Spatio temporal Dynamics of Face and Recognition for Brain and Behaviour/19970... · famous faces...
Transcript of Spatio temporal Dynamics of Face and Recognition for Brain and Behaviour/19970... · famous faces...
Cerebral Cortex May 2008;18:997--1009
doi:10.1093/cercor/bhm140
Advance Access publication August 22, 2007
Spatio temporal Dynamics of FaceRecognition
Emmanuel J. Barbeau1, Margot J. Taylor2, Jean Regis3,4,5,
Patrick Marquis3,4,5, Patrick Chauvel3,4,5 and
Catherine Liegeois-Chauvel3,4,5
1Centre de recherche Cerveau et Cognition, Universite Paul
Sabatier Toulouse 3, Centre National de la Recherche
Scientifique, Toulouse, France, 2Diagnostic Imaging, Research
Institute, Hospital for Sick Children, Toronto, Canada, 3Institut
National de la Sante et de la Recherche Medicale U751,
Marseille, France, 4Assistance Publique - Hopitaux de Marseille
Timone, Marseille, France and 5Universite Aix-Marseille,
Marseille, France
To better understand face recognition, it is necessary to identify notonly which brain structures are implicated but also the dynamics ofthe neuronal activity in these structures. Latencies can then becompared to unravel the temporal dynamics of informationprocessing at the distributed network level. To achieve high spatialand temporal resolution, we used intracerebral recordings inepileptic subjects while they performed a famous/unfamiliar facerecognition task. The first components peaked at 110 ms in thefusiform gyrus (FG) and simultaneously in the inferior frontal gyrus,suggesting the early establishment of a large-scale network. Thiswas followed by components peaking at 160 ms in 2 areas alongthe FG. Important stages of distributed parallel processes ensuedat 240 and 360 ms involving up to 6 regions along the ventralvisual pathway. The final components peaked at 480 ms in thehippocampus. These stages largely overlapped. Importantly, event-related potentials to famous faces differed from unfamiliar facesand control stimuli in all medial temporal lobe structures. Thenetwork was bilateral but more right sided. Thus, recognition offamous faces takes place through the establishment of a complexset of local and distributed processes that interact dynamically andmay be an emergent property of these interactions.
Keywords: electrophysiology, epileptic patients, intracerebral recordings,model of face recognition, neural network
Introduction
Face recognition is a fundamental skill that has received con-
siderable attention because of its importance for social inter-
actions. Furthermore, the ability to recognize a face can be
impaired in isolation suggesting that it relies on dedicated
neural circuits (Charcot 1883; Wilbrand 1892; Bodamer 1947).
Subsequent to the report of spatially and temporally local-
izable brain responses to faces (Allison, Ginter, et al., 1994),
there has been an enormous interest in revealing the char-
acteristics of these neural circuits via various functional
neuroimaging and electrophysiological methods as well as
neuropsychological case studies (De Renzi et al. 1994; Bentin
et al. 1999; Haxby et al. 1999; McCarthy et al. 1999; Halgren
et al. 2000; Itier and Taylor 2004a). Several brain areas have
been identified as being involved in face processing, such as the
inferior occipital gyrus, the posterior fusiform gyrus (FG), or
the temporal poles (Sergent et al. 1992; Evans et al. 1995; Puce
et al. 1995; Kanwisher et al. 1997). This has led to the proposal
that face recognition relies on a distributed neural network
(Haxby et al. 2000; Ishai et al. 2005). In the temporal domain, a
negative component peaking around 170 ms (N170) is re-
corded on the scalp surface and has been shown to be
consistently larger to face stimuli (Bentin et al. 1996; Itier and
Taylor 2004b). This N170 is preceded by an earlier component
(P1) that also shows some modulation by faces (Halgren et al.
2000; Taylor et al. 2001) and is followed by later responses
around 250 and 400 ms that are modulated by face familiarity
(Bentin and Deouell 2000; Schweinberger et al. 2002).
Despite progress in identification of the neural substrate and
timing underlying face recognition, it has been difficult to
relate both those dimensions because most investigation
methods currently available allow satisfactory precision in
either the spatial distribution (positron-emission tomography
[PET], functional magnetic resonance imaging [fMRI]) or the
temporal course of the information (scalp electroencephalog-
raphy [EEG], magnetoencephalography), but not the 2 simul-
taneously. However, precise knowledge of the temporal course
of the neural activation in each brain area involved in face
recognition is critical for any model of face recognition. It is
only with this knowledge that how the different brain areas
interact can be properly understood (Nowak and Bullier 1997;
Bullier 2001; Bullier et al. 2001; Bar 2003). The aim of the
present study was to analyze the temporal dynamics of the
different brain areas involved in face recognition.
Precise measures of where and when activity occurs in the
brain can be obtained with intracranial recordings. The classic
studies by Allison, Ginter, et al. (1994), Allison, McCarthy, et al.
