- Email: [email protected]
Size and distribution of movement fields in the monkey superior colliculus
Brain Research,
113 (1976) 21-34 0 Elsevier Scientific Publishing Company,
SIZE AND DISTRIBUTION SUPERIOR COLLICULUS
Amsterdam
- Printed in
OF MOVEMENT
The Netherlands
FIELDS
21
IN THE MONKEY
DAVID L. SPARKS, RICHARD HOLLAND and BARTON L. GUTHRIE Department of Psychology and Neurosciences mingham, Ala. 35294 (U.S.A.)
Program,
University of Alabama in Birmingham,
Bir-
(Accepted January 20th, 1976)
SUMMARY
A gradient of response magnitude was observed across the movement fields (the range of eye movements which alter the discharge frequency of a neuron) of neurons in the intermediate and deeper layers of the superior colliculus. A vigorous discharge preceded movements with a particular direction and amplitude but reduced responses preceded movements which deviated from this direction and/or amplitude. Movement field size is a function of the amplitude of the optimal movement. Neurons discharging prior to small saccades have small and sharply tuned fields. Neurons discharging prior to large saccades have large movement fields and tuning is relatively coarse. Movement fields are topographically organized within the superior colliculus. Neurons discharging prior to small saccades are located anteriorly; neurons firing before large saccades are found caudally. Neurons near the midline discharge prior to up movements and neurons located laterally fire before downward movements. Movement fields of superior colliculus neurons are also characterized by a temporal gradient. The interval between spike discharge and the onset of a saccade is greater for movements near the center of the movement field than for movements to the periphery of the field. Results are interpreted as supporting the foveation hypothesis of superior colliculus function. It is suggested that precise saccadic movements are not produced by the discharge of a small population of finely tuned neurons but result from the weighted sum of the simultaneous movement tendencies produced by the activity of a large population of less finely tuned neurons.
INTRODUCTION Earlier suggestionslpsg7 that the superior colliculus plays an important role in the control of eye movements are supported by two lines of recent evidence. Firstly,
22 electrical stimulation of the superior colliculus in alert monkeys elicits short-latency saccadic eye movementsr3v1a. The amplitude and direction of these movements are a function of the site of stimulation within the superior colliculus rather than parameters of the stimulus train. Saccades have up components if elicited by stimulation of medial regions of the superior colliculus and have down components if elicited by stimulation of lateral regions. Elicited movements are small rostrally and large caudally. Secondly, chronic microelectrode recordings have shownJ+l*J0sa1+2” that the maximal discharge of neurons in the intermediate and deeper layers of the superior colliculus occurs prior to eye movements with a particular direction and amplitude. These recent observations were the basis of the proposal17 that the superior colliculus is part of a system which acquires visual targets for fovea1 viewing. According to this ‘foveation’ hypothesis, the superior colliculus is involved in coding the location of an object relative to the fovea and in eliciting saccadic movements which produce fovea1 acquisition of the object. One of the criticismsa” of the foveation hypothesis is that the movement fields (the range of eye movements which alter the discharge frequency of a neuron) of superior colliculus neurons are relatively large. If, however, the response of superior colliculus neurons is sharply tuned to particular movements, this argument may not be critical. The major purpose of this experiment was to describe, in more detail, the movement field properties of neurons in the deeper layers of the superior colliculus. Secondary purposes were: (1) to explore, further, the topographical organization of the movement fields and (2) to examine the temporal relationship between the spike discharge and the onset of associated saccades. Preliminary reports of these findings have been presented elsewhere20J1. METHODS
Five rhesus monkeys (Macaca mulatta) weighing from 2.5 to 4 kg were used. Each animal underwent 3 aseptic surgical procedures. First, 4 stainless steel screws were implanted in the skull to permit immobilization of the animal’s heads during training and recording sessions. Approximately one month later, a coil of fine wire was implanted beneath the bulbar conjunctiva and recti insertions. The ends of the coil were led subcutaneously to a connector mounted on the skull. After 4-6 weeks of behavioral training, a stainless steel receptacle for a microdrive was secured to the skull. During training and recording sessions, monkeys were seated in a primate chair located in an electrostatically shielded and sound-resistant room. Horizontal and vertical eye position signals were obtained (with a sensitivity of 0.25”) using the magnetic coil technique of Fuchs and Robinson4. Extracellular unit potentials were recorded with tungsten microelectrodes insulated with teflon and glass. Electrodes were advanced through a 16-gauge stainless steel cannula, extending 2-3 mm below the dura, with a hydraulic microdrive (Kopf). Recordings were filtered for frequencies above 5 kHz to reduce contamination with the 18 kHz signal of the alternating magnetic fields. Spikes were displayed on an oscilloscope, fed into a discriminator trigger unit, and transcribed on tape (band pass of 40 Hz-19 kHz). Other
23 FM tape channels (band pass D C to 1250 Hz) were used to record trial, stimulus and response events, vertical and horizontal eye position, and vertical and horizontal target position. Electrolytic lesions were made at the bottom of electrode paths in which responsive units were encountered. Fifty micron sections through the brain stem were mounted and stained with thionin. Recording sites were determined from histological sections and measurements of electrode positions taken at the time of the recordings.
Behavioral training Monkeys were trained to track a visual target, subtending a visual angle of less than 0.1 °, presented on a Hewlett-Packard 1310A oscilloscope (CRT viewing area of 27 by 37 cm). Target position was controlled by a PDP-8/I computer. Appropriate programs z2 permitted the target to be displayed at any point on a 4096 by 4096 matrix on the screen or to be moved at specified rates from one position to another. On training and recording days, animals were water-deprived for 23 h. Reinforcement was 0.1 ml of water delivered through a tube near the animal's mouth. During initial phases of training, the animal's eye movements were observed by the experimenter. Intermittently, the target was presented at the center of the oscilloscope screen. If the animal appeared to 'look at' the target, a reinforcement was given. When the animal reliably oriented the onset of the target with short latencies, the required fixation time was increased, gradually, to 2 sec. Next, the target was presented at the center of the screen and after a variable fixation time (1-3 sec) was moved instantaneously to a second position. If the animal followed the target with an eye movement o f short latency and fixated it for 2 sec, a reinforcement was delivered. These tracking movements were used to calibrate the eye movement recording system. During subsequent sessions, target presentations and reinforcements were under computer control. Horizontal and vertical eye position signals were sampled each msec and compared with target position.
