Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
The Relationship of Monkey Frontal Eye Field Activity to Saccade Dynamics MARK A. SEGRAVES AND KATHRYN PARK Department ofNeurobiology and Physiology, Northwestern University, Evanston, Illinois 60208-3.520; and Laboratory Sensorimotor Research, National Eye Institute, National Institutes of Health, Bethesda, Maryland 60892 SIJMMARY AND CONCLUSIONS I. In this study, we compared the temporal waveforms of the activity of monkey frontal eye field movement neurons with the dynamics of saccadic eye movements. 2. Movement neurons in the frontal eye field were selected according to previously published criteria. They had little or no response to visual stimuli in a fixation task, and equivalent activity before visually guided and memory-guided saccades. We studied corticotectal neurons and corticopontine neurons identified by antidromic excitation, as well as neurons whose projections were not identified. 3. These neurons had a peak activation at a mean of 13 ms before the saccade began. However, rather than falling off rapidly as the saccade ended, most neurons continued to fire after the saccade. returning to baseline at a mean of 93 ms after the end of the saccade. 4. We measured the decrement in activity for these neurons during the saccade. Although a few neurons showed decrements of ~60% of their peak activity level, the average activity dropped only 16.9%, with some neurons actually showing a rise in activity during the saccade. If we ignored the latency between peak in activity and saccade start and measured the fall in activity for a period equal to one saccade duration after the peak, the average drop in activity was still only 34.9%. Thus the activity of these neurons did not appear to be closely related to dynamic motor error, which falls from its maximum value to zero over the time course of a saccade. 5. These results suggest that a focus of movement activity within the topographic map in the frontal eye field specifies the amplitude and direction for an impending saccade, whereas the peak of movement activity signals the time to initiate a saccade. 6. Unlike the superior colliculus, the activity of frontal eye field movement neurons does not appear to be related to dynamic events that occur during the saccade, such as motor error. INTRODUCTION The frontal eye field has an important role in the control of saccadic eye movements. Microstimulation of this portion of frontal cortex produces saccades with short latency and low threshold (Bruce et al. 1985; Robinson and Fuchs 1969); its movement neurons discharge before purposive saccades, including those made to remembered target locations ( Bruce and Goldberg 1985 ). Frontal eye field movement neurons project to both the superior colliculus and to oculomotor regions of the pons, including the region of omnipause neurons in the paramedian pontine reticular formation ( Segraves 1992; Segraves and Goldberg 1987 ). Although it is likely to participate in the generation of all voluntary saccades, the frontal eye field appears to be essential for the generation of cognitively difficult forms of saccades, including saccades to remembered target locations 1880 of (Deng et al. 1986) and antisaccades-saccades directed away from the spatial location of a visual cue (Guitton et al. 1985). Within cerebral cortex, several regions are involved in the control of saccades besides the frontal eye field, including posterior parietal cortex (Andersen and Gnadt 1989; Barash et al. 199 la,b; Duhamel et al. 1992; Kurylo and Skavenski 199 1 ), the supplementary eye field ( Schall 199 1; Schlag and Schlag-Rey 1987) and the cortex surrounding the principal sulcus (Both and Goldberg 1989; Funahashi et al. 199 1). These regions share many connections and appear to form a cortical network involved in the control of saccades. Recently, it has been reported that lesions of the frontal eye field or parietal regions alone do not have a substantial effect on visually guided saccades, whereas combined lesions reduce the accuracy and i ncrease the latency for saccades to visual targets (Lynch 1,992). These monkeys, however, recovered from th e effects of th ese lesions after several weeks. Schi ller and colleagu es (1 980) have shown that combined bilateral lesions of the superior colliculus and frontal eye field cause substantial reductions in the range of eye movements, as well as decreases in the frequency and velocity of saccades. These deficits do not diminish over time, and suggest that the frontal eye field generates an essential output from the cortical network. The frontal eye field has many features in common with the superior colliculus. Both regions have neurons that discharge before movements; both can support saccades in the absence of the other; and of all cortical regions studied, only the frontal eye field can evoke saccades by electrical stimulation in the absence of the superior colliculus (Keating et al. 1983; Schiller 1977). Although they seem to have parallel functions in the generation of saccades, the frontal eye field does have the ability to influence the colliculus in the generation of saccades in the normal monkey. It sends a powerful presaccadic signal to the colliculus (Segraves and Goldberg 1987), and may also help to control the basal ganglia circuits, which themselves exert an influence on the colliculus (Hikosaka et al. 1989). It is believed to exert a cognitive influence on a more stimulus-bound colliculus, helping to select the appropriate eye movement to make from a variety of possibilities (Goldberg and Segraves 1987). Many characteristics of saccade-related activity in the superior colliculus have been known for some time. The most prominent feature of the firing pattern of neurons with saccade-related activity is a burst of firing that begins ~20 ms before the start of the eye movement (Mays and Sparks 0022-3077/93 $2.00 Copyright 0 1993 The American Physiological Society FRONTAL EYE FIELD AND SACCADE DYNAMICS 1980; Mohler and Wurtz 1976; Sparks 1978). For some neurons, this burst is preceded by a gradual increase in firing rate beginning shortly after the signal to make a saccade-several hundred milliseconds before the beginning of the movement. Because the focus of this activity arises from a location within the movement map in the intermediate and deep layers of the colliculus, this activity could signal the amplitude and direction of an impending movement. The burst of activity occurs at a time appropriate to trigger brain stem saccade-generating circuitry. Recent evidence, however, suggests that the activity of subpopulations of collicular neurons helps to control dynamic aspects of the saccadic eye movement. Focal lesions in the superior colliculus result in decreased saccadic velocity as well as dysmetria (Hikosaka and Wurtz 1986; Lee et al. 1988), suggesting that the superior colliculus is