Latent Stereopsis for Motion in Depth in

Strabismic Amblyopia
Robert F. Hess,
1
Behzad Mansouri,
1,2
Benjamin Thompson,
1,3
and Elena Gheorghiu
1
PURPOSE. To investigate the residual stereo function of a group
of 15 patients with strabismic amblyopia, by using motion-in-
depth stimuli that allow discrimination of contributions from
local disparity as opposed to those from local velocity mecha-
nisms as a function of the rate of depth change.
METHODS. The stereo performance (percentage correct) was
measured as a function of the rate of depth change for dynamic
random dot stimuli that were either temporally correlated or
uncorrelated.
RESULTS. Residual stereoscopic function was demonstrated for
motion in depth based on local disparity information in 2 of the
15 observers with strabismic amblyopia. The use of a neutral-
density (ND) lter in front of the xing eye enhanced motion-
in-depth performance in four subjects randomly selected from
the group that originally displayed only chance performance.
This nding was true across temporal rate and for correlated
and uncorrelated stimuli, suggesting that it was disparity based.
The opposite occurred in a group of normal subjects. In a
separate experiment, the hypothesis was that the benecial
effect of the ND lter is due to its contrast and/or mean
luminance-reducing effects rather than any interocular time
delay that it may introduce and that it is specic to motion-in-
depth performance, as similar improvements were not found
for static stereopsis.
CONCLUSIONS. A small proportion of observers with strabismic
amblyopia exhibit residual performance for motion in depth,
and it is disparity based. Furthermore, some observers with
strabismic amblyopia who do not display any signicant stereo
performance for motion in depth under normal binocular view-
ing may display above-chance stereo performance if the degree
of interocular suppression is reduced. The authors term this
phenomenon latent stereopsis. (Invest Ophthalmol Vis Sci.
2009;50:50065016) DOI:10.1167/iovs.09-3551
O
ur ideas on the nature of the binocular visual decit in
strabismic amblyopia are in a state of ux. It was once
believed that observers with strabismic amblyopia lack binoc-
ular cortical mechanisms due to the effects of a binocular
competitive imbalance early in life during the critical period of
visual development. These ideas were based, on the one hand,
on the neurophysiological nding that animals deprived by
way of surgically induced strabismus early in life develop a
cortex with less responsive binocular cells
13
and, on the other
hand, the nding that observers with strabismic amblyopia do
not exhibit signicant binocular summation of contrast sig-
nals.
46
Clinically, it has been argued that observers with
strabismic amblyopia are stereoblind to static disparities when
tested with random dot stimuli devoid of the gural artifacts
contained in some clinical stereo tests.
7
The picture was of a
cortex devoid of binocular cells with the consequence of
irretrievable loss of all stereo function.
We now know that not all binocular cells are affected by
surgically induced strabismus early in life; binocular cells re-
sponding to motion in depth (i.e., dynamic stereo) may be
spared.
8
Complimentary evidence in observers with strabismic
amblyopia who are blind to static disparities shows that they
can detect dynamic disparities in motion-in-depth stimuli.
911
Furthermore, the loss of the binocular responsiveness of cor-
tical cells in strabismic animals is largely reversible
12
by iono-
phoretic applications of bicuculline (selective blocker of GABA
A receptors), suggesting an active suppression rather than a
loss of cellular function.
13
Furthermore, there is reason to
doubt the claim that observers with strabismic amblyopia do
not possess binocular mechanisms, since Baker et al.
14
showed
normal binocular contrast summation in observers with stra-
bismic amblyopia when the signal attenuation by the ambly-
opic eye is taken into account, suggesting that the lack of
summation found previously was due to the imbalance in the
monocular signals before the point of summation. Taken to-
gether, these ndings suggest that observers with strabismic
amblyopia do have binocular mechanisms. More recently, it
has been shown that the reason that binocular combination
does not normally occur for suprathreshold motion and orien-
tation tasks in observers with strabismic amblyopia is because
of interocular suppression.
15
A reduction in suppression leads
to normal levels of binocular combination in these observers,
revealing the presence of binocular cortical mechanisms. Thus,
there is converging evidence for the conjecture that persons with
strabismic amblyopia possess cortical cells with binocular con-
nections, but that during binocular viewing suppressive mecha-
nisms render the cortex functionally monocular. A new picture is
emerging that suggests that the loss of stereo function is incom-
plete with some sparing for motion in depth and that it is depen-
dent on active suppression rather than loss of cells.
In this study, we asked two questions: First, is the residual
stereo function in strabismic amblyopia that has been reported
for motion in depth due to preserved motion processing or to
preserved disparity processing? Second, is its absence in some
persons with strabismic amblyopia relative (i.e., present but
