The Development of Binocular Summation
in Human Infants
Eileen E. Birch* and Richard Heldf
The difference in pupil diameter between monocular and binocular viewing conditions was used as
a measure of binocular luminance summation in 97 infants aged 0 to 12 months. No significant
difference in pupil diameter under monocular versus binocular viewing conditions was found until the
end of the fourth month of life. The age of onset was positively correlated with the age of onset for
stereopsis as assessed by preferential looking. By the end of the 6th month, the difference in pupil
diameter under the two viewing conditions was adult-like in magnitude. The procedure provides an
objective measure of binocular function that is not affected by fixation error. The data suggest that
the visual pathways by which signals from the two eyes converge are not functional until the 4th
month of life. Invest Ophthalmol Vis Sci 24:1103-1107, 1983
cessing of relative binocular positional information.
The key finding obtained from adults is that when
equal luminance stimuli are presented to each eye,
binocular brightness exceeds monocular brightness.9"12 A recordable physiologic parallel to binocular luminance summation was first described by
Bartley13 and later described in detail in a series of
papers by ten Doesschate and Alpern.14"17 They
found that, under steady state conditions, the pupil
is smaller when both eyes are illuminated than when
either eye is occluded. Moreover, with unequal illumination of the two eyes, a qualitative and quantitative parallel to the brightness summation data of
Fechner18 and Levelt19 was obtained.
The present report describes results obtained with
a procedure for assessing binocular luminance summation in the pupillary response of infants. Both the
time course of development for binocular summation
and the relation between binocular summation and
stereopsis were determined.
Recent studies of the development of binocular
vision in infancy, using both psychophysic and electrophysiologic techniques, indicate that the onset of
stereopsis occurs during the 3rd to 5th months of
life.1"8 These data suggest that the visual pathways
through which signals from the two eyes converge for
disparity processing are not developed at birth but,
rather, require a period of maturation. In addition to
disparity processing, successful performance on a test
for stereopsis also depends upon the capacity for bifoveal fixation. Special conditions may bypass this
requirement3 but, in general, a failure to demonstrate
stereopsis may reflect immaturity either of disparity
processing or of control of vergence.
Other measures of binocular function exist that do
not confound maturity of vergence with maturity of
the binocular visual pathways. Tests of binocular luminance summation, for example, assess the degree
of convergence of the luminance signals from the two
eyes. Full-field tests of luminance summation are not
affected by fixation error and do not require pro-
Materials and Methods
From the Vision Research Center, Retina Foundation of the
Southwest, Dallas, Texas and Department of Ophthalmology, University of Texas Health Science Center, Dallas, Texas,* and the
Department of Psychology, Massachusetts Institute of Technology,
Cambridge, Massachusetts^
Supported by a postdoctoral research fellowship from Fight for
Sight, Inc., New York City, in Tribute to the Memory of Hermann
M. Burian, MD (to Eileen E. Birch). Supported in part by grants
from The National Institutes of Health (No. 5P30-EY02621 and
No. 2R01-EY01191) and from the Spencer Foundation (LTRDTD-71373).
Submitted for publication: November 29, 1982.
Reprint requests: Eileen E. Birch, PhD, Vision Research Center,
Retina Foundation of the Southwest, 8220 Walnut Hill Lane, Suite
012, Dallas, TX 75231.
Subjects
Ninety-seven infants, aged 0-12 post-term months,
were recruited by letter through birth records available at the City Hall in Cambridge, Massachusetts.
All infants who participated were healthy, orthophoric, and within the normal range of refractive error as assessed by near-retinoscopy.20 T-wenty-three
infants were tested monthly for binocular summation
with the first visit at 0-3 months of age and the final
visit at 6 months of age. Twelve infants were tested
twice per month for both binocular summation and
stereopsis during the 3rd-6th months of life. Sixty-
0146-0404/83/0800/1103/$ 1.05 © Association for Research in Vision and Ophthalmology
1103
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933113/ on 06/18/2017
1104
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Augusr 1983
two infants, aged 0-12 months, participated in a
cross-sectional sample. Ten normal adults, aged 2050 years, were also tested.
Binocular Summation
All testing was conducted in a dark room. The
infant was held over its parent's shoulder with its head
at the entrance of a full-field, uniformly illuminated
(1.92 cd/m2) white dome 35.5 cm diameter. The
dome contained an 8° aperture through which a low
light level video camera monitored the infant's right
eye. The video camera was employed in conjunction
with a telephoto lens (approx. 5X magnification), a
videotape recorder, and a black and white display
monitor.
The infant's left eye was occluded with an opaque
black patch and the right eye was adapted to the background illumination of the sphere for 5 min. Dim red
light-emitting diodes located next to the camera lens
flashed alternately to help maintain the infant's attention. Since the infant was relatively free to move
about during the taping, a minimum of 2 min of
videotape of the infant's right eye was obtained with
the left eye occluded (monocular viewing condition).
