RAPID COMMUNICATION
Compensation for Gaze Perturbation During Inactivation of the Caudal
Fastigial Nucleus in the Head-Unrestrained Cat
LAURENT GOFFART, ALAIN GUILLAUME, AND DENIS PÉLISSON
Espace et Action, Institut National de la Santé et de la Recherche Medicale Unité 94, 69500 Bron, France
after inactivation of the caudal fastigial nucleus (cFN)
(head-fixed monkey: Ohtsuka et al. 1994; Robinson et al.
1993; head-free cat: Goffart and Pélisson 1998) could result
either from an inadequate feedback control (i.e., a mismatch
between the actual movement and its internal representation)
or from an impaired specification of the reference signal of
desired gaze displacement.
To distinguish between these two possibilities, we tested
the ability of the cFN-inactivated cat to compensate for a
perturbation in gaze position while orienting toward a visual
target. Gaze perturbation was induced by electrically microstimulating the deep layers of superior colliculus (SC) during the reaction time of a gaze shift toward a flashed target
(Pélisson et al. 1989; Sparks and Mays 1983). To test the
effectiveness of compensations independently from the dysmetria related to cFN inactivation, we compared the compensatory gaze shift of such perturbed trials to the primary gaze
shift of unperturbed trials with a similar desired gaze displacement. The predictions are as follow. If the dysmetria
induced by cFN inactivation results from inadequate feedback signals, gaze perturbations may be miscoded and compensation failures are expected. On the other hand, if the
dysmetria results from a change in the reference signal, full
compensations for gaze perturbations are expected.
METHODS
INTRODUCTION
It was proposed that saccadic shifts of the visual axis are
controlled by a feedback mechanism. First formulated for
head-restrained saccadic eye movements (Jürgens et al.
1981; Robinson 1975), this hypothesis was later extended
to gaze shifts resulting from combined eye and head movements (for review see Guitton 1992). The displacement of
the line of sight would be driven by a reference signal specifying the gaze displacement required to foveate the target
(desired gaze displacement command). The basic idea of
this feedback control hypothesis is that gaze accuracy is
preserved regardless of how the movement is executed, as
long as the feedback mechanisms accurately inform about
the actual performance. Compatible with this prediction is
the demonstration in behaving animals of a preserved accuracy of eye or gaze saccades despite perturbations induced
by intracerebral electrical microstimulation (e.g., Keller et
al. 1996; Pélisson et al. 1989, 1995; Sparks and Mays 1983).
The contribution of the medioposterior cerebellum to the
control of gaze-shift accuracy is well established (Leigh and
Zee 1991), but its specific role in the context of the feedback
control hypothesis is still unclear. The loss of gaze accuracy
A detailed description of the methods can be found in previous
papers (Goffart and Pélisson 1998; Pélisson et al. 1989). Briefly,
two cats were prepared for the experiments under general anesthesia and aseptic conditions following the guidelines from the French
Ministry of Agriculture (87/848) and from the European Community (86/609/EEC). Two coils were implanted for the recording
of gaze and head position by the search-coil-in-magnetic-field technique (Robinson 1963). Signals from eye and head coils were
linearized (see equation in Judge et al. 1980) and scaled on-line
by a custom-written computer program with calibration parameters
defined before implantation and checked after surgery in the behaving animal (details in Goffart and Pélisson 1998). A recording
chamber was implanted stereotaxically over the cerebellum. A second chamber was implanted over the SC in one animal (cat L),
whereas in the other animal (cat H), four bundles of four microwires each were stereotaxically implanted in each deep SC. Finally,
a plastic head-holding device was fixed to the skull with cement.
