Visual Neuroscience (1992), 9, 303-312. Printed in the USA. Copyright © 1992 Cambridge University Press 0952-5238/92 $5.00 + .00

Binocular depth perception following early experience with interocular torsional disparity

PAUL G. SHINKMAN,1 BRIAN TIMNEY,2 AND MICHAEL R. ISLEY1 1 2

Department of Psychology and Brain and Development Research Center, University of North Carolina, Chapel Hill Department of Psychology, University of Western Ontario, London, Canada

(RECEIVED June 3, 1991; ACCEPTED January 28, 1992)

Abstract

The relationship between the behavioral and physiological consequences of rearing with optically induced cyclotropia was assessed. Beginning at the age of 4 weeks, kittens wore goggles that rotated the visual field in opposite directions in each eye for several hours each day over a period of several weeks. The amounts of interocular rotation were 0 deg (control), 16 deg, and 32 deg. Subsequently, they were tested to determine their monocular and binocular depth thresholds and, in some cases, visual acuity. In several kittens recordings were also made from the visual cortex. Binocular performance of all kittens in the 0-deg condition and three out of six kittens in the 16-deg condition was comparable to, although slightly lower than, that of normally reared kittens. In contrast, none of the 32-deg kittens showed any evidence of the binocular superiority that would suggest the presence of stereopsis. Extracellular unit recordings from the visual cortex confirmed our earlier results with goggle-reared kittens. In 16-deg kittens, the distribution of the cells' preferred interocular disparities (IOD) in receptive-field orientation showed a compensating shift so that the mean matched the experienced rotational disparity. In the 32-deg kittens, binocularity was greatly disrupted and there was no compensatory shift in the IOD distribution. Two 32-deg kittens were afforded 3 years of subsequent normal visual experience. Both the behavioral and the physiological findings were unaffected by normal visual exposure in adulthood. Control measurements of acuity indicated that any deficits in depth perception were not due to reduced spatial-resolution abilities. The data indicate that the kitten visual system is able to maintain functional binocularity sufficient to subserve a moderate level of stereoacuity with interocular rotations of up to at least 16 deg. Keywords: Binocular depth perception, Cyclotropia, Stereopsis, Visual cortex, Visual development

Introduction The quality and quantity of visual experience an animal receives early in life may have a profound effect on its later visual development. From the earliest studies (Hubel & Wiesel, 1970; Wiesel & Hubel, 1965), it has been known that even brief periods of monocular occlusion lead to a reduction in the proportion of neurons serving the deprived eye. Similarly, if the input to the two eyes is discordant, as occurs in strabismus, there is a reduction in the number of neurons that have binocular connections (Hubel & Wiesel, 1965). Previously, we have reported on the cortical physiology of kittens reared with torsionally disparate visual input to the two eyes (Bruce et al., 1981; Isley et al., 1990; Podell et al., 1982; Shinkman & Bruce, 1977; Shinkman et al., 1983a). Between the ages of 1 and 3 months, kittens wore prism goggles that rotated the visual field in opposite directions for the two eyes. They viewed a normal environment through the goggles for several hours daily and spent the rest of the time in the dark. At the end Reprint requests to: Paul G. Shinkman, Department of Psychology, CB 3270, University of North Carolina, Chapel Hill, NC 27599-3270, USA. 303

of the rearing period the physiological organization of the visual cortex was assessed on a variety of measures, including the distribution of binocular cells' preferred interocular orientation disparities. These experiments showed that after early experience with a relatively small interocular field rotation (8 deg in each eye for a total interocular orientation disparity of 16 deg), visual cortical physiology was essentially normal with respect to binocularity, orientation specificity, the distribution of receptive-field types, and an orderly columnar arrangement. The effects of the small interocular rotational disparity were evident, however, in the distribution of binocular cells' preferred interocular disparities in stimulus orientation: The mean was shifted away from 0 deg and corresponded to the experienced rotation of 16 deg. Kittens that experienced an interocular orientation disparity of intermediate magnitude (24 deg) had normal ocular-dominance distributions and showed a partial but incomplete shift in the distribution of preferred interocular disparities of stimulus orientation. This rearing condition also had more generalized effects on visual cortical physiology than were seen in kittens that wore 16-deg goggles, although the differences were not great. In contrast, binocularity was severely disrupted in kittens

P.G. Shinkman, B. Timney, and M.R. Isley

304 that experienced a large interocular rotational disparity (32 deg). These subjects showed greater variability in the distributions of cells' interocular orientation disparities, and there was no evidence of a compensatory shift in the means of these distributions. Also, contralaterally dominated cells with preferred stimulus orientations near horizontal or vertical were overrepresented in the visual cortices of these kittens (Podell et al., 1982; Shinkman et al., 1985). Thus, the limiting interocular orientation disparity for the development of matched orientation requirements in binocular cells' receptive fields in the two eyes appears to lie between 16-24 deg, while for the development of normal binocularity as reflected in the ocular-dominance distribution, the limiting disparity falls between 24-32 deg. All of the physiological effects described above were permanent and not subject to reversal by several months of normal visual experience without the goggles after the end of the rearing period (Isley et al., 1979; Shinkman et al., 1985). These results were interpreted in terms of permanent changes in the synaptic organization of the visual cortex consequent upon early experience with rotationally disparate visual input to the two eyes. The role of binocular neurons in the maintenance of stereopsis is well documented. Blake and Hirsch (1975) reported that kittens that had been subjected to a period of alternating monocular deprivation in the first few months of life showed deficits in binocular depth perception that were correlated with a reduction in the proportion of cortical binocular neurons. Related findings were obtained by Packwood and Gordon (1975), who compared normally reared cats, Siamese cats with a substantial proportion of crossed fibers from the temporal retina and a corresponding reduction in cortical binocularity, and cats reared with alternating monocular deprivation. They reported evidence for stereopsis in the normal cats but none in the Siamese or deprived cats. More recently, Altmann et al. (1987) tested monocular and binocular depth perception of kittens that had been reared with rapidly alternating monocular occlusion. Altmann et al. used the jumping-stand technique (Mitchell et al., 1979). They reported that kittens raised with alternation rates greater than 500 ms showed a deterioration in binocular depth thresholds to the level of the monocular thresholds. These kittens also failed to show evidence of binocular summation in pupillary responsiveness. Single-unit recordings were made and indices of cortical binocularity were compared with the ratio of monocular and binocular depth thresholds. A significant correlation of 0.87 was found between the physiological and behavioral measures. Although a loss of binocular neurons is likely to result in a disruption of stereopsis, the presence of binocular neurons does not necessarily guarantee functional stereopsis. For example, Kaye et al. (1982) raised kittens in total darkness for the first 4 months of life. They found that although some of the animals were orthophoric and had a large proportion of binocular neurons, these kittens showed no evidence for stereoscopic depth perception. Thus, it seems reasonable in the present instance to ask whether the binocularity observed following rearing with small rotations of the visual field might be adequate to subserve binocular depth perception. Specifically, we wished to determine whether the relatively normal organization of the visual cortex in the 16-deg kittens would be reflected in correspondingly good binocular depth perception, and whether the disruption of physiological binocularity produced by early experience with 32-deg

