Intracerebral dynamics of saccade generation in the human frontal eye field and supplementary eye field

www.elsevier.com/locate/ynimg

NeuroImage 30 (2006) 1302 – 1312

Intracerebral dynamics of saccade generation in the human

frontal eye field and supplementary eye field

Jean-Philippe Lachaux,a,d,* Dominique Hoffmann,e Lorella Minotti,b

Alain Berthoz,c and Philippe Kahaneb

aMental Processes and Brain Activation, Unite 280, INSERM 151, cours Albert Thomas, 69424 Lyon cedex 03, FrancebDepartment of Neurology, Grenoble Hospital, INSERM JE2413, Grenoble, FrancecCNRS Laboratoire de Physiologie de la Perception et de l’Action, College de France, Paris, FrancedCNRS UPR640, LENA, Paris, FranceeDepartment of Neurosurgery, Grenoble Hospital, INSERM 318, Grenoble, France

Received 18 April 2005; revised 9 November 2005; accepted 11 November 2005

Available online 18 January 2006

Recent functional imaging and electrical stimulation studies have

localized in humans two frontal regions critical for the production of

saccadic and anti-saccadic eye movements: the frontal and supple-

mentary eye fields (FEF and SEF, respectively). We investigated the

time course of their activations during the generation of pro- and anti-

saccades from direct intracranial EEG recordings of three human

epileptic patients. We found the preparation and the production of the

saccades to be coincident with focal and transient increases of EEG

power above 60 Hz. Those were produced in very specific brain sites

distributed in the FEF and the SEF (as identified by previous human

studies at a coarser time resolution). Furthermore, the spatio-temporal

resolution of those recordings turned out to be sufficient to

discriminate anatomically between several types of neural responses,

determined either by the visual or by the motor components of the

saccade tasks, and within this second category of responses, between

some associated with the preparation of the saccades and others

associated with their execution. Altogether, this study provides the

first evidence of high-frequency neural responses in the generation of

saccades in humans, and provides a firm basis for other studies

detailing further the functional organization of the human oculomotor

system at this level of spatial and temporal resolution.

D 2005 Elsevier Inc. All rights reserved.

Keywords: Saccades; Intracerebral EEG; Gamma band; Supplementary eye

field; Frontal eye field

1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.neuroimage.2005.11.023

* Corresponding author. Mental Processes and Brain Activation, Unite

280, INSERM 151, cours Albert Thomas, 69424 Lyon cedex 03, France.

Fax: +33 4 72 68 19 02.

E-mail address: [email protected] (J.-P. Lachaux).

Available online on ScienceDirect (www.sciencedirect.com).

Introduction

How does the human brain generate saccadic eye movements to

explore its visual environment? Converging lesional, stimulation

and functional imaging studies have shown that saccades are

produced by local generators in the reticular formation of the brain

stem and in the mesencephalic reticular formation, and prepared

and triggered in the superior colliculus, the basal ganglia, the

hippocampal formation and several areas (‘‘eye or oculomotor

fields’’) of the cerebral cortex (O’Driscoll et al., 1995; Petit et al.,

1993, 1995, 1996). In the human cortex, two ‘‘eye fields’’, the

frontal eye field (FEF) and the supplementary eye field (SEF), are

crucial for the final shaping of the ending commands for the

execution of saccades: the FEF is located in the premotor

dorsolateral cortex (Brodmann’s area 6) and is subdivided in two

subregions, the deep and lateral FEF (Lobel et al., 2001; Paus,

1996; Petit et al., 1995), while the SEF is located on the medial

wall of the premotor cortex at the vicinity of the paracentral sulcus

(Grosbras et al., 1999).

While the static architecture of this cortical network is now well

described, the dynamics of its activation, during saccade produc-

tion, is still largely unknown, essentially because this dynamics can

only be observed with (extremely rare) direct recordings of the

neural activity in the human FEF and SEF. To our knowledge, only

two studies have benefited from such circumstances to report the

time course of neural activation in the frontal eye fields during

saccades in humans (Sakamoto et al., 1991; Yamamoto et al.,

2004).

In both studies, the focus was solely on the low-frequency

(typically less than 10 Hz) EEG activities associated with visually

guided saccades; while high-frequency activities (faster than 60

Hz) were not studied. Our assumption was that in fact high-

frequency EEG activities may be the best neural correlates of

saccade preparation because this had been shown to be the case for

other types of movements: Crone et al. (1998) had found neural

http://www.sciencedirect.com

mailto: [email protected]

http://dx.doi.org/10.1016/j.neuroimage.2005.11.023

J.-P. Lachaux et al. / NeuroImage 30 (2006) 1302–1312 1303

activations above 60 Hz that precisely matched the temporal

dynamics of unilateral limb movements and the sensorimotor

topography of the motor cortex established by electrical stimula-

tions. Drawing from those results, and from congruent observa-

tions at the scalp level (EEG/MEG) (Pfurtscheller and Lopes da

Silva, 1999), we made the hypothesis that similar gamma

activations occur in the oculomotor fields and constitute an

important aspect of the network dynamics that produces saccades.

We benefited from the cooperation of three epileptic patients

with implanted intracranial electrodes in the SEF and FEF to test

this hypothesis in two simple paradigms requiring two types of

saccadic eye movements: visually guided saccades (directed

towards an incoming visual stimulus) and anti-saccades (directed

towards the mirror-position of the incoming stimulus). Since the

fast dynamics of the saccade generation system had never been

described at this resolution in humans, the main objective of this

initial study was to describe the spatio-temporal organization of the

neural activity within this system.

Methods

Subjects

Three patients (P1, P2, P3) suffered from drug-resistant partial

epilepsy and were candidates for surgery. Magnetic resonance

imaging (MRI) of the brain was normal in two cases (P1, P3),

while it showed in the third patient (P2) a left dysembryoplastic

neuroepthelial tumor well-located in the posterior part of the third

frontal gyrus, which did not involve nor modify the FEF and SEF

regions. Because the location of the epileptic focus could not be

identified using noninvasive methods, the three patients under-

went intracerebral recordings by means of stereotactically

implanted multilead depth electrodes (SEEG) (for explanation

of this methodology, see Kahane et al., 2004), on the basis of

which the epileptogenic zone proved to be right mesial frontal

(P1), left inferior frontal (P2) and right frontal dorsolateral (P3).

Selection of sites to implant was made entirely for clinical

purposes with no reference to the present experimental protocol;

however, patients who entered this protocol were selected

because their implantation sampled the frontal oculomotor areas.

