around its meridian axis in each eye, the directional relation between ..... alternation of normal and reversed depth directions; (iv) the perception of the slant in.
Perception, 1993, volume 22, pages 971-984
How is depth perception affected by long-term wearing of left-right reversing spectacles? Makoto Ichikawa Department of Psychology, Faculty of Letters, Osaka City University, Sumiyoshi-ku, Osaka 558, Japan Hiroyuki Egusa Department of Psychology, Faculty of Letters, Chiba University, Yayoi-cho, Chiba 263, Japan Received 30 January 1991, in revised form 3 September 1992
Abstract. The plasticity of binocular depth perception was investigated. Six subjects wore leftright reversing spectacles continuously for 10 or 11 days. On looking through the spectacles, the relation between the direction of physical depth (convex or concave) and the direction of binocular disparity (crossed or uncrossed) was reversed, but other depth cues did not change. When subjects observed stereograms through a haploscope and were asked to judge the direction of perceived depth, the directional relation between perceived depth and disparity was reversed both in the two line-contoured stereograms and in the random-dot stereogram in the middle of the wearing period, but the normal relation often returned late in the wearing period. When subjects observed two objects while wearing the spectacles and were asked which appeared the nearer, veridical depth perception increased as the wearing-time passed. These results indicate that the visual transformation reversing the direction of binocular disparity causes changes both in binocular stereopsis and in processes integrating different depth cues. 1 Introduction When one wears left-right reversing spectacles, which reverse the whole retinal image around its meridian axis in each eye, the directional relation between physical depth and disparity is reversed. That is, a convex object produces uncrossed disparity and a concave object, crossed disparity. But other depth cues, eg motion parallax, occlusion, and linear perspective, are not influenced by this visual transformation and present veridical depth information. Consequently, the left-right reversal of retinal images introduces conflicts between binocular disparity and other depth cues with respect to the same physical object. Thus it is to be expected that the spectacles will produce abnormalities of depth perception, which interfere with adaptive behaviour. How does the perceptual system resolve this informational conflict? Several studies have indicated that binocular depth perception undergoes adaptive changes in a few minutes when the visual input is optically transformed so that the amount of disparity is changed, thus causing an informational conflict between binocular disparity and other depth cues (eg Epstein 1975; Fisher and Ciuffreda 1990; Fisher and Ebenholtz 1986; O'Leary and Wallach 1980; Wallach et al 1963). However, it has also been shown that a visual transformation which reverses the direction of disparity could not cause the adaptive changes of stereopsis, such that crossed disparity produces an impression of concave depth (eg Brown 1928; Stratton 1898). It has been accepted that the relation between the direction of disparity (crossed or uncrossed) and perceived-depth direction (convex or concave) is stable and that its reversal is beyond the plasticity of stereopsis. Therefore, few studies have examined the effects on stereopsis of a visual transformation which reverses the direction of disparity. Shimojo and Nakajima (1981) reported that a continuous visual transformation could cause the adaptive reversal of binocular stereopsis. One of their two subjects wore left-right reversing spectacles for 9 days and observed line-contoured stereograms (LCSs) and random-dot stereograms (RDSs). On day 3 of the wearing period,
972
M Ichikawa, H Egusa
the relation between the direction of the disparity of LCSs and the direction of perceived depth was reversed completely. This tendency was maintained until day 4 after the removal of the spectacles, and had not disappeared completely even on the 79th day. In contrast, such effects were not found for RDSs. Their other subject showed similar tendencies, but the occurrence of reserved-depth perception depended on the point of fixation. From these results, Shimojo and Nakajima (1981) assumed that the process of stereopsis consisted of two subprocesses: one tuned to LCSs which underwent adaptive changes, and the other tuned to RDSs which was independent of visual experience. Furthermore, they suggested that the changeable process might be the process of 'motor fusion'. They assumed that this motor fusion was the process governed by minute eye movements which scanned linear segments in stereograms. The study of Shimojo and Nakajima (1981) has many implications for understanding depth perception, but these workers investigated only binocular stereopsis during the observation of stereograms. The purpose of our study was to examine the effects of reversal of the retinal images, not only on binocular stereopsis but also on the whole depth-perception system. We conducted two tests of depth perception on six subjects who wore left-right reversing spectacles continuously for 10 or 11 days. We designed test 1 to confirm Shimojo and Nakijima's (1981) results. In test 2, we investigated depth perception in physical space under a condition in which conflicts between binocular disparity and other depth cues occurred. 