unable to be used due to suppression) or absolute (not present
at all)? Since most previous tests of stereo sensitivity in ambly-
opia involve static disparities, to obtain a more comprehensive
assessment of the depth-processing capabilities of observers
with strabismic amblyopia, we measured their stereo perfor-
mance for stimuli moving at different rates in depth. Our
stimuli were composed of dynamic random dots (a dynamic
random dot stereogram or DRS) devoid of any gural clues as
to depth.
From
1
McGill Vision Research, Department of Ophthalmology,
McGill University, Montreal, Quebec, Canada; the
2
Division of Neurol-
ogy, Department of Internal Medicine, University of Manitoba, Win-
nipeg, Manitoba, Canada; and the
3
Department of Optometry and
Vision Science, University of Auckland, Auckland, New Zealand.
Supported by Canadian Institute for Health Research (CIHR)
Grants MT108-18 and MOP53346 (RFH).
Submitted for publication February 10, 2009; revised April 1 and
28, 2009; accepted July 23, 2009.
Disclosure: R.F.Hess, None; B. Mansouri, None; B. Thompson,
None; E. Gheorghiu, None
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be marked advertise-
ment in accordance with 18 U.S.C. 1734 solely to indicate this fact.
Corresponding author: Robert F. Hess, McGill Vision Research,
Department of Ophthalmology, McGill University, 687 Pine Avenue W
(H4-14), Montreal, Quebec H3A 1A1, Canada; [email protected].
Investigative Ophthalmology & Visual Science, October 2009, Vol. 50, No. 10
5006 Copyright Association for Research in Vision and Ophthalmology
Two mechanisms have been proposed for stereo motion-in-
depth detection: (1) a mechanism that is sensitive to temporal
changes of binocular disparity.
1622
This mechanism rst de-
tects disparities and then computes motion in depth from its
change over time. This mechanism can contribute to motion in
depth in both temporally correlated and temporally uncorre-
lated DRSs. (2) A mechanism sensitive to interocular velocity
differences
2325
: This mechanism rst detects monocular mo-
tion (velocity signals) separately in the two eyes and then
computes motion in depth from the interocular velocity differ-
ences. This mechanism can contribute to motion in depth only
in temporally correlated DRSs.
Our use of temporally correlated and temporally uncorre-
lated DRSs allowed us to determine the nature of any mecha-
nism that supports any residual motion-in-depth stereo perfor-
mance in strabismic amblyopia. If residual performance is
observed only for temporally correlated DRSs then the mech-
anism must compute the motion in depth from a comparison
of the monocular velocity signals. If similar performance is
observed for both correlated and uncorrelated DRS, then a
disparity-based mechanism must be responsible.
METHODS
Observers
Fifteen subjects with strabismic amblyopia were tested: The results are
displayed in Table 1. All provided informed consent, in accordance
with the Declaration of Helsinki.
Four of those subjects (GH, AR, AM, and KS) and four normal subjects
participated in the experiment in which we examined the effects of a
neutral-density (ND) lter or an interocular delay on stereo motion per-
ception. All normal subjects had normal or corrected-to-normal visual
acuity and good stereoscopic vision. Clinical details of the four amblyopic
observers participating in the experiment are given in Table 1.
Stimuli and Procedure
The stimuli were DRSs consisting of red and green randomly distributed
bright dots (square-like elements) on a dark background (5% dot density).
The dot size was 6 arcmin. The DRSs were generated in real time on a
computer (Macintosh G4; Apple Computer, Cupertino, CA), using
OpenGL libraries graphic software (http://www.opengl.org/ provided in
the public domain by Khronos Group, Beaverton, OR) and presented to
the subjects on a monitor (1280 1024 pixels at 75 Hz; LaCie IV, LaCie,
Hillsboro, OR). Stereoscopic vision was made possible by placing red and
green lters in front of the observers eyes so that each image was visible
to one of the eyes only (light separation between red and green lters that
were matched to the r and g monitor guns was better than 99%). The
experiments were performed in an otherwise dark room.
Two types of DRSs were used: temporally uncorrelated and tempo-
rally correlated. The temporally uncorrelated DRS consisted of the ongo-
ing alternation of two image pairs: (A, A) and (B, B) (Fig. 1a). Before each
trial, the computer generated two stereograms with left and right image
pairs (L, R), which we denote by (A, A) and (B, B), that were fully
correlated. The patterns A and B were uncorrelated. Two disparity-dened
shapes (squares) were hidden on the left and right sides of each image
(Figs. 1a, 1b). Successive images contained different disparities, such that,
when displayed in alternation, they resulted in a square-wavelike modu-
lation of disparity over time. The size of the disparity-dened squares was
50 50 dots and the two disparities were D
1
0 arcmin and D
2
48
arcmin (8 dots 6 arcmin) crossed and uncrossed disparity. In successive
TABLE 1. Clinical Data for the 15 Amblyopic Subjects at the Initial Screening
Obs Age/Sex Type Refraction Acuity Squint History, Stereo
RD 49/F LE strab 3.25 DS 20/15 XT 5 Detected age 6 y, glasses since 6 y, no other treatment,
local stereopsis (160 arc second) 4.750.7545 20/40
