dunn's pro-cedure revealed that the infants showed a significantlygreater amount of binocular VEP summation than ei-ther the stereonormal or the stereodeficient groups onthe only comparable check size of 26 min (overall P< .05).

The group differences in binocular VEP summationshown in Figure 1 were also reflected in the individualsubject scores. These scores were divided into the threecategories of binocular VEP summation described byNuzzi and Franchi28 (Table 2), and revealed differentdistributions across the normal adults, stereodeficientadults, and the combined infant groups (X2 = 9.493,P < .05 for the 26' check size common to all groups).Most subjects, regardless of group, showed summation(scores between 0 and 100%), indicating a binocularVEP amplitude that was greater than the mean mon-ocular amplitude, but less than twice as large. It is im-portant to note that high summation scores (near 100%)were neither the result of large interocular differencesin monocular VEP amplitude nor the result of a zero-amplitude VEP in one eye (monocular VEP amplitudeswere all > 1 /xv). However, the groups differed in thatmore stereodeficient adults showed inhibition (scoresless than 0%), indicating a binocular VEP amplitude

No. 2 BINOCULAR SUMMATION OF VEPs / Shea er al. 361

that was less than the mean monocular amplitude, andmore young infants showed facilitation (scores greaterthan 100%), indicating a binocular VEP amplitude thatwas more than double the mean monocular amplitude.Unlike Nuzzi and Franchi's28 results, some stereonor-mal adults showed inhibition scores, but only on the26 min check size, and these scores were only slightlybelow zero (range: -0.40% to -12.91%).

Separation of the stereodeficient adults into the threeclinical classifications (small angle, ET, and XT) re-vealed that none of these subgroups showed a signifi-cant amount of binocular VEP summation. || Separa-tion of the infant scores into a younger and an olderage group revealed that the younger infants tested onthe 52 and 26 min check sizes showed either sum-mation (n = 7) or facilitation (n = 7). In contrast, noneof the older infants showed facilitation on either checksize and one showed inhibition.

Figure 2 presents a further illustration of this differ-ence between the two age groups of infants. The youn-ger infants showed amounts of binocular VEP sum-mation that were significantly different from 0 on boththe 52 min (/ = 2.66, df = 6, P < .05) and 26 min (t= 3.00, df = 6, P < .05) check sizes, while the olderinfants showed significant binocular VEP summationon only the smaller 26 min check size (t = 3.27, df= 6, P < .02). Multiple comparisons using Dunn'sprocedure indicated that the two infant age groups dif-fered significantly from each other in the amount ofbinocular VEP summation shown on the 26 min checksize (overall P < .05), but not on the 52 min checksize. In addition, stereonormal adults showed a signif-icantly smaller amount of binocular VEP summationthan the younger group of infants on the 26 min checksize (overall P < .05).

The much greater binocular VEP summation scoresobtained from the infants could be the result of asmaller monocular amplitude or a larger binocularamplitude (or both). While the amplitude of VEPstends to be higher in infants than in adults,47 Figure 3shows that, under the present testing conditions, mon-ocular VEP amplitudes did not differ by age. This sim-

|| Multiple comparisons using Dunn's procedure, and its poolederror term, indicated that even though none of the individualsubgroups' summation scores differed significantly from zero, therewere some significant between-group differences because of the pres-ence of positive and negative summation scores. Specifically, the ETadults showed a significantly higher amount of binocular VEP sum-mation on the 26 min check size than the small angle, XT and stereo-normal adults (overall P < .05). On the 10 min check size, the ster-eonormal adults had significantly higher summation scores than theXT adults (overall P < .05). The infants' scores did not differ fromany of the stereodeficient groups, primarily because of the relativelylarge variability in all subgroups.

Table 2. Number of subjects showing differentcategories of VEP summation

Group

Stereonormal(n = 12)

Young Infant(n = 9)

Older Infant(n = 10)

Stereodeficient(n = 19)

Check size

52'26'10'52'26'10'52'26'10'52'26'10'

Facilitation

_0134

—00

—21

Summation

71143

—37

—1111

Inhibition

5000—10

—57

Note: Missing data points are due to infants' sessions which were discardedbecause of poor attention, infants who contributed one good session and couldnot return to the laboratory because of scheduling conflicts or illness, and oneadult's session which was interrupted by mechanical failure and could not berescheduled.

ilarity in monocular VEP amplitudes provides a base-line across groups from which the differences in thebinocular amplitudes can be compared. As shown inFigure 3 for the 26 min check size common to allgroups, it appears that the higher levels of infant bin-ocular VEP summation are the result of larger binoc-ular VEP amplitudes (F(3, 40) = 11.30, P < .001).Monocular VEP amplitudes were not significantly dif-ferent between groups (F(3, 40) = 1.64, P = . 19). Mul-tiple comparisons using Dunn's procedure revealed

oO

200 -,

160-

120 -

8 0 -

I I 52 minESS 26 mini ^ H 10 min

StereonormalAdults

1 I

YoungInfants

OlderInfants

Fig. 2. The percentage of binocular VEP summation shown bynormal adults, young infants (<140 days), and old infants (> 160days). Bars = 1 SE.