(1994), Allison et al. (1999), McCarthy et al. (1999), and Puce
et al. (1999) using grids on the cortical surface identified
a series of components evoked by faces. A potential peaking at
200 ms specific to faces was found in the posterior FG as well
as the posterior middle temporal gyrus regions. Other
components such as a P350 were recorded in more widespread
regions in the posterior (VP350) and anterior (AP350) ventral
temporal surface as well as the posterior lateral temporal
surface (LP350). Using intracerebral electrodes implanted within
brain structures, Halgren, Baudena, Heit, Clarke, Marinkovic
(1994) and Halgren, Baudena, Heit, Clarke, Marinkovic, Chauvel
(1994) confirmed and extended these findings identifying
a series of components evoked by a face recognition task with
a first sequence of N130--P180--N240 followed by another of
N310--N430--P630. These potentials did not have the same
spatial distribution. Some were recorded in a limited set of
brain areas, such as the 180 component in the ventral pos-
terior temporal region, whereas others were recorded from
different areas, such as the 130 and 240 components in the
temporal, parietal, and frontal lobes. Using still a different
intracerebral approach, with an electrode passing through the
grand axis of the hippocampus, Trautner et al. (2004) and Dietl
et al. (2005) focused on a potential peaking around 400 ms
� 2007 The Authors
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.0/uk/) which
permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
(facial anterior medial temporal lobe-N400) recorded in the
depth of anterior subhippocampal structures and on a potential
recorded within the hippocampus peaking around 600 ms
(hippocampal P600).
These studies confirmed that face recognition is carried out
in different brain regions and helped reveal the underlying
distributed network. The temporal dynamics of face recogni-
tion are also highlighted, with components found as early as
110 ms, up to 600 ms. However, except in a figure of ours
(Halgren and Chauvel 1993), there has been little attempt to
specifically compare the latencies of the intracerebral compo-
nents recorded from different regions with one another. Such
comparison can help in meeting our objective of providing
a comprehensive framework of the spatiotemporal dynamics of
face recognition at the whole-brain level.
In order to elicit potentials related to the recognition of
faces, a famous face/unfamiliar face paradigm was used. To
achieve simultaneous high spatial and temporal resolution, this
paradigm was presented to subjects with drug-refractory
epilepsy who had 6 --10 depth electrodes implanted stereotax-
ically orthogonal to the interhemispheric plane. Each electrode
contained 10--15 contacts along its length. It was thus possible
to record from many different brain regions simultaneously,
including some that had seldom been studied. We first
investigated the spatiotemporal dynamics of the recognition
of famous faces. We then investigated whether the recognition
of faces differed from the recognition of another type of
stimuli. Lastly, we analyzed the lateralization of face processing.
Material and Methods
Stimuli and TasksAll subjects underwent a famous face/unfamiliar face recognition task
using Eprime v1.1 (Psychology Software Tools Inc., Pittsburgh, PA).
Accuracy of the delivery of the visual stimulus and the trigger recorded
simultaneously with the EEG was controlled using a photodiode on the
screen the subject was looking at. Famous faces (actors, singers, or
politicians) and unknown faces (Fig. 1) were presented on the screen
for 396 ms in a random order, and subjects had to indicate verbally
whether they knew them. Note that our procedure was not intended to
allow assessing how the patients recognized the faces (i.e., based on
a feeling of familiarity, on semantic retrieval, or on naming). All
photographs were gray scale and corresponded to an angular size of
about 6� 3 6�. Interstimulus interval (1992 ms) was filled by a fixation
cross. Mean luminance of famous and unknown face photographs was
equivalent. There were 48 pictures of each category of faces. Overall
performance was 83% correct responses (detailed in Table 1). After
correcting for errors (false alarms and omissions) and artifact rejection,
the number of epochs per condition (famous faces correctly
recognized and unfamiliar faces correctly rejected) was equated.
Following these procedures, there were on average 38.1 (standard
deviation = 7.8) epochs used to compute event-related potentials
(ERPs) in each condition per patient.
In order to differentiate processes related to the recognition of faces
from processes related to the recognition of other kinds of stimuli,
patients also completed a visual recognition memory task in which
trial-unique abstract patterns consisting of colorful clip arts (Fig. 1)
were used as stimuli. In this task, subjects first had to learn a set of 15
stimuli and, after an interfering task of 3 min, had to recognize them
among distracters in an old/new paradigm. Patients underwent several
blocks according to their availability.
Subjects and RecordingsEighteen patients who had drug-refractory epilepsy and were un-
dergoing evaluation of possible surgical intervention were studied.
Stereoelectroencephalographic (SEEG) recording was performed in
order to define the epileptogenic zone (Talairach and Bancaud 1973).
The choice of electrode location was based on pre-SEEG clinical and
video-EEG recordings and made independently of the present study.
This study did not add any invasive procedure to depth EEG recordings.
All subjects were fully informed about the aim of investigation before
giving consent. Subjects had from 6 to 10 intracerebral electrodes
implanted stereotaxically orthogonal to the midline vertical plane. Each
electrode was from 33.5- to 51-mm long, had a diameter of 0.8 mm, and
contained from 10 to 15 contacts 2-mm long separated by 1.5 mm
(Alcis, Besancxon, France). Seventy-two to 126 intracerebral contacts
were simultaneously recorded in each patient.
ERP recordings were part of the functional mapping procedure
(language, memory, and vision depending on the epilepsy) carried out
in each subject. Anticonvulsant therapy was reduced or withdrawn
Figure 1. Examples of stimuli used in this study. Top: famous and unknown faces. Bottom: trial-unique abstract stimuli.
