Participation
of Medial
Eye Movement
Pontine
Generation
Reticular
Formation
in
in Monkey
E. L. KELLER Department Research
of Electrical Laboratory,
Engineering University
and
Computer
of California,
THE DEEP CORE Of the brain stem, and in particular the med ial pontine reticu lar formation (PRF), has long been considered to be directlv involved in the immediate supranuclear control of eye movements. This belief is based on the demonstration of direct anatomic connections from this area to all the oculomotor nuclei (ZZ), on the severe and permanent disruption of eye movement that results from focal lesions placed in this area (3), and on the shortlatency eye movements that are evoked by electrical stimulation in this area (6). Single-neuron recordings in this or closely adjacent brain stem areas (5, 7, 8, 15, 24) have revealed several categories of neural activity that is closely and differentially correlated with various types of eye movemen ts. These previous studies have either been conducted on animals without the primate’s repertory of eye movements (7, S), have been brief, qualitative reports concerned only with segregating the units by type (5, 24), or have not considered unit behavior during smooth pursuit or vestibularly driven movements (15). It has been previously shown that each oculomotor neuron participates in all the types of primate eye movement (21). Therefore, in working out the details of the supranuclear organization of the oculomotor system the question that remains to be clarified is how and at what level signals from the various separate eye-movement-generating subsystems are brought together. In this respect it is important to have .more precise and detailed recordings on each type of reticular formation whose activity is correlated with
Received for publication 316
July 19, 1973.
Sciences
Berkeley,
California
and
the
Electronics
94720
all the types of eye movement in the primate repertory. This line of attack is initiated by this report on the PRF. Following the detailed recording of a unit’s behavior, the recordin .g site was el ecthe microel ectricall v stimulated through trode. The eye movements resulting from these focal stimulations were compared with the type of unit activity previously recorded at the site. Previous stimulation studies of this area carried out with gross, implanted electrodes at much higher levels only smooth eve of current (6) evoked movements. While t was no t expected that the small currents utilized in this studv would stimulate just the unit being recorded, the technique of microstimulation can eff ectivelv limit the area and, hence, number of neurons excited to very small, discrete colonies in cortical areas (1). In a structure like the reticular formation the affected area might be expected to be larger due to the extensive array and spread of connecting fibers throughout the area (22). Nevertheless, since the types of units recorded in this area respond very differently for various types of eye movements, some reflection of unit characteristics in the type induced bv stimulation of eye movement was expected. To a limited extent this expectation was fulfilled. METHODS
Six young monkeys (Macaca mulatta) were prepared for these studies by chronically impentobarbital planting three devices under sodium anesthesia and aseptic conditions. A coil of Teflon-covered, stainless steel wire was implanted on one globe. This coil, placed on an animal put in alternating magnetic fields kept in spatial and temporal quadrature, pro-
RETICULAR
FORMATION
vided a signal which is a function of vertical and horizontal eye position with a sensitivity of 15’ of arc. Details of this method of measuring eye movements are the same as those previously given (11) except that fields alternating at a frequency of 20 kHz were utilized. Steep attenuation of the gain of the microelectrode recording electronics above a frequency of 10 kHz prevented contamination of the neural recordings with signals from the magnetic fields. During the preliminary stages of the experiments, eye movements in the first two monkeys were measured with direct-coupled electrooculograms. Data from these two animals are pooled with that from the four later monkeys. A stainless steel chamber was implanted stereotaxically on the skull above the pontine reticular formation. Tungsten microelectrodes were driven through the chamber using an eccentric hydraulic-drive system. The dura was left intact and electrodes were passed through it within a sterile guard tube. A light aluminum crown was bolted to the skull to provide a platform for restraining the monkey’s head during recording sessions. Following recovery from these surgical procedures the monkeys were placed in a primate chair with heads immobilized for the daily recording sessions. The animals had been previously trained to attend to a visual display and to fixate a variable-position target lamp. With this simple target display a controlled and reproducible sequence of both small and large saccades and steady fixations at selected positions could be quickly elicited whenever a unit correlated with eye movement was located. Smooth movements were produced by rotating a large mirror in front of the monkey. Vestibular movements were obtained by manual rotations of a turntable on which the primate chair and magnetic field coils were rigidly mounted. In the last three monkeys, after unit activity had been recorded, the microelectrode was switched from the recording preamplifier to the output of a constant-current stimulator. The site was stimulated with brief pulse trains of variable-intensity cathodal current and the resulting eye movements were recorded. The stimulus parameters could be varied from currents of 5-40 PA, train lengths of 10 ms to several seconds, and intratrain pulse frequencies of 100-1,000/s. Pulse width was constant at 0.3 ms. Neural activity was coupled through an FET preamplifier to a conventional amplifier with a band pass of 300-10 kHz. Unit activity and voice annotations were recorded on direct chan-
AND
EYE
MOVEMENT
317