362 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Februory 1987 Vol. 28

2 5 i

20-

uj 1 5 -Q

10 -

5-

1 IN\N BinocularI I Monocular

1I X

YoungInfants

OldInfants

Stereonormal StereodeficientAdults Adults

Fig. 3. Mean VEP amplitudes of young infants, old infants, normaladults, and stereodeficient adults at the check size of 26 min as afunction of viewing condition. Bars = 1 SE.

significant differences between the binocular VEP am-plitudes of (1) young and old infants, (2) young infantsand stereonormal adults, (3) young infants and stereo-deficient adults, and (4) old infants and stereodeficientadults (overall P < .05). The difference between ster-eonormal adults and older infants was not significant,nor were the differences between the two groups ofadults. None of the differences between the meanmonocular amplitudes were significant. Finally, acrossall adult subjects, median tests indicated that the levelof stereoacuity was not associated with the magnitudeof binocular VEP summation at either check size (X2

= .27, n.s., X2 = .31,n.s.).

Discussion

The binocular VEP summation scores obtained inthe present study support two major conclusions. First,infants showed more binocular VEP summation thaneither stereonormal or stereodeficient adults. Further-more, younger infants showed more binocular VEPsummation than older infants, despite the fact thatmany were younger than the reported mean onset agefor stereopsis. This larger amount of binocular VEPsummation is not due to the larger monocular andbinocular VEP amplitudes generally found in infants,47

because the binocular VEP summation index is ex-pressed as a ratio that is uninfluenced by absolute VEPamplitudes. These large binocular VEP summation

res, particularly in the younger infants, contrastmarkedly with the findings of Amigo et al.23 Only oneof their nine infants under 5 months of age showedscores greater than 100%, whereas five of our nineyounger infants showed facilitation of the binocularVEP (two infants on both check sizes, one on only the

52 min size, and two on only the 26 min size). Thisdiscrepancy is not solely the result of the different in-dices used to compute binocular VEP summation. Us-ing Amigo et al's23 more conservative index, four ofour younger infants still showed facilitation (two infantson both check sizes and two more on only the 26 minsize).

It is apparent from Figure 3 that the higher levels ofbinocular VEP summation among the infants are pri-marily the result of larger absolute binocular VEP am-plitudes. These larger binocular VEP amplitudes areunlikely to be solely the result of attentional mecha-nisms. First, the reduction in binocular VEP ampli-tudes with age would imply a significant developmentaldecrease in attention. However, this implication is notsupported by the behavioral ratings in the binocularcondition, which averaged 1.25 in the younger infantsand 1.41 in the older infants (t = 1.47, df = 75, P= n.s.). Second, the younger infants' monocular VEPamplitudes may have been depressed more than theolder infants' because the eyepatch was more distract-ing. If so, the younger infants' behavioral ratings shouldhave been poorer than the older infants'. In fact, thereverse was true, as 33% of the young infants' ratingsin the monocular viewing conditions were assigned thehighest score of " 1 , " while none of the older infantsachieved this rating while wearing the patch.

The second conclusion of the present study is thatstereonormal and stereodeficient adults do not showthe same pattern of binocular summation as measuredby VEPs. The stereonormal adults showed both goodstereoacuity and a significant amount of binocular VEPsummation, which was greatest at the smaller checksize. The group of stereodeficient adults did not showa significant amount of binocular VEP summation,and these summation scores were only weakly corre-lated with stereoacuity or with a history of strabismus,a finding which supports some earlier reports.28'30 Therange of binocular VEP summation scores observed inthe stereonormal adults (—12.91-103.6%) was com-pletely subsumed by the range of scores from the ste-reodeficient adults (Table 1). This variability, in com-bination with the weak correlation between binocularVEP summation and stereoacuity or binocular history,renders binocular VEP summation a poor prospect asa clinical test of functional binocularity. These findingsindicate that the type of binocular summation mea-sured by VEPs may be an index of fusion independentof bifoveal fixation and strabismus,9 and that binocularVEP summation may not be a reliable predictor of thelevel of stereopsis.28'48

Amigo et al23 did not find significant amounts ofbinocular VEP summation in their sample of 14 ste-reodeficient children and adults, but their overall scores

No. 2 BINOCULAR SUMMATION OF VEPs / Shea er al. 363

were lower than those in the present report. There areseveral possible explanations for this. First, Amigo etal23 used slightly different contrast and luminance lev-els. Second, as noted earlier, Amigo et al23 used a con-servative statistic which was less likely to detect signif-icant amounts of binocular VEP summation. Finally,Amigo et al23 did not report whether their subjects weresuppressors, although none were exotropes.