After recovery from surgery, each cat was placed in a hammock
that gently restrained the body, without constraint on natural movements of the head. The visual target was a spoon filled with a
food puree and fitted with two infrared diodes that permitted us to
continuously record its position (Urquizar and Pélisson 1992). The
sudden presentation of this food target on either side of an opaque
panel placed in front of the animal (41-cm distance) elicited coor-
0022-3077/98 $5.00 Copyright q 1998 The American Physiological Society
1552
/ 9k2c$$se10
J197-8RC
08-19-98 18:00:47
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 15, 2017
Goffart, Laurent, Alain Guillaume, and Denis Pélisson. Compensation for gaze perturbation during inactivation of the caudal
fastigial nucleus in the head-unrestrained cat. J. Neurophysiol. 80:
1552–1557, 1998. Muscimol injection in the caudal part of the
fastigial nucleus (cFN) leads, in the head-unrestrained cat, to a
characteristic dysmetria of saccadic gaze shifts toward visual targets. The goal of the current study was to test whether this pharmacological cFN inactivation impaired the ability to compensate for
unexpected perturbations in gaze position during the latency period
of the saccadic response. Such perturbations consisted of moving
gaze away from the target by a transient electrical microstimulation
in the deep layers of the superior colliculus simultaneously with
extinction of the visual target. After injection of muscimol in the
cFN, targets located in the contralesional hemifield elicited gaze
shifts that fell short of the target in both ‘‘perturbed’’ and ‘‘unperturbed’’ trials. The amplitude of the compensatory contraversive
gaze shifts in perturbed trials coincided with the predicted amplitude of unperturbed responses starting from the same position.
Targets located in the opposite hemifield elicited hypermetric gaze
shifts in both trial types, and the error of compensatory responses
was not statistically different from that of unperturbed gaze shifts.
These results indicate that inactivation of the cFN does not interfere
with the ability of the head-unrestrained cat to compensate for
ipsiversive or contraversive perturbations in gaze position. Thus
the gaze-related feedback signals that are used to compute a reference signal of desired gaze displacement are not impaired by cFN
inactivation.
COMPENSATORY GAZE SHIFTS DURING
RESULTS
Muscimol-induced inactivation of the cFN produced a
marked hypermetria of ipsiversive gaze shifts and hypometria of contraversive ones, as described in previous reports
(Goffart and Pélisson 1994, 1998). Because these modifications in accuracy occurred predominantly in the horizontal
plane, the analysis of the compensation abilities after muscimol injection in the cFN is restricted to the horizontal component of gaze shifts. We report on the compensatory responses elicited by a target presented in the hemifield contralateral and ipsilateral, respectively, to the inactivated cFN.
Target in the contralesional visual hemifield
Compensatory gaze shifts were still observed after muscimol injection in the left cFN of cat H (experiment H/lcFN/
rSC), as illustrated in Fig. 1A, showing representative responses elicited by a target located 277 to the right. The
temporal courses of horizontal gaze and head position are
plotted for two unperturbed (Fig. 1A, left panel) and one
perturbed trials (Fig. 1A, right panel). Stimulation of the
right SC shifted gaze to the left (‘‘perturbation’’) and was
followed by a rightward gaze shift (‘‘compensation’’) toward the remembered target location. Note that because of
the cFN inactivation, all three target-elicited gaze shifts
markedly undershot the target location. Note further that
the compensatory gaze shift was more hypometric than the
unperturbed gaze shift that started from a gaze position corresponding to the preperturbation position (Fig. 1A, trajectory a). Nevertheless, the amount of hypometria (error Å
/ 9k2c$$se10
J197-8RC
INACTIVATION
1553
–15.77 ) was similar to that of the other unperturbed gaze
shift (error Å –15.67 ), which started from a position comparable with the gaze position reached at the end of the perturbation (Fig. 1A, trajectory b). For a given target position
when gaze starts from the same position, the similarity between gaze-shift amplitude in the perturbed and unperturbed
trials suggests that the gaze perturbation was adequately
taken into account in the production of the compensatory
gaze shift. To evaluate the consistency of this observation,
the amplitude of each unperturbed and compensatory gaze
shift recorded during cFN muscimol inactivation is plotted
as a function of the amplitude of the desired horizontal gaze
displacement (Fig. 1B). For unperturbed trials ( s ) the desired gaze displacement corresponds to the distance between
target and gaze during the target presentation time, and for
perturbed trials ( m ) it represents the new distance between
target and gaze after the electrically induced change in gaze
position. The regression analysis that was performed for
unperturbed gaze shifts confirmed the typical gain reduction
of contraversive gaze shifts (regression equation in legend).