The three patients had previously given their informed consent to

participate in the experiment. All had normal vision without

corrective glasses.

Electrodes implantation

12 to 15 semi-rigid electrodes were implanted per patient, in

cortical areas which varied depending on the suspected origin of

seizures. Each electrode had a diameter of 0.8 mm and comprised

10 or 15 leads of 2 mm length, 1.5 mm apart (Dixi, Besancon,

France), depending on the target region. Therefore, various mesial

and lateral cortical areas were evaluated in the frontal lobe,

including sulcal cortex (Fig. 1). The electrode contacts were

identified on each individual stereotactic scheme, and then

anatomically localized using the proportional atlas of Talairach

and Tournoux (1988). In addition, the computer-assisted matching

of postimplantation CT-scan with a preimplantation 3-D MRI

(VOXIM R, IVS Solutions, Germany) provided a direct visuali-

zation of the electrodes contacts with respect to the brain anatomy

of each patient.

Paradigm

Each trial started with the appearance of a white fixation cross at

the center of a black computer screen (located 60 cm away from the

patient, in front of him/her), the cross remaining in this central

position for a random duration [1000 ms to 1500 ms] before jumping

to one of the two lateral positions (+ or �15- along the horizontal

axis) where it stayed for 1000 ms before jumping back to the central

position to initiate the beginning of a new trial. In pro-saccades

blocks, patients were instructed to perform as quickly as possible a

saccade to the visual cue when it moved to a lateral position, while

maintaining their head and body still. In anti-saccades blocks, patients

were instructed to saccade to the mirrored position of the visual cue.

In all the blocks, patients were instructed to avoid blinking

when the cue was in the central position. Each block consisted of

60 cue-jumps to the right and 60 cue-jumps to the left. Pro- and

anti-saccades blocks alternated for a total of 6 to 8 blocks.

The visual stimuli were delivered on a CRT monitor with a

refresh rate of 60 Hz, controlled by a PC (Pentium 133, Dos) with a

stimulus presentation software ‘‘stimulat’’ designed at the Cogni-

tive Neuroscience and Brain Imaging laboratory (LENA, CNRS

UPR640). To control the timing of stimulus delivery, a TTL pulse

was sent by the stimulation PC to the EEG acquisition PC each

time the first pixel of a stimulus frame was displayed on the screen,

with a time uncertainty less than 1 ms. The stimulus presentation

timing could therefore be measured with a precision less than 16

ms, the time necessary to display an entire frame at 60 Hz.

Recordings and stimulation

The SEEG studies were performed according to our routine

procedure, over a period of 10 to 17 days, with a reduction of the

patients medication. The aims were first to record spontaneous

seizures and second to perform intracerebral electrical stimulation

(IES). The goal of these latter was to reproduce part or all of the

patient’s ictal clinical symptomatology and to map eloquent areas

of the brain functionally.

Intracerebral recordings were conducted using an audio–

video–EEG monitoring system (Micromed, Treviso, Italy), which

allowed the simultaneous recording of 63 depth-EEG channels

sampled at 512 Hz [0.1–200 Hz bandwidth] during the experi-

mental paradigm (and 256 Hz [0.1–100 Hz bandwidth] during the

electrical stimulation session). One of the contact sites in the white

matter was chosen as reference. This reference has the same

impedance as the other contact sites, and is located in a region with

no or little source of electrical field, in addition, it is not

contaminated by eye movement artifacts or electromyographic

activity from subtle muscle contractions.

Saccade protocol

The patients performed the task of the protocol 4 days after the

implantation of the electrodes. Saccades were sorted in four

categories prior to analysis: pro-saccades to the right and to the

left (ProRight and ProLeft) performed during the guided saccades

blocks; and successful anti-saccades to the left (i.e. after a right cue)

and to the right (after a left cue) (AntiLeft and AntiRight).

Oculomotor performance was followed online using horizontal

and vertical electro-oculograms (EOG) together with an infra-red

based systemmounted on the patient’s head (Veonys, Paris, France).

This device measured the horizontal and vertical position of the left

Fig. 1. TF representation of a saccade in the frontal and supplementary eye fields. The top panel shows the entry point of the implanted electrodes for the

patients. The gray box indicates the region where most gamma activations were found. Gray (respectively black) dots indicate electrodes in the left

(respectively right) hemisphere (AC = anterior commissure, PC = posterior commissure). The three maps surrounding the brain show the TF energy modulation

recorded in three sites averaged across all the saccades of a given type (bottom left: nV1 in the FEF of P1, for AntiRight saccades; bottom right: a1 in the SEF of

P3 for ProRight saccades; top right: mV6 in the FEF of P2, for AntiLeft saccades). For each frequency, the energy has been normalized relative to the [�500:�300 ms] baseline (z-transform) (that is, for instance, the baseline of the (P2, mV6) map was computed only from AntiLeft saccades). Note the strong

deactivation in the beta range (around 20 Hz), for each site. The plot in the bottom left shows for comparison the event-related potential of nV1 (P1) for

AntiRight saccades.

J.-P. Lachaux et al. / NeuroImage 30 (2006) 1302–13121304

eye pupil 50 times per second. It was used to monitor the amplitudes

of the saccades. The precise latency of saccades onsets were

extracted from the EOG with a precision better than 5 ms. Only

correct saccades and anti-saccades were kept for analysis, with the

additional constraint that the eyes had to have been still within 600

ms prior to the saccade and 200ms after the saccade. In particular, all

trials showing blinks were excluded from analysis. Among the

remaining trials, only those with no sign of epileptiform activity, on

any of the analyzed electrodes, were considered.

Intracerebral electric stimulations (IES)

IES was performed a few days later under video–SEEG control.

Following our standard clinical practice (Kahane et al., 1993),

stimulations at 1 Hz (pulse width = 3ms) and 50 Hz (pulse width = 1

ms) were applied between contiguous contacts at various levels of

the electrodes axis. Bipolar stimuli were delivered using a constant

current rectangular pulse generator designed for a safe diagnostic

stimulation of the human brain (Micromed, Treviso, Italy), accord-

ing to parameters proven to produce no structural damage (Gordon et

al., 1990). Attention was focused on the clinical responses elicited

by the stimulation of the anatomical sites in which high-frequency

EEG activations during saccade generation were observed.