Data collection and analysis During data collection trials, the animal was permitted 500 msec to acquire the initial target presented at the center of the screen. After a variable fixation interval (1-3 sec), the target was moved to a second position. If this target was acquired within 500 msec and fixation maintained for two additional seconds, a reinforcement was given. If not, the trial was terminated, and after a delay, a new trial was initiated. The origin of the target and the angle and radius of target movements were controlled by the experimenter via on-line interaction with the computer. A computer-generated display of horizontal and vertical eye position and spike activity was presented at the end of each trial. The movement fields of units were plotted by requiring animals to track target movements which varied systematically in radius and angle. These movements were always initiated from a 'straight ahead' position. An attempt was made to record unit activity during at least 3 movements to each sampled position. To ensure that movement-related discharges were not the result of retinal image displacements, the
24 activity of each unit was also observed during spontaneous eye movements occurring in darkness. The origin of the target movement was varied occasionally to determine whether or not saccade-associated activity was independent of initial eye position. For each reinforced tracking trial, the difference between the number of spikes occurring during a 500 msec period of center target fixation and the number of spikes occurring in the 500 msec following target movement was computed. This difference is referred to as the burst-index and provides an indication of the strength of neural discharge associated with saccades initiated during the 500 msec following target movement. Computer-generated plots of instantaneous spike frequency (accurate to the nearest 100/~sec) and horizontal and vertical eye position were used to determine the temporal relationship between spike activity and changes in eye position. RESULTS The activity of 117 superior colliculus neurons was recorded on magnetic tape. Subsequent analysis indicated that 86 of these were both clearly isolated from background noise and had firing patterns related to eye movements. Electrolytic lesions made through the recording electrodes at the site of movement-related neurons were located in the intermediate and deeper layers of the superior colliculus. Movement-related neurons were sorted into two major categories: cells with both visual and movement fields (56 ~o) and cells with only movement fields (44 ~o). Cells which discharged in response to visual stimuli and prior to eye movement had overlapping, but not coextensive, movement and visual receptive fields 17,25. The movement-related component of the discharge of these cells was similar in all respects to the response of neurons with only a movement field. Thus, characteristics of the movementrelated discharge described below refer to both categories of movement-related neurons. Detailed maps of the movement fields of 23 neurons in the intermediate and deeper layers of the superior colliculus were obtained. The angle and amplitude associated with the most vigorous discharge was determined for 47 additional units but the movement fields were plotted in less detail. We were unable to obtain detailed maps of the movement fields of 16 units which discharged vigorously to large amplitude saccades since animals rarely made saccades greater than 20° during visual tracking. If the target was displaced 25 °, the target was acquired with two saccades, one of approximately 20 ° and a second of approximately 5°. In general, our results confirm the findings of previous investigators 16-18,2°, 24,~5. Unit discharge occurred prior to saccadic movements, was specific to the direction and amplitude of contralateral saccades, and was independent of the initial position of the eye in the orbit. These neurons discharged prior to movements in total darkness, indicating that the movement-related activity was not the result of retinal image displacements. A gradient of response magnitude was observed across the movement field. The response of a neuron located in the anterior tip of the superior colliculus is illustrated in Fig. 1. Fig. 1B plots burst-index (see methods) as a function of angle of eye movement. The data for 4 different movement amplitudes (0.5, 1, 3, and 5°) are shown. A
25
A
~
B
S'rAt4DAN)"n~,ACKIN6TASK
C
•
.
13, B50.
O=1/2 = =,l °
&
.0"
225
• ~
315"
02 4 0
,"
280
3eO
360
ANGLE (OI[G,qlrES)
270"
D
20
270*
ANGLE
240*
320 °
20*
i
I
I
I/2"
A
M P L I T U D
E
I°
LUI. u
3°
t
,
l
.
t
I
I=~
Fig. 1. A: the visual tracking task. If subject fixated the center dot for 2 sec, the target was moved to a second position. If the target was acquired within 500 msec and maintained for 2 sec, a reinforcement was given. B: burst index (see text) as a function of angle and amplitude of movement. Each point represents the median value of 3 observations. C: three-dimensional representation of burst index as a function of angle and amplitude of eye movement. The maximal burst index value is 48. D: response of a superior colliculus unit to a series of saccades at different angles and amplitudes. The arrow beneath each tracing represents the onset of the target movement.
3-dimensional plot of the same data is shown in Fig. 1C. The maximal discharge of this neuron occurred prior to small right saccades with a downward component (1 ° in amplitude at an angle of 320°). Movements within the movement field but less than or greater than 1° in amplitude were preceded by a less vigorous response. Similarly, if the angle of movement deviated from 320 °, less vigorous responses were observed. Spike activity recorded during movements of different amplitudes (0.5, 1, 3, and 5°)
26 and at different angles (240, 320 and 20'-) are shown in Fig. 1D. It should be noted that although the maximal discharge of this neuron occurred prior to small right saccades, less vigorous responses also accompanied small left saccades (at angles of 240 and 260°). Although we did not isolate superior colliculus neurons which displayed maximal discharge during vertical saccades (90 or 270°), the movement fields of neurons with optimal movements near 90 or 270 ° included both left and right saccades. The movement fields of many superior colliculus neurons were not symmetrical but elongated in the direction of larger amplitude movements. For a fixed movement amplitude, deviations from the optimal angle resulted in a symmetrical reduction of response magnitude (Fig. 2). However, for saccades in the optimal direction, the reduction in response magnitude which accompanied deviations from the optimal amplitude were asymmetrical. Response magnitude decayed sharply as amplitudes became smaller than optimal whereas, a more gradual decline in the neural response was observed as amplitudes became larger than optimal (Fig. 2). Thus, the shape of the movement fields of neurons in the intermediate and deeper layers of the superior colliculus, plotted in visual field coordinates, is similar to the ellipsoidal visual receptive fields of neurons in more superficial layersS, 10. As previously reported 17, the size of the movement fields increases as a function of the amplitude of the optimal movement. Fig. 3 plots the response of 3 neurons which displayed maximal discharge prior to eye movements of 3, 6 and 10° in amplitude. The optimal direction of movement for the discharge of these neurons was 50, 35 and 140 °, respectively. The unit illustrated in Fig. 3C discharged prior to or during
50' R:$ °
9-18"74
5O
MS" U8
B 40.
e:35o o
8 40
U
U R
A
SR30
~ 3o
~EO D E × 10
~ ~o
T
~ tO
0
,
300
5O R=3 °
3EO
,
,
,
,
. . . . .
340 360 EO T H E T A (DES.)
•
40 AMPLITUDE (DE@,)
2- 2.7-T5
50
_
e=350 °
B 40
U R
~ ,o
S 30 T
~ 2o
~
~ eo
-
X tO"
X 100
oJ 3
3OO
THETA (DE@.)
5
B
I0
AMPUTUCIE (DES.)
Fig. 2. Burst index (for movements at the optimal amplitude) as a function of direction of movement and burst index (for movements in the optimal direction) as a function of the amplitude of movement for two neurons, A and B.
270 °
270" 90"
270 ° Fig. 3. Burst index as a function of direction and distance of saccadic movements for 3 neurons. Plots were smoothed by an interpolation routine in the graphics program. The visual field has been rotated 65 ° along the horizontal axis. Maximum values for burst index were 15, 48 and 52 for graphs A, B and C, respectively.
28 movements ranging in direction from 85 to 190° and in amplitude from less than 5 to 20 °. Response magnitude was reduced by 5 0 ~ at angles of 120 and 170° and was reduced by 80 ~o or greater at amplitudes of 2 or 20 °. Much sharper tuning of movement fields was seen in the anterior superior colliculus. Fig. 3A illustrates the response of a unit which gave a vigorous discharge prior to up and right movements of 3 amplitude but did not reliably discharge prior to 0.5 or 5° movements in the same direction. All the movement-related units we have encountered in the intermediate and deeper layers of the superior colliculus show response tuning similar to the neurons illustrated in Fig. 3. We examined, in greater detail, the temporal characteristics of the movementrelated discharge of 22 neurons. Examination of neuronal activity associated with optimal movements revealed two types of movement-related discharge. The response of class I neurons (N =: 7) was characterized by a relatively discrete burst beginning approximately 20 msec (range of 18-32 msec; median of 20 msec) prior to saccade onset (Fig. 4A). The response of class II neurons (N =: 15) was characterized by a gradual increase in spike frequency beginning approximately 80-100 msec before saccade onset which reached a maximum approximately 10-20 msec before the saccade (Fig. 4B). A discrete burst of spike activity was never observed in these neurons. The movement fields of class II neurons are characterized by a temporal gradient as well. The interval between the onset of the neural discharge associated with a movement and the onset of the saccade is greater for movements near the center of the
R = I0" 0=40" H ----------..---- -
~ i J ,-
H
v
:
v
1200
R=5* 0=345 *
1200
2-17-75
J
2-28-75 M4-U4
F I000 R E 800 Q U 6O0 E N 400 C Y 200
F I000 R E 800 O U 600 E N 400 C Y 200 0
0 SO
40 0 TIME (msec.)
A
40
80
40 0 T I ME (mse¢.)
40
B
Fig. 4. Discharge patterns recorded from two movement-related neurons in the superior colliculus. H: horizontal eye position. V: vertical eye position. Middle tracing: spike discharge. Bottom graph: instantaneous spike frequency as a function of time. The dotted line represents the onset of the eye movement.
