Download Orbital position dependency is different for the gain of externally and

Motor Systems
NeuroReport
NeuroReport 10, 2665±2670 (1999)
THE gain of visually triggered saccades depends on
orbital position. Centrifugal saccades have smaller gains
and are slower than centripetal saccades elicited by the
same target amplitude. We determined whether internally triggered saccades, e.g. scanning or memory saccades, exhibit the orbital position dependency evident in
visually guided saccades. The search coil technique was
used to record eye movements of healthy subjects while
they performed horizontal 12.58 saccades under three
paradigms (gap, scanning and memory saccades). Orbital position in¯uenced externally triggered gap saccades
but not the gain or peak eye velocity of scanning or
memory saccades. These ®ndings do not support the
idea that position dependency caused by orbital mechanics is compensated for at the level of the common
brainstem burst generator. Instead our results are
consistent with the view that cortical output re¯ects
the differences evident in the gain of visually triggered
centrifugal and centripetal saccades. NeuroReport
10:2665±2670 # 1999 Lippincott Williams & Wilkins.
Key words: Centrifugal; Centripetal; Craniotopic; Orbital
eye position; Orbital mechanics; Retinotopic
Introduction
The gain of visually triggered saccades partly depends on the position of the eye at the start of the
saccade. Centripetal saccades are signi®cantly larger
than centrifugal saccades. In addition, the peak eye
velocities of centrifugal saccades . 158 are smaller
than those of centripetal saccades of the same size
[1,2]. This position dependency may be caused by
the elastic force of the extraocular muscles and
tendons which tend to pull the eyeball back to the
primary position near straight ahead [3]. The elastic
force assists centripetal saccades and opposes centrifugal saccades. The neuronal signal driving the eye
muscles must compensate for the effect of the elastic
force on the eye. An incomplete compensation for
this force could explain the gain and velocity differences between centrifugal and centripetal saccades.
Compensation for orbital position is thought to
occur in two regions of the cerebellum: the oculomotor vermis and the caudal fastigial nucleus
(CFN). Lesions of these areas cause an abnormally
large difference in the gains of centrifugal and
centripetal saccades [4,5]. The CFN receives input
from the oculomotor vermis and projects to the premotor network in the paramedian pontine reticular
formation, which is common to all saccades, i.e. the
brain stem burst generator. Thus, current evidence
suggests that visually triggered saccades are depen0959-4965 # Lippincott Williams & Wilkins
Orbital position
dependency is different
for the gain of externally
and internally triggered
saccades
Thomas Eggert, Frank Mezger,
Farrel Robinson1 and
Andreas StraubeCA
Department of Neurology, Ludwig-Maximilians
University, 81377 Munich, Germany; 1 Department
of Biological Structure, University of Washington,
Seattle, USA
CA
Corresponding Author
dent on eye position because the oculomotor vermis
and CFN provide incomplete compensation via the
brain stem burst generator.
We could infer more about the origin of position
compensation if we knew more about the position
dependency of internally triggered saccades, such as
memory and scanning saccades. There is increasing
evidence that different circuits in the cerebral cortex
are responsible for generating externally (visually
triggered or re¯exive) saccades and internally triggered saccades [6]. Re¯exive saccades probably depend mostly on pathways from the posterior
parietal cortex to the superior colliculus via the
posterior part of the internal capsule. Internally
triggered saccades probably depend more on direct
pathways from the frontal or supplementary eye
®elds to the brain stem [6,7,8].
Currently no data are available from normal
subjects which indicate if position dependency is the
same for externally and internally triggered saccades.
There are some signs that the cerebellum may be less
involved in internally triggered than externally triggered saccades. Straube et al. [9] examined a patient
with a large cerebellar midline lesion due to tumor
surgery who showed a pronounced hypermetria for
the re¯exive saccades but normal internally triggered
scanning saccades. This suggests that the position
dependency may be stronger in internal saccades. In
this study, we investigated whether different types
Vol 10 No 12 20 August 1999
2665
NeuroReport
of saccades exhibit different position dependency in
healthy subjects.
Materials and Methods
Six healthy subjects (ages 27±40 years, one female
and ®ve males) participated in the study. Each
subject was recorded at least twice on different days.
