Download Primate Frontal Eye Fields. II. Physiological and Anatomical

JOURNALOF NEUROPHYSIOLOGY
Vol. 54, No. 3, September 1985. Printed
Primate Frontal Eye Fields. II. Physiological
and Anatomical Correlates of Electrically
Evoked Eye Movements
CHARLES
J. BRUCE, MICHAEL
M. CATHERINE
BUSHNELL,
AND
E. GOLDBERG,
GREGORY
B. STANTON
Laboratory of SensorimotorResearch,National Eye Institute, National Institutes of Health,
Bethesda,MaryZand20205; Sectionof Neuroanatomy, Yale University Schoolof Medicine,
New Haven, Connecticut06510; Departmentof Neurology, GeorgetownUniversitySchoolof Medicine,
Washington,DC 20007; and Departmentof Anatomy, Howard University, Washington,DC 20059
but corresponds with the union of Walker’s
cytoarchitectonic
areas 8A and 45.
1. We studied single neurons in the frontal
6.
Saccade
amplitude
was topographically
eye fields of awake macaque monkeys and
compared their activity with the saccadic eye organized across the frontal eye fields. Ammovements elicited by microstimulation
at the plitudes of elicited saccades ranged from < lo
to >30”. Smaller saccades were evoked from
sites of these neurons.
the
ventrolateral portion, and larger saccades
2. Saccades could be elicited from electrical
stimulation in the cortical gray matter of the were evoked from the dorsomedial portion.
frontal eye fields with currents as small as 10 Within the arcuate sulcus, evoked saccades
PA. Low thresholds for eliciting saccades were were usually larger near the lip and smaller
found at the sites of cells with presaccadic ac- near the fundus.
7. Saccade direction had no global orgativity. Presaccadic neurons classified as visnization
across the frontal eye fields; however,
uomovement or movement were most assosaccade
direction
changed in systematic prociated with low (~50 PA) thresholds.
gressions with small advances of the micro3. High thresholds (> 100 ,uA) or no elicited
saccades were associated with other classes of electrode, and all contralateral saccadic directions were often represented in a single elecfrontal eye field neurons, including neurons
trode
penetration down the bank of the arcuate
responding only after saccades and presaccadic
sulcus. Furthermore, the direction of change
neurons, classified as purely visual.
4. Throughout the frontal eye fields, the in these progressions periodically reversed, allowing particular saccade directions to be
optimal
saccade for eliciting presaccadic
neural activity at a given recording site pre- multiply represented in nearby regions of
dicted both the direction and amplitude of the cortex.
8. These experiments
support the hysaccades that were evoked by microstimulapotheses
that
frontal
eye
field presaccadic
tion at that site. In contrast, the movement
neurons have a causal role in the generation
fields of postsaccadic cells were usually differof voluntary saccades and that microstimuent from the saccades evoked by stimulation
lation
in the frontal eye fields elicits eye moveat the sites of such cells.
ments
by artificially activating these cells, and
5. We defined the low-threshold frontal eye
hence their subcortical targets.
fields as cortex yielding saccades with stimulation currents 150 PA. It lies along the pos- INTRODUCTION
terior portion of the arcuate sulcus and is
largely contained in the anterior bank of that
It has been known since 1870 (15-17) that
sulcus. It is smaller than Brodmann’s area 8 electrical stimulation of the frontal lobes can
SUMMARY
I14
AND
CONCLUSIONS
ELECTRICAL
STIMULATION
elicit contraversive eye movements. Recent
studies that have used unanesthetized monkeys (67) and measured eye position accurately
(54) have established that these elicited movements are conjugate saccades, indistinguishable from naturally occurring saccades. Deep
anesthesia raises the threshold for eliciting
saccades, and light anesthesia can give an illusion of elicited nystagmus because the eyes
persistently drift back after each saccade. A
given electrode site yields saccades with a
characteristic direction and amplitude that are
largely independent of stimulation parameters.
For example, prolonged stimulation
repetitively elicits the characteristic saccade for a
site (54). Simultaneous stimulation
of two
frontal eye field loci evokes a saccade that is
approximately the mean of the characteristic
saccades at the two sites (54).
Although the phenomenology
of these
evoked saccades is well described, it has not
been shown how, if at all, they are related to
the functional physiology of this cortex. Because there is little presaccadic activity in the
frontal eye fields when monkeys make spontaneous eye movements (5, 6, 8), it has been
suggested that the frontal eye fields have no
role in generation of saccades and that the
elicited eye movements reflect the antidromic
activation of a corollary discharge pathway
(5 1). However, it is now known that substantial frontal eye field activity precedes purposive
saccadic eye movements, particularly saccades
directed at visual targets (82 1,43,46, 72). In
a recent study of the frontal eye fields (8) we
identified three distinct types of neural activity
preceding visually guided saccades: anticipatory activity, visual activity, and movement
activity. Different neurons had different combinations of these activities, and we distinguished three major classes of presaccadic
neurons based on them. 1) Visual cells have
visual activity only and do not discharge before
saccades made without visual guidance. 2)
Movement cells have predominantly
movement activity; they discharge comparably before purposive saccades made with or without
visual guidance but have little response to visual stimuli alone. 3) Visuomovement
cells
have both visual and movement activity and
discharge best before visually guided saccades.
Both movement and visuomovement cells often show anticipatory activity in a repetitive
saccade task.
OF FRONTAL
EYE FIELDS
715
In the present study we first analyzed the
activity of single neurons, classifying their activity and determining their response fields.
We then electrically stimulated through the
recording microelectrode to examine whether
the properties of neurons were correlated with
the eye movements evoked by low-level electrical stimulation at that site. We also used
electrical stimulation to map the anatomical
location of the low-threshold frontal eye fields.
Preliminary reports of these experiments have
been published elsewhere (19, 22).
METHODS
The activity of singleneuronsand the effectsof
electricalstimulation through the recordingmicroelectrodewere studied in awake monkeystrained
to perform severalvisuomotor tasks for a liquid
reward. An on-line digital computer (PDP-1l/34)
wasusedto monitor the monkey’sbehavior,control
stimulusand rewardpresentation,measureeyepositionand singleneuronactivity, computeperievent
rastersand histograms,and storeneuronalandeyeposition information for subsequentanalysisoffline. Most of thesemethods have been fully describedelsewhere(8, 18).
Monkeys weretrained on four basicvisuomotor
tasks: 1) In the no-saccade(or fixation) task the
monkeysgazedat a stationary spotof light and did
not make eyemovementswhen a secondspot was
flashedon the screen.This task wasusedto study
the visual propertiesof neurons.It wasalsousedto
test microstimulationduring visual fixation by delivering current in lieu of the secondspot. 2) In the
visually guidedsaccadetask the fixation point disappearedwhen the target light went on, and the
monkey madea saccadeto fixate the target light.
Generally this wasthe initial task usedwhen a cell
wasfirst isolated,and it alsowasthe principal task
for mappingresponsefields.3) In the stabletarget
saccadetaskthe peripheralspotwasalwayspresent,
but the monkey madea saccadeto it only whenthe
fixation point disappeared.This task was usedto
dissociate
phasicvisualactivity from activity related
to visuallyguidedsaccades.
4) In the learnedsaccade
taskthe monkey madea purposivesaccadewithout
visual guidance.The monkey made saccades
to a
briefly flashedtargetthat appearedwhenthe fixation
point went out; however, the flashedtarget was
omitted on randomly selectedtrials. The monkey
wasrewardedfor makingthe samesaccadeon those
trials without a visual target. This taskwasusedto
dissociatevisual activity from movement activity.
Subjectsfor the single-cellrecordingexperiments
werethree adult rhesusmonkeys(Macaca mdatta)
and onecynomolgusmonkey(Macaca fascicularis).
The rhesusmonkeyswere highly trained and used
716
BRUCE,
GOLDBERG,
BUSHNELL,
AND
STANTON
each stimulation test. One record contained highfor several months in experiments directed at analyzing both neuronal activity and electrically elicprecision samples of horizontal and vertical eye poited eye movements (8). The cynomolgus monkey
sition taken every millisecond for 500 ms following
the beginning of the stimulation. The other record
was minimally trained and studied for only 1 mo.
contained lower-precision
samples of the eye poIn three additional untrained cynomolgus monkeys
we mapped the frontal eye fields with electrical
sition taken every 4 ms over a 2 s epoch surrounding
stimulation without studying neuronal activity ex- the stimulation. One bit represented 0.0825” in the
high-precision
record and 0.33” in the lower-precept to judge if the electrode were in gray or white
cision one. When very small eye movements were
matter.
Conventional techniques for recording extracelstudied, the gain was increased by a factor of two
or five.
lular neural activity with glass-coated platinum-irWe recorded from each rhesus monkey over a
idium microelectrodes
were used. The electrode
signal went to a BAK DIS-1 window discriminator,
period of several months. A few weeks before terand the resultant unit pulses were sampled at a fremination of the experiments we began making small
quency of 1 Khz by the computer. Perievent hiselectrolytic marking lesions (20 PA, electrode negative, 30 s) at sites with low thresholds for eliciting
tograms were calculated on-line and stored on
saccades. The cynomolgus monkeys were studied
magnetic media. The data were used to construct
plots of neuronal discharge frequency versus saccade
for shorter periods of time, and electrolytic lesions
amplitude and direction.
were made at all low-threshold
(~50 PA) sites.