The finding that some stereodeficient adults do notshow significant amounts of binocular VEP summationis consistent with psychophysical evidence supportingthe hypothesis that binocular summation is mediatedby binocular cortical neurons. However, the largeamounts of binocular VEP summation found in theinfants is inconsistent with that conclusion if thesebinocular cortical neurons are necessary and sufficientfor stereopsis, and if the reported onset age of 4 monthsfor stereopsis is correct. If binocular VEP summationwere highly correlated with stereopsis, or if they sharedthe same neural mechanism, then young infants underthe onset age for stereopsis should have less binocularVEP summation than older infants or stereonormaladults. Moreover, if binocular VEP summation werecorrelated with stereopsis, then binocular VEP sum-mation would increase, not decrease, with age, and ste-reopsis and binocular VEP summation would be atleast moderately correlated in stereonormal and ste-reodeficient adults.

Several researchers28'31 have suggested an alternativeto the hypothesis that the larger VEP signal under bin-ocular conditions is due to the excitation of binocularcortical neurons. Because there are cortical neuronsthat respond to inputs delivered to only the right orleft eye, one could hypothesize two monocular poolsof neurons whose overall firing pattern under binocularconditions is greater than under monocular conditions.This hypothesis is supported by the finding that bin-ocular VEP summation was either absent or greatlyreduced in subjects who suppressed one eye duringbinocular viewing conditions. While the present studydid not test suppression during the collection of VEPs,the absence of binocular VEP summation in suppres-sors parallels findings by other researchers.49 Alternat-ing suppression should result in the activation of, atmost, one pool of monocular neurons, regardless ofviewing condition. Thus, the expected binocular VEPamplitude in suppressors should never exceed eitherof the two monocular VEP amplitudes.

The hypothesis that binocular VEP summation maysimply be due to the activation of a larger pool of cor-tical neurons under binocular viewing conditions couldaccount for the results from our infants, stereonormaladults, and stereodeficient adults if we make the fol-lowing assumptions. First, in early infancy, prior to the

onset of stereopsis, the amplitude of the binocular VEPis approximately equal to the algebraic sum of the twomonocular VEP amplitudes. This complete summa-tion (equal to a binocular VEP summation score of100%) results from the activation of two independentpools of monocularly driven cortical neurons. Second,after the fourth postnatal month, as the neural mech-anism for stereopsis emerges, the contribution of thetwo pools of monocular cortical neurons to the am-plitude of the binocular VEP changes. Specifically,there is an increase in the magnitude of interocularVEP inhibition elicited under binocular conditions. Asa result, the binocular VEP in the normal, mature vi-sual system is characterized by an interocular inhibitionthat effectively saturates the combination of the twomonocular VEP amplitudes, resulting in binocularVEP summation scores below 100% but greater than0%. Finally, some individuals with ocular anomaliesdevelop even greater interocular inhibition than nor-mals, resulting in binocular VEP summation scores ofapproximately 0%. The neural mechanism assumed tounderlie this interocular cortical inhibition does notreside in the number or percentage of binocular corticalneurons (which mediate psychophysical binocularsummation and stereopsis), but rather in the gating onand off of excitatory and inhibitory neurons in the cor-tex. Recent data on single unit and VEP recordingsfrom cats indicate quite clearly that binocular VEPsummation does not require the presence of binocularcortical neurons.50

In conclusion, we have shown that binocular VEPsummation is not correlated with the presence or levelof stereopsis in infants, stereonormal adults, and ste-reodeficient adults. We have also shown that binocularVEP summation is considerably greater in younger in-fants than in older infants or adults. Finally, we haveshown that some, but not all, stereodeficient adultsshow significant binocular VEP summation, despitethe absence of stereopsis. One could account for thesefindings by proposing that binocular cortical neuronsare required for binocular VEP summation, and thatyoung infants and some strabismic adults who showbinocular VEP summation do not show evidence ofstereopsis because of factors other than the absence ofbinocular cortical neurons. However, the much greateramount of binocular VEP summation in young infantssuggests that two independent pools of monocularlydriven neurons are sufficient for significant levels ofbinocular VEP summation. We have hypothesized thatthe amplitude of the binocular VEP signal saturates atprogressively lower levels in later infancy and adult-hood as interocular cortical inhibition increases.