2 General method 2.1 Subjects The subjects were five male undergraduates and one of the authors (MI). All of them except MI were volunteers. Subjects SW, KW, and HO had normal acuity, and the others corrected-to-normal acuity. They all had stereoscopic acuity of at least 40 s as measured in the stereo fly test. They wore left-right reversing spectacles continuously for 10 days (subjects MI, SW, and HY) or 11 days (subjects KW, HH, and HO). To avoid any visual experience without the spectacles, subjects removed the spectacles and wore an eyemask during sleep and rest periods. The subjects in the experimental group carried on with their normal daily lives while wearing the left-right reversing spectacles. They participated in two tests on depth perception before, during, and after the same wearing period. Table 1 shows the schedule of the two tests for each subject. In addition, they recorded impressions of depth in everyday life, with a micro tape recorder. This experiment formed part of a systematic study of various aspects of adaptation to left-right reversed vision, and the subjects of the experimental group also participated in some other experiments. They were encouraged to be as active as possible in everyday life. On the first day of the wearing period they all had serious sickness and had trouble in some actions, such as walking about, orienting their face to someone in order to converse, and eating with a knife and fork. On about the third day of the wearing period, their sickness gradually disappeared, and they mastered everyday actions step-by-step. However, other more delicate tasks such as riding a bicycle and catching a bouncing ball, were very difficult for all of them, even on the last day of the wearing period. The control group consisted of five other subjects who had no experience of leftright reversed vision. They all had normal or corrected-to-normal acuity and the same stereoscopic acuity as the experimental subjects. All of the control group underwent six sessions of test 1 on different days. One of them also completed five sessions with test 2 on different days.
973
Effects of retinal-image reversal on depth perception
Table 1. Time schedule of the tests for each subject in the experimental group. Asterisks represent the day on which the subject was tested. The days of the pre-wearing period (pre-WP) are counted backwards from the start of the wearing period (WP). Day 0 of the prewearing period is the same day as day 1 of the wearing period. Day 0 of the post-wearing period (post-WP) is the same day as the last day of the wearing period. Subject and test number
Time period Pre-WP
WP
Post-WP
day 2 1 0 1 2 3 4 5 6 7 8 9 10 11 0 1 4 8 9 11 16 17 24 32 51 107
KW 1 2
HH 1 2
SW 1 2
HY 1 2
MI 1 2
HO 1 2
* * * * *
* *
* *
* *
* *
*
* *
* * * *
*
*
*
*
*
*
*
* * * *
*
* *
*
He
* *
* *
* * * *
*
*
*
*
*
*
*
*
*
*
*
*
*
*
* *
*
* *
* * *
* *
*
*
*
*
*
*
*
*
* *
* * * *
2.2 Apparatus The spectacles were constructed out of two triangular prisms fixed on aluminium or balsa frames. The prisms were set in front of each eye and brought about left-right reversal of retinal images (figure 1). The frames and the prisms restricted the
(b) Figure 1. Schematic plan diagram of the relation between object depth and retinal images. (a) When a convex object is observed without wearing spectacles, the observer's retinal images are represented by three dots and a line on the right side of each eye. (b)When the same convex object is observed through left-right reversing spectacles constructed from a pair of triangular prisms, the retinal images are reversed.
974
M Ichikawa, H Egusa
subject's field of view to about 50 deg height x 70 deg width. For subjects MI, HY, and HH, corrective lenses were attached to the spectacles in front of the prisms. 3 Test 1 The main aim of test 1 was to confirm the results of Shimojo and Nakajima (1981) with the same types of stereograms as they had used: a line stereogram of an LCS pyramid and an RDS. In addition, our subjects were tested on a new type of LCS, a stereogram consisting of two vertical lines, as shown in figure 2a, to examine differences in the influence of reversal of retinal images on depth perception of different types of stereograms. Our new type of stereogram was expected to be less affected by linear perspective depth information than is the pyramid stereogram. We also compared depth perception with free eye movements with that with restricted eye movements, so testing Shimojo and Nakajima's (1981) hypothesis that the adaptive changes are explicable in terms of minute eye movement during observation of the stereograms. 3.1 Method 3.1.1 Apparatus and stimuli. We used three types of stereogram (figure 2). Stereogram a was composed of two vertical lines, whose length was 6.8 deg. The right line had 20 min or 10 min binocular disparity relative to the left line. There was a black point at the centre of the left line. Stereogram b was composed of a square 10 deg x 10 deg and diagonals. The intersection of the diagonals had 40 min or 10 min binocular
(a)
Figure 2. Examples of stereograms used in test 1. When the figure on the right is presented to the right eye and the figure on the left to the left eye, ie when the stereogram has a crossed disparity, (a) a line with a dot appears nearer in stereogram a, (b) a convex pyramid appears in stereogram b, and (c) a flat rectangle appears to protrude from its surround in stereogram c.