nels of a magnetic tape recorder with a band pass of 300-11 kHz, while simultaneous eye positions and a signal proportional to chair position were recorded on separate FM channels (band pass DC to 1,250 Hz) of the same recorder. For analysis the recorded data were transcribed on photosensitive paper by an ultraviolet mirror galvanometer recorder. The overall system bandwidth for eye movements was 250 Hz and for neural data was 5,000 Hz. The recorder was run at paper speeds of 125 or 250 mm/s for the analysis of unit discharge rates during fixation and 500 or 1,000 mm/s for saccadic activity analysis. The accuracy of the time measurements, determined by recording the output of a precision oscillator through the entire system, was within t 37& Electrolytic lesions were made through the tip of the recording electrode at the site of units At least two such lesions of particular interest. were made in each animal, but generally not more than three. Following the final track the animals were heavily anesthetized with pentobarbital and then perfused with normal saline and formalin. Subsequent histological examination (25-p sections stained with cresyl violet) verified the location of the lesions and the recording sites on other electrode tracks were reconstructed from the marked locations. RESULTS
in this study were run that portion of the brain stem reticular formation from about 2 mm anterior to 2 mm posterior of the rostra1 and caudal poles of the abducens nucleus. The lateral limit of the region explored extended out to about 3 mm from the midline. The ventral border of the area studied was generally established by the olivocerebellar tract. Some of the more posterior tracks passed through the vestibular nuclei, but eye-movement-related units isolated in this complex will not be described in this report. The tracks, run at a 25” angle from the sagi t tal plane, usually passed through the brachium conjunctivum. Soon after leaving this dense fiber track units related to eye movements began to be encountered. Most of the isolated units were judged to be cell bodies on the basis of their spike shape and duration (15). The behavior of 233 eye-movement-correlated units were recorded and analyzed in this study. Since the area of the brain stem studied and the qualitative response
Recording uniformly
tracks
into
318
E. L. KELLER
of the recorded units correspond very closely to that previously described by Luschei and Fuchs (15), the units will be divided for descriptive purposes into the same four main categories established by these investigators. Using their nomenclature we isolated 120 burst units, which exhibited high-frequency bursts of activity in association with rapid eye movements in a single or several directions (called the on-direction) but no sustained tonic activity, and 37 tonic units whose activity was related to steady eye position during fixation and, in some cases, eye velocity during pursuit or ves tibularly driven m9vements. Twenty units were isolated that paused during all rapid eye movements or for rapid movements in a particular direction. During periods of fixation between saccades these units exhibited a tonic discharge that was only poorly or not at all correlated with eye position or velocity. We also isolated 41 burst-tonic units which exhibited bursts of firing during rapid eye movements in a particular direction and tonic activity related to fixation positions in the same direction. In their behavior such units resemble oculomotor neurons (21) but were located outside recognized motor nuclei. In addition, track reconstructions always located these units outside the PRF, either in the several small nuclei located at this level in the central gray, in the vestibular nuclei, or in the medial longitudinal fasciculus. Therefore these responses were not considered as representative of activitv in the reticular formation and are not further described. Finally, 15 additional units were isolated in the reticular formation that displayed eye-movement-correlated behavior that could not be categorized under these four main headings and will not be discussed further. Single-unit
recording
UNITS. Bursting units could be further subdivided according to the preferred directions of the movements associated with the burst and according to the temporal relationship of the burst to the onset of the rapid eye movement (15). Medium-lead burst units. Sixty-eight
BURSTING
units were isolated in the current study which exhibited medium-lead bursts. These units invariably discharge a high-frequency burst of activity prior to the initiation of any rapid eye movement either visually or vestibularly evoked with a component of the movement in the proper direction. The defining criteria of proper direction was variable among the units, but the required component of the movement was strictly ipsilateral for only 16 units or about 25y0 of the medium-lead units observed. Figure IA illustrates the behavior of this type of burst-lead unit for saccadic movements, while Fig. 1B shows similar behavior for the same unit during the quick phase of rotatory nystagmus. It should be particularly noted that these u .ni ts never discharge during pure vertical or medial rapid movements whi ch ser ves to differentiate them from the larger population of medium-lead burst units to be described later. The lengths of the burst lead in milliseconds, the duration of the burst, and the intraburst firing frequencies were closely examined to determine if any quantitative differences existed for visually evoked or vestibulary driven rapid eye movements. No detectable difference could be determined in the burst lead. The mean burst lead-determined from at last 25 rapid movements for each unit-was the same for both saccades and quick phases for all 16 units, and varied from 7 to 12 ms with an overall mean of 9 ms for the group as a whole. This is very similar to the burst leads found by Luschei and Fuchs (15) for this type of unit for saccadic eye movements. In comparison the average burst lead for abducens motoneurons is 6 ms (14). Likewise there was no detectable difference in burst length for rapid eye movements of either type for movements of similar duration. In confirmation of the findings of Luschei and Fuchs (15) the burst length and rapid eye movement duration were almost identical for movements of duration greater than 15 ms (corresponding to a movement of about 5”). A comparison of the intraburst discharge rate during ipsilateral saccadic and quick phase eye movements was obtained by separately pooling the data for 10 saccades and
RETICULAR
FORMATION
AND
EYE MOVEMENT
319
/-------
LAT.