Pilot work showed this to be the minimum duration
within which at least ten frames that were clearly in
focus and nearly straight ahead in viewing angle could
be obtained. Actual recording time sometimes exceeded this and depended on the cooperation of the
infant. The left eye was then unpatched and a break
in videotaping of 60 sec was given since pilot work
showed that up to 20-30 sec readaptation was necessary for pupil size to reach equilibrium after the
patch was removed. At least 2 min of videotape was
then obtained of the right eye with the left eye open
(binocular viewing condition).
Pupil and iris diameters were measured from single
frames displayed on the black and white monitor.
Pupil diameter was converted to a percentage of iris
diameter in order to eliminate variations associated
with changes in head position during taping. Ten such
measures of right eye pupil size were obtained for
both the monocular and binocular viewing conditions. The measures were taken from the first 10
frames of the right eye that were clearly in focus,
nearly straight ahead in viewing angle, and separated
by at least 5 sec.
Stereopsis
Stimuli and apparatus previously described in detail2 were employed. Briefly, stereograms were rearprojected onto two Polacoat Lenscreens (3M Company) that were 11.5° in diameter with 34° center-
Vol. 24
to-center separation. These screens were located to
the left and right of two small light emitting diodes.
Two half fields of a 45 min crossed disparity stereogram were superimposed on one screen; two identical half-fields of a zero-disparity stereogram were superimposed on the second screen. All patterns were
composed of three 1.9° wide black bars spaced at
1.9°, except that in the stereogram the two outside
bars were shifted laterally in one half field to create
the disparity. The views of the two eyes were separated by means of linear polarizers (extinction ratio
= 10~4) mounted in the stereoprojector (Hawk MK
VI) and in lightweight goggles (Speedo) worn by the
infant. The patterns had a mean luminance of 7.5 cd/
m2 and 92% contrast.
The infant was held on its parent's lap or over its
parent's shoulder at a distance of 50 cm from the
screens. The parent was asked to maintain the infant's
head in an upright position and an observer, watching
through a peephole between the screens, was able to
monitor head position throughout the experiment.
The room was dark so that the only light was provided
by the screens. On each trial, the observer centered
the infant's gaze byflashingthe light-emitting diodes.
The stimuli were then presented and the observer,
without knowledge of the position of the disparate
stimulus, made a forced choice judgment as to the
side at which the infant preferred to look. The position of the disparate stimulus was randomized over
trials. Within a session, lasting approximately 10 min,
20 trials were obtained. Presence of stereopsis was
defined as 75% preference (p < 0.02) for disparate
stimulus.
During some of the sessions in which the infants
demonstrated at least 75% preference for 45 min
crossed disparity over zero disparity, control trials
were also included. Nine infants were each tested
once with 45 min vertically disparate horizontal
stripes paired with zero disparity horizontal stripes
as a mean age of 17.9 weeks (SD = 4.2 weeks). These
stimuli contain all of the same monocular cues and
nonstereoscopic binocular cues as the stimuli used to
assess stereopsis. However, unlike horizontal binocular disparity, vertical binocular disparity is not a cue
for stereopsis. The mean % preference for 45 min
horizontal disparity over zero disparity in these sessions was 78.3% (SD = 4.1%). In contrast, there was
no significant preference for 45 min vertical disparity
over zero disparity (mean = 49.4%; SD = 11.2%).
Eleven infants (mean age =18.1 weeks; SD = 4.7
weeks) were each tested once with 80 min crossed
disparity vs zero disparity. This horizontal binocular
disparity is so large as to fall outside the range for
stereopsis. Normal adults report rivalry in response
to this stimulus. The mean % preference for 45 min
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933113/ on 06/18/2017
1105
BINOCULAR SUMMATION ./• Dirch and Held
No. 8
disparity over zero disparity in these sessions was
79.5% (SD = 5.4%) while the mean percent preference
for 80 min disparity over zero disparity was 40.0%
(SD= 18.7%).
In both of these control conditions, stimuli contained the same monocular and binocular cues as that
in the test for stereopsis except for a binocular horizontal disparity cue within the range required for
stereopsis. Without the disparity cue for stereopsis,
the infants failed to demonstrate a significant visual
preference. Thus, we conclude that the visual preference observed in the test for stereopsis is based on
horizontal binocular disparity.