Effects of retinal-image reversal on depth perception
975
disparity relative to the square. These disparity levels were smaller than the largest (48 min) and the smallest (12 min) disparities used by Shimojo and Nakajima (1981). Stereogram c was the RDS in which a horizontal (2.9 deg x 7.8 deg) or a vertical (7.8 deg x 2.9 deg) rectangle can be perceived (a 50%-reduced copy from Julesz 1971, p 280). The rectangle had 17.6 min binocular disparity relative to its background. This disparity was smaller than the 23 min disparity used by Shimojo and Nakajima (1981). Each disparity value of stereograms a and b or each shape of type c was presented both crossed and uncrossed. Furthermore, we prepared versions of the stereograms both with and without a fixation point. The fixation point was a letter a, and was placed near the right edge of the stereograms. The size of the fixation point was about 30 min x 30 min. The stereograms were drawn on transparent acrylic sheets. These sheets were inserted in a synoptiscope (Inami L-250), which is a type of haploscope, and were illuminated from the rear. The luminance of the stimulus surface was about 2.4cdm~ 2 . The distance from the subject's eyes to the stereogram surface was 17 cm. 3.1.2 Procedure. All subjects observed the stereograms through the synoptiscope with the spectacles removed. Thus, the effects of wearing the spectacles on stereopsis were measured as aftereffects. The subjects took off their reversing spectacles just before starting the test. They immediately looked into the synoptiscope so that they did not see anything except the stereogram without the spectacles. We compared a free-observation condition with a restricted-observation condition. In the free-observation condition, subjects observed stereograms without a fixation point, and actively moved their eyes over the stereogram. In the restricted-observation condition, they gazed at the fixation point while observing the stereograms. The subjects judged whether the line with the black point appeared nearer than the other line for stereogram a; whether the diagonal lines appeared convex or concave for stereogram b; and whether the rectangle was vertical or horizontal and whether it appeared in front of or behind its surround for stereogram c. In addition, they reported any other impressions while observing the stereograms. Subjects SW, MI, and HY estimated the perceived depth in centimetres, using the observation distance as a standard of 20 cm. The three types of stereograms a, b, and c were presented in alphabetical order to all of the subjects. Subjects SW, MI, and HY and the control group observed the stereograms in both observation conditions. The 8 conditions (2 levels of disparity or 2 shapes x 2 directions of disparity x 2 observational conditions) for each type of stereogram were presented four times in random order. The observational condition was announced before each presentation. Subjects KW and HH observed only the stereograms without the fixation point in the free-observation condition. The four conditions (2 levels of disparity or 2 shapes x 2 directions of disparity) were presented four times for stereograms a and b, and twice for stereogram c in random order. 3.2 Results and discussion The subjects in the experimental group reported several depth percepts which were different from those of the control group: (i) a percept which included a reversed relationship between the direction of perceived depth and the direction of binocular disparity; (ii) the perception of a plane surface; (iii) the perception of the sudden alternation of normal and reversed depth directions; (iv) the perception of the slant in depth of line segments in stereograms a and b or a rectangle in stereogram c; and (v) the perception of depth of some magnitude but without a clear direction. These percepts frequently occurred in this order.