998 Spatio temporal Dynamics of Face Recognition d Barbeau et al.
during the SEEG exploration in order to facilitate seizures. However, no
subject had seizures in the 12 h before ERP recordings. Signal was
acquired using SynAmps amplifiers and Neuroscan software (Compu-
medics, El Paso, TX). The sampling frequency of EEG depth recording
was 1000 Hz with an acquisition filter band-pass of 0.15--200 Hz.
Reference was a surface electrode located in Fz. Implantation of
electrodes, localization of epilepsy, and behavioral performance on the
famous/unknown face recognition task are reported in Table 1.
ERP ProcessingERPs were computed off-line using Brainvision v1.05 (Brain Products
GmbH, Munchen, Germany) between 0 and 1500 ms with a prestimulus
baseline of 150 ms. ERPs to famous faces and unknown faces as well as
old/new abstract figures were computed after the rejection of
omissions and false alarms (behavioral data were lost for 4 subjects
due to a hard disk failure. In this case, all responses were considered
correct, Table 1). Visual inspection of the EEG as well as artifact-
rejection procedures enabled the rejection of periods with interictal
activities. A filter with a high band pass of 0.53 Hz (time constant = 0.30,
12 dB/oct) and a low band pass of 40.00 Hz (12 dB/oct) was applied to
the resulting ERPs.
Statistical AnalysesThere were from 5 to 7 subjects’ data per brain area, so a nonparametric
matched-paired signed-rank Wilcoxon test on individual averaged ERPs
was used to assess significant differences. The Wilcoxon test was run
on each millisecond in the 100--600 time windows. Performing tests on
multiple time points increases the probability of a false positive. We
therefore only considered differences significant that were present for
10 consecutive time points (Molholm et al. 2006). In contrast, gaps of
less than 10 ms between adjacent periods were considered non-
physiological and the periods were combined.
Location of the Electrode ContactsContact location has been described in detail elsewhere (Barbeau,
Wendling, et al. 2005). Each electrode was inserted stereotaxically
under general anesthesia. A routine postoperative computerized
tomography (CT) scan without contrast was performed to check for
the absence of bleeding and for the precise location of each contact.
Magnetic resonance imaging (MRI) with 3-dimensional reconstruction
was performed after the removal of the electrodes in order to locate
the trace of each depth probe in the brain. The fusion of the
postoperative CT scan with this MRI allowed precise anatomical
localization of contacts. The distance from the midline vertical plane of
a given contact could be calculated on the axial CT scan. After the trace
of the electrode had been found on the postoperative axial MRI, the
distance from the midline vertical plane could be determined and the
position of the contact could then be viewed in the coronal or sagittal
plane. After this procedure was completed, the anatomical structures in
which contacts were located were identified using 3-dimensional
verification.
Results
More than 2000 sites were recorded in 18 subjects in total, an
average of 111 sites recorded simultaneously per subject. ERPs
to famous faces recorded from these contacts constitute the
basis of this study.
ERP Identification
A major aim of the present study was to identify the pattern of
ERPs underlying the temporal processing of face recognition.
We thus analyzed the ERPs region by region to identify those
that met the 2 criteria: 1) exhibiting a similar waveform and
latencies in the same region across and 2) being generated
focally. Criteria for local generation were polarity reversal
between adjacent contacts and/or steep voltage gradient and
high voltage fluctuations between adjacent contacts (Halgren,
Baudena, Heit, Clarke, Marinkovic 1994; Fernandez et al. 1999).
A supplementary criterion was to reach a threshold of 5 sub-
jects per region across our population of 18 patients in order to
be able to perform statistical comparisons at least 5 subjects.
ERPs corresponding to the criteria defined above were
identified in 7 regions: the posterior fusiform gyrus (FG), the
middle FG 1--2 cm anterior to the previous region, the posterior
parahippocampal gyrus, the perirhinal cortex within the
collateral sulcus, the medial temporal pole, the hippocampus
(posterior and anterior), and the inferior frontal gyrus. Figure 2
depicts all brain regions explored and shows in black the 7
regions from which the focal ERPs reproducible across subjects
were identified. Focal ERPs were recorded from other brain
regions (shown in grey), for example, from the amygdala, the
orbitofrontal cortex, the lateral temporal lobe (LTL) along the
Table 1Subjects, clinical details, and epileptic data
Subject Sex Age Handedness Languagelateralization
Epileptic focus Etiology Number ofelectrodes
Number ofcontacts
Analyzed in this study(r: right, l: left)
Taskperformance (%)
1 M 23 L L R-LTL CD 10 120 rpFG, rPG, rPC, rTP, rIFG 862 M 24 L L L-MTL OGD 6 78 — 673 M 32 R L L-MTL CD 8 90 lTP, lH 894 M 34 R R L-MTL HS 10 126 lPG, lPC, lTP, lH, lIFG 985 M 43 R L R-LTL Unknown 9 120 rPG, rPC, rTP, rH, rIFG n/a6 F 35 L B R-MTL CD 9 117 — n/a7 F 33 R L L-MTL Unknown 8 114 rPC 848 M 33 R L R-PF CD 10 120 — 909 M 36 R L R&L-MTL Unknown 6 72 lH 9510 M 29 R L R-LOE CD 10 109 lpFG n/a11 M 35 R L L-MTL HS 8 110 lmFG n/a12 M 42 L L R-MTL Unknown 9 117 rmFG, rPG, rTP, rH, rIFG 8913 M 25 R B R-MTL Unknown 9 114 rPG, lPC, lTP, rH, rIFG 6914 F 26 L L L-LOE Unknown 9 120 lpFG, lmFG 8115 F 24 R L L-LTL-LOE Unknown 10 121 rPC 7216 F 48 A L L&R-PA Unknown 9 123 rpFG, rmFG, lPG, rTP 7517 F 41 R L R-LOE CD 8 120 rpFG, rmFG 9618 M 56 R L R-MTL HS 8 115 rIFG 76
Note: R, right; L, left; A, ambidextrous; MTL, medial temporal lobe epilepsy; LTL, lateral temporal lobe epilepsy; PF, prefrontal lobe epilepsy; LOE, lateral occipital epilepsy; PA, parietal lobe epilepsy;
pFG, posterior fusiform gyrus; mFG, middle fusiform gyrus; PG, posterior parahippocampal gyrus; PC, perirhinal cortex; TP, temporal pole; H, hippocampus; IFG, inferior frontal gyrus; r, right; l, left; CD,
cortical dysplasia; OGD, oligodendroglioma; HS, hippocampal sclerosis; n/a, non available.
