 |
INTRODUCTION |
Optimal vision requires that the eyes be
stationary immediately after saccades, the rapid eye movements used to
redirect gaze. Recordings of activity from ocular motoneurons indicate
that a characteristic pattern of innervation is responsible for
postsaccadic eye stability: the slide, a decaying exponential of
innervation that is thought to compensate for relaxation of the orbital
visco-elastic forces (Collins et al. 1975
; Fuchs
and Luschei 1970), and the step, the tonic innervation that
opposes the elastic recoiling forces of the orbital tissues
(Robinson 1970). To minimize postsaccadic drift, these
neural signals must be matched correctly to the saccadic pulse, and in
the long-term, they must be adjusted to compensate for the mechanical
changes in the oculomotor plant that occur during growth and aging and
for acute pathological changes, such as extraocular muscle pareses. As
these changes may not affect the two eyes symmetrically, the adaptive
mechanism should be capable of modifying the motor output differently
for each eye (disconjugate adaptation) rather than just
changing the innervation equally for both eyes (conjugate adaptation).
Two afferent signals could indicate the presence of postsaccadic drift
and drive adaptive modification of the slide and step of innervation:
image motion on the retina (retinal slip) and proprioceptive input from
the spindles and tendon organs of the extraocular muscles. Retinal slip
clearly has a prominent effect on conjugate and disconjugate adaptation
of postsaccadic drift. In subjects with unilateral extraocular muscle
palsies, chronic monocular viewing with the paretic eye leads to the
suppression of drift in that eye and to the induction of drift in the
opposite direction in the normal, covered eye ("conjugate"
adaptation) (Abel et al. 1978; Kommerell et al.
1976; Optican and Robinson 1980). Chronic
binocular viewing can result in suppression of drift in the paretic eye
without producing drift in the normal eye ("disconjugate"
adaptation) (Viirre et al. 1988). In these experiments,
the paretic eye drifted after saccades, so both retinal and
proprioceptive afferents potentially could have contributed information
about eye motion to the brain. Modification of the retinal signal
alone, however, was sufficient to adaptively alter postsaccadic drift.
The extraocular muscles of primates contain abundant muscle spindles
and tendon organs (Lukas et al. 1994; Ruskell
1978), and proprioceptive afferents carry signals that encode
both the position and the velocity of the eyes (Fahy and
Donaldson 1998). Although we reported that proprioception
contributes to the regulation of the amplitude of the saccadic pulse
and the static ocular alignment in animals with unilateral vertical
muscle pareses (Lewis et al. 1994), little is known
about the possible role of proprioceptive afference in the control of
eye motion immediately after saccades. A potential role for
proprioception is suggested by the finding that manipulation of
proprioceptive afferents by passively rotating the eye modifies
activity of Purkinje cells in the flocculus (Kimura and Maekawa
1981; Miyashita 1984), a region of the
cerebellum that is crucial for the adaptation of postsaccadic drift
(Optican et al. 1986; Zee et al. 1981).
In the current study, we have examined the effects of modifying visual
afference and of deafferenting the extraocular muscles on postsaccadic
drift in monkeys with unilateral weakness of a vertical extraocular
muscle. The purpose was to analyze the respective roles of retinal and
proprioceptive afference on the regulation of postsaccadic drift.
| |
METHODS |
General experimental procedures
The movements of both eyes were recorded using the magnetic
field search coil technique in three juvenile rhesus monkeys with unilateral vertical eye muscle pareses. The output signal of the coil
system was filtered at 90 Hz, sampled at 250 Hz, and saved to a digital
computer. Coil system resolution was ~0.05°. Targets were presented
on a bar that was oriented vertically or horizontally and that was
located 1.47 m in front of the animal's head. Light-emitting diodes (LEDs) were spaced at 2.5° intervals along the bar. Eye movements were recorded in total darkness, except for the illuminated LED target.
A vertical and horizontal calibration for the position of each eye was
obtained for each recording session by having the animal monocularly
fixate a series of LEDs in 2.5° increments, ranging from down 20 to
up 20° and from left 20 to right 20°. Vertical saccades then were
recorded during monocular viewing with the normal eye for target jumps
from 0 to up 10°, up 10° to 0, 0 to down 10°, and down 10° to
0. Horizontal saccades were recorded for target jumps from 0 to right
10°, right 10° to 0, 0 to left 10°, and left 10° to 0. During
the saccade paradigm, the animals fixated the LED target for 1.0 s
before the target moved to a new position. Thirty sets of vertical
saccades and 10 sets of horizontal saccades were recorded during each
experimental session, and data were acquired 3 days/wk in each
visuoproprioceptive state, beginning 3 days after the state was modified.
Surgical procedures
All surgical procedures were performed under pentobarbital
anesthesia (30 mg/kg iv), and all animal care complied with the Johns
Hopkins Medical School veterinary guidelines. Each animal was implanted
with a head holder and binocular scleral search coils (Judge et
al. 1980). In two animals (SO1 and SO2),
a superior oblique paresis was produced by sectioning the trochlear
nerve intracranially. In the third monkey (IR), the inferior
rectus muscle was weakened by sectioning its tendon.
Proprioceptive inputs from the paretic eye were eliminated at a later
date by sectioning the ophthalmic division of the trigeminal nerve
immediately distal to the Gausserian ganglion (Porter et al.
