Exp Brain Res (2001) 141:349–358 DOI 10.1007/s002210100876

R E S E A R C H A RT I C L E

R.F. Lewis · D.S. Zee · M.R. Hayman · R.J. Tamargo

Oculomotor function in the rhesus monkey after deafferentation of the extraocular muscles

Received: 12 February 2001 / Accepted: 30 July 2001 / Published online: 29 September 2001 © Springer-Verlag 2001

Abstract The function of extraocular muscle proprioception in the control of eye movements remains uncertain. In this study, we examined the effect of bilateral proprioceptive deafferentation of the extraocular muscles on eye movements in two rhesus monkeys. Before and after deafferentation, we analyzed baseline ocular alignment, saccades, pursuit, and vestibular eye movements. We also examined visually mediated adaptation of ocular alignment, saccades, and pursuit. Deafferentation of the eye muscles did not affect baseline ocular motor control, either acutely or over a 5-week period of study. Furthermore, visually mediated adaptation of the eye movement subtypes was also unaffected by deafferentation. These results suggest that ocular proprioception in primates is not used in the immediate, on-line control of eye movements and does not interact with visual cues in the adaptive modification of ocular motor function. We conclude that the efferent command (efference copy) provides sufficient information about eye kinematics to the brain for accurate eye movement control in normal monkeys, and that this information is modified by visual feedback independently of proprioception. We hypothesize that proR.F. Lewis (✉) Department of Otolaryngology, Harvard Medical School, Boston, Mass., USA e-mail: [email protected] Tel.: +1-617-5733620, Fax: +1-617-5734154 R.F. Lewis Department of Neurology, Harvard Medical School, Boston, Mass., USA D.S. Zee Department of Neurology, Johns Hopkins Medical School, Baltimore, Md., USA M.R. Hayman Department of Pharmacology, University of Edinburgh, Edinburgh, UK R.J. Tamargo Department of Neurosurgery, Johns Hopkins Medical School, Baltimore, Md., USA R.F. Lewis 243 Charles St., Boston, MA 02114, USA

prioception may be used to calibrate the efference copy during development and in response to perturbations by signaling potential mismatches between eye movement information derived from the efferent command and the actual motion of the eye transduced by the proprioceptive organs. Keywords Eye movements · Proprioception · Adaptation · Monkey

Introduction The extraocular muscles (EOMs) of primates contain proprioceptive end-organs (reviewed by Ruskell, 1999), and information from these receptors projects to many brain regions concerned with ocular motor and visual function (Donaldson and Long 1980; Ashton et al. 1984). The ocular motor system in primates lacks a stretch reflex (Keller and Robinson 1971), however, and the role played by proprioceptive information in the control of eye movements remains uncertain and speculative (Weir et al. 2000). Three hypotheses have been proposed based on prior experimental results. First, proprioception might contribute to the immediate, on-line control of eye movements. Two types of experiments support this hypothesis. If the proprioceptive signal is perturbed by applying external forces to the globe, the pigeon vestibulo-ocular reflex (VOR) is altered (Donaldson and Knox 2000), as are eye movements such as saccades (Knox et al. 2000), pursuit (van Donkelaar et al. 1997), and the VOR (Knox and Donaldson 1993) in humans. Similarly, vibration of EOM tendons modifies spindle activity and produces illusionary motion of visual targets, suggesting an on-line role for proprioception in the coding of orbital eye position (Roll et al. 1991). In the frog, however, experiments using external forces applied to the globe have not identified any on-line effects of proprioception on eye movement control (Daunicht and Dieringer 1986). The proprioceptive signal can also be eliminated by sectioning the ophthal-

350

mic division of the trigeminal nerve (Porter et al. 1983), and this deafferentation procedure impairs the VOR of the pigeon (Hayman and Donaldson 1995) and the rabbit (Kashii et al. 1989). The effect of proprioceptive deafferentation on ocular motor control in foveate animals that are capable of voluntary eye movements has not been carefully investigated, although prior experiments in the monkey have not suggested a significant on-line role for proprioception (Guthrie et al. 1983; Lewis et al. 1994, 1999). If proprioception does contribute to the on-line control of eye movements in primates, then deficits would be expected after deafferentation, as have been observed in lower species. This was the principal hypothesis tested in the current study. A second possibility is that proprioceptive and visual cues interact during adaptive modification of eye movements. There is extensive evidence that visual-proprioceptive interactions are necessary for the development of certain characteristics of the visual cortex such as stereopsis and orientation selectivity (reviewed by Buisseret, 1995). In the ocular motor system, experiments in human subjects using external forces applied to the globe have suggested possible proprioceptive-visual interactions in the adaptation of ocular alignment (Gauthier et al. 1994) and pursuit (van Donkelaar et al. 1997). If such interactions are necessary for visually mediated adaptation of eye movements, then deafferentation of the EOMs should impair the ocular motor adaptation driven by retinal cues. This hypothesis was also investigated in the current study. A third possibility is that proprioception provides sensory feedback that is used by the brain to “model” the physical characteristics of the ocular motor plant (Jurgens et al. 1981; Steinbach 1986). In this formulation, the kinematic information required for the on-line control of eye movements is derived from the motor command (efference copy or corollary discharge), but the efference copy is calibrated over time by both retinal and extraretinal afferent signals. This hypothesis, supported by prior deafferentation studies in rhesus monkeys with extraocular muscle pareses (Lewis et al. 1994, 1999), predicts that ocular deafferentation in normal, developed animals would not affect the on-line control of eye movements, since an accurate efference copy remains available, and also suggests that visual-driven adaptation would occur independently of proprioception. In this study, we investigated these hypotheses in normal rhesus monkeys. Alignment, saccades, pursuit, and vestibular eye movements were examined before and after bilateral proprioceptive deafferentation of the EOMs, and visually mediated adaptation of the different eye movement subtypes was also studied in both proprioceptive states. Particular emphasis was placed on the possible role of proprioception in maintaining eye movement conjugacy and in tailoring eye movements to different orbital positions, as prior studies and theoretical concerns suggest a potential function for proprioception in these aspects of ocular motor control (O’Keefe and Berkley 1991; Lewis et al. 1994).