MM 21/F RE strab 20/40 ET 5 Detected age 5 y, patching for 1 y, recent surgery, no
static stereopsis 20/25
YC 36/M LE Strab 2.00 DS 20/15 ET 5 Detected age 2 y, patching for 4 years, glasses for 18 y,
no static stereopsis 2.00 DS 20/40
AR* 47/M RE strab Plano 20/20 ET 2 Detected age 6 y, no patching, no surgery, no static
stereopsis Plano 20/50
GH* 45/M RE mixed 1.250.5180 20/20 ET 6 Detected age 10 y, patching for 1 m, glasses for 1 y, no
static stereopsis 1.25DS 20/63
ED 43/F LE strab 0.75DS 20/16 ET 3 Detected age 6 y, patching for 1 y, local stereopsis (40
arc second) 0.75DS 20/63
PH 33/M LE mixed 2.00DS 20/25 ET 5 Detected age 4 y, patching for 6 m; surgery age 5 y, no
static stereopsis 0.50DS 20/83
GN 32/M RE mixed 5.002.00120 20/70 ET 8 Detected age 5 y, patching for 3 m, no glasses, 2
strabismus surgeries RE age 1012, no static stereopsis 3.501.0075 20/20
KS* 40/M RE strab 0.501.00180 20/70 EX 4 Detected age 10 y, patching for 1 m, glasses for 1 y, no
static stereopsis 0.50DS 20/20
SY 39/F LE mixed 8.00D 20/20 ET 1 Detected age 5 y, patching, no surgery, no static
stereopsis 4.002.50180 20/70
MB 50/M RE strab 1.00 DS 20/32 ET 3 Steady central xation, no surgery, rst glasses at 32
years, no static stereopsis 1.00 DS 20/80
ML 23/F RE mixed 1.00.7590 20/80 ET 6
Detected age 5 y, patching for 2 y, no static stereopsis 3.25 DS 20/25
MG 33/F RE strab 0.50 DS 20/100 ET 1 Detected age 4 y, patching for 6 m, no surgery, no static
stereopsis 0.50 DS 20/15
XL 33/F LE strab 12.50 DS 20/20 ET 15
Detected age 13 y, no treatment, no static stereopsis 12.75 0.75110 20/400
AM* 44/M RE strab Plano 20/20 ET 20 Detected age 4 y, no patching, no surgery, no static
stereopsis Plano 20/600
The angle of strabismus was measured with a major amblyoscope, and stereopsis was measured with the Randot test. The squint angle is given
in degrees where 1 equals 1.75 prism diopter. strab, strabismus; Mixed, strab and aniso; Sq, squint; Obs, observers; RE, right eye; LE, left eye; ET,
esotropia; XT, exotropia; ortho, orthotropic alignment; DS, diopter sphere.
* Subjects who underwent a more detailed investigation.
IOVS, October 2009, Vol. 50, No. 10 Motion in Depth in Strabismic Amblyopia 5007
images, the two squares were displaced in opposite directions in
depth (i.e., one square alternated between 0 and 48 arcmin crossed
disparity and the other between 0 and 48 arcmin uncrossed
disparity. The temporally correlated DRS consisted of a single-
image pair (A, A) that alternated between the two disparity values,
0 and 48 arcmin (Fig. 1b). Because the same textural pattern was
used in the temporally correlated DRS, the disparity-dened shape
steps in depth relative to a static background, whereas in temporally
uncorrelated DRS, the depth of the disparity-dened square is judged
inrespect witha dynamic background. To make the relative depthjudgments
as comparable as possible for the two types of DRS, we used also a dynamic
background for temporally correlated DRS (i.e., the background had a differ-
ent random-dot textural pattern on every frame).
Experiment 1. We varied the presentation times of the two
images in steps of 13.3 ms, between 13.3 and 226 ms. This method
resulted in temporal frequencies between 37.5 and 2.2 Hz. The step
size of 13.3 ms for the presentation times was dictated by the frame
rate of the monitor (1/75 Hz). Different frame durations (i.e., tem-
poral frequencies) were presented in a random order during each
session. Each of the 17 temporal frequency conditions was mea-
sured 20 times.
Experiment 2. We repeated these measurements on four observ-
ers with strabismic amblyopia who exhibited chance performance in
experiment 1 but with an ND lter (ND 1.00) in front of the fellow xing
eye. Similar measurements were undertaken in four normal observers.
Experiment 3. In the experiment in which we examined the
effects of interocular delay, the DRSs consisted of the continuous alterna-
tion of two image pairs: image pair (A, B) and image pair (B, A) (Fig. 1c).
The two alternated images were uncorrelated in two ways: spatially and
temporally (i.e., the left and right images were uncorrelated at all times
and also each eyes image consisted of the alternation of two uncorrelated
images). The stimulus (A, B), (B, A) constituted a DRS in which correlation
was 0 within each of the periods T
AB
and T
BA
and thus, throughout the
presentation time. In this stimulus, the sequence of patterns was identical
in the left and right eyes (i.e., A, B, A, B,), but the patterns were out of
phase. However, if the visual system would tolerate the delay of one eyes
image by one time period, then this stimulus would become binocularly
indistinguishable from the temporally uncorrelated DRS stimulus (A, A),
(B, B), resulting in high correlation at all times. We varied the presentation
times of the two images in steps of 13.3 ms, between 13.3 and 186 ms.
This variation resulted in temporal frequencies between 37.5 and 2.6 Hz.
Again, different frame durations (i.e., temporal frequencies) were pre-
FIGURE 1. Stimuli used in the ex-
periments: (a) temporally uncorre-
lated DRS consisting of the ongoing
alternation of two image pairs, (A, A)
and (B, B), each having one of two
disparity values: 0 and 48 arcmin.
The patterns A and B were uncorre-
lated. (b) Temporally correlated
DRS in which a single image pair (A,
A), alternated between two disparity