3 W
‘--\----=
MED.
r
I
I
I
0 TIME
(MSEC)
FIG. 1. Activity of an ipsilateral medium-lead burst unit during saccadic eye movements (A) and the movements of rotationally induced vestibular nystagmus (23). In A and B upper trace is quick-phase vertical and middle trace is horizontal eye position. The time calibration for both shown below B. Insets in A and B are high-speed records of one horizontal rapid eye movement and associated unit activity. Time calibration shown below B the same for both insets. Eye-movement calibration the same as in lowspeed records.
for 10 quick phases of 40°, 20°, and 5” deviations for each of the ipsilateral medium-lead burst units. Reproducible saccades of these sizes (within -F- 0.5’) were readily available because the animals were trained to fixate the target positions. As a general rule, quick-phase movements of reproducible size were not obtained and
therefore the averaged data were obtained from 10 movements of the above sizes -F- 2”. The onset of the burst was abrupt and the discharge rate reached levels as high as 950 spikes/s which were sustained until the final 20-30 ms of the burst, during which time a decreasing and more variable discharge obtained. For rapid eye movements
320
E. L. KELLER
greater than loo an identical high level of rather steady discharge was reached. For movements less than 5O the discharge rate was significantly lowered. Statistical analysis of these averaged burst rate profiles indicated that there was no significant difference in intraburst firing frequency for visually evoked and vestibularly induced rapid movements. Finally it was found that the burst rate was not dependent on the initial eye position within arange of -t- 20” from the primary position. Although burst rate increased very little for saccades larger than lo”, it was related nonlinearly to saccadic velocity. Measurements of maximum saccadic velocities for a range of movement amplitudes were pooled for all the animals utilized in these experiments. The curve drawn through the plotted means of these measurements shown in Fig. 2A illustrates the well-known dependence of saccadic velocity on movement amplitude (9). On this same figure are plotted the overall mean burst rates (and ranges of the means) for the 16 ipsilateral bursting units during saccades of the same amplitude for which maximum velocities of eye movement are plotted. It is apparent that burst frequencies show a much more complete and earlier saturation for larger saccades than does saccaclic velocity. To clarify the relationship between burst frequency and saccadic velocity the means from Fig. 24 are replotted as the filled circles in 2e, which shows that saccadic velocity depends mono tonically on burst frequency. The amplitude of the saccade associated with each datum point is shown in parentheses. The data plotted in Fig. 2A were obtained with the monkey training and attempting to fixate the randomly appearing target lights. Both maximum saccadic velocity and movement duration showed the expected relationship-for fully motivated monkeys-to saccade amplitude. If the target display was turned off, the monkey would continue to make spontaneous saccades to various objects of interest around the experimental room. Under these conditions it was always possible to find saccades in the records with abnormally long durations and lower maximum velocities when compared to movements of corresponding
1000
1000
0
0
I
I
I
IO
20
30
AMPLITUDE
L
1
40
( DEG 1
6
.(40)
800.(30) l (20)
C 8 co \
--28'
ii! Y
-33
4 (15) .
A
::CJ
. (IO)
4W
-
>
A
8. IO'
A
-I
o(6)
014)
so -A l (2) 0
f
0
I
1
200
400
BURST
DISCHARGE
I
I
600 RATE
800 ( SPIKES
I
1000
/ SEC)
2. A: relationship between saccadic amplitude and maximum saccadic velocity (filled circles). Each point represents the pooled mean from at least 10 saccadic movements of each size for each of the six monkeys. Also shown is the relationship between saccade size and the associated burst frequency for medium-lead burst units (open circles). Each point represents the grand mean of the burst rates associated with the 10 saccades of each size for all 16 ipsilateral burst units (bars are ranges of means for the 16 units). B: means (burst rate and velocity) from A replotted to show the relationship between intraburst firing frequency and saccadic velocity (filled circles). Numbers in parentheses are amplitudes of saccades associated with each point. Preceding data all obtained with animals executing changes in fixation angle in response to target movement. Triangles show maximum saccadic velocity and associated burst rate for several abnormally slow spontaneous saccades of the size shown on the interrupted horizontal lines. FIG.