Methods General experimental procedures The methodology used in these experiments has been detailed in previous publications (Lewis et al. 1994, 1999). Briefly, binocular eye movements were recorded with magnetic search coils in two juvenile rhesus monkeys (M1 and M2). One experiment was also performed in a third monkey (M3). The output signal of the coil system was filtered at 90 Hz, sampled at 500 Hz, and saved to a computer for off-line analysis. Statistical comparisons were made with t-tests and analysis of variance (ANOVA), and comparisons were deemed to be significant if the P-value was less than 0.05. A horizontal calibration for each eye was obtained for each recording session by having the animal monocularly fixate a series of targets ranging from left to right 20°. Visual targets were presented on a video monitor 24 cm in front of the animal. Eye movements were recorded in total darkness, except for the target. Eye movements were recorded three or four times per week for 2 weeks prior to bilateral ocular deafferentation and for a 5-week period after the deafferentation procedure. The postoperative recording sessions began 6 days after the second eye was deafferented, to allow adequate time for recovery from surgery. Saccades and ocular alignment were recorded sequentially throughout the pre- and postdeafferentation periods, and pursuit and vestibular eye movements were studied twice before and after deafferentation. Surgical procedures were performed under pentobarbital anesthesia and all animal care complied with the guidelines of the Johns Hopkins Medical School and the National Institutes of Health. Each animal was implanted with a head holder and binocular scleral search coils. Proprioceptive afference was eliminated by sectioning the ophthalmic division of the trigeminal nerve (Porter et al. 1983) immediately distal to the gasserian ganglion. The ophthalmic division was identified at surgery anatomically and physiologically (since electrical stimulation produced a blink but no eye movement response), and the corneal reflex was absent throughout the postoperative period. The two eyes were deafferented in sequential surgical procedures separated by approximately 2 weeks. Experimental protocols and data analysis All of the experiments described, except where specifically noted, were performed in monkeys M1 and M2 before and after bilateral deafferentation of the EOMs. One experiment, disconjugate position-dependent adaptation of saccades and alignment with a prism combination, was also performed in a third monkey (M3) after bilateral deafferentation. Alignment Horizontal ocular alignment during steady monocular fixation with the right eye (phoria) was measured for targets ranging from left to right 10°, in 5° increments. The size of the phoria was calculated as the difference between the horizontal positions of the right eye (RE) and left eye (LE), and the dependence of the phoria on orbital eye position was calculated by estimating the slope (with a linear regression) of the plot relating phoria size to the position of the fixating RE (see Fig. 1). The phoria was measured three or four times per week for 2 weeks before ocular deafferentation, and for 5 weeks after deafferentation. Position-independent adaptation of the phoria was studied by measuring ocular alignment (viewing monocularly with the RE) before and after the animal viewed binocularly in the light for 1 h while wearing a 20-D base-out prism over the left eye. Position-dependent adaptation of the phoria was studied by placing a combination of prisms over the left eye for 3 days. This consisted of a 2-D base-in prism, which covered the LE when it was directed more than 5° to the left, and a 2-D base-out prism,