Cerebral Cortex May 2008, V 18 N 5 999
superior temporal sulcus, the anterior insula, or the dorsolat-
eral frontal lobes, but did not fulfill the criteria mentioned
above and so were not analyzed further.
Spatiotemporal Dynamics of Face Recognition
Figure 3 shows ERPs averaged across patients recorded from
the posterior to anterior regions mentioned above in response
to correctly recognized famous faces.
These results are also reported for a representative subject
(subject 1) in whom most brain areas of interest in this study
were recorded simultaneously, allowing an intrasubject com-
parison (Fig. 4). It demonstrates that the different components
reported in the group analysis are robust as they can also be
observed at the individual level.
The ERPs recorded from the posterior and middle FG were
characterized by a N110--P160--N240 with a mean onset at ~80ms. The N110, but not the P160, peaked slightly earlier in the
posterior than the middle region. The P160 could be recorded
from both the posterior and the middle FG simultaneously in
some subjects and sometimes from the lateral occipitotemporal
cortex, suggesting the involvement of several occipitotemporal
brain areas.
A delayed N240--P300--N360 triphasic complex was
recorded from mesial structures, notably from the posterior
parahippocampal gyrus, the perirhinal cortex, and the medial
temporal pole (Fig. 3). As shown in the intrasubject compar-
ison, this complex was also recorded from other brain areas
such as the lingual gyrus (Fig. 4), suggesting that it is a common
complex that may concern more regions than those detailed in
this study. Despite the similar morphology of this triphasic
complex, differences in onset and amplitude were observed.
The N240 was prominent in the perirhinal cortex and started
earlier in this region, whereas the N360 was prominent in the
temporal pole.
ERPs to faces recorded from the hippocampus had a quite
different waveform from the other structures, displaying a large
and slow positive component starting at ~160 ms, peaking at
480 ms, and finishing around 750--800 ms. This distinct pattern
suggests that the hippocampus plays a different role in face
processing.
Finally, a low-amplitude negative component (–19 mV)
peaking at 117 ms was recorded from the posterior inferior
frontal gyrus and displayed an interesting analogy with the
N110 recorded from the FG. They were recorded in both
regions in 2 subjects. The N110 from the inferior frontal gyrus
was slightly delayed compared with the N110 from the FG
(Fig. 5). Also, the onset of the N240 recorded in the perirhinal
cortex was found to coincide with the peak of the N110 in
both the group study and the intrasubject comparison (Figs 3
and 4).
In summary, we demonstrated early components peaking at
110 ms poststimulus not only in the FG but also in the inferior
frontal gyrus, a region not usually included in early visual
processing. The N110 is followed by 2 stages of widespread
parallel processing peaking at 240 and 360 ms. Given the
limited spatial sampling of intracerebral electrodes, it is
probable that still other regions participate in this network.
The 240- and 360-ms stages involved many aspects of the visual
ventral stream. In contrast, none of these components were
recorded in the hippocampus indicating that this region is
clearly involved in a different type of processing. These
components do not index independent serial stages but largely
overlap. The P160 is encompassed within the rise of the N240.
Likewise, the N240 and N360 components are encompassed
within the rise of the P480. These results suggest strong
interactions between these stages and, together with the
finding that an early N110 could be recorded in the inferior
frontal gyrus, argue against sequential processing stages.
Rather, they suggest that face recognition may be an emergent
property of these interactions.
Face Recognition
In this section, we compare ERPs to famous and unfamiliar
faces for each time point between 100- and 600-ms post-
stimulus onset using a nonparametric matched-pair signed-
rank Wilcoxon test on individual averaged ERPs (Fig. 6).
No difference between these conditions was found in the
posterior or middle FGs or in the inferior frontal gyrus.
However, significant differences (P < 0.05) were found in all
medial temporal lobe (MTL) regions that were investigated,
starting at 156 ms in the perirhinal cortex (156- to 216-ms
poststimulus onset) followed by the posterior parahippocampal
gyrus (220--248 and 268--326 ms), the medial temporal pole
(315--426 and 450--510 ms), and the hippocampus (322--541
ms). This effect was prominent in the temporal pole and the
hippocampus where 100% of the subjects showed a difference
during some periods (338--406 ms, n = 7, in the temporal pole
and 372--465 ms, n = 6, in the hippocampus). Famous faces
Figure 2. Schematic representation of all brain areas explored. Focal ERPs were recorded from different regions (in gray and black) but were found reliably with reproduciblemorphology at similar latencies only in the areas depicted in black. ERPs recorded in the areas in gray did not meet study criteria. No ERPs were recorded from the areas in white.