values: 0 and 48 arcmin. (c) DRS
presented with interocular delay.
The stimulus (A, B), (B, A) consists of
the ongoing alternation of two image
pairs: (A, B) and (B, A).
5008 Hess et al. IOVS, October 2009, Vol. 50, No. 10
sented in a random order within each session. Each temporal frequency
was measured 20 times.
Experiments 1, 2, and 3. We used a forced-choice para-
digm and a depth discrimination task in which the subject indicated
(by pressing a key) whether the disparity-dened square in the left
or in the right side of the screen was closest to him or her. The DRSs
were presented on the screen until the subject gave an answer. The
subjects were free to make eye movements. We used a level of 75%
correct answers, indicated by a dotted horizontal line in the graph
averages, as the criterion for reliable depth detection.
RESULTS
Experiment 1: Temporally Correlated and
Uncorrelated Motion-in-Depth Sensitivity
From our initial screening, which involved the measurement
of performance (percentage correct) as a function of the
temporal frequency (or frame duration in milliseconds), we
found only two subjects who exhibited above-chance per-
formance (ED, RD). In these subjects, performance was near
ceiling level and did not show any dependence on either the
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
MM
GN
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
ED
XL
Frame duration (ms)
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
RD
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
SY
YC
MG
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms)
FIGURE 2. Percentage correct as a
function of the reciprocal of the tem-
poral frequency (frame durations)
for eight amblyopic subjects for tem-
porally correlated () and uncorre-
lated (F) DRSs.
IOVS, October 2009, Vol. 50, No. 10 Motion in Depth in Strabismic Amblyopia 5009
correlated/uncorrelated nature of the stimuli or the tempo-
ral frequency. The data from 11 observers with strabismic
amblyopia, including the two subjects with stereo function
are displayed in Figures 2 and 3 for the temporally correlated
and temporally uncorrelated DRSs.
Experiment 2: Effect of the ND Filter
We choose four strabismic subjects whose screening results
suggested no stereoscopic function and retested their stereo
performance for the temporally correlated and temporally un-
correlated DRSs with an ND lter (density 1.0, with an atten-
uating effect of a factor of 10) in front of the fellow xing eye.
Our reasoning was to reduce the suppressive inuence of the
normal eye to test the conjecture that there is latent stereo
function in these subjects under these viewing conditions. The
individual and group results from these four subjects are
shown in Figure 4 for uncorrelated DRSs and Figure 5 for
correlated DRSs.
These results are plotted in the same way as described in
Figure 2. It is apparent that the application of an ND lter in
front of the xing eye improved the stereo performance in
some of the observers with strabismic amblyopia. To test
whether the performance was signicantly better with the
1-ND lter versus without 1-ND lter with temporally cor-
related and uncorrelated DRSs, we performed a three-way
ANOVA (analysis of variance), with type of DRS (temporally
correlated versus temporally uncorrelated DRS), combina-
tion (with 1-ND lter vs. without 1-ND lter), and frame
duration or temporal frequency (13.3226 ms) as factors on
the data shown in Figures 4b and 5b. The main effect of
combination (with versus without 1-ND lter) was signi-
cant (F
(1,1)
254.27, P 0.05). The effect of temporal
frequency was not signicant (F
(1,16)
0.91, P 0.05; P
0.5579). The effect of type of DRS was also not signicant
(F
(1,1)
3.01, P 0.05; P 0.0891). Thus, the improve-
ment of the group was signicant for uncorrelated and
correlated DRSs when an ND lter was used.
For comparative purposes, we undertook a similar compar-
ison in four normal subjects, in which the ND lter was placed
in front of the dominant eye. These results are shown in
Figures 6 (uncorrelated DRS) and 7 (correlated DRS). These
results show that the placement of an ND lter in front of one
eye reduced stereo performance; again, some subjects were
affected more than others. Performance for both correlated
and uncorrelated DRSs was reduced.
Experiment 3: Effect of Interocular Delay
It remains a possibility that the reason that stereopsis was
not seen when the DRS was used by observers with strabis-
mic amblyopia is that there was a delay between the outputs
of the two eyes and that the juxtaposition of an ND lter in
front of the xing eye improves the synchronization of the
monocular inputs at the point of binocular combination.
Such a timing difference may be also the result of suppres-
sion or may have nothing whatsoever to do with suppres-
sion. To investigate this possibility, we reassessed stereo
performance of our four observers with strabismic amblyo-
pia, but this time with a physical delay between the inputs
of the two eyes. This delay varied with the temporal rate,
and we wanted to know whether we could mimic the effects
we had previously found with the ND lter using stimuli that
were simply delayed. We compared uncorrelated DRS (as in
Fig. 1a) with uncorrelated delayed DRS (as in Fig. 1c) inter-
leaved within the same run. The results are displayed in
Figure 8. Subjects GH and AM exhibited the same baseline