1983). The ophthalmic division of the trigeminal nerve was identified at surgery anatomically and physiologically (with electrical stimulation, which evoked a blink but no eye movement), and the corneal
reflex was absent throughout the postoperative period. It has been
suggested that a portion of the afferent innervation of the extraocular
muscles travels to the brain stem in the ocular motor nerves
(Gentle and Ruskell 1997) rather than the trigeminal nerve. Nevertheless the cell bodies of the afferent neurons that innervate the extraocular muscles are located in the trigeminal ganglion (Billig et al. 1997; Porter
1986) so that section of the ophthalmic division of the
trigeminal nerve immediately distal to the ganglion would deafferent
the extraocular muscles even if some sensory fibers cross to the ocular
motor nerves proximal to the ganglion.
Experimental protocol
After the vertical muscle paresis was induced, the paretic eye
was covered immediately with an opaque patch, and the animals viewed
monocularly with the normal eye for 4 wk. For monkeys IR and
SO1 (see Table 1), the patch
then was removed and the animals were viewed binocularly for 2 wk. A
base-down wedge prism, with strength that approximated the size of the
vertical misalignment with the normal eye straight ahead, then was
placed in front of the paretic eye for 2 wk to promote binocular
fusion. For monkey SO2, the opaque patch was replaced with a
base-down prism for 2 wk. Monkey SO1 subsequently had its
paretic eye covered with an opaque patch for 2 wk to allow deadaptation
from the binocular/prism state, and then the patch was switched to the
normal eye for 2 wk, forcing it to view monocularly with the paretic
eye. At the completion of these experiments, the paretic eye of each
animal was covered with the opaque patch for 2 wk. The paretic eye then was deafferented proprioceptively, and the preceding protocol was
repeated.
Data analysis
Data were analyzed off-line with an interactive computer
program. The position of each eye was calibrated with a third-order polynomial linearization program to compensate for any nonlinearities in the search coil signal. The amplitude of the saccadic pulse was
determined by subtracting the eye position at the end of the pulse (p),
defined as the position at which eye velocity first dropped <45°/s,
from the eye position at the start of the saccade (the point at which
eye velocity 1st exceeded 20°/s). The step position (s) for each eye
was determined as the position of the eye when eye velocity returned to
a steady value of zero, before the subsequent saccade. The vertical
"vergence" angle (V) was defined as the vertical
position of the paretic eye minus the vertical position of the normal
eye, and the vertical retinal disparity was defined as the vertical
retinal error of the paretic eye minus the vertical retinal error of
the normal eye.
For drift waveforms that were monophasic (Fig.
1, monkey IR), the amplitude
of the postsaccadic drift was defined as (s-p). For drift waveforms
that had more than one component (Fig. 1, monkey SO1), the
amplitude of each drift component was measured by placing marks at the
inflection points in the drift waveform, defined as the points where
eye velocity changed sign. Because the amplitude of the saccadic pulse
in the paretic eye changed after deafferentation (Lewis et al.
1994) and alterations in pulse size potentially could affect
the amplitude of the postsaccadic drift (Optican and Miles
1985), each measurement of drift amplitude was normalized by
dividing it by the gain of the preceding pulse for that eye. The pulse
gain was defined as (pulse amplitude)/(target displacement).
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 1.
Representative vertical saccades made by monkeys IR and
SO1, normal-eye-viewing state, predeafferentation,
demonstrating the waveforms of the postsaccadic drift in the paretic
eye. Vertical axes indicate vertical eye position (in °) and
corresponding vertical eye velocity (°/s), but the traces have been
offset for clarity. P, pulse; s, step; x and d are inflection points in
the drift movement, as described in the text. Horizontal lines indicate
velocity of 0°/s.
|
|
To determine the disconjugacy of the changes in postsaccadic drift
induced by the visual and proprioceptive manipulations, the change in
the drift of the paretic eye (y axis) was plotted against
the change in the drift of the normal eye (x axis; see Fig.
2). If the change in drift was limited to
the paretic eye, the points for the four saccade types studied would be
located on the y axis; if the change was conjugate (equal in
the 2 eyes), the points would fall on the y = x line. The overall nearness of the data points to these two
lines is therefore a way to measure the disconjugacy of the change in
postsaccadic drift and was calculated for each animal by summing the
squared error of the four data points (corresponding to the 4 saccades
studied) about the y axis and the y = x line.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 2.
Plots of the average change in postsaccadic drift of the paretic eye
(PE) vs. the average change in the drift of the normal eye (NE) for
monkeys IR ( ) and SO1
( ), resulting from the transition from the
normal-eye-viewing (NEV) state to the binocular viewing (BEV) state
before deafferentation. Each icon represents the average change in
drift for 1 of the 4 types of saccades measured and is derived from all
of the saccade data recorded in the 2 states (as displayed in Table 2).
Drift measurements of the paretic eye are (s-p) for monkey
IR and (s-d) for monkey SO1. Drift in the normal
eye is (s-p) for both monkeys.
|
|
The average velocity of the drift of the paretic eye in the 100-ms
period after the initial, rapid postsaccadic movement (x-p, d-p) was quantified as the mean of the absolute value of the
eye velocity measured every 4 ms during this period. Eye velocity was
calculated by differentiating and filtering the position signal with a
seven-point Gaussian filter. When drift movements could be approximated
by an exponential, a time constant was determined by fitting a single
exponential (a + be
t/TC) to
the eye movement trace with the least-squared-error. Statistical analysis was performed with two-way ANOVA and Student's
t-test. Unless otherwise specified, all data presented below
are vertical eye movements recorded during monocular viewing with the
normal eye.
| |
RESULTS |
Effect of muscle paresis, normal-eye viewing
As previously reported, the unilateral vertical muscle
paresis produced a vertical misalignment of the eyes that increased with down gaze, and hypometric vertical saccades in the paretic eye
relative to the normal eye (Lewis et al. 1994). The
muscle paresis also resulted in vertical postsaccadic drift in the
paretic eye. The monkey with an inferior rectus tenotomy
(IR) displayed monophasic drift movements (s-p) in the
paretic eye in the direction opposite to that of the antecedent saccade
(Fig. 1, Table 2). The two monkeys with a
paresis of the superior oblique muscle (SO1 and
SO2) had vertical postsaccadic drift movements in the paretic eye that consisted of three components after upward saccades (x-p, d-x, s-d) and two components after downward saccades
(d-p, s-d) (Fig. 1, Table 2). In all three animals, the postsaccadic drift in the normal, viewing eye was small in amplitude and its waveform was monophasic (s-p).