goggles would in turn correlate with deficits in binocular depth perception. Preliminary reports in abstract form have been made of portions of these findings (Shinkman, 1983; Shinkman et al., 1986o,b\ Shinkman et al., 19836). Methods

Kittens Sixteen kittens were used, drawn from nine litters born and reared in the Psychology Department colony at Chapel Hill. Following completion of the rearing procedures (see below), they were shipped to the University of Western Ontario for behavioral testing. After this stage of the experiment, they were returned to North Carolina where electrophysiological recordings were made. Rearing procedures The methods have been described in detail previously (Bruce et al., 1981). Mothers and their litters were transferred to a dark room during the first postnatal week before eye opening. Beginning at 28 days of age and continuing for 8 weeks, kittens were given 2-4 h of visual experience each day, viewing a normal environment while wearing goggles fitted with prisms that rotated the images in the two eyes about the visual axes. The rotations, which were always equal but opposite in the two eyes, were achieved by cementing together two right-angle prisms in each goggle eyepiece. While they were wearing the goggles, the kittens were kept in a well-illuminated room containing toys and obstacles suitable for climbing and jumping, in the presence of littermates and an experimenter. Kittens spent the rest of the time in the dark. There were three experimental groups. Two were used to test the effects of small and large rotations. The third group was intended to examine potential recovery consequent upon normal visual experience following the goggle rearing. The eight kittens in the 16-deg group wore goggles that imposed 8 deg of image rotation in each eye. For five of these kittens the rotations were counterclockwise in the left eye and clockwise in the right eye (the + 16-deg condition), and for three kittens the rotations were clockwise in the left eye and counterclockwise in the right eye (the —16-deg condition). There were two kittens in the 32-deg group, one in the +32-deg condition, and one in the —32-deg condition. Finally, two kittens in the 32-deg + N group received the same treatment as kittens in the 32-deg group: one +32-deg and one -32-deg. At the end of the period of goggle exposure, these kittens were tested neurophysiologically as described below, and then returned to the main colony where they had normal visual experience without goggles. After 6-8 months, they were retested neurophysiologically, and after 3 years of normal visual exposure they were tested behaviorally as described below. The four kittens in the 0-deg or control group received the same treatment as kittens in the 16- and 32-deg groups, but wore goggles whose prisms introduced no torsional disparity between the left and right eyes' visual fields. Behavioral assessment Two classes of visual function were measured in these animals. All of the kittens had their binocular and monocular depth threshold assessed and, for some, measurements of visual acu-

Stereopsis in cats following cyclotropic rearing ity were also made. Testing was conducted using a blind procedure: workers at Western Ontario were unaware of the rearing conditions of kittens they tested. Depth thresholds were measured using a jumping-stand apparatus which has been described in detail elsewhere (Mitchell et al., 1979). The jumping stand permits two-choice discrimination tasks to be carried out and allows psychophysical thresholds to be obtained. The apparatus consists of a box with a transparent Plexiglas top which serves as a landing surface for kittens when they jump. Inside the box is a balance beam, on whose arms are placed sheets of translucent acetate covered with a random array of dots of different sizes. These dot surfaces may be moved up and down so that one or the other rests against the Plexiglas landing surface. The separation between the dot surfaces may be varied by moving the whole balance beam up and down. To minimize the number of available monocular cues to the position of the targets, an opaque mask may be placed beneath the Plexiglas sheet such that the kitten's view of the targets is restricted to an aperture 14 x 19 cm. Kittens were required to jump from a raised platform towards the closer of the two surfaces. If they were correct they were rewarded with a small amount of commercial baby food and social reinforcement. An incorrect choice was signaled by either a loud, high-frequency tone, or a loud burst of white noise. Kittens were trained first with both eyes open. Once they had mastered the task, they were retrained with an opaque scleral occluder preventing vision in one eye. For some cats, different eyes were tested on alternate days, for others the same eye was tested repeatedly. No differences in performance were observed between cats that were tested with different monocular schedules and all monocular data were pooled in the final summaries. After initial training, preliminary threshold estimates were obtained both binocularly and monocularly using a modified method of limits. The cats were run in blocks of five or ten trials. If they obtained at least 70% correct the separation was reduced for the next trial block. This was continued until performance fell below 70% correct and the series was started again. Threshold was taken as the smallest separation at which the cat could discriminate consistently with 70% accuracy. Binocular and monocular training for all cats took place over a period of several months before final threshold measurements were obtained. Final thresholds were obtained using a method of constant stimuli. Five separations were chosen to bracket the threshold estimate obtained using the method of limits. The cats were then tested in blocks of five trials at a given separation with a different separation chosen randomly for each block. This procedure was repeated over a number of sessions until a minimum of 50 trials had been obtained for each separation. Occasionally, if performance extended beyond the range of separations chosen at first, the range was shifted up or down to provide data points at additional separations. Typically, final binocular thresholds were obtained first and then monocular thresholds. In some instances, if the cats performed particularly well monocularly, several additional binocular sessions were run. For visual-acuity tests, the stimuli consisted of a set of cards containing high-contrast square-wave gratings. The spatial frequency of the gratings was varied by changing the cards. Another card containing a very high spatial-frequency grating (> 12 cycle/deg) that was beyond the ability of the kittens to resolve was used as the negative stimulus. The space-averaged luminance of the gratings was matched at 80 cd/m 2 so that they