Monkeys were anesthetized with a large dose of
After analyzing the response properties of a single
neuron, we studied the effect of electrical stimulapentobarbital
and perfused with 0.85% NaCl foltion at the site of that neuron by microstimulation
lowed by 10% formalin in 1% NaCl. Sections, taken
to the arcuate sulthrough the recording microelectrode.
Trains of either coronally or perpendicular
cus at its posterior limit, were stained with cresyl
pulse pairs were programmed on a Grass S88 stimviolet, thionine, or Merker’s silver stain (42). Markulator. Each pulse was 0.25 ms in duration, stimulation frequency was 350 Hz, and the standard
ing lesions were located on lateral reconstructions
of the periarcuate region and transferred to photrain duration was 70 ms. The stimulator outputs
tographs of this region.
were connected to a pair of constant-current
stimulus isolation units (Grass PSIUQ, which were
linked together to provide biphasic (negative-posiRESULTS
tive) pulses. Current was monitored by the voltage
drop across a 1 KQ resistor in series between the
Relationship between electrically elicited
output of the isolation units and the microelectrode.
saccades and cell type
Currents were read as half peak-to-peak values, and
At 366 frontal eye field sites we studied an
the S88 controls were adjusted to equate positive
isolated cell and then microstimulated through
and negative current. Currents tested ranged from
5 to 150pA.
the recording electrode. Using several behavWe electrically stimulated the monkey while it
ioral paradigms, we classified these cells into
was alert but not performing a task, and took care
eight categories as listed in Table 1. These catto stimulate while the monkey was looking roughly
egories were put into three more general
at the center of the screen. Threshold, defined as groups based on the presence or absence of
the lowest current yielding eye movements on half
presaccadic movement activity. The 71 cells
the applications, was determined, and representative
in
the first group all had presaccadic moveeye movements were stored. Most records were
ment activity. Movement cells (n = 19) distaken using suprathreshold
currents sufficient to
charged briskly before saccades, even learned
evoke saccades on almost every trial. This was ususaccades made without visual guidance, and
ally twice the threshold level. The saccade dimenhad little or no response to visual stimuli given
sions for each site were taken as the median amplitude and direction of the elicited saccades in these
in the no-saccade task. Visuomovement
cells
stored records. Most sites yielding saccades were
(n = 52) also discharged before learned sacfurther tested using a paradigm that applied micades but discharged better before visually
crostimulation
while the monkey was fixating a guided saccades and also responded to visual
small spot of light at the center of the screen. Difstimuli in the no-saccade task.
ferential effects of these two stimulation conditions
The 2 13 cells in the second group all lacked
are reported elsewhere (20).
presaccadic
movement activity. Visual cells
Eye movements were recorded with a search coil
(n = 30) discharged in response to visual stim(50) surgically implanted subconjunctivally
in one
uli, whether or not the stimuli were saccade
eye (33). The computer system triggered the stimtargets, and did not discharge before learned
ulation train and collected two digital records of
ELECTRICAL
TABLE
Relationship
1.
STIMULATION
OF
FRONTAL
EYE
717
FIELDS.
of cell type to threshold
Current Threshold
Cell Type
40 /LA
50- 150 PIA
34
11
45
7
6
13
11
2
13
52
19
71
3
4
3
6
15
9
17
21
20
55
59
30
36
68
79
Total without movement activity
11
47
155
213
Ambiguous cells
Uncharacterized presaccadic
Pre-post
Total ambiguous
32
3
35
20
21
21
5
26
73
9
82
Total all cells
91
81
194
366
Cells with movement activity
Visuomovement
Movement
Total with movement activity
Cells without movement activity
Visual
Postsaccadic
Not saccade-related
Not driven
saccades in the dark. Postsaccadic cells (n =
36) discharged after or during saccades but not
before them. Cells not related to saccades (n =
68) responded to fovea1 visual stimuli, auditory
stimuli, or in association with smooth pursuit
eye movements or particular directions of
gaze. Also included in this group were 79 neurons we could not drive.
For the third group (n = 82) the presence
of movement activity was ambiguous. Uncharacterized presaccadic cells (n = 73) discharged before visually guided saccades but
were not further characterized; presumably
some but not all of these cells had movement
activity. Pre-post cells (n = 9) discharged before visually guided saccades-in one direction
but also discharged after saccades in roughly
the opposite direction.
After studying each neuron we observed eye
movements elicited by microstimulation
at the
recording site. Saccades were elicited by currents as low as 10 PA. To analyze the relationship between cell type and electrically
elicited saccades, we classified each stimulation
site according to its threshold. Sites with
thresholds 150 PA were designated as low
threshold, sites with thresholds between 50 and
150 PA, as intermediate threshold, and sites
that required currents > 150 PA, or from which
saccades could not be elicited, as high threshold. Most high-threshold sites were ones at
>150 PA
Total
which no saccades were evoked because we
rarely used currents > 150 PA in order to minimize tissue and electrode damage.
Table 1 lists the number of cells of each
type recorded at low-, intermediate-, and highthreshold sites. Movement
activity was
strongly associated with low thresholds; 63%
(45/7 1) of visuomovement
and movement
cells were at low-threshold sites, and only 18%
were at high-threshold sites. Conversely, the
absence of movement activity was associated
with high thresholds; only 5% (1 l/2 13) of the
cells lacking movement activity were at lowthreshold sites, whereas 73% ( 155/2 13) were
at high-threshold sites. Uncharacterized presaccadic cells were intermediate; 43% (35/82)
were at low-threshold sites, and 3 1% (26/82)
were at high-threshold sites. This suggests that
the uncharacterized presaccadic neurons were
a mixture of cells with visual and movement
activity.
Figure 1 shows an example of eye movements elicited by electrical stimulation at a
low-threshold site and the activity of a visuomovement cell recorded at that site during
different tasks. The cell discharged briskly on
the visually guided saccade task (top) and responded weakly when the same visual stimulus
was given in the fixation task (middle-right).
These activities are not necessarily indicative
of a visuomovement cell as visual cells with
718
BRUCE, GOLDBERG,
BUSHNELL,
stable trgt
.
.
Sl
.
AND STANTON
fixation
..--
.
.
m
.
.
.
.
.
.
”
.
.
.
I
13-491
Electrical
Stimulation
3 10”
- I,.,....IIII,,~,,~~~~
I. ..,,,,I,.., ..I..,,.,,,
2oIC;S
- I1,,,.a,,.,,,,.,,., ,,,,I
1..,,I,,,,,,,.,,,,b
FIG. 1. Activity of a visuomovement neuron during different oculomotor tasks (top two rows) and saccades evoked
by low-level microstimulation (50 PA) at the site of this cell (bottom TOW).H labels the horizontal eye position trace,
V, the vertical, Sl, the fixation point (up - on; down - off), and S2, the saccade target. Neuron discharges are shown
as dots between the eye and stimulus traces. The top two records show the standard saccade task: Sl disappears, and
S2 appears simultaneously, and the monkey makes a saccade recognized by the computer as appropriate and marked
by ticks on the vertical eye trace. In the top left record, the saccade had a short latency and in the top right, a much
longer latency. The stable target saccade task is shown in the left middle record: the target (S2) was continuously
present, and the monkey saccaded to it in response to the disappearance of the fixation light (3). The right middle
record shows the fixation task: Sl did not disappear, and no saccade was made. The bottom records show the elicited
saccade at the site of this neuron. The duration of electrical stimulation is marked by a dark bar beneath the eye traces.
enhancement may have similar responses (8,
21, 72). However, the visuomovement
cell
burst immediately
before saccades made to
stable targets (middle-left). Movement cells
also respond before saccades made in this paradigm, but purely visual cells either do not
respond or respond tonically whenever the
target lies in their receptive field and do not
change their discharge rate before the saccade
(8). Therefore, the stable target saccade task
distinguishes visuomovement and movement
cell types from purely visual ones, and hence
is very diagnostic for low-threshold sites.
High thresholds were associated with a diverse grouping of cells lacking presaccadic
movement activity. Two cell types in this
group are of special interest because their responses are nonetheless related to saccades.
Visual cells are active before visually guided
saccades, yet they lack movement activity as
determined by other tasks. Only 10% of visual
cells (3/30) were at low-threshold sites, whereas
70% (2 l/30) were at high-threshold sites. Cells
with postsaccadic activity are also important
as they discharge in conjunction with particular sets of saccades, and it has been hypothesized that electrical stimulation of the frontal
eye fields evokes saccades by antidromically
activating afferents to such cells (5 1). However,
thresholds at postsaccadic sites did not support
this hypothesis. Only 3% of 36 sites with purely
postsaccadic activity had low thresholds,
whereas 56% had high thresholds (Table 2).