M Ichikawa, H Egusa
976
The frequencies of reversed-depth perception, ie percept (i), were summed across two levels of disparity or two shapes of rectangle because there were no evident differences in the frequency of reversed-depth perception between these conditions. These total frequencies are plotted, as a function of the number of wearing days for each subject, in figures 3 and 4. Reversed-depth perception occurred not only during observation of LCSs, but also during observation of RDSs, in contrast to Shimojo and Nakajima's (1981) results. From binomial tests, in each test session for each subject with eight observations, less than one and more than seven occurrences of reverseddepth perception show statistically significant tendencies towards the normal and reversed relations between binocular disparity and direction of perceived depth, respectively (p < 0.05). Generally, reversed-depth perception tended to occur with high frequency early in the wearing period (for example, it was significant for subject KW with the uncrossed stereogram a during the 4th day), and it often decreased late in the wearing period for any stereogram type, unlike the results of Shimojo and Nakajima (1981). After removal of the reversing spectacles, reverseddepth perception persisted for several days for subjects H H , SW, and HY. There were salient individual variations in the conditions under which reverseddepth perception occurred. Subjects K W and H Y experienced reversed depth in all stereograms. Subjects H H and SW reported reversed depth mainly in stereograms a and b. Subject MI perceived reversed depth chiefly in stereogram c. T h e frequency of reversed-depth perception during the wearing period was significantly higher with uncrossed than with crossed disparities, for stereogram a for subjects SW {x2 = 6.92, p < 0.01) and H Y [%2 = 16.55, p < 0.005), and stereogram b for subjects K W (x2 = 11.11, p < 0.005) and H Y (x2 = 56.72, p < 0.005). Subject KW
Subject HH
61 § 4 & 2 | 0 lei
i *T3
C3
8
•
4
•
2
•
0
•
•
6
:
4 '• 2 •
0
• o»o»o*
b
•
6
8-
a
:
o
: ° : • * •i : ojo*
•
: °!:°
° ° * o**i««o»«o»o»o«o b
8 •
o
6 •
• • o
• • •
• o
o o o o
o
4 •
•
•
•
•
o«
•
o o o»o»
2 • o» 0
• •
• •
o o
• •
• •
o » o o o o o o o
S3
cr £ 4 2 0
4 2 io»o» o o o o o» o»o»
0 i o.i;0»
• O* 0» OtjO* 0» O* 0» O* 0» 0» 0»
2 6 10 8 16 32 107 6 10 8 16 pre4 8 1 9 pre4 8 1 9 24 51 post-wearing wearing wearing wearing wearing post-wearing period period period period period period Time of testing/days Figure 3. Frequencies of reversed-depth perception in test 1. Frequencies of convex perception for the stereogram with an uncrossed disparity (open circles) and concave perception for the stereogram with crossed disparity (filled circles) for stereograms a, b, and c, for subjects KW and HH. The two horizontal lines in each panel indicate the statistically significant frequencies of reversal (top) or normal (below) directional relation between disparity and perceived depth. 2
977
Effects of retinal-image reversal on depth perception
!
8 •
o
:
•
•: o :•
a¥
o
6 •
•
4 •
•
8•
jo
6 •
:
! °
o
0 •
J
4 •
So* o 2 •
< c¥
I b¥
o o
0
•• • o •
•
2 •
•o o o
0 •
• o»j*
•
•
•
•
o
o o
1 °* •
• •
o»o» o»o> o»o» o»
1 1 '
bR
aR
8 6
:
4
jo
o o
: o
;
2 2
5 3
8•
10 11 8 4 17
:
!
i
o
a¥
•
o: : • jo» o» • o» o»-o» o» o»
m 1 •1
° • o » o » • o» o» cm cm o
2
5 3
(b)
10 11 8 4 17
:
•I 1 1 •I •J •
• oj : ° : :° : j • • J o i • :• • o
5 3
8 •
8
!• •
4 2 • 0 •
b
10 11 4 17 ccF
1 • o»-
6 •
•
o
i
o
i
: • :o o j cm
: o :o ! • • o»
cR
:
8 •
o
•
m
•
C
• j
6 •
*
o 4 • 0
2
• o
? 0 ;
• !o
2 bV
i ° °
8
o » j « o» o> o* o»;ot o» cm
10 11 8 4 17
1
6 4
]
!o o o
•I o ! •
t aR
o
5 3
4 •
0 •
; •
2
6•
7 •
i • • i
o»l» • • •
0 ]
cR
0»
•
•
2
• • 5
3
o • o» • o
: o o j
:
o:
• 0 o» O O
10 11 8 4 17 b¥
aF
2
5 3
8
;0»
o
0»
10 11 8 4 17 cF
••
6 4 1
4 •
2
2 •
0
• cm cm cm cm]
cm-cm cm cm cm cm:cm i • i
i
i
i
0 •
;•
j
•
•
•
o»|o o o o
;
ojo*
i • i
aR
bR
cR
8
•:
6 4
41 2 0
•j
•
? • • o» • cm cm; • * i
i
2
cm-.cmcm o o » o » | o »
i
5
10
2
5
10
0
:•
•
o s j o cm o o i
2
5
i
ojo» i * i
10