Subjects were seated in front of a screen at a distance
of 1.40 m in darkness. A red laser spot, which was
controlled by a mirror galvanometer (General Scanning G120D, USA), served as visual target. Eye
movements were recorded with the magnetic search
coil technique. A 1.5 3 1.5 3 1.5 m three-®eld coil
was used (Remmel System, USA) and a 2-dimensional scleral coil (Scalar, Delft, Netherlands) was
attached to the right eye of the subjects. Eye movement signals were sampled at 1 kHz and were stored
on the hard disk of the computer.
For calibration of the eye movements a thirdorder polynomial calibration was applied. In each
session intervals of stable eye positions of .300 ms
were collected from the late ®xation periods (within
a period of 500 ms immediately before the target
stepped). These ®xations were used to compute the
coef®cients of the polynomial. This method provided a reliable calibration within the examined
range of horizontal eccentricities ( 258). Saccades
were detected and marked on the basis of velocity
criteria. All movements with peak absolute eye
velocity . 1008/s were analyzed. The start and the
end of the saccade were de®ned as the points where
the saccade velocity rose to or dropped below,
respectively, 10% of the peak eye velocity. Thus,
the computer automatically marked the onset, the
end, and the peak velocity of the saccades. From
these marks the gain (ratio eye to target amplitude),
the saccade duration, the peak velocity, and the
skewness of the velocity pro®le were calculated offline. The skewness was computed according to the
method of Van Opstal and Van Gisbergen [10];
positive values indicate a deceleration period longer
than the acceleration period.
In all paradigms the subject had to perform
saccades to one of ®ve target locations that were at
the center, ‡12.58, and 258 on the horizontal
meridian at eye level. The sequence of target locations was a pseudo random sequence with length of
200 and was identical for all experiments. We
elicited three different types of saccade. (1) Gap
saccades were elicited when the ®xation spot was
switched off and 100 ms later the laser spot reappeared at the new target position. (2) Scanning
saccades were elicited when the subject could always
see the ®ve targets. The sequence of the targets to be
scanned was indicated by a slight increase of the
2666
Vol 10 No 12 20 August 1999
F. Eggert et al.
luminance of the next target. (3) Memory guided
saccades were elicited when the next target was
¯ashed for 50 ms while the ®xated target stayed on.
After a delay period of 1 s the ®xated target was
switched off, and the subject had to perform a
saccade to the remembered position of the brie¯y
illuminated target. After this saccade, the laser spot
reappeared at the location where it had brie¯y
appeared. The new trial started from this position 2 s
later, an interval which was suf®cient for corrective
saccades.
The latency of the saccade was de®ned as the
interval between the blanking of the ®xation spot
and the start of the primary saccade (gap and
memory-guided saccades) or the increase of the
luminance of the target location and the start of the
saccade (scanning saccades).
Each subject was examined in two recording
sessions. In each session we ®rst elicited 100 gap
saccades and then 200 scanning or memory-guided
saccades. Finally, we recorded 100 gap saccades
again to measure the effect of fatigue on our measurements. We discarded all saccades that crossed the
midline from our analysis. Our statistical comparisons always used a signi®cance level of 0.01.
Results
The data were divided into two saccade directions
and ®ve different starting positions (08, 12.58,
258).This resulted in eight subgroups because
saccades starting at 258 were in only one direction,
i.e. centripetal. In each of these eight resulting
subgroups, the median of the number of saccades
analyzed per subject was 27 gap saccades, 17 memory saccades, and 10 scanning saccades. Figure 1
shows for all subjects the dependence of the gain,
the peak velocity and the skewness on initial eye
position for gap saccades, memory, and scanning
saccades with an amplitude of 12.58. We discuss each
type of saccade separately.