Relationship between electrically elicited
saccades and movement fields
of single neurons
We studied each cell with presaccadic activity using the visually guided saccade task to
determine its visual receptive field or movement field. We varied the target location,
thereby eliciting visually guided saccades of
ELECTRICAL
STIMULATION
OF FRONTAL
different directions and sizes, and found the
“optimal”
dimensions associated with the
greatest presaccadic discharge. The dimensions
we had judged to be optimal usually matched
the dimensions
of saccades subsequently
evoked at that recording site. This relationship
is illustrated in Fig. 2. The top row shows saccades elicited by microstimulation,
and the
middle row shows activity of the cell recorded
at that site while the monkey made visually
guided saccades using the target location we
had judged to be optimal for that cell. Note
that the saccades elicited by microstimulation
are quite similar, in both size and amplitude,
to those elicited in the saccade task by the optimal target placement. The bottom row confirms that the cell was much less responsive
when the target location, and hence the saccade direction, differed from the optimal.
We quantitatively
investigated this relationship by making tuning curves of cell ac-
719
tivity as a function of direction and amplitude.
We compiled histogram sets for target locations systematically differing in either direction
or eccentricity. Plots of cell discharge as functions of saccade amplitude or direction were
later fit with Gaussian curves, and the parameters corresponding to the peaks of the curves
were taken as the optimal direction or amplitude for the cell (8). Figure 3 shows four pairs
of plots and curves with the median dimensions of electrically elicited saccades indicated
for comparison. On the right are representative
examples of both the optimal saccades with
associated cell discharge and the electrically
elicited saccades. As with the sites mapped
qualitatively, the elicited saccade dimensions
usually matched the optimal saccade as obtained from the tuning curves. Complete direction tuning curves were made for 16 cells
at whose sites we subsequently elicited saccades with microstimulation.
Figure 4A shows
Al
A2
A3
Bl
82
83
/
Hi
VI
4
.-
Sl
s2
EYE FIELDS
,
..
I,
.-I
II
-.‘J
I
I1
.
-.- -
. .
.
I
------A
I
I
I
I
I,
I
J I
I
*
I
I
I,,
I
I
I
..w.-I
I
I
*
. . ..
*
T I.
x.
12-374
FIG. 2. Relationship of a frontal eye field neuron’s movement field to saccade evoked by electrical stimulation. Al,
A2, and A3 show 3 different saccades evoked by electrical stimulation at a single site in the frontal eye fields. H and V
denote horizontal and vertical eye position, and the ticks on the vertical eye trace indicate the beginning and end of
saccades, as recognized by the computer. The electrical stimulation occurred at the artifact line in the stimulation
trace. Note that the eye movement was nearly the same in all 3 records, even though the eye position at the time of
stimulation differed considerably. Rows B and C show activity of the neuron studied at this stimulation site. Bl, B2,
and B3 show activity during the visually guided saccade judged optimal for the cell. Note that the electrically elicited
saccades in A resemble this saccade. Each dot below the eye movement trace represents a neuronal discharge. Sl
denotes the fixation point, and S2 denotes the peripheral stimulus to which the monkey made a saccade. When the
associated trace is up, the stimulus was on; when it is down, the stimulus was off. Cl shows a much weaker response
to the same peripheral stimulus during a fixation task. C2 and C3 show weak (C2) or nonexistent (C3) responses to
saccades other than that shown in row B.
720
BRUCE, GOLDBERG,
BUSHNELL, AND STANTON
8,
1
8:
13-4e1
---0
-iL-
“+-
8:
$
.
.
,-no-.
.
/
52:
Ri*
.
0,
/
/6
0
1..
.,....,
0
5
.I.,,...
.-,.,....,....
10
15
20
. . . .
25
30
,_
35
40
45
AMPLITUDE
10
15
20
25
30
35
40
is
AMPLITUDE
---
,‘...l..“,....,...,,....,
10
15
20
25
30
35
40
45
AMPLITUDE
,....I....,...-,....,....
0
DIRECTION
5
1Q
15
20
_... ..
25
30
35
I‘“.,
90
45
AMPLITUDE
FIG. 3. Correlation of movement fields of neurons with saccades evoked by electrical stimulation at the site of the
neurons. The left sraph in each row shows the relationship between cell discharge and direction of eye movement at
the optimum amplitude for each cell. Direction is in degrees of arc, with 0 representing horizontal rightward by
convention. The vertical line through the graph shows the median direction of electrically elicited saccades at the site
of the cell. The right graph in each row shows the relationship between cell discharge and saccade amplitude at the
optimum direction, with the vertical line indicating the median amplitude of electrically elicited saccades. The top,
right records show visually guided saccades with horizontal and vertical eye traces and cell discharges (dots below).
Saccade recognition ticks are on the vertical eye traces. Saccades evoked by electrical stimulation are shown directly
beneath, with the duration of stimulation marked.
ELECTRICAL
STIMULATION
the relationship between the optimum saccade
directions determined by tuning curves and
the median saccade evoked by suprathreshold
electrical stimulation. The correlation between
optimal saccade and evoked saccade direction
was +0.93. Complete amplitude tuning curves
were made for 15 cells at successful stimulation
sites. Figure 4B shows the relationship between
the optimal saccade amplitude determined by
the tuning curves and the median, electrically
evoked saccade amplitude. Amplitude
was
plotted with logarithmic axes because saccade
amplitude tuning in the frontal eye fields, like
visual eccentricity in visual cortical areas, appears to be logarithmically
scaled (59). The
correlation between optimal and elicited amplitudes, both logarithmically
scaled, was
+0.8 1.
Although the evoked saccade dimensions
and saccade dimensions optimal for presaccadic activity generally were similar, there were
exceptions and qualifications to this relationship. First, there was a systematic trend involving cells that preferred fairly vertical saccades, either up or down. Stimulation at the
sites of such cells usually evoked saccades
somewhat more contralateral (i.e., rotated
away from vertical) than the cell’s optimal
saccade. In the bottom example of Fig. 3 the
cell’s best response preceded vertical (upward)
saccades, but elicited saccades from that site
were oblique. Second, the saccades evoked by
electrical stimulation tended to be shorter than
the optimal saccade amplitude for the cells.
This was particularly evident at sites with neurons having large, eccentric movement fields
(see Fig. 4B). Nevertheless, the saccade evoked
by electrical stimulation was one before which
the cell at that site had discharged vigorously.
Finally, there were some instances of poor
correspondence between evoked and optimal
saccades besides the two special cases just discussed. Such mismatches were seen, whether
the cell’s optimal saccade was judged qualitatively or quantitatively, as evidenced by discrepant points in Fig. 4. Mismatches were
more likely at higher-threshold sites and at sites
mapped with visual cells.
Relationship of elicited saccade dimensions
and postsaccadic movement fields
We also stimulated at the sites of cells for
which we had mapped the movement field for
OF FRONTAL
EYE FIELDS
721
z2d
g
P
5
5
E
5QDo _
z
00
$
z
3
6
/
n
V
I
Neuron
Optimum
Direction
(Deg)
1
Neuron
Optimum
Amplitude
(Des)
FIG. 4. Relationship between amplitude and direction
of the saccade associated with the greatest neural activity
for quantitatively studied cells and saccades evoked by
electrical stimulation at site of those cells. A: the scatter
plot of median evoked saccade direction to the neural optimal direction as computed from the amplitude tuning
curve for each cell. Dotted line, ideal x = y function; unbr&en line, regression line calculated by the method of
least squares. B: the scatter plot of median evoked saccade
amplitude versus the optimal saccade amplitude for neuronal activity. Both axes are logarithmic. Both correlation
coefficients (r) are significant at P < 0.001.
722
BRUCE,
GOLDBERG,
BUSHNELL,
AND
STANTON
postsaccadic activity. The results did not sup- ments with both types of saccade-related acport the hypothesis that electrical stimulation
tivity. The optimal movements for each of the
of the frontal eye fields evokes saccades by an- component activities of these cells are usually
tidromically
activating the afferents of cells in opposite directions (8). The postsaccadic
discharge is obligate, occurring after all sacwith postsaccadic activity. This antidromic
hypothesis would predict that the evoked sac- cades of the right dimensions, whereas the
cade should match the postsaccadic movepresaccadic component is conditional and is
ment field of nearby cells, but this was not the absent or minimal in conjunction with sponcase. When saccades were electrically elicited
taneous saccades made in the dark (8). As ilat such sites, their dimensions usually did not lustrated in Fig. 5, the dimensions of elicited
agree with the movement field for postsaccadic saccades from the sites of such cells were found
activity; in fact, many postsaccadic cells had to be in agreement with the presaccadic field
ipsilateral movement fields yet the elicited eye and opposite the postsaccadic movement field.
movements were contralateral.
Therefore, these combination
cells dramatiCells having both pre- and postsaccadic re- tally show that elicited eye movements in the
sponses allowed for a direct comparison of the frontal eye fields are related to presaccadic,
relationship of electrically elicited eye movebut not postsaccadic, neural activity.
PRE-SACCADIC
DIRECTION
(trigger line = saccade
POST-SACCADIC
DIRECTION
(trigger line = saccade
start)
end)
Visually
Guided
Saccades
Spontaneous
Saccades
in Dark
Electrically
Elicited
Saccades
I
,-
--d--- -
/’
__1_1_
] 10”
13-481
FIG. 5. Comparison of neuronal activity and the saccades evoked by electrical stimulation near a cell with pre- and
postsaccadic activity. The left column shows activity for saccades in the presaccadic optimum direction, the rig&
column for saccades in the optimal postsaccadic direction. Representative eye-position traces are shown above, with
the computer saccade recognition ticks on the vertical trace. In the rasters, each dot is a cell discharge, and successive
epochs of cell activity are all synchronized on the onset of the saccade. The histograms sum the activity in the rasters
and the histogram calibration line signifies a discharge rate of 100 Hz. Eye movements evoked by electrical stimulation
at the recording site are shown at the bottom. Note that the electrically evoked saccades are in the presaccadic direction.