performance (Fig. 8) as previously described in Figures 4
and 5, even though these measurements were performed 6
to 9 months apart. For these two subjects, it is clear that a
simple delay (Fig. 8) does not in itself improve stereo per-
formance in strabismic eyes. A similar conclusion can be
reached for the results of subject KS although baseline
measurements (Fig. 8) are worse than measured previously
(Figs. 4, 5). The results of subject AR do not bear on this
issue, because in the time between the initial measurements
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
MB
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
ML
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms)
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
PH
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms)
Temporally correlated DRS
Temporally uncorrelated DRS
FIGURE 3. Percentage correct as a
function of the reciprocal of the tem-
poral frequency (frame durations)
for three amblyopic subjects for tem-
porally correlated and uncorrelated
DRSs.
5010 Hess et al. IOVS, October 2009, Vol. 50, No. 10
(Figs. 4, 5) and the current ones (Fig. 8), he underwent an
antisuppression treatment, which resulted in the restoration
of binocular fusion with subsequent establishment of static
stereopsis.
26
This improvement in static stereopsis (80 sec-
onds, according to the Randot test) also occurred for stereo-
in-depth performance (Fig. 8, compare baseline posttreat-
ment performance with previous pretreatment performance
in Figs. 4, 5). On the basis of previous results in normal
subjects,
27
we were expecting to see an artifactual improve-
ment at the shortest duration (i.e., at the highest repetition
rate) as the DRS is perceived to be static at a 50% correla-
tion, a nding present only in the results of amblyopic
subject AR for frame durations shorter than 50 ms. Overall,
stereo performance was not improved by introducing a time
delay across the temporal range, as we had seen previously
with the ND lter.
DISCUSSION
Static stereopsis in observers with strabismic amblyopia is rare,
but it can occur. The screening data of McKee et al.
28
suggest
an incidence of 2%. It has been reported that a greater
fraction of people with strabismic amblyopia exhibit some
stereo performance for dynamic stimuli. Rouse et al.
10
report
that half of their observers with strabismic amblyopia who
were stereoblind to static disparities performed above chance
with dynamic stereo stimuli. We found 2 (13%) of 15 cases
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms) Frame duration (ms)
a
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
AR
GH
AM
KS
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
No filter
With filter
Temporally Uncorrelated DRS
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Average across subjects
Frame duration (ms)
b
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
No filter
With filter
FIGURE 4. Percentage correct as a
function of the reciprocal of the tem-
poral frequency (frame durations)
for four strabismic subjects for tem-
porally uncorrelated DRSs, both
with and without an ND lter in
front of the xing eye. Individual (a)
and group (b) results are displayed.
IOVS, October 2009, Vol. 50, No. 10 Motion in Depth in Strabismic Amblyopia 5011
from our initial screening (i.e., RD and ED) with motion-in-
depth stimuli, and these were the same two individuals who
exhibited stereopsis for static stimuli (i.e., the Randot test).
Clinically, there was nothing that set these two observers with
strabismic amblyopia apart from the other subjects. The task
used by Rouse et al. did not involve a depth judgment, and it is
possible that velocity-based cues could have determined the
above-chance performance that they reported. We set out to
use an approach that would specically address this question.
Our task was based on motion in depth and our two stimulus
manipulations were designed to answer the question, is any
above-chance performance for motion in depth based on in-
terocular velocity differences or disparity differences? Compar-
ing performance for temporally correlated and uncorrelated
DRSs allowed us to answer this question, because although
both stimuli have disparity-based information, the monocular
velocity-based information is disrupted in the temporally un-
correlated DRS. Since we found comparable performance for
these two different stimulus manipulations, we conclude that
the residual motion in depth performance in these two sub-
jects could not be ascribed to monocular velocity mechanisms.
In addition, in four randomly selected cases who did not
exhibit above-chance motion-in-depth performance, we found
improved stereo performance across all temporal rates tested
when an ND lter was put in front of the xing eye. A similar
comparison of static stereopsis using the Randot and TNO
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250 0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms) Frame duration (ms)
a
AR
GH
AM
KS
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
No filter
With filter
Temporally Correlated DRS
b
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms)
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
No filter
With filter
Average across subjects
FIGURE 5. Percentage correct as a
function of the reciprocal of the tem-
poral frequency (frame durations)