It is remarkable that the amplitude values of the compensatory gaze shifts ( m ) are distributed along the regression
line fitting the unperturbed responses. This observation was
analyzed further by computing for each perturbed trial the
inaccuracy of compensation as defined relative to comparable unperturbed responses. This value of compensation error
was obtained by subtracting from the compensatory gazeshift amplitude the amplitude of the predicted unperturbed
gaze response to the same target (i.e., distance between each
perturbed data point and the regression line in Fig. 1B). A
positive value corresponds to an overcompensation and a
negative one to an undercompensation. As seen in Fig. 1C,
the compensation error did not depend on the perturbation
size (
, regression); in addition, it was not statistically
different from zero ( –0.7 { 2.97, mean { SD; t (df Å 17) Å
1.0, P ú 0.05). The values of compensation error differed
clearly from the theoretical values predicted from the hypothesis of no compensation ( – – – ).
Target in the ipsilesional visual hemifield
Figure 2 shows the data obtained after muscimol injection
in the left cFN of cat L (experiment L/lcFN/lSC). The
time course of the horizontal component of gaze and head
movements toward a target located 197 to the left are shown
in Fig. 2A. All gaze shifts were hypermetric and ended at
the same position beyond the target. In the perturbed trial,
the left SC stimulation quickly shifted gaze to the right
(perturbation). This perturbation was shortly followed by a
leftward gaze shift (compensation) toward the remembered
target location. This compensatory gaze shift was markedly
hypermetric and overshot the target location by 13.57, an
error relative to the target virtually identical to that of the
two unperturbed gaze shifts. The amplitude of each unperturbed and compensatory gaze shift was plotted as a function
of the amplitude of the desired horizontal gaze displacement
(Fig. 2B). First, note that the ‘‘unperturbed relationship’’
is displaced toward negative values of actual gaze displacement (see equation of regression in Fig. 2 legend). This
result illustrates the characteristic hypermetria of gaze shifts
directed toward the inactivated cFN, which is largely related
08-19-98 18:00:47
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 15, 2017
dinated eye and head movements that shifted gaze (eye-in-space)
in a saccadic manner. Different panels were used to elicit gaze
shifts toward {19, {27, and {357 targets in the horizontal plane.
The ambient lights were turned off Ç80 ms after target presentation
by an electronic shutter so that the orienting response was triggered
and completed in darkness. In about one-third of the trials (perturbed trials), initial gaze position was perturbed by applying a
short microstimulation train to the SC at the time of target offset;
in these trials the target location was chosen to elicit compensatory
gaze shifts in a direction opposite to the gaze perturbation. Both
perturbed and unperturbed trials were randomly presented during
a 1–2 h recording session initiated 15–20 min after the onset of
muscimol injection in the cFN. In one animal, control recordings
were performed several days before injection to verify that the
accuracy of gaze shifts was preserved despite the stimulation-induced gaze perturbation (Pélisson et al. 1989).
Stimulation of the deep SC layers was delivered through a chronically implanted microwire (stainless steel, 25-mm diam) or a
tungsten microelectrode (50- to 80-ms train duration, 500 Hzfrequency, 0.2- to 0.5-ms pulse duration, 1.2–2 times threshold
intensity). Unilateral injections of a saline solution of muscimol
(Sigma, 8.8 mM, 0.2–0.3 ml) were performed by lowering a cannula (230 mm OD) into the cFN of the head-fixed animal. The
injection sites were chosen on the basis of stereotaxic coordinates
and electrophysiological recordings. We report here on three separate experiments. In the first (cat H), the inactivated cFN was
contralateral to the stimulated SC (left cFN/right SC: experiment
labeled H/lcFN/rSC), whereas in the others (cat L) the inactivated
cFN and stimulated SC were on the same side (right cFN/SC and
left cFN/SC: experiments labeled L/rcFN/rSC and L/lcFN/lSC).
All sites of injection in the cFN and of stimulation in the SC were
confirmed by postmortem histological reconstruction.
C FN
1554
L. GOFFART, A. GUILLAUME, AND DENIS PELISSON
to a bias of the response end points relative to the target.
