BINOCULARITY IN PRISM-REARED MONKEYS
M. L. J. CRAWFORD\ R. S. HARWERTH2, Y. M. CHIN02 E. L. SMITH, 1112 and Houston, Texas
story to be presented here, by establishing that early SUMMARY Prismatic binocular dissociation in infant monkeys abnormal visual experience played a role in deter mining cortical circuitry and binocular function. mimicked a concomitant squint. Within weeks, the 4 5 numbers of binocular neurons in the primary3 visual Contemporary research by Bela Julesz , showed cortex were reduced by half and did not recover with up that horizontal disparities in the binocular view of to 5 years of subsequent unrestricted binocular visual random-dot stereograms were alone sufficient to experience. The monkeys failed to show binocular induce not only the perception of depth but the vivid summation for spatial contrast sensitivity tasks and perception of form as well. Concurrently, clinicians were unable to utilise horizontal binocular disparities in began to incorporate the findings of basic research 8 random-dot stereograms - two indices of stereoblind into their theory and practice in the clinic.6- In short, ness. Electrophysiological analysis of the and the confluence of these theoretical, empirical and cortices showed a dramatic reduction inVI binocular V2 clinical efforts comes down squarely on the binocular neurons. Analysis of interocular spatial phase tuning neurons of the visual cortex as the neural substrate functions showed a conspicuous loss of excitatory underlying binocularity, the general term which binocular drive in neurons which was sufficient to subsumes the definition of stereopsis, the sensing of VI account for many of the defects in binocular function. horizontal disparities leading to the appreciation of depth.9-12 It is akin to bringing coals to Newcastle to assert that Having a basic interest in the role of early visual circuits of binocular neurons of the visual cortex con experience in moulding the function of the visual stitute the neurophysiological substrate for binocu system in children, several years ago we began larity and the associated function of stereopsis. Since controlled studies modifying the quality of visual the beginning of written history, it has been known experience in infant monkeys to simulate the visual that somewhere in the brain the visual information disorders commonly encountered in the paediatric coming from the two eyes must be united in order to clinic.13-15 Among the methods employed, we put have a single perception of the world, and that prisms before the eyes of infant monkeys to simulate unified single perception was reduced in quality and the optical conditions attendant on concomitant sensitivity if one eye was lost or disabled. Wheat 1 strabismus. We devised a lightweight, padded helmet stone's historical demonstration that binocular view for holding base-in wedge prisms, with one prism ing of identical flat images set at a slight horizontal rotated downward to create a condition of chronic disparity induced the perception of the depth diplopia.16-18 The helmet was put on the monkeys at equivalent to a real-life scene, set investigators on a different ages, from birth up to 4 months of age, and course of speculation and search to find the locus in worn continuously for durations between 1 and 12 the brain where monocular union could take place; weeks. This rearing strategy was selected to deprive where binocular vision arises with the full-blown 2 3 the binocular neurons of V1 cortex, present in expression of stereopsis. Wiesel and Hubel , gave monkeys at birth,t9 of normal binocular stimulation the search enormous impetus with their benchmark during infancy, with the further aim of relating the work, not only by describing the phenomenon of numbers and functions of binocular neurons to the binocular convergence but, more relevant for the behavioural abilities of these young monkeys. l ! That such abnormal early visual experience is From: The Department of Ophtha mology & Visual Science of The Medical School at Houston, and The College of Optometry, ruinous for cortical binocular neurons is illustrated in The University of Houston, Houston, Texas, USA. Fig. 1a (0 years, open bars) which shows the ocular Correspondence to: M.L.I. Crawford, 6420 Lamar Fleming, dominance distribution obtained immediately at the Houston, TX 77030, USA. Fax: +1-713-7924513. e-mail: 16 [email protected]. end of the period of wearing the prisms. This eye-
Eye © 1996 Royal College of Ophthalmologists (1996) 10,161-166 M. L. 1. CRAWFORD 162 ET AL. Normal Monkey Prism-reared Monkey 45 0 w 40 ..J Il. 35 ,f;' 100 100 :E > ct "" 30 · iii en c Q) en 25 (/) z
Examples of monocular and binocular contrast 70 2. Fig.sensitivity functions from a normal and from a prism-reared C monkey. w ...J 60 Il. dominance histogram shows that binocular neurons :!:« in VI cortex are reduced by 60%, and that this en 50 en reduction occurs within a matter of days, being z almost complete within 60 days (Fig. Ib). Note that 0 40· a: the balance in the numbers of monocular neurons is ;:) . . __ .. . _ ...... _--_.. . . .- ..... �.-- ...... - ...... -.-- w -+_ ...... _\. --. - - - retained, suggesting that the prism rearing procedure Z 30 u.. did not create an amblyopia in either eye.