for four strabismic subjects for tem-
porally correlated DRSs, both with
and without an ND lter in front of
the xing eye. Individual (a) and
group (b) results are displayed.
5012 Hess et al. IOVS, October 2009, Vol. 50, No. 10
stereo tests with and without the ND lter did not reveal any
improved stereopsis for any of these four individuals. This
suggests a degree of latent stereopsis that is specic for dy-
namic stimuli in strabismic amblyopia.
Effect of an ND Filter on Contrast and Mean Lumi-
nance. It has been shown that persons with strabismic ambly-
opia have binocular mechanisms, because they exhibit normal
levels of binocular contrast summation if the contrast threshold
decit of the amblyopic eye is rst accounted for.
14
Further-
more, it has been shown that when the suppression normally
exerted by the fellow xing eye is reduced, suprathreshold
information from the two eyes of observers with strabismic
amblyopia can be combined normally.
15
A reduction in the
interocular contrast is sufcient to effect such a reduction in
suppression.
15
All this argues for there being intact binocular
hardware within the visual cortex of observers with strabismic
amblyopia that is rendered functionally monocular due to ac-
tive suppression. More recently, we have shown that suppres-
sion results in a reduction in the perceived mean luminance by
the amblyopic eye rather than a reduction in contrast percep-
tion (Maehara G et al., manuscript in preparation) and that the
suprathreshold decit for contrast discrimination in amblyopia
can be modeled by a normal eye viewing through an ND
lter.
29
With the stimuli used in the present study, an ND lter
will reduce both the contrast and the mean luminance since
they were displayed as bright elements on an otherwise dark
background. Therefore, the benecial effects of placing an ND
lter in front of the xing eye is likely to be the result of
reducing the xing eyes suppressive inuence via a reduction
in contrast, mean luminance, or both.
Effect of an ND Filter on Timing. The effect of an ND
lter placed over one eye is also to introduce an interocular
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms)
Frame duration (ms)
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250 0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
No filter
With filter
AV CA
BT EN
a
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Average across subjects
Frame duration (ms)
b
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
No filter
With filter
Temporally Uncorrelated DRS
FIGURE 6. Percentage correct as a
function of the reciprocal of the tem-
poral frequency (frame durations)
for four stereonormal subjects for
temporally uncorrelated DRSs, both
with and without an ND lter in
front of the dominant eye. Individual
(a) and group (b) results are dis-
played.
IOVS, October 2009, Vol. 50, No. 10 Motion in Depth in Strabismic Amblyopia 5013
delay between the signals to the two eyes. Several studies have
been undertaken to investigate the effects of interocular delays
in normal subjects,
3032
and the reports have shown that the
stereoscopic system tolerates a time difference between bin-
ocular correlated random-dot images of up to 50 ms. Using
DRSs, Julesz and White
30
investigated the delay hypothesis and
reported that a disparity-dened square of 48 arcmin was not
perceived with one frame delay (80 ms). Cumming and
Read
33
recorded responses of disparity-selective V1 neurons
using DRSs in which each image was refreshed for every frame
and showed that an interocular delay of only one video frame
(14 ms) reduced dramatically the magnitude of the disparity-
selective response. Using different DRSs consisting of contin-
uous alternation of two image pairs, Gheorghiu and Erkelens
27
showed that disparities from simultaneously presented monoc-
ular inputs dominate those from interocularly delayed inputs
and thus, interocular time delays between correlated images
are hardly tolerated in the visual system.
To address whether the reason that stereopsis is not seen
using DRS in observers with strabismic amblyopia is because
there is a delay between the two eyes outputs and that the
juxtaposition of an ND lter in front of the xing eye improves
the synchronization of the monocular inputs at the point of
binocular combination, we reassessed stereo performance for
our four observers with strabismic amblyopia, but this time
with a delay between the two eyes inputs. This delay varied
with the temporal rate, and we wanted to know whether we
could mimic the effects we had previously found with the ND
lter using stimuli that were simply delayed. We compared
uncorrelated DRS (as in Fig. 1a) with uncorrelated delayed DRS
(as in Fig. 1c) interleaved within the same run. The results (Fig.
8) show that this explanation is not tenable, as stereo perfor-
mance was not improved as we had seen previously for the ND
lter by introducing a time delay across the temporal range.
In conclusion, along with other researchers,
911
we have
found that stereo performance for motion-in-depth stimuli is
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Frame duration (ms)
Frame duration (ms)
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
a
Temporally Correlated DRS
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
No filter
With filter
CA
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
AV
BT
EN
P
e
r
c
e
n
t