Visually mediated adaptation before deafferentation
EFFECT OF BINOCULAR VISION WITHOUT A PRISM.
When the patch was removed from the paretic eye for monkeys
IR and SO1, the animals viewed binocularly but were not
able to fuse the images from the two eyes. The absence of fusion was
inferred from the persistence of a large vertical deviation of the eyes (which ranged from 4 to 15°) during binocular viewing (tropia), which
was equal to the deviation during monocular viewing (phoria). No
vertical fusional movements were observed when the animals fixated a
target in the dark or during spontaneous fixation in a lit room. The
animals generally fixated with the normal eye but occasionally
alternated the fixating eye and foveated the target with the paretic
eye for several seconds.
Binocular viewing resulted in a reduction of the amplitude of the
monophasic (s-p) drift movements in the paretic eye for monkey
IR and reduced the amplitude of the drift components in the
paretic eye that followed downward saccades (d-p, s-d) for monkey
SO1 (Table 2, Fig. 3;
t-test: P < 0.001 for the drift components
in both monkeys). In monkey SO1, binocular viewing did not
reduce the amplitude of the drift movements in the paretic eye that
followed upward saccades (Table 2).
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 3.
Representative saccades from up 10 to 0° for monkeys
IR and SO1 in the normal-eye-viewing (NEV,  )
and binocular-viewing (BEV, · · · ) states,
before deafferentation of the paretic eye. PE, paretic eye; NE, normal
eye; V, PE  NE (vertical vergence trace). Eye traces in this
figure (and all subsequent figures of saccade waveforms) are offset to
clearly present the drift movements and were recorded during monocular
viewing with the normal eye. The position of the normal eye at the end
of the saccade approximates the position of the target. Because the
paretic eye is higher than the normal eye in all animals and
visuo-proprioceptive conditions, the vergence (V)
values are always positive, and the direction of decreasing vergence
angle is downward. Calibration bars for the time and position axes are
located in the corner of each figure.
|
|
For both animals, only small changes occurred in the postsaccadic
drift of the normal eye in the binocular viewing condition (Table 2,
Fig. 3). Plotting the change in the drift of the paretic eye versus the
change in the drift of the normal eye for monkey IR (Fig. 2)
illustrates that chronic binocular viewing resulted in changes in drift
that were disconjugate and primarily affected the paretic eye. Because
the waveform of the postsaccadic drift in the paretic eye had several
components in monkey SO1 (and SO2), assessing the
conjugacy of the drift change induced by binocular viewing was less
straightforward than in monkey IR. As discussed in the
section describing the effect of viewing with the paretic eye, when the
normal eye was habitually patched in monkey SO1, that eye
developed drift directed upward. This suggests that the change in
postsaccadic drift in the normal eye resulted primarily from the
alterations in innervation that suppressed the slow, downward drift
movement in the paretic eye (s-d). Therefore the change in the (s-d)
component of the drift in the paretic eye was used to quantify the
conjugacy of the change in drift in monkeys SO1 and
SO2. Plotting the change in the (s-d) component of drift in
the paretic eye versus the change in drift in the normal eye for
monkey SO1 (Fig. 2) indicates that the changes associated with chronic binocular viewing were disconjugate but less so than in
monkey IR. The data points are closer to the y
axis than the y = x line for
SO1 but are further from the y axis than are the data points
for IR.
In summary, despite the apparent absence of fusion, chronic
binocular viewing resulted in a reduction in the amplitude of postsaccadic drift in the paretic eye for monkey IR (upward
and downward saccades) and SO1 (downward saccades). The
change in postsaccadic drift was disconjugate, as the reduction in the
drift of the paretic eye was accompanied by smaller changes in
the drift of the normal eye.
EFFECT OF BINOCULAR VISION WITH A PRISM.
Eye movements of all three monkeys were studied during a 2-wk period of
binocular viewing with a base-down prism in front of the paretic eye.
The strength of the prism for each animal approximated the size of the
vertical deviation when the normal eye viewed a target straight ahead,
thereby reducing the retinal disparity between the eyes and promoting
binocular fusion. As previously reported, binocular viewing with a
prism led to adaptive changes in the saccadic pulse (an increase in the
size of the pulse in the paretic eye) and adaptive changes in static
alignment (a decrease in the position-dependence of the phoria)
(Lewis et al. 1994).
In monkeys SO1 and SO2, chronic binocular viewing
with a disparity-reducing prism resulted in a reduction of the
amplitude of drift movements in the paretic eye that followed downward
saccades (d-p, s-d) and reduced the amplitude of the slow, downward
component of the drift (s-d) that followed upward saccades (Table 2,
Fig. 4; t-test:
P < 0.001 for these drift components in both animals). Little change occurred in the postsaccadic drift of the normal eye
(Fig. 5). These changes in drift reduced
the retinal slip of the paretic eye but increased the amplitude of the
postsaccadic fixation disparity in monkeys SO2 (Fig.
6) and SO1.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 4.
Representative saccades from 0 to up 10° for monkeys
IR and SO2 in NEV state () and binocular/prism
state (BEV/pr, · · · ) before deafferentation.
|
|
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 5.