amplitude in Fig. 2A. Nevertheless, all 16 units continued to burst for every saccadic movement with an ipsilateral component. Moreover, the burst continued to be of almost the same duration as the ipsilateral
RETICULAR
FORMATION
AND
Values
Mean 1.9 1.7 1.4 1.5 in parentheses
321
number of spikes per degree of eye movement (mean burst units during horizontal, ipsilateral saccades
Saccade Size, O 5 10 20 30
MOVEMENT
units also discharged only in association with rapid eye movements but with a more complicated determinate for the on-direction of movement. Each unit in this group displayed a burst for every rapid movement with an ipsilateral component, but the ipsilateral component was not a necessary condition for a burst. No units were found that were exclusively active for pure vertical saccades. Specifically, 10 units were isolated that burst for both ipsilateral and contralateral rapid movements. In every case the burst frequency was greater for movements with an ipsilateral component. Twenty-one additional units were located that burst for all rapid movements, but nine of these showed the greatest discharge frequency for movements with an ipsilateral component, five units the greatest frequency for movements that contained a down component, and six the greatest frequency for movements that contained an up component. Ten units burst for saccades in all directions except pure contralateral movements. For each of these 10 units the greatest discharge frequency was associated with rapid movements having an ipsilateral component. Finally, 11 units produced a burst with every ipsilateral or vertical saccade or combination of the two, however, almost complete inhibition of the burst, which normally accompanied the verticle component, was a concomitant of every saccade that contained even the slightest contralateral component. Long-lead burst units. Forty units were isolated that began to discharge substantially before rapid eye movements with a variable interval of uneven or ragged firing. The length of this interval was extremely variable from saccade to saccade for a given unit and among different units. Figure 3A shows a unit whose behavior is
component even for saccades of very low velocity that occurred as the animal became less alert. Such movements are probably the result of a continuous scale of central nervous system alertness from fully alert and motivated to drowsy, and would not normally be of interest for study. However, in this case, the characteristic behavior of the burst units is clarified by the depressed state of alertness. Not only do such units continue to discharge with bursts of the same duration as the movement, but the intraburst discharge rate remains correlated with velocity and not amplitude of the movement. This is shown on Fig. ZB for a few selected “slow saccades.” The triangles show the paired velocity and burst frequency associated with saccades of the size shown on the interrupted horizontal lines. When this number is compared with the number in parentheses for the nearest filled circle (fully motivated saccade) it is apparent that these movements have much lower velocities than “normal saccades” for the given amplitudes and yet the associated burst frequencies continue to plot in correspondence with the velocity of the movement. Since the mean burst rate increased very little with saccade size and movement duration increased almost linearly, the number of spikes per degree of movement calculated by dividing the total number of spikes in a burst by the size of the associated movement, might be expected to be rather constant over a wide range of movements. In fact this relationship (gain factor) was surprisingly constant, as shown in Table 1, for a sample of the ipsilateral medium-lead burst units recorded. Spot checks on the remaining 10 ipsilateral bursting units showed that they behaved similarly. The remaining 52 medium-lead burst TABLE 1. Average for six medium-lead
EYE
(.3) (.2) (.3) (.5) are SD. Data
2.5 2.4 2.7 2.4
(.4) (2) (.5) (.4)
are taken
Gain
2.2 2.3 2.5 2.2 from
(.4) (.3) (.6) (.4)
at least
Factor,
Spikes/ 1.8 1.9 2.1 1.9
five saccades
gain
factor)
O
(.4) (.3) (.4) (.7)
2.2 2.3 2.3 2.1
of each size.
(.2) (.4) (.2) (.5)
c.1 3.4 2.9 3.2
(.3) (.6) (.5) (.7)
E.L. KELLER
322
typical of this group. The prelude of uneven discharge, which was not correlated in frequency or duration with saccade size, began between 30-300 ms before the onset of a saccade and terminated with a more intense burst of activity which also preceded the movement. The high-speed record in Fig. 3A illustrates the large variability in interspike interval during the low-frequency prelude and during the more intense burst. This variability makes definition of the burst duration and initiation time somewhat arbitrary. For 22 units in this group the duration of the period of more intense activity corresponded very roughly with the duration of the saccade. For the remaining 18 units the period of intense activity was of much shorter duration than the saccade, so that the burst ended almost synchronously or slightly after the onset of the movement. The burst lead was more variable among these units than had been the case for the medium-lead group, but was fairly consistent for any one unit. The distribution of the burst lead ranged from 28 to 8 ms with the group mean at 14 ms. Thus, in this group the period of more intense activity tends to be initiated before the burst in the medium-lead group. The on-direction of the units in the longlead group included 27 ipsilateral, 5 ipsilatera1 and down, 3 ipsilateral and up, and 2 omnidirectional. z
LAT.