1000 Spatio temporal Dynamics of Face Recognition d Barbeau et al.
showed higher amplitudes in all these regions except in the
posterior parahippocampal gyrus where unfamiliar rather than
famous faces showed higher amplitude.
The difference between familiar and unfamiliar faces was
found in the mesial structures for the components N240, N360,
and P480. Together with the finding that the N240 and N360
reflect parallel processing in many different regions, this is
a further indication that recognition may be a property of
a network of activated brain regions rather than depend on a
single region. All these medial structures play a well-known
role in declarative memory, and the different responses in these
regions probably reflect the various aspects of recognition such
as familiarity detection and access to long-term semantic or
episodic memory.
Face Recognition versus Recognition of Other Stimuli
These data demonstrate that a network of brain areas is
involved in face recognition. The question we address in this
section is whether this neural system is a general system
involved in the recognition of any kind of stimulus. Thus, we
compared the ERPs to the famous/unfamiliar faces with the
ERPs to familiar and unfamiliar stimuli in a visual recognition
memory task, in the same series of patients. In this task, the
patients learned trial-unique abstract patterns (Fig. 1) during an
encoding phase, followed a 3-min interfering phase, and then
underwent a recognition phase. The patients had to recognize
the familiar stimuli (learned during the encoding phase) and
reject the unfamiliar stimuli (trial-unique distracters). We did
not try to match faces and the abstract stimuli on any low-level
visual characteristic, our goal in this section being to compare
the recognition processes involved in the 2 tasks.
ERPs to recognized abstract targets are shown in Figure 7.
Early components to familiar abstract patterns and famous faces
are not compared as visual characteristics of the 2 sets of
stimuli differ too much. However, note that the N110 and P160
are of strikingly similar amplitude in the middle FG indicating
a similar level of activity. In contrast, the amplitude to abstract
patterns was significantly lower in all MTL structures, whereas
the difference did not reach significance in the inferior frontal
gyrus.
Computing the difference between familiar and unfamiliar
abstract patterns yielded very different results than for faces as
Figure 3. Comparison of the averaged ERPs to famous faces across regions using a color-coding scheme. The vertical lines are drawn to facilitate comparison of the differentpeaks.
Cerebral Cortex May 2008, V 18 N 5 1001
differences were mainly found in both regions of the FG during
the N160 and N240 (matched-pair signed-rank Wilcoxon test,
P < 0.05). On the other hand, no differences were found in the
perirhinal cortex, hippocampus, and inferior frontal gyrus.
Differences during relatively late periods were seen in the
posterior parahippocampal gyrus (308--322 ms poststimulus)
and medial temporal pole (443--495 ms).
Abstract patterns elicited clearly formed ERPs in all brain
regions in which ERPs to faces were recorded (Fig. 7) with
similar morphologies. Amplitude between the categories of
stimuli was comparable in the middle FG and the inferior
frontal gyrus. In this sense, the network involved in the
recognition of faces is also involved in the recognition of
abstract patterns.
However, significant amplitude differences were found in
the posterior FG and all MTL structures, indicating a degree of
specificity for faces possibly related to additional processes. For
example, the posterior FG shows some early specificity to faces
as areas in this region appear to be preferentially involved in
the processing of faces over other kinds of stimuli (Puce et al.
1995; Kanwisher et al. 1997). Likewise, the amplitude differ-
ences found in MTL structures could be related to the fact that
famous faces carry strong semantic content, whereas abstract
Figure 4. Intrasubject comparison of ERPs to famous faces across regions (subject 1, right hemisphere). ERPs recorded in the lingual gyrus and lateral temporal pole arepresented to demonstrate that the N240--P300--N360 complex and N360 may be even more common than reported in the group study.
Figure 5. ERPs to famous faces recorded from the inferior frontal gyrus and the FG(posterior FG in subject 1, middle FG in subject 12). The N110 peaks slightly earlier inthe FG. Note the different amplitude scales.
1002 Spatio temporal Dynamics of Face Recognition d Barbeau et al.
patterns were newly learned stimuli and carry little semantic
content. In support of this hypothesis, a striking finding was
that a recognition effect for faces was found in all MTL
structures, whereas this effect was considerably smaller for
abstract patterns. In contrast, a recognition effect was seen
mainly in the FG for abstract patterns, whereas no such effect
was found for faces. These results further indicate that
recognition per se can rely on many brain regions and that
this may differ with stimulus category and processes involved.
Hemispheric Asymmetries
In our previous sections, we combined data from both left and
right hemispheres to focus on the general network underlying
face recognition and because data per region were obtained
from small samples (the fact that it was usually unilateral
implantation and the small number of patients per group
prevented direct right/left statistical comparisons). However,
numerous studies have found greater activity for faces in the
right hemisphere. In order to investigate this aspect, the same
data were analyzed for each hemisphere.