305 were indistinguishable on the basis of differential brightness. The psychophysical procedure was similar to that used for testing depth thresholds. Pupillography Photographs were made of the kittens' eyes to measure the angle formed between the pupils, constricted either by topically applied ophthalmic pilocarpine HC1 or by sunlight. Pictures were taken at the beginning and end of the period of prism goggle experience, at the end of the period of behavioral assessment, and at the beginning of electrophysiological recording sessions before and after Flaxedil paralysis. Several pictures were made on each occasion, and those yielding well-constricted pupils with appropriately centered corneal reflexes were used to estimate interpupillary angles. Vergence ratio {VR) was calculated as the ratio of the distance between corneal reflexes (ICRD) to the distance between the centers of the two pupils (IPD), a dimensionless quantity comparable across kittens regardless of magnification differences. A limitation of the photographic technique is that it does not control for variations in angle kappa, the angle between the optical and visual axes. In young kittens this poses significant difficulties because it appears that angle kappa changes systematically with age, leveling off after about 10 weeks (Olson & Freeman, 1978). In older cats there are some individual differences in angle kappa which means that pupillary convergence is not a certain indicator of a misalignment of the visual axes. In normally reared kittens, VR is somewhat variable, with an average value of 0.95 (Von Griinau, 1979). Divergent and convergent strabismus are thus represented in the kitten by values of VR less than or greater than 0.95. Recording procedures Standard methods were used for extracellular single-unit recording (Bruce et al., 1981). Kittens were prepared under surgical anesthesia; following ether induction the kitten was intubated with an endotracheal tube coated with a long-lasting local anesthetic (Zyljectin [discontinued]), placed on a heating pad, and secured in a stereotaxic instrument. A small craniotomy was made 2.5-4.0 mm posterior to instrument zero and 1.0-1.5 mm lateral to the midline. The kitten was then repositioned in the stereotaxic instrument using a small anteriorly mounted atraumatic head holder so as to be firmly held without pressure points or discomfort and with an unobstructed frontal view. Wound margins were infiltrated with Zyljectin. A solution of one part gallamine triethiodide (Flaxedil, 20 mg/ml) and two parts of 5% dextrose in 0.9% saline was administered during the remainder of the session. The kitten was respired with a 70/30% N 2 O/O 2 mixture. Electrocardiogram, rectal temperature, and end-tidal CO 2 were monitored and maintained in the normal range. Ophthalmic atropine sulfate was used to relax accommodation and dilate the pupils, the corneas were anesthetized, and ophthalmic phenylephrine HC1 was applied to retract the nictitating membranes. The eyes were focused on a tangent screen using corneal contact lenses, and the optic disk, its surrounding blood vessels, and the area centralis were mapped separately for each eye and marked on a plastic overlay on the tangent screen. Artificial pupils were positioned directly in front of each contact lens so that the area centralis remained clearly visible through the ophthalmoscope.

P.G. Shinkman, B. Timney, and M.R. Isley

306 Glass micropipettes were used to record the activity of neurons in the visual cortex. When a cell was isolated, determinations of the optimal receptive-field characteristics were made for each eye separately with the other eye occluded; each eye was retested at least once before proceeding to another cell. To test the reliability of the measurements of receptive-field orientation, a small number of cells were studied using a blind procedure. One experimenter made a determination of the cell's preferred orientation, after which a second experimenter rotated a prism interposed between the eye and the tangent screen so as to introduce a rotation of the retinal image whose amount and direction were unknown to the first experimenter, who then repeated the determination of preferred receptive-field orientation. The difference between the two measurements was compared to the alteration in retinal image orientation produced by the prism rotation. Three experimenters repeated this procedure in pairs. The results were always very close, and because the procedure greatly increased the time required to study a cell, only a few cells were tested in this fashion. At the end of semichronic experiments the scalp was sutured and Flaxedil was discontinued. The kitten was given local anesthetics and antibiotics, and returned to the dark room following recovery to preclude any effects of subsequent normal visual experience; antibiotic treatments, both local and systemic, were continued for several days. At the end of acute experiments, the animal was given an overdose of Nembutal intravenously and then perfused with 0.9% saline followed by 10% Formalin. Gyral patterns were photographed and 50-^m sections were made parallel to the measured angle of electrode penetration for reconstruction of electrode tracks. Results General Kittens adapted well to wearing the goggles, and after a few days showed relatively normal visual behaviors. There were qualitative differences between the 0- and 16-deg kittens, on the one hand, and the 32-deg kittens on the other hand; the former showed generally better visuomotor coordination. At about 3 months of age (3 years for the 32-deg + N condition) kittens were sent to the University of Western Ontario for quantitative behavioral testing. Subsequent to behavioral testing, electrophysiological recordings were made from seven kittens in the three experimental rearing conditions and from two kittens in the 0-deg control condition. For the 32-deg + N kittens, recordings were made 3 years prior to behavioral testing, as described above. Responses to visual stimulation were studied in 528 visual cortical neurons near the 17/18 border; the great majority were visually responsive and nearly all had receptive fields within 10 deg of the area centralis. 0-deg control condition Depth perception Table 1 shows the monocular and binocular depth thresholds for kittens reared under the 0-deg goggle condition. The binocular thresholds for these kittens ranged from 9-20 min of retinal disparity. With the exception of the kitten with a 9-min threshold, these values are somewhat greater than those found