c
o
r
r
e
c
t

r
e
s
p
o
n
s
e
s
Average across subjects
Frame duration (ms)
b
0 50 100 150 200 250
0
10
20
30
40
50
60
70
80
90
100
No filter
With filter
FIGURE 7. Percentage correct as a
function of the reciprocal of the tem-
poral frequency (frame durations) in
four stereonormal subjects for tem-
porally correlated DRSs, both with
and without an ND lter in front of
the xing eye. Individual (a) and
group (b) results are displayed.
5014 Hess et al. IOVS, October 2009, Vol. 50, No. 10
possible in strabismic amblyopia although we found it in the
same individuals who exhibited static stereopsis (i.e., RD and
ED). Above-chance performance for motion-in-depth stimuli
does not depend on the temporal rate, and it is based on
disparity detection not monocular velocity detection. In cases
in which it is not present under normal binocular viewing, it
may be revealed under conditions in which suppressive inu-
ences of the fellow xing eye are reduced. This can be
produced by interocular reductions of contrast and mean lu-
minance. The use of an ND lter can be benecial. These
results suggest that in some cases of strabismic amblyopia (all
the cases studied here were of microstrabismus except two),
there is a latent stereopsis for dynamic stimuli. Our use of
stimuli of xed disparities does not permit us to speculate
about the range of disparities over which this latent stereopsis
operates nor on the relationship between the size of suppres-
sion scotomata and residual motion-in-depth performance.
References
1. Baker F, Grigg P, von Noorden G. Effects of visual deprivation and
strabismus on the response of neurons in the visual cortex of the
monkey, including studies of striate and prestriate cortex in the
normal animal. Brain Res. 1974;66:185208.
2. Bennett MJ, Smith EL 3rd, Harwerth RS, Crawford ML. Ocular
dominance, eye alignment and visual acuity in kittens reared with
an optically induced squint. Brain Res. 1980;193:3345.
3. Hubel DH, Wiesel TN. Binocular interaction in striate cortex of
kittens reared with articial squint. J Neurophysiol. 1965;28:1041
1059.
4. Levi DM, Harwerth RS, Smith EL 3rd. Observers deprived of normal
binocular vision have binocular interactions tuned to size and
orientation. Science. 1979;206:852854.
5. Levi DM, Harwerth RS, Smith EL 3rd. Binocular interactions in
normal and anomalous binocular vision. Doc Ophthalmol. 1980;
49:303324.
6. Vedamurthy I, Suttle CM, Alexander J, Asper LJ. Interocular inter-
actions during acuity measurement in children and adults, and in
adults with amblyopia. Vision Res. 2007;47:179188.
7. Cooper J, Feldman J. Random-dot-stereogram performance by stra-
bismic, amblyopic, and ocular-pathology patients in an operant-
discrimination task. Am J Optom Physiol Opt. 1978;55:599609.
8. Cynader M, Gardner JC, Mustari M. Effects of neonatally induced
strabismus on binocular responses in cat area 18. Exp Brain Res.
1984;53:384399.
9. Kitaoji H, Toyama K. Preservation of position and motion stereop-
sis in strabismic subjects. Invest Ophthalmol Vis Sci. 1987;28:
12601267.
10. Rouse MW, Tittle JS, Braunstein ML. Stereoscopic depth per-
ception by static stereo-decient observers in dynamic displays
with constant and changing disparity. Optom Vis Sci. 1989;66:
355362.
11. Sireteanu R, Fronius M, Singer W. Binocular interaction in the
peripheral visual eld of observers with strabismic and anisome-
tropic amblyopia. Vision Res. 1981;21:10651074.
12. Mower GD, Christen WG, Burchel JL, Duffy FH. Microionto-
phoretic bicuculline restores binocular responses to visual cortical
neurons in strabismic cats. Brain Res. 1984;309:168172.
13. Sengpiel F, Jirmann K-U, Vorobyov V, Eysel U. Strabismic suppres-
sion is mediated by interactions in the primary visual cortex.
Cerebral Cortex. 2006;16:17501758.
14. Baker DH, Meese TS, Mansouri B, Hess RF. Binocular summation of
contrast remains intact in strabismic amblyopia. Invest Ophthal-
mol Vis Sci. 2007;48:53325338.
15. Mansouri B, Thompson B, Hess RF. Measurement of suprathresh-
old binocular interactions in amblyopia. Vision Res. 2008;48(28):
27752784.
Frame duration (ms)
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200
0
AM P
e
r
c
e
n
t
a
g
e