involved in the actual motor programming of the saccade. Berthoz and colleagues ( 1986) showed that part of the collicular discharge resembles the velocity profile of the ongoing saccade, and Waitzman and colleagues ( 1988, 1991) have shown that the discharge of some collicular eye movement neurons decreases as a linear function of the distance remaining in the saccade, the motor error. Because of its close relationship to the superior colliculus in the generation of movement, we thought it likely that the frontal eye field would also be involved in the control of saccade dynamics. Such a relationship would be particularly important in neurons that project to the pons, neurons that are likely to drive saccadic eye movements in the absence of the superior colliculus. In these experiments, we compared the relationship between neuronal discharge and saccadic eye movements to see if the frontal eye field, like the superior colliculus, was important in the dynamic programming of saccades. We studied three populations of movement neurons: identified corticotectal neurons, identified corticopontine neurons, and neurons that could not be excited antidromically from either the superior colliculus or the pons. The presaccadic component of the activity of these neurons was similar to the original descriptions of saccade-related neurons in the colliculus. This activity began to rise above baseline - 100 ms before the beginning of the saccade, and peaked just before the start of the eye movement. This signal might I) relay the amplitude and direction of an impending saccade to the superior colliculus and pons and 2) initiate the trigger to make a saccade. We failed to show a consistent relationship between saccade dynamics and neuronal discharge for these cortical neurons. In particular, frontal eye field neurons did not appear to emit a signal that could be related to motor error or velocity during the saccade. A preliminary report of these experiments has been published elsewhere (Segraves et al. 1990). 1881 V H StimuIusOnset 0 2ooms Visual response of corticopontine movement neuron. Activity in visual no-saccade task. The monkey fixed his gaze on a fixation point (FP) directly in front of him. During the time that he was fixating, a light was turned on for 50 ms at a location with radius of 18’ and angle of 109O (S). The monkey was rewarded for keeping his gaze on the fixation point. The stimulus light was used to test the neuron’s visual responsiveness. Below the schematic of fixation point and stimulus light positions are the recordings of vertical (V) and horizontal (H) eye position during the trial. Below the eye position traces are a raster and histogram showing the activity of this neuron (sampled at 250 Hz) during a period of 200 ms before and after the onset of the peripheral stimulus. Each line in the raster contains the activity from a single trial, and each small vertical line indicates the time at which the neuron fired. The data from 16 consecutive trials included in this raster are aligned on the onset of the stimulus. This time is indicated by the large vertical line drawn through the center of the raster. The histogram shows the sum of the activity from these trials, with the vertical line again indicating the onset of the stimulus. The vertical bracket at the bottom l@ of the histogram indicates the amplitude of a firing rate of 100 spikes per second. Note the faint visual response beginning -80 ms after the onset of the stimulus. The same general configuration used in this figure is used in Figs. 2 and 3. FIG. 1. neuronsencounteredduring experimentsdesignedto identify and characterize neuronswith projections to the superior colliculus and pons.The proceduresfor training, surgery,electrophysiology, METHODS and histology for these monkeys appear in the original reports The data describedin this paper were obtained during single- (Segraves1992;Segravesand Goldberg 1987). unit recordingsfrom four behavingrhesusmonkeys(A4acacamulcltfcr)identifiedasLSR30,LSRSO,LSRS1, and MASO 1.All exper- Behavioral tasks imental proceduresfor monkeysLSR30,LSRSO,and LSRS1 were performedat the National Eye Institute. The experimentalproceTo obtain the data describedin this report, behavioraltaskswith duresfor monkey MASOl and subsequentdata analysisfor all and without saccadeswere used. I) Visual no-saccadetask: the four monkeys was performed at Northwestern University. The monkey fixated the stationary light spot in the center of the tanneuronsdescribedin this report are a subsetof frontal eye field gent screen(FP in Fig. 1) for the duration of the trial. During the 1882 M. A. SEGRAVES A AND K. PARK B C V H Target Onset Saccade Beginning II II Saccade End IIII II I I' I I' II I I I I I I II I I IN I 1" I I :'I": II I I : I I I II I I I' ,'I ml11 I I :I I' : II llrn III I I I I II II 11 I' ' l I I I : 'I: II III* I II I l'a IY II II Ill I c 0 2ooms 2. Activity in the visually guided saccade task. While the monkey was fixating, the fixation point was turned off and a target light (T) turned on for 50 ms at the same location used in the visual no-saccade task illustrated in Fig. 1. The rasters and histograms illustrate activity occurring 200 ms before and after the onset of the target light (A), the beginning of the saccade ( B), and the end of the saccade ( C). In this task, the saccade begins ~200 ms after the onset of the target, and so it is difficult to separate visual- from movement-associated activity. It is clear, however, that the activity is much stronger than when the monkey did not make an eye movement (Fig. 1 ), and that the increase in activity peaks near the beginning of the saccade. FIG. trial, at an unpredictabletime, a secondlight spot(S in Fig. 1) was flashedfor a variable period of time. The position of the second light wascontrolled by mirror galvanometersand could be placed anywherewithin the monkey’s visual field. The monkey wasrewardedfor maintainingfixation of the center light throughout the trial. The peripheralstimulushad no behavioral significance.Its position wasusedto map the visual receptivefieldsof frontal eye field neurons. 