Table 1. Behavioral and pupillographic data Depth thresholds (nominal retinal disparity in min of arc) Kitten

Condition (deg)

84.1 86.1 86.2 87. l a 91.1 96. l a 96.2 a 109.1 109.2 91.2 a 109.3 109.4 79.1 79.2 52.1° 54. l a

0 0 0 0 + 16 + 16 + 16 + 16 + 16 -16 -16 -16 + 32 -32 +32-deg + N -32-deg + N

Monocular

Binocular

44 44

20 16 15 9 20 47 42 18 10 36 20 9 43 >70 34 40

20 9 44 b

46 18 10 40 34 34 b

52 38 48

Vergence ratios After prism goggle rearing

After depthperception testing

0.97 0.98 0.99 0.95 0.96 0.94 1.00 0.85 — 0.96 0.95

— 0.97 1.02 0.95 1.03 0.98 1.01 0.97 1.04 0.97 0.96 0.94 0.96 0.98 c 0.99 c

— 0.94 0.91

a

Subjects that received delayed depth-perception testing as described in the text. b No responses could be elicited under the monocular condition. c After 6-8 months normal visual experience following prism-goggle rearing.

in most normally reared kittens (Timney, 1990), although Altmann et al. (1987) reported a binocular threshold of 18 min for one of their normal kittens. There are several possible explanations for the performance of these kittens. First, the higher binocular thresholds in the 0-deg kittens may be a consequence of their prolonged intervals in the dark. Although these kittens spent several hours each day in the light, there are no data available to indicate how much normal exposure is necessary to ensure complete development of normal stereopsis, nor is it possible to ascertain the precise effects of wearing goggles throughout their periods in the light. Second, older kittens and adult cats typically are more difficult to test on the jumping stand than young animals, and so their lower performance may simply be a reflection of their general recalcitrance. The measure of interest, however, is the difference between binocular and monocular thresholds. In assessing stereopsis using the jumping-stand procedure, inferences are made about its presence or absence based on relative monocular and binocular thresholds. For normal kittens this usually presents no difficulty, because monocular thresholds are typically several times greater than binocular thresholds. In our own work the average monocular/binocular ratio is between 4 and 5 (Timney, 1990). The comparison of binocular and monocular thresholds has two limitations, however. First, if binocular performance is reduced somewhat from that of the typical normal kitten (5-10 min), then comparisons of binocular and monocular thresholds will show a reduced ratio. It should be remembered that stereopsis is not an all-or-none phenomenon; it is possible to have reduced stereoacuity without losing stereopsis. Thus, kittens with low stereoacuity who nevertheless perform better

Stereopsis in cats following cyclotropic rearing

307

with two eyes than one may be presumed to have at least some stereoscopic ability. The second limitation occurs if a kitten achieves a high level of monocular performance, so that both monocular and binocular thresholds are low, and fall within the range of normal binocular thresholds. In our experience this does happen occasionally, both in normal and visually deprived kittens; interpretation of data of this type is equivocal. It is possible that the kitten possesses stereopsis, but is also adept at using an accurate monocular cue such as motion parallax. Alternatively, the animal may be using monocular cues under both testing conditions. Examination of the training data indicated that while there was some difference in the speed with which different cats learned the discriminations both monocularly and binocularly, there was no indication that kittens that received more monocular trials had better monocular thresholds. It has been our experience that level of monocular performance is not linked systematically with amount of monocular practice. In the present experiment, one 0-deg kitten (87.1) and two 16-deg kittens (109.1 and 109.2) showed relatively good performance under both monocular and binocular testing conditions. Because of the difficulties of interpretation no conclusions may be drawn from their performance. For the remaining three kittens in the 0-deg condition, however, performance was superior in the binocular testing condition than in the monocular testing

CONTROL CONDITIONS DEPTH THRESHOLDS

100

90

condition. Fig. 1 shows psychometric functions for monocular and binocular threshold determinations for a 0-deg gogglereared kitten (86.1) and, for comparison, a normally reared kitten from an earlier study. Neurophysiological results Most cells could be classified into ocular-dominance (OD) categories 1-7 (Hubel & Wiesel, 1962) with the exception of those that were visually unresponsive (VU) or whose response properties were highly variable or unmappable (UM). In the 0-deg condition, the OD distribution of 70 cells obtained from two kittens (86.2 and 87.1) resembled that found in normally reared kittens (Fig. 2). Of these, 61% (43 cells) had binocular oriented receptive fields; the mean of the distribution of optimal interocular orientation disparities was —1.9 deg (Fig. 2), which does not differ from the expected mean value of 0 deg for normally reared kittens (/ = -0.85, P > 0.20). 16-deg condition Depth perception Five kittens exhibited good binocular performance, with thresholds falling within the range of the 0-deg control kittens. Of these, three showed superior thresholds for binocular viewing; Fig. 3 shows the threshold functions for one of these animals (109.4). Two kittens (109.1 and 109.2), as mentioned earlier, performed well under both conditions. Three kittens in the 16-deg condition showed relatively poor binocular depth perception; their monocular and binocular thresholds were similar and fell in the range of monocular values obtained from normally reared kittens studied in previous experiments (Timney,

0° CONDITION (Kitten 86.1)

0° CONDITION

80 70

20-

n = 70

60

rr § o

50 -

" 60

2100 o UJ 0.

w 50

40

30

20

10

0

10--

_i ULJ

—

o

2

3

4

5

6

7

VU

UM

OCULAR DOMINANCE

NORMAL

90

2 30

80

LU O DC

70

£20

s = 18° n = 43

60

10

50 60 50 40 30 20 10 NOMINAL RETINAL DISPARITY (min of arc)

0

Fig. I. Monocular (open circles) and binocular (filled circles) depth thresholds for one kitten in the 0-deg rearing condition (top) and one normally reared kitten from an earlier experiment (bottom).