3 8 1 3 8 1 preprewearing wearing post-wearing post-wearing wearing wearing post-wearing period period period period period period period Time of testing/days Figure 4. Frequencies of reversed depth perception in test 1 for subjects SW(a), H Y ( b ) , and MI (c) under conditions of free observation (F) and restricted observation (R). T h e frequencies are of convex perception for the stereogram with an uncrossed disparity (open circles) and concave perception for the stereogram with crossed disparity (filled circles). T h e two horizontal lines in each panel indicate the statistically significant frequencies of reversal (top) or normal (below) directional relation between disparity and perceived depth. 3
prewearing wearing period period
aF
aR ? 1 I 4 (a)
4
4 *
2 3 5 8 10 4 1117
1 15
2 3 5 8 10 4 1117
2 3 5 8 10 4 1117
2 3 5 8 10 4 1117
2 3 5 8 10 4 1117
aF
a a 73
> aR : * « A A
(b)
AA
AA
2 3 5 8 1 0 4 1117
cF
10 aF
A
*A
*
A
*
cR
10 i aR 8
51
A
2 3 5 8 10 1
2 3 5 8 10 1
2 3 5 8 10 1
(C)
prewearing period
prepost prepost wearing wearing wearing wearing wearing wearing period period period period period period Time of testing/days Figure 5. Mean estimated depth magnitudes in test 1 for subjects SW(a), HY(b), and MI (c), under conditions of free observation (F) and restricted observation (R) of stereograms: a with disparity of 20 min uncrossed (open circles) or crossed (filled circles), or disparity of 10 min uncrossed (open triangles) or crossed (filled triangles); b with 40 min disparity uncrossed (open circles) or crossed (filled circles), or disparity of 10 mm uncrossed (open triangles) or crossed (filled triangles); and c with uncrossed (open circles) or crossed (filled circles) disparity. wearing period
post wearing period
Effects of retinal-image reversal on depth perception
979
But, it was significantly lower with uncrossed than with crossed disparities for stereogram c for subjects KW [%2 = 6.49, p < 0.05), SW (x2 = 4.48, p < 0.05), and HY (X2 = 34.36,p < 0.005). Although reversed-depth perception occurred in both observation conditions, the frequency of reversed-depth perception during the wearing period was significantly higher with free observation than with restricted observation for stereograms a (X2 = 21.20, p < 0.005) and b {x2 = 4.48, p < 0.05) for subject SW, and stereogram c for subject HY (%2 = 7.25, p < 0.01). These results imply that restricting eye movements by fixation on a point can interrupt the occurrence of reversed-depth perception. Figure 5 shows the mean estimated magnitude of the depth in the normal direction for each subject. Each data point in the stereogram c panel represents the mean for the two shapes of rectangle. For stereogram a, no subject showed any consistent changes in perceived-depth magnitude with increase in wearing time. On the other hand, the magnitude of perceived depth in stereograms b and c tended to reduce with increasing wearing time for subjects MI and HY. These individual variations did not correspond with those in the direction of perceived depth. This asynchrony of the direction and the magnitude of perceived depth indicates that the visual transformation which reverses the direction of disparity interferes with the perception of depth direction and that of depth magnitude separately. The aforementioned reversed-depth perception and reduction of perceived-depth magnitude shown by the experimental subjects did not occur for the five control subjects. Hence, these changes cannot be attributed to the repetition of tests. 4 Test 2 In test 2 we investigated the effects of long-term reversed vision on the depth perception produced by several concurrent depth cues and the suggested adaptive changes in the process integrating depth information from these cues. To this end, in the experimental setup binocular disparity conflicted with linear perspective or occlusion. 4.1 Method 4.1.1 Apparatus and stimuli. Two luminous objects were presented binocularly at the level of the subject's eyes in a completely dark room. The right hand object was the standard stimulus, Ss, and was placed 3 m from the subject. The left hand object was the comparison stimulus, Sc, and was placed 2.5, 2.8, 3.2, or 3.5 m from the subject (seefigure 6). comparison stimulus standard stimulus
"3.5 m
subject Figure 6. Apparatus used in condition (ii) in test 2.