c
o
r
r
e
c
t

a
n
s
w
e
r
s

Frame duration (ms)
0 50 100 150 200
0
10
20
30
40
50
60
70
80
90
100
P
e
r
c
e
n
t
a
g
e

c
o
r
r
e
c
t

a
n
s
w
e
r
s

GH
Frame duration (ms)
P
e
r
c
e
n
t
a
g
e

c
o
r
r
e
c
t

a
n
s
w
e
r
s

0 50 100 150 200
0
10
20
30
40
50
60
70
80
90
100
AR
Frame duration (ms)
0 50 100 150 200
0
10
20
30
40
50
60
70
80
90
100
P
e
r
c
e
n
t
a
g
e

c
o
r
r
e
c
t

a
n
s
w
e
r
s

KS
Temporally uncorrelated
Temporally correlated
With interocular delay
DRS
FIGURE 8. Percentage correct per-
formance for correlated and uncor-
related motion-in-depth stimuli, with
and without a temporal delay that
scaled with temporal frequency. The
results were obtained by temporally
interleaving these three conditions
within the same run in four observ-
ers with strabismic amblyopia.
IOVS, October 2009, Vol. 50, No. 10 Motion in Depth in Strabismic Amblyopia 5015
16. Cumming BG. The relationship between stereoacuity and stereo-
motion thresholds. Perception. 1995;24:105114.
17. Cumming BG, Parker AJ. Binocular mechanisms for detecting mo-
tion-in-depth. Vision Res. 1994;34:483495.
18. Lages M, Mamasian P, Graph EW. Spatial and temporal tuning of
motion in depth. Vision Res. 2003;43:28612873.
19. Norcia AM, Tyler CW. Temporal frequency limits for stereoscopic
apparent motion processes. Vision Res. 1984;24:395401.
20. Regan DM. Binocular correlates of the direction of motion in
depth. Vision Res. 1993;33:23592360.
21. Tyler CW. Stereoscopic depth movement: two eyes less sensitive
than one. Science. 1971;174(12):958961.
22. Gheorghiu E, Erkelens CJ. Differences in perceived depth for
temporally correlated and uncorrelated dynamic random-dot ste-
reograms. Vision Res. 2005;45:16031614.
23. Regan D, Beverley KI. Binocular and monocular stimuli for motion
in depth: changing disparity and changing size feed the same
motion-in-depth stage. Vision Res. 1979;19:13311342.
24. Regan DM, Beverley KI, Cynader M. Separate subsystems for po-
sition in depth and for motion in depth. Proceed Roy Society
London B. 1979;204:485501.
25. Shioiri S, Saisho H, Yaguchi H. Motion in depth based on inter-
ocular velocity differences. Vision Res. 2000;40:25652572.
26. Mansouri B, Thompson B, Hess RF. Improvements in binocular
vision following a new binocularly-based treatment for amblyopia
(abstract). Abstract. Washington, DC: Society for Neuroscience:
2008:Program 454.4.
27. Gheorghiu E, Erkelens CJ. Temporal properties of disparity pro-
cessing revealed by dynamic random-dot stereograms. Perception.
2005;34:12051219.
28. McKee SP, Levi DM, Movshon JA. The pattern of visual decits in
amblyopia. J Vis. 2003;3:380405.
29. Baker DH, Meese TS, Hess RF. Contrast masking in strabismic
amblyopia: attenuation, noise, interocular suppression and binoc-
ular summation. Vision Res. 2008;48(15):16251640.
30. Julesz B, White B. Short term visual memory and the Pulfrich
phenomenon. Nature. 1969;222:639641.
31. Mitchell DE, OHagan S. Accuracy of stereoscopic localization of
small line segments that differ in size or orientation for the two
eyes. Vision Res. 1972;12:437454.
32. Ross J, Hogben JH. Short-term memory in stereopsis. Vision Res.
1974;14(11):11951201.
33. Cumming BG, Read JC. Effect of interocular delay on disparity-
selective v1 neurons: relationship to stereoacuity and the pulfrich
effect. J Neurophysiol. 2005;94:15411553.
5016 Hess et al. IOVS, October 2009, Vol. 50, No. 10