Key words: binocular summation, VEP, infants

364 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / February 1987 Vol. 28

Acknowledgments

The authors are indebted to David Link for hardware sup-port and Mary E. Buuck for assistance in running some ofthe subjects, and R. Blake, V. Dobson, R. Fox, W. Makous,C. Nelson, and S. Sokol for their detailed and helpful com-ments.

References

1. Blake R and Fox R: The psychophysical inquiry into binocularsummation. Percept Psychophys 26:161, 1973.

2. Blake R, Sloane M, and Fox R: Further developments in bin-ocular summation. Percept Psychophys 30:266, 1981.

3. Lema SA and Blake R: Binocular summation in normal andstereoblind humans. Vision Res 17:69, 1977.

4. Levi D, Harwerth R, and Manny RE: Suprathreshold spatialfrequency detection and binocular interaction in strabismic andanisometropic amblyopia. Invest Ophthalmol Vis Sci 18:716,1979.

5. Wade NJ: On interocular transfer of the movement aftereffectin individuals with and without normal binocular vision. Per-ception 5:113, 1976.

6. Westendorf D, Langston A, Chambers D, and Allegretti D: Bin-ocular detection by normal and stereoblind observers. PerceptPsychophys 24:209, 1978.

7. Odom J V and Harter MR: Interocular suppression in adults andinfants using anaglyphic stimuli: Visually evoked potential mea-sures. Electroencephalogr Clin Neurophysiol 56:232, 1983.

8. Schor CM: Visual stimuli for strabismic suppression. Perception6:583, 1977.

9. Campos E and Chiesi C: Binocularity in comitant strabismus:II. Objective evaluation with visual evoked responses. DocOphthalmol 55:277, 1983.

10. Fox R, Aslin RN, Shea SL, and Dumais ST: Stereopsis in humaninfants. Science 207:323, 1980.

11. Held R, Birch EE, and Gwiazda J: Stereoacuity of human infants.Proc Natl Acad Sci USA 77:5572, 1980.

12. Petrig B, Julesz B, Kropfl W, Baumgartner G, and Anliker M:Development of stereopsis and cortical binocularity in humaninfants: Electrophysiological evidence. Science 213:1402, 1981.

13. Birch EE, Gwiazda J, and Held R: Stereoacuity development forcrossed and uncrossed disparities in human infants. Vision Res22:507, 1982.

14. Birch EE, Gwiazda J, and Held R: The development of vergencedoes not account for the onset of stereopsis. Perception 12:331,1983.

15. Birch EE and Held R: The development of binocular summationin human infants. Invest Ophthalmol Vis Sci 24:1103, 1983.

16. Birch EE, Shimojo S, and Held R: Preferential-looking assessmentof fusion and stereopsis in infants aged 1-6 months. InvestOphthalmol Vis Sci 26:366, 1985.

17. Shea SL, Doussard-Roosevelt JA, and Aslin RN: Pupillary mea-sures of binocular luminance summation in infants and stereo-blind adults. Invest Ophthalmol Vis Sci 26:1064, 1985.

18. Romano PE, Romano JA, and Puklin JE: The development ofnormal binocular single vision in childhood. In Orthoptics: Past,Present, Future, Moore S, Mein J, and Stockbridge L, editors.New York, Intercontinental Book Co., 1975, pp. 11-18.

19. Reinecke RD and Simons K: A new stereoscopic test for am-blyopia screening. Am J Ophthalmol 78:714, 1974.

20. Cooper J and Feldman J: Operant conditioning and assessmentof stereopsis in young children. Am J Optom Physiol Opt 55:532, 1978.

21. Walraven J: Amblyopia screening with random-dot stereograms.Am J Ophthalmol 80:893, 1975.

22. von Noorden GK: Burian-von Noorden's Binocular Vision andOcular Motility. St. Louis, CV Mosby, 1980, pp. 270-274.

23. Amigo G, Fiorentini A, Pirchio M, and Spinelli D: Binocularvision tested with visual evoked potentials in children and infants.Invest Ophthalmol Vis Sci 17(9):910, 1978.

24. Apkarian PA, Nakayama K, and Tyler CW: Binocularity in thehuman visual evoked potentials: facilitation, summation andsuppression. Electroencephalogr Clin Neurophysiol 51:32, 1981.

25. Apkarian PA, Levi D, and Tyler CW: Binocular facilitation inthe visual-evoked potential of strabismic amblyopes. Am J OptomPhysiol Opt 58:820, 1981.