Time-frequency analysis

Fig. 2 provides a schematic illustration of the time-frequency

analysis of the data. For each single saccade, bipolar derivations

computed between adjacent electrode contacts were analyzed in the

time-frequency (TF) domain by convolution with complex Gaussian

Morlet’s wavelets (Tallon-Baudry et al., 1997), thus providing for

each saccade a TF power map P (t, f) = Aw (t, f) * s(t)A2, where w(t,

f) was for each time t and frequency f a complex Morlet’s wavelet

w(t, f) =A exp(� t2 / 2rt2). exp(2ipft), withA ¼ rt

ffiffiffi

pp

ð Þ�1=2 and rt =

1 / (2prf) and rf a function of the frequency f: rf = f / 7. With this

value of rf, the temporal resolution of the wavelet transform is 30ms

at 100 Hz and 60 ms at 50 Hz (as measured by the temporal extent of

the wavelet transform of a Dirac function).

Significant spectral modulations caused by the stimuli were

detected using a Wilcoxon nonparametric test that compared across

the saccades of a given type, the total energy in a given TF tile, with

that of a tile of similar frequency extent, but covering a prestimulus

baseline period [from �500 ms to �300 ms before the saccade

onset] (typically in this study, the frequency extent of the tiles was

[110–140 Hz] to detect very high gamma responses). Significant

responses were defined by a P value less than 0.001.

Comparison between saccade types (e.g. pro-saccades to the left

vs. anti-saccades to the left) was done via a Kruskal–Wallis

nonparametric analysis (in the text: KW) applied on the raw TF

values of energy, on a set of TF tiles [100 ms � 30 Hz] covering a

[0:1000 ms] � [110:140 Hz] domain (one test per tile comparing

the values obtained for all the trials in the two conditions).

Statistical comparisons were performed individually for each

patient, and within each patient, individually for each recording

site, because of the possible heterogeneity in neural behavior

across different patients and anatomical locations. For each

Fig. 2. Schematic illustration of the data analysis procedure. For each recording site, the EEG activity recorded during each saccade (or trial) is converted into

its time-frequency representation using a wavelet transform. The average across trials of those time-frequency maps can be baseline corrected to reveal the

gamma band response induced by the saccade (bottom map). The average across trials of the raw EEG traces is the event-related potential (ERP). Note that the

wavelet transform of the ERP is not equal to the average of the wavelet transforms of the individual raw EEG traces (red cross). Also, the saccade-related

gamma band oscillations are usually invisible in the raw EEG traces (and in their wavelet transform) because of their small amplitude, as compared with the

amplitude in the lower frequency bands. In this example (P1, m7 in the right FEF, AntiRight saccades), the most visible effect in the raw traces is the rebound

of the beta band oscillations after the saccades. To compare the spectral energy of the EEG response between two experimental conditions, we (a) define a time-

frequency region of interest (TFROI, as identified for instance by the green box A in the upper two TF maps), (b) extract the mean energy value in this region of

interest for each trial, then (c) use a Kruskal–Wallis nonparametric test to compare the list of values obtained in the two conditions. The procedure is similar to

identify significant spectral energy increases relative to the baseline, except that we use a Wilcoxon nonparametric test to compare across the trials of a single

saccade type the energy in the TFROI with the energy in the baseline (green box B). (For interpretation of the references to colour in this figure legend, the

reader is referred to the web version of this article.)


Table 2

Anatomical locations of the sites where high-frequency oscillations were

recorded

Pt Electrode

contacts

Talairach

coordinates

Anatomical

location

P1 i13 +45/�11/+36 R lateral FEF

n12 +45/+4/+48 R lateral FEF

m7 +25/�16/+50 R deep FEF

mV7 �26/�20/+49 L deep FEF

nV8 �27/+3/+46 L deep FEF

n1 +5/+4/+48 R SEF

nV1 �1/+3/+46 L SEF

xV1 �9/+28/+21 L VMFCx

P2 mV6 �23/�6/+49 L deep FEF

mV7 �27/�6/+49 L deep FEF

mV1 �6/�6/+49 L SEF

P3 a8 +31/�1/+53 R lateral FEF

v11 +38/�3/+41 R lateral FEF

vV12 �45/�3/+41 L lateral FEF

v8 +27/�3/+41 R deep FEF

vV6 �26/�3/+41 L deep FEF

a1 +4/�1/+53 R SEF

y3 +5/+1/+49 R SEF

v1 0/�3/+41 R SEF

vV2 �10/�2/+47 L SEF

The Talairach coordinates of the deep and lateral FEF and the SEF are

drawn from electric stimulations and functional neuroimaging studies: L

lateral FEF (�41 < x <�48; �13 < y < 7; 35 < z < 52), R lateral FEF: (36 <

x < 44; �8 < y < 7; 37 < z < 48), L deep FEF(�25< x < �32; �4 < y <

�12; 49 < z < 56), R deep FEF (23 < x < 34; �3 < y < �8; 44 < z < 56)

(Beauchamp et al., 2001; Grosbras et al., 1999; Heide et al., 2001; Lobel et

al., 2001; Luna et al., 1998). SEF: (�12 < x < 8; �36 < y < �2; 48 < z <

54) (Grosbras et al., 1999). VMFCx = Ventro-Medial Frontal Cortex.


recording site, we analyzed the effects of two factors (saccade

direction, ‘‘left or right’’, and saccade instruction, ‘‘pro or anti-

saccade’’) on the EEG spectral energy measured during the

saccades (the repeated measures).

For display purposes, some figures show normalized version of

the average TF maps. Those correspond to TF maps showing

frequency by frequency, the relative energy variation with respect to

a baseline chosen [�500 ms: �300 ms] prior to the saccade

instruction. That is, the mean and standard deviation of the TF

energy during the baseline period were calculated for each

frequency, and used to z-transform the average TF maps (i.e.

averaged across the saccades of a given type) frequency-wise (i.e.

P(t, f) is replaced by (P(t, f)�meanbaseline( f)) / stdbasline( f)). Unless

specified otherwise, the baseline is computed from the average TF

map that is normalized; when this is not the case, the types of

saccades included in the baseline computation are specified in the

legend of the figure. Note that the statistical comparisons between

saccades types were not based on those normalized maps but on the

original (not baseline corrected) TF maps.

The EEG signals were evaluated with the software package for

electrophysiological analysis (ELAN-Pack) developed in the

INSERM U280 laboratory.

Amplitude correlations across recording sites

We tested for possible cross-site correlations between the

energy of well-defined components of the TF domain. This was

done by defining for each site a TF region of interest (TFROI),

centered around the saccade onset [�100 ms: 100 ms], and then

computing a Spearman rank correlation coefficient between the

series of energy values obtained for each site within their TFROI

across the trials (Lachaux et al., 2003).