Gap saccades: As can be seen in the left-most
column of Fig. 1, there was a signi®cant correlation
of the eye position and the gain (leftward saccades:
r ˆ 0.7, p , 0.001; rightward saccades: r ˆ 0.7, p ,
0.001) with centripetal saccades larger than centrifugal saccades. A similar highly signi®cant relationship
was also seen for the peak velocity, which was
slower for the centrifugal saccades than for the
centripetal saccades (leftward: r ˆ 0.79, p , 0.001;
rightward: r ˆ ÿ0.78, p , 0.001; leftmost panel,
Fig. 1). Table 1 lists the average gain and the average
peak velocity of all subjects for saccades to 12.58
targets at larger orbital eccentricities (i.e. saccades
between 12.58 and 258), where the difference be-
Dependency of saccades on orbital position
A
Gap
NeuroReport
Memory
Scanning
140
130
Gain (%)
120
110
100
90
80
70
218.75 26.25 0 6.25 18.75
218.75 26.25 0 6.25 18.75
218.75 26.25 0 6.25 18.75
218.75 26.25 0 6.25 18.75
218.75 26.25 0 6.25 18.75
218.75 26.25 0 6.25 18.75
B
600
Peak velocity (deg/s)
500
400
300
200
100
C
0.5
0.4
Skewness
0.3
0.2
0.1
0.0
20.1
20.2
218.75 26.25 0 6.25 18.75
218.75 26.25 0 6.25 18.75
218.75 26.25 0 6.25 18.75
FIG. 1. The individual means of the six subjects (open circles) as well as the group means ( s.d.) for saccades starting at the center (08), at 12.58 or
at 258 eccentricity on the left (negative values) or on the right (positive values) with petal direction (gray symbols) or fugal direction (dark symbols) for
the gap, memory, and scanning saccades are presented. For more clarity the mean eccentricity of the saccades (18.758 and 6.258) are plotted slightly
offset from each other. (A) Saccadic gain plotted against orbital eye position. (B) Peak eye velocity against orbital eye position. (C) Skewness against
orbital eye position. The linear regression slopes of the gain and peak eye velocity were much steeper for the gap saccades than for the memory and
scanning saccades. These differences were highly signi®cant.
tween centrifugal and centripetal saccades was largest. For the skewness the correlation with orbital
position was not signi®cant (leftmost panel Fig. 1C;
Table 1).
Memory saccades: In contrast to gap saccades,
memory saccades showed no position dependency.
There was no signi®cant correlation between gain
and eye position (leftward: r ˆ ÿ0.27, p , 0.2; rightVol 10 No 12 20 August 1999
2667
NeuroReport
F. Eggert et al.
Table 1. Values (mean s.d.) of the 12.58 saccades between 12.58 and 258 eccentricity
Condition
Petal abduction
Petal adduction
Fugal abduction
Fugal aduction
Gain (%)
Gap
Scanning
Memory
102.9 5.8
95.2 2.5
85.9 5.2
103.4 8.8
89.8 6.9
85.2 8.3
87.8 3.0
97.6 2.8
93.6 20.7
86.7 6.1
93.2 4.3
91.5 7.7
Velocity (8)
Gap
Scanning
Memory
431.4 26.9
415.9 25.2
309.5 29.9
401.9 37.4
354.3 39.7
268.4 39.7
316.3 39.9
320.4 74.8
230.8 64.9
301.6 26.6
298.4 45.3
237.7 48.8
Skewness
Gap
Scanning
Memory
0.12 0.04
0.12 0.07
0.29 0.08
0.04 0.1
0.08 0.08
0.24 0.1
0.02 0.1
0.13 0.14
0.31 0.14
ÿ0.016 0.06
0.01 0.1
0.19 0.15
ward: r ˆ 0.18, p , 0.38, middle panel, Fig. 1A). The
dependency of the peak velocity on the orbital
position was less than that for gap saccades (leftward: r ˆ 0.26, p , 0.21; rightward: r ˆ ÿ0.57, p ,
0.004; middle panel, Fig. 1B). Finally, skewness did
not show a correlation with the orbital position
(leftward: p , 0.6; rightward: p , 0.79; middle panel,
Fig. 1C).
Scanning saccades: Like memory saccades, but in
contrast to the gain differences evident in the gap
saccades, there was no consistent correlation between gain and orbital position (leftward: r ˆ ÿ0.18,
p , 0.4; rightward: r ˆ 0.47, p , 0.02; rightmost panel, Fig. 1A). The small tendency observed for
rightward saccades was in the direction opposite to
that for gap saccades. There was a slightly stronger
in¯uence of the orbital position on the peak eye
velocity than on memory saccades (leftward: r ˆ 0.5,
p , 0.02; rightward: r ˆ ÿ0.6, p , 0.003; rightmost
panel, Fig. 1B). The skewness showed no signi®cant
correlation (leftward: p , 0.2, rightward: p , 0.85;
rightmost panel, Fig. 1C). The overall mean of the
latency of the primary saccade was 114 ms in the
gap paradigm, 233 ms in the memory paradigm, and
585 ms in the scanning paradigm.