351 the right and left, and the target moved during the saccade by 5° toward the starting position of the eye. 220 double-step saccades were used in the training session, and baseline saccades were measured before and after the training period for 5° and 10° rightward and leftward centrifugal saccades. For the position-dependent conjugate paradigm, the training period consisted of 250 double-step saccades limited to one direction and orbital position. The target jumped from 0° to right 5°, and then returned to 0° while the saccade was in progress. Before and after the training period, 5° rightward horizontal saccades were measured with starting positions of left 5°, 0°, and right 5°. Disconjugate saccade adaptation was produced with the 2-0-2 prism combination already described. The prisms produced a retinal disparity that varied with orbital eye position, and, based on past reports (Oohira and Zee 1992), it was expected that part of the alignment change required by the prisms would be incorporated into the saccades by altering the relative size of the pulse in each eye. Horizontal saccades were measured for rightward and leftward target jumps, viewing monocularly with the RE, prior to wearing the prism combination and after wearing it for 1 day and 3 days. The intrasaccadic change in ocular alignment was calculated as the RE pulse minus the LE pulse. Fig. 1 Top row: The phoria during monocular viewing with the right eye (RE), before and after bilateral deafferentation of the extraocular muscles. Postdeafferentation data were recorded 6 days after the second trigeminal nerve was sectioned. The difference in the horizontal position of the two eyes is plotted against the position of the fixating RE. Predeafferentation data is indicated by the open squares and dotted lines, and postdeafferentation data by the filled squares and solid lines, with the lines indicating a linear regression of the alignment data. Each icon represents the mean of 10–15 measurements ±1 SD. Middle row and bottom row show the effect of deafferentation on the size and slope of the RE viewing phoria in monkey M1. The size of the phoria was measured looking straight ahead, and the slope was calculated with a linear regression of the phoria plot. Icons to the left of the vertical line (day 0) indicate the mean values predeafferentation and the icons to the right of the line indicate the mean values from each recording session postdeafferentation (error bars indicate 1 SD for the phoria size data). Horizontal dotted lines are the mean value of the postdeafferentation data, and do not differ significantly from the predeafferentation means

Pursuit

which covered the LE when it was directed more than 5° to the right. No prism covered the central field of vision of the LE (Oohira and Zee 1992). This prism combination (referred to as a 2-0-2 prism) required that the eyes diverge on leftward gaze and converge on rightward gaze, and therefore provided a retinal disparity that depended on orbital eye position. The phoria viewing with the RE was measured before the prism combination was added, and after it was worn for 1 day and 3 days.

Pursuit was studied before and after deafferentation in M1 and after deafferentation in M2. Baseline pursuit was measured with a horizontal step-ramp stimulus. The animal viewed a fixation target straight ahead; the target jumped 3.8° to the left and then moved to the right at a constant velocity of 20°/s. The target passed through the fixation point after 190 ms, which eliminated saccades at the onset of the pursuit response. The beginning of the pursuit movement was determined with a velocity criterion (when eye velocity first exceeded 2°/s). Two aspects of the pursuit response were analyzed. The open-loop response was characterized as the mean acceleration of the eye during the first 100 ms of the pursuit movement (Takagi et al. 2000); the closed-loop gain was defined as the mean slow-phase eye velocity during the period of time 500–600 ms after the onset of the eye movement, normalized for the target velocity of 20°/s. Thirty pursuit trials were recorded prior to the adaptation paradigm. Then 150 adaptation trials were performed and the openloop response was remeasured using the paradigm described. For the adaptation trials, the same step-ramp paradigm was used, but the target velocity increased to 40°/s when it passed the midline fixation point. The eye was therefore exposed to a retinal slip velocity of 20°/s before it began to move, but was required to accelerate to 40°/s to pursue the target. Previous experiments with this paradigm in normal monkeys have demonstrated an increase in the initial acceleration of the eye, indicating an adaptive change in the open-loop pursuit response (Takagi et al. 2000).

Saccades

Vestibular

Saccades were recorded for horizontal target jumps from 0° to right or left 5°, 10°, and 15°. Data were collected during binocular and monocular (RE) viewing conditions. The amplitude of the saccadic pulse was determined by subtracting the eye position at the end of the pulse (defined as the point where eye velocity first drops below 45°/s) from the eye position at the start of the pulse (defined as the position where eye velocity first exceeds 20°/s). The conjugacy of the pulse in the two eyes was calculated as the ratio of the RE pulse to the LE pulse, and the amplitude of the postsaccadic drift was defined as the size of the eye movement that occurred in the first 200 ms after the end of the pulse. Saccades were recorded three or four times per week for 2 weeks prior to deafferentation and for 5 weeks after deafferentation. Conjugate saccade adaptation was induced with a double-step paradigm (Deubel et al. 1982). In the position-independent conjugate paradigm, animals made 5° and 10° centrifugal saccades to

Vestibular eye movements were recorded in the dark during sinusoidal oscillations about the earth-vertical (yaw) axis. Sinusoids had a frequency of 0.2 Hz and a peak velocity of 42°/s. The eye movement traces were manually desaccaded with an interactive program, and the slow-phase eye velocity data were fit to a sinusoid with the least-mean-squared error. The gain and phase of the vestibular response were calculated by comparing the amplitude and timing of the eye movement responses with those of the chair.