Plots of the average change in the drift of the PE vs. the average
change in the drift of the NE, resulting from the transition from the
NEV state to the BEV/pr state, predeafferentation. Icons represent the
4 saccade types for monkeys IR (),
SO1 (), and SO2
( ) and are derived from all of the saccade data
recorded in the 2 states. Paretic eye drift measurements are (s-p) for
monkey IR and (s-d) for monkeys SO1 and
SO2.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 6.
Effect of the adaptation produced by chronic binocular viewing
with a disparity-reducing prism on the postsaccadic drift in the
paretic eye, and on the postsaccadic retinal disparity,
predeafferentation. Plotted are the average values of the PE drift
(s-p) and the postsaccade disparity for the 4 saccade types in
monkey SO2 ( ) and IR
(), PRE ( and ) and
POST ( and ) visually mediated
adaptation. Postsaccade disparity is determined at the step position
and is defined as (PE retinal error NE retinal error) while viewing
through the base-down prism. Changes in postsaccadic disparity
resulting from visually mediated adaptation are due to modifications in
both the saccadic pulse (Lewis et al. 1994) and the
postsaccadic drift.
|
|
In contrast, the changes in drift that occurred in the binocular
viewing/prism state in monkey IR always brought the eyes toward the alignment that allowed binocular fusion, but in some cases
large drift movements were induced in the paretic eye. For example,
because the effective strength of the prism increased with upward gaze
positions in monkey IR, saccades from 0 to up 10° required
an increase in the paretic eye hyperdeviation to maintain bifoveal
fixation. Chronic viewing in the binocular/prism condition resulted in
the induction of upward drift in the paretic eye for saccades from 0 to
up 10° (Table 2, Fig. 4), decreasing the postsaccadic retinal
disparity. For monkey IR, the amplitude of the drift in the
paretic eye, and the consequent retinal slip, increased in the
binocular viewing/prism state for three of the four saccades types
studied, although the postsaccadic retinal disparity was reduced in
each case (Fig. 6).
In summary, monkeys SO1 and SO2 had large drift
movements in the paretic eye but small postsaccadic disparities in the
binocular viewing/prism condition before adaptation, and chronic
binocular viewing with the disparity-reducing prism resulted in a
disconjugate reduction of the amplitude of the postsaccadic drift in
the paretic eye but an increase the size of the postsaccadic fixation
disparity (Fig. 6). Conversely, monkey IR had relatively
large postsaccadic disparities and small drift movements in the
binocular viewing/prism condition before adaptation and responded to
the chronic binocular viewing/prism state by reducing the fixation
disparity but generally increasing the amplitude of the drift in the
paretic eye (Fig. 6).
EFFECT OF VIEWING WITH THE PARETIC EYE.
In one monkey (SO1), eye movements were studied during a
2-wk period of viewing with the paretic eye. In this state, the retina of the paretic eye but not of the normal eye is exposed to the image
motion associated with postsaccadic drift. Chronic viewing with the
paretic eye caused a reduction of the amplitude of the postsaccadic
drift movements in the paretic eye and an induction of upward drift in
the normal, covered eye (t-test: P < 0.001 for the changes in both eyes; Table 2, Fig.
7). The upward direction of the drift in
the normal eye implies that it resulted from the innervational change
that reduced the slow, downward movement (s-d) in the paretic eye.
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 7.
Representative saccades for monkey SO1 in the NEV ()
and paretic-eye-viewing (PEV, · · · ) states,
predeafferentation. Left: saccade from 0 to up 10;
right: saccade from up 10 to 0.
|
|
Effect of deafferentation of the paretic eye
NORMAL-EYE-VIEWING CONDITION.
In all three animals, when the paretic eye was covered with an opaque
patch and that eye was deafferented proprioceptively, the vertical
postsaccadic drift of the paretic eye was modified by a "directional
bias" in the amplitude of the drift movements. This bias in
postsaccadic drift generally affected each component of the drift
waveform, increasing the amplitude of drift movements in the same
direction as the bias and decreasing the amplitude of drift movements
in the direction opposite the bias. The direction of the drift bias
appeared to be idiosyncratic, however, as it could be directed upward
or downward for different animals and saccade directions. For example,
in monkey IR, the monophasic postsaccadic drift in the
paretic eye was biased upward after deafferentation for both upward and
downward saccades [Figs. 8 and
9; t-test: P < 0.001 for (s-p) amplitude]. The amplitude of the drift movements
that followed downward saccades also was biased upward in
SO1 but was biased downward in SO2 (Figs. 8 and
9; t-test: P < 0.05 for each drift
component in both animals). The drift that followed upward saccades was
biased down for monkey SO1 (P < 0.001 for
each drift component) but was not directionally biased by
deafferentation in SO2 (Figs. 8 and 9).
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 8.
Representative saccades for the 3 monkeys in the NEV predeafferentation
() and postdeafferentation (· · ·) states.
Top: saccades from 0 to up 10°; bottom:
saccades from up 10 to 0°.
|
|
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 9.
Average postsaccadic drift in the PE and NE for each monkey in the NEV
state before () and after ()
deafferentation of the PE. Error bars indicate 1 SD. Displayed data are
derived from all of the saccades recorded in the 2 proprioceptive
states, and the number of saccades for each state is indicated in
Tables 2 and 6.
|
|
Before deafferentation, horizontal saccades for the three monkeys were
followed by a monophasic postsaccadic drift in the paretic eye that was
in the opposite direction of the antecedent saccade. The amplitude of
these horizontal postsaccadic drift movements was generally small
(<0.39° for 11 of the 12 saccades studied), but trigeminal nerve
section produced small changes in the amplitude of the horizontal drift
in the paretic eye (<0.2° for 11 of the 12 saccades studied) that
were significant statistically (t-test: P < 0.05).