Second, the overlap between the perturbed data points and
the unperturbed ones indicates that the relationship between
the amplitude of compensatory gaze shifts and the desired
gaze displacement is similar to that of unperturbed responses. The average compensation error was not statistically different from zero [0.73 { 3.927, t (15) Å 0.74, P ú
0.05]. In addition, as shown in Fig. 2C, there was no correlation between compensation error and perturbation size. In
the same cat we injected muscimol in the right cFN, and
compensatory responses to a target presented in the ipsilesional hemifield (experiment L/rcFN/rSC) provided the same
/ 9k2c$$se10
J197-8RC
results. Table 1 shows that the average value of compensation error and its correlation with perturbation size did not
reach a significant level in any experiment.
DISCUSSION
These results provide some clues regarding the origin of
the head-unrestrained gaze dysmetria observed after muscimol inactivation of the cFN (Goffart and Pélisson 1994,
1998; Goffart et al. 1998). It is shown that inactivation
of the cFN did not interfere with the ability of the headunrestrained cat to compensate for collicular-induced pertur-
08-19-98 18:00:47
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 15, 2017
FIG . 1. Compensatory responses toward a target located in the contralesional visual hemifield (experiment H/lcFN/rSC).
A: time course of horizontal gaze and head position during representative responses elicited by a target briefly presented
(hatched column) at 277 to the right and recorded during inactivation of the left caudal fastigial nucleus (cFN). In the
unperturbed trials (A, left) gaze and head movements markedly undershot the target position. In the perturbed trial (A, right)
the perturbation in gaze position evoked by superior colliculus (SC) stimulation (darker shaded column) was shortly followed
by a compensatory gaze shift that also was hypometric. Note that the amplitude of the compensatory gaze shift matched that
of the unperturbed gaze shift initiated from a similar starting position (trajectory b). B: relationship between actual and
desired amplitude of horizontal gaze displacement achieved during the primary response of unperturbed trials ( s ) or during
the compensatory response of perturbed trials ( m ). Desired gaze displacement amplitude refers to the horizontal distance
, regression fitting the unperturbed data (y Å
between starting gaze position and target position. – – – , unitary slope;
0.65x – 2.41, R 2 Å 0.77). C: relationship between gaze compensation error and perturbation size (absolute value of gaze
horizontal perturbation). Compensation horizontal error corresponds to the distance between the amplitude value of each
perturbed data point and the regression line shown in B. Regression lines (intercept set to 0) are shown for the theoretical
relationship predicted from the hypothesis of no compensation ( – – – , slope corresponding to that of regression line in B)
, y Å –0.12x, slope not statistically different from 0, t (17) Å –1.7,
and for the relationship fitting the actual data (
P ú 0.05). B and C, arrow: compensatory gaze shift shown in A. Ipsi, ipsiversive; contra, contraversive.
COMPENSATORY GAZE SHIFTS DURING
C FN
INACTIVATION
1555
bations moving gaze away from the target of an impending
gaze shift (Pélisson et al. 1989). Irrespective of whether the
perturbation was directed toward or away from the inactivated cFN, the amplitude of the compensatory gaze shifts
to a given target was inaccurate but nevertheless comparable
with the amplitude of unperturbed gaze shifts starting from
similar positions. In both cases, the compensation error was
not related to the size of the perturbation within the tested
range, and its average value was virtually zero. These find1. Perturbation size and compensation horizontal error
computed for each experiment
TABLE
Experiment
Perturbation
Size (7)
Compensation
Horizontal
Error (7)
Theoretical
Slope
Actual
Slope
H/lcFN/rSC
L/lcFN/lSC
L/rcFN/rSC
9.4 { 3.7
11.5 { 4.6
10.7 { 3.6
0.7 { 2.9 NS
00.7 { 3.9 NS
00.7 { 5.5 NS
00.65
01.24
01.30
00.12 NS
0.06 NS
00.09 NS
Values are means { SD. Number of responses is 18, 16, and 24 for rows
1, 2, and 3, respectively. Last two columns are slopes of regressions (intercept set to zero) between the following parameters: 1) theoretical relationship predicted from the hypothesis of no compensation and 2) actual relationship. NS, differences not statistically significant from zero (Student’s
t-test, P ú 0.05).