352

Results Ocular alignment Baseline alignment Prior to deafferentation, the static horizontal alignment measured during monocular viewing with the RE (phoria) was mildly exophoric (eyes slightly diverged) in both animals (Fig. 1, top row, open squares). The phoria was only minimally dependent on orbital eye position, as the slope of the phoria plot was generally less than 0.05 in magnitude (Fig. 1, top row, dotted lines). Following bilateral deafferentation of the EOMs, no early changes occurred in the size (ANOVA, P>0.05) or the positiondependence (t-test on regression slopes, P>0.05) of the phoria in either animal (Fig. 1, top row, filled squares, solid lines). No gradual changes occurred in the size or the position-dependence (slope) of the phoria in the 5-week period that followed bilateral deafferentation (Fig. 1). Adaptation of alignment Position-independent adaptation of the static alignment was induced with a 20-D base-out prism worn over the LE for 1 h, and required an ocular convergence of approximately 11° to maintain bifoveal fixation. Prior to deafferentation, the RE viewing phoria converged by approximately 9° in monkey M1 and 10° in monkey M2 after the period of adaptation (Fig. 2, top row, open diamonds). Bilateral proprioceptive deafferentation did not affect the size of the adaptive change in the phoria induced by the prism (Fig. 2, top row, filled diamonds; ANOVA, P=0.12 for M1, P=0.29 for M2). Position-dependent adaptation of the static alignment was studied by measuring the effect of the 2-0-2 prism combination on the RE viewing phoria. Monkey M2 failed to adapt significantly with this paradigm before or after deafferentation. Monkey M1 did adapt, but the response to the paradigm was variable, so the pre- and postdeafferentation results could not be compared quantitatively. The postdeafferentation results in M1, however, demonstrate that the adaptive change in the RE viewing phoria after wearing the prism combination for 3 days closely followed the position-dependent disparity induced by the prisms, and similar findings were obtained in a third monkey (M3) studied after bilateral deafferentation (Fig. 2, bottom row). The results indicate that after deafferentation, these animals could adaptively tailor their ocular alignment to specific orbital eye positions. A comparison of these animals with normal monkeys studied previously in our laboratory with the identical protocol (Oohira and Zee 1992) suggests that deafferentation did not substantially alter this form of positiondependent phoria adaptation.

Fig. 2 Top row: The horizontal RE viewing phoria, before (squares) and after (diamonds) phoria adaptation with a 20-D base-out prism. Open icons are predeafferentation and filled icons are postdeafferentation. The phoria was measured looking straight ahead, and each icon is the mean of 40–80 measurements ±1 SD. Bottom row: The change in the horizontal RE viewing phoria (RE – LE position) after wearing the 2-0-2 prism combination for 3 days. Filled squares show the change in the phoria measurements (day 3 prism minus preprism data) at different orbital positions, with each data point the mean of 6–12 measurements. Crosses indicate the position-dependent disparity induced by the prism combination, which was calculated as the change in the position of the LE produced by the prism combination during monocular LE viewing

Saccades Baseline saccades Before deafferentation, saccades in both monkeys were approximately orthometric and conjugate, were followed by minimal postsaccadic drift (Fig. 3, top row, Table 1), and displayed a normal relationship between pulse amplitude and peak velocity (Fig. 3, bottom row). In the first data set recorded after bilateral deafferentation, neither the mean nor the variance of these saccade parameters changed significantly in either animal (Table 1, Fig. 3). During the five-week period of study that followed deafferentation, there were no gradual changes in the amplitude or conjugacy of saccades, or in the size of the postsaccadic drift in either monkey (Fig. 4). Saccade adaptation Conjugate, position-independent saccadic adaptation was produced with a double-step paradigm. Prior to deafferentation, the training period resulted in a significant reduction (t-test, P<0.001) in saccadic pulse amplitude in both animals (Fig. 5, top row, position-independent). After deafferentation, the training protocol produced a significant reduction (t-test, P<0.001) in pulse amplitude in both animals, and the size of the adaptive change in the pulse produced by the double-step para-

353

Fig. 3 Top row: Characteristic saccades for monkey M1 viewing monocularly with the RE, before and after deafferentation. Saccades are from 0° to right 10°, and the position of the right (R) and left (L) eyes are illustrated, as is the vergence trace (R-L). Bottom row: Peak saccade velocity plotted against pulse amplitude for M1, before (open squares) and after (filled squares) deafferentation

Table 1 Saccade parameters for monkeys M1 and M2, before [Prop (+)] and after [Prop (–)] bilateral proprioceptive deafferentation. Data were acquired during monocular viewing with the right eye (REV) for the horizontal saccades indicated. The mean and 1 SD are shown for each saccade parameter. The postdeafferentation data were recorded 6 days after the second eye was deafferented. No significant changes occurred in the mean or the variability of these saccade parameters after bilateral deafferentation (RE right eye,LE left eye)

RE pulse M1, REV 0 to R10 Prop (+) Prop (–) 0 to R15 Prop (+) Prop (–) M2, REV 0 to L10 Prop (+) Prop (–) 0 to L15 Prop (+) Prop (–)

Fig. 4 Effect of deafferentation (vertical line, day 0) on the LE pulse (0 to left 15° saccades), RE/LE pulse-pulse ratio (0 to right 5° saccades), and postsaccadic drift in the LE (0->right 10° saccades). Data were recorded during monocular RE viewing in monkey M1, and icons represent means ± SD

LE pulse

RE/LE pulse

RE drift

LE drift

n

10.3±0.8 10.4±0.7

10.2±0.8 10.5±0.7

1.01±0.01 0.99±0.02

–0.5±0.1 –0.5±0.1

0.0±0.1 –0.1±.15

19 20

15.0±0.6 14.9±0.8

15.1±0.5 15.0±0.7

0.99±0.01 0.99±0.02

–0.5±0.1 –0.6±0.1

0.0±0.2 0.0±.1

20 19

–11.7±0.7 –11.6±0.6

–12.1±0.6 –11.9±0.7

0.97±0.02 0.97±0.02

0.4±0.1 0.5±0.1

1.1±0.1 1.3±0.2

30 19

–14.9±1.0 –14.9±0.7

–15.2±1.0 –15.4±0.8

0.98±0.02 0.97±0.02

0.6±0.1 0.5±0.1

1.3±0.2 1.2±0.1

30 22

digm was independent of the proprioceptive state (ANOVA, P=0.36). Position-dependent, conjugate saccadic adaptation was induced with a double-step paradigm, where the training saccades were limited to one orbital position and direction (0 to right 5°). Prior to deafferentation, the training procedure reduced the pulse gain significantly (t-test, P=0.002) in monkey M1 for the three saccades that were tested (Fig. 5, top row, position-dependent), and the change in gain was greatest at the site of training (0 to right 5°). In this animal, the training session in-

duced a significant dependence of the pulse size on orbital eye position (ANOVA, P=0.01). After deafferentation in monkey M1, the adaptation paradigm similarly reduced the pulse gain (t-test, P=0.03) and produced a significant dependence of the pulse size on orbital eye position (ANOVA, P=0.02). Deafferentation did not affect the position dependency of the adaptive change in pulse size produced by the training paradigm in monkey M1 (ANOVA, P=0.7). Monkey M2 responded to this training paradigm with a position-independent change in pulse gain before and after deafferentation.

354

Fig. 5 Top row: Conjugate saccade adaptation using a double-step paradigm. The percent change in pulse gain is indicated for the position-independent and position-dependent paradigms, before (open squares) and after (filled squares) deafferentation. For the position-independent paradigm, the icons represent the mean gain change (n=30–50) for the four saccades studied (0 to right 5°, right 10°, left 5°, and left 10°). For the position-dependent paradigm, the icons represent rightward saccades in different orbital positions (n=10–20). Bottom row: The change in ocular alignment produced by the 2-0-2 prism combination in monkeys M1 and M3 after deafferentation. Filled bars (PP) are the change in the intrasaccadic alignment of the eyes (RE pulse–LE pulse) recorded during monocular viewing with the RE, that was produced after wearing the prism combination for 3 days. Leftward saccades became divergent and rightward saccades became convergent for the gaze shifts indicated at the bottom. Open bars (PH) indicate the change in RE viewing phoria associated with the same shifts in gaze, and the hatched bars (PR) indicate the change in retinal disparity produced by the prism combination for these shifts in gaze. Each icon represents the mean of 15–20 measurements

Disconjugate saccadic adaptation was elicited with the 2-0-2 prism combination described. The two animals that developed a position-dependent change in phoria after deafferentation (M1, M3) also developed disconjugate horizontal saccades, such that the eyes diverged during leftward saccades and converged during rightward saccades (Fig. 5, bottom row, solid bars). The intrasaccadic change in alignment paralleled, but was generally less than the alignment change required to maintain bifoveal fixation when making a horizontal gaze shift (Fig. 5, bottom row, hatched bars). The extent of the disconjugate saccadic adaptation varied considerably in monkeys M1 and M3 when the paradigm was repeated pre- and postdeafferentation, so it was difficult to precisely quantify any change in response that may have resulted from deafferentation. A comparison of the postdeafferentation results in monkeys M1 and M3 (Fig. 5, bottom row) to those of normal monkeys studied in our laboratory with the same paradigm (Oohira and Zee 1992), however, suggests that this form of disconjugate, posi-

Fig. 6 Pursuit position and velocity traces for monkey M1, preand postdeafferentation. Preadaptation eye movements are indicated by the dark dotted lines, postadaptation responses by the light dotted lines, and the target is indicated by the solid lines

tion-dependent saccadic adaptation was not substantially influenced by the presence or absence of ocular proprioception. Pursuit Baseline pursuit Horizontal pursuit was studied before and after deafferentation in M1 and after deafferentation in M2. Prior to deafferentation, pursuit movements in M1 had normal position and velocity profiles (Fig. 6, dark dotted lines), a mean closed-loop gain of 0.98, and an mean open-loop acceleration of 316°/s2. After deafferentation, the eye movement waveforms did not change (Fig. 6, dark dotted lines), and no significant changes occurred in the closed-loop gain (t-test, P=0.67) or the open-loop acceleration (t-test, P=0.78; Fig. 7). The pursuit parameters measured in monkey M2 after deafferentation were similar to those recorded in M1 (Fig. 7) and to those measured in normal monkeys using the identical protocol (Takagi et al. 2000). Pursuit adaptation Prior to deafferentation in M1, the adaptation protocol produced a significant (t-test, P=0.005) increase in the acceleration of the eye during the initial 100 ms of the eye movement (Fig. 7, right panel). This caused the eye

355

the initial acceleration remained significant (Fig. 11; t-test, P=0.001). The magnitude of the change in openloop acceleration produced by the adaptation paradigm was independent of the proprioceptive state in M1 (ANOVA, P=0.77). Monkey M2, studied only after deafferentation, had a significant increase in initial acceleration following the adaptation paradigm (Fig. 7; t-test, P=0.018), which was similar in magnitude to the adaptive change recorded in M1 and to that observed in normal monkeys studied with this protocol (Takagi et al. 2000). Vestibular

Fig. 7 Baseline closed-loop pursuit gain (left) and open-loop initial acceleration (middle) in the two animals, before (filled bars) and after (open bars) deafferentation. The means and one SD are illustrated for 15–25 pursuit movements. The right panel illustrates the percentage change in open-loop acceleration produced by the adaptation paradigm

The VOR was recorded in both monkeys before and after deafferentation. As illustrated in Fig. 8 for M2, deafferentation had no affect on the eye position or velocity waveforms recorded during sinusoidal rotation. The VOR gain and phase lead before deafferentation were 0.85, 36° (M1) and 0.81,40° (M2), and neither the gain nor the phase were affected by deafferentation in either animal (P>0.05).

Discussion The results demonstrate that bilateral deafferentation of the EOMs does not affect ocular alignment, saccades, pursuit, or the VOR in rhesus monkeys. Eye movements did not change acutely after deafferentation, nor were there gradual changes in saccades or alignment over a 5-week period of study. Visually mediated adaptation of alignment, saccades, and pursuit were also unaffected by proprioceptive deafferentation, and, more specifically, adaptation that altered the conjugacy of eye movements or induced changes tailored to specific orbital positions was unaffected by deafferentation. On-line ocular motor control

Fig. 8 Horizontal VOR of monkey M2 during sinusoidal rotation about the earth-vertical yaw axis (frequency 0.2 Hz, peak velocity 42°/s), recorded in the dark. The position and velocity of the right eye is shown before and after bilateral proprioceptive deafferentation. Quick-phase velocities are clipped for clarity

position and velocity to overshoot those of the target, and a corrective saccade was needed to refoveate the target (Fig. 6, light dotted lines). After deafferentation, the adaptive changes in the eye movements were qualitatively similar (Fig. 6, light dotted lines) and the change in

Proprioceptive afferents in the pigeon encode eye position and velocity during passive eye motion (Fahey and Donaldson 1998) and such kinematic information could potentially contribute to the immediate, on-line control of eye movements. We found, however, that bilateral proprioceptive deafferentation of the EOMs did not alter ocular motor function in rhesus monkeys. This finding confirms and extends the results of prior deafferentation studies in primates (Guthrie et al. 1983; Lewis et al. 1994, 1999) and suggests that proprioception does not play a significant role in the on-line control of eye movements in the monkey. Similar conclusions were also reached in human studies of spatial localization after strabismus (Steinbach and Smith 1981) or enucleation (Steinbach 1986) surgery, as the putative proprioceptive effects were not immediately apparent but rather developed slowly over a period of days.

356

There are several alternate explanations for our findings, however, which must be considered. Due to an obligatory period of recovery following surgery, eye movements were not recorded until 6 days after the second eye was deafferented. Hence, it is possible that transient changes occurred after deafferentation but were no longer evident 6 days later. This possibility could only be refuted with a less invasive method of deafferentation, which would allow eye movement measurements acutely after proprioception was removed. It should be noted, however, that the changes in the VOR of the pigeon and rabbit that occurred after deafferentation were not transient, but rather persisted for the duration of the respective studies (Kashii et al. 1989; Hayman and Donaldson 1995). A second, related possibility is that changes in ocular motor function did occur after deafferentation, but were completely compensated for by retinal afference (Weir et al. 2000). Data were recorded in both binocular and monocular viewing conditions in our study, and the nonviewing eye did not receive any immediate visual feedback in the later condition. The animals viewed binocularly during the postoperative recovery period, however, so we cannot exclude the possibility that visually mediated adaptation occurred during that period which eliminated the putative deficits produced by ocular deafferentation. In a prior study in rhesus monkey, however, the deafferented eye was not allowed any visual afference and no changes were observed in horizontal saccades or ocular alignment (Lewis et al. 1994). Furthermore, the changes in vertical eye movements that occurred after deafferentation in monkeys with vertical muscle pareses were not corrected by chronic binocular vision (Lewis et al. 1994). Together, these findings suggest that adaptation driven by retinal cues did not compensate for deafferentation-induced deficits in the current study. More generally, it is unlikely that eye movement changes akin to the marked disruption of the pigeon VOR (Hayman and Donaldson 1995) occurred in the monkey after deafferentation, since it is doubtful that such extreme deficits could be readily corrected by visually mediated adaptation. A third possibility is that proprioception and efference copy normally provide identical, redundant information about eye movement parameters, and that both are used in on-line ocular motor control. This hypothesis suggests that proprioception contributes to the control of eye movements, but that deafferentation has no effect because the redundant information derived from the efferent command remains available. This formulation cannot be refuted on the basis of our results. Due to the inherent noise in all biological signals, however, if proprioception and the efferent copy provide redundant information, one would expect that the variability in eye movements would be reduced by the convergence of two separate signals coding identical parameters (as has been described in other neural systems; Meredith and Stein 1986). The finding that deafferentation does not increase the variable error of eye movement parameters or point-