A
All the units of this group continue to display their characteristic uneven interval of discharge and period of more intense activity prior to every rapid movement in the on-direction during vestibular nystagmus. The extreme variability in unit discharge even for similar saccades made any quantitative comparison of the possible differences occurring during quick phases and saccades very difficult. Figure 3B illustrates that the pattern of discharge was essentially the same with respect to the length of the uneven discharge interval and the burst lead. Following-lead burst units. The consistent characteristic of this group of 12 units was that they never initiated their burst prior to the onset of rapid eye movements, but instead began several milliseconds after the movement. The group continued to show similar behavior during rotatory nystagmus. That is they still fired bursts of irregular activity at variable short intervals after contralateral quick-phase movements. UNITS. Thirty-seven units were isolated which discharged steadily during fixation at rates that increased monotonically with ipsilateral eye position. Figure 4 shows the behavior of a unit representative of this group. The majority of units in this group (26 units) had thresholds-the eye position at which the neuron began to discharge tonically-in the nasal field of movement or were never totally inhibited even
TONIC
B
FIG. 3. Activity of a long-lead burst unit during saccadic eye movements (A) and the quick-phase movements of vestibular nystagmus (B). Eye movement and time calibration the same in A and B. Only the horizontal component of eye position shown. Insets below A and B are high-speed records of one horizontal rapid eye movement of each type and associated unit activity.
RETICULAR
-
FORMATION
EYE
MOVEMENT
323
A
UP 20
I
g -
AND
/
01
2OJ
k
DOWN
C
1
0 FIG.
tional Time
1
TIME
I
(MSEC)
1
1
400
4. Activity of tonic units during gaze fixation (A), optokinetic nystagmus (B), and vestibular rotanystagmus-in total darkness (C). Only horizontal-component of eye movement shown in B and C. and eye movement calibration the same for all traces.
for extreme medial gaze fixations. The remainder of the cells in this group (11 units) had thresholds in the lateral field of movement. The distribution of units into these two groups was far from discrete however, as demonstrated in Fig. 5B which shows a distribution of tonic unit thresholds from the most medial fixation positions and beyond (cross-hatched area) to about 25” into the lateral field. There was tendency for the units with lateral thresholds to have more linear frequency-eye position curves of greater slope. About 75% of the units with nonlinear curves showed a saturation effect (decreased slope at more lateral positions) and the other quarter had increased slopes at more lateral eye positions. Figure 5A
shows three rate-position plots for units from this group that are representative of the range of behavior displayed during fixation by tonic units. Whenever possible tonic units were tested carefully for response during pursuit movements. A consistent dichotomy emerged for the 12 units that were held long enough to ascertain their behavior during pursuit movements. The relationship of firing rate to eye velocity was determined by plotting the instantaneous firing rate at a selected position (usually for a position on the most linear portion of the rate-position curve for the unit, but in any case always the same position for each rate-velocity analysis for any one unit) for a number of different
324
E.L.KELLER
velocities through the chosen position. Instantaneous firing rate was estimated by averaging about six interspike intervals which most closely symmetrically bracketed that point in time at which the eye passed through the selected position. An example of the data from which these measurements were made is shown in Fig. 4B where firing rates were determined during a pursuit movement directed in one case through a position in the lateral direction (upward arrow) and then through the same position in the medial direction (downward arrow). When a number of such firing rates were plotted for different velocities, curves similar to those shown in Fig. 6 resulted. Three units, all of which had extreme medial thresholds during fixation, were similar to unit 1 with essentially no change in firing rate due to velocity for either laterally or medially directed pursuit movements. The other nine units, all of which had lateral thresholds, discharged at rates greater by an amount proportional to eye velocity in the lateral direction but a constant rate for various velocities in the medial direc-
0
/
f
I
I
I
I
I
I
0 HORIZONTAL
B
EYE
POSITION
40 ( DEG)
-6 I 40
30
20
IO
0
IO
MEDIAL
I
1
!
30
40 LATERAL
TONIC
FIG. 5.
20
FIRING
THRESHOLD
(DEG)
A : relationship of unit discharge rate during fixation to horizontal eye position for three Data points (filled circles) obtypical tonic units. tained during visual fixations are shown for only one unit (unit 1). Curves that indicate the trend of data are drawn by hand. Instantaneous firing rates measured for a number of eye positions during the slow-phase of rotatory nystagmus in the dark (filled circles) are shown for unit 1 for comparison. 8: distribution of thresholds of 37 tonic units recorded in this study. Gaze deviation required to initiate tonic discharge is divided into 50 bins on the absissca. Cross-hatched bin on the extreme left contains all units that were still active at the limit of contralateral gaze.