980
M Ichikawa, H Egusa
There were three stimulus conditions. In condition (i), the two luminous stimuli used as the Ss and Sc were horizontal cylinders of 5 mm diameter and 6 mm length coated with luminous paint for subjects KW and HH, and light-emitting diodes of 4 mm diameter for the remaining subjects. The Ss and Sc were presented 5 cm from the median plane. Binocular disparity was available as a cue in this condition. In condition (ii), a board 180 cm long and 50 cm wide was positioned horizontally 30 cm below the same Ss and Sc as in condition (i). There were two rows of eighteen luminous discs of 2 cm diameter on the board with an interdisc separation of 10 cm and interrow separation of 50 cm. In this condition, binocular disparity and linear perspective were available as cues. In condition (iii), the Ss was a square card whose sides were 10 cm long, and the Sc an equilateral triangle with 10 cm sides. The centres of both the Ss and the Sc were presented at 4 cm from the median plane so that one appeared to overlap the other. In this condition, binocular disparity and occlusion were available as cues. Nothing was visible to the subject except for the aforementioned objects. 4.1.2 Procedure. The subject positioned his head on a chinrest and viewed the stimulus. He reported whether the Sc appeared nearer than the Ss or not. The subjects in the experimental group viewed the stimuli through left-right reversing spectacles during wearing sessions and without the spectacles in the post-wearing sessions. The single control subject viewed the stimuli through the reversing spectacles condition (i)
• •
condition (ii) •
«o
o
•
condition (iii^ •
1
o
•
-1 J
o o
• o
•
«0
KW
o
o o
"11 •°
• o
4
o
•
2
0
so 8•
•
•
j «o
1
6 |
•
4 •
4 I
o
•:
HH
•o
HO
o
2 |
2 •
0•
«o «o
*°
• • • •
8
6 •
•
• o
«o «o
\
0 1 *°
•0
•o
•0
• o
6
• •0
•
1
o •
o
o •
o
o
1
3 5 10 11 17
o
wearing period
: o i
post wearing period
0]
o
o
o
1 3 5 10 11 17 post wearing wearing period period Time of testing/days
3 5 10 11 17 post wearing wearing period period
Figure 7. Frequencies of veridical depth perception in test 2 in conditions (i), (ii), and (iii) for the six experimental subjects (KW, HH, HO, SW, MI, and HY, and the control subject KK. Open symbols are judgments "Sc further" and filled symbols are judgments "Sc nearer". The two horizontal lines in each panel indicate the statistically significant frequencies of veridical (top) or nonveridical (lower) depth perception.
981
Effects of retinal-image reversal on depth perception
in every session. The three conditions were given in numerical order. The Sc were presented four times at each of the four positions in random order in each condition. 4.2 Results and discussion Figure 7 shows the frequencies of veridical judgments—judgments that correspond with the physical depth relation between the Ss and the Sc. Frequencies of veridical judgments for the Sc at 2.5 and 2.8 m and those for the Sc at 3.2 and 3.5 m were summed in "Sc nearer" and "Sc further" respectively. From binomial tests, less than one and more than seven occurrences of veridical judgments in each session of each subject show a statistically significant tendency towards nonveridical and veridical depth perception, respectively (p < 0.05). Subjects KW and HH showed a tendency for the frequency of veridical judgments with left-right reversed vision to increase with wearing time for conditions (i) and (ii). Such a tendency was also shown for condition (ii) by subject SW, and for all three conditions by subject ML Subjects HY and HO responded veridically only under condition (iii). For the control subject, KK, the frequency of veridical response remained in conditions (i) and (ii). Therefore, the increase in veridical judgment shown by the subjects of the experimental group in conditions (i) and (ii) is not due to repetition of the tests.
•
•
•
6
o
o
0 8
o
o
•
•o
•
8 •
•
4 2 1
condition (iii)
condition (ii)
condition (i)
o
•
• o
«o
O
6
6 •
4 •
•
2 •
0
• • • •
8 •
o
0
•
o
4 •
o
2•
0
0•
MI
•
6
SW
0
o
o o
4 o
2
o
0
•
8
•
o •
•
0
o
•
•o
•
•
5
8
10
HY
6 4
•
0i
•
•o
2 0
o
•0
1
3
5
•o 8
10
•
o
o
o
•o
o
•
1
3
5
8
10
1
wearing period
wearing period 8 •
8•
8 •
6 •
6
6
4•
4•
4 •
• •
2 • 0 •
o
1 2
o
o
«o «o
3 4
Figure 7 (continued)
5
0 •
o
o
• • •O
O
O
• 90
O
1 2 3 4 5 Time of testing /days
0
o
o
o
•
•
•
•
2•
2 •
•
3
wearing period
• 1 2
3 4
5
KK
982
M Ichikawa, H Egusa
5 General discussion 5.1 Changes in binocular stereopsis Some authors have proposed that binocular stereopsis is divided into processes tuned to different spatial frequencies (eg Bishop and Henry 1971; Frisby and Mayhew 1977; Julesz and Miller 1975; Marr 1982; Marr and Poggio 1979). This notion distinguishes a process tuned to low spatial frequencies which participates in the perception of LCSs and a process tuned to high spatial frequencies which participates in the perception of RDSs. Shimojo and Nakajima (1981) argued that the wearing of reversing spectacles influences only stereopsis tuned to low spatial frequency because they found the reversal of depth perception only for LCSs. However, the results in test 1 in the present study showed that left-right reversed vision could change the direction not only of low-frequency-tuned stereopsis but also that of high-frequencytuned stereopsis. This indicates that the high-frequency-tuned stereopsis is not independent of the experience of visual transformation. There were salient individual variations in the kinds of stereogram for which depth reversal occurred. Our subjects HH, SW, and MI, and the subjects of Shimojo and Nakajima (1981), reported reversed depth only for either LCSs or RDSs, and our remaining subjects in the experimental group reported reversed depth for both LCSs and RDSs. These individual variations are consistent with the notion that binocular stereopsis is divided into separate processes tuned to different spatial frequencies, and indicate that these processes are influenced separately by visual transformation. Moreover, the discrepancies of changes in perceptions of depth direction and depth magnitude between stereograms a and b suggest that pictorial depth cues, for