29 3-7-75 M4-U7
I10"
. . . . . . . . . . . . .
. . . . . . . . . . .
j,7.
. . . . . . . . .
,, i
120" 1
150" .
A N G L E
.
.
t
.
140"
:,
; il i ~ J H I t i i l l l l l l
i I
1
150" 160" i
170 °
b~
L.
.
.
.
.
.
.
.
.
............
"
'-
71'-
r
*
1
180" I
50
msec.
I
Fig. 5. Response of a superior colliculus neuron to a series of saccades 16° in amplitude but varying in direction. The arrow and dotted line represent the onset of the saccade.
field than for movements to the periphery of the field. Fig. 5 illustrates the response of a superior colliculus unit to a series of saccades at the optimal amplitude (16 °) but varying in direction. At the best angle (150°), the onset of the discharge occurs approximately 125 msec prior to the onset of the eye movement. In contrast, discharge onset occurs approximately 25 msec prior to movement at an angle of 180 °. A similar temporal gradient was not apparent in the discharge of class I movement-related neurons. We observed systematic changes in the size and location of movement fields as a function of recording site in the superior colliculus. As successive neurons were encountered along a single electrode penetration, two characteristics of the eye movements producing maximal discharge rates showed a systematic change. There was an increase in the downward component of the optimal direction of movement accompanied by an increase in the amplitude of the optimal eye movements. Also, the amplitude of the eye movements producing maximal neuronal responses increased with more caudal electrode placements. Our electrode penetrations were not normal to the surface of the superior colliculus, but rather, passed through the superior colliculus at an angle of approximately 45 ° . Hence, the electrodes moved caudally in the superior colliculus as depth increased. When corrections for electrode orientation were made, our data were consistent with the stimulation data of Robinson 14 and provide further evidence of a topographical mapping of eye movement neurons in the intermediate and
30 deeper layers of the superior colliculus. Neurons discharging prior to small saccades are located in the anterior superior colliculus; neurons firing before large saccades are found caudally. Movement fields of neurons near the midline have up components and the movement fields of neurons located in the lateral superior colliculus have down components. When different cells were compared, burst duration was not correlated with optimal saccade amplitude. Burst duration of neurons firing maximally to 3° saccades was as great as (or greater than) burst duration of neurons firing maximally to 10° movements. Nor was cell discharge specifically correlated with either the vertical or horizontal components of associated saccades. This is in contrast to what is observed in the burst neurons of the pontine reticular formationS,9, 23. Thus, anatomical location was the only property of superior colliculus neurons which was related to saccade amplitude and direction. This is consistent with the view 15 that saccade direction and amplitude are spatially coded in the superior colliculus and signals appropriate for each extraocular muscle are extracted at subsequent synaptic junctions. DISCUSSION We observed two categories of movement-related neurons in the superior colliculus. Class I neurons are characterized by a discrete spike burst which is tightly coupled to saccade onset. Class II neurons exhibit a gradual increase in spike frequency which reaches a maximum 10-20 msec before saccade onset. A discrete burst of activity was never observed in class II neurons. We suggest, tentatively, that the axons of class I neurons (generating a discrete spike burst) comprise the major efferent pathway from the superior colliculus to subsequent oculomotor premotor neurons. This suggestion is supported by two observations. (1) Robinson 14 observed that the average latency of a saccade following stimulation of the deep layers was approximately 20 msec. The onset of the spike burst of class I neurons preceeds the onset of a saccade by a comparable time. (2) The pattern of spike activity of these neurons resembles the pattern of activity recorded from the long-lead burst units of the pontine reticular formationS, 9. It should be noted that the collicular burst precedes the burst of pontine reticular formation neurons and motor neurons by only 11 and 14 msec, respectively. If the burst observed in the superior colliculus neurons is involved in initiation of saccades, the time available for extraction of signals appropriate for innervation of the extraocular muscles may be less than previously suspected.
Hypotheses concerning the role of the superior eolliculus in eye movement Schiller and Koerner 17 contend that the superior colliculus is involved in the mechanism of foveation. Wurtz and Goldberg 25 argue against this view. They suggest that the colliculus is concerned with shifting attention to one area of the visual field and facilitating movement to that area but that it does not play a critical role in the guidance of eye movements. Available evidence, in our opinion, supports the foveation hypothesis. (1)
31 Electrical stimulation of discrete regions of the superior colliculus elicits saccades with specific amplitudes and directions. The variability in elicited movements is small TM. (2) A burst of spike activity occurring in one class of superior colliculus neurons is tightly coupled to saccade onset. (3) The interval between burst onset and saccade onset is comparable to the latency of saccadic movements following stimulation of the deeper layers of the superior colliculus. (4) In monkeys, we varied the probability of saccade occurrence by varying target duration. We found that the probability of a spike burst occurring in class I superior colliculus neurons was perfectly correlated with the occurrence of a saccade. In these neurons, a spike burst always occurred if a saccade was produced, but never occurred when the animal was able to cancel the saccade (Sparks, unpublished observations). One argument against the foveation hypothesis is that the receptive and movement fields of superior colliculus neurons are too large to direct precise movements. In an attempt to assess the validity of this argument, we measured the response gradient across the movement fields of superior colliculus neurons. We found that the gradient is not sharply tuned to specific movements, particularly for neurons discharging prior to saccades larger than 5° in amplitude. The finding that a single movement-related neuron discharges prior to a wide range of saccades also indicates that a large region of the superior colliculus will be active prior to a specific saccade. (We were unable to determine the exact distribution of this activity). We conclude that precise information about saccade direction and amplitude (if present in the colliculus) must be encoded in the collective discharge of a large number of cells since the discharge of any one movement cell is ambiguous as to the exact saccade direction and amplitude. Does this mean that the superior colliculus does not participate in the initiation and elaboration of precise saccades? Electrophysiological evidence 1~ suggests that at least two, and usually more, synapses intervene between superior colliculus and motor neurons. Thus, extraction of more precise signals from the simultaneous activity of a large number of superior colliculus neurons may occur at subseqent synapses. One possible mechanism by which precise saccades could be initiated by superior colliculus neurons with large movement fields is presented below. This scheme is similar to the one proposed by Mcllwain 10. Assume that activity of a single movement-related cell in the superior colliculus, through subsequent synaptic connections, produces a fixed ratio or pattern of excitation in motoneurons innervating the extracoular muscles. Activity of this neuron would produce a fixed 'movement tendency' which can be represented conceptually by a vector. Because of the large receptive and movement fields, cellular activity may be elicited by stimuli in several regions of the visual field and this cell will be active during a range of saccades. However, the movement tendency produced by activity of this cell remains constant and will always be represented by the same vector. Now, assume that the region of movement-related neurons in the intermediate and deeper layers of the superior colliculus which discharge prior to a saccade to acquire a small visual target occupies a symmetrical, approximately circular area as shown in Fig. 6. Also, referring to Fig. 6, assume that the excitation of a small subset of neurons at point A in the center of the activated region produces a movement
32
ACTIVE AREA 2*-
\ X X,
,, ",
''
-60"
-20"
,i.
z"
I
~,
B ~.-...~R=4*,O=IO* A
C ~
>
_
~
R=5*,e= O*
B÷C t
=
A (R=S*,@=O*)
~>...
R=7O,e=.iO.