300,
UNIT
I
UNIT
2 4
-20
EYE
0
0
VELOCITY
(DEG
/ SEC1
FIG. 6. Relationship between unit discharge rate and eye velocity for two tonic units. Discharge rates for visual pursuit velocities (open) and vestibular slow-phase velocities (filled) in both the ipsilateral (+) and contralateral (-) directions at instants when the eye passed through a given position for each unit.
tion. The slope of the rate-velocity relationship for lateral velocities was correlated with threshold in that units with more lateral threshold also had the largest increase in rate for lateral following movements. Nine of the twelve tonic units studied during pursuit movements were also tested for their respon.se during rotatory nystagmus in total darkness. All the units participated by smoothly increasing or decreasing their firing rate during lateral or medial slow-phase movements, respectively, as shown in Fig. 4C. The instantaneous firing rates at a number of positions were calculated by averaging about six interspike intervals about the point in time at which the eye crossed each position. For three of these units, when their firing rates were plotted on rate-position graphs for the same unit (one example in Fig. 5A), they fell with some slight scatter on the same estimated curve for the unit obtained during visual fixation. These were the same three units which had not shown any change in discharge rate related specifically to the velocity of pursuit movements. The remaining six units, which had displayed a velocity proportional increment in firing rate for lateral pursuit movements, also showed a similar relationship during slowphase movements (unit 2, Fig. 6). However, when the discharge rate-slow phase velocity relationship was plotted on the same coordinates as the rate-velocity relationship
RETICULAR
FORMATION
for pursuit movements, the rate data points for slow-phase velocities always plotted below the estimated curve for rate-pursuit velocity relationship as shown in Fig. 6. PAUSING UNITS. These 20 units fired rather steadily during fixation but stopped firing completely slightly before and during rapid An example of the beeye movements. havior of this type of unit is shown in Fig. 7. The discharge rate during fixation was not at all correlated with eye position for 12 of these units, but for the other 8 there was a consistent trend toward higher discharge rates for more ipsilateral gaze deviation. The typical standard deviation of the discharge rates for repetitive fixations at the same angle was ZO-257& of the mean rate. This wider range of scatter was in marked contrast to similar measurements on the tonic units in this study and on abducens motoneurons (14) where typical standard deviations of discharge rate of 10 and 57& were found, respectively. For the pausing group discharge rates ranged from 180-250 spikes/s for the units not correlated with eye position, while those that displayed maximum rates for ipsilateral positions never increased to more than 150 spikes/s. Thus discharge rate served to further differentiate these two subtypes. The occurrence of a pause was directionally sensitive in that the 8 position-correlated units paused only for saccades with ipsilateral components, while the remaining 12 units paused for all rapid eye movemen ts.
AND
EYE
325
MOVEMENT
The onset of the pause for both subtypes was abrupt and preceded the initiation of the rapid movement by an interval that ranged from 12 to 25 ms for the 20 units with no systematic difference between the sub types. The length of the pause was almost equal to the duration of the rapid movement, which confirms the findings of Luschei and Fuchs (15) in opposition to the power-law relationship found by Cohen and Henn (5). There was no consistent difference in either pause lead or duration for saccades or quick-phase movements in any of these pausing units. None of the high-frequency, non-position-related units showed any detectable difference in discharge rate correlated with slow-phase eye movements during rotatory nystagmus. The eight other pausing units changed rate smoothly during rotation (and the concomitant compensatory slow-phase eye movements) in addition to pausing for ipsilateral quick-phase movements. Since the discharge rate-position relationship was so inconsistent for these units it was difficult to make a quantitative measure of discharge rate during rotation. However the discharge rate during one direction of rotation could be compared to the immediately succeeding rotation in the opposite direction during sinusoidal rotation. When this was done it was found that the maximum instantaneous discharge rate always occurred in phase or slightly lagging maximum contralateral rotational velocity and, hence, maximum ipsilateral slow-phase eye movement.
UP
20-l DOWN
LAT.
MED.
FIG. 7. Activity of pause units during a sequence of eye movements. A: unit that Unit firing rate only for saccades with an ipsilateral component of eye movement. fixation is also correlated with ipsilateral position. Time calibration in milliseconds.
pauses during
completely periods of
326
E. L. KELLER
Microstimulation Luschei and Fuchs (15) in their singleunit study of a wider area of the alert monkey brain stem that included the pontine reticular formation attempted without clear success to anatomically segregate the anterior-posterior locations of units of similar discharge characteristics on the basis of several coronal slices of the brain stem through the area bracketing the abducens nucleus. This question has been reexamined here on the basis of the units’ mediolateral location instead. Since the electrode tracks all penetrated the reticular formation at a 25” angle from the vertical and the point at which the electrode crossed the midline was usually easy to determine from the background activity, we could calculate the approximate lateral location with respect to the midline of each unit that we recorded. Sagittal sections of the brain stem of 0.5 mm width were graphically constructed and the responses recorded in each section were grouped according to unit type as shown in the histogram of Fig. 8. While considerable overlap exists, it does appear that medium-lead burst units with complexly conditioned on-directions, are grouped closer to the midline than medium-lead burst types that discharge only for saccades with an ipsilateral component. On the other hand, tonic types were almost evenly distributed throughout this 3 mm extent of the PRF. Long-lead burst types tended to be distributed similarly to medium-lead typesthat is, ipsilateral on-direction units were located more lateral than bilateral, vertical, and omnidirectional types-but they are not included in the histogram for clarity. The location of the subtype of pausing cell that interrupted its high frequency, steady discharge for all rapid eye movements was more discrete. All these units were recorded in a small circumscribed area just anterior to the rostra1 pole of the abducens nucleus. Once a pausing unit of this type was found, others were always located nearby, in contrast to the situation elsewhere in the reticular formation where medium-lead and long-lead burst units were found closely intermingled with each other and tonic units.