The effect of deafferentation on the average velocity of vertical
postsaccadic drift in the paretic eye paralleled the effect on the
amplitude of the drift (Table 3, Fig. 9).
In monkey SO2, the net amplitude of the vertical
postsaccadic drift increased after deafferentation as did the mean
drift velocity. For monkeys IR and SO1, the net
amplitude and mean velocity of the drift increased for some saccades
(for example: IR,10 to >0; SO1,0 to >10) and decreased for other saccades (IR,0 to >10;
SO1,10 to >0) (Table 3, Fig. 9). The time constant of the
vertical drift movement, when it could be approximated by a negative
exponential [(s-p) for monkey IR; (s-d) for monkeys
SO1 and SO2), became lower after deafferentation of the
paretic eye (Table 4).
The changes in vertical postsaccadic drift that followed
deafferentation were largely limited to the paretic eye for each monkey; little change occurred in the amplitude of the postsaccadic drift of the normal eye (Fig. 10).
Furthermore the change in the drift of the paretic eye did not develop
gradually after deafferentation; it was evident when the first data
after deafferentation was recorded (3 days after trigeminal nerve
section), and it did not vary substantially over the subsequent 4-wk
period of data collection (Fig. 11).
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 10.
Plots of the average change in the postsaccadic drift of the PE vs.
that of the NE resulting from deafferentation of the paretic eye, NEV
condition. Drift is defined for these plots as the net change in eye
position after the pulse (s-p). Data are derived from all the saccades
recorded in the 2 states. , monkey IR;
, monkey SO1; and ,
monkey SO2.
|
|
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 11.
Postsaccadic drift in the PE for monkey SO1, down 10 to
0 saccades, NEV condition before and after deafferentation (indicated
by the vertical line at day 0). Displayed are the average amplitudes of
the drift components before deafferentation ±1 SD (large icons to the
left of the vertical lines), and the average amplitudes
recorded in each session after deafferentation (n = 20-30 saccades/session), plotted as a function of time after
deafferentation. Dotted horizontal lines indicate the average amplitude
of each drift component after deafferentation.
|
|
In summary, deafferentation altered the amplitude, velocity, and time
constant of the postsaccadic drift movements of the paretic eye. The
time constants typically were shorter after deafferentation in all
three animals, and the drift bias was directed upward for all saccade
types in monkey IR. No other consistent pattern to the
changes was evident in the three animals, and no systematic differences
occurred between animals with the two types of plant lesions.
Deafferentation did not consistently increase the amplitude of the
drift movements in the paretic eye, did not have a consistent effect on
the conjugacy or position-dependency of the postsaccadic drift, and did
not consistently alter the magnitude or position dependence of the
static ocular misalignment (Table 5).
VISUALLY MEDIATED ADAPTATION AFTER DEAFFERENTATION.
Although deafferentation of the paretic eye altered the drift in that
eye when it did not receive visual input, deafferentation did not
affect the changes in postsaccadic drift that were produced by altering
the viewing condition. After deafferentation, chronic binocular viewing
without a prism, binocular viewing with a prism, and chronic viewing
with the paretic eye resulted in the same pattern of changes in the
drift of the paretic eye as occurred predeafferentation (Table
6, Fig.
12). Analysis with a two-way ANOVA
indicated that the amplitude of each drift component in the paretic eye
was significantly affected by the visual (P < 0.001)
and proprioceptive (P < 0.01) states but that no
consistent interaction existed between the visual and proprioceptive
states.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 12.
Representative saccades for the 3 monkeys after deafferentation in the
4 visual states. , NEV condition; · · ·, BEV, BEV/pr, or PEV
conditions. Downward saccades are from up 10 to 0, and upward saccades
are from 0 to up 10.
|
|
The disconjugacy of the changes in drift produced by altering the
viewing condition was not affected by deafferentation in a consistent
manner. Before and after deafferentation, the two binocular viewing
conditions resulted in changes in the amplitude of the drift of the
paretic eye, but much smaller changes in the amplitude of the drift of
the normal eye (Figs. 2, 5, and 13). The summed square errors about the y (disconjugate) axis in
these diagrams were therefore small in both proprioceptive states (Fig. 14). The conjugacy of the change in
drift produced by chronic viewing with the paretic eye in monkey
SO1, assessed as the summed square error about the
y = x (conjugate) axis, was also similar
before and after deafferentation (pre: 0.31 deg2;
post: 0.55 deg2).
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 13.
Average change in the drift of the PE vs. that of the NE for the 4 saccade types in all 3 monkeys, resulting from the transition from the
NEV state to the BEV state or the BEV/pr state postdeafferentation.
Icons are derived from all the data recorded in these states. ,
monkey IR; , monkey SO1;
and , monkey SO2.
|
|
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 14.
Plots of the summed square errors (in deg2) for the 4 vertical saccade types, measured about the disconjugate
(y axis) and the conjugate (y = x line; see Fig. 2), resulting from the transition from
the NEV state to the BEV and BEV/pr states. Filled icons are before and
open icons are after deafferentation. Values are derived from the means
of all the data recorded in each visual and proprioceptive state.
|
|
| |
DISCUSSION |
Our results demonstrate several new findings regarding the roles
of retinal and proprioceptive afference in the control of postsaccadic
drift. When binocular fusion was not possible, retinal slip information
from both eyes was an adequate stimulus to promote disconjugate
adaptation of postsaccadic drift. When fusion was possible, either
retinal disparity or slip could be the dominant visual error signal
used to modify postsaccadic eye motion. In addition, proprioceptive
deafferentation of the paretic eye altered the amplitude, velocity, and
time constant of the postsaccadic drift of that eye but did not
influence visually mediated adaptation of postsaccadic drift.