-60 -72 -64 -56 -«8 -40 -32 -24 -16 -6

0

16 24 32 40 46 96 64 72 80

INTEROCULAR ORIENTATION DIFFERENCE (degrees) Fig. 2. Distributions of cells' ocular dominance (top) and preferred interocular disparities of cells' receptive-field orientations (bottom) from two kittens in the 0-deg rearing condition.

308

P.G. Shinkman, B. Timney, and M.R. Isley

16° CONDITION

16° CONDITION DEPTH THRESHOLDS

20

90 r (Kitten 109.4)

n- 99

80 CO

70 -

10

_i

LU

60-

o

UJ

40

orr

50

40

30

20

10

0

3 4 5 6

7

VU

UM

OCULAR DOMINANCE

O |

DC

cr 8

1 2

LL

50 h

20

s = 18° n=61

(Kitten 91.2)

10 -80 -72 -64 -56 -48 -40 -32 -24 -16 -8

0

6

16 24 12 40 48 56 64 72 SO

INTEROCULAR ORIENTATION DIFFERENCE (degrees) Fig. 4. Distributions of cells' ocular dominance (top) and preferred interocular disparities of cells' receptive-field orientations (bottom) from three kittens in the + 16-deg rearing condition. 50 40 30 20 10 0 NOMINAL RETINAL DISPARITY (min of arc)

Fig. 3. Monocular (open circles) and binocular (filled circles) depth thresholds for two kittens in the 16-deg rearing condition. Kitten 109.4 showed normal vergence while kitten 91.2 was esotropic (see Table 1).

1985). Fig. 3 shows threshold functions for one of these kittens (91.2). The variability in binocular depth threshold among kittens in the 16-deg condition was related to pupillographic changes as described below. Neurophysiological results Fig. 4 shows the OD distribution of 99 cells from three + 16-deg kittens (91.1, 96.1, and 109.1). Overall, 63% were binocular, which does not differ significantly from the 0-deg control condition (x 2 = 0.09, P > 0.75). Among cells assigned to OD categories 1-7, 75% were binocular, which also does not differ significantly from the value of 70% observed in the 0-deg condition (x 2 = 0.36, P > 0.50). As we have reported previously (Bruce et al., 1981; Shinkman & Bruce, 1977), experience with 16-deg prism goggles altered the distribution of visual cortical cells' preferred interocular orientation disparities. Fig. 4 shows the distribution of 61 cells' optimal disparities, recorded from the same three kittens; the mean was 16.0 deg, which differs significantly from 0 deg (t = 6.88, P < 0.002) but not from 16 deg (/ = 0.01, P > 0.80). No differences were apparent among 16-deg kittens in terms of the physiological results, and the variances of the 0- and 16-deg interocular orientation disparity distributions did not differ significantly (F = 1.00, P > 0.05).

32-deg and 32-deg + N conditions Depth perception The four kittens in the 32- and 32-deg + N conditions all exhibited severe deficits in binocular depth perception. Fig. 5 shows threshold depth functions for the two kittens in the 32deg condition (79.1 and 79.2); one of these did not perform under monocular testing and had a binocular depth threshold comparable to the monocular thresholds of normally reared kittens. Similarly the two kittens in the 32-deg -I- N condition (52.1 and 54.1) showed no superiority of binocular over monocular depth perception despite 3 years of normal visual experience in the colony following their initial rearing from 3-12 weeks of age wearing 32-deg goggles. Their data are presented in Fig. 6. Overall, the binocular depth thresholds of kittens in the 32and 32-deg + N conditions were significantly higher than those of kittens in the 0- and 16-deg conditions ( P < 0.05, two-tailed Mann-Whitney U test), despite the fact that the latter kittens' binocular thresholds were somewhat elevated compared to normally reared kittens, as described above. Neurophysiological results Kittens in the 32- and 32-deg + N conditions showed a marked loss of binocular cells in the visual cortex, both at the end of the prism-goggle rearing period and also after extended normal visual experience (Figs. 5 and 6). Specifically, among cells in OD categories 1-7, the proportion falling in extreme OD categories 1, 2, 6, and 7 was 48% in the 0-deg condition, compared with 94% in the -t-32-deg condition (x 2 = 20.61, P < 0.001), 98% in the -32-deg goggle

Stereopsis in cats following cyclotropic rearing

309

+ 32° CONDITION (Kitten 79.1)

- 3 2 ° CONDITION (Kitten 79.2) DEPTH THRESHOLDS 80

O

£40 rr O

60

50

40

0 60 50 40 30 20 10 NOMINAL RETINAL DISPARITY (min of arc)

30

20

10

0

VISUAL ACUITY THRESHOLD

90 r 80 70 60 50 3 4 1 SPATIAL FREQUENCY (cycles/degree)

40 III

40

n = 42

30

30

2 20

20

u.