/ 9k2c$$se10
J197-8RC
ings indicate that the gaze perturbation is taken into account
to update the command specifying the required gaze displacement for foveating a visual target. This result in turn
suggests that neither the feedback mechanism informing
about the gaze position perturbation nor the combination of
this gaze perturbation with the memorized retinal signals of
initial gaze error, to update the desired gaze displacement
command, are affected by the muscimol cFN inactivation.
On the basis of the present findings, the origin of the
gaze dysmetria induced by cFN inactivation can be discussed
further in relation to two possible feedback control systems
for gaze shift. On the one hand, if the feedback signals
discussed above (hEGD for estimated horizontal gaze displacement) are used for the dynamic control of the gaze shift
(Fig. 3A), gaze dysmetria cannot result from an inadequate
dynamic feedback control. Instead, gaze dysmetria would
result from a modification of the desired horizontal gaze
displacement command (hDGD), which serves as a reference signal to the dynamic feedback mechanisms. This hypothesis was proposed previously to account for the bias of
ipsiversive gaze shifts and the ensuing fixation offset (Goffart and Pélisson 1994, 1998) and for the unchanged dynamics of contraversive gaze shifts (Goffart et al. 1998). On
the other hand, if distinct gaze-related feedback signals are
used for updating the reference signal and for the dynamic
08-19-98 18:00:47
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 15, 2017
FIG . 2. Compensatory responses toward a
target located in the ipsilesional visual hemifield
(experiment L/lcFN/lSC). Same conventions as
in Fig. 1. A: time course of horizontal gaze and
head position during representative responses
elicited by a target briefly presented at 197 to the
left and recorded during inactivation of the left
cFN. B: relationship between actual and desired
amplitude of horizontal gaze displacement
achieved during the primary gaze shift of unperturbed trials ( s ) or during the compensatory gaze
shift of perturbed trials ( m ). Equation of regression: y Å 1.24x – 10.36 (R 2 Å 0.89). C: relationship between gaze compensation horizontal
, equation: y Å
error and perturbation size [
0.06x, slope not statistically different from 0, t
(15) Å 0.8, P ú 0.05]. B and C, arrow: compensatory gaze shift shown in A.
1556
L. GOFFART, A. GUILLAUME, AND DENIS PELISSON
control of gaze shift (Fig. 3B), the current findings have
some implication regarding the hypometria of contraversive
gaze shifts during cFN inactivation. Indeed, the correct updating of the desired gaze displacement after gaze perturbation (see long feedback pathway in Fig. 3 B) implies that
the gain reduction of contraversive gaze shifts takes place
downstream from this updating. In addition, the unchanged
dynamics of contraversive gaze shifts (Goffart et al. 1998)
would again place this gain reduction upstream from the
dynamic feedback loop.
We thank M. Riley and C. Urquizar for writing some of the display/
parameter extraction software, N. Boyer-Zeller for providing essential support for histology, and M.-L. Loyalle for taking care of the animals.
This study was supported by Human Frontier Science Program Organization Grant RG58/92B. L. Goffart was supported by a fellowship from the
French Ministry of Research and Technology.
Present address of L. Goffart: Division of Neuroscience, Baylor College
of Medicine, One Baylor Plaza, Houston, Texas 77025.
/ 9k2c$$se10
J197-8RC
Address for reprint requests: D. Pélisson, Espace et Action, INSERM U
94, 16 Avenue Doyen Lépine, 69500 Bron, France.
Received 17 March 1998; accepted in final form 14 May 1998.
REFERENCES
GOFFART, L. AND PÉLISSON, D. Cerebellar contribution to the spatial encoding of orienting gaze shifts in the head-free cat. J. Neurophysiol. 72:
2547–2550, 1994.
GOFFART, L. AND PÉLISSON, D. Orienting gaze shifts during muscimol inactivation of the caudal fastigial nucleus in the cat. I. Gaze dysmetria. J.
Neurophysiol. 79: 1942–1958, 1998.
GOFFART, L., PÉLISSON, D., AND GUILLAUME, A. Orienting gaze shifts during
muscimol inactivation of the caudal fastigial nucleus in the cat. II. Dynamics and eye-head coupling. J. Neurophysiol. 79: 1959–1976, 1998.
GUITTON, D. Control of eye-head coordination during gaze shifts. Trends
Neurosci. 15: 174–179, 1992.