0 ...... -._...... their ... do binocular neurons recover .. Once lost, -1 ...... �- ..------20 inputs when the prisms are removed? (The monkeys '#. showed no evidence of strabismus.) Fig. illustrates 10 the contrast sensitivity functions of a normal2 and a prism-reared monkey. In the typical data set from a 0 normal monkey of about 4 years of age, one can see 0 10 20 30 40 50 60 that the monocular sensitivity for the two eyes is well DAYS OF PR balanced with peak sensitivities of about 4 cycles/deg, (b) and a high spatial frequency cut-off of about 25-30 cycles/deg_ Note that the binocular function is significantly higher than the monocular curves
1. (a) Eye dominance histograms for neurons (about 0.2 log units) at all spatial frequencies. This recordedFig. from five groups of monkeys. Group VI0 years was phenomenon of binocular summation is considered recorded immediately at the end of the period of prismatic to reflect the contribution of the cortical binocular binocular dissociation. Groups 4 and years were 5 neurons to the monkey's superior binocular contrast recorded after the indicated 3,intervening period of O 21 post-treatment normal binocular experience. The histo sensitivity? , In comparison, the curves to the right grams from the controls are connected by the continuous in Fig. 2 show that the prism-reared monkey gained line to indicate the normally large numbers of binocular no increase in sensitivity by binocular viewing neurons, which are significantly reduced in all the prism reared groups. The right-eye dominant categories range compared with the best monocular curve, suggesting from the monocular to the weakly binocular to the that there was no binocular advantage to sensitivity strongly binocular RI, group. E4 represents theR2, equally from viewing with two eyes. The absence of normal balanced and stronglyR3 binocular cells. The L categories binocular summation suggests that binocular func represent the neurons similarly dominated by the left eye. treatment had not (b) The average relative percentage of remaining binocular tions lost during the experimental neurons as a function of days of continuous prism recovered, even with 2 years of subsequent binocular rearingVI stimulation.22 (PR). BINOCULARITY IN PRISM-REARED MONKEYS 163 had not recovered after as much as 3 years of 100 IZ,L.,£ &� � oQi£ binocular stimulation. These data are shown in the 90 � 4: III eye-dominance histogram of Fig. 1a (3 years), which Z 0 80 is virtually identical to histograms obtained immedi � u 70 ately after binocular dissociation (0 years). More LIJ I- 60 LIJ over, in another cohort of monkeys recorded 4 years C 50 I- following prism rearing (Fig. 1a; 4 years) there was U ?4 LIJ 40 still no significant recovery of binocular neurons II II 30 In a recent set of experiments on a subset of these 0 u, 20 .it stereoblind monkeys, we recorded not only the eye � • 'I. )'L " 10 dominance profiles (Fig. la, 5 years; compare with • IJ J J J .. A 0 B& C Controls shown in this figure: there was no significant N N P P P P P P P P P P P P N N MONKEYS NORMAL (N) PRISM MONKEYS (P) A Orientation Response Function B Spabal Frequency Response Funcbon 3. Form detection (A) and crossed and uncrossed 90deg 31 Fig.disparity detection (B&C) in dynamic random-dot stereo grams by 4 normal (N) and prism-reared monkeys. o nght eye • Iefteye The horizontal line indicates12 chance performance(P) in the task. 180 dog
o nghleye We next questioned the significance of the loss of • lefteye 270 de9 binocular cells from the V1 cortex for functional 01 1 10 stereopsis. Could these monkeys detect horizontal Spabal Frequency (cy/deg)
disparities, the universally accepted functional role of C Dlspanty Tuning Funcbon cortical binocular neurons? We tested these, and 100 Maxlmonoc = 3 6 control monkeys, using random-dot stereograms, and 80 found that indeed these prismatically dissociated 811 = 1 22 u'" SlN=59 monkeys were unable to use disparity cues in a ., 60 en behavioural task. The monkeys were trained to press .><., '0.. 40 and hold a lever while watching a colour video � = ., Averagetmonoc 1 6 C/1 20 display for the appearance (or disappearance) of a c: o Q. stereoscopic form (a square, or a grating) embedded C/1 ., within a pattern of dynamic random dots, the form a: M1IlImonoc = .0 4 and apparent depth being determined by the -20 �------�------�// horizontal disparity of a subset of dot-pairs. Fig. 3 90 180 270 360 summarises the performances of 4 normal and 12 Relative Phase (deg) experimental monkeys on a normally easy disparity