NUMBER OF UNITS
1 ,.
0 UNIT
LOCATION
(MM)
FIG. 8. Schematic drawing of a coronal section through the brain stem at the level of the abducens nucleus. Electrode tracts were run into the pontine reticular formation within the approximate area shown by the dashed lines throughout 5-mm-thick sections. The histogram below the section was constructed by summing the number of omnidirectional medium-lead burst units (solid bars), the ipsilateral medium-burst units (open bars), and the tonic units (crosshatched bars), respectively, encountered in each 0.5-mm sagittal section of brain stem throughout the entire 5-mm-thick area. BC, brachium conj unctivum; Gn VII, genu of the facial nerve; VI, abducens nucleus; IV, 4th ventricle; IO, inferior olive.
The other type of pausing unit, which discharges at lower tonic rates interrupted only for ipsilateral saccades, was found scattered throughout the PRF. This may explain the difference in our results from that of Luschei and Fuchs (15) who found no clustering tendency for pausing units, since thev did not differentiate between the locations of the two subtypes. When the sites of units located about 1 mm or more from the midline were stimulated, one stereotyped form of eye movement resulted regardless of the type of unit -medium-lead, long-lead, tonic, or ipsilateral pausing type-which had been recorded at the site. These movements were qualitatively similar to those reported to be evoked by order-of-magnitude higher currents from gross electrodes permanently
RETICULAR
FORMATION
implanted in the PRF (6). An example of the eye movement resulting from these stimulations is shown in Fig. 9. The movement began about 10 ms after the onset of the stimulus, which was a typical latency that did not depend on the type of unit recorded at the site. The eye accelerates smoothly through a transient phase to reach a constant velocitv of rotation which is maintained continuously for as long as the stimulus is presented. The transient phase of the evoked movement may be closely compared to that of a normal pursuit movement (upper trace) evoked in the same animal by rotating a mirror in front of its eyes. The eyes are accelerated in both cases to constant velocity in 50 ms, which is about one-half the time constant of a first-order approximation of the eye muscle to orbital mechanical system (13). This observation indicates that an excess rate of change of force above the constantly increasing force required to maintain the steady-state rotation must be present during the initial stage of the stimulated movement in analogy to that present during the initiation of a pursuit movement (16). Thus the hypothesis that the eye movements evoked by stimulation in this area are the result of a neural integration of the stimulation train (6) is an oversimplification. The velocity of the smooth, stimulusevoked movement increased linearly with stimulation frequencies up to 500/s, then gently saturated for frequencies up to 800/s, and finally decreased sharply at 1,000/s-a frequency above that previously investigated (6) but of importance since PRF units fire at rates above 800/s. The movement
1
I STIMULUS
TIME
(MSEC)
9. Stimulus-evoked eye movement (middle trace) compared to smooth pursuit movement (upper trace) elicited in the same animal by mirror rotation. The lower trace shows the envelope of the stimulus which was a lOO-ms train of 500/s, 20 PA, 0.3 ms pulse width, current impulses. FIG.
AND
EYE
MOVEMENT
327
velocity was also dependent on stimulus current strength, a parameter not previously investigated. Evoked-eye-movement velocity increased linearly with currents from 5 to 40 PA. At this higher level of current strength other facial movements began to be evoked, so the effect of currents above this level were not explored. Finally it was noted that the velocity of the evoked smooth movements were not position dependent for a range of initial positions -+ 20” on either side of the primary position, which confirms previous observations (6). When sites were stimulated in the most medial area of the reticular formation extending out about 1 mm from the midline, some differences in the evoked eye movement were noted. In all cases (12 sites) when the unit recorded at the site had a vertical component in its on-direction the stimulated movement, although still a smooth movement, also had a vertical component in the same direction. Eye movements with vertical components were never obtained with gross stimulation of the PRF (6). The ipsilateral, horizontal component always predominated, so that the stimulated movements were never at an angle of more than 20” from the horizontal except as noted below. The second difference that was noted on stimulating medial locations, and this was true regardless of the type of unit recorded at the site, was that the evoked eye movements had become initial-position sensitive. The horizontal component of the movement was most noticeably affected. With the eye positioned initially at extreme medial deviations the highest ipsilateral, constant velocities were obtained. When the initial position was at smaller medial deviations the horizontal velocity was lower and, hence, for a constant-duration stimulus the amplitude of the evoked movement was smaller. The effect became more pronounced as the initial eye position moved to lateral deviations, until at some angle of lateral gaze (typically about 20”), no further ipsilateral component could be evoked and the stimulated movement became purely vertical. When the stimulated site was located in the area just rostra1 to the abducens nucleus
E.L.