Fig. 6. Scheme by which a large population of poorly tuned superior colliculus neurons produces accurate saccadic movements. Top: dorsal view of left superior cotliculus showing hypothetical region of active neurons in the intermediate and deeper layers. The active region is plotted on Robinson's motor map TM. Bottom: hypothetical summation of movement tendencies produced by the simultaneous activation of a large region of the superior colliculus. See text for further details. tendency which may be described as a vector with R = 5°, 0 = 0 °. Similarly, suppose that the activity of neurons located near point B produces a tendency to move the eye along a vector with R ~-- 4 °, 0 = 10°. The spike activity occurring at point B, when a movement of R ~ 5°, 0 = 0 ° has been programmed, will influence the movement to be made and could result in an inaccurate movement. Note, however, that neurons at point C, an equal distance from point A in neural space, will also be active and will produce a movement tendency with a vector of R = 7°, 0 ~- - - 1 0 °. Robinson 14 has shown that when two points in the superior colliculus are stimulated simultaneously, the resulting movement will be to a point on the straight line joining the tips of the vectors representing the saccades evoked by stimulation of each point alone. Thus, the sum of the movement tendencies induced by neural activity at points B and C will be a movement tendency with a vector of R = 5°, 0 = 0 °, the same movement tendency produced by neurons at point A in the center of the active region. Thus, for each subset of active neurons which produces a movement tendency along a vector (V2) other than the desired or programmed vector (V1), there will be a second subset of active neurons producing a movement tendency along a vector (V3) such that V2 -F V3 = V1. (This is not, of course, vector summation in the mathematical sense). According to this argument, saccade direction and amplitude are determined by the spatial location of the population of neurons active in the intermediate and deeper layers of the superior colliculus. The precision or accuracy of a saccade results from the summation of the movement tendencies produced by the population of neurons
33 rather than the discharge of a small number of finely tuned neurons. The contribution of each neuron to the direction and amplitude of the movement is relatively small. Consequently, the effects of variability or 'noise' in the discharge frequency of a particular neuron are reduced by averaging over many neurons. By reducing the effects of 'noise' in the discharge frequency of individual neurons, the large movement fields (which result in large populations of neurons being active during a specific movement) may contribute to, rather than detract from, the accuracy of a saccade. A second argument against the foveation hypothesis is that lesions of the superior colliculus affect the latency but not the accuracy of saccades 26. One reply to this argument is that parallel pathways are involved in the control of visually guided eye movements and that accurate movements can occur in the absence of one, but not all, of the pathways. This is consistent with recent data 11 demonstrating accurate visual tracking after removal of the visual cortex, but not after removal of both visual cortex and superior colliculus. It should be emphasized that the foveation hypothesis and attentional hypothesis are not necessarily incompatible. Although available data support the foveation hypothesis, novel and/or biologically important signals are potent stimuli for eliciting eye movements. Enhancement of the response of visual neurons in the dorsal layers of the superior colliculus 6 may be one mechanism by which selectivity occurs. In summary, we conclude that our results are compatible with the hypothesis that the superior colliculus is involved in coding the location of an object relative to the fovea and in eliciting saccadic movements which produce foveal acquisition of the object. Although movement fields of superior colliculus neurons may be large and coarsely tuned, the direction and amplitude of saccades are coded spatially. Each programmed saccade will be preceded by neuronal activity originating from a population of neurons located in a specific region of the superior colliculus. Accurate movements are produced by the summation of the movement tendencies produced by a population of coarsely tuned neurons, not by the discharge of a small group of finely tuned ones. Important, but unanswered, questions are (1) what are the anatomical destinations of the signals conveyed by the neurons in the intermediate and deeper layers of the superior colliculus and (2) what changes occur in the response properties of neurons at subsequent synaptic junctions? ACKNOWLEDGEMENTS We thank S. Watson for technical asistance during the experiments. This investigation was supported by NIH Grant EY-01189 and by UAB grant 80-9431. REFERENCES 1 Apter, J. T. Projection of the retina on superior colliculus of cats, J. Neurophysiol., 8 (1945) 123-134. 2 Apter, J. T., Eye movements followingstrychninizationof the superior colliculus of cats, J. Neurophysiol., 9 (1946) 73-86.
34 3 Evarts, E. V., A technique for recording activity of subcortical neurons in moving animals, Electroenceph. clin. Neurophysiol., 24 (1968) 83-86. 4 Fuchs, A. F. and Robinson, D. A., A method for measuring horizontal and vertical eye movement chronically in the monkey, J. appl. Physiol., 21 (1966) 1068-1070. 5 Goldberg, M. E. and Wurtz, R. H. Activity of superior colliculus in behaving monkey. I. Visual receptive fields of single neurons. J. NeurophysioL, 35 (1972) 542-559. 6 Goldberg, M. E. and Wurtz, R. H., Activity of superior colliculus in behaving monkey. II. Effect of attention on neuronal responses, J. Neurophysiol., 35 (1972) 560-574. 7 Hess, W. R., Diencephalon. Autonomic and Extrapyramidal Functions, Grune and Stratton, New York, 1954. 8 Keller, E. L., Participation of medial pontine reticular formation in eye movement generation in monkey, J. Neurophysiol., 37 (1974) 316-331. 9 Luschei, E. S. and Fuchs, A. F., Activity of brain stem neurons during eye movements of alert monkeys, J. NeurophysioL, 35 (1972) 445-461. 10 McIlwain, J. T., Visual receptive fields and their images in superior colliculus of the cat, J. Neurophysiol., 38 (1975) 219-230. 11 Mohler, C. W. and Wurtz, R. H., Role of striate cortex and superior colliculus in visual guidance of saccadic eye movements in monkey, Soc. Neurosci. 4th Ann. Meet., 1974, abstract. 12 Precht, W., Schwindt, P. and Magherini, P., Tectal influences on cat ocular motoneurones, Brain Research, 82 (1974) 27-40. 13 Rizzolatti, G., Camarda, R., Grupp, L. A. and Pisa, M., Inhibitory effect of remote visual stimuli on visual responses of cat superior colliculus: spatial and temporal factors, J. Neurophysiol., 37 (1974) 1262-1275. 14 Robinson, D. A., Eye movements evoked by collicular stimulation in the alert monkey, Vision Res., 12 (1972) 1795-1808. 15 Robinson, D. A., Models of the saccadic eye movement control system, Kybernetik, 14 (1973) 71 83. 16 Robinson, D. L. and Jarvis, C. D., Superior colliculus neurons of single units in superior colliculus of the alert rhesus monkey, J. Neurophysiol., 34 (1971) 925-936. 17 Schiller, P. H. and Koerner, F., Discharge characteristics of single units in superior colliculus of the alert rhesus monkey, J. Neurophysiol., 34 (1971) 920-924. 18 Schiller, P. H. and Stryker, M., Single-unit recording and stimulation in superior colliculus of the alert rhesus monkey, J. NeurophysioL, 35 (1972) 915-924. 19 Schiller, P. H., Stryker, M., Cynader, M. and Berman, N., Response characteristics of single cells in the monkey superior colliculus following ablation or cooling of visual cortex. J. NeurophysioL, 37 (1974) 181-194. 20 Sparks, D. L., Response properties of eye movement-related neurons in the monkey superior colliculus, Brain Research, 90 (1975) 147 152. 21 Sparks, D. L., Movement field size and distribution in the monkey superior colliculus, Assoc. Res. Vision OphthahnoL, Sarasota, Fla., 1975. 22 Sparks, D. L. and Holland, R., Computer control of eye position and velocity, Behav. Res. Meth. lnstrum., 7 (1975) 115-119. 23 Sparks, D. L. and Travis, R. P., Jr., Firing patterns of reticular formation neurons during horizontal eye movements, Brain Research, 33 (1971) 477-481. 24 Updyke, B. V., Characteristics of unit responses in superior colliculus of the Cebus monkey, J. NeurophysioL, 37 (1974) 896-909. 25 Wurtz, R. H. and Goldberg, M. E., Activity of superior colliculus in behaving monkey. 11I. Cells discharging before eye movements, J. NeurophysioL, 35 (1972) 575-586. 26 Wurtz, R. H. and Goldberg, M. E., Activity of superior colliculus in behaving monkey. IV. Effects of lesions on eye movements. J. NeurophysioL, 35 (1972) 587-596.