26. Arden GB and Barnard WM: Effect of occlusion on the visualevoked response in amblyopia. Trans Ophthalmol Soc UK 99:419, 1979.

27. Levi DM: Do visual evoked potentials reveal amblyopic abnor-malities not readily apparent in psychophysical tests? Ann NYAcad Sci 388:615, 1982.

28. Nuzzi G and Franchi A: Binocular interaction in visual-evokedresponses: Summation, facilitation and inhibition in a clinicalstudy of binocular vision. Ophthalmic Res 15:261, 1983.

29. Oguchi Y, Katsumi O, and Kawara T: Binocular VECP withand without fusion. Doc Ophthalmol Proc Ser 31:415, 1982.

30. Srebro R: The visually evoked response. Binocular facilitationand failure when binocular vision is disturbed. Arch Ophthalmol96:839, 1978.

31. Wanger P and Nilsson B: Visual evoked responses to pattern-reversal stimulation in patients with amblyopia and/or defectivebinocular functions. Acta Ophthalmol 56:617, 1978.

32. Perry NW, Childers DG, and McCoy JG: Binocular addition ofthe visual evoked response at different cortical locations. VisionRes 8:567, 1968.

33. Ciganek L: Binocular addition of the visually evoked responsewith different stimulus intensities in man. Vision Res 10:479,1970.

34. Barnet A, Friedman S, Weiss I, Ohlrich E, Shanks B, and LodgeA: VEP development in infancy and early childhood: a longi-tudinal study. Electroencephalogr Clin Neurophysiol 49:476,1980.

35. Blom JL, Barth PG, and Visser SL: The visual evoked potentialin the first six years of life. Electroencephalogr Clin Neurophysiol48:395, 1980.

36. Harter MR, Deaton FK, and Odom JV: Pattern visual evokedpotentials in infants. In Visual Evoked Potentials In Man: NewDevelopments, Desmedt JE, editor. Oxford, Clarendon Press,1977, pp. 332-352.

37. Marg E, Freeman DN, Peltzman P, and Goldstein PJ: Visualacuity development in human infants: evoked potential meas-urements. Invest Ophthalmol 15:150, 1976.

38. Moskowitz A and Sokol S: Development of changes in the humanvisual system as reflected by the latency of the pattern reversalVEP. Electroencephalogr Clin Neurophysiol 56:1, 1983.

39. Pirchio M, Spinelli D, Fiorentini A, and Maffei L: Infant contrastsensitivity evaluated by evoked potentials. Brain Res 141:179,1978.

40. Sokol S and Jones K: Implicit time of pattern evoked potentialsin infants: An index of maturation of spatial vision. Vision Res19:747, 1979.

41. Spekreijse H: Maturation of contrast EP's and development ofvisual resolution. Arch Ital Biol 116:358, 1978.

42. Moskowitz A and Sokol S: Spatial and temporal interaction ofpattern-evoked cortical potentials in human infants. Vision Res20:699, 1980.

No. 2 BINOCULAR SUMMATION OF VEPs / Shea er al. 365

43. Sokol S, Hansen VC, Moskowitz A, Greenfield P, and TowleVL: Evoked potential and preferential looking estimates of visualacuity in pediatric patients. Ophthalmology 90:552, 1983.

44. MacKay DM and Jeffreys DA: Visually evoked potentials andvisual perception in man. In Handbook of Sensory Physiology,Vol. VII(3), Visual Centers in the Brain. Jung R, editor. NewYork, Springer-Verlag, 1973, pp. 647-678.

45. Dobson V and Teller DY: Visual acuity in human infants: Areview and comparison of behavioral and electrophysiologicalstudies. Vision Res 18:1469, 1978.

46. Teller DY: Measurement of visual acuity in human and monkeyinfants: the interface between laboratory and clinic. Behav BrainRes 10:15, 1983.

47. Salapatek P and Nelson CA: Event-related potentials and visualdevelopment. In Measurement of Audition and Vision in theFirst Year of Postnatal Life: A Methodological Overview. GottliebG and Krasnegor N, editors. New Jersey, Ablex, 1985, pp. 419-453.

48. Lovasik JV and Bishop D: Short-term effects of induced anisei-konia on the binocular pattern VER. Am J Optom Physiol Opt61:513, 1984.

49. Fiorentini A, Maffei L, Pirchio M, and Spinelli D: An electro-physiological correlate of perceptual suppression in anisometro-pia. Vision Res 18:1617, 1978.

50. Sclar G, Ohzawa I, and Freeman RD: Contrast gain control inthe kitten's visual system. J Neurophysiol 54:668, 1985.