JUDGE, S. T., RICHMOND, B. H., AND CHU, F. C. Implantation of magnetic
search coils for measurement of eye position: an improved method. Vision
Res. 20: 535–538, 1980.
JÜRGENS, R., BECKER, W., AND KORNHÜBER, H. H. Natural and drug-in-
08-19-98 18:00:47
neupa
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 15, 2017
FIG . 3. Hypothetical block diagrams of the control system for gaze shift showing the postulated level of cFN action
(shaded areas). These highly simplified diagrams represent 2 putative structures of the gaze control system but do not address
the exact nature of the feedback signals (position vs. displacement, gaze vs. eye and head). The 2 dynamic controllers
(premotor centers / feedback generators) are each recruited either by the presentation of the visual target or by the SC
electrical microstimulation. Eye and head premotor centers are driven by a signal of horizontal dynamic motor error (hDME),
which itself results from the permanent comparison between a reference signal specifying the desired horizontal gaze
displacement (hDGD) and feedback signals estimating the current horizontal displacement of gaze (hEGD). In A, the
feedback signals (hEGD) that accurately encode the SC microstimulation-induced perturbation of gaze position are involved
in the dynamic control of the gaze shift. Gaze dysmetria induced by muscimol inactivation of the cFN results from an
impairment in the translation of retinal signals into the hDGD reference signal (gain reduction for the hypometria of
contraversive gaze shifts, addition of bias K for the hypermetria of ipsiversive gaze shifts). In B, the feedback signals
(hEGD) encoding gaze perturbation and allowing to update the reference signal are distinct from those used for the dynamic
control of gaze shift, constraining the site of gain reduction for the hypometria of contraversive gaze shifts (see text).
COMPENSATORY GAZE SHIFTS DURING
duced variations of velocity and duration of human saccadic eye movements: evidence for a control of the neural pulse generator by local
feedback. Biol. Cybern. 39: 87–96, 1981.
KELLER, E. L., GANDHI, N. J., AND SHIEH, J. M. Endpoint accuracy in saccades interruped by stimulation of the omnipause region in the monkey.
Vis. Neurosci. 13: 1059–1067, 1996.
LEIGH, R. J. AND ZEE, D. S. The Neurology of Eye Movements. Philadelphia,
PA: Davis, 1991.
OHTSUK A, K., SATO, H., AND NODA, H. Saccadic burst neurons in the
fastigial nucleus are not involved in compensating for orbital nonlinearities. J. Neurophysiol. 71: 1976–1980, 1994.
PÉLISSON, D., GUITTON, D., AND GOFFART, L. On-line compensation of gaze
shifts perturbed by micro-stimulation of the superior colliculus in the
head-free cat. Exp. Brain Res. 106: 196–204, 1995.
PÉLISSON, D., GUITTON, D., AND MUNOZ, D. P. Compensatory eye and head
C FN
INACTIVATION
1557
movements generated by the cat following stimulation-induced perturbations in gaze position. Exp. Brain Res. 78: 654–658, 1989.
ROBINSON, D. A. A method of measuring eye movement using a scleral
coil in a magnetic field. IEEE Trans. Biomed. Elect. 10: 137–145, 1963.
ROBINSON, D. A. Oculomotor control signals. In: Basic Mechanisms of
Ocular Motility and Their Clinical Implications, edited by G. Lennerstrand and P. Bach-Y-Rita. Oxford, UK: Pergamon, 1975, p. 337–378.
ROBINSON, F. R., STRAUBE, A., AND FUCHS, A. F. Role of the caudal fastigial
nucleus in saccade generation. II. Effects of muscimol inactivation. J.
Neurophysiol. 70: 1741–1758, 1993.
SPARKS, D. L. AND MAYS, L. E. Spatial localization of saccade targets. I.
Compensation for stimulation-induced perturbations in eye position. J.
Neurophysiol. 49: 45–63, 1983.
URQUIZAR, C. AND PÉLISSON, D. A non-contact system for 2-dimensional
trajectory recording. J. Neurosci. Methods 43: 77–82, 1992.
Downloaded from http://jn.physiology.org/ by 10.220.32.247 on June 15, 2017
/ 9k2c$$se10
J197-8RC
08-19-98 18:00:47
neupa
LP-Neurophys