Results

Behavioral data

Table 1 summarizes the amplitude, reaction times, standard

deviations and number of analyzed saccades for each patient, after

artifact rejection. For all three patients, reaction times were faster

Table 1

Number of correct and clean saccades for each patient and saccade types

Patient Saccade

type

#

saccades

Mean

RT

(ms)

Standard

RT (ms)

Mean

amplitude

(degree)

Standard

amplitude

(degree)

P1 ProLeft 186 206 55 15.6 2.2

P1 ProRight 196 197 51 13.6 1.8

P1 AntiLeft 119 228 67 19.0 6.4

P1 AntiRight 120 239 80 15.0 4.9

P2 ProLeft 183 168 84 16.6 1.3

P2 ProRight 194 160 34 14.5 1.0

P2 AntiLeft 83 257 59 18.6 4.5

P2 AntiRight 70 254 62 13.8 2.7

P3 ProLeft 135 189 47 18.5 3.5

P3 ProRight 52 275 90 13.4 3.3

P3 AntiLeft 46 319 96 18 5

P3 AntiRight 40 229 42 15.5 4.5

The four columns on the right contain the mean and standard deviations of

the reaction times and of the saccades amplitudes, in degrees.

for saccades than for anti-saccades, a well-known effect (Munoz

and Everling, 2004) (P1, ANOVA, F = 38.5, P < 0.001), (P2,

ANOVA, F = 225.5, P < 0.001), (P3, ANOVA, F = 21.6, P <

0.001). Also, for two patients, the anti-saccades had a larger

amplitude than the pro-saccades (P1, Anti > Pro, t test, P < 0.001)

(P2, Anti > Pro, t test, P = 0.0017). This is partly due to some

constraints associated with working with patients (we could not

impose them an effortful training) but also, and mostly, to our

choice to provide no target for the anti-saccades. An alternative

was to display a little mark symmetrical to the cue position in

order to reduce the anti-saccades variability; however, we

deliberately chose not to do so: we believed that this alternative

paradigm would amount to involve the selection of one, among

two, guided pro-saccades of opposite direction, rather than the

generation of an anti-saccade in its strictest sense. This amplitude

difference between pro and anti-saccades must be taken into

consideration when comparing the electrophysiological effects

associated with both kinds of saccades.

High-frequency EEG activations during saccade generation

The analysis was focused on the recordings sites exploring

the FEF and SEF regions, as assessed by several functional

imaging studies (cf. Table 2). The TF analysis revealed that the

generation of saccades was simultaneous with energy modu-

lations in several frequency bands. Those modulations extended

in a broad frequency range up to 200 Hz (the upper limitation

of the analysis, given the 512 Hz sampling rate). The visual

inspection of all the TF energy modulations led us to follow a


previous study (Crone et al., 1998) and to consider separately

several frequency bands: beta [15–30 Hz], low gamma (LG)

[30–45 Hz], high gamma (HG) [60–90 Hz] and very high

gamma (VHG) [110–140 Hz]. The event-related potentials

(ERP) were also computed, to examine EEG modulations

phase-locked to the saccade onset. Given the short time span

of interest in this study (mostly the 100 to 200 ms separating

the cue from the saccade onset), the alpha band was not

examined.

Fig. 3 shows the time course of energy modulation in each

frequency band for all the sites in the SEF and FEF. A visual

inspection of that figure indicated that the timing of the

responses in the VHG band, compared to the timing in the

other bands, seemed (a) more reproducible within each brain

region and (b) more closely related to the saccade timing (i.e.

characterized by a gradual increase of activity peaking at the

Fig. 3. Energy profiles during saccade generation. Each graph represents,

for the ERP or for a specific frequency range, the energy modulation around

the saccade onset for all the sites located around the FEF (left graphs) and

the SEF (right graphs). Except for ERP, all plots correspond to the energy

(a) averaged in the specific frequency range, then (b) averaged across all the

AntiRight saccades and finally (c) normalized (z-transform) with respect to

the mean energy in the [�500: �300 ms] presaccade baseline (estimated

across all saccades types). ERP plots were computed by applying directly

steps b and c to the raw signals. In the bottom graphs, a star indicates the

sites showing atypical behaviors (v8 and v11 in the FEF of P3).

saccade onset). In an attempt to quantify those impressions, we

calculated within each frequency band, including the ERP (and

separately for the SEF and the FEF), the correlation coefficients

between each individual response (i.e. each of the 7 responses

recorded in the SEF or the 10 responses recorded in the FEF

displayed in the individual boxes of Fig. 3) and either (a) the

average of those (7 or 10) responses (to quantify the

reproducibility of the responses) or (b) a model gaussian

function peaking at saccade onset (to quantify the correlation

with the saccade timing). The standard deviation of the gaussian

was chosen so that 99% of the distribution occurred within 225

ms of the saccade onset (that is, the average reaction time across

all saccade types as calculated from Table 1 values). A Mann–

Whitney nonparametric test revealed that, in the FEF, the

correlation coefficients measuring the timing reproducibility were

significantly larger (P < 0.01) in the VHG band than in each of

the other four frequency bands. In the SEF, the reproducibility

was significantly larger (P < 0.01) for the VHG band than for

the beta band. No frequency band yielded reproducibility

measures significantly larger than in the VHG band. Similar

results were found for the correlation coefficients measuring the

correlation with the saccade timing (VHG larger than all the

other frequency bands in the FEF, and VHG larger than the ERP

in the SEF). For those reasons, the analysis was focused on the

behavior of the VHG band, in agreement with a previous study

of limb movements (Crone et al., 1998). One should note,

however, that the smallest degree of reproducibility in the other

frequency bands as compared with the VHG should not be taken

as an indication that they do not play an important role in the

saccade generation. Simply, we could not compensate this lack

of reproducibility across sites by reproducibility across several

patients with the same electrode locations (a recurrent limitation

in such rare clinical contexts).

While in all three patients, the electrodes sampled widely

distributed portions of the frontal lobes (see Fig. 1), significant

activations in the VHG frequency band were only found (to the

exception of only one site in the left ventro-medial frontal cortex)

in the FEF and the SEF, as defined in several studies (Beauchamp

et al., 2001; Grosbras et al., 1999; Heide et al., 2001; Lobel et al.,

2001; Luna et al., 1998). Fig. 4 shows for comparison the Talairach

coordinates drawn from a meta-analysis of FEF and SEF

activations during saccades (Lobel, 1999) and the coordinates of

the VHG activation sites found in this study.