Differences between the conditions: It is critical to
determine if the differences in position dependency
of the different types of saccades are signi®cant. To
test this, we compared the slopes of the relationships
between initial position and gain, velocity and skewness for the three types of saccades tested. The initial
position vs gain slopes for gap saccades were signi®cantly different from those for memory ( p ,
0.003) and scanning saccades ( p , 0.0001). Thus,
position dependency is signi®cantly higher for the
gain of gap saccades than for the gain of either type
of internally guided saccades (Fig. 1A). In contrast,
there was no signi®cant difference between the three
2668
Vol 10 No 12 20 August 1999
types of saccades we tested in the position dependency of peak eye velocity or skewness. In summary, the gain, but not peak eye velocity or
skewness, of gap saccades depends more on initial
eye position than does the gain of memory and
scanning saccades.
Discussion
Our data for the visually guided re¯exive saccades
(gap saccades) are in agreement with previous published results [1,2,11] and show that there is strong
in¯uence of the orbital position on the amplitude
and velocity of the saccade. Centrifugal saccades
generally have a smaller gain and a slower peak eye
velocity than centripetal saccades of the same desired amplitude. This difference may be related to
the mechanical properties of the tissue around the
globe in the orbit. The elastic force of the orbital
tissue tends to pull the eye toward a primary
position [3]. Thus, this force is larger at eccentric
orbital position and should add to the forces
obtained by the muscle activation, giving a net result
of a higher acceleration and peak velocity for
saccades towards the primary eye position. Pelisson
and Parblanc [12] propose another explanation.
They speculate that the force accelerating the eyes
(agonist force minus antagonist force) is highest at
eccentric orbital positions, because here the step
activity of the agonist is largest and the antagonist is
deactivated during a saccade.
Our ®ndings that there is no centrifugal±centripetal difference in the gain of internally triggered
scanning and memory-guided saccades cannot be
explained by such an effect. A position dependency
of the signals to the extraocular muscles would,
contrary to our ®ndings, induce the same centripetal±centrifugal differences in gain for all types of
saccades.
The variability of internally triggered saccades
Dependency of saccades on orbital position
cannot explain the absence of centrifugal±centripetal
gain differences. The gain of scanning saccades
showed less variability than did the gain of gap
saccades. Even memory saccades were only slightly
more variable than gap saccades.
Functional consequences: Our results are not consistent with the idea that the dependency of saccade
gain on initial eye position is compensated for at the
level of the brainstem burst generator. If this were
true, we would expect that the difference in the gain
of centrifugal and centripetal saccades would be the
same for all types of saccades. This is clearly not the
case. Our results are also not compatible with our
initial hypothesis that the cerebellum is less involved
in the compensation of position dependency of
internal saccades. That the gap saccades show a
dependency on position and scanning/memory saccades do not, implies that the compensation for
initial eye position is done upstream from the brain
stem burst generator where the saccade signals are
still handled by separate channels. Our results do
not reveal whether the position dependency of gap
saccades originates with orbital tissue and/or the
position dependency of the signals to the extraocular
muscles. Whatever the origin of position dependency, the signals diving the burst generator must be
different to explain our results.
In contrast to the gain the position dependency of
the peak velocity and the skewness was nearly
identical for all saccade types. This indicates that the
position dependency of the peak eye velocity does
not have the same origin as the position dependency
of the saccade amplitude and that the coupling
between both is not fully explained by the well
known amplitude±velocity relationship of saccades,
i.e. the main sequence. Peak velocity is not exclusively determined by the saccade amplitude but also
by the saccade type, the initial position, and the
direction. This is supported by our results concerning the reduced peak velocities in the memory task
which are in agreement with the results of Smit et
al. [13] and by the observation that scanning
saccades with identical amplitudes but different
positions and different directions have different peak
velocities (see Fig. 1).
Previous evidence indicates that position-sensitive
signals are present in several locations in the brain
upstream from the brainstem burst generator. The
brain stem burst generator receives direct input from
the superior colliculus. Eye position in¯uences the
activity of presaccadic neurons in the superior
colliculus of monkeys in such a way that the peak
®ring rate of the neurons is changed [14]. Whether
this position dependency of the neuronal discharge
re¯ects the cortical input or is a consequence of
NeuroReport
positional feedback signals from the nucleus prepositus hypoglossi to the superior colliculus is not clear.
There is also increasing evidence that the discharge of neurons in the lateral intraparietal sulcus
and in some other areas is in¯uenced by the orbital
eye position [15±21]. Microstimulation at different
locations of the intraparietal sulcus evoked saccades
with similar direction but different amplitudes depending on the eye position (vector saccades) [15].