DC ID 0-

1 2 3 4 5 6 7

VU

D

72

10

UM

I D•

—1 , 1 2 3 4 5 6 7

vu

UM

OCULAR DOMINANCE

condition (x 2 = 40.49, P < 0.001), 76% in the +32-deg + N condition immediately following goggle rearing (x 2 = 7.99, P < 0.005), 79% in the +32-deg + N condition following 6-8 months normal visual experience (x 2 = 10.50, P < 0.005), 76% in the —32-deg + N condition after goggle rearing (x 2 = 11.43, P < 0.001), and 79% in the -32-deg + N condition after normal visual exposure (x 2 = 12.88, P < 0.001). For each neuron, preferred stimulus orientation was plotted as described above for the receptive field in one or both eyes (monocular and binocular cells), and subsequently corrected for the intorsional changes in ocular position caused by Flaxedil paralysis. Most cells had oriented receptive fields in one or both eyes, although the incidence of non-oriented receptive fields was generally greater in the 32-deg conditions than in the 0-deg control condition. In the 32- and 32-deg + N conditions, there was also an overrepresentation of receptive fields with preferred orientations near horizontal or vertical (HV). This meridional bias was also observed, although to a much lesser extent, in the 0-deg condition. The meridional effect was linked to both oc-

Fig. 5. Behavioral and physiological results in the +32-deg (left) and -32-deg (right) conditions. Top: monocular (open circles) and binocular (filled circles) depth thresholds (no responses could be elicited from kitten 79.1 under the monocular condition). Middle: acuity thresholds from kittens 79.1 and 79.2. Bottom: distributions of cells' ocular dominance for kittens 79.1 and 79.2.

ular dominance and orientation selectivity. The majority of HV receptive fields were encountered in monocular cells or in the dominant eye of binocular cells, and often reflected strong contralateral input. Furthermore, cells with narrowly tuned receptive-field orientations tended to prefer horizontally or vertically oriented stimuli rather than oblique orientations. The distributions of cells' optimal interocular orientation disparities in the 32- and 32-deg + N conditions showed no evidence of compensatory shifts towards the experienced rotations; none of the means differed significantly from 0 deg. The variances of these distributions, however, were all greater than the corresponding variances observed in the 0- and 16-deg conditions, as we have reported previously (Isley et al., 1990). Pupillographic

measurements

Vergence ratios were calculated for 15 of the 16 kittens in these experiments. Results of these measurements are shown in Table 1. Vergence ratios of kittens in the 16-deg rearing condition

310

P.G. Shinkman, B. Timney, and M.R. Isley

+ 32° + N CONDITION (Kitten 52.1)

- 32° + N CONDITION (Kitten 54.1)

After Goggle Exposure 30

3

20

20

10

10 1 2 3 4 5 6 7

W O

VUUM

1 2 3 4 5 6 7

VUUM

OCULAR DOMINANCE

After 6 to 8 Mos. Subsequent Normal Visual Experience 40 r

ID O rr HI

a.

n = 78

30

n = 48

n = 63

n = 56

30 20

20

LJ]

10

1 2 3

4 5 6 7

•

10

VUUM

1 2 3 4 5 6 7

VUUM

OCULAR DOMINANCE

After 3 Yrs. Subsequent Normal Visual Experience DEPTH THRESHOLDS 90 r 80

rr

70

OC70

O

60 60

50 40

UJ40 50

40

60 50 40 30 20 30 20 10 0 NOMINAL RETINAL DISPARITY (min of arc)

were of special interest because of the variation in the depthdiscrimination thresholds. It is apparent that kittens in the 16-deg condition that showed better binocular depth perception differed systematically from kittens that did not. Specifically, the kittens with poor binocular depth perception in the 16-deg condition were esotropic at the time of behavioral testing. Indeed, in the 0- and 16-deg conditions there was no overlap in the vergence ratios of kittens that showed superiority of binocular over monocular performance, which ranged from 0.95-0.97, and those of kittens exhibiting poor binocular performance, which ranged from 0.98-1.04 (Table 1). Kittens in the 32-deg conditions had normal vergence ratios, while the two 32-deg + N kittens were esotropic following several months of normal visual experience. Subsequent inspection of the experimental records showed that three 16-deg kittens (91.2, 96.1, and 96.2) and one 0-deg kitten (87.1) received their behavioral testing 2 months or more after arriving in Canada, during which time they lived in a normally illuminated laboratory environment. The other 0-, 16-,

10

0

Fig. 6. Behavioral and physiological results in the+32-deg + A'(left) and -32-deg + N (right) conditions. Ocular-dominance distributions obtained at age 3 months following prism goggle rearing (top) and after 6-8 months subsequent normal visual experience in the main colony (middle). Bottom: monocular (open circles) and binocular (filled circles) depth thresholds measured after 3 years subsequent normal visual exposure.

and 32-deg kittens began their testing immediately. Within the 16-deg group, the three kittens tested late all showed reduced depth perception when tested binocularly, whereas none of the other five showed such deficits (/> = 0.05, Fisher's exact probability test). Furthermore, vergence ratios around the time of behavioral testing (Table 1, right-hand columns) differed overall between the six kittens that received delayed testing (footnote a in Table 1, A / = 1.01) and eight that did not ( M = 0.97); the former were significantly esotropic (P < 0.01, two-tailed MannWhitney (/test). This difference was not present in the vergence ratios measured right after the period of goggle rearing (for kittens tested immediately, M = 0.95; for kittens that received delayed testing, M = 0.95). Acuity It was important to determine the degree, if any, to which deficits in binocular depth perception might be attributable to