example, line-size in stereogram a and linear perspective in stereogram b, may interfere with the directional and quantitative functions of low-spatial-frequency-tuned stereopsis. The results of some studies indicate that crossed and uncrossed disparity are processed by different mechanisms (eg Regan and Beverley 1973; Richards 1971). Mustillo (1985) maintained that crossed disparities are processed more readily and more efficiently than are uncrossed ones. Our results in test 1 showed that the reversed-depth perception appeared more frequently for LCSs with uncrossed disparities than for LCSs with crossed disparities, but the opposite tendency appeared for RDSs. This result suggests that the mechanisms processing different disparity directions in high-frequency-tuned stereopsis are organized in a different way from those in low-frequency-tuned stereopsis. Shimojo and Nakajima (1981) assumed that the motor-fusion process governed by eye movements corresponds to the adaptively changeable subprocess of binocular stereopsis participating in the perception of LCSs. Certainly, it was confirmed that restricting eye movements by fixation on a point did influence the frequencies of reversed-depth perception in our test 1. This suggests that the changes in the eyemovement process are responsible for the directional changes of binocular stereopsis. We also found evidence, in the results from test 1, for another cause of reverseddepth perception in stereogram observation: the sudden alternation of normal with reversed-depth perception, [percept (iii)], for RDSs. Fender and Julesz (1967) demonstrated that the depth impression from an RDS was not much influenced by eye movements because of cortical registration. So, the alternation of normal and reversed depth perception for RDSs suggests that the reversed depth perception may be caused not only by the changes of the oculomotor process but also by the modification of the cortical process which registers and operates on binocular-disparity information. The possibility of such modification of cortical processes with experience has been supported by some neurophysiological studies (eg Bruce et al 1981).
Effects of retinal-image reversal on depth perception
983
5.2 Changes in the rank of depth cues Our results show that the directional changes of depth perception do not occur concurrently in the two tests. For example, in test 2, the veridical response under leftright reversed vision, shown by subject KW, increased with the number of spectaclewearing days, and its frequency rose to over 85% of all responses on the 10th day of the wearing period. But, in test 1, the frequencies of reversed-depth perception initially increased but then decreased. This decrease in perception of reversed-depth in the stereograms indicates that, under left-right reversed vision, the binocular disparity cue became nonveridical again. The discrepancies between the results of the two tests indicate that the changes of depth perception due to multiple depth cues cannot be explained exclusively in terms of changes of binocular stereopsis. Therefore, we must discuss the changes in the relationship between binocular stereopsis and other depth cues independently of the changes within binocular stereopsis. We assume that the left-right reversal of retinal images changes not only binocular stereopsis but also processes integrating depth cues so that the salience of binocular disparity cues as determinants of depth perception falls and that of pictorial cues rises. This assumption can explain the discrepancy between the directional change of perceived depth in test 1 and test 2. To complete veridical depth perception when the optical transformation reverses the direction of disparity, it is necessary to reverse the directional relation between disparity and perceived depth, because the salience of binocular stereopsis in the integration process is high early in the wearing period. But, because the salience of stereopsis is reduced later in the wearing period, the visual system ignores the binocular disparity cue and produces veridical depth perception using other depth cues. Consequently, the reversed stereopsis does not need to be maintained, and vanishes. Moreover, this interpretation is supported by two observations from daily life reported by subjects in the experimental group. First, although the reversed (nonveridical) depth impression often occurred for objects which present unambiguous disparity cues (eg branches of trees, legs of a table, and poles on the roadside) early in the wearing period, such impressions decreased as wearing-time increased in spite of the imperfection of direction reversal seen in the stereograms. Second, exaggerated depth impressions were reported during binocular observation of two-dimensional pictures during the wearing period. These impressions did not occur with monocular observation. Moreover, immediately after the removal of the reversing spectacles, at the end of the wearing period, the subjects perceived unnaturally exaggerated depth. This impression lasted for some hours. These exaggerated depth impressions suggest that transformed vision raised the sensitivities of binocular processes operating on pictorial cues, and that the relative salience of depth cues and the sensitivity to each cue when adapted to reversed vision was not suited to normal vision. In summary, this study has added the following new knowledge about the effects of retinal-image reversal on binocular depth perception. First, the reversal of retinal images causes reversed depth perception in LCSs and RDSs. Second, the results of the present two tests suggest that the conflict between binocular disparity and other depth cues changes the process integrating depth information from multiple depth cues so that the weight of binocular disparity in depth perception falls. In conclusion, the human depth-perception system is very flexible, and the relationship between stimulus information and perceived depth is constantly updated to adapt to the environment. This updating seems attributable not only to the eye-movement system but also to cortical processes.