113 (1976) 21-34 0 Elsevier Scientific Publishing Company,
SIZE AND DISTRIBUTION SUPERIOR COLLICULUS
Amsterdam
- Printed in
OF MOVEMENT
The Netherlands
FIELDS
21
IN THE MONKEY
DAVID L. SPARKS, RICHARD HOLLAND and BARTON L. GUTHRIE Department of Psychology and Neurosciences mingham, Ala. 35294 (U.S.A.)
Program,
University of Alabama in Birmingham,
Bir-
(Accepted January 20th, 1976)
SUMMARY
A gradient of response magnitude was observed across the movement fields (the range of eye movements which alter the discharge frequency of a neuron) of neurons in the intermediate and deeper layers of the superior colliculus. A vigorous discharge preceded movements with a particular direction and amplitude but reduced responses preceded movements which deviated from this direction and/or amplitude. Movement field size is a function of the amplitude of the optimal movement. Neurons discharging prior to small saccades have small and sharply tuned fields. Neurons discharging prior to large saccades have large movement fields and tuning is relatively coarse. Movement fields are topographically organized within the superior colliculus. Neurons discharging prior to small saccades are located anteriorly; neurons firing before large saccades are found caudally. Neurons near the midline discharge prior to up movements and neurons located laterally fire before downward movements. Movement fields of superior colliculus neurons are also characterized by a temporal gradient. The interval between spike discharge and the onset of a saccade is greater for movements near the center of the movement field than for movements to the periphery of the field. Results are interpreted as supporting the foveation hypothesis of superior colliculus function. It is suggested that precise saccadic movements are not produced by the discharge of a small population of finely tuned neurons but result from the weighted sum of the simultaneous movement tendencies produced by the activity of a large population of less finely tuned neurons.
INTRODUCTION Earlier suggestionslpsg7 that the superior colliculus plays an important role in the control of eye movements are supported by two lines of recent evidence. Firstly,
22 electrical stimulation of the superior colliculus in alert monkeys elicits short-latency saccadic eye movementsr3v1a. The amplitude and direction of these movements are a function of the site of stimulation within the superior colliculus rather than parameters of the stimulus train. Saccades have up components if elicited by stimulation of medial regions of the superior colliculus and have down components if elicited by stimulation of lateral regions. Elicited movements are small rostrally and large caudally. Secondly, chronic microelectrode recordings have shownJ+l*J0sa1+2” that the maximal discharge of neurons in the intermediate and deeper layers of the superior colliculus occurs prior to eye movements with a particular direction and amplitude. These recent observations were the basis of the proposal17 that the superior colliculus is part of a system which acquires visual targets for fovea1 viewing. According to this ‘foveation’ hypothesis, the superior colliculus is involved in coding the location of an object relative to the fovea and in eliciting saccadic movements which produce fovea1 acquisition of the object. One of the criticismsa” of the foveation hypothesis is that the movement fields (the range of eye movements which alter the discharge frequency of a neuron) of superior colliculus neurons are relatively large. If, however, the response of superior colliculus neurons is sharply tuned to particular movements, this argument may not be critical. The major purpose of this experiment was to describe, in more detail, the movement field properties of neurons in the deeper layers of the superior colliculus. Secondary purposes were: (1) to explore, further, the topographical organization of the movement fields and (2) to examine the temporal relationship between the spike discharge and the onset of associated saccades. Preliminary reports of these findings have been presented elsewhere20J1. METHODS
Five rhesus monkeys (Macaca mulatta) weighing from 2.5 to 4 kg were used. Each animal underwent 3 aseptic surgical procedures. First, 4 stainless steel screws were implanted in the skull to permit immobilization of the animal’s heads during training and recording sessions. Approximately one month later, a coil of fine wire was implanted beneath the bulbar conjunctiva and recti insertions. The ends of the coil were led subcutaneously to a connector mounted on the skull. After 4-6 weeks of behavioral training, a stainless steel receptacle for a microdrive was secured to the skull. During training and recording sessions, monkeys were seated in a primate chair located in an electrostatically shielded and sound-resistant room. Horizontal and vertical eye position signals were obtained (with a sensitivity of 0.25”) using the magnetic coil technique of Fuchs and Robinson4. Extracellular unit potentials were recorded with tungsten microelectrodes insulated with teflon and glass. Electrodes were advanced through a 16-gauge stainless steel cannula, extending 2-3 mm below the dura, with a hydraulic microdrive (Kopf). Recordings were filtered for frequencies above 5 kHz to reduce contamination with the 18 kHz signal of the alternating magnetic fields. Spikes were displayed on an oscilloscope, fed into a discriminator trigger unit, and transcribed on tape (band pass of 40 Hz-19 kHz). Other
23 FM tape channels (band pass D C to 1250 Hz) were used to record trial, stimulus and response events, vertical and horizontal eye position, and vertical and horizontal target position. Electrolytic lesions were made at the bottom of electrode paths in which responsive units were encountered. Fifty micron sections through the brain stem were mounted and stained with thionin. Recording sites were determined from histological sections and measurements of electrode positions taken at the time of the recordings.
Behavioral training Monkeys were trained to track a visual target, subtending a visual angle of less than 0.1 °, presented on a Hewlett-Packard 1310A oscilloscope (CRT viewing area of 27 by 37 cm). Target position was controlled by a PDP-8/I computer. Appropriate programs z2 permitted the target to be displayed at any point on a 4096 by 4096 matrix on the screen or to be moved at specified rates from one position to another. On training and recording days, animals were water-deprived for 23 h. Reinforcement was 0.1 ml of water delivered through a tube near the animal's mouth. During initial phases of training, the animal's eye movements were observed by the experimenter. Intermittently, the target was presented at the center of the oscilloscope screen. If the animal appeared to 'look at' the target, a reinforcement was given. When the animal reliably oriented the onset of the target with short latencies, the required fixation time was increased, gradually, to 2 sec. Next, the target was presented at the center of the screen and after a variable fixation time (1-3 sec) was moved instantaneously to a second position. If the animal followed the target with an eye movement o f short latency and fixated it for 2 sec, a reinforcement was delivered. These tracking movements were used to calibrate the eye movement recording system. During subsequent sessions, target presentations and reinforcements were under computer control. Horizontal and vertical eye position signals were sampled each msec and compared with target position.