Fig. 4 illustrates the spatio-temporal dynamics of the SEF and

FEF VHG activations for ProLeft and AntiLeft saccades. The

energy increases gradually within 100 to 200 ms to reach a

maximum around the precise latency of the saccade before

returning to a baseline level over the course of 400 ms.

SEF vs. FEF activations

We found no clear-marked differences between the time courses

of the SEF and FEF VHG activations, or between their frequency

extents. However, the only sites which clearly started to activate

before the cue onset were in the SEF (P1, nV1 and n1), while all the

FEF sites started to activate around the cue onset (as determined

from the mean reaction time). In the FEF, the VHG time courses

were consistent across recording sites, with the exception of two

sites (P3, v8 and v11) which activated early after the cue onset.

However, those two sites seemed to respond to the specific location

of the cue, rather than to the saccade properties (see below).

Fig. 4. Dynamics of saccade generation in the SEF and FEF. Top panel shows for each patient the sites that showed an energy modulation in the [100–150 Hz]

during saccade generation (large spots). Small spots did not show any gamma modulation. Patients are color-coded. Superimposed are FEF and SEF activation

sites drawn from a meta-analysis of imaging data in similar protocols (Grosbras et al., 1999; Lobel, 1999). Axis indicate x and y Talairach coordinates around

the plane (z = 45mm). Left and right bottom panels show, on the same map, the normalized energy levels in the [100–150 Hz] bands for ProLeft saccades (left

panel) and AntiLeft saccades (right panel), at several time steps relative to saccade onset (t = 0 ms). For both panels, the energy was normalized relative to the

mean energy level during the [�500: �300 ms] � [100–150 Hz] baseline computed across all ProLeft saccades. In the left panel, bold circles indicate

activations that are significantly greater than the average energy level in the precue baseline (Wilcoxon, P < 0.001). In the right panel, bold circles indicate

activations that are significantly greater for the AntiLeft saccades than for the ProLeft saccades. This allows the direct comparison of pro- vs. anti-saccades, for

the same latency relative to the saccade onset. Crossed circles indicate instances where pro-saccades yield a stronger gamma activation than anti-saccades.

Triangles indicate negative values. Red symbols are nonsignificant. (For interpretation of the references to colour in this figure legend, the reader is referred to

the web version of this article.)


Bilateral activation

To the exception of two FEF sites (P3, v8 and v11), saccades

to the left and saccades to the right elicited comparable VHG

activations in the FEF and SEF; and this was true for pro- and for

anti-saccades (Kruskal–Wallis tests, P > 0.5). The two exceptions

were the two neighboring sites in the right FEF that yielded an

early VHG activation after the cue onset (P3, v8 and v11). There,

the VHG energy around the saccade onset [�100 ms: +100 ms]

was stronger for the ProLeft than for the ProRight saccades (KW,


P < 0.001). However, the effect was exactly the opposite for anti-

saccades. This was due to the fact that the two sites yielded

similar activations for similar cue positions (that is, similar

activations for Proleft and AntiRight saccades, for instance), so

that their activation seemed not related to the motor aspects of the

saccade preparation, but to the analysis of the cue position or to

the attentional shift triggered by the cue onset.

Saccade preparation vs. saccade execution

In some of the FEF and SEF sites showing a gradual VHG

energy increase before the saccade onset, we found a significant

correlation across the trials between the saccade latency and the

level of VHG activation before the cue onset (i.e. the mean

energy in the [cue � 200 ms: cue] � [110–140 Hz] TF region of

interest). Those significant correlations were always negative, that

is, reaction times were shorter for higher levels of VHG

activations (P1, nV1: ProLeft (Spearman rank correlation; rho =

�0.37, P < 0.001), ProRight: rho = �0.42, P < 0.001); AntiLeft,

rho = �0.21, P = 0.022; AntiRight, rho = �0.21, P = 0.022) (P3,

V11, AntiLeft, rho = �0.47, P < 0.001). A much more precise

correlation with the saccade timing was found with an atypical

response in the FEF: site n12, that lied lateral to the other right

FEF sites, in P1, activated almost solely during the saccade

production and in a narrow frequency range around 80 Hz. As

can be seen in Fig. 5, the time course of activation in this site

followed very closely the time course of the first time derivative

of the horizontal EOG. It is thus possible that the saccade

preparation requires a broad gamma activation in the SEF and

deep FEF, while its triggering is mediated by a lower frequency

burst in a more lateral region which could correspond to the

lateral FEF identified by Lobel et al. (2001).

Fig. 5. Saccade-triggering in the lateral part of the FEF. The top panel

shows the TF map of energy as it was recorded in the lateral portion of

the FEF (P1: n12) during an anti-saccade to the right (AntiRight). The

map has been normalized for each frequency relative to the [�500: �300ms] baseline (z-transform). The maps were essentially the same for the

other types of saccades. The bottom panel shows the good correlation

between the cross-section of this activation map in the 60�90 Hz

frequency range (bold line) and the timing of the saccade as it can be

quantified by the horizontal EOG (gray line) and by its first time-

derivative (dotted line).

Anti-saccades vs. pro-saccades

We compared systematically pro- and anti-saccades moving the

eyes in the same direction and observed that in all sites but two (v8

and v11, P3, for the reasons mentioned above) anti-saccades

yielded either a stronger or an equivalent VHG activation than pro-

saccades (the anti-saccades yielded a stronger VHG activation in 9/

17 sites for the saccades to the left, and 8/17 sites for the saccades

to the right). For the saccades to the left, the activation was

stronger (KW: P < 0.001) for anti- than pro-saccades in 4/7 SEF

sites (P1: n1, nV1; P2: mV1; P3: vV2) and 5/10 FEF sites (P1: m6,

mV7, nV8; P2: mV6; P3: vV6), while for the saccades to the right, the

activation was stronger for anti- than pro-saccades in 3/7 SEF sites

(P1: n1, nV1; P2: mV1) and 5/10 FEF sites (P1: m7, mV7, nV8; P3: v8,v11). The right panel of Fig. 4 illustrates the differences between

pro- and anti-saccades to the left (results are essentially the same

for saccades to the right). The strongest difference between pro-

saccades and anti-saccades activations was observed in the SEF,

where they could start even before the cue onset (P1: n1, nV1) (inthis block design, patients knew in advance whether they had to do

a pro- or an anti-saccade), but they also occurred in the FEF. Note

that the difference in VHG amplitude may simply be due to the fact

that anti-saccades had larger amplitudes than pro-saccades, if in

general, the saccade’s amplitude correlates with the VHG

activation level. We tested this possibility for the anti-saccades:

we calculated for each site the Spearman correlation coefficient

between the saccade amplitude and the level of VHG activation

before the saccade onset across all the anti-saccades (once again,

the VHG activation was defined as the mean energy in the [cue

�200 ms: cue] � [110–140 Hz] TF region of interest). However,

we found contradictory results (a couple of significant correlations

which sign varied across sites and patients).