As can be observed from Figure 8, both hemispheres are
involved in face recognition. Further, the neuronal populations
involved in face recognition perform comparable processes as
ERPs with similar morphology can be found in both hemi-
spheres. There are 2 notable exceptions, however. First, a large
N240 was observed in the middle FG on the right, whereas this
component was virtually absent on the left. An N240 larger in
the right than in the left was also observed in the perirhinal
cortex. Second, the N110 was very small in the left inferior
frontal gyrus (in one patient, however) compared with the
N110 in the right. Some areas thus appear to be more specific
on the right than on the left. A notable exception was the
finding that the ERP to famous faces showed higher amplitudes
on the left than the on the right in the temporal pole.
Furthermore, 50% of the 18 patients included in this study
had right-hemisphere implantation (left: 39%, bilateral: 11%).
For all regions, we found more ERPs meeting our criteria (focal
and similar morphology across subjects) on the right hemi-
sphere than on the left, that is, the proportion of local ERPs
retained for further analysis was higher on the right than
expected by the number of right implantations (posterior and
middle FG: 60%, posterior parahippocampal gyrus and peri-
rhinal cortex: 67%, medial temporal pole: 57%, hippocampus:
67%, inferior frontal gyrus: 83%). This difference was maximal
in the inferior frontal gyrus with 5 ERPs found in the right
hemisphere for 1 ERP found in the left hemisphere.
Figure 6. ERPs to famous and unfamiliar faces. Asterisk indicates P\ 0.05. For clarity, peak labels are indicated for some brain areas only.
Cerebral Cortex May 2008, V 18 N 5 1003
Consistent with a large literature (Sergent et al. 1992; Allison,
Ginter et al. 1994; Allison et al. 1999; Puce et al. 1996;
Kanwisher et al. 1997; McCarthy et al. 1999; Haxby et al. 1999;
Rossion et al. 2000; Ishai et al. 2005), these data demonstrate
that face recognition relies on a bilateral network with a right
dominance. This was true not only for posterior, visuopercep-
tive regions but also for middle and anterior regions that may
be more involved in memory processing (e.g., posterior para-
hippocampal gyrus, perirhinal cortex, and inferior frontal gyrus).
Discussion
The time course of face recognition processes has been studied
extensively using surface electrophysiological methods. How-
ever, it has generally been difficult to relate these activities to
precise brain regions due to the poor spatial resolution of these
techniques. The brain areas involved in face recognition have
been extensively studied using fMRI and PET, but these
techniques have poor temporal resolution. Here, we meet
these limitations using intracerebral recordings, allowing the
assessment of the precise temporal course of the electrophys-
iological activity from ‘‘within’’ the brain regions involved in
face recognition. We recorded from a large sample of brain
regions across a group of 18 patients and report only the
electrophysiological activities that were similar in morphology
and latency across patients. The brain regions and electrophys-
iological activities identified using this approach are thus
hypothesized to be representative of the main regions and
activities involved in face recognition in the general population.
From these data, we propose a model of the electrophysiolog-
ical activities indexing face recognition, with the particularity
that they are detailed simultaneously in both time and space.
Figure 9 shows a schematic representation of this model,
which confirms and extends the current knowledge on face
recognition. 1) An early (110-ms poststimulus onset) electro-
physiological activity can be recorded not only from the
posterior regions but also from the inferior frontal gyrus. 2)
Around 160 ms, there is activation in both the middle and the
posterior FGs. 3) A stage of massive parallel processing occurs
around 240 ms poststimulus, involving different areas in the
visual ventral stream and closely followed by a second stage
of parallel processing around 360 ms. 4) Long-lasting electro-
physiological activities are recorded from the hippocampus
that appear unrelated to the activities recorded from the visual
ventral stream. 5) There are strong interactions between these
stages because some are encompassed in others, for example,
Figure 7. ERPs to (old) recognized abstract patterns. Same patients as in Figure 5. Left and right hemispheres combined.
1004 Spatio temporal Dynamics of Face Recognition d Barbeau et al.
the N160 on the rise of the N240. 6) Recognition effects
between famous and unfamiliar faces are found in all 4 MTL
regions investigated, suggesting that these effects are distrib-
uted. On the other hand, recognition of abstract patterns mainly
involved posterior regions where no such effect was found for
faces. Overall, these data support the view that face recognition
relies on a distributed network of brain areas, involving both
hemispheres but more prominent in the right, and illustrate the
dynamics of the information flow among these areas.
The finding that face recognition is mediated by a distrib-
uted network of brain structures is in accordance with several
electrophysiological and fMRI studies (Puce et al. 1999;
Leveroni et al. 2000; Paller et al. 2000; Ishai et al. 2002,
2004, 2005; Ishai and Yago 2006). Leveroni et al. (2000)
reported activations in parietal and frontal brain regions for
newly learned faces compared with trial-unique faces. These
authors also compared the recognition of famous faces with
the recognition of these newly learned faces. They found an
even larger network to the famous faces, including areas
in the LTL and MTL, the frontal, and parietal lobes as well as
in cingular and extrastriate structures. This network was
Figure 8. ERPs to famous faces recorded from the left and right hemispheres. For clarity, peak labels are indicated for some brain areas only.
Figure 9. Schematic model of the main electrophysiological activities underlying facerecognition in different brain regions. Components’ names are indicated on theabscissa. The rectangles indicate the main peaks of the components in each region.The horizontal lines provide an idea of the components’ onsets. The rectangles in darkblue indicate the peaks during which a recognition effect was found.