984
M Ichikawa, H Egusa
Acknowledgements. This work was supported by the Japanese Ministry of Education, Science, and Culture Grant-in-Aid for Scientific Research No 61301016 to Professor Takayuki Mori of Chukyo University (chair). We wish to thank Professor Takara Tashiro of Osaka City University and Miss Kaoru Sekiyama of Kanazawa University for practical help with this study, and Professor Tadasu Oyama of Nihon University for his helpful comments on an earlier version of this article. We also thank Mr Hiroyuki Horikawa, Mr Hiroyuki Oishi, Mr Kouichi Watanabe, Mr Susumu Watanabe, and Mr Hisashi Yoshida who participated in this study as experimental subjects. References Bishop P O, Henry G H, 1971 "Spatial vision" Annual Reviews ofPsychology 22 119 - 1 6 0 Brown G G, 1928 "Perception of depth with disoriented vision" British Journal of Psychology 19 117-146 Bruce C J, IsleyMR, ShinkmanPG, 1981 "Visual experience and development of interocular orientation disparity in visual cortex" Journal of Neurophysiology 46 215 - 228 Epstein W, 1975 "Recalibration by pairing: a process of perceptual learning" Perception 4 59-72 Fender D H , Julesz B, 1967 "Extension on Panum's fusional area in binocularly stabilized vision" Journal of the Optical Society ofAmerica 57 819-830 Fisher S K, Ciuffreda K J, 1990 "Adaptation to optically-increased interocular separation under naturalistic viewing conditions" Perception 1 9 1 7 1 - 1 8 0 Fisher S K, Ebenholtz S M, 1986 "Does perceptual adaptation to telestereoscopically enhanced depth depend on the recalibration of binocular disparity?" Perception & Psychophysics 40 101-109 FrisbyJP, MayhewJEW, 1977 "Global processes in stereopsis: some comments on Ramachandran and Nelson (1976)" Perception 6 195 - 206 Julesz B, 1971 Foundations of Cyclopean Perception (Chicago, IL: University of Chicago Press) Julesz B, Miller J E, 1975 "Independent spatial-frequency-tuned channels in binocular fusion and rivalry" Perception 4 1 2 5 - 1 4 3 MarrD, 1982 Vision: A Computational Investigation into the Human Representation and Processing of Visual Information (New York: W H Freeman) MarrD, Poggio T, 1979 "A computational theory of human stereo vision" Proceedings of the Royal Society ofLondon Series 5 204 301-328 Mustillo P, 1985 "Binocular mechanisms mediating crossed and uncrossed stereopsis" Psychological Bulletin 97 187-201 O'LearyA, WallachH, 1980 "Adaptation in stereoscopic depth constancy" Perception & Psychophysics 27 403 - 408 Regan D, Beverley K I, 1973 "Disparity detectors in human depth perception: Evidence for directional selectivity" Science 181 8 7 7 - 8 7 9 Richards W, 1971 "Anomalous stereoscopic depth perception" Journal of the Optical Society of America 61 410-414 Shimojo S, NakajimaY, 1981 "Adaptation to the reversal of binocular depth cues: effects of wearing left-right reversing spectacles on stereoscopic depth perception" Perception 10 391-402 StrattonGM, 1898 "A mirror pseudoscope and the limit of visible depth" Psychological Review 2 632-638 WallachH, Moore M E, Davidson L, 1963 "Modification of stereoscopic depth perception and the kinetic depth" American Journal of Psychology 76 4 2 9 - 4 3 5
p
© 1993 a Pion publication printed in Great Britain