Cross-sites correlations within oculomotor fields

We found that the amount of VHG energy around the saccade

onset ([�100 ms: 100 ms] � [110–140 Hz]) was correlated

between SEF and FEF sites (P1, nV1 and m7: ProLeft, Spearman

rank coefficient rho = 0.31, P < 0.0001; ProRight, rho = 0.20,

P = 0.0051; AntiLeft, rho = 0.17, P = 0.051) (P1, n1 and m7:

ProLeft, Spearman rank coefficient rho = 0.27, P = 0.000186;

ProRight, rho = 0.17, P = 0.013; AntiRight, rho = 0.19, P =

0.034). Those correlations were always positive, indicating that

the saccades for which the VHG activations were the strongest

(respectively weakest) in one site tended to be also those for

which they were the strongest (respectively weakest) for the other

one. We also tested for cross-sites phase-correlations between

sites (in the sense of phase-synchrony, as quantified in Lachaux et

al., 2002), but found no significant phase correlations in this

frequency range.

Comparison with the effects of electric stimulations

In patient P3, IES at 50 Hz had oculomotor effects in three FEF

sites (vV6: right tonic eye version; v8: left tonic eye version; v11: asubjective sensation that the eyes moved to the left) as described in

a previous study (Lobel et al., 2001). In the first of those three sites

(P3, vV6), individual pulses at 1 Hz produced clonic eye versions. It

is not clear how IES can produce eye movements; however, one

may formulate from this study the hypothesis that IES elicits

saccades because they induce gamma activations in the FEF and

Fig. 6. Effect of an electric stimulation in the FEF. Micro-pulse electric

stimulations in the FEF of Patient P3 (vV8, P3) triggered saccades to the

right. Interestingly, those pulses also induced above 60 Hz a gamma

energy increase between 100 and 400 ms in a FEF site nearby (vV11) andin the contra-lateral FEF (v11). Those increases can be seen after the

stimulation artifact in the TF representations of the responses in those

sites (highlighted boxes in the bottom displays). Both maps were

normalized frequency by frequency relative to a [�400: �300 ms]

baseline (z-transform).


the SEF. 50 Hz IES could not be used to test this hypothesis,

because they produced artifacts in the gamma range; however,

since in P3, low-frequency IES of the left FEF also triggered

saccades, to the right (vV6, 1Hz, 3 mA), we had the rare opportunity

to test in the SEF/FEF for the presence of high-frequency

responses to the pulses that produced saccades. We could not

prove our hypothesis, but we found that the pulses produced

indeed a burst of energy above 70 Hz, and at latencies between 200

and 400 ms (Wilcoxon test, comparison across 40 consecutive

pulses between the TF energy of the burst, and the prepulse

baseline level in the same frequency range (Fig. 6). Note that those

signals were recorded in a different session at a lower sampling

frequency, 256 Hz).

Discussion

The results essentially confirmed our working hypothesis: the

generation of pro- and anti-saccades coincided with focal energy

increases in the gamma range in frontal lobe regions which are

known to be involved in pro- and anti-saccades tasks, i.e. the FEF

and SEF regions. The time course of the increase matched well the

timing of the saccades, particularly in the high portion of the

gamma range (typically above 100 Hz, what we called very high

gamma or VHG).

Our recordings provided for the first time in humans a glimpse

into the saccade generation system at a millimeter/millisecond

spatio-temporal resolution. This resolution turned out to be

sufficient to identify several distinct types of neural responses

within this system, and this is, in our sense, the major outcome of

this study.

In the FEF, we could distinguish (a) VHG activations that

responded to the direction of the saccades and (b) VHG activations

that responded to the direction of the cue. This is consistent with

the existence of several classes of neurons in the FEF, eye

movement-related neurons that respond to the motor components

of the saccade and others responding to its visual components

(Amador et al., 2004; Moschovakis et al., 1996; Murthy et al.,

2001; Schall, 2002). In fact, we did not expect to be able to

separate between these two kinds of responses with our millimetric

spatial resolution, since this implied a topographical separation

between the two classes on this coarse millimeter scale. Still, this

observation now paves the way for experimental designs that

would investigate differentially the visual and motor responses in

the human FEF, and possibly its implication in spatial attention

(Corbetta et al., 1998; Schall, 2004).

Within the FEF neurons, population that responded to the motor

component of the saccade, we could use their time course to

separate between (i) an activation that precisely matched the timing

of the saccade execution, from (ii) a group of activations starting

200 ms before the saccade onset to end 200 ms after. Since the first

activation (i) was in a more lateral aspect of the FEF than the

second group (ii), it is possible that they originated from two

functionally distinct portions of the FEF, one involved in the

programming of the saccade and one involved in its execution.

This dichotomy could match the distinction proposed between a

lateral FEF and a deep FEF (Lobel et al., 2001), although the

functional distinction reported between the two regions was

different from the one we observed.

In the SEF, we found that a majority of VHG activations had a

time profile similar to the FEF activations. However, two sites

started to activate several hundreds of milliseconds before the cue

onset for the anti-saccades. This is compatible with a role of the

SEF in the inhibition of incorrect reflex pro-saccades in the anti-

saccade paradigm (Munoz and Everling, 2004).

In addition, we observed that some SEF and FEF VHG

activations were correlated in terms of their energy. This can be

interpreted in at least two ways: (a) the two regions may be under

the influence of a common, possibly remote, driver that determines

for each saccade their level of VHG activation, (b) alternatively,

the correlation may be due to reciprocal, or unilateral, connections

between the two fields and reveal a cooperation between the two

regions during the generation of saccades. For instance, the

emergence of a local synchronization in one of these regions

may facilitate the synchronization in the other. We found similar

correlations between the ventral and the dorsal visual pathways

during the perception of visual objects (Lachaux et al., 2005a,b). If

this is confirmed, then it should be possible to test in the future the

implication of this functional connectivity in the production of

saccades in humans.

Besides those main results, our recordings revealed several

aspects that are open for discussion.