Data collection and analysis During data collection trials, the animal was permitted 500 msec to acquire the initial target presented at the center of the screen. After a variable fixation interval (1-3 sec), the target was moved to a second position. If this target was acquired within 500 msec and fixation maintained for two additional seconds, a reinforcement was given. If not, the trial was terminated, and after a delay, a new trial was initiated. The origin of the target and the angle and radius of target movements were controlled by the experimenter via on-line interaction with the computer. A computer-generated display of horizontal and vertical eye position and spike activity was presented at the end of each trial. The movement fields of units were plotted by requiring animals to track target movements which varied systematically in radius and angle. These movements were always initiated from a 'straight ahead' position. An attempt was made to record unit activity during at least 3 movements to each sampled position. To ensure that movement-related discharges were not the result of retinal image displacements, the
24 activity of each unit was also observed during spontaneous eye movements occurring in darkness. The origin of the target movement was varied occasionally to determine whether or not saccade-associated activity was independent of initial eye position. For each reinforced tracking trial, the difference between the number of spikes occurring during a 500 msec period of center target fixation and the number of spikes occurring in the 500 msec following target movement was computed. This difference is referred to as the burst-index and provides an indication of the strength of neural discharge associated with saccades initiated during the 500 msec following target movement. Computer-generated plots of instantaneous spike frequency (accurate to the nearest 100/~sec) and horizontal and vertical eye position were used to determine the temporal relationship between spike activity and changes in eye position. RESULTS The activity of 117 superior colliculus neurons was recorded on magnetic tape. Subsequent analysis indicated that 86 of these were both clearly isolated from background noise and had firing patterns related to eye movements. Electrolytic lesions made through the recording electrodes at the site of movement-related neurons were located in the intermediate and deeper layers of the superior colliculus. Movement-related neurons were sorted into two major categories: cells with both visual and movement fields (56 ~o) and cells with only movement fields (44 ~o). Cells which discharged in response to visual stimuli and prior to eye movement had overlapping, but not coextensive, movement and visual receptive fields 17,25. The movement-related component of the discharge of these cells was similar in all respects to the response of neurons with only a movement field. Thus, characteristics of the movementrelated discharge described below refer to both categories of movement-related neurons. Detailed maps of the movement fields of 23 neurons in the intermediate and deeper layers of the superior colliculus were obtained. The angle and amplitude associated with the most vigorous discharge was determined for 47 additional units but the movement fields were plotted in less detail. We were unable to obtain detailed maps of the movement fields of 16 units which discharged vigorously to large amplitude saccades since animals rarely made saccades greater than 20° during visual tracking. If the target was displaced 25 °, the target was acquired with two saccades, one of approximately 20 ° and a second of approximately 5°. In general, our results confirm the findings of previous investigators 16-18,2°, 24,~5. Unit discharge occurred prior to saccadic movements, was specific to the direction and amplitude of contralateral saccades, and was independent of the initial position of the eye in the orbit. These neurons discharged prior to movements in total darkness, indicating that the movement-related activity was not the result of retinal image displacements. A gradient of response magnitude was observed across the movement field. The response of a neuron located in the anterior tip of the superior colliculus is illustrated in Fig. 1. Fig. 1B plots burst-index (see methods) as a function of angle of eye movement. The data for 4 different movement amplitudes (0.5, 1, 3, and 5°) are shown. A
25
A
~
B
S'rAt4DAN)"n~,ACKIN6TASK
C
•
.
13, B50.
O=1/2 = =,l °
&
.0"
225
• ~
315"
02 4 0
,"
280
3eO
360
ANGLE (OI[G,qlrES)
270"
D
20
270*
ANGLE
240*
320 °
20*
i
I
I
I/2"
A
M P L I T U D
E
I°
LUI. u
3°
t
,
l
.
t
I
I=~
Fig. 1. A: the visual tracking task. If subject fixated the center dot for 2 sec, the target was moved to a second position. If the target was acquired within 500 msec and maintained for 2 sec, a reinforcement was given. B: burst index (see text) as a function of angle and amplitude of movement. Each point represents the median value of 3 observations. C: three-dimensional representation of burst index as a function of angle and amplitude of eye movement. The maximal burst index value is 48. D: response of a superior colliculus unit to a series of saccades at different angles and amplitudes. The arrow beneath each tracing represents the onset of the target movement.
3-dimensional plot of the same data is shown in Fig. 1C. The maximal discharge of this neuron occurred prior to small right saccades with a downward component (1 ° in amplitude at an angle of 320°). Movements within the movement field but less than or greater than 1° in amplitude were preceded by a less vigorous response. Similarly, if the angle of movement deviated from 320 °, less vigorous responses were observed. Spike activity recorded during movements of different amplitudes (0.5, 1, 3, and 5°)
26 and at different angles (240, 320 and 20'-) are shown in Fig. 1D. It should be noted that although the maximal discharge of this neuron occurred prior to small right saccades, less vigorous responses also accompanied small left saccades (at angles of 240 and 260°). Although we did not isolate superior colliculus neurons which displayed maximal discharge during vertical saccades (90 or 270°), the movement fields of neurons with optimal movements near 90 or 270 ° included both left and right saccades. The movement fields of many superior colliculus neurons were not symmetrical but elongated in the direction of larger amplitude movements. For a fixed movement amplitude, deviations from the optimal angle resulted in a symmetrical reduction of response magnitude (Fig. 2). However, for saccades in the optimal direction, the reduction in response magnitude which accompanied deviations from the optimal amplitude were asymmetrical. Response magnitude decayed sharply as amplitudes became smaller than optimal whereas, a more gradual decline in the neural response was observed as amplitudes became larger than optimal (Fig. 2). Thus, the shape of the movement fields of neurons in the intermediate and deeper layers of the superior colliculus, plotted in visual field coordinates, is similar to the ellipsoidal visual receptive fields of neurons in more superficial layersS, 10. As previously reported 17, the size of the movement fields increases as a function of the amplitude of the optimal movement. Fig. 3 plots the response of 3 neurons which displayed maximal discharge prior to eye movements of 3, 6 and 10° in amplitude. The optimal direction of movement for the discharge of these neurons was 50, 35 and 140 °, respectively. The unit illustrated in Fig. 3C discharged prior to or during
50' R:$ °
9-18"74
5O
MS" U8
B 40.
e:35o o
8 40
U
U R
A
SR30
~ 3o
~EO D E × 10
~ ~o
T
~ tO
0
,
300
5O R=3 °
3EO
,
,
,
,
. . . . .
340 360 EO T H E T A (DES.)
•
40 AMPLITUDE (DE@,)
2- 2.7-T5
50
_
e=350 °
B 40
U R
~ ,o
S 30 T
~ 2o
~
~ eo
-
X tO"
X 100
oJ 3
3OO
THETA (DE@.)
5
B
I0
AMPUTUCIE (DES.)
Fig. 2. Burst index (for movements at the optimal amplitude) as a function of direction of movement and burst index (for movements in the optimal direction) as a function of the amplitude of movement for two neurons, A and B.
270 °
270" 90"
270 ° Fig. 3. Burst index as a function of direction and distance of saccadic movements for 3 neurons. Plots were smoothed by an interpolation routine in the graphics program. The visual field has been rotated 65 ° along the horizontal axis. Maximum values for burst index were 15, 48 and 52 for graphs A, B and C, respectively.
28 movements ranging in direction from 85 to 190° and in amplitude from less than 5 to 20 °. Response magnitude was reduced by 5 0 ~ at angles of 120 and 170° and was reduced by 80 ~o or greater at amplitudes of 2 or 20 °. Much sharper tuning of movement fields was seen in the anterior superior colliculus. Fig. 3A illustrates the response of a unit which gave a vigorous discharge prior to up and right movements of 3 amplitude but did not reliably discharge prior to 0.5 or 5° movements in the same direction. All the movement-related units we have encountered in the intermediate and deeper layers of the superior colliculus show response tuning similar to the neurons illustrated in Fig. 3. We examined, in greater detail, the temporal characteristics of the movementrelated discharge of 22 neurons. Examination of neuronal activity associated with optimal movements revealed two types of movement-related discharge. The response of class I neurons (N =: 7) was characterized by a relatively discrete burst beginning approximately 20 msec (range of 18-32 msec; median of 20 msec) prior to saccade onset (Fig. 4A). The response of class II neurons (N =: 15) was characterized by a gradual increase in spike frequency beginning approximately 80-100 msec before saccade onset which reached a maximum approximately 10-20 msec before the saccade (Fig. 4B). A discrete burst of spike activity was never observed in these neurons. The movement fields of class II neurons are characterized by a temporal gradient as well. The interval between the onset of the neural discharge associated with a movement and the onset of the saccade is greater for movements near the center of the
R = I0" 0=40" H ----------..---- -
~ i J ,-
H
v
:
v
1200
R=5* 0=345 *
1200
2-17-75
J
2-28-75 M4-U4
F I000 R E 800 Q U 6O0 E N 400 C Y 200
F I000 R E 800 O U 600 E N 400 C Y 200 0
0 SO
40 0 TIME (msec.)
A
40
80
40 0 T I ME (mse¢.)
40
B
Fig. 4. Discharge patterns recorded from two movement-related neurons in the superior colliculus. H: horizontal eye position. V: vertical eye position. Middle tracing: spike discharge. Bottom graph: instantaneous spike frequency as a function of time. The dotted line represents the onset of the eye movement.