ELECTRICAL
STIMULATION
Saccade Iatencies
Saccade latencies were measured from the
onset of the microstimulation
train until the
beginning of the evoked saccade. Each site had
a consistent, characteristic latency, and site latencies, based on a median of 10 or more successful trials, ranged from 20 to 60 ms, with
30-45 ms being typical. There was no obvious
basis for this variation: Short and long latencies
were found for sites with large and small saccades, sites with low and high thresholds, sites
in different monkeys, and even at different
sites within individual electrode penetrations.
Likewise, although latencies were somewhat
longer and more variable at threshold, further
increases in suprathreshold current had little
effect. We should emphasize that these observations were made using currents 5 150 PA.
Using much higher currents, Robinson and
Fuchs (54) found that frontal eye field saccades
had uniformly short latencies of 15-25 ms.
Relationship of threshold to behavioral state
Low saccade thresholds were critically dependent on the state of the monkeys. A lack
of alertness raised thresholds for eliciting saccadic eye movements with microstimulation
and elicited saccades suffered from the same
slowness and postsaccadic drift that characterized spontaneous saccades when the monkeys were not alert. Waking the monkeys with
noise or rewards eliminated these effects, at
least temporarily. Thresholds also were raised
when the animals were tested under fixation,
as described elsewhere (20).
Anatomical location of the frontal eyeJields
We used electrical stimulation to study the
extent and topography of the frontal eye fields,
both in the untrained monkeys, used solely
for such mapping experiments, and in trained
monkeys, which were primarily used to study
neuronal activity. The emphasis on stimulation in the untrained monkeys enabled us to
proceed more rapidly and place more marking
lesions than we could in the trained monkeys.
We designated cortex where saccades were
elicited with currents 5 50 PA as the lowthreshold frontal eye fields and placed most
electrolytic marking lesions at low-threshold
sites. In all monkeys this low-threshold area
lay along the posterior portion of the arcuate
sulcus. We saw no difference in frontal eyefield location between Macaca mulatta and
Macaca fascicularis, although the exact loca-
OF FRONTAL
EYE FIELDS
723
tion relative to landmarks such as the principal
sulcus varied among monkeys. Figure 6 shows
the location of the low-threshold frontal eye
fields in five hemispheres. These surface views
do not portray the full expanse of the lowthreshold region as it is largely buried in the
anterior bank of the arcuate sulcus and extends
only a few millimeters onto the surface of the
prearcuate gyrus. Based on its approximate
extent along the arcuate (10 mm) and extension into this sulcus (8 mm), the low-threshold
frontal eye fields encompass -80 mm2 of cerebral cortex in each hemisphere.
The frontal eye fields defined by the lowthreshold criterion are considerably smaller
than Brodmann’s cytoarchitectonic
area 8.
Examination of stained sections (cresyl violet,
thionin, and Merker’s silver Nissl stain) indicated that the frontal eye fields better correspond to the union of Walker’s area 8A and
area 45 (68). Both of these areas are characterized by clusters of very large pyramidal cells
in layer V, and area 45 is further distinguished
by large pyramids in layer III as well. Both
areas have a dysgranular layer IV, which becomes agranular at their posterior border in
RIGHT
LEFTVIEVERSED)
Ml7
1 CM
FIG. 6.
Location of the frontal eye fields. The lowthreshold frontal eye fields in 5 different hemispheres of
3 cynomolgus monkeys. The surface projection of the arcuate sulcus in which low-threshold saccades were found
in the anterior bank is shown as the darkened line between
two arrows. Each hemisphere is shown in a view looking
directly dotin on the principal sulcus, with the central
sulcus shown as the far-left landmark of each drawing,
and the front of the brain on the right.
724
BRUCE, GOLDBERG,
BUSHNELL,
the arcuate fundus and strongly granular at
their anterior border on the prearcuate gyrus.
Figure 7 shows the cytoarchitecture of areas
8A and 45 in cresyl material using sections
with low-threshold marking lesions. In agreement with the overall topography described
AND STANTON
below, saccades elicited from area 8A were
generally large, whereas saccades from area 45
were generally small.
The views shown in Figs. 6,7,8, and 9 were
taken normal to the surface of the prearcuate
gyrus. This view portrays the shape of the ar-
Ill
IV
lmm
v
FIG. 7. Photomicrographs of the frontal eye fields showing the cytoarchitecture of Walker’s areas 8A and 45. The
line drawing insert shows the location, extent, and orientation of the 2 sections. The micrographs, right, are enlargements
of the rectangles delineated on the let%micrographs. Micrographs, top, show area 45 in a section near the ventral border
of the frontal eye fields. In the low-magnification photograph, notice the strongly granular cortex on the top of the
brain, which becomes dysgranular near the sulcal lip. In the high-magnification photograph, notice the clusters of large
cells in both layers III and V. Immediately above the region selected for enlargement are part of 2 parallel electrode
tracks, each with a pair of microlesions marking low thresholds for elicited saccades, which were small in amplitude.
Set of micrographs, bottom, shows area 8A from a section near the dorsal border of the frontal eye fields. In the highmagnification photograph, notice that the clusters of large cells are mostly confined to layer V. Large saccades were
elicited along the tracks visible in the low-magnification section.
ELECTRICAL
STIMULATION
OF FRONTAL
EYE FIELDS
725
FIG. 8. Eye movements evoked by electrical stimulation at different sites showing the increase in saccade amplitude
with dorsal-to-medial movement across the frontal eye fields. Several eye movements evoked while the monkey fixated
near the center of the screen are superimposed for each site and are shown in a 2-dimensional coordinate system whose
origin is the fixation point. Eye position was sampled at 1 KHz, and the raw sampled values are shown as dots without
interpolation. Each set of saccades is connected by a line to the square on the surface of the brain marking the entry
point of the penetration in which the saccades were evoked. The median saccade amplitude is written on the line; note
that the scales for the different sets of saccades are adjusted from 15O per tick for the largest saccades (upper right) to
3’ per tick for the shortest saccades (lower right).
cuate sulcus veridically and emphasizes the
broad posterior portion that contains the
frontal eye fields. In some monkeys this portion of the arcuate sulcus is gently rounded,
as a “(“. In other monkeys it is more squared,
as a “[“. In contrast, the commonly used lateral view of the macaque brain offers a sharply
angled perspective of the arcuate region, which
artificially foreshortens this posterior portion
and often causes the arcuate sulcus to appear
as a “<” pointing posteriorly. Based on its appearance in the lateral view, the arcuate sulcus
has been exhaustively divided into superior
and inferior limbs. We think it is misleading
to ascribe the frontal eye fields to either of these
lrm bs.
BRUCE, GOLDBERG,
I
I
I
MILLIMETERS
I
I
1
BUSHNELL,
AND STANTON
-90”
- 1
I
I
1
1
MILLIMETERS
15-19
FIG. 9. Systematic changes in the direction of eye movements evoked by electrical stimulation over long electrode
penetrations through the frontal eye fields. A and B each show the angle of saccade evoked by electrical stimulation at
different sites along a single electrode penetration. In the top graphs, the ordinate is saccade direction, and the abscissa
is distance along the electrode track. In the brain section tracings the direction is indicated by the slant of the arrows
and marking lesions by dots on the penetration line. Saccades could not be evoked from the 2 sites labeled X in
penetration B. The center drawing is a view normal to the cortex in the vicinity of the frontal eye fields with the 2
electrode penetrations marked. CS, central sulcus, PS, principal sulcus.
Topography of the frontal eye fields
45. There was also a tendency within individual penetrations that larger saccades were elicNearly any saccade can be elicited by stimulation at some site in the frontal eye fields. ited near the lip of the sulcus and smaller ones
near the fundus.
Saccade amplitudes ranged from < 1Oto >30”,
In contrast to the organization of saccade
and all directions of saccades were represented
with microstimulation
in each frontal eye field amplitude, there was no overall topographic
yielding saccades into the contralateral field. mapping of saccade direction. Instead, direcThe typical elicited saccade was oblique and tion was well organized at a local level in that
no particular direction, such as along the plane the direction of elicited saccades systematically
shifted as the electrode was advanced small
of action of the eye muscles, was dominant.
Saccade amplitude was topographically or- distances within individual penetrations. This
ganized. The largest saccades were elicited
shift of saccade direction was most evident in
from the most dorsolateral portion of the penetrations that followed the rostra1 bank of
frontal eye fields, and the smallest saccades the arcuate, remaining in the low-threshold
frontal eye fields for several millimeters. Figure
were elicited from its ventrolateral portion.