First of all, we found that for a majority of SEF and FEF

sites, VHG activations were equivalent for saccades to the left

and saccades to right. This is in contradiction with several

studies indicating that the activation should be stronger for


saccades that aim at positions contralateral to the recording side

(Munoz and Everling, 2004). One possible explanation is that

our recordings may average the activity of saccade and fixation

neurons, but that would imply that the contributions of the two

groups are exactly equal in our macro-recordings. Another

possibility is that each frontal lobe is able to mediate saccades

in both directions, and indeed, Herter and Guitton (2004)

recently reported that a single hemicortex can mediate accurate

bidirectional saccade control.

Also, we found that VHG activations were stronger before anti-

saccades than before pro-saccades, particularly in the SEF (and

with the exception of the two FEF sites that responded to the cue

location). This is in agreement with several studies using different

level of observations (Amador et al., 2004; Everling et al., 1997;

Munoz and Everling, 2004). Although one should bear in mind that

our patients produced anti-saccades with larger amplitudes than the

pro-saccades, it is possible that the SEF participates in the

inhibition of the unwanted reflex pro-saccade (Munoz and Ever-

ling, 2004) and this is consistent with the fact that the VHG

difference between the two kinds of saccades exists even before the

cue onset, in the SEF.

Finally, we found no strong sign of functional dissociation

between the FEF and the SEF, since a majority of SEF and FEF

sites had similar VHG activations. However, this may be due to a

limitation of our simple paradigm; more complex experimental

designs, introducing gaps in the anti-saccade task (Dorris and

Munoz, 1995) or oculomotor sequences (Lu et al., 2002) could lead

us to a clearer dissociation between the two regions. However, the

choice of a simple, but classic, protocol was deliberate as a starting

point in the investigation of the oculomotor system with human

intracranial EEG. It is likely that more sophisticated tasks will lead

to a wider-spread activations, including for instance the Cingulate

Eye Field and the Dorsolateral Pre Frontal Cortex for tasks

involving decisions (Pierrot-Deseilligny et al., 2003).

Altogether, this study demonstrated the potential of intracranial

EEG for the understanding of oculomotricity in humans. Since

most of the studies investigating eye movements in humans have

relied on metabolic imaging, such as fMRI and PET, it is worth

mentioning that one advantage of SEEG is to provide a

multidimensional description of the local neural activity, while

fMRI and PET project this activity down to a single activation

value. This multidimensional view extends along a time and a

frequency axis, and reveals the coexistence of simultaneous

activations and deactivations in several frequency ranges. We

made a deliberate choice to focus solely on the gamma band, but

this should not mask the involvement of lower frequency activities,

in particular the beta band. In several FEF and SEF sites, we

observed for instance after the saccade a rebound of energy in the

beta band, in which precise function should be investigated in the

future, in conjunction with similar rebounds observed during visual

perception after the presentation of visual objects (Bouyer et al.,

1987; Lachaux et al., 2005a,b).

Our study also emphasized the usefulness of ‘‘dynamical

spectral brain imaging’’, that is the anatomically-resolved quanti-

fication of the activations and deactivations in all the EEG

frequency ranges (necessarily at the intracerebral level). It is quite

remarkable that similar gamma band activations appear to be

directly involved in production of saccades in the FEF/SEF and in

the visual cortex during the perception of visual objects, as we

have shown in another study (Lachaux et al., 2005a,b). This opens

the possibility to use dynamical spectral brain imaging in a

systematic way to map out the large-scale neural networks

mediating cognition in humans.

Acknowledgments

We thank Pierre Leboucher and Michel Ehrette for their help

regarding the infrared oculography device. We also thank Valerie

Balle, Patricia Boschetti, Carole Chatelard, Veronique Dorlin,

Eliane Gamblin and Martine Juillard and Laurent Hugueville for

their invaluable help. JPL was supported by the Fondation

Fyssen.

References

Amador, N., Schlag-Rey, M., Schlag, J., 2004. Primate antisaccade: II.

Supplementary eye field neuronal activity predicts correct performance.

J. Neurophysiol. 91 (4), 1672–1689.

Beauchamp, M.S., Petit, L., Ellmore, T.M., Ingeholm, J., Haxby, J.V., 2001.

A parametric fMRI study of overt and covert shifts of visuospatial

attention. NeuroImage 14 (2), 310–321.

Bouyer, J.J., Montaron, M.F., Vahnee, J.M., Albert, M.P., Rougeul, A.,

1987. Anatomical localization of cortical beta rhythms in cat.

Neuroscience 22 (3), 863–869.

Corbetta, M., Akbudak, E., Conturo, T.E., Snyder, A.Z., Ollinger, J.M.,

Drury, H.A., Linenweber, M.R., Petersen, S.E., Raichle, M.E., Van

Essen, D.C., Shulman, G.L., 1998. A common network of functional

areas for attention and eye movements. Neuron 21 (4), 761–773.

Crone, N.E., Miglioretti, D.L., Gordon, B., Lesser, R.P., 1998. Functional

mapping of human sensorimotor cortex with electrocorticographic

spectral analysis: II. Event-related synchronization in the gamma band.

Brain 121 (Pt. 12), 2301–2315.

Dorris, M.C., Munoz, D.P., 1995. A neural correlate for the gap effect

on saccadic reaction times in monkey. J. Neurophysiol. 73 (6),

2558–2562.

Everling, S., Krappmann, P., Spantekow, A., Flohr, H., 1997. Influence of

pre-target cortical potentials on saccadic reaction times. Exp. Brain Res.

115 (3), 479–484.

Gordon, B., Lesser, R.P., Rance, N.E., Hart Jr., J., Webber, R., Uematsu, S.,

Fisher, R.S., 1990. Parameters for direct cortical electrical stimulation in

the human: histopathologic confirmation. Electroencephalogr. Clin.

Neurophysiol. 75 (5), 371–377.

Grosbras, M.H., Lobel, E., Van de Moortele, P.F., LeBihan, D., Berthoz, A.,

1999. An anatomical landmark for the supplementary eye fields in

human revealed with functional magnetic resonance imaging. Cereb.

Cortex 9 (7), 705–711.

Heide, W., Binkofski, F., Seitz, R.J., Posse, S., Nitschke, M.F., Freund, H.J.,

Kompf, D., 2001. Activation of frontoparietal cortices during memo-

rized triple-step sequences of saccadic eye movements: an fMRI study.

Eur. J. Neurosci. 13 (6), 1177–1189.

Herter, T.M., Guitton, D., 2004. Accurate bidirectional saccade control by a

single hemicortex. Brain 127 (Pt. 6), 1393–1402.