.n=4
20
38
47
26
28
10
10
2
3
4
5
6
Post-term Age (Months)
7-12 Adult
2
3
4
5
6
Post-term Age (Months)
7-12 Adult
Results
Cross-sectional
Mean right eye pupil diameter as a percentage of
iris diameter in the monocular viewing condition
(filled triangles) and in the binocular viewing condition (open triangles) obtained with 62 infants aged
0-12 months is shown in Figure 1. Pupil diameter
on the monocular viewing condition increases with
age over months 1-6 (Pearson r = 0.94, p < 0.01),
but pupil diameter in the binocular viewing condition
is not correlated with age over this same time period
(Pearson r = 0.16, ns). A difference score was calculated for each infant and the mean difference in
pupil diameter under the two viewing conditions is
also shown in Figure 1. T-tests for paired comparisons
confirm that there is no significant difference in right
eye pupil diameter for monocular vs binocular viewing during months 1-4. However, right eye pupil diameter is significantly larger when the left eye is occluded during months 5-12 (p < 0.02) and in adults
(p < 0.05).
Fig. 1. Mean right eye pupil diameter as a percentage of iris
diameter with the left eye occluded (A) and with the left eye open
(A). The number of infants tested in each age group is indicated
along the top of thefigure.Mean difference in pupil diameter under
the two viewing conditions (•) is also shown. Vertical bars represent one standard error of the mean.
50r A
I
Left eye occluded 1
Left eye open
/ ^ X
-
b
V
I-
?>-
A
g
Longitudinal
** Longitudinal data from a representative infant are
shown in Figure 2. Mean right eye pupil diameter as
a percentage of iris diameter with the left eye occluded
(filled triangles) and with the left eye open (open triangles) is shown in Figure 2A. The infant's data are
typical in that pupil diameter in the binocular viewing
condition does not change significantly with age while
pupil diameter in the monocular viewing condition
increases with age. Figure 2B shows the differences
computed from the data in Figure 2A by subtracting
mean pupil diameter for binocular viewing from
mean pupil diameter for monocular viewing at each
age. It is not until the 14th week (4th month) that
this infant shows a significant difference between the
two test conditions.
Differences in pupil diameter under the two view-
k
>/
i
i
i
i
i
5
10
15
20
Post-term Age (Weeks)
25
5
10
15
20
Post-term Age (Weeks)
25
Fig. 2. Mean right eye pupil diameter of a representative infant
at six post-term ages with the left eye occluded (A) and with the
left eye open (A). Vertical bars represent one standard error. The
difference in mean pupil diameter under the two viewing conditions (•) is also shown.
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933113/ on 06/18/2017
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Augusr 1983
1106
Vol. 24
Fig. 3. Difference in right
eye pupil diameter when the
left eye is occluded versus
when the left eye is open for
23 infants tested during the
first 6 months of life.
5 0 6 2 0 2 5 0 5
6 2 0 25 0 5 1 0 6 2 0 25 0
Post-term Age (Weeks)
ing conditions for the 23 infants who participated in
the longitudinal study are shown in Figure 3. These
difference values were calculated as described for Figure 2. The mean standard deviation of measurement
for this group of infants is 2.10% of iris diameter, so
that the minimum significant difference between test
conditions in a t-test for paired comparisons is 4.75%
of iris diameter (p < 0.05). Using this criterion, the
mean age at which a significant difference in right eye
25
n = l2
r = 0,775 (p<0.0l)
Mean onset age difference=0,33wks
20
•S 15
i</>
'g.
o
10
0)
a>
CO
tr
6
5
o
5
10
15
20
25
Onset of Binocular Summation (Age in Weeks)
Fig. 4. Scatterplot of the ages of onset for binocular summation
and for stereopsis. The solid line is the best fit as determined by
linear regression.
pupil diameter is found under the two viewing conditions is 15.83 wks (SD = 3.64 wks). For these infants, the mean change in pupil diameter with monocular viewing is +11.05% of iris diameter
(SD = 4.91%) over the first 6 months of life. In
comparison, pupil diameter with binocular viewing
changes little over the same time period (mean change
= +1.00% of iris diameter; SD = 5.8%).
Figure 4 presents data obtained from the 12 infants
who were tested twice per month for both binocular
summation and stereopsis over the 2nd through 6th
months of life. An age of onset for each measure of
binocularity was defined. For binocular summation,
the age of onset was defined as post-term age at the
test session during which the infant first showed a
difference in pupil diameter with monocular versus
binocular viewing of at least 4.75% of iris diameter.
Age of onset for stereopsis was defined as post-term
age at the first session in which the infant demonstrated 75% preference for the disparate stimulus over
the nondisparate stimulus. Each data point in Figure
4 represents a single infant's ages of onset for both
tests. Five of the infants have exactly the same age
of onset with both procedures, three show the pupil
diameter difference earlier, and four show evidence
of stereopsis earlier. Mean age of onset for binocular
summation is 16.1 wks (SD = 4.2 wks) and for stereopsis is 15.5 wks (SD = 3.8 wks). These mean ages
of onset are not significantly different (mean difference = 0.33 wks or 0 sessions). The ages of onset
obtained from the procedures are positively correlated (Pearson r = 0.775; t10 = 3.878; p < 0.01).