Although this topography was not always ev- 9 shows direction progressions for two penetrations that yielded low-threshold, elicited
ident from comparisons of saccades elicited
from closely spaced penetrations, it was always saccades for nearly the entire depth of the arobserved in an extended series of penetrations
cuate sulcus. Although the direction of the first
along the rostra1 lip of the arcuate, such as saccade elicited in a penetration seemed unthose shown in Fig. 8. This topographical or- predictable (see Fig. 8), thereafter saccade direction varied continuously with small shifts
ganization dictates that saccades elicited from
area 8A are larger than those elicited from area generally being evident after electrode move-
ELECTRICAL
STIMULATION
ments of 0.25 or 0.5 mm. The direction of
change usually held for several millimeters but
often reversed on reaching a vertical direction
(up or down). Most long penetrations showed
at least one reversal in direction of change;
consequently, many penetrations encountered
large sets of directions more than once.
Most lesions marking low-threshold sites
were found to lie in infragranular laminae, and
electrode penetrations with low thresholds for
an extended distance usually were found to
have traversed the infragranular cortical laminae (see Fig. 9). At many sites we retested
microstimulation
after making the electrolytic
lesion. Threshold was invariably raised, and
often no eye movement could be elicited by
150-PA currents at what had been a lowthreshold site (40 PA). Subsequent electrode
movements of 0.25-0.5 mm usually restored
low-threshold, elicited saccades, as well as
bringing back a normal level of background
neuronal activity.
Slow eye movements and other
evoked behaviors
Microstimulation
elicited slow, smooth eye
movements at 21 sites. These movements
lacked the all-or-none property of elicited saccades; they continued for the duration of
stimulation (up to 2 s) without evidence of a
staircase. Furthermore, eye velocity increased
Visually
Guided’
Smooth
Pursuit
OF FRONTAL
EYE FIELDS
727
with increases in current (up to 150 PA). These
movements cannot be characterized as either
slow saccades or nystagmus produced by a lack
of alertness because they were elicited both
during alert spontaneous behavior and even
during visual-fixation paradigms. Nor were
they vergence movements; although a search
coil was present in only one eye, it was clear
from direct observation that these movements
were conjugate. Finally, differentiation of the
eye-position trace showed that the eye velocity
was continuously low, and hence the movement was not a stream of minute saccades undetectable in the position domain.
We think these slow movements are artificial elicitations of smooth pursuit eye movements that enable monkeys to track moving
visual targets. Neuronal activity supported this
hypothesis; at 12 of these sites, cells discharged
preferentially during smooth pursuit of a small
target moving in particular directions. Figure
10 shows the activity of a cell during visually
guided pursuit eye movements and the slow
eye movements subsequently elicited by electrical stimulation at that site. Smooth movement sites were very deep within the arcuate
sulcus; however, small saccadic eye movements were also elicited at, or near most of,
these sites. Therefore, it is not clear whether
slow movements are interspersed within the
saccadic frontal eye fields or if they comprise
V-
. . . . .. ..... . .. .. ........ . .. .... ..
Stimulation
H
(50 pa)
v
.
.
.
. .
.
.
0.
Stimulation
(100 pa)
17-025
loo1
lOi&
FIG. 10. Smooth pursuit eye movements evoked by electrical stimulation in the frontal eye fields. The top pair of
records show horizontal (H) and vertical (V) eye-position traces and the neuron discharges for that movement (dots
below). The cell discharged more during upward-to-left pursuit than during pursuit in other directions. The left record
shows activity in this direction, and the right diagram shows activity associated with pursuit down and to the right.
The middle trace shows the smooth eye movements elicited by stimulation (50 PA current, 500 ms duration) through
the recording microelectrode at the site of this cell. Stimulation is i ndicated by the dark line beneath the eye-position
traces. The bottom traces show higher velocity movements evoked by increasing the current to 100 /LA.
728
BRUCE,
GOLDBERG,
BUSHNELL,
a contiguous, but separate, area adjacent to
the small saccade region.
Stimulation in the posterior bank of the arcuate sulcus did not evoke eye movements.
Instead, arm and hand movements, mouth
movements, or blinks were often evoked from
this cortex. Stimulation anterior to the frontal
eye fields usually evoked saccades at high
thresholds or failed to elicit any overt behavior.
AND
STANTON
ison with thresholds near neurons with presaccadic activity. Furthermore, postsaccadic
movement fields did not match the elicited
eye movements. For example, many postsaccadic neurons were selective for ipsilaterally
directed eye movements, whereas the direction
of both presaccadic response fields and elicited
saccades were nearly always contralateral. For
cells with both pre- and postsaccadic activity,
the elicited saccade went into the presaccadic
movement field. We therefore suggest that the
DISCUSSION
antidromic hypothesis is untenable. Because
of the close relationship between presaccadic
Relationship between elicited saccades and
activity and elicited eye movements, it is also
neuronal activity
unnecessary. Our data indicate that electrically
These results indicate that saccades elicited
evoked saccades reflect the activation of preby electrical stimulation
of the frontal eye saccadic cells in the frontal eye fields and orfields reflect a functional role of this cortex in thodromic conduction to their targets in the
triggering saccadic eye movements, as origisuperior colliculus and brain stem preocunally postulated by Ferrier, Holmes, and oth- lomotor centers.
ers (12, 15, 16,26,27,62).
The lowest threshLike cells with postsaccadic activity, cells
olds for eliciting saccades were found near with visual activity alone were rarely found at
neurons that discharged before saccades, in low-threshold sites. This implies that frontal
particular, visuomovement
and movement
eye-field visual activity is not directly relayed
cells. Moreover, the dimensions of saccades to the subcortical oculomotor targets of the
evoked by electrical stimulation matched the frontal eye fields. Instead, we think that this
movement fields of the cells at the site stimvisual activity is processed within the frontal
ulated. We postulate that activity of visuo- eye fields and has the potential to be transmovement and movement cells constitutes the formed into movement activity destined for
frontal eye-field signal that triggers particular
subcortical oculomotor centers. Recent work
saccades and that electrical stimulation of the from this laboratory directly supports this hyfrontal eye fields elicits saccades of the same pothesis. Visuomovement
and movement
direction and amplitude by exciting these cells, but not visual cells, can be activated anneurons and their efferent processes.
tidromically
by electrical stimulation
of the
Contrasted with thresholds near visuo- intermediate
layers of the superior collicumovement and movement cells, thresholds
lus (60).
near visual cells, postsaccadic cells, and other
Lesions marking low thresholds for micell types lacking presaccadic movement ac- crostimulation
saccades were usually located
tivity were much higher. It is significant that in infragranular laminae. Cortical projections
low thresholds for electrically eliciting eye to the colliculus, thalamus, and brain stem
movements were not found at sites having
originate in the pyramidal cells of laminae V
postsaccadic neuronal activity. This was the and VI; therefore the low thresholds probably
first functional activity reported in the frontal
reflect the proximity of the electrode to these
eye fields (5,6) and has been postulated to be output cells. Because low thresholds were asa corollary discharge signal originating in sub- sociated with visuomovement and movement
cortical eye movement structures (5, 6, 66). cells, we infer that these cell types are preferBecause activity preceding saccades was not entially located in infragranular cortical layers
yet reported, it was hypothesized that the eye and that visual and postsaccadic cells are more
movements evoked by frontal eye-field stimconcentrated in supragranular laminae. More
ulation were caused by antidromic conduction
definitive experiments are needed to determine
over fibers carrying this corollary discharge to exactly how cell types are distributed across
the frontal eye fields (5 1). Our study argues laminae and which types project to the differagainst this antidromic hypothesis. Thresholds
ent subcorticai and cortical targets of the fronnear postsaccadic cells were high in compartal eye fields.
ELECTRICAL
STIMULATION
Although we have correlated the effects of
electrical stimulation with the properties of the
cell nearest the recording microelectrode, our
levels of microstimulation
obviously excited
more than this one cell. There are two ways
that the effects of electrical stimulation can
spread: I) By passive current spread or 2) by
active spread over neural processes. Workers
in this field (47) begin by analyzing passive
current spread. One model holds that the
threshold current needed to excite a cell body
or axon with monopolar stimulation obeys a
distance square law
i= a + kr2
where i is the threshold current needed to excite a neuron that is r microns away from the
electrode, and a and k are constants. Stony,
Thompson, and Asanuma (64) investigated
the parameters of this equation for the motor
cortex of the cat using glass-coated electrodes
similar to those used in this study. They calculated that single pulses of 10 PA excite cells
up to 110 pm distant, 50 PA excites cells up
to 300 pm away, and 100 PA, to 450 pm. We
accept these distances as estimates of our
sphere of effective passive current spread. Because electrolytic lesions, which were roughly
250 p in extent, raised thresholds for eliciting
saccades considerably, these estimates may be
large.
Active spread over neural axes, for example,
via cortical interneurons, is more difficult to
estimate; however, for the current levels used
here, passive current spread may predominate.
Cortical organization is predominantly
radial
(1 l), and except for layer I, the horizontal
spread of neural processes is usually ~300 p
(40). Therefore the horizontal spread of activation from currents 250 PA should largely
reflect passive current spread. Because lowthreshold sites were predominantly
confined
to infragranular laminae, it appears that active
spread is limited radially as well.
Therefore we estimate that microstimulation at 50 JLA, our criterion for the low-threshold frontal eye fields, excites a cortical neighborhood on the order of 300 pm in radius.