2) Visually guided saccadetask: after a variable period of fixation, the center light wasextinguishedat the same momentthat a peripheraltarget light (T in Fig. 2) wasturned on. When the peripheraltarget appeared,the monkey wasrequiredto make a saccadeto the target’slocation within 500 ms. After the saccade,the monkey wasrewardedeither for maintainingeye position within a window with -3-6’ radiussurroundingthe new targetposition or for having madea saccadewhoseend point was within the window. The movable light spotwasagainusedfor the peripheraltarget, and could be positionedanywhere on the tangent screen.The duration of the target flashwasvariable. For the trials illustratedin Fig. 2, the target wason for only 50 ms,and the saccadewasactually madein darknessafter the disappearanceof the target. In another version of this task, the target remainedon until the saccadewascomplete and the trial ended.3) Memoryguided saccadetask: in a second,more difficult form of saccade task, the center light cameon to start the trial, asbefore, and the monkey wasrequiredto beginand maintain fixation of the center light until it wasturned off. During the central fixation period, a peripheraltarget wasturned on for 50-300 ms.When the center light wasturned off, the monkey wasrequiredto makea saccade to the position of the flashedperipheraltarget light. The duration and time of occurrenceof the target flashwereadjustedsothat the target light wasturned off during the fixation period, requiring the monkey to maintain fixation for 5250 ms after the target light’s disappearance and then make a saccadeto a rememberedtarget location. This task enabledusto distinguishbetweenvisual- and movement-associated neuronal activity during the trial. In this study, the location of visual stimuli and targetsare expressedin polar coordinates,radius, and angle, where radius is equal to the distance,in degreesof arc, of the target or stimulus from the central fixation point. An angleof 0” describesa rightward horizontal direction, and a 90” angledescribesan upward vertical direction. Data analysis Rastersand histogramswere constructed with neuron activity alignedon important eventswithin a trial. We classifiedneuronson the basisof their activity during visuomotor tasks,usingthe criteria of Bruceand Goldberg( 1985). Our presentanalysiswasconfined to neuronswith movement activity. Movement neuronshave little or no responseto visualstimuli, but very strong activity before both visually guided and memoryguided saccades.Movement neuronsin the frontal eyefield have definable movement fields. Their peak dischargeoccurs for saccadesof a specificamplitude and direction, with theseparameters dependingon the neuron’slocation in the frontal eyefield’stopographic map. To facilitate the comparisonbetween neuron activity and the changesin eye position and velocity during saccades,the static rasteror histogramplot of activity wasconverted to a continuous spike density function (Richmond et al. 1987; Sandersonand Kobler 1976). This wasdone in two ways.For most neurons( n = 33), both neuron firing activity and eye position data during eye movementtrials werestoredat 250or 1,000Hz. A Gaussiancurve with 0 = 8 mswasfitted to the time of occurrenceof eachneuron spike during a trial. For each trial, Gaussiancurves representing each spike were summed,providing a continuous function ex- FRONTAL EYE FIELD AND pressing the expected spike frequency for any point in time during the trial. This activity was then averaged across trials to produce the average spike density for the neuron in a particular task. For these data, the beginning and end of the saccade were defined as the times at which eye velocity exceeded and then decreased below 100’ /s. For the remaining neurons ( II = 15 ), continuous records of eye position during each trial were not available. For these cells, spike density functions were generated from rasters of neuron activity that contained timing marks to indicate the beginning and end of the saccade. These rasters were generated during the recording experiment with a sampling frequency of 250 Hz. The beginning and end of the saccade were determined on-line at the time of the experiment, and corresponded to the times at which eye velocity exceeded and then decreased below 3O”/s. For these data, a histogram of activity averaged across 16 trials was generated and then a spike density function, with CI again equal to 8 ms, was derived from the histogram. Although two different velocity criteria were used to distinguish the beginning and end of saccades, these times, as well as the durations of saccades obtained with both techniques, were very similar. RESULTS We examined the activity of a total of 48 frontal eye field movement neurons. These included 35 identified neurons: 26 corticotectal neurons and nine corticopontine neurons. In addition, our analysis included 13 movement neurons whose projections were not identified: one neuron that could not be antidromically excited by a collicular stimulating electrode, and 12 neurons that were not antidromically excited by a pontine electrode. The activity of a corticopontine movement neuron is illustrated in Figs. l-4. This particular neuron was chosen for illustration because the timing of the beginning, peak, and end of its saccade related activity was similar to the average times for the entire sample. During the visual nosaccade task (Fig. 1 ), this movement neuron gave only a few spikes per trial in response to the 50-ms flash of a spot of light in the center of its movement field. In contrast, it gave brisk bursts of activity that preceded and continued through the duration of both visually guided saccades (Fig. 2) and saccades made in darkness to the remembered location of the target spot (Fig. 3). Although the rasters show some variability from trial to trial, the histograms illustrate that, overall, the activity of this cell began to rise - 150 ms before the start of the saccade, peaked near the beginning of the saccade, and fell back to a baseline level - 100 ms after the end of the eye movement. The cell’s activity for both visually guided and memory-guided saccades was quite similar. Figure 4 illustrates the mean spike density for the same neuron shown in Figs. l-3. The mean spike density is superimposed on plots of vertical eye position and velocity. Velocity was computed by digital differentiation. The spike density is compared to saccades made to both visually (Fig. 4A ) and memory-guided saccades (Fig. 4B). A qualitative examination of the spike density profile confirms what was seen in the histograms of Figs. 2 and 3. The rise in spike density related to the saccade begins and ends > 100 ms before and after the saccade, and peaks near the start of the eye movement. Note that the spike density profile was similar for firing activity generated in both visually guided ( Fig. 4A ) and memory-guided ( Fig. 4 B) saccade tasks. SACCADE DYNAMICS 1883 For each of the 48 neurons included in our sample, we made quantitative measurements of the timing of the rise in spike density relative to the saccade. Figure 4A illustrates how these measurements were made. The baseline level of spike activity was determined by examining the activity for 5 1,000 ms before and after the start of the saccade. The beginning and end of the spike density rise were defined as the times at which spike density rose above and later returned to the baseline level of activity. The time for the beginning and peak of the rise in spike density was measured relative to the start of the eye movement. The time for the end of saccade-related spike density activity was measured relative to the end of the saccade. Figure 