29 3-7-75 M4-U7
I10"
. . . . . . . . . . . . .
. . . . . . . . . . .
j,7.
. . . . . . . . .
,, i
120" 1
150" .
A N G L E
.
.
t
.
140"
:,
; il i ~ J H I t i i l l l l l l
i I
1
150" 160" i
170 °
b~
L.
.
.
.
.
.
.
.
.
............
"
'-
71'-
r
*
1
180" I
50
msec.
I
Fig. 5. Response of a superior colliculus neuron to a series of saccades 16° in amplitude but varying in direction. The arrow and dotted line represent the onset of the saccade.
field than for movements to the periphery of the field. Fig. 5 illustrates the response of a superior colliculus unit to a series of saccades at the optimal amplitude (16 °) but varying in direction. At the best angle (150°), the onset of the discharge occurs approximately 125 msec prior to the onset of the eye movement. In contrast, discharge onset occurs approximately 25 msec prior to movement at an angle of 180 °. A similar temporal gradient was not apparent in the discharge of class I movement-related neurons. We observed systematic changes in the size and location of movement fields as a function of recording site in the superior colliculus. As successive neurons were encountered along a single electrode penetration, two characteristics of the eye movements producing maximal discharge rates showed a systematic change. There was an increase in the downward component of the optimal direction of movement accompanied by an increase in the amplitude of the optimal eye movements. Also, the amplitude of the eye movements producing maximal neuronal responses increased with more caudal electrode placements. Our electrode penetrations were not normal to the surface of the superior colliculus, but rather, passed through the superior colliculus at an angle of approximately 45 ° . Hence, the electrodes moved caudally in the superior colliculus as depth increased. When corrections for electrode orientation were made, our data were consistent with the stimulation data of Robinson 14 and provide further evidence of a topographical mapping of eye movement neurons in the intermediate and
30 deeper layers of the superior colliculus. Neurons discharging prior to small saccades are located in the anterior superior colliculus; neurons firing before large saccades are found caudally. Movement fields of neurons near the midline have up components and the movement fields of neurons located in the lateral superior colliculus have down components. When different cells were compared, burst duration was not correlated with optimal saccade amplitude. Burst duration of neurons firing maximally to 3° saccades was as great as (or greater than) burst duration of neurons firing maximally to 10° movements. Nor was cell discharge specifically correlated with either the vertical or horizontal components of associated saccades. This is in contrast to what is observed in the burst neurons of the pontine reticular formationS,9, 23. Thus, anatomical location was the only property of superior colliculus neurons which was related to saccade amplitude and direction. This is consistent with the view 15 that saccade direction and amplitude are spatially coded in the superior colliculus and signals appropriate for each extraocular muscle are extracted at subsequent synaptic junctions. DISCUSSION We observed two categories of movement-related neurons in the superior colliculus. Class I neurons are characterized by a discrete spike burst which is tightly coupled to saccade onset. Class II neurons exhibit a gradual increase in spike frequency which reaches a maximum 10-20 msec before saccade onset. A discrete burst of activity was never observed in class II neurons. We suggest, tentatively, that the axons of class I neurons (generating a discrete spike burst) comprise the major efferent pathway from the superior colliculus to subsequent oculomotor premotor neurons. This suggestion is supported by two observations. (1) Robinson 14 observed that the average latency of a saccade following stimulation of the deep layers was approximately 20 msec. The onset of the spike burst of class I neurons preceeds the onset of a saccade by a comparable time. (2) The pattern of spike activity of these neurons resembles the pattern of activity recorded from the long-lead burst units of the pontine reticular formationS, 9. It should be noted that the collicular burst precedes the burst of pontine reticular formation neurons and motor neurons by only 11 and 14 msec, respectively. If the burst observed in the superior colliculus neurons is involved in initiation of saccades, the time available for extraction of signals appropriate for innervation of the extraocular muscles may be less than previously suspected.
Hypotheses concerning the role of the superior eolliculus in eye movement Schiller and Koerner 17 contend that the superior colliculus is involved in the mechanism of foveation. Wurtz and Goldberg 25 argue against this view. They suggest that the colliculus is concerned with shifting attention to one area of the visual field and facilitating movement to that area but that it does not play a critical role in the guidance of eye movements. Available evidence, in our opinion, supports the foveation hypothesis. (1)
31 Electrical stimulation of discrete regions of the superior colliculus elicits saccades with specific amplitudes and directions. The variability in elicited movements is small TM. (2) A burst of spike activity occurring in one class of superior colliculus neurons is tightly coupled to saccade onset. (3) The interval between burst onset and saccade onset is comparable to the latency of saccadic movements following stimulation of the deeper layers of the superior colliculus. (4) In monkeys, we varied the probability of saccade occurrence by varying target duration. We found that the probability of a spike burst occurring in class I superior colliculus neurons was perfectly correlated with the occurrence of a saccade. In these neurons, a spike burst always occurred if a saccade was produced, but never occurred when the animal was able to cancel the saccade (Sparks, unpublished observations). One argument against the foveation hypothesis is that the receptive and movement fields of superior colliculus neurons are too large to direct precise movements. In an attempt to assess the validity of this argument, we measured the response gradient across the movement fields of superior colliculus neurons. We found that the gradient is not sharply tuned to specific movements, particularly for neurons discharging prior to saccades larger than 5° in amplitude. The finding that a single movement-related neuron discharges prior to a wide range of saccades also indicates that a large region of the superior colliculus will be active prior to a specific saccade. (We were unable to determine the exact distribution of this activity). We conclude that precise information about saccade direction and amplitude (if present in the colliculus) must be encoded in the collective discharge of a large number of cells since the discharge of any one movement cell is ambiguous as to the exact saccade direction and amplitude. Does this mean that the superior colliculus does not participate in the initiation and elaboration of precise saccades? Electrophysiological evidence 1~ suggests that at least two, and usually more, synapses intervene between superior colliculus and motor neurons. Thus, extraction of more precise signals from the simultaneous activity of a large number of superior colliculus neurons may occur at subseqent synapses. One possible mechanism by which precise saccades could be initiated by superior colliculus neurons with large movement fields is presented below. This scheme is similar to the one proposed by Mcllwain 10. Assume that activity of a single movement-related cell in the superior colliculus, through subsequent synaptic connections, produces a fixed ratio or pattern of excitation in motoneurons innervating the extracoular muscles. Activity of this neuron would produce a fixed 'movement tendency' which can be represented conceptually by a vector. Because of the large receptive and movement fields, cellular activity may be elicited by stimuli in several regions of the visual field and this cell will be active during a range of saccades. However, the movement tendency produced by activity of this cell remains constant and will always be represented by the same vector. Now, assume that the region of movement-related neurons in the intermediate and deeper layers of the superior colliculus which discharge prior to a saccade to acquire a small visual target occupies a symmetrical, approximately circular area as shown in Fig. 6. Also, referring to Fig. 6, assume that the excitation of a small subset of neurons at point A in the center of the activated region produces a movement
32
ACTIVE AREA 2*-
\ X X,
,, ",
''
-60"
-20"
,i.
z"
I
~,
B ~.-...~R=4*,O=IO* A
C ~
>
_
~
R=5*,e= O*
B÷C t
=
A (R=S*,@=O*)
~>...
R=7O,e=.iO.
Fig. 6. Scheme by which a large population of poorly tuned superior colliculus neurons produces accurate saccadic movements. Top: dorsal view of left superior cotliculus showing hypothetical region of active neurons in the intermediate and deeper layers. The active region is plotted on Robinson's motor map TM. Bottom: hypothetical summation of movement tendencies produced by the simultaneous activation of a large region of the superior colliculus. See text for further details. tendency which may be described as a vector with R = 5°, 0 = 0 °. Similarly, suppose that the activity of neurons located near point B produces a tendency to move the eye along a vector with R ~-- 4 °, 0 = 10°. The spike activity occurring at point B, when a movement of R ~ 5°, 0 = 0 ° has been programmed, will influence the movement to be made and could result in an inaccurate movement. Note, however, that neurons at point C, an equal distance from point A in neural space, will also be active and will produce a movement tendency with a vector of R = 7°, 0 ~- - - 1 0 °. Robinson 14 has shown that when two points in the superior colliculus are stimulated simultaneously, the resulting movement will be to a point on the straight line joining the tips of the vectors representing the saccades evoked by stimulation of each point alone. Thus, the sum of the movement tendencies induced by neural activity at points B and C will be a movement tendency with a vector of R = 5°, 0 = 0 °, the same movement tendency produced by neurons at point A in the center of the active region. Thus, for each subset of active neurons which produces a movement tendency along a vector (V2) other than the desired or programmed vector (V1), there will be a second subset of active neurons producing a movement tendency along a vector (V3) such that V2 -F V3 = V1. (This is not, of course, vector summation in the mathematical sense). According to this argument, saccade direction and amplitude are determined by the spatial location of the population of neurons active in the intermediate and deeper layers of the superior colliculus. The precision or accuracy of a saccade results from the summation of the movement tendencies produced by the population of neurons
33 rather than the discharge of a small number of finely tuned neurons. The contribution of each neuron to the direction and amplitude of the movement is relatively small. Consequently, the effects of variability or 'noise' in the discharge frequency of a particular neuron are reduced by averaging over many neurons. By reducing the effects of 'noise' in the discharge frequency of individual neurons, the large movement fields (which result in large populations of neurons being active during a specific movement) may contribute to, rather than detract from, the accuracy of a saccade. A second argument against the foveation hypothesis is that lesions of the superior colliculus affect the latency but not the accuracy of saccades 26. One reply to this argument is that parallel pathways are involved in the control of visually guided eye movements and that accurate movements can occur in the absence of one, but not all, of the pathways. This is consistent with recent data 11 demonstrating accurate visual tracking after removal of the visual cortex, but not after removal of both visual cortex and superior colliculus. It should be emphasized that the foveation hypothesis and attentional hypothesis are not necessarily incompatible. Although available data support the foveation hypothesis, novel and/or biologically important signals are potent stimuli for eliciting eye movements. Enhancement of the response of visual neurons in the dorsal layers of the superior colliculus 6 may be one mechanism by which selectivity occurs. In summary, we conclude that our results are compatible with the hypothesis that the superior colliculus is involved in coding the location of an object relative to the fovea and in eliciting saccadic movements which produce foveal acquisition of the object. Although movement fields of superior colliculus neurons may be large and coarsely tuned, the direction and amplitude of saccades are coded spatially. Each programmed saccade will be preceded by neuronal activity originating from a population of neurons located in a specific region of the superior colliculus. Accurate movements are produced by the summation of the movement tendencies produced by a population of coarsely tuned neurons, not by the discharge of a small group of finely tuned ones. Important, but unanswered, questions are (1) what are the anatomical destinations of the signals conveyed by the neurons in the intermediate and deeper layers of the superior colliculus and (2) what changes occur in the response properties of neurons at subsequent synaptic junctions? ACKNOWLEDGEMENTS We thank S. Watson for technical asistance during the experiments. This investigation was supported by NIH Grant EY-01189 and by UAB grant 80-9431. REFERENCES 1 Apter, J. T. Projection of the retina on superior colliculus of cats, J. Neurophysiol., 8 (1945) 123-134. 2 Apter, J. T., Eye movements followingstrychninizationof the superior colliculus of cats, J. Neurophysiol., 9 (1946) 73-86.
34 3 Evarts, E. V., A technique for recording activity of subcortical neurons in moving animals, Electroenceph. clin. Neurophysiol., 24 (1968) 83-86. 4 Fuchs, A. F. and Robinson, D. A., A method for measuring horizontal and vertical eye movement chronically in the monkey, J. appl. Physiol., 21 (1966) 1068-1070. 5 Goldberg, M. E. and Wurtz, R. H. Activity of superior colliculus in behaving monkey. I. Visual receptive fields of single neurons. J. NeurophysioL, 35 (1972) 542-559. 6 Goldberg, M. E. and Wurtz, R. H., Activity of superior colliculus in behaving monkey. II. Effect of attention on neuronal responses, J. Neurophysiol., 35 (1972) 560-574. 7 Hess, W. R., Diencephalon. Autonomic and Extrapyramidal Functions, Grune and Stratton, New York, 1954. 8 Keller, E. L., Participation of medial pontine reticular formation in eye movement generation in monkey, J. Neurophysiol., 37 (1974) 316-331. 9 Luschei, E. S. and Fuchs, A. F., Activity of brain stem neurons during eye movements of alert monkeys, J. NeurophysioL, 35 (1972) 445-461. 10 McIlwain, J. T., Visual receptive fields and their images in superior colliculus of the cat, J. Neurophysiol., 38 (1975) 219-230. 11 Mohler, C. W. and Wurtz, R. H., Role of striate cortex and superior colliculus in visual guidance of saccadic eye movements in monkey, Soc. Neurosci. 4th Ann. Meet., 1974, abstract. 12 Precht, W., Schwindt, P. and Magherini, P., Tectal influences on cat ocular motoneurones, Brain Research, 82 (1974) 27-40. 13 Rizzolatti, G., Camarda, R., Grupp, L. A. and Pisa, M., Inhibitory effect of remote visual stimuli on visual responses of cat superior colliculus: spatial and temporal factors, J. Neurophysiol., 37 (1974) 1262-1275. 14 Robinson, D. A., Eye movements evoked by collicular stimulation in the alert monkey, Vision Res., 12 (1972) 1795-1808. 15 Robinson, D. A., Models of the saccadic eye movement control system, Kybernetik, 14 (1973) 71 83. 16 Robinson, D. L. and Jarvis, C. D., Superior colliculus neurons of single units in superior colliculus of the alert rhesus monkey, J. Neurophysiol., 34 (1971) 925-936. 17 Schiller, P. H. and Koerner, F., Discharge characteristics of single units in superior colliculus of the alert rhesus monkey, J. Neurophysiol., 34 (1971) 920-924. 18 Schiller, P. H. and Stryker, M., Single-unit recording and stimulation in superior colliculus of the alert rhesus monkey, J. NeurophysioL, 35 (1972) 915-924. 19 Schiller, P. H., Stryker, M., Cynader, M. and Berman, N., Response characteristics of single cells in the monkey superior colliculus following ablation or cooling of visual cortex. J. NeurophysioL, 37 (1974) 181-194. 20 Sparks, D. L., Response properties of eye movement-related neurons in the monkey superior colliculus, Brain Research, 90 (1975) 147 152. 21 Sparks, D. L., Movement field size and distribution in the monkey superior colliculus, Assoc. Res. Vision OphthahnoL, Sarasota, Fla., 1975. 22 Sparks, D. L. and Holland, R., Computer control of eye position and velocity, Behav. Res. Meth. lnstrum., 7 (1975) 115-119. 23 Sparks, D. L. and Travis, R. P., Jr., Firing patterns of reticular formation neurons during horizontal eye movements, Brain Research, 33 (1971) 477-481. 24 Updyke, B. V., Characteristics of unit responses in superior colliculus of the Cebus monkey, J. NeurophysioL, 37 (1974) 896-909. 25 Wurtz, R. H. and Goldberg, M. E., Activity of superior colliculus in behaving monkey. 11I. Cells discharging before eye movements, J. NeurophysioL, 35 (1972) 575-586. 26 Wurtz, R. H. and Goldberg, M. E., Activity of superior colliculus in behaving monkey. IV. Effects of lesions on eye movements. J. NeurophysioL, 35 (1972) 587-596.