328
that contained the high-frequency omnilateral pause units (10 sites), the results were quite dramatic. Even at very low current levels the outcome was the almost complete inhibition of further voluntary saccadic movements for the duration of the stimulation. Figure 1OA shows an example of one such stimulation. Prior to the stimulus onset the monkey was looking spontaneously at objects around the laboratory with saccades occurring on the average about every 0.5 s. When the stimulus was applied and maintained almost all saccades were eliminated although the viewing conditions had not changed. In fact, all attempts to evoke saccadic movements, including moving threatening objects near the monkey’s
B ^h
-
c -
II 0
TIME
(SEC)
5
Results of high-frequency stimulation (200/s) in the area containing saccade pause units. In each case the upper trace is vertical eye position, the middle trace horizontal eye position, and the lower trace the stimulus envelope. Eye-movement calibrations are the same for all traces. In B the lowest trace is primate-chair (animal-head) position -calibration given on the left is 500. Time calibration is the same for all traces. A: stimulation delivered with the animal making - spontaneous saccades about the experimental room. -Usually no further saccades would occur for the duration of the stimulation. B: stimulation delivered during rotational vestibular nystagmus. Only the quick-phase movements were affected by being completely eliminated. C: stimulation delivered during optokinetic nystagmus driven by rotating a mirror in front of the monkey’s face. Quick-phase movements are eliminated and the gain of the smooth-phase system reduced. FIG.
10.
KELLER face and making loud noises at eccentric positions, failed. The quick phases of vestibular and optokinetic nystagmus were also totally eliminated by stimulation of the pausing area as shown in Fig. 1OB and C. The slow phase of vestibular nystagmus was not effected by the stimulation, but the velocity of the smooth phase of the optokinetic nystagmus was also greatly reduced during the period of stimulation, even though the mirror rotation inducing the optokinetic nystagmus continued at the same rotational velocities. DISCUSSION
Saccadic pulse generation Since the transmittance characteristics between presynaptic terminals and oculomotor neurons is unknown, the form of the input signal required to produce the pulselike increase or decrease in motoneuron firing rate during saccadic movements (10, 17, 23) is uncertain. Nevertheless, it has been demonstrated that oculomotor neurons respond to 55-ms steady depolarizing current pulses injected through an intracellular electrode with a steady discharge of spikes of up to 400/s or higher for the same duration as the current pulse (2). This activity resembles very closely the activity observed in typical oculomotor neurons during saccades. Thus, it seems highly likely that the input required by the motoneuron to produce its saccade-associated pulse of activity is a pulse of presynaptic activity of the same duration. Luschei and Fuchs (15) have hypothesized that the unilateral, medium-lead burst units provide the input which generates this high-frequency burst of activity in motoneurons during saccades. This hypothesis was based on the relative time of onset of burst unit discharge and their burst duration. The results presented in this study on the burst lead and duration of similar units supports this hypothesis. In addition, the detailed analysis of intraburst frequency during saccades of different velocities adds additional weight to the argument. The consistent relationship between saccadic velocity and burst frequency for a variety of levels of alertness also argues in favor of the directness of the connections to oculomotor neurons.
RETICULAR
FORMATION
The question now arises concerning the nature and location of the neural mechanism that generates the precise, saccadelinked burst of activity in the medium-lead burst units themselves. It is clear that a spatial-temporal translation of information must occur between cortical or collicular sites where saccadic movements are coded in terms of anatomical location (19, 20) and the PRF where they seem to be coded by the duration of the burst of medium-lead burst units. The activity of the pausing units, also recorded in this area, sheds some light on intermediate form of this the possible mechanism. On the basis of the single-unit behavior and on the results obtained by stimulation, we propose that the pulse-gating mechanism includes an input from the high-frequency pausing units. The input from this group of neurons could exert a powerful inhibitory influence on all medium-lead burst units and prevent them from discharging except during rapid eye movements when input from the pausing group is interrupted for the brief interval of the movement. This hypothesis is consistent with the observation that the average onset of the pause in this type of unit occurs before the average initiation of the burst in medium-lead burst units. Also stimulation of the pausing group, which presumably prevents the group from pausin