5 illustrates the distributions of times for the beginning, peak, and end of the change in spike density associated with saccades. For 27 neurons, the times used for these distributions were measured from data obtained during the visually guided saccade task. For the remaining 21 neurons, sufficient data for visually guided saccades did not exist, and the times for these distributions were measured from data obtained during the memory-guided saccade task. These movement neurons began to increase their activity from 3 1 to 2200 ms before the start of the saccade (Fig. 5A ). The mean time for activity to rise above the baseline level was 148 ms before the beginning of the saccade. Their activity peaked from 120 ms before to 93 ms after the start of the saccade (Fig. 5 B), with 58% of the neurons having a peak of activity within 40 ms of the start of the saccade. The mean time for activity to peak was 13 ms before the saccade began. Some significant differences ( t test, P < 0.05) were found for the mean times of peak activity for the different groups of neurons included in this study. The peak of activity for corticotectal neurons (-34 ms) occurred earlier than it did for both corticopontine (-2 ms) and unidentified (+20 ms) movement neurons. All neurons, except three corticotectal cells, finished firing after the saccade was completed (Fig. 5C). Most neurons continued firing for a substantial period of time after the end of the eye movement, with a mean of 93 ms after the end of the saccade. In the superior colliculus, neurons that are believed to code dynamic motor error showed a sharp decrement in their firing activity during a saccade (Waitzman et al. 199 1). These collicular neurons had peak activity at the beginning of a saccade, and their activity decreased nearly 100% during the course of the eye movement. We examined the decrement in activity during saccades for the frontal eye field neurons included in this study. For each neuron, we determined the magnitude of the change in activity during the saccade and expressed this value as a percentage of the peak firing activity. A change in activity from peak to baseline level during the saccade was represented as a 100% drop. Increases in activity during the saccade were expressed as negative percent drops. For example, for the neuron illustrated in Fig. 4, activity dropped an amount equal to 33% of the peak activity during the visually guided saccade task ( Fig. 4A), and dropped 28% during the memory-guided saccade task (Fig. 4 B). Figure 6 is a plot of these values for the 48 frontal eye field neurons included in this study. Each value is referenced to the latency of the neuron’s peak discharge to the beginning of the saccade. Overall, the activity of these neurons dropped 16.9 t M. A. SEGRAVES AND K. PARK B C V Saccade Beginning Saccade End Ill I I' I I '2 ' I' II I I I' II I I I II FIG. 3. Activity in the memory-guided saccade task. While the monkey was fixating, a target light was turned on for 350 ms and then turned off. Fifty milliseconds after the target was extinguished, the fixation point was turned off, signaling to the monkey that he could make a saccade. The saccade was made, in darkness, to the remembered location of the flashed target. The monkey was rewarded for making a saccade of appropriate amplitude and direction. Rasters and histograms are arranged as in Fig. 2. In this task, the onset of the target and the beginning of the saccade were separated by -500 ms. As a result, it is easyto distinguish the faint visual response in A from the substantial movement-related activity shown in R and (‘. 3 1.3% (mean t SD) during the eye movement. Note that only a few neurons showed drops >60%, and many neurons actually increased their activity during the saccade. The activity of unidentified movement neurons tended to drop less during the saccade than did other neurons, although this difference was only significant when comparing unidentified with corticotectal neurons (t test, P < 0.05 ). We attempted to allow for delays between the drop in activity after the peak and the actual time of generation of the movement by measuring the drop in activity during a time period equal to one saccade duration immediately after the peak in discharge. For the neuron illustrated in Fig. 4, this value was equal to 40% for visually guided saccades (Fig. 4A) and 27% for memory-guided saccades (Fig. 4B). This procedure eliminates the generation of negative percent drops. Figure 7 plots the distribution of these values for all 48 neurons. This manipulation ignores the latency required for frontal eye field activity to have an effect on eye movement dynamics, and produces a narrower distribution of changes in activity. We found an average of 34.9 t 15.4% (mean t SD) drop in activity. Again, only a small number of neurons showed a drop in activity of >60% of the peak activity level. Corticopontine neurons tended to drop less quickly after the peak of activity, but this tendency was only significant when comparing the means for corticopontine with unidentified movement neurons (t test, P < 0.05). DISCUSSION We have examined the activity of movement neurons in the frontal eve field and have comnared that activitv with the dynamics of saccadic eye movements. The population of movement neurons that we studied included cells with identified projections to the superior colliculus or to the pons, as well as neurons whose termination sites were unknown. As a group, these neurons began to increase their activity > 100 ms before the start of a saccade. Their activity peaked with a mean of 13 ms before the beginning of the saccade, and then gradually fell toward baseline levels. Their activity reached baseline 93 ms after the completion of the saccade. On average, the activity of these neurons fell during the saccade by an amount equal to only 17% of the peak activity level. There was a great deal of variability in this measurement, with a few neurons showing decrements of >60% during the saccade and a number of neurons showing increases in activity during the eye movement. This discussion will consider the probable role of frontal eye field movement activity in the generation of saccades, and the relationship of that activity to feedback control models of the saccadic system, and will make comparisons with the role performed by the superior colliculus in this process. Frontal eye jeld movement neurons signal the amplitude and direction qf an intended movement, and the time to initiate that movement The scope of this study was restricted to an examination of neurons with movement-related activity. Movement neurons form the largest population of cells that project to the superior colliculus and pons (Segraves 1992; Segraves and Goldberg 1987). Thus their activitv generates a maior out- FRONTAL EYE FIELD AND SACCADE DYNAMICS 1885 V. Positian V. Velocity Saccade Beginning -200 End 0 