detection task (15 min element disparity). The 4. Examples of the measurements made of simple and normal monkeys could readily detect the form from complexFig. neurons taken from control monkeys, and the disparities alone, while most of the experimental experimentalVI monkeys some 5 years after prism-rearing. (A) monkeys could not. This finding was consistent with Simple cell with narrow orientation tuning and balanced sensitivity. (B) Simple cell with binocularly matched narrow the functional absence of binocular neurons in the 16 22 spatial frequency tuning. (Neither orientation nor spatial visual cortex, , and was interpreted as an indica frequency tuning have been shown to be affected by prism tion of stereoblindness. Concurrently, we tested rearing.) (C) Binocular spatial phase tuning for a simple human subjects on the same apparatus and found neuron. Two sets of response data are shown, with a best-fitVI that visually normal children could detect the sine wave from which are derived indices of binocular interaction. The binocular interaction index (Bll) '" disparities while those with histories of strabismus 23 Amplitude of sine wave/Average binocular response. SIN during infancy could not. These findings lent '" Amplitude of the sine wave/the variance of the fit of the credence to our testing procedures, and suggested sine wave to the data. Max/monoc '" Amplitude of sine that similar defects in binocular cell function were wave/average monocular response rate. A verage/monoc '" likely to be present in the visual cortices of the Average binocular response rate/average monocular re sponse rate. Min/monoc '" Sine wave minimum/average children. monocular response rate. The average monocular response To relate the behavioural performance of the rates (triangles) for the left and right stimulations are shown stereoblind monkeys to the loss of binocular neurons, on the ordinate to the right, along with the response rate we did electrophysiological studies in most of these when stimulated by a homogeneous field of space-averaged monkeys and found essentially the same deficit in luminance equal to the gratings. This example of a simple neuron shows a high degree of modulation by the binocular neurons as we had found immediately after driftingVI gratings, with a complete suppression of response the prismatic treatment, i.e. the binocular neurons when the gratings were in antiphase. M. L. J. eRAWFORD 164 ET AL. recovery of binocular neurons), but in addition, the shown as a function of the relative interocular spatial sensitivity of V1 neurons to the relative interocular phase of the drifting gratings presented dichoptically spatial phase of optimal grating patterns. In our prior at the optimal grating orientation, velocity and recordings we had used the traditional receptive field spatial frequency. The two sets of data, collected mapping techniques introduced by Hubel and about 45 minutes apart, have been fitted by a Wiesel, mapping left- and right-eye receptive fields sinusoid, using a least squares method. Symbols on sequentially. With this method, any binocular inter the right-hand ordinate show the responses for actions would have been missed. Therefore, we independent monocular stimulation, as well as the employed a methodology similar to that used by maintained rate when the receptive field was Ohzawa and Freeman25,26 on the cat, using dichoptic stimulated by a uniform field comparable to the binocular stimulation of V1 neurons by drifting sine space-averaged luminance of the grating. This cell wave gratings. Fig. 4 illustrates the methods, showing maintained a stable response over many hours, with in Fig. 4A a polar-coordinates plot of grating clear binocular interactions characterised by binocu orientation tuning for a V1 neuron, indicating a lar facilitation at one relative phase (0) and complete narrow range of tuning around the 45 X 225 deg axis, binocular suppression of the response when the with both receptive fields having the same orienta spatial phase of the grating in one eye was shifted by tion. Fig. 4B shows that this neuron had a narrow about 180 deg. Five indices of binocular interaction spatial frequency tuning around 2 cycles/deg, with are indicated on this figure: (1) the binocular the left eye being marginally more responsive. In interaction index (BII);25 the peak amplitude of the general, neither stimulus orientation nor spatial sine wave divided by the average binocular response, frequency tuning were found to be substantially where a value of 0 indicates no binocular spatial altered with prism-rearing. phase interaction and a value of 1 indicates a high Fig. 4C illustrates the analysis method for relative degree of interaction and complete suppression at binocular disparity tuning and the results for a V1 non-optimal phases; (2) the signal-to-noise (SIN) simple cell. The neuron's response rate (spikes/s) is ratio, the variance about the sine wave fitted to the data, (3) the ratio of the maximum binocular Normal Monkeys Prism-reared Monkeys response divided by the maximum monocular (simple cells) (simple cells) response (Max/monoc); (4) the average binocular
Unit 188L5.el Unit 197L46 e5 A. D . 30 30 Normal Monkeys Prism-reared Monkeys � 1. __ monOC= 3�_O �. (complex cells) (complex cells) r 20 �� 20 . 11 Unit 19717.e3 � A, Unit 198115.el D. 60 6 minimonoc=Q.86 mmlmonoc=O.88 maxlmono,=1 39 10 10 811=0.12 811=0,19 811=0.14 40 S/N=2.69 S/N=1.40 �
20 Unit 190L26.e5 B. Unit 184l4S.el E. U 9 811=0 09 Q) SIN=1 15 :�1 � !
U B Unit 184L16.e1 E, Unit 195L63.el �'0. 10,0 � 811=020 3 SfN=O 63 7 5 . c: 8))=0 31 maxlmo"o,=0.5' 3lo S/N=2.34 c. 5.0 0 I a::::G Unit 190L2.eS 2.5 C. Unit 208L30.el F 24 4 811=0.80 maxlmonoc= 1 77 mm/monoc=O 36 S/N=4.64 0.0 18 a::: :fl Uni1194L1S.e1 Unit 197L57.e1 12 C. F. Il�,.�· 24 75
16 50 mm/monoc=-O 17 1 I ·6 r�1 I , , , ;' 7' 25 90 180 270 360
iii' Relative Phase (deg) Relative Phase (deg) 360 90 180 270 360
5. Examples of the binocular spatial phase analysis of Fig.six simple cells drawn from the first quartile (top), from the Relative Phase (deg) Relative Phase (deg) median range (middle), and from the third quartile 6. Examples as in Fig. 4, but for similar samples from (bottom) of the BlI distribution. Most simple cells were complexFig. cells. Note that the response modulation is some well modulated by the drifting gratings and showed a what less than that for simple cells. and (E) are binocular response facilitation at some spatial phase (E examples of suppression of the binocular(D) response below shows a rare example of a response modulation, but that of the best monocular response - more frequently suppressed below the best monocular response rate). found in the complex than in the simple cells. BINOCULARITY IN PRISM-REARED MONKEYS 165 response divided by the best monocular response; Differences in the binocular cell profilesfor the and (5) the minimum binocular response divided by normal and prism-reared monkeysVl become evident the minimum monocular response. when the neuronal distributions are compared. Fig. 7 Figs. 5 and 6 present three samples each of simple shows the distribution of the BII for the simple and and complex cells from normal and from the prism complex cells from the normal and the prism-reared reared monkeys. The samples from top to bottom monkeys. Recalling that the BII ratio indicates the represent cells of increasing levels of spatial phase strength of the binocular spatial phase interaction (0 modulation: the top panel represents cells with indicates no phase-specificbinocular interaction, and relatively low levels of binocular interaction, typical a value of 1 indicates a high degree of interaction) it of those within the first quartile of the BII distribu is seen that normally simple cells tend to have more tion; those in the middle panel represent cells near of the high BII values. If one compares the numbers the median of the BII distribution; and the bottom of simple cells having BII values greater than 0.6 (an panel represents cells with high levels of binocular arbitrary selection), the relative percentage for modulation from the third quartile of the BII simple cells is twice (45% of the sample; 47/104) distribution. The pattern of BII of the simple cells that for complex cells (21 % of the sample; 27/129). (Fig. 5) shows an overlapping range of binocular Prism-rearing has the effect of reducing the relative spatial phase interactions in both the normal and the percentages of neurons with high BII values by prism-reared monkeys, with the BII being somewhat half (20% (9/44),Vl for the simple cells; 9% (5/54), for smaller in the middle and lower panels for the prism the complex cells). That is, neurons best suited reared monkeys. for discriminating binocular spatialVl phase differences In a similar manner Fig. 6 compares the complex are greatly reduced. neurons in the normal and the prism-reared The binocular facilitation in excitation which Vl monkeys. As was the case for the simple cells, the comes from binocular stimulation is indicated in the examples from the normal and the prism-reared ratio of the average binocular response divided by monkeys show considerable overlap. A hint of what the average monocular response from the best eye, is to come in the overall group comparisons is seen in and is shown as a histogram in Fig. 8. Prism-reared the upper-right panels, where suppression is evident monkeys show two feature differences in the in that the binocular modulation at all phase angles is distributions of both simple and complex neurons. significantly smaller than the best monocular mod Relative to the control animals, there are fewer ulation. neurons with high ratios (cells which show the Comparisons of Figs. 5 and 6 suggest that complex highest excitatory facilitation) in the distributions of neurons generally show less modulation to binocular stimulation than do simple cells, consonant with the Simple Cells Complex Cells findings in normal cats.26 Normal Monkeys Normal Monkeys '"c 0.3 0.3 I Simple Cells Complex cells e" I