Stereopsis in cats following cyclotropic rearing inadequate visual acuity. Three kittens (two in the 32-deg conditions and one 0-deg kitten) were tested for visual acuity as described above. The 0-deg kitten had an acuity threshold of 3.11 cycle/deg, and the two 32-deg kittens had comparable acuities (2-3 cycle/deg; Fig. 5). The diameters of the dots comprising the pattern for testing depth perception ranged from 0.5-1.8 cm, so that even the smallest dots had angular subtense greater than 20 min from the kittens' viewing distance of 80 cm. Thus, although the acuity of these kittens was lower than that found for most normally reared kittens, it was clearly more than adequate for performance of the depth-discrimination task, since the kittens reliably discriminated targets subtending less than 15 min of arc. Discussion Studies of changes in the proportion of binocular neurons that result from different rearing procedures such as monocular deprivation, strabismus, and so forth provide crucial information about the plasticity of synaptic connectivity. However, such studies do not address the issue of the instructive role of experience in tuning an immature cortex. This question is particularly important in considering the development of cooperative interactions between the two eyes. Accurate stereoscopic depth judgments require that retinal disparity values be calibrated with respect to the distance of a target from the fixation plane. Retinal disparity itself is a function of both fixation plane and target distance, and also of the lateral separation of the eyes. In the developing animal, this separation increases rapidly (Timney, 1988). Pettigrew (1983), among others, has argued that a potential role for the sensitive period may be to permit the continuous calibration of the stereoscopic mechanism during development. Our previous experiments on the effects on cortical physiology of early experience with interocular rotations of the visual field (e.g. Bruce et al., 1981) provided support for such an instructive role of visual input during development when the rotational perturbation is small (16 deg). Large rotations, on the other hand, represent a degree of discordant or mismatched inputs to the two eyes too great to allow such instructive adaptation to occur (Blakemore et al., 1975; Crewther et al., 1980; Isley et al., 1990; Washington et al., 1977). Considered together, the present results strongly support the overall hypothesis that guided these experiments: the behavioral findings on depth perception generally showed a clear and consistent relation to the neurophysiological findings on the synaptic organization and receptive-field properties of the visual cortex. Furthermore, in those cases where the behavioral and physiological findings seemed at first not to be in accord, unanticipated underlying consistencies emerged involving pupillographic changes secondary to the timing of the experimental measurements. The 32-deg prism-goggle rearing condition, like other rearing paradigms that disrupt visual cortical binocularity, produces a total deficit in binocular depth perception or stereopsis. In other experiments, we have shown that the cortical effects of prism-goggle rearing are permanent and not subject to reversal even with extended periods of subsequent normal visual input (Shinkman et al., 1980). The present findings from the 32-deg + N condition establish that the permanence of these physiological effects is paralleled by a permanent behavioral change as manifested in these subjects' lack of stereoscopic depth perception after 3 years of normal visual experience.

311 The mixed behavioral results in the 16-deg condition were initially puzzling. These eight kittens were divided into two sets, three that exhibited reduced binocular depth perception and five that did not. Reexamination of pupillographic data showed that these two sets differed in that the former kittens were esotropic when tested behaviorally while the latter kittens had normal ocular position. Further reexamination of the experimental protocols revealed that, for logistical reasons deemed unimportant at the time, some of these kittens were tested quite soon after their prism-goggle rearing had been completed, while others were kept waiting for 2 months or more. As can be seen from Table 1, all of the kittens tested late were esotropic and/or showed esotropic changes in ocular position, these changes corresponding closely to those reported for dark-reared kittens upon being returned to the light (Cynader, 1979). The following principal conclusions may be drawn. First, the presence of a substantially normal complement of binocular visual cortical cells is necessary for the normal development of binocular depth perception or stereopsis. This conclusion is supported by the relatively good binocular performance of most kittens in the 0- and 16-deg conditions, whose ocular-dominance distributions were normal, and by the uniformly poor binocular performance of kittens in the 32- and 32-deg + N conditions, all of whom showed severe deficits in stereopsis and marked loss of binocular cortical neurons. Second, while the disruption of cortical binocularity results in a loss of stereopsis, the presence of binocular neurons does not guarantee that stereopsis will be retained, since there were kittens in the 0- and 16-deg conditions whose cortical binocularity appeared normal but who showed reduced stereopsis. These latter kittens may have lacked disparity detecting neurons among the population of binocular cells observed, an interpretation (as yet untested) that may also apply to the deficient stereopsis in kittens reared under conditions of binocular deprivation (Kaye et al., 1982), whose visual cortices likewise retain a substantial proportion of binocular cells. Third, esotropia is similarly sufficient but not necessary to disrupt stereopsis, since three kittens in the 16-deg conditions with esotropia or esotropic changes at or during the time of behavioral testing showed poor stereopsis, while the +32- and —32-deg kittens were not esotropic (vergence ratios of 0.94 and 0.96, respectively) yet failed to show any evidence of stereopsis. Fourth, the absence of stereoscopic depth perception was not due to deficits in visual acuity, since all three kittens tested showed acuity thresholds in the range of 2-3 cycle/ deg; these included two 32-deg kittens without stereopsis and one 0-deg control kitten with good stereoscopic depth perception. Fifth, as has been reported previously for other rearing paradigms (Timney, 1988), the capacity for stereopsis, once lost, cannot be regained even after extensive subsequent normal visual experience. Thus, in terms of the organization of the visual cortex, including especially the subsystem of orientation disparity detecting neurons, and also of stereoscopic depth perception, the developing visual system appears capable of adapting both physiologically and behaviorally to small (16 deg) but not to large (32 deg) optically induced rotations of the two eyes' visual fields.

Acknowledgments Grateful acknowledgment is made to Diane Rogers-Ramachandran and to Cathy Bryans for assistance in conducting the experiments, to Vicki

312 Fowler for assistance in preparing the manuscript, to Betty Lloyd for assistance in preparing the figures, and to Dr. William Beel for assistance with the optometric measurements. M.R. Isley received fellowship support from USPHS training grants to the Experimental Psychology Program and to the Neurobiology Program, University of North Carolina. This research was supported by USPHS Grant MH-17570 to P.G. Shinkman, by the Brain and Development Research Center, University of North Carolina, and by Medical Research Council of Canada Grant MA-7125 to B. Timney.

References ALTMANN, L., LUHMANN, H.J., GREUEL, J.M. & SINGER, W. (1987).

Functional and neuronal binocularity in kittens raised with rapidly alternating monocular occlusion. Journal of Neurophysiology 58, 965-980. BLAKE, R. & HIRSCH, H.V.B. (1975). Deficits in binocular depth perception in cats after alternating monocular deprivation. Science 190, 1114-1116. BLAKEMORE, C , VAN SLUYTERS, R.C., PECK, C.K. & HEIN, A. (1975).

Development of cat visual cortex following rotation of one eye. Nature 257, 584-586. BRUCE, C.J., ISLEY, M.R. & SHINKMAN, P.G. (1981). Visual experience

and development of interocular orientation disparity in visual cortex. Journal of Neurophysiology 46, 215-228. CREWTHER, S.G., CREWTHER, D.P., PECK, C.K. & PETTIGREW, J.D.

(1980). Visual cortical effects of rearing cats with monocular or binocular cyclotorsion. Journal of Neurophysiology 44, 97-118. CYNADER, M. (1979). Interocular alignment following visual deprivation in the cat. Investigative Ophthalmology and Visual Science 18, 726-741. HUBEL, D.H. & WIESEL, T.N. (1962). Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. Journal of Physiology 160, 106-154. HUBEL, D.H. & WIESEL, T.N. (1965). Binocular interaction in striate cortex of kittens reared with artificial squint. Journal of Neurophysiology 28, 1041-1059. HUBEL, D.H. & WIESEL, T.N. (1970). The period of susceptibility to the physiological effects of unilateral eye closure in kittens. Journal of Physiology 206, 419-436.

P.G. Shinkman, B. Timney, and M.R. fsley PETTIGREW, J.D. (1983). Teleology of the critical period for binocular vision. Presented at the satellite symposium on Development of Visual Pathways in Mammals, International Union of Physiological Sciences, Newport, Australia. PODELL, M., ISLEY, M.R., SHINKMAN, P.G. & ROGERS, D.C. (1982). Vi-

sual development: Early experience with torsionally disparate images. Metabolic, Pediatric and Systemic Ophthalmology 6, 273-283. SHINKMAN, P.G. (1983). Effects of interocular torsional image disparity on visual cortical development. Invited presentation, Australasian Winter Conference on Brain Research, Queenstown, New Zealand. SHINKMAN, P.G. & BRUCE, C.J. (1977). Binocular differences in cortical receptive fields of kittens after rotationally disparate binocular experience. Science 197, 285-287. SHINKMAN, P.G., ISLEY, M.R. & ROGERS, D.C. (1980). Cortical binoc-

ular receptive fields in developing kittens: Effects of early visual field rotation followed by normal visual exposure during adulthood. Proceedings of the International Union of Physiological Sciences 14, 700. SHINKMAN, P.G., ISLEY, M.R. & ROGERS, D.C. (1983a). Prolonged dark

rearing and development of interocular orientation disparity in visual cortex. Journal of Neurophysiology 49, 717-729. SHINKMAN, P.G., ISLEY, M.R. & ROGERS, D.C. (1985). Development of

interocular relationships in visual cortex. In Advances in Neural and Behavioral Development, Vol. 1, ed. ASLIN, R., pp. 187-268. New Jersey: Ablex Publishing Corp. SHINKMAN, P.G., TIMNEY, B. & ISLEY, M.R. (1986a). Binocular depth

perception and pupillographic changes in kittens reared with interocular torsional disparity. Society for Neuroscience Abstracts 12, 785. SHINKMAN, P.G., TIMNEY, B. & ISLEY, M.R. (19866). Perceptual, pu-

pillographic, and physiological effects in cats reared with interocular torsional disparity. Presented at First International Congress of Neuroethology, Tokyo, Japan. SHINKMAN, P.G., TIMNEY, B., ISLEY, M.R. & ROGERS, D.C. (19836).

eye alignment and cortical ocular dominance of dark-reared cats. Developmental Brain Research 2, 37-54.

Physiological and behavioral consequences of optically-induced interocular torsional disparity in early visual input. Proceedings of the International Union of Physiological Sciences 15, 445. TIMNEY, B. (1985). Visual experience and the development of depth perception. In Brain Mechanisms and Spatial Vision, ed. INGLE, D.J., JEANNEROD, M. & LEE, D., pp. 147-174. The Netherlands: M. Nijhoff. TIMNEY, B. (1988). The development of depth perception. In Advances in Neural and Behavioral Development, Vol. 3, ed. SHINKMAN, P.G., pp. 153-208. Norwood, New Jersey: Ablex Publishing Corp. TIMNEY, B. (1990). Effects of brief monocular deprivation on binocular depth perception in the cat: A sensitive period for the loss of stereopsis. Visual Neuroscience 5, 273-280. VON GRUNAU, M.W. (1979). The role of maturation and visual experience in the development of eye alignment in cats. Experimental Brain Research 37, 41-47.

MITCHELL, D.E., KAYE, M. & TIMNEY, B. (1979). Assessment of depth

WASHINGTON, I.M., ISLEY, M.R. & SHINKMAN, P.G. (1977). Effects of

perception in cats. Perception 8, 389-396. OLSON, C. & FREEMAN, R.D. (1978). Eye alignment in kittens. Journal of Neurophysiology 41, 848-859. PACKWOOD, J. & GORDON, B. (1975). Stereopsis in normal domestic cat, Siamese cat, and cat raised with alternating monocular occlusion. Journal of Neurophysiology 38, 1485-1499.

large interocular rotational disparities on the development of binocularity in kitten visual cortical neurons. Society for Neuroscience Abstracts 3, 580. WIESEL, T.N. & HUBEL, D.H. (1965). Comparison of the effects of unilateral and bilateral eye closure on cortical unit responses in kittens. Journal of Neurophysiology 28, 1029-1040.

ISLEY, M.R., PLUMMER, K.R. & SHINKMAN, P.G. (1979). Cortical bin-

ocular receptive fields after visual field rotation in developing kittens: Effects of subsequent normal visual exposure during adulthood. Society for Neuroscience Abstracts 5, 790. ISLEY, M.R., ROGERS-RAMACHANDRAN, D.C. & SHINKMAN, P.G. (1990).

Interocular torsional disparity and visual cortical development in the cat. Journal of Neurophysiology 64, 1352-1360. KAYE, M., MITCHELL, D.E. & CYNADER, M. (1982). Depth perception,