Apparently the dependency of these saccades on the
orbital eye position was not completely compensated. In contrast, at distinct locations (e.g. the ¯oor
of the sulcus) the evoked saccades were independent
of the starting position (goal-directed saccades) and
convergent toward a goal in head-centered space
(compensating for the orbital position) [15]. Neuronal network models have been suggested to explain
how a combination of visual information in retinotopic coordinates and eye position signals can form
a representation in head-centered coordinates [16].
In these networks the vector saccade neurons represent an earlier step of the saccade processing, the
goal directed saccade neurons a later step. The
superior colliculus receives projections from the
lateral intraparietal area as well as from the frontal
eye ®eld [22]. Thus, a possible explanation for the
differences in the compensation for the orbital
position we observed is that the visual triggered
re¯exive saccades, which in our experiment always
had much shorter latencies than the internally
guided saccades, are generated by signals originating
more from neurons representing the target in retinal
coordinates, whereas the internally triggered saccades originated more from neurons representing a
target position in head coordinates.
Conclusion
The gain of externally triggered saccades depends
more on the orbital position of the eye than the gain
of internally triggered saccades. In contrast, the
position dependency of the peak velocity and the
skewness was similar for all saccade types. The
contribution of the cerebellum to the compensation
of orbital position dependencies of saccades cannot
explain these ®ndings. It seems more likely that the
dependence of the saccadic gain on the saccade type
is due to different cortical signals, which re¯ect
either more retinotopic or more craniotopic organization of the input to the brain stem burst generator.
References
1. Becker W. Metrics. In: Wurtz R and Goldberg M, eds. The Neurobiology of
Saccadic Eye Movements. Amsterdam: Elsevier 1989: 13±67.
2. Collewijn H, Erkelens CJ and Steinman R. J Physiol (Lond) 404, 157±182
(1988).
3. Robinson DA. Control of eye movements. In: Brooks VB, ed. Handbook of
Vol 10 No 12 20 August 1999
2669
NeuroReport
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
Physiology The Nervous System. Vol 11 Part 2. Baltimore: Williams and Wilkins
1981: 1275±1320.
Ritchie L. J Neurophysiol 39, 1246±1256 (1976).
Robinson FR, Straube A and Fuchs AF. J Neurophysiol 70, 1723±1740 (1993).
Pierrot-Deseilligny C, Rivaud S, Gaymard B et al. Ann Neurol 37, 557±567
(1995).
Wurtz RH. IVOS 37, 2131±2145 (1996).
Fuchs AF, Kaneko CRS, Scudder CA. Ann Rev Neurosci 8, 307±337 (1985).
Straube A, Deubel H, Spuler A and Biittner U. Neuro-ophthalmology 15, 67±74
(1995).
Van Opstal AJ and Van Gisbergen JA. Vision Res 27, 731±745 (1987).
JuÈrgens R, Becker W and Komhuber HH. Biol Cybernet 39, 87±96 (1981).
Pelisson D and Prablanc C. Vision Res 28, 87±94 (1988).
Smit AC, Van Gisbergen JAM and Cools AR. Vision Res 27, 1745±1762 (1987).
Van Opstal AJ, Hepp K, Suzuki Y and Henn V. J Neurophysiol 74, 1593±1610
(1995).
Thier P and Andersen RA. Proc Natl Acad Sci 93, 4963±4967 (1996).
Zipser D and Andersen RA. Nature 331, 679±684 (1988).
2670
Vol 10 No 12 20 August 1999
F. Eggert et al.
17. Stricanne B, Andersen RA and Mazzoni P. J Neurophysiol 76, 2071±2076
(1996).
18. Andersen RA, Bracewell RM, Barash S et al. J Neurosci 10, 1176±1196 (1990).
19. Bremmer F, Ilg UJ, Thiele A et al. J Neurophysiol 77, 944±961 (1997).
20. Bremmer F, Distler C and Hoffmann KP. J Neurophysiol 77, 962±977 (1997).
21. Bremmer F, Graf W, Ben Hamed S and Duhamel JR. NeuroReport 10, 1±6
(1999).
22. Blatt GJ, Andersen RA and Stoner GR. J Comp Neurol 299, 421±445 (1990).
ACKNOWLEDGEMENTS: We are grateful to J. Benson for her help in editing the
paper. This work was supported by the Deutsche Forschungsgemeinschaft (DGF)
and NIH grants RR-00166 and EY10578.
Received 21 April 1999;
accepted 24 June 1999