Cerebral Cortex May 2008, V 18 N 5 1005
hypothesized to reflect participation of these regions in long-
term memory.
We identified a distributed cortical network of at least 7
structures involved in face recognition. These included the
posterior and middle FGs, posterior parahippocampal gyrus,
perirhinal cortex, medial temporal pole, hippocampus, and
inferior frontal gyrus. Haxby et al. (2000) reported that face
processing is mediated by a distributed neural system, which
consists of a core and an extended system. The core system
includes the inferior occipital gyrus and FG and the superior
temporal sulcus. The FG is proposed to process invariant
aspects of faces, whereas the superior temporal sulcus would
process changeable aspects such as eye or mouth movements.
The extended system comprises several regions depending on
the information extracted from the face. Pertinent to the
present work, person identification would depend on anterior
temporal structures. The neural network identified in our study
largely supports the model of Haxby et al. as it includes both
the FG as well as anterior temporal lobe structures such as the
perirhinal cortex and temporal pole, expected when subjects
process famous faces. That the perirhinal cortex participates in
face recognition has previously been hypothesized by other
authors (Halgren, Baudena, Heit, Clarke, Marinkovic 1994;
Allison et al. 1999; Trautner et al. 2004). Our data confirm this
assumption as they are the first direct recordings from this area.
Components were also identified in the superior temporal
sulcus in our study, but not reliably across subjects. This
appears consistent with the fact that subjects did not have to
process changeable aspects of the face to perform the task. In
an fMRI study, Ishai et al. (2005) found that face perception is
mediated, independently of the nature of the task, by a cortical
network that includes the inferior occipital gyrus, FG, superior
temporal sulcus, hippocampus, amygdala, inferior frontal gyrus,
and orbitofrontal cortex. Interestingly, this study as well as
others (Leveroni et al. 2000; Ishai et al. 2002) report on 2 brain
structures, the hippocampus and the inferior frontal gyrus,
which we also identified but which were not reported in the
model of Haxby et al. On the other hand, anterior temporal lobe
structures are not reported in Ishai’s study, which is probably
related to the methodology used. These authors searched for
areas activated through various face tasks, rather than
exclusively tasks of face recognition. In contrast, our study
confirms that anterior lobe structures are part of a complemen-
tary system specifically involved in face recognition. Note that
Ishai et al. (2005) also reported on several brain areas that were
not identified in our study (inferior occipito gyrus, amygdala,
and orbitofrontal cortex). This discrepancy is probably related
to our study criteria to report only ERPs that were comparable
in at least 5 subjects and as such does not refute the possibility
that these other brain regions may also contribute to the
network involved in face recognition. It remains, however, to
characterize the temporal course of the neuronal activity in
these regions.
Although in most of the above studies activations were ob-
served in different brain regions, several pieces of information
critical to characterize a network were missing. These include
the notion of connectivity (is there a relation between the
different brain areas?), the direction or causality between these
areas, and the time when these relations occur. Fairhall and
Ishai (2006) recently attempted to meet these issues using
dynamic causal modeling on fMRI data while subjects
processed emotional and famous faces. They found that the
core system is mediated by a feed-forward architecture with
the inferior occipital gyrus exerting influence on the FG and
superior temporal sulcus. Furthermore, the FG was found to
exert a strong causal influence on the orbitofrontal cortex
when processing famous faces and on the amygdala and
inferior frontal gyrus when processing emotional faces.
However, the timing of the FG influence on other brain regions
was not determined and could occur either during the N110,
P160, or the N240, which can be recorded in this region.
The present study identified an N110 in the FG; an N110 was
also identified in the inferior frontal gyrus. However, it was
shown that this component was slightly delayed compared
with the N110 recorded in the posterior FG. It may be
compatible with the notion that rapid feed forward from the
FG occurs during this period, although an alternative is that
a third region (i.e., the lateral occipital gyrus) projects to both
simultaneously. The involvement of prefrontal cortex in face
processing has already been reported in primates (Wilson et al.
1993) and humans in intracranial and fMRI studies (Halgren,
Baudena, Heit, Clarke, Marinkovic, Chauvel, 1994; Allison et al.
1999; Vignal et al. 2000; Ishai et al. 2002, 2005). Marinkovic
et al. (2000) described large ERPs selective to faces in the right
anterior prefrontal region beginning 150 ms after stimulus
onset. However, although an N110 recorded from the FG
had already been reported (Halgren, Baudena, Heit, Clarke,
Marinkovic, Chauvel, 1994; Allison et al. 1999), this is the first
time that a component as early as 110 ms is reliably reported in
the inferior frontal gyrus.
Besides the N110, the causal influence of the FG on other
brain regions could also occur during the P160, which was
recorded in several posterior regions. This component prob-
ably corresponds to the P180 recorded using intracerebral
electrodes (Halgren, Baudena, Heit, Clarke, Marinkovic, 1994)
or to the N200 recorded on the cortical surface (Allison,
McCarthy, et al. 1994; Allison et al. 1999) as well as to the N170
recorded from the scalp surface. The polarity inversion
between intracerebral and surface recordings is related to the
fact that different sides of the generators are captured. The
N110--P160--N240 intracerebral complex thus probably corre-
sponds to the P1--N170--P2 components recorded from the
brain surface. (Note, however, that the N240--P300--N360 that
we also report is probably not polarity inverted and corre-
sponds, at least partly, to the N400 recorded from the surface.