Mechanism of postsaccadic drift
Stability of the eye after vertical saccades requires that the
torques applied to the globe by the four vertically acting extraocular
muscles compensate accurately for the relaxation of the orbital
visco-elastic forces (Goldstein 1983) and for the elastic recoiling forces of the orbital tissues (Robinson
1970). Postsaccadic drift therefore can occur if the torques
applied to the globe are altered, for example, by changing the strength of an extraocular muscle or by changing the mechanical effect of a
muscle on the globe or if the visco-elastic properties of the
oculomotor plant are modified. In our experimental paradigm, the
superior oblique muscle was paralyzed in two animals, and the
mechanical action of the inferior rectus muscle was reduced in one
animal. Both types of lesions alter the torques applied to the globe
during and after vertical saccades and also change the mechanical
properties of the orbital tissues.
The superior oblique and inferior rectus normally produce different
vertical torques on the globe, and the changes in the visco-elastic
properties of the plant associated with a paresis of the superior
oblique or a tenotomy of the inferior rectus also should be quite
different (Miller and Robinson 1984). Although the
waveform of the postsaccadic drift was the same in the two animals with
superior oblique pareses, the inferior rectus tenotomy and superior
oblique paresis produced different patterns of postsaccadic drift. It
is difficult, however, to correlate specific morphological features of
the composite drift waveforms with the mechanical consequences of the
two lesions (Inchingolo and Bruno 1994). The rapid drift
movements that immediately follow and are in the direction opposite to
the saccadic pulse (d-p, x-p), in particular, are of uncertain origin.
Similar movements are observed in normal humans and monkeys and are
referred to as dynamic overshoots (Kapoula et al. 1986).
These movements may be due to a braking pulse supplied by the brainstem
saccadic burst neurons (Van Gisbergen et al. 1981), or
they may represent a passive response of the ocular motor plant to the
pulse-slide-step of innervation (Goldstein 1987).
Visually mediated adaptation before deafferentation
DISCONJUGATE ADAPTATION WITHOUT BINOCULAR FUSION.
In the binocular viewing condition, postsaccadic drift in the paretic
eye was suppressed without inducing postsaccadic drift in the normal
eye, despite the absence of binocular fusion. These results indicate
that motion of images on the two retinas is an adequate stimulus to
promote disconjugate adaptation of postsaccadic drift and
that binocular fusion is not mandatory for this process. In contrast,
disconjugate adaptation of the saccadic pulse and static alignment
appears to require binocular foveal (Lewis et al. 1995;
Oohira and Zee 1992) or perifoveal (Kapoula et
al. 1996a, 1998) fusion. These findings are consistent with the
tenotomy studies of Viirre et al. (1988), in which
animals with large static deviations (that presumably did not allow
binocular fusion) were able to suppress postsaccadic drift in the
paretic eye with chronic binocular viewing but were not able to
disconjugately adapt their static misalignment or saccadic pulse dysmetria.
In normal human subjects, if postsaccadic retinal image motion is
presented to one eye while the other eye views a stationary, nonfusible
pattern, a conjugate pattern of postsaccadic drift adaptation occurs (i.e., changes in drift are approximately equal in
the 2 eyes) (Kapoula et al. 1990). These authors
suggested that binocular fusion is required for disconjugate adaptation of drift, although subsequent similar experiments using fusible dichoptic visual stimuli failed to induce substantial vertical disconjugate postsaccadic drift (Kapoula et al. 1996b).
The reason for the discrepancy between their results and our findings
is uncertain. Although the incongruency between retinal and
proprioceptive afferent signals in the paradigm used by Kapoula
et al. (1990) may be partially responsible for their findings,
we have demonstrated that ocular deafferentation does not inhibit
visually mediated disconjugate adaptation of drift. Thus this form of
oculomotor adaptation does not require a congruence between retinal and
extraretinal afferent signals.
DISCONJUGATE ADAPTATION WITH BINOCULAR FUSION.
When the animals viewed binocularly with the disparity-reducing prism,
two potentially conflicting adaptive stimuli were present after each
saccade. The paretic eye was exposed to postsaccadic image motion and a
potentially fusible retinal disparity was present. To optimize vision,
the animals ideally would minimize retinal slip by reducing the
postsaccadic drift of each eye and minimize retinal disparity by moving
the eyes toward the alignment where bifoveal fixation is achieved. The
two animals with superior oblique pareses (who had large drift
movements in the paretic eye but small postsaccadic disparities)
responded to the chronic binocular/prism state by suppressing drift in
the paretic eye; this led to an increase in the postsaccadic retinal
disparity. The animal with the inferior rectus tenotomy (who had large
postsaccadic disparities but small drift movements in the paretic eye)
responded by moving the eyes toward the alignment that allowed bifoveal
fixation, although this required the induction of drift in the paretic
and normal eyes for some saccade types.
These results suggest that postsaccadic retinal slip or retinal
disparity can be used to adaptively modify motion of the eyes after
saccades and imply that the larger of the two visual error signals may
dominate the pattern of adaptation. In addition, superior oblique
pareses are associated with more prominent cyclodeviation than are
pareses of the vertical recti, and the extrafoveal disparity present in
the two monkeys with superior oblique pareses could have limited their
ability to generate fusional vergence movements. Reduction of
postsaccadic retinal slip (which does not require fusion) therefore may
have been the primary objective in these animals. In contrast, the
monkey with the inferior rectus tenectomy probably lacked a substantial
cyclodeviation and hence may have made stronger fusional vergence
movements when the vertical deviation was reduced by the prism. In this
animal, bifoveal fixation may have been the principle goal, resulting
in the induction of postsaccadic drift that reduced the fixation disparity.
Previous work in primates suggests that adaptation of the slide and
step of innervation minimizes postsaccadic retinal slip but does not
correct conjugate or monocular retinal position errors at the end of
the saccade (Optican and Miles 1985; Optican and Robinson 1980). Although slow postsaccadic movements that bring the eye toward the visual target have been reported in cats, on the
basis of their dynamic characteristics, these movements most likely do
not result from the efferent (slide-step) signals that normally control
postsaccadic eye motion (Missal et al. 1993).
During disconjugate adaptation of saccades using optical devices such
as prisms (Oohira and Zee 1992) or anisometropic
spectacles (Lemij and Collewijn 1991), a retinal
disparity is present at the end of saccades, which promotes a
postsaccadic fusional vergence movement. In these studies, disconjugate
postsaccadic movements persisted during monocular viewing and moved the
eyes toward the alignment that was required for bifoveal fixation
during the binocular training. Slow, disconjugate movements therefore
can be induced adaptively to correct postsaccadic retinal disparity,
but postsaccadic drift movements are not induced adaptively in primates
to correct monocular or conjugate position errors. Because fusional
vergence movements occur in response to retinal disparity but are not
produced when the retinal positional errors are conjugate, these
findings suggest that the disconjugate eye movements observed in
monkey IR that reduced the postsaccadic retinal disparity
are due to adaptation involving the vertical vergence system. These
movements may result, for example, if the animal learned to preprogram
a corrective vergence movement with each vertical saccade (Ygge and Zee 1995).
Effect of deafferentation on postsaccadic drift
Proprioceptive afferents carry information about eye position and
velocity during passive eye motion (Fahy and Donaldson
1998) and could provide the brain with feedback signals that
encode these parameters during volitional eye movements. These afferent signals could function potentially in the immediate, on-line control of
eye movements or could contribute to long-term, adaptive oculomotor control. In the vestibular system of the pigeon, there is evidence that
proprioception functions in an on-line fashion, as passive motion of
the eye modifies vestibular slow phases (Knox and Donaldson 1993) and deafferentation of the extraocular muscles alters
vestibular eye movements (Hayman and Donaldson 1995).
In normal monkeys, however, ocular deafferentation does not affect
saccadic eye movements (Guthrie et al. 1983) or the eye position information used to encode visual space in craniotopic coordinates (Lewis et al. 1998).
In contrast, in monkeys with a vertical muscle paresis, ocular
deafferentation produced a gradual decrease in the amplitude of the
saccadic pulse in the paretic eye (Lewis et al. 1994). In accordance with prior suggestions (Jürgens et al.
1981; Steinbach and Smith 1981), we hypothesized
that the feedforward command normally provides the brain with adequate
information for immediate, on-line oculomotor control and that
proprioception provides an error signal used in the long-term, off-line
calibration of the efferent command (Lewis et al. 1994).
In the current study, we extended our investigation to evaluate the
short and long-term effects of ocular deafferentation on postsaccadic
eye motion in animals with unilateral vertical muscle pareses. The
results indicate that proprioceptive deafferentation modifies
postsaccadic drift in animals with vertical muscle pareses, as the
amplitude, mean velocity, and time constant of the postsaccadic drift
movements in the paretic eye were altered after deafferentation. Postsaccadic drift in the normal eye was not affected by
deafferentation of the paretic eye, despite suggestions that
proprioceptive afference from one eye may affect movements of the other
eye (O'Keefe and Berkley 1991).
EFFECT OF DEAFFERENTATION ON THE STEP AND SLIDE.
As previously described, deafferentation decreased the amplitude of the
pulse in the paretic eye for all vertical saccade conditions except
upwards saccades in monkey SO2 (Lewis et al. 1994). If the step had not been affected by deafferentation,
the change in drift of the paretic eye would have been onward in the direction of the antecedent pulse. This pattern was not consistently observed, suggesting that the drift bias was not simply a passive result of changes in the pulse, but that the step of innervation also
was affected by deafferentation of the paretic eye.
The time constant of postsaccadic drift depends on the gain and time
constant of the slide of innervation and on the mechanical response of
the ocular plant to a step of innervation (Optican and Miles
1985). Assuming that the mechanical properties of the plant
were not affected by deafferentation, then the changes in time constant
observed after section of the trigeminal nerve reflect a modification
of the gain or time constant of the slide of innervation.
EYE VELOCITY AND POSITION INFORMATION.
Although proprioceptive afferents carry eye velocity and position
signals (Fahy and Donaldson 1998), the changes in
postsaccadic drift that follow deafferentation do not appear to result
from a loss of corrective velocity or position feedback. If
proprioception provided information about postsaccadic eye
velocity that was used to minimize eye motion after
saccades, the velocity and amplitude of the components that make up the
drift waveform should have increased systematically after
deafferentation, but this pattern was not observed.
We previously demonstrated that the amplitude of the saccadic pulse in
the paretic eye decreased after deafferentation (Lewis et al.
1994), suggesting that proprioception partially corrected the
position-error associated with the gaze shift. If proprioception provided information about eye position that was used to
modify postsaccadic eye motion to correct errors in the amplitude of the gaze shift, then deafferentation should have biased systematically the drift in the direction opposite the preceding hypometric pulse. This also was not consistently observed. Our results therefore suggest
that whereas proprioception influences the efferent slide and step
commands that follow the saccadic pulse, it does not serve to minimize
postsaccadic eye velocity or to correct postsaccadic position error in
a consistent fashion.
TEMPORAL COURSE OF CHANGES AFTER DEAFFERENTATION.
Exactly when the bias in postsaccadic drift developed after
deafferentation is uncertain, as it was evident in the first set of
postdeafferentation data recorded 3 days after trigeminal nerve section
and did not change during the subsequent weeks of recording. This
differs from the changes in the saccadic pulse, which evolved during a
period of several weeks after deafferentation (Lewis et al.
1994). These results suggest that proprioception may
contribute to the control of postsaccadic eye motion in a more
immediate, on-line manner or via a short-term adaptive process that is
completed within hours to a few days.
EXPERIMENTAL LESIONS OF THE OCULOMOTOR PLANT.
After the trochlear nerve or the inferior rectus tendon was sectioned,
the paretic eye drifted after vertical saccades, and the three normal
vertically acting extraocular muscles could provide the brain with
meaningful proprioceptive feedback about postsaccadic eye motion.
Although the quality of the afferent signal from the paretic muscle
likely differed in these two experimental models, the consequences of
deafferenting the three normal, vertically acting muscles should have
been comparable with both lesions.
Because the eye movement data acquired after deafferentation was
recorded chronologically later than the predeafferentation data, it is
possible that spontaneous, mechanical alterations in the oculomotor
plant may have contributed to the changes in postsaccadic drift we
observed. Although these mechanical contributions cannot be excluded,
the drift components in the paretic eye were not uniformly reduced in
amplitude after deafferentation for any monkey or any saccade type. The
usual pattern of change was an increase in the amplitude of some drift
components and a decrease in others. In some cases, deafferentation was
associated with a change in the direction of the postsaccadic drift,
which would not be produced by a recovery of eye muscle function.
DEAFFERENTATION AND VISUALLY MEDIATED ADAPTATION.
Deafferentation of the paretic eye did not interfere with the adaptive
changes in postsaccadic drift in the paretic eye caused by changing the
chronic viewing state. Furthermore although there is evidence that
proprioception contributes to the binocular coordination of the eyes
(O'Keefe and Berkly 1991), the disconjugate adaptation of postsaccadic drift induced by visual experience also was unaffected by deafferentation. These results are consistent with our finding that
visually mediated adaptation of the saccadic pulse and of the static
ocular alignment does not depend on proprioception from the paretic eye
(Lewis et al. 1994) and with the finding of
Optican and Miles (1985) that postsaccadic drift can be
induced adaptively by moving the visual surround after saccades in
normal animals. On the basis of our results, however, we cannot exclude the possibility that deafferentation might have slowed the early rate
of visually mediated adaptation, as we did not record eye movements
until 3 days after the visual state was modified.
Although proprioceptive and retinal afferent signals interact at
the single-unit level in several brain areas involved in vision and the
control of eye movements (Ashton et al. 1984;
Donaldson and Long 1980; Lal and Friedlander
1990), our results suggest that proprioceptive and visual
information can act independently in the adaptive control of eye
movements. These findings contrast with results in the visual system
(Buisseret 1995), in which interactions between retinal
and extraretinal afference appear to be necessary for the development
of properties such as orientation (Buisseret et al.
1988) and disparity selectivity (Trotter et al.
1993) in cells within the visual cortex.
Conclusions
In summary, deafferentation of the paretic eye produced a bias in
the postsaccadic drift, suggesting that the step was altered, and a
change in the time constant of the drift, suggesting that the slide was
modified. Visually mediated adaptation of postsaccadic drift was not
affected by proprioceptive deafferentation. The direct cause of the
changes in postsaccadic drift that occurred after deafferentation is
uncertain. Our results are not consistent with the hypothesis that
these changes are simply due to a loss of feedback information that is
used to minimize postsaccadic eye velocity or to correct errors in eye
position. The change in drift after deafferentation appeared to be
idiosyncratic, as no consistent pattern occurred that points to an
identifiable function for the proprioceptive signal in the control of
postsaccadic eye motion (see Table 5).
The finding that visually mediated adaptation was not affected by
deafferentation suggests a functional segregation between the afferent
retinal and extraretinal signals in the control of eye movements. One
possible hypothesis is that proprioception, which transduces length and
tension information in the intrinsic coordinates of the eye muscles,
provides information that is used by the brain to model the mechanical
characteristics of the ocular motor plant. Visual afference, in
contrast, provides direct feedback about saccadic accuracy (via the
retinal error and retinal slip), and the brain may optimize eye
movement control by minimizing these error signals.
Examining larger saccades or saccades in more eccentric orbital
positions potentially could help clarify the function of the proprioceptive signal because its role may be more evident when the
mechanical nonlinearities of the plant are more prominent. Furthermore
examining the early rate of visually induced adaptive changes could be
informative because it might be slowed after deafferentation if
proprioception provides information that guides the adaptive process
(i.e., by signaling the anatomic locus of the abnormality responsible
for the eye movement inaccuracy).
We thank D. Roberts, P. Kramer, M. Shelhamer, C. Bridges, and A. Lasker.
This work was supported by National Institutes of Health Grants
EY-06273 and NS-01656 to R. F. Lewis and EY-01849 to D. S. Zee.
Present address and address for reprint requests: R. F. Lewis,
Massachusetts Eye and Ear Infirmary, 243 Charles St., Boston, MA 02114.
The costs of publication of this article were defrayed in part by the
payment of page charges. The article must therefore be hereby marked
"advertisement" in accordance with 18 U.S.C. Section
1734 solely to indicate this fact.
TOP
ABSTRACT
INTRODUCTION