ing responses (Lewis et al. 1998) therefore suggests that proprioception does not provide redundant information used for on-line ocular motor control. Another consideration is the assertion that sectioning the ophthalmic division of the trigeminal nerve produces a complete or nearly complete proprioceptive deafferentation of the EOMs. The anatomic studies of Porter et al. (1983) indicate that in rhesus monkey all afferent information from the EOMs enters the brain via V1, but these findings have recently been questioned by Gentle and Ruskell (1997), who suggested that proprioceptive afference may enter the brain stem in the ocular motor nerves. The cell bodies of the afferent nerves that innervate the EOMs are located in the trigeminal ganglion (Porter 1986; Billig et al. 1997), however, so section of V1 immediately distal to the ganglion would eliminate most or all of the afferent innervation of the EOMs, even if some afferent fibers cross back to the ocular motor nerves proximal to the ganglion. Finally, our experiments measured eye movements within the ocular range of left to right 20°. It is clear, however, that nonlinearities in the ocular motor plant increase markedly beyond 30°s of eccentricity (Porrill et al. 2000). If proprioception contributes to the central adjustment of eye muscle innervation that is needed to compensate for these mechanical nonlinearities, then this effect may not be observable within the smaller ocular range we examined. Visually mediated adaptation Interactions between visual and proprioceptive signals have been documented in the visual system at the level of the lateral geniculate (Lal and Friedlander 1990) and visual cortex, and are critical for the development of cortical properties such as orientation selectivity and stereopsis (reviewed by Buisseret, 1995). It has been suggested that similar interactions between retinal and extraretinal afference are necessary for visually mediated adaptation of eye movements. In particular, passive eye motion appears to influence adaptation of pursuit (van Donkelaar et al. 1997) and produces persistent changes in ocular alignment (Gauthier et al. 1994). Our results demonstrate, however, that visually mediated adaptation of alignment, saccades, and pursuit does not depend on coincident retinal and extraretinal afferent signals, since deafferentation had no clear effect on any of the adaptation paradigms studied. These included several protocols designed to alter the conjugacy or position-dependence of eye movements. It should be noted that we did not examine the rate of adaptation in this study, but rather focused on the magnitude of the adaptive change produced by the different paradigms. While it remains possible that deafferentation altered the time course of adaptation without modifying the final size of the adaptive change, our results strongly suggest that vision and proprioception can function independently in the adaptive control of eye movements in the monkey.

357

Long-term calibration of eye movements It has been hypothesized that proprioception contributes to the control of eye movements via long-term, parametric calibration of the efference copy signal (Ludvigh 1952; Jurgens et al. 1981). Our results, which suggest that the efferent command provides adequate information to the brain for accurate eye movement control, is consistent with this proposal. Specifically, in normal animals such as those studied here, the efference copy position and velocity signals should closely coincide with the actual motion of the eye in the orbit, which is sensed by the proprioceptive organs. If proprioception provides an error signal indicating a discrepancy between the intended and actual motion of the eye, then deafferentation should not affect eye movements when such a discrepancy is negligible. In contrast, if the ocular motor plant is altered during development or following perturbations, the efferent command would no longer reflect the true motion of the eye. In this setting, deafferentation would eliminate the putative error signal which would lead to changes in eye movements, as we previously demonstrated in monkeys with eye muscle pareses (Lewis et al. 1994, 1999). These findings parallel those observed in spatial localization experiments, as deafferentation of normal animals did not affect pointing accuracy (Lewis et al. 1998), but dissociation of the proprioceptive and efferent signals did modify reaching behavior (Gauthier et al. 1990; Lewis and Zee 1993). We suggest, therefore, that proprioception contributes to the long-term calibration of the efference copy and that this calibration may be driven in part by an error signal derived from a mismatch between the efferent command and the proprioceptive feedback that reflects the true motion of the eye. Acknowledgements We thank A. Lasker, C. Bridges, P. Kramer, and D. Roberts. Supported in part by National Institutes of Health grants NS-01656 (to R.F.L.) and EY-01849 (to D.S.Z.).

References Ashton JA, Boddy A, Donaldson IML (1984) Directional selectivity in the responses of units in cat primary visual cortex to passive eye motion. Neuroscience 13:653–662 Billig I, Buisseret C, Buisseret P (1997) Identification of nerve endings in cat extraocular muscles. Anat Rec 248:566–575 Buisseret P (1995) Influence of extraocular muscle proprioception on vision. Physiol Rev 75:323–338 Daunicht WJ, Dieringer N (1986) Extraocular muscle proprioceptive signals affect ocular motor activity neither directly nor parametrically in the presence of optokinetic or vestibular stimulation in the frog. Exp Brain Res 64:535–540 Deubel H, Wolf W, Hauske G (1982) Corrective saccades: effect of shifting the saccadic goal. Vision Res 22:353–364 Donaldson IML, Knox PC (2000) Afferent signals from the extraocular muscles affect the gain of the horizontal vestibulo-ocular reflex in the alert pigeon. Vision Res 40:1001–1011 Donaldson IML, Long AC (1980) Interactions between extraocular proprioceptive and visual signals in the superior colliculus of the cat. J Physiol (Lond) 298:85–110