The close relationship we found between neuronal activity and elicited saccades indicates
that the one cell we happened to study was
usually representative of cells in the stimulated
ensemble. Use of larger stimulating currents
mav Preclude or weaken such a relationshin.
OF FRONTAL
EYE FIELDS
729
For example, much higher currents are required to elicit saccades in posterior parietal
cortex, and the correlation between elicited
saccade dimensions and receptive fields there
(6 1) appears to be much lower than it is in the
frontal eye fields.
Extent and topography of the frontal
eye fields
The frontal eye fields have been mapped
many times since their discovery over a century ago (9, 12, 15-17,54,62,67,72).
In some
reports, elicited eye movements were obtained
over large expanses of the frontal lobe; however, it is difficult to compare our results with
these previous studies because radically different techniques were used: for example,
surface stimulation,
macroelectrodes,
high
currents, anesthetized subjects, and informal
observations of eye movements. The lowthreshold frontal eye fields, which we define
as that region from which intracortical microstimulation at (50 PA evokes saccades, is
a more discrete region nestled in the posterior
aspect of the arcuate sulcus. It largely agrees
with Robinson and Fuchs’s map (54), which
also was based on intracortical stimulation in
awake subjects, except that the low-threshold
frontal eye fields are less extensive anteriorly
and are largely confined to the anterior bank
of the arcuate sulcus. These differences probably reflect the criteria employed. Robinson
and Fuchs (54) used currents up to 3 mA,
whereas our map is based on 50 PA thresholds.
Furthermore, we ignored sites subsequently
found to be in white matter, as the deep sulci
in this region make it uncertain where such
fibers originate.
These low-threshold frontal eye fields do not
extend into the strongly granular cortex of primate prefrontal cortex and hence do not encompass all of Brodmann’s area 8, which historically has been equated with the frontal eye
fields. They better correspond to the union of
Walker’s areas 8A and 45 (68), both of which
are characterized by large pyramidal cells in
layer V and a dysgranular layer IV. Thus the
rostra1 border of the low-threshold frontal eye
fields is at the transition to the granular cortex
of Walker’s area 46, where more intense stimulation is required to elicit saccades, It is unclear whether these higher-threshold saccades
are mediated by the subcortical projections of
area 46 (24. 38). short association moiections
730
BRUCE,
GOLDBERG,
BUSHNELL,
back to the frontal eye fields, by current spread
to frontal eye field efferents in the white matter
below area 46, or even by spread to the frontal
eye fields themselves or to the supplementary
oculomotor area recently described by Schlag
and Schlag-Rey (5 7).
The posterior boundary within the fundus
of the arcuate approximately coincides with
the transition to the agranular cortex of Brodmann’s area 6. Electrical stimulation in the
posterior bank of the arcuate produced hand,
face, or mouth movements. Cells in area 6 often respond to somesthetic stimuli, visual
stimuli near the monkey’s body (48, 49), or
visual stimuli that cue limb movements (69,
70). The frontal eye fields could be considered
the eye-movement specialization of this premotor region.
The pattern of larger saccades being elicited
from the most dorsomedial portion of the
frontal eye fields and the smaller ones from
the ventrolateral portion is consistent with the
mapping of Robinson and Fuchs (54) and also
parallels the topography of visual receptive
fields in granular cortex immediately anterior
to the frontal eye fields (65). Furthermore, this
amplitude topography accords with the afferent and efferent connections of the frontal eye
fields and adjacent prefrontal cortex. Several
anatomical studies (4,3 1,32) indicate that the
ventrolateral frontal eye fields receive cortical
afferents preferentially from visual areas with
central receptive fields (e.g., TE, TEO),
whereas visual areas with peripheral receptive
fields project preferentially to dorsomedial
frontal eye fields and neighboring prefrontal
cortex. We have recently found that tectal
projections of the frontal eye fields (63) further
reflect this frontal eye-field topography. The
dorsomedial frontal eye fields project preferentially to the caudomedial superior colliculus,
where large saccades are represented, whereas
the ventrolateral frontal eye fields project more
to the rostrolateral colliculus, where small saccades are represented (13, 23, 52, 7 1).
Saccade direction did not have an overall
topographic organization. In this respect the
frontal eye fields differ from the superior colliculi, where upwardly and downwardly
oblique saccades are segregated, with the representation of purely horizontal (contralateral)
saccades dividing the colliculus in half. In
contrast, upwardly oblique and downwardly
oblique saccades were interspersed throughout
AND
STANTON
the frontal eye fields. Suzuki and Azuma (65)
reported a similar indeterminancy for the direction of visual receptive fields in the prearcuate cortex, anterior to the frontal eye fields.
The point-to-line topography of visual fields
in the lateral suprasylvian gyrus of the cat (45)
is another example of such directional indeterminancy.
Saccade direction was well organized at a
more local level of observation. Small advances of the microelectrode produced small
shifts in the direction of elicited saccades, and
these small shifts added up to systematic progressions through a large range of directions,
with periodic reversals in the direction of
change. This organization of saccade direction
resembles the organization
of orientation
specificity in primary visual cortex, where all
orientations are systematically represented
over a set of columns, termed a hypercolumn,
and orientation hypercolumns are repeated
throughout the area (28-30). Specificity for the
direction of stimulus movement in prestriate
area MT is organized in a similar fashion (1).
However, the complete concept of a cortical
column also reflects the observation that cells
along a radial penetration are similar with respect to a physiological feature, e.g., orientation or ocular dominance in visual cortex. Because few of our penetrations traversed the
frontal eye-field cortex radially, we can only
infer that the frontal eye fields may have a
columnar organization analogous to those
found in visual (28, 29, 30), somatosensory
(44), and motor (2) cortices.
That direction is organized locally, but lacks
an overall topographic organization, may explain why some previous reports (9,54) found
a preponderance of horizontal (contralateral)
saccades and also why Robinson and Fuchs
(54) observed that saccades became more horizontal as stimulation current was increased.
If the saccade elicited from the frontal eye
fields reflects an average of all sites stimulated,
then large currents would stimulate over all
contralateral directions, and the average would
tend to be a horizontal saccade, with the upward and downward components canceling
each other. In contrast, a global topography,
such as in the superior colliculus, should have
larger neighborhoods
of similar cells and
therefore allow elicitation of oblique saccades
even with high currents.
A similar argument may explain why mi-
ELECTRICAL
STIMULATION
crostimulation
near cells responsive before
vertical saccades elicited movements slightly
more oblique than the optimal saccade for the
cell. Movement fields in the frontal eye fields
are quite broad, as indicated by tuning curves
for amplitude and direction both here and in
our previous report (8). Therefore, natural
saccades will be preceded by activity in a large
ensemble of frontal eye-field cells and for saccades having a significant vertical component,
this ensemble will include many cells in both
hemispheres. If unilateral microstimulation
does not effectively stimulate cells in the contralateral hemisphere, then it will not mimic
the bilateral activity that precedes natural vertical saccades, and the elicited saccades will be
oblique because the horizontal components
fail to cancel. Others have found that bilateral
stimulation of the frontal eye fields readily
elicits vertical eye movements (9).
Finally, we were surprised that stimulation
of some sites elicited slow eye movements that
were clearly distinct from the rapid saccades
typically obtained and resembled the smooth
pursuit eye movements that are used to track
moving visual targets. A small population of
neurons also discharged in association with
pursuit at these sites. The frontal eye fields
have not generally been thought to participate
in smooth pursuit, although there is a report
of patients having smooth pursuit disorders
following damage to the frontal lobes (41).
Bizzi (5) described neurons that discharged
during smooth pursuit, but these neurons (type
II) also discharge in association with tonic eye
position (6) and are therefore distinct from the
smooth pursuit neurons described here. Nonsaccadic eye movements and other behaviors
(vergence, centering, pupil dilation, blinking,
and nystagmus), previously described as elicited by frontal eye-field stimulation, appear to
reflect the use of extremely high currents and
anesthetized subjects (7, 9, 12, 62). However,
slow movements in this study were elicited
with currents as low as 25 PA in fully awake
subjects and were even obtained while the
monkeys fixated a stationary light and therefore were quite alert. Moreover, neuronal activity at slow-movement sites suggested a relationship between cell function and elicited
eye movements analogous to that we showed
for rapid, saccadic eye movements. Too few
of these cells were studied to answer several
basic questions about them, such as if they
OF FRONTAL
EYE FIELDS
731
discharge in response to the visual surround
motion created by eye movements ( 10, 55),
and if they respond during smooth eye movements elicited by optokinetic or vestibular
stimuli (34).
Saccade latencies and frontal
eye-jield eflerents
The wide range of evoked saccade latencies
found here could reflect the different pathways
whereby the frontal eye fields can effect saccadic eye movements. Three principal sets of
frontal eye-field efferents project to subcortical
oculomotor structures, and all originate in the
infragranular layers where low thresholds for
evoked saccades were found. First, there is a
massive projection to the intermediate layers
of the superior colliculus (3, 35, 36, 38). This
collicular zone is strongly related to saccadic
eye movements (7 1). The frontal eye-field
projection to the caudate also can influence
the colliculus via the substantia nigra. Second,
the frontal eye fields project to the extreme
lateral zone of the mediodorsal nucleus of the
thalamus (35). Neural activity in this thalamic
region is also related to eye movements (56,
58). Third, there are direct projections to several premotor nuclei in the brain stem (37,39,
63), including the paramedian pontine reticular formation, which contains the saccade
generator of the brain stem (25, 53).