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933113/ on 06/18/2017
BINOCULAR SUMMATION / Birch and Held
No. 8
Discussion
The results of this study demonstrate that binocular summation is measurable by the 4th month of
life and is adult-like in magnitude by the 6th month
of life. The onset age for binocular summation is
highly correlated with the onset age for stereopsis.
The correlation between binocular summation and
stereopsis found in this study is consistent with the
absence of binocular summation in the pupillary response of strabismic adults.141617
The agreement between the onset ages for binocular summation in the pupillary response and for stereopsis provides additional evidence that development of bifoveal fixation is not the sole factor determining the onset of stereopsis. Rather, the visual
pathways by which signals from the two eyes converge
must undergo maturation. This hypothesis has already been advanced on the basis of data obtained
both in a study of the onset ages of crossed versus
uncrossed stereopsis2 and in a study using stereo stimuli specifically designed to minimize the effects of
vergence inaccuracy on stereo testing.3 Taken together, these results indicate that binocular neural
development occurs during the first few months of
life.
The objective procedure for assessing binocular
summation described here may be particularly suitable for use in testing strabismic infants. It requires
a minimum of time and cooperation from the infant
and can assess the integrity of binocular visual pathways even in the presence of large angle tropias.
Key words: binocular summation, infants, stereopsis, pupillary response
Acknowledgments
The authors thank Joseph A. Bauer, Jr. for technical
assistance and David Birch for helpful suggestions on earlier
drafts of this manuscript.
References
1. Bechtoldt HP and Hutz CS: Stereopsis in young infants and
stereopsis in an infant with congenital esotropia. J Pediatr
Ophthalmol Strabismus 16:49, 1979.
1107
2. Birch EE, Gwiazda J, and Held R: Stereoacuity development
for crossed and uncrossed disparities in human infants. Vision
Res 22:507, 1982.
3. Birch E, Gwiazda J, and Held R: The development of vergence
does not account for the onset of stereopsis. Perception, in
press.
4. Fox R: Stereopsis in animals and human infants: A review of
behavioral investigations. In Development of Perception; Psychobiological Perspectives. Vol. 2: The Visual System, Aslin
RN, Alberts JR, and Petersen MR, editors. New York, Academic Press, 1981, pp. 335-381.
5. Fox R, Aslin RN, Shea SL, and Dumais ST: Stereopsis in
human infants. Science 207:323, 1980.
6. Held R, Birch E, and Gwiazda J: Stereoacuity of human infants. Proc Natl Acad Sci USA 77:5572, 1980.
7. Petrig B: Nachweis von Stereopsis bei Kindern mittels stochastischer Punktstereogramme und der zugehorigen evozierten Potentiale. Dissertation, Eidgenossischen Technischen
Hochschule, Zurich, 1980.
8. Petrig B, Julesz B, Kropfl W, Baumgartner G, and Anliker M:
Development of stereopsis and cortical binocularity in human
infants: Electrophysiological evidence. Science 213:1402, 1981.
9. Curtis DW and Rule SJ: Binocular processing of brightness
information: A vector-sum model. J Exp Psychol [Hum Percept] 4:132, 1978.
10. Engel GR: The visual process underlying binocular brightness
summation. Vision Res 7:753, 1967.
11. Fry GA and Bartley SH: The brilliance of an object seen binocularly. Am J Ophthalmol 16:687, 1933.
12. Stevens JC: Brightness function: Binocular versus monocular
stimulation. Perception Psychophys 2:451, 1967.
13. Bartley S: Some parallels between pupillary reflexes and brightness discrimination. J Exp Psychol 32:110, 1943.
14. Alpern M: The pupillary light reflex and binocular interaction.
In Early Experience and Visual Information Processing in
Perceptual and Reading Disorders, Young F and Lindsley DB,
editors. Washington, DC, National Academy of Science, 1970,
pp. 119-129.
15. Ten Doesschate J and Alpern M: Response of the pupil to
steady-state retinal illumination: Contribution of cones. Science 149:989, 1965.
16. Ten Doesschate J and Alpern M: Effect of photoexcitation of
the two retinas on pupil size. J Neurophysiol 30:562, 1967.
17. Ten Doesschate J and Alpern M: Influence of asymmetrical
photoexcitation of the two retinas on pupil size. J Neurophysiol 30:577, 1967.
18. Fechner G: Elements of Psychophysics, Translated by H. Adler. New York, Holt, Rinehart, and Winston, 1966.
19. Levelt WJM: Some demonstrations of the complementary
functioning of the eyes. Perception Psychophys 1:39, 1966.
20. Mohindra I: A non-cycloplegic refraction technique for infants
and young children. J Am Optom Assoc 48:518, 1977.
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933113/ on 06/18/2017