Kahane, P., Tassi, L., Francione, S., Hoffmann, D., Lo Russo, G., Munari,

C., 1993. Manifestations electro-cliniques induites par la stimulation

electrique intra-cerebrale par ‘‘chocs’’ dans les epilepsies temporales.

Neurophysiol. Clin. 22, 305–326.

Kahane, P., Minotti, L., Hoffmann, D., Lachaux, J., Ryvlin, P., 2004.

Invasive EEG in the definition of the seizure onset zone: depth

electrodes. In: Rosenow, F., Luders, H.O. (Eds.), Handbook of Clinical

Neurophysiology. Pre-surgical Assessment of the Epilepsies with

Clinical Neurophysiology and Functional Neuroimaging. Elsevier

Science.

Lachaux, J.P., Lutz, A., Rudrauf, D., Cosmelli, D., Le Van Quyen, M.,

Martinerie, J., Varela, F., 2002. Estimating the time-course of coherence


between single-trial brain signals: an introduction to wavelet coherence.

Neurophysiol. Clin. 32 (3), 157–174.

Lachaux, J.P., Chavez, M., Lutz, A., 2003. A simple measure of

correlation across time, frequency and space between continuous brain

signals. J. Neurosci. Methods 123 (2), 175–188.

Lachaux, J.P., George, N., Tallon-Baudry, C., Martinerie, J., Hugueville, L.,

Minotti, L., Kahane, P., Renault, B., 2005a. The many faces of the

gamma band response to complex visual stimuli. Unpublished

observations.

Lachaux, J.-P., George, N., Tallon-Baudry, C., Martinerie, J., Hugueville,

L., Minotti, L., Kahane, P., Renault, B., 2005b. The many faces of the

gamma band response to complex visual stimuli. NeuroImage 25 (2),

491–501.

Lobel, E., 1999. Orientation dans l’espace: imagerie fonctionnelle des

reseaux corticaux impliques dans la perception spatiale et le controle

des mouvements d’orientation du regard. SHFJ. (Paris VI).

Lobel, E., Kahane, P., Leonards, U., Grosbras, M., Lehericy, S., Le Bihan,

D., Berthoz, A., 2001. Localization of human frontal eye fields:

anatomical and functional findings of functional magnetic resonance

imaging and intracerebral electrical stimulation. J. Neurosurg. 95 (5),

804–815.

Lu, X., Matsuzawa, M., Hikosaka, O., 2002. A neural correlate of

oculomotor sequences in supplementary eye field. Neuron 34 (2),

317–325.

Luna, B., Thulborn, K.R., Strojwas, M.H., McCurtain, B.J., Berman, R.A.,

Genovese, C.R., Sweeney, J.A., 1998. Dorsal cortical regions subserv-

ing visually guided saccades in humans: an fMRI study. Cereb. Cortex

8 (1), 40–47.

Moschovakis, A.K., Scudder, C.A., Highstein, S.M., 1996. The microscop-

ic anatomy and physiology of the mammalian saccadic system. Prog.

Neurobiol. 50 (2–3), 133–254.

Munoz, D.P., Everling, S., 2004. Look away: the anti-saccade task and

the voluntary control of eye movement. Nat. Rev., Neurosci. 5 (3),

218–228.

Murthy, A., Thompson, K.G., Schall, J.D., 2001. Dynamic dissociation

of visual selection from saccade programming in frontal eye field.

J. Neurophysiol. 86 (5), 2634–2637.

O’Driscoll, G.A., Alpert, N.M., Matthysse, S.W., Levy, D.L., Rauch, S.L.,

Holzman, P.S., 1995. Functional neuroanatomy of antisaccade eye

movements investigated with positron emission tomography. Proc. Natl.

Acad. Sci. U. S. A. 92 (3), 925–929.

Paus, T., 1996. Location and function of the human frontal eye-field: a

selective review. Neuropsychologia 34 (6), 475–483.

Petit, L., Orssaud, C., Tzourio, N., Salamon, G., Mazoyer, B., Berthoz, A.,

1993. PET study of voluntary saccadic eye movements in humans:

basal ganglia-thalamocortical system and cingulate cortex involve-

ment. J. Neurophysiol. 69 (4), 1009–1017.

Petit, L., Tzourio, N., Orssaud, C., Pietrzyk, U., Berthoz, A., Mazoyer, B.,

1995. Functional neuroanatomy of the human visual fixation system.

Eur. J. Neurosci. 7 (1), 169–174.

Petit, L., Orssaud, C., Tzourio, N., Crivello, F., Berthoz, A., Mazoyer, B.,

1996. Functional anatomy of a prelearned sequence of horizontal

saccades in humans. J. Neurosci. 16 (11), 3714–3726.

Pfurtscheller, G., Lopes da Silva, F.H., 1999. Event-related EEG/MEG

synchronization and desynchronization: basic principles. Clin. Neuro-

physiol. 110 (11), 1842–1857.

Pierrot-Deseilligny, C., Muri, R.M., Ploner, C.J., Gaymard, B., Demeret, S.,

Rivaud-Pechoux, S., 2003. Decisional role of the dorsolateral prefrontal

cortex in ocular motor behaviour. Brain 126 (Pt. 6), 1460–1473.

Sakamoto, A., Luders, H., Burgess, R., 1991. Intracranial recordings of

movement-related potentials to voluntary saccades. J. Clin. Neuro-

physiol. 8 (2), 223–233.

Schall, J.D., 2002. The neural selection and control of saccades by the

frontal eye field. Philos. Trans. R. Soc. London, Ser. B Biol. Sci. 357

(1424), 1073–1082.

Schall, J.D., 2004. On the role of frontal eye field in guiding attention and

saccades. Vision Res. 44 (12), 1453–1467.

Talairach, J., Tournoux, P., 1988. Co-planar Stereotaxic Atlas of the Human

Brain. 3-Dimensional Proportional System: An Approach to Cerebral

Imaging. Thieme.

Tallon-Baudry, C., Bertrand, O., Delpuech, C., Permier, J., 1997.

Oscillatory gamma-band (30–70 Hz) activity induced by a visual

search task in humans. J. Neurosci. 17 (2), 722–734.

Yamamoto, J., Ikeda, A., Satow, T., Matsuhashi, M., Baba, K., Yamane, F.,

Miyamoto, S., Mihara, T., Hori, T., Taki, W., Hashimoto, N., Shibasaki,

H., 2004. Human eye fields in the frontal lobe as studied by epicortical

recording of movement-related cortical potentials. Brain 127 (Pt. 4),

873–887.

Intracerebral dynamics of saccade generation in the human frontal eye field and supplementary eye field

Documents