The longer-latency saccades are consistent
with a midbrain or thalamic route; however,
the shorter-latency saccades are not consistent
with such an indirect route because eye movements elicited by collicular stimulation are at
least as long (52). Therefore, we hypothesize
that short latencies were obtained when the
electrode was near neurons projecting to the
brain stem and, conversely, that longer latenties were obtained when the electrode was near
cells projecting to the colliculus or thalamus.
The frontal eye-field saccade latencies found
by Robinson and Fuchs (54) complement ours
regarding this hypothesis. With the use of
much higher stimulation currents they report
uniformly short latencies of 15-25 ms. High
currents may always recruit frontal eye-field
cells with direct projections to the brain stem,
whereas low currents may activate these
quicker pathways only if the electrode is in the
immediate vicinity of cells projecting to the
brain stem. The scattered clusters of large pyramidal cells in layer V of the frontal eye fields
732
BRUCE, GOLDBERG,
BUSHNELL, AND STANTON
suggest a patchy organization of frontal eye
field efferents, which might underlie the different saccade latencies.
In conclusion, this study strongly supports
the hypothesis that the frontal eye fields have
a causal role in initiating saccades. The classic
phenomenon of eye movements elicited by
frontal eye-field stimulation is most directly
explained as electrical activation of presaccadic
neurons in this cortex. All possible saccades
are represented within the frontal eye fields;
but this representation is not a simple topographic map like the representation of the visual field in structures where visual stimuli are
analyzed. Because the organization of saccades
in the brain stem is also complex (25, 53), we
are still far from understanding how particular
patterns of frontal eye-field activity translate
into particular eye movements.
ACKNOWLEDGMENTS
We are grateful to Jean Steinberg and Pam Brown for
manuscript preparation, Allison Abood, Geary Fitzpatrick,
and Rashid Mahdi for assistance with the monkeys, Laura
Cooper, Alvin Ziminsky, and George Creswell for technical
assistance, and to Joshua Goldberg for flicking the switch
between stimulate and record countless times.
This paper was supported in part by the National Eye
Institute EY-04740 to Charles J. Bruce, and EY-03763 to
Gregory B. Stanton.
Dr. Bushnell’s current address is: Faculte de Medecine
Dentaire, Universite de Montreal, C.P. 6209, Succ. A,
Montreal, Quebec H3C 3T9, Canada.
Address reprint requests to Michael E. Goldberg, M.D.,
Laboratory of Sensorimotor Research, National Eye Institute, Building 10, Room 1OB-09, Bethesda, MD 20205.
Received 3 1 December 1984; accepted in final form 12
April 1985.
REFERENCES
1. ALBRIGHT, T.D., DESIMONE, R., ANDGROSS,C.G.
Columnar organization of directionally selective cells
in visual area MT of the macaque. J. ~eur~~~y~~oz.
51: 16-31, 1984.
2. ASANUMA,
H. Recent developments in the study of
the columnar arrangement of neurons within the motor cortex. Physiol Rev. 55: 143-156, 1975.
3. ASTRUC, J. Corticofugal connections of area 8 (frontal
eye field) in Macaca mulatta. Brain Res. 33: 24 l-256,
1971.
4. BARBAS,
5.
6.
7.
8.
9.
10.
H. AND MESULAM, M-M.Organization
of
afferent input to subdivisions of area 8 in the rhesus
monkey. J. Comp. Neural. 200: 407-43 1, 198 1.
BIZZI, E. Discharge of frontal eye field neurons during
saccadic and following eye movements in unanesthetized monkeys. Exp. Brain Res. 6: 69-80, 1968.
BIZZI, E. AND SCHILLER, P. H. Neuronal activity in
the frontal eye fields of unanesthetized monkeys during head and eye movement. Exp. Brain Res. 10: 15 l158, 1970.
BLUM, B., KULIKOWSKI,
J. J., CARDEN, D., AND
HARWOOD,
D. Eye movements induced by electrical
stimulation of the frontal eye fields of marmosets and
squirrel monkeys. Brain Behav. Evol. 2 1: 34-41, 1982.
BRUCE,~.
J. ANDGOLDBERG,
M.E.Primate frontal
eye fields. I. Single neurons discharging before saccades. J. Neurophysiol.
53: 603-635,
1985.
BRUCHER, J. M. The frontal eye field of the monkey.
Int. J. Neural. 5: 262-28 1, 1966.
BUSHNELL,
M. C., GOLDBERG,
M. E., AND ROBINSON,
D. L. Behavioral enhancement of visual responses in
monkey cerebral cortex. I. Modulation in posterior
parietal cortex related to selective visual attention. J.
Neurophysiol.
11. COLONNIER,
tex. In: Brain
46: 755-772,
198 1.
M. The structural design of the neocorand Conscious Experience,
edited by
J. C. Eccles. New York: Springer-Verlag, 1966, p. l23.
E.C.,Yoss,R.E., ANDHENDERSON,
J. W.
The mammalian midbrain and isthmus regions. Part
II. The fiber connections. D. The pattern for eye
movements on the frontal eye field and on discharge
of specific portions of this field to and through midbrain levels. J. Comp. Neural. 97: 357-383, 1952.
13. CYNADER, M. AND BERMAN, N. Receptive-field organization of monkey superior colliculus. J. Neuro12. CROSBY,
physiol. 35: 187-201,
1972.
14. DESIMONE, R., ALBRIGHT,
T.D., GROSS,C.G.,AND
BRUCE, C. J. Stimulus selective properties of inferior
temporal neurons in the macaque. J. Neurosci. 4:
2051-2062,
1984.
15. FERRIER, D. Experiments on the brains of monkeys.
Philos. Trans. London 165: 433-488, 1875. (Croonian
Lect .)
16. FERRIER, D. The localization of function in the brain.
Proc. R. Sot. London Ser. B 22: 229-232,
1874.
17. FRITSCH, G. AND HITZIG,
E. Ueber die elektrische
Erregbarkeit des Grosshirns. Arch. F. Anat. Physiol
u. Wissenach. Med. 37: 300-332, 1870. Cited in: Some
Paperes on the Cerebral Cortex, edited by G. von Bonin.
Springfield, IL: Thomas, 1960.
18. GOLDBERG, M. E. Studying the neurophysiology of
behavior: methods for recording single neurons in
awake behaving monkeys. In: Methods in Cellular
Neurobiology,
edited by J. L. Barker and J. F. McKelvy. New York: Wiley, 1983, vol. 3, chapt. 8, p.
225-248.
19. GOLDBERG,
M.E. ANDBRUCE, C.J.Therelationship
of movement fields of frontal eye field neurons and
the saccades evoked by electrical stimulation at the
site of those neurons. Invest. Ophthalmol.
Visual Sci.
Suppl. 24: 272, 1983.
20. GOLDBERG,
M. E. AND BUSHNELL,
M. C. Visual behavior modifies the effects of frontal eye field stimulation in the rhesus monkey. Invest. Ophthalmol.
VisualSci.Suppl.
19: 175-176,
1980.
ELECTRICAL
STIMULATION
21. GOLDBERG, M.E. ANDBUSHNELL,
M.C.Behavioral
enhancement of visual responses in monkey cerebral
cortex. II. Modulation in frontal eye fields specifically
related to saccades. J. Neurophysiol.
46: 773-787,
1981.
22. GOLDBERG, M. E. AND BUSHNELL, M. C. Role ofthe
frontal eye fields in visually guided saecades.In: Progress in Oculomotor
Research, edited by A. Fuchs and
W. Becker. New York: Elsevier/North Holland, 198 1,
p. 185-192.
23. GOLDBERG,
M. E. AND WURTZ, R. H. Activity of
superior colliculus in behaving monkey. I. Visual receptive fields of single neurons. J. Neurophysiol.
35:
542-559, 1972.
24. GOLDMAN, P. S. AND NAUTA, W. J. Autoradiographic
demonstration of a projection from prefrontal association cortex to the superior colliculus in the rhesus
monkey. Brain Res. 116: 145-149, 1976.
25. HENN, V., HEPP,K.,ANDBUTTNER-ENNEVER,
J.A.
The primate oculomotor system. II. premotor system-a synthesis of anatomical, physiological, and
clinical data. Hum. Neurobiol.
1: 87-95.
26. HOLMES, G. Disturbances of visual orientation. Brit.
J. Ophthalmol.
2: 449-468,
1918.
27. HOLMES, G. The cerebral integration of the ocular
movements. Br. Med. J. 2: 107-l 12, 1938.
28. HUBEL, D. H. AND WIESEL, T. N. Receptive fields
and functional architecture of monkey striate cortex.
J. Physiol. London
195: 2 15-243, 1968.
29. HUBEL, D. H. AND WIESEL, T. N. Ferrier lecture.
Functional architecture of macaque monkey visual
cortex. Proc. R. Sot. Lond. Ser. B, 198, l-59, 1977.
30. HUBEL, D. H. AND WIESEL, T. N. Sequence regularity
and geometry of orientation columns in the monkey
striate cortex. J. Comp. Neural. 158: 267-294, 1974.
31. JACOBSON, S. AND TROJANOWSKI, J. Q. Prefrontal
granular cortex of the rhesus monkey. I. Intrahemispheric cortical afferents. Brain Res. 132: 209-233.
32. JONES, E. G. AND POWELL, T. P. S. An anatomical
study of converging sensory pathways within the cerebral cortex of the monkey. Brain 93: 793-820, 1970.
33. JUDGE, S. J., RICHMOND,
B. J., AND CHU, F. C. Implantation of magnetic search coils for measurement
of eye position: an improved method. Vision Res. 20:
535-538, 1980.
34. KAWANO, K., SASAKI, M., AND YAMASHITA, M. Response properties of neurons in posterior parietal cortex of monkey during visual-vestibular stimulation.
I. Visual tracking neurons. J. Neurophysiol.
5 1: 340351, 1984.
35. KUNZLE, H. AND AKERT, K. Efferent connections of
cortical, area 8 (frontal eye field) in Macaca fascicularis. A reinvestigation using the autoradiographic
technique. J. Comp. Neural. 173: 147-164, 1977.
36. KUNZLE, H., AKERT, K., AND WURTZ, R. H. Projection of area 8 (frontal eye field) to superior colliculus
in the monkey. An autoradiographic study. Brain Res.
117: 487-492, 1976.
37. LEICHNETZ, G. R., SMITH, D. J., AND SPENCER,
R. F. Cortical projections to the paramedian pontine
reticular tegmental and basilar pons in the monkey.
J. Comp.
38. LEICHNETZ,
Neural.
228: 388-408,
1984.
G. R., SPENCER, R. F., HARDY, S. G. P.,
AND ASTRUC, J. The prefrontal cortico-tectal projection in the monkey. An anterograde and retrograde
OF FRONTAL
EYE FIELDS
733
horseradish peroxidase study. J. Neurosci. 2: 10231031, 1982.
39: LEICHNETZ, G. R., SPENCER, R. F., AND SMITH,
D. J. Cortical projections to nuclei adjacent to the
oculomotor complex in the medial dienmesencephalic
tegmentum in the monkey. J. Comp. Neural. 288:
359-387, 1984.
40. LUND, J. S., LUND, R. D., HENDRICKSON,
A. E.,
BUNT, A. H., AND FUCHS, A. F. The origin of efferent
pathways from the primary visual cortex, area 17, of
the macaque monkey as shown by retrograde transport
of horseradish peroxidase. J. Comp. Neural. 164: 287304, 1976.
41. LURIA, A. R., KARFQV, B. A., AND YARBUS, A. L.
Disturbances of active visual perception with lesions
of frontal lobes. Cortex 2: 202-2 12, 1966.
42. MERKER, B. Silver staining of cell bodies by means
of physical development. J. Neurosci. Methods 9: 235241, 1983.
43. MOHLER, C. W., GOLDBERG, M. E., AND WURTZ,
R. H. Visual receptive fields of frontal eye field neurons. Brain Res. 6 1: 385-389, 1973.
44. MOUNTCASTLE,
V. B. Modality and topographic
properties of single neurons of cat’s somatic sensory
cortex. J. Neurophysiol.
20: 408-434,
1957.
45. PALMER, L. A., ROSENQUIST,
A. C., AND TUSA,
R. J. The retinotopic organization of lateral suprasylvian visual areas in the cat. J. Comp. Neural. 177:
237-256, 1978.
46. PIGAREV, I. N., RIZZOLATTI, G., AND SCANDOLARA,
C. Neurons responding to visual stimuli in the frontal
lobe of macaque monkeys. Neurosci. Lett. 12: 207212, 1979.
47. RANCK, J. B., JR. Which elements are excited in electrical stimulation of mammalian central nervous system: a review. Brain Res. 98: 4 17-440, 1975.
48. RIZZOLATTI, G., SCANDOLARA, C., MATELLI, M., AM>
GENTILUCCI,
M. Afferent properties of periarcuate
neurons in macaque monkeys: I. Somato-sensory responses. Behav. Brain Res. 2: 125-146, 198 1.
49. RIZZOLA‘TTI, G., !~CANDOLARA, C., MATELLI, M., AND
GENTILUCCI,
M. Afferent properties of periarcuate
neurons in macaque monkeys: II. Visual responses.
Behav. Brain Res. 2: 147-163, 1981.
50. ROBINSON, D. A. A method of measuring eye movement using a scleral search coil in a magnetic field.
IEEE Trans. Biomed. Eng. 10: 137-145, 1963.
5 1. ROBINSON, D. A. Quoted by A. Roucoux and M.
Crommelink. Influence of supranuclear structures on
brain stem neurons. B. Collicular, thalamic, and cortical influence on brain stem neurons. In: Control of
Gaze by Brain Stem Neurons, edited by R. Baker and
A. Berthoz. New York: Elsevier/North Holland, 1977,
p. 487.
52. ROBINSON, D. A. Eye movements evoked by collicular
stimulation in the alert monkey. Vision Res. 12: 17951808, 1972.
53. ROBINSON, D. A. Control of eye movements. In:
Handbook
of Physiology.
The Nervous System. Bethesda, MD: Am. Physiol. Sot., 198 1, sect. 1, vol. II,
part 2, chapt. 28, p. 1275-1320.
54. ROBINSON, D. A. AND FUCHS, A. F. Eye movements
evoked by stimulation of frontal eye fields. J. Neurophysiol.
32: 637-648,
1969.
55. ROBINSON, D. L., GOLDBERG, M. E., AND STANTON,
734
BRUCE, GOLDBERG,
BUSHNELL,
G. B. Parietal association cortex in the primate: sensory mechanisms and behavioral modulations. J.
Neurophysiol.
41: 910-932, 1978.
56. SCHLAG, J. AND SCHLAG-REY,
M. Visuomotor functions of central thalamus in monkey. II. Unit activity
related to visual events, targeting, and fixation. J.
Neurophysiol.
5 1: 1175- 1195, 1984.
57. SCHLAG, J. AND SCHLAG-REY, M. Visuo-motor units
in frontal dorsomedial cortex of monkey. Sot. Neurosci. Abstr. 10, pt. 1: 989, 1984.
58. SCHLAG-REY, M. AND SCHLAG, J. Visuomotor functions of central thalamus in monkey. I. Unit activity
related to spontaneous eye movements. J. Neurophysiol. 5 1: 1149- 1174, 1984.
59. SCHWARTZ, E. Computational anatomy and functional architecture of striate cortex: a spatial mapping
approach to perceptual coding. Vision Res. 20: 645699, 1980.
60. SEGRAVES, M. A. AND GOLDBERG, M. E. Functional
properties of tectal projection neurons in the monkey
frontal eye fields. Sot. Neurosci. Abstr. In press.
61. SHIBUTANI, H., SAKATA, H., AND HYVARINEN,
J.
Saccades and blinking evoked by microstimulation of
the posterior parietal association cortex of the monkey. Exp. Brain Res. 55: 1-8, 1984.
62. SMITH, W. K. The frontal eye fields. In: The Precentral
Motor
Cortex, edited by P. C. Bucy. Urbana: Univ.
of Illinois Press, 1949.
63. STANTON, G. B., BRUCE, C. J., AND GOLDBERG,
M. E. Organization of subcortical projections from
AND STANTON
saccadic eye movement sites in the macaque’s frontal
64.
65
66
67.
68.
69.
70.
71.
72.
eye fields. Sot. Neurosci. Abstr. 8: 293, 1982.
STONEY, S. D., JR., THOMPSON,
W. D., AND ASANUMA, H. Excitation of pyramidal tract cells by intra-
cortical microstimulation: Effective extent of stimulating current. J. Neurophysiol.
3 1: 659-669,
1968.
SUZUKI, H. AND AZUMA, M. Topographic studies on
visual neurons in the dorsolateral prefrontal cortex of
the monkey. Exp. Brain. Res. 53: 47-58, 1983.
TEUBER, H. L. Unity and diversity of frontal lobe
function. Acta Neurobiol. Exp. 32, Suppl.: 6 15-656,
1972.
WAGMAN,
I. H., KRIEGER, H. P., PAPATHEODOROU,
C. A., AND BENDER, M. B. Eye movements elicited
by surface and depth stimulation of the frontal lobe
of Macaque mulatta. J. Comp. Neural. 117: 179- 188,
1961.
WALKER, A. E. A cytoarchitectural study of the prefrontal area of the macaque monkey. J. Camp. Neural.
73: 59-86, 1940.
WEINRICH,
M. AND WISE, S. P. The premotor cortex
of the monkey. J. Neurosci. 2: 1329-l 345, 1982.
WEINRICH,
M., WISE, S. P., AND MAURITZ, K-H. A
neurophysiological analysis of the premotor cortex of
the monkey. Brain 107: 385-414, 1984.
WURTZ, R. H. AND GOLDBERG,
M. E. Activity of
superior colliculus in behaving monkey. III. Cells discharging before eye movements. J. Neurophysiol.
35:
575-586, 1972.
WURTZ, R. H. AND MOHLER,
C. W. Enhancement
of visual responsesin monkey striate cortex and frontal
eye fields. J. Neurophysiol.
39: 766-772,
1976.