200 B -190 ms -200 +llO ms I 1 I 1 1 I I I 1 0 1 200 Time (ms) 4. Spike density during visually guided and memory-guided saccades. A plot of mean spike density is superimposed on the vertical component of eye position and eye velocity for a period of 200 ms before and after the beginning of the saccade. A: mean spike density for 10 trials of the visually guided saccade task. B: mean spike density for 1 1 trials in the memory-guided saccade task. A illustrates how the quantitative measurements used for this study were made: the neuron’s activity was examined for 5 1,000 ms before and after the beginning of the saccade, and a line drawn to indicate the activity baseline. The beginning and end of the rise in spike density were marked at the points where spike density rose above and later fell below the baseline level of activity. Eye velocity was computed by digital differentiation of the eye position trace, and was used to identify the beginning and end of the saccade. The beginning and end of the saccade were defined as the points at which the rise and fall of eye velocity passed the 100” /s level. The time of the beginning and peak of the spike density rise were measured relative to the beginning of the saccade. The end of the spike density rise was measured relative to the end of the saccade. Note that the changes in spike density associated with the saccade are quite similar for both visually guided and memory-guided saccades. In this example, neuron activity was sampled at 1 kHz; thus a spike density value ofO.1 signifies an expected firing frequency of 100 Hz. FIG. put signal that is sent from frontal eye field to the brain stem. We have not included fovea1 neurons in this study. Frontal eye field fovea1 neurons are responsive to visual stimulation of the fovea and, in addition, have activity related to fixation (Bizzi 1968; Bruce and Goldberg 1985; Suzuki and Azuma 1977; Suzuki et al. 1979). Fovea1 neu- rons compose the second-largest group of neurons that project to the superior colliculus and pons. Their activity is believed to contribute to the maintenance and release of fixation (Segraves 1992; Segraves and Goldberg 1987). Figure 8 summarizes our current understanding of the signals sent by frontal eye field movement neurons to both 1#86 M. A. SEGRAVES AND K. PARK 507 A 100% Beginning Standard of Spike Deviation I :2ob Density = 54 Saccad; Beginning Peak of Spike Mean = -13 Standard Density Deviation 1 Rise 0 Corticotectal l Corticopontine A Unidentified 2 +200 Rise = 49 I l 7~ ---.- ~-~-r---T- t -200 Saccade Latency 5 1200 End Mean Standard Saccad; Beginning of Spike Density 2 +200 Rise +200 Beginning of Peak Discharge to Start of Saccade (ms) FIG. 6. Drop in activity during saccade. Drop in activity during the saccade is expressed as a percentage of peak activity for the cell, and referenced to the latency of the cell’s peak discharge to the beginning of the saccade. Negative percentage values signify a rise in activity during the saccade. = +93 Deviation = 66 Saccade End Time (ms) 5. Timing of the beginning, peak, and end of the rise in spike density associated with the saccade. Measurements taken from plots such as those illustrated in Fig. 4 were used to produce histograms of the distribution of the times for beginning (A), peak (R), and end (C) of the rise in saccade-related spike density. These data were put into bins 20 ms wide. Note that for A and B, time = 0 marks the beginning ofthe saccade, and for C‘, time = 0 marks the end of the saccade. Time values less than -200 ms were included in the -200-ms bin: values greater than +200 ms were included in the +200-ms bin. Values for 48 neurons are included in these histograms. FIG. the superior colliculus and pons. Our interpretation of the present results is that the movement neurons in the frontal eye field provide two signals relevant to the generation of a saccadic eye movement. I) The peak of their activity signals the time to initiate the saccade. It is a trigger signal. The peak in movement neuron activity, at a mean of 13 ms before the saccade, occurs at about the time that omnipause neurons are turned off at lo- 15 ms before the saccade (Henn and Cohen 1976; Keller 1974; Luschei and Fuchs 1972; Robinson 1975 ). On the basis of the latencies of corticopontine, corticotectal, and tectopontine cells, this signal could directly excite pontine cells within 1.4-4.0 ms and within 2.3-4.9 ms via a projection from cortex to colliculus to pons (Raybourn and Keller 1977; Segraves 1992; Segraves and Goldberg 1987). Because it is unlikely that the corticopontine neurons are directly inhibitory (Schwartz et al. 1985, 1988) if frontal eye field neurons turn off pause neurons, they probably do so via inhibitory neurons in the vicinity of the pause cells. 2) The sustained increase in activity output from a group of movement neurons signals the amplitude and direction for the impending saccade. This activity would originate from a restricted area within the movement map of the frontal eye field that represents large eye movements dorsomedially and small eye movements ventrolaterally ( Bruce et al. 1985 ). Our study does not rule out the possibility that some frontal eye field neurons send a dynamic motor error signal to the midbrain and pons; our sample may have missed those neurons. The present analysis, however, makes it clear that the frontal eye field is sending both a trigger and a desired change in eye position signal to the midbrain and pons. These signals are very similar to those ascribed by Sparks and colleagues to saccade-related burst neurons in the superior colliculus. Those collicular neurons gave a burst of activity that began - lo-20 ms before the start of a saccade with optimal amplitude and direction. About one fourth of superior colliculus movement neurons showed a gradual rise during the period between the signal to make a saccade and the beginning of the saccade (Mays and Sparks 1980; Sparks et al. 1976). This collicular activity may arise, in part, from frontal eye field input, but is probably also due to the integration of input from many other sources. Concerning the slow rise in fronf$ 10096, t 2 g li ! 3 z g o .i:F g g & ’ 0 0 0 5. 0 0. 0 . 0. l A . 0 l 0 0. 0 oi ’ 1 0 *a 0 lA . A .* Ir Corticopontine A Unidentified l dA l *. A A .+ 0 l 0 l .