Donkelaar P van, Gauthier GM, Blouin J, Vercher J-L (1997) The role of ocular muscle proprioception during modification in smooth pursuit output. Vision Res 37:769–774 Fahey FL, Donaldson IML (1998) Signals of eye position and velocity in first order afferents from pigeon extraocular muscles. Vision Res 38:1795–1804 Gauthier GM, Nommay D, Vercher J-L (1990) The role of ocular muscle proprioception in visual localization of targets. Science 249:58–61 Gauthier GM, Vercher J-L, Zee DS (1994) Changes in ocular alignment and pointing accuracy after sustained passive rotation of one eye. Vision Res 34:2613–2627 Gentle A, Ruskell G (1997) Pathway of the primary afferent nerve fibres serving proprioception in monkey extraocular muscles. Ophthalmic Physiol Opt 17:225–231 Guthrie BL, Porter JD, Sparks DL (1983) Corollary discharge provides accurate eye position information to the oculomotor system. Science 221:1193–1195 Hayman MR, Donaldson IML (1995) Deafferentation of pigeon extraocular muscles disrupts eye movements. Proc R Soc Lond B Biol Sci 261:105–110 Jurgens R, Becker W, Kornhuber HH (1981) Natural and drug-induced variations of velocity and duration of human saccadic eye movements: evidence for control of the neural pulse by local feedback. Biol Cybern 39:87–96 Kashii S, Matsui Y, Honda Y, Ito J, Sasa M, Takaori S (1989) The role of extraocular proprioception in vestibulo-ocular reflex of rabbits. Invest Ophthalmol Vis Sci 30:2258–2264 Keller EL, Robinson DA (1971) Absence of a stretch reflex in extraocular muscles of the monkey. J Neurophysiol 34:908– 919 Knox PC, Donaldson IML (1993) Do extraocular muscle afferent signals play a role in the human vestibulo-ocular reflex? Soc Neurosci Abstr 19:858 Knox PC, Weir CR, Murphy PJ (2000) Modification of visually guided saccades by a nonvisual afferent feedback signal. Invest Ophthalmol Vis Sci 41:2561–2565 Lal R, Friedlander MJ (1990) Effect of passive eye movement on retino-geniculate transmission in the cat. J Neurophysiol 63: 523–538 Lewis RF, Zee DS (1993) Abnormal spatial localization with trigeminal-oculomotor synkinesis: evidence for a proprioceptive effect. Brain 116:1105–1118 Lewis RF, Zee DS, Gaymard BM, Guthrie BL (1994) Extraocular muscle proprioception functions in the control of ocular alignment and eye movement conjugacy. J Neurophysiol 72:1028– 1031 Lewis RF, Gaymard BM, Tamargo RJ (1998) Efference copy provides the eye position information required for visually guided reaching. J Neurophysiol 80:1605–1608 Lewis RF, Zee DS, Goldstein HP, Guthrie BL (1999) Proprioceptive and retinal afference modify postsaccadic ocular drift. J Neurophysiol 82:551–562 Ludvigh E (1952) Possible role of proprioception in the extraocular muscles. Arch Ophthalmol 48:436–441 Meredith MA, Stein BE (1986) Visual, auditory, and somatosensory convergence on cells in superior colliculus results in multisensory integration. J Neurophysiol 56:640–653 O’Keefe LP, Berkley MA (1991) Binocular immobilization induced by paralysis of the extraocular muscles of one eye: evidence for an inter-ocular proprioceptive mechanism. J Neurophysiol 66:2022–2033 Oohira A, Zee DS (1992) Disconjugate ocular motor adaptation in rhesus monkey. Vision Res 32:489–497 Porrill J, Warren PA, Dean P (2000) A simple central law generates Listing’s positions in a detailed model of the extraocular muscle system. Vision Res 40:3743–3758 Porter JD (1986) Brainstem terminations of extraocular muscle primary afferent neurons in the monkey. J Comp Neurol 247: 133–143 Porter JD, Guthrie BL, Sparks DL (1983) Innervation of monkey extraocular muscles: localization of sensory and motor neu-

358 rons by retrograde transport of horseradish peroxidase. J Comp Neurol 218:208–219 Roll R, Velay JL, Roll JP (1991) Eye and neck proprioceptive messages contribute to the spatial coding of retinal input in visually oriented activities. Exp Brain Res 85:423–431 Ruskell GL (1999) Extraocular muscle proprioceptors and proprioception. Prog Retinal Eye Res 18:269–291 Steinbach MJ (1986) Inflow as a long-term calibrator of eye position in humans. Acta Psychol 63:297–306

Steinbach MJ, Smith DR (1981) Spatial localization after strabismus surgery: evidence for inflow. Science 213:1407– 1409 Takagi M, Zee DS, Tamargo RJ (2000) Effects of lesions of the oculomotor cerebellar vermis on eye movements in primates: smooth pursuit. J Neurophysiol 83:2047–2062 Weir CR, Knox PC, Dutton GN (2000) Does extraocular muscle proprioception influence oculomotor control? Br J Ophthalmol 84:1071–1074