In this case, it is the component (P350) reported by Allison
et al. (1999) which was polarity inverted because they
recorded from the temporal ventral surface and not from
above the source (dipole) as we did or as is done with surface
recordings.) The fact that the P160 was seen at several
occipitotemporal regions is in agreement with previous studies
which showed that this component could be recorded from
a large area covering the ventral occipital and posterior
temporal region (Allison, McCarthy, et al. 1994; Halgren,
Baudena, Heit, Clarke, Marinkovic 1994; Allison et al. 1999),
including simultaneously in different regions in the same
subject. Interestingly, Klopp et al. (2000) showed that during
160- to 230-ms poststimulus onset, a face-selective increase in
coherence was found between the FG and regions in the
temporal, parietal, and frontal lobes. This mechanism may
underlie the causal influence found by Fairhall and Ishai (2006).
The fact that the fusiform P160 was encompassed in the rise of
the N240 components found in different regions corroborates
this analysis.
1006 Spatio temporal Dynamics of Face Recognition d Barbeau et al.
On the other hand, the N110 was also found to be
synchronous with the onset of the N240 in the perirhinal
cortex. These data suggest that the N110 could be a signal
triggering further processing. Specifically, the frontal N110
could be a top-down signal on the visual ventral stream. Such
a mechanism has been postulated to be necessary to allow
more accurate and faster computation (Humphreys et al. 1997;
Tomita et al. 1999; Adolphs 2002; Bar 2003; Bar et al. 2006). Bar
(2003) proposed that low spatial frequencies of an image could
be projected rapidly through the magnocellular pathway from
occipital regions to the inferior frontal gyrus in order to trigger
expectations about the identity of the image itself. This top-
down modulation has been proposed to occur around 250 ms
(Ishai et al. 2006), although the current data suggest that this
can happen considerably earlier. This rapid activation could
then have projections on the visual ventral stream, possibly on
the perirhinal cortex, for early activations.
Furthermore, a N240--P300--N360 complex following the
P160 was observed in different brain regions. In the perirhinal
cortex and temporal pole, this complex may be equivalent to
the facial AMTL-N400 described by Trautner et al. (2004) and
Dietl et al. (2005). It has a comparable latency and polarity,
a similar topography in anterior subhippocampal structures,
elicits a component of higher amplitude to famous faces in the
300- to 600-ms time frame, and is composed of 2 subpeaks (see
Fig. 1a in Dietl et al. 2005 for example). However, the data
reported here indicate that the AMTL-N400 is not a single
component but is clearly divided into 2 subcomponents. A
further important finding of our study is that the N240 and the
N360 were observed in different regions (>5 or 6) where they
were generated locally, suggesting parallel processing. These
regions were located in the visual ventral pathway (Ungerleider
and Mishkin 1982; Ungerleider and Haxby 1994), although the
term pathway is improper here because simultaneous, rather
than serial, activation was observed. Therefore, along with feed-
forward activations, phases of possible top-down signal and
parallel processing are also identified in our study.
The recognition effects found in the present study were all
in MTL structures. This is in accordance with a study that
identified single neurons with invariant representations of
famous faces in several regions of the human MTL including the
amygdala, entorhinal cortex, parahippocampal gyrus, and
hippocampus (Quian Quiroga et al. 2005). It is important to
note that recognizing a face can mean different things: having
a sense of familiarity with a face, activating semantic
knowledge related to a person who has been recognized, or
retrieving the name of that person. In this study, a sense of
familiarity with the face was enough to perform the task, but
complementary processing related to identification may have
been performed. Along this line of thought, we hypothesize
that the activities recorded in MTL structures primarily reflect
access to different aspects of long-term memory. A putative
model could be that the earliest signal from the perirhinal
cortex during the N240 could signal that the stimulus is
familiar, based mainly on the visual characteristics of the
stimulus. This would be congruent with the role of this struc-
ture in visual processing and familiarity detection (Aggleton
and Brown 1999; Murray and Bussey 1999; Barbeau, Felician,
et al. 2005; Barense et al. 2005). The N360 amplitude to famous
faces was higher than to unfamiliar faces in the temporal pole
and, although seen bilaterally, was more prominent in the left
hemisphere. It may indicate access to a distributed semantic
system involving person identification and attempts to name
the faces (Joubert et al. 2006). On the other hand, because the
P480 recorded from the hippocampus showed a component
unrelated to those of the visual ventral pathway, a possibility is
that this component is related to a different system, for
example, episodic (personal) memory. We also suggest that the
MTL is where the major recognition effects are found as the
unique nature of famous faces can trigger memory retrieval. In
accordance with this hypothesis, we found that ERPs to famous
faces showed higher amplitude than to recently learned
abstract patterns. These control stimuli carry little semantic
content or relation to episodic memory, which may explain
why they elicit less neuronal activation in this brain region.
There is a continuing debate over the period during which
recognition effects (i.e., differences between familiar and
unfamiliar faces) can be found. The most robust effects are
seen 250- to 500-ms poststimulus onset when comparing
famous and unfamiliar faces in scalp recordings (Bentin and
Deouell 2000; Eimer 2000; Jemel et al. 2003). The N170, in
contrast, was not modulