Download Preferential-looking assessment of fusion and stereopsis in

Preferential-Looking Assessment of Fusion and
Stereopsis in Infants Aged 1-6 Months
Eileen E. Birch,* Shinsuke Shimojo,t and Richard Heldf
The ability of infants to discriminate zero-disparity stimuli from both reverse contrast (rivalrous) and
disparate (stereoscopic) stimuli was investigated in a two-alternative, forced-choice, preferentiallooking paradigm. Few infants under 4 months of age demonstrated discrimination for any stimulus
pairing. Of the infants tested at 4 months of age, approximately 70% preferred zero-disparity stimuli
to reverse contrast stimuli, and 82% preferred stereoscopic stimuli to zero-disparity stimuli. Nearly
100% of 5- and 6-month-old infants exhibited these preferences. These findings suggest that sensory
fusion is not present at birth but develops rapidly over the first 6 months of life. The time course for
the development of sensory fusion was similar to the time course for the development of stereopsis
in nine infants tested longitudinally. Invest Ophthalmol Vis Sci 26:366-370, 1985
Previous studies using classical1 5 and random-dot
stereograms5"7 have shown that the onset of a visual
preference for stereoscopic stimuli is abrupt in a
given infant and occurs on the average at 4 months
of age. Infants over 4 months of age show consistent
preference for line stereograms containing binocular
disparities over zero disparity, with percent preference
rising from chance to approximately 100% over a
range of disparities from 0 to 60 min. For even larger
disparities, perceived as rivalrous or lustrous by adults,
percent preference decreases to chance or reverses to
a preference for the zero-disparity stimulus. A similar
pattern of responses was noted in a study employing
random-dot stereograms of 45- and 313-min or 447min disparity.6 One interpretation of this reversal is
that the disparate elements of the stereogram are
perceived as rivalrous and that infants over 4 months
of age prefer a fusion stimulus over one that has
rivalrous elements. However, the 75-min disparity
grating must be considered a weak stimulus for
eliciting a preferential response since less than half of
the infants had a preference. This may be due to the
fact that only alternate bars were binocularly disparate.
The purpose of the present study was to determine
at what ages fusion stimuli are preferred over rivalrous
stimuli. In addition, the onset age of preference for
fusion and the onset age for preference for stereoscopic
depth were compared. For these purposes, two separate but similar stimulus configurations were developed independently by two investigators and presented
to separate populations of infants. The stimulus configurations were designed to enhance the dichotomy
between rivalry and fusion, based on the findings in
adults that (1) reverse-contrast classical stereo halfimages are fused poorly 89 and (2) in the case of
random-dot stereograms, they cannot be fused at
all. 1 0 " Because of the similarities in results and
conclusions, we report the two studies together, despite
differences in stimuli and procedures.
Materials and Methods
General
From the Vision Research Center, Retina Foundation of the
Southwest, and the Department of Ophthalmology, University of
Texas Health Science Center, Dallas, Texas*; and the Department
of Psychology, Massachusetts Institute of Technology, Cambridge,
Massachusetts.!
Presented in part at the meeting of the Association for Research
in Vision and Ophthalmology, Sarasota, Florida, May 1983.
Supported in part by NIH grants EY01191 and EY02621, by a
postdoctoral fellowship to E. Birch awarded by Fight for Sight, Inc.
(New York City) in Tribute to the Memory of Hermann M.
Burian, MD, and by a grant from the Anne Burnett and Charles
Tandy Foundation.
Submitted for publication: August 25, 1983.
Reprint requests: E. Birch, PhD, Vision Research Center, Retina
Foundation of the Southwest, 8220 Walnut Hill Lane #012, Dallas,
TX 75231.
Healthy infants aged 1-6 post-term months were
solicited through letters to parents listed in birth
records. All infants were examined by a pediatric
optometrist, and only those who were orthophoric
and within the normal range of refractive error 1213
were recruited for participation. Infants were assigned
to one of six 1-month age groups. The number of
infants participating in each age group and experimental paradigm is shown in Table 1. Informed
consent was obtained from all parents after the procedure had been explained fully. Six adults aged 2050 years also were tested.
,366
Downloaded From: http://iovs.arvojournals.org/ on 05/03/2017
367
INFANT FUSION AND STEREOPSIS / Birch er ol.
No. 3
Table 1. Age distribution of participating infants
Control: both halffields to one eye
Post-term
age (months)
Stereo pair
and checkerboard
1
2
3
4
5
6
4
6
15
17
9
4
Total
55
Random dot
Control: monocular
occlusion
Stereo
Check
Stereo
Check
1
3
3
7
7
6
0
0
4
4
5
4
0
0
4
4
3
4
0
0
4
5
3
4
0
0
4
5
2
4
0
0
2
5
6
6
27
17
15
16
15
19
Paired stimuli were rear-projected (Hawk MK.V
or Hawk MK..VI Stereoprojector) onto two 38 deg
X 30 deg Polacoat Lenscreens, which were located to
the left and right of a 4-deg-wide opaque panel
containing two small light-emitting diodes. The two
channels of the stereoprojector were polarized orthogonally, and infants wore goggles containing polarizing filters while viewing the images. The extinction
ratio for the filters was 10~4. Since all stimuli were
comprised of two polarized half-fields whether or not
binocular disparity was present, head rotation could
not provide a differential intensity cue by which the
disparate stimulus could be discriminated from the
nondisparate stimulus. The three stimulus pairs employed in the study are shown schematically in Figure
1. Each stimulus pair consisted of a zero-disparity
pattern (identical half-fields for each eye) and a
binocularly disparate pattern or reverse contrast pat-
Screen I
Binocular Disparity
Left eye
Right eye
A. Line Stereogram
(45 min disparity)
Control: one halffield to both eyes
Random dot
tern. Each infant was held by its parent at 50 cm
from the screens. The parent was instructed to maintain the infant's head in an upright position and an
observer, watching through a peephole from behind
the screens, monitored head position throughout the
experiment. The room was dark so that the only light
came through the projection screens. Thus, the only
stimulus cues present for accommodation and convergence were located at the plane of the screens. On
each trial, an observer centered the infant's gaze by
flashing the light-emitting diodes. The stimulus pair
then was presented, and the observer, without knowledge of the positions of the stimuli, made a forcedchoice judgment of the side at which the infant
preferred to look. The stimulus positions were randomized over trials differently for each subject. The
disparate stimulus appeared on the same side no
more than three times in a row. No feedback was
Perception of
Normal Adults
Screen 2
Zero Disparity
Left eye
Right eye
Depth vs. Fusion
B. Checkerboard
Rivalry vs. Fusion
C. Random Dot
Rivalry vs. Fusion
Fig. 1. Schematic diagram of the stimulus pairings used in assessing stereopsis (A) and fusion (B, C). The stimuli are not drawn to scale.
Downloaded From: http://iovs.arvojournals.org/ on 05/03/2017
368
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / March 1985
given to the observer about stimulus location, since
the judgment was one of preference and not of
stimulus location.
Paradigm 1
This study employed bar sterogram and checkerboard stimulus patterns. The bar stereogram was
identical to the ones that were previously employed
to assess the development of stereopsis.2 Briefly, these
stereograms were composed of seven 2.0-deg-wide
vertical black bars spaced 2.0 deg apart (0.92 contrast).
To create 45-min crossed binocular horizontal disparity, alternate bars on one half-field were shifted
laterally. When normal adults viewed this stimulus
through polarizing filters, all seven bars on one side
appeared to lie in the plane of the screen while the
second, fourth, and sixth bars on the other side
appeared to stand out in front of the screen. The
checkerboard stimulus pair (check size = 2.0 deg;
0.92 contrast) consisted of checks that were of the
same contrast for the two eyes on one screen, paired
with checks that were reversed in contrast for the two
eyes on the second screen. When viewing through
polarizing goggles, normal adults reported black and
white checks on one screen and checks that appeared
alternately black and white on the second screen
(rivalry).
The classical stereogram pair contained two monocular cues that possibly could have mediated discrimination and served as the basis for any preference
shown. The first monocular cue (due to incomplete
polarization) was the presence of faint gray stripes,
which appeared on either side of the black bars only
in the disparate pattern. This difference could be
enhanced by tilting the goggles and/or head. To
determine whether this monocular cue alone was
sufficient for the infants to show a preference, the
polarizing filters in the projector were oriented parallel
to one another. This had the effect of occluding one
eye while the other eye viewed both half-fields presented on both screens. A second monocular cue on
which discrimination may have been based was the
difference in spacing of the bars in the two patterns.
In the disparate pattern, alternate bars were shifted
laterally while, in the zero-disparity pattern, the bars
were spaced regularly. In order to examine whether
the irregular spacing itself was attractive to the infant
in the absence of binocular horizontal disparity, both
polarizers in the goggles were oriented in parallel,
while the filters in the projector remained orthogonal.
In this case, both eyes viewed only one-half of the
stereo pair present on each screen. Two control
conditions for the checkerboard stimuli were generated
in the same way.
Downloaded From: http://iovs.arvojournals.org/ on 05/03/2017
Vol. 26
In this experimental paradigm, each of 55 infants
participated in a single test session of 20 trials with
the stimulus pair designed to assess stereopsis (45min vs 0-min crossed disparity) and 20 trials with
the checkerboard stimulus pair. Twenty-six infants
viewed bar stereograms first, and 29 viewed checkerboards first, with the order of the stimulus pairs
unknown to the observer. In addition, 45 of these
infants aged 3-6 months received 10 trials in each of
one or more of the control conditions during the
same test session. The order of experimental and
control conditions was counterbalanced. Four other
infants failed to complete 40 trials, and their data
have been excluded from analyses. Nine additional
infants were tested bi-weekly beginning at age 1-11
post-term weeks (mean age at first test session = 7.9
weeks, SD = 3.0 weeks). These infants were tested
until they reached criterion on both stimulus pairs.
Criterion performance was 75% preference in the
experimental conditions. Since the hypothesis tested
predicted which one of the stimulus pair would be
preferred (one-tailed test), the binomial probability
associated with criterion performance was 0.021. Criterion performance in the control conditions was
80% preference (one-tailed binomial probability
= 0.055).
Paradigm 2
The second study employed a stimulus pair consisting of random-dot patterns (dot size = 30 min;
density = 50%; contrast = 0.92). The patterns presented to each eye were identical on one screen. On
the other screen, random-dot patterns were opposite
in contrast for each eye. When viewed through polarizing filters by normal adults, one screen appeared
as a fused black-and-white pattern, while the other
side appeared rivalrous or lustrous. Twenty-six infants
participated in a single session in which they received
10 trials with the random-dot stimulus pair. Nineteen
of these infants aged 3-6 months also were tested
monocularly on 10 trials during the same test session.
This control condition was used to determine whether
contrast differences between the two stimuli might
mediate preference. The order of experimental and
control conditions was counterbalanced.
Criterion performance for both the experimental
and control conditions was 80% preference. Since the
hypothesis predicted which one of the stimulus pair
would be preferred (one-tailed test), the binomial
probability associated with criterion performance was
0.055.
Results
The proportion of infants attaining a criterion
preference for binocularly identical patterns (Fig. 2A)
369
INFANT FUSION AND STEREOPSIS / Dirch er ol.
No. 3
and for binocular disparate patterns (Fig. 2B) varied
as a function of age. With both the checkerboard and
random-dot patterns, over 60% of the 4-month-old
infants preferred zero disparity. All 10 of the 6month-old infants preferred the fusional stimulus.
None of the infants tested preferred reverse contrast
stimuli at any age. In general, infants who reached
criterion with the checkerboard stimulus pair also
reached criterion with the line stereogram pair, while
infants who failed to reach criterion with the checkerboard stimulus pair also tended to fail to reach
Table 2. Age of onset for fusion and stereopsis
Infant
1
2
3
4
5
6
7
8
9
Mean
SD
First test
(week)
8
8
1
11
5
8
9
11
10
7.9
3.0
Fusion onset
(week)
Stereo onset
(week)
Difference
(weeks)
10
10
7
13
9
10
13
13
20
10
10
13
13
9
10
15
13
18
0
0
6
0
0
0
2
0
-2
11.7
3.5
12.3
2.7
0.7
2.1
1.0
A. Fusion
o
I 0.8
D Checkerboard
•
Random dot
I"§ 0.6
a>
</>
|
0.4
•s
g
I
a.
a.g
0.2
0.0
2
3
4
Post-term age (months)
Fig. 2. A. The proportion of infants who reached criterion
preference for matched contrast stimuli over reverse contrast stimuli
as a function of post-term age.
1.0
c
o
B. Stereopsis
I 0.8
o
c
o 0,6
o
k_
o 0.4
I 0.2
criterion with the line stereogram pair (contingency
coefficient C = 0.41, P < 0.01; see reference 14). For
the line stereogram stimulus pair, none of the infants
preferred the zero-disparity stimulus at any age. The
proportion of infants preferring 45-min crossed disparity was larger in the older age groups. Over -80%
of the infants aged 4-6 months preferred the binocularly disparate stimulus. All six adults were capable
of making discriminations for each of the three
stimulus pairs on 100% of 10 trials.
Data from the nine infants who were tested biweekly with both the line stereogram pair and the
checkerboard pair are summarized in Table 2. The
age at which each infant first reached criterion with
each stimulus pair was defined as the onset age for
stereopsis (line stereograms) and fusion (checkerboards). Six of the nine infants first reached criterion
for fusion and stereopsis during the same test session.
Two infants reached criterion with the checkerboard
stimulus pair first, and one reached criterion with the
line stereogram first. The mean ages of onset for
stereopsis (12.3 weeks) and for fusion (11.7 weeks)
were not significantly different (t-test for paired comparisons, t = 0.7, P > 0.05).
The performance of 3-6-month-old infants in each
of the control conditions is compared with their
performance on the same day in the experimental
conditions in Figure 3. Over 70% of 3-6-month-old
infants reached criterion in each of the three experimental conditions (stereogram, checkerboard, and
random-dot patterns). In contrast, less than 11% of
the infants reached criterion in each of the three
control conditions.
o
Discussion
1 2
3
4
Post-term age (months)
5
6
Fig. 2. B. The proportion of infants who reached criterion
preference for 45-min binocular disparity over 0-min binocular
disparity as a function of post-term age.
Downloaded From: http://iovs.arvojournals.org/ on 05/03/2017
The results with line stereograms in this study are
consistent with previous behavioral and electrophysiologic studies,1"7 which have shown that most infants
aged 4-6 post-term months can discriminate binocularly disparate stereoscopic stimuli from zero-dis-
370
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / March 1985
I.Or
ocular vision and hyperacuity also is supported by
studies of convergent eye movements,16 binocular
summation,17 evoked-potential responses to binocular
correlation, 718 amblyopia,19 and vernier acuity.20
0.8
ii
Key words: fusion, stereopsis, preferential looking, infants,
development
g o>
I*
0.4
References
0.2
0.0L
Stimulus:
Vol. 26
[=•
A B C
One halffield to
each eye
A B
Both halffields to
one eye
A B
One halffield to
both eyes
n
C
Monocular
occlusion
Fig. 3. The proportion of infants aged 3-6 months who reached
criterion in the experimental (one half-field to each eye) and three
control conditions. Stimuli are those shown in Figure 1: A, line
sterograms; B, checkerboards; C, random-dot patterns.
parity stimuli. A few infants aged 1-3 months also
reached criterion in the stereopsis paradigm. This
result is in accord with previous reports that 10-30%
of infants aged 2 months are capable of this discrimination. 1 - 5715
The novel finding in the present study is that
infants over 4 months of age tested with checkerboard
and random-dot patterns discriminate between binocularly identical and reverse contrast stimuli. However, in this case, the preference is reversed. Infants
prefer a binocularly disparate steroscopic stimulus to
a zero-disparity stimulus but prefer the zero-disparity
stimuli to reverse-contrast stimuli. The longitudinal
data presented here show an agreement between the
average ages at which discrimination of these two
distinct classes of binocular stimuli first occurs (within
2 weeks). The qualitatively different responses that
infants produce for stimuli which adults perceive in
depth (preferred) and for stimuli which adults perceive
as rivalrous (nonpreferred), suggest that the 4-6month-old infant has approximately the same upper
limit in fusional range as adults.
The time course for the development of sensory
fusion described in the present study is similar to the
time course for the development of stereopsis previously reported in both behavioral and electrophysiologic studies.1"7 That the fourth month of life represents an important age for the development of bin-
Downloaded From: http://iovs.arvojournals.org/ on 05/03/2017
1. Birch EE, Gwiazda J, and Held R: Stereoacuity development
for crossed and uncrossed disparities in human infants. Vision
Res 22:507, 1982.
2. Birch EE, Gwiazda J, and Held R: The development of
vergence does not account for the onset of stereopsis. Perception
12:331, 1983.
3. Held R, Birch EE, and Gwiazda J: Stereoacuity of human
infants. Proc Natl Acad Sci USA 77:5572, 1980.
4. Bechtoldt HP and Hutz CS: Stereopsis in young infants. J Ped
Ophthalmol Strab 16:49, 1979.
5. Teller DY: Scotopic vision, color vision, and stereopsis in
infants. Curr Eye Res 2:199, 1983.
6. Fox R, Aslin RN, Shea SL, and Dumais ST: Stereopsis in
human infants. Science 207:323, 1980.
7. Petrig B, Julesz B, Kropfl W, Baumgartner G, and Anliker J:
Development of stereopsis and cortical binocularity in human
infants: Electrophysiological evidence. Science 213:1402, 1981.
8. Julesz B: Stereopsis and binocular rivalry from dichoptic
stereograms. J Opt Soc Am 53:994, 1963.
9. Julesz B: Foundations of Cyclopean Perception. Chicago, University Press, 1971, pp. 157-159.
10. Levy MM and Lawson RB: Stereopsis and binocular rivalry
from dichoptic stereograms. Vision Res 28:239, 1978.
11 Treisman A: Binocular rivalry and stereoscopic depth perception.
Vision Res 7:89, 1962.
12 Mohindra I, Held R, Gwiazda J, and Brill S: Astigmatism in
infants. Science 202:329, 1978.
13. Mohindra I: A non-cycloplegic refraction technique for infants
and young children. J Am Optom Assoc 48:518, 1977.
14. Siegel S: Non-parametric Statistics for the Behavioral Sciences.
New York, McGraw-Hill Book Company Inc., 1956, pp. 196202.
15. Gwiazda JE, Birch EE, and Held R: Le developpement de la
vision chez I'enfant. La Recherche 128:1348, 1981.
16. Aslin RN: Development of binocular fixation in human infants.
J Exp Child Psychol 23:133, 1977.
17. Birch EE and Held R: The development of binocular summation
in human infants. Invest Ophthaimol Vis Sci 24:1103, 1983.
18. Braddick O, Atkinson J, Julesz B, Kropfl W, Bodis-Wollner I,
and Raab E: Cortical binocularity in infants. Nature 288:363,
1980.
19. Jacobson SG, Mohindra I, and Held R: Age of onset of
amblyopia for infants with esotropia. Doc Ophthalmol 30:210,
1981.
20. Shimojo S, Birch EE, Gwiazda J, and Held R: Development
of vernier acuity in infants. Vision Res 24:721, 1984.