(__ ~~ ~7 ~~~~r-~--~7-----, Saccade Latency Corticotectal l 0 l &fmp.mm -200 0 of Peak Discharge +200 Beginning to Start of Saccade (ms) FIG. 7. Drop in activity during the time beginning with the peak in activity and lasting for a period equal to the duration of the saccade. See text for additional details. FRONTAL EYE FIELD AND SACCADE DYNAMICS 1887 demonstrated that some neurons in the monkey superior colliculus appear to have activity that codes dynamic motor error (Waitzman et al. 1988, 199 1). They propose that the colliculus generates this signal by continually subtracting a current change in eye position signal from the initial desired change in eye position signal. The current change in eye position signal would be obtained by integrating the eye SUPERIOR PONTINE OMNIPAUSE ! velocity command output by burst neurons (Waitzman et & BURSTNEURONS -( coLLIcuLus + al. 1988, 199 1). Munoz and colleagues have proposed that neurons in the cat superior colliculus lie within a gaze feedback loop and exhibit changes in activity that are related to the dynamics of movements of the eyes and head (Munoz FIG. 8. Input of frontal eye field movement neurons to the superior et al. 199 1). In this case, brain stem feedback is used to shift colliculus and pons. The signals output by the frontal eye field are based on the center of activity across the collicular movement map, the present study. Signals sent between superior colliculus and pons are compiled from a number of works cited in the text. See text for additional from the region coding the desired saccade amplitude and detail. direction toward the fixation zone at the rostra1 pole of the colliculus. Their model also requires that the feedback sigtal eye field activity before the movement, it is interesting to nal pass through a neural integrator (Guitton 199 1; Munoz note that Sparks and colleagues ( 1987) have reported that et al. 1991). activity within the paramedian pontine reticular formation The superior colliculus may also help to control the velocconcerning an impending eye movement builds up graduity of saccades (Berthoz et al. 1986; Hikosaka and Wurtz ally over a IOO-ms period before the start of the saccade. 1986; Lee et al. 1988). We did not, however, find evidence This corresponds closely to the rise time for frontal eye field in this study for a relationship between eye velocity and the movement activity. activity of movement neurons in the frontal eye field. In most cases, frontal eye field activity declined slowly during Frontal evefield does not contribute tofiedback control or the saccade-an inappropriate signal for coding the dythe specljication of eye velocity during saccades namics of eye velocity. Waitzman and colleagues ( 199 1) Feedback is an important component of the control sys- have pointed out that a steadily declining signal could not tem for generating saccades [see Van Gisbergen and Van code the increase and decrease of velocity that occur during Opstal ( 1989) for review]. Early experimental evidence for a saccade. If it did, equal velocities during the acceleration this feedback came from patients with spinocerebellar de- and deceleration phases of the movement would be coded generation whose saccades were sufficiently slow to allow by different levels of neuron activity. Unfortunately, our intrasaccadic modification of their trajectories (Zee et al. experiments did not allow us to address the possibility that 1976). Other evidence came from normal human subjects frontal eye field activity might code the peak velocity for an (Becker and Jurgens 1979; Hallett and Lightstone 1976) impending saccade. Although it has been reported that the and monkeys (Sparks and Mays 1983). Robinson ( 1975, velocity of equal-length saccades can vary depending on the 198 1) proposed a local feedback control model where de- eye movement task (Rohrer et al. 1987) we noted very sired eye position is compared with an efference copy of little variation in the velocity of saccades generated in our visually versus memory-guided saccade tasks. One possible actual eye position to yield a motor error signal. The motor error signal drives the burst neurons, which in turn produce explanation is that the gap between target disappearance and the cue to make a saccade tended to be relatively short the pulse of eye velocity that provides the major component in our memory task (5250 ms). of the motor neuron discharge. In subsequent modifications of the feedback model (Jiirgens et al. 198 1; Scudder 1988; Van Gisbergen and Van Opstal 1989), desired sac- Relative contributions of’the fkontal eye field and superior cade amplitude is compared with the distance traversed by colliculus in the genera&z of ‘saccad& * the current saccade to produce the motor error signal. The Placing the superior colliculus, but not the frontal eye present results suggest that the frontal eye field does not participate in the feedback process, because its signals are field, within a feedback loop involved in the dynamic control of saccades is difficult to reconcile with the finding that relatively constant during the saccade. Thus the feedback loop must occur at a level downstream from the frontal eye monkeys can make visually guided saccades after bilateral field cortex. Scudder ( 1988) has proposed that the feedback removal of the superior colliculus (Schiller et al. 1980). occurs at the level of long lead burst neurons in the pons, This lesion study emphasized the parallel nature of the parwith frontal eye field and superior colliculus providing simiticipation of colliculus and frontal eye field in the generalar inputs specifying the saccade target and the trigger to tion of saccades. If, as recent findings suggest, the superior initiate the movement. The activity of many saccade-re- colliculus is within a feedback loop that supplies the oculolated neurons in the superior colliculus seems appropriate motor system with a dynamic motor error signal, how is this signal derived in the absence of the superior colliculus? for this function, and is in many ways similar to that found in the frontal eye field (Mays and Sparks 1980; Mohler and One possibility is that the frontal eye field, as well as other areas, modify their output in the abWurtz 1976; Sparks 1978). Recent experiments, however, oculomotor-related sence of the superior colliculus and provide the needed moprovide evidence that would place the superior colliculus within the feedback loop. Waitzman and colleagues have tor error signal to the brain stem. Anatomic projections TRIGGER DESIRED CHANGE IN EYE POSITION %rg,gcr Dewed Change III Eye Position Velocity VdOCl~ - Dvnamrc Current hfomr Chunge Error rn Eye Posrrmn 1888 M. A. SEGRAVES from oculomotor cortex to the brain stem that bypass the superior colliculus are known to exist. Both the frontal eye field and supplementary eye field project directly to oculomotor regions of the pons, including the region of omnipause neurons in the nucleus raphe interpositus; the general vicinity of burst neurons in the paramedian pontine reticular formation; and nuclei forming relays with the cerebellum, including the nucleus reticularis tegmenti pontis and basal pontine nuclei (Huerta and Kaas 1990; Huerta et al. 1986; Leichnetz et al. 1984; Schnyder et al. 1985; Shook et al. 1990; Stanton et al. 1988). An alternative explanation is that the dynamic motor error signal observed in the superior colliculus is not necessary to make accurate saccades, but is perhaps part of a corollary discharge generated by the colliculus. It should be noted that the anatomic targets of the collicular dynamic motor error cells described by Waitzman and colleagues have not been identified. Finally, it may be possible that a desired change in eye position signal from the frontal eye field is sufficient for other structures such as the cerebellum to drive the saccade-generating mechanism in the absence of a motor error signal from the colliculus. Although it is unclear how saccades are generated in the absence of the superior colliculus, under normal circumstances the superior colliculus and frontal eye field, as well as other regions including posterior parietal cortex, the supplementary eye field, and periprincipal cortex operate via both serial and parallel pathways to control the generation of saccades. Our examination of frontal eye field activity combined with the work of other laboratories investigating superior colliculus activi tY reveals a significant difference between the superior colliculus and frontal eye field in their roles in the generation of saccades. Whereas the colliculus appears to code signals that are related to dynamic aspects of the movement, like velocity and motor error, the frontal eye field functions at a higher level in the saccade generation process, signaling the target for the saccade and triggering the movement. This emphasizes a purported role for the frontal eye field as a link between cortical cognitive processing preceding the initiation of an eye movement and the more machine-like organization of the midbrain and brain stem designed to execute the motor command (Goldberg and Segraves 1987). As such, the frontal eye field may select the target for the next eye movement from a variety of external visual stimuli and internal cues, but it is not involved in the guidance of the movement once it has begun. We are grateful to the following for skillful support: A. Hayes for computer software; A. Ziminsky and Northwestern’s Chemistry Electronics Shop for electronic hardware; C. Crist, T. Ruffner, and Northwestern’s Physics Instrument Shop for machining; M. Smith for histology; C. Polchow for histology and photography; and A. Hanson, H. Macgruder, K. Powell, and the staff of Northwestern’s Center for Experimental Animal Resources for animal care. We are particularly grateful to M. Goldberg for providing the computer software to generate the spike density functions and for innumerable discussions and insights concerning this work. We thank D. Burman for his comments and discussion concerning this manuscript. This work was supported by National Eye Institute Grant EY-082 12, Division of Research Resources-Biomedical Research Development Grant-BRSG Award RR-07028, and The Sloan Foundation. Address for reprint requests: M. A. Segraves, Dept. of Neurobiologv and AND K. PARK Physiology, 0. T. Hogan Hall, Northwestern 60208-3520. University, Evanston, IL Received 24 July 1992; accepted in final form 29 January 1993. REFERENCES R. A. AND GNADT, J. W. Posterior parietal cortex. In: The Neurobiology qf’Succadic Eye Movements. Reviews of Oculomotor Research, edited by R. H. Wurtz, M. E. Goldberg. Amsterdam: Elsevier, 1989, vol. III, p. 3 15-335. BARASH, S., BRACEWELL, R. M., FOGASSI, L., GNADT, J. W., AND ANDERSEN, R. A. Saccade-related activity in the lateral intraparietal area. I. Temporal properties: comparison with area 7a. J. Neurophysiol. 66: 1095-l 108, 1991a. BARASH, S., BRACEWELL, R. M., FOGASSI, L., GNADT, J. W., AND ANDERSEN, R. A. Saccade-related activity in the lateral intraparietal area. II. Spatial properties. J. Neurophysiol. 66: 1109- 1 124, 199 1b. BECKER, W. AND J~~RGENS, R. An analysis of the saccadic system by means of double step stimuli. Vision Res. 19: 967-983, 1979. BERTHOZ, A., GRANTYN, A., AND DROULEZ, J. Some collicular efferent neurons code saccadic eye velocity. Ncurosci. Lett. 72: 289-294, 1986. BIZZI, E. Discharge of frontal eye field neurons during saccadic and following eye movements in unanesthetized monkeys. Exp. Bruin Res. 6: 6980, 1968. BOCH, R. A. AND GOLDBERG, M. E. Participation of prefrontal neurons in the preparation of visually guided eye movements in the rhesus monkey. J. Neurophysio1. 6 1: 1064- 1084, 1989. BRUCE, C. J. AND GOLDBERG, M. E. Primate frontal eye fields. I. Single neurons discharging before saccades. J. Neurophysiol. 53: 603-635, 1985. BRUCE, C. J., GOLDBERG, M. E., STANTON, G. B., AND BUSHNELL, M. C. Primate frontal eye fields. II. Physiological and anatomical correlates of electrically evoked eye movements. J. Neurophysiol. 54: 7 14-734, 1985. DENG, S.-Y ., GOLDBERG, M. E., SEGRAVES, M. A., UNGERLEIDER, L. G., AND MISHKIN, M. The effect of unilateral ablation of the frontal eye fields on saccadic performance in the monkey. In: Adaptive Process in the Visual and Oculomotor Systems, edited by E. Keller and D. S. Zee. Oxford, UK: Pergamon, 1986, p. 20 l-208. DUHAMEL, J. R., COLBY, C. L., AND GOLDBERG, M. E. The updating ofthe representation of visual space in parietal cortex by intended eye movements. Science W’ash. DC 255: 90-92, 1992. FUNAHASHI, S., BRUCE, C. J., AND GOLDMAN-RAKIC, P. S. Neuronal activity related to saccadic eye movements in the monkey’s dorsolateral prefrontal cortex. J. IvczIr~)~~lI?‘.sio/. 65: 1464- 1483, 199 1. GOLDBERG, M. E. AND SEGRAVES, M. A. Visuospatial and motor attention in the monkey. Nellroys!v’t7ologiu 25: 107- 118, 1987. GUITTON, D. Control of saccadic eye and gaze movements by the superior colliculus and basal ganglia. In: Eye IMovements, Vision and Visual Dys/unction, edited by R. H. S. Carpenter. Boca Raton, FL: CRC, 199 1, vol. VIII, p. 244-276. GUITTON, D., BUCHTEL, H. A., AND DOUGLAS, R. M. Frontal lobe lesions in man cause difficulties in suppressing reflexive glances and in generating goal-directed saccades. E-1-p.Brain Res. 58: 455-472, 1985. HALLETT, P. E. AND LIGHTSTONE, A. D. Saccadic eye movements towards stimuli triggered by prior saccades. Vision Rex 16: 99- 106, 1976. HENN, V. AND COHEN, B. Coding of information about rapid eye movements in the pontine reticular formation of alert monkeys. Brain Res. 108: 307-325, 1976. HIKOSAKA, O., SAKAMOTO, M., AND USUI, S. Functional properties of monkey caudate neurons. I. Activities related to saccadic eye movements. J. NeurophysioI. 6 1: 780-798, 1989. HIKOSAKA, 0. AND WURTZ, R. H. Saccadic eye movements following injection of lidocaine into the superior colliculus. Exp. Bruin Res. 61: 53 l-539, 1986. HUERTA, M. F. AND KAAS, J. H. Supplementary eye field as defined by intracortical microstimulation: connections in macaques. J. Comp. Neural. 293: 299-330, 1990. HUERTA, M. F., KRUBITZER, L. A., AND KAAS, J. H. Frontal eye field as defined by intracortical microstimulation in squirrel monkeys, owl monkeys, and macaque monkeys. I. Subcortical connections. J. Comp. Neural. 253: 4 15-439, 1986. J~~RGENS, R., BECKER, W., AND KORNHUBER, H. H. Natural and druginduced variations of velocitv and duration of human saccadic eve ANDERSEN, FRONTAL EYE FIELD AND movements: evidence for a control of the neural pulse generator by local feedback. Biol. Cyhern. 39: 87-96, 198 1. KEATING, E. G., GOOLEY, S. G., PRATT, S. E., AND KELSEY, J. E. Removing the superior colliculus silences eye movements normally evoked from stimulation of the parietal and occipital eye fields. Brain Res. 269: 145-148, 1983. E. L. Participation of medial pontine reticular formation in eye movement generation in the monkey. .I. Neuroph-vsiol. 37: 3 16-332, KELLER, 1974. SACCADE SCHLAG, 1253, 1991. SCHWARTZ, SEGRAVES, G., SMITH, D. J., AND SPENCER, R. F. Cortical projections to the paramedian tegmental and basilar pons in the monkey. J. Comp. LEICHNETZ, 228: 388-408, 1984. E. S. AND FUCHS, A. F. Activity of brain stem neurons during eye movements of alert monkeys. J. Neuruphysiol. 35: 395-403, 1972. LYNCH, J. C. Saccade initiation and latency deficits after combined lesions of the frontal and posterior eye fields in monkeys. J. Neurophysiol. 68: 1992. 1980. C. W. AND WURTZ, R. H. Organization of monkey superior colliculus: intermediate layer cells discharging before eye movements. J. MOHLER, 1992. M. A. AND GOLDBERG, M. E. Functional properties of corticotectal neurons in the monkey’s frontal eye field. J. Neurophysiol. 58: 1387-1419, 1976. 66: 1642-1666, 1991. RAYBOURN, M. S. AND KELLER, SPARKS, E. L. Colliculoreticular organization in primate oculomotor system. .I. Neurophysiol. 40: 86 l-878, 1977. RICHMOND, B. J., OPTICAN, L. M., PODELL, M., AND SPITZER, H. Temporal encoding of two-dimensional patterns by single units in primate inferior temporal cortex. I. Response characteristics. J. Neurophysiol. 1987. D. A. Oculomotor control signals. In: Basic A4echanisms of und Their Clinical Implications, edited by G. Lennerstrand and P. Bach-y-Rita. Oxford, UK: Pergamon, 1975, p. 337-374. ROBINSON, D. A. The use of control systems analysis in the neurophysiology of eye movements. Annu. Rev. Neurosci. 4: 463-503, 198 1. ROBINSON, D. A. AND FUCHS, A. F. Eye movements evoked by stimulation of frontal eye fields. J. Neuroph~~siol. 32: 637-648, 1969. ROHRER, W. H., WHITE, J. M., AND SPARKS, D. L. Saccade-related burst cells in the superior colliculus: relationship of activity with saccadic velocity. Sot. Netrrosci. Ahstr. 13: 1092, 1987. SANDERSON, A. C. AND KOBLER, B. Sequential interval histogram analysis of nonstationary neuronal spike trains. Biol. Cybern. 22: 6 l-7 1, 1976. SCHALL, J. D. Neuronal activity related to visually guided saccadic eye movements in the supplementary motor area of rhesus monkeys. J. Motility 66: 530-558, 21-34, 1976. D. L. AND MAYS, L. E. Spatial localization of saccade targets. I. Compensation for stimulation-induced perturbations in eye position. J. Neurobiol. 49: 45-63, 1983. D. L., MAYS, L. W., AND PORTER, J. D. Eye movements induced by pontine stimulation: interaction with visually triggered saccades. J. SPARKS, Neurophysiol. 58: 300-3 18, 1987. G. B., BRUCE, C. J., AND GOLDBERG, M. E. Frontal eye field efferents in the macaque monkey. II. Topography of terminal fields in midbrain and pons. J. Comp. Neural. 27 1: 493-506, 1988. SUZUKI, H. AND AZUMA, M. Prefrontal neuronal activity during gazing at a light spot in the monkey. Bruin Res. 126: 497-508, 1977. SUZUKI, H., AZUMA, M., AND YUMIYA, H. Stimulus and behavioral factors contributing to the activation of monkey prefrontal neurons during gazing. Jpn. J. Ph~wol. 29: 47 l-489, 1979. VAN GISBERGEN, J. A. M. AND VAN OPSTAL, A. J. Models. In: The Neuro- STANTON, ROBINSON, Neurophysiol. 1978. D. L., HOLLAND, R., AND GUTHRIE, B. L. Size and distribution of movement fields in the monkey superior colliculus. Bruin Res. 1 13: SPARKS, Ocular 16: 899, 1990. B. L., SCHLAG-REY, M., AND SCHLAG, J. Primate supplementary eye field. I. Comparative aspects of mesencephalic and pontine connections. J. Comp. Neural. 30 1: 6 18-642, 1990. SPARKS, D. L. Functional properties of neurons in the monkey superior colliculus: coupling of neuronal activity and saccade onset. Brain Res. SHOOK, 156: 1-16, D. P., GUITTON, D., AND PELISSON, D. Control of orienting gaze shifts by the tectoreticulospinal system in the head-free cat. III. Spatiotemporal characteristics of phasic motor discharges. J. Neurophysiol. 1987. M. A., PARK, K. L., AND GOLDBERG, M. E. Relationship of monkey frontal eye field activity to saccade dynamics. Sot. Neurosci. SEGRAVES, MUNOZ, 57: 132-146, 1988. C. A. A new local feedback model of the saccadic burst genera- tor. J. Neurophysiol. 59: 1455- 1475, 1988. M. A. Activity of monkey frontal eye field neurons projecting to oculomotor regions of the pons. J. Neurophysiol. 68: 1967- 1985, Abstr. L. E. AND SPARKS, D. L. Dissociation of visual and saccade-related responses in superior colliculus neurons. J. Neurophvsiol. 43: 207-232, MAYS, 39: 722-744, P. S. Periodicity J. Neuroscience SEGRAVES, LUSCHEI, Neurophysiol. D. S., AND GOLDMAN-RAKIC, cells in primate prefrontal cortex. of GABA-containing Land. 1913-1916, 1 1: 503, 1985. M. L., ZHENG, 8: 1962-1970, 1988. M. Evidence for a supplementary eye field. 57: 179-200, 1987. REISINE, H., HEPP, K., AND H., HENN, V. Frontal eye field projection to the paramedian pontine reticular formation traced with wheat germ agglutinin in the monkey. Bruin Res. 329: 15 l-160, 1985. SCHWARTZ, M. L., ZHENG, D. S., AND GOLDMAN-RAKIC, P. S. Laminar and tangential variation in the morphology and distribution of GABAcontaining neurons in rhesus monkey prefrontal cortex. Sot. Neurosci. SCUDDER, 332: 357-360, 1889 SCHLAG-REY, SCHNYDER, C., ROHRER, W. H., AND SPARKS, D. L. Population coding of saccadic eye movements by neurons in the superior colliculus. Nature LEE, Neural. J. AND J. Neurophysiol. Abstr. D. D. AND SKAVENSKI, A. A. Eye movements elicited by electrical stimulation of area PG in the monkey. J. Neurophysiol. 65: 1243- KURYLO, DYNAMICS 199 1. P. H. The effect of superior colliculus ablation elicted by cortical stimulation. Bruin Res. 122: 154- 156, SCHILLER, P. H., TRUE, S. D., AND CONWAY, J. L. Deficits ments following frontal eye field and superior colliculus SCHILLER, N~~zrr~~~l~??).siol. 44: 1 175- 1 189, 1980. biology Brain on saccades in eye moveablations. J. Eve Movements. Reviews of Oculomotor Research, Res. 72: 649-652, D. M., MA, 1988. T. P., OPTICAN, L. M., AND WURTZ, R. H. Superior colliculus neurons mediate the dynamic characteristics of saccades. WAITZMAN, 1977. oj’Saccadic edited by R. H. Wurtz and M. E. Goldberg. Amsterdam: Elsevier, 1989, vol. III, p. 69- 10 1. WAITZMAN, D. M., MA, T. P., OPTICAN, L. M., AND WURTZ, R. H. Superior colliculus neurons provide the saccadic motor error signal. Lkp. J. Neurophysiol. 66: I7 16- 1737, 199 1. D. S., OPTICAN, L. M., COOK, J. D., ROBINSON, D. A., AND ENGEL, W. K. Slow saccades in spinocerebellar degeneration. .4rch. Neural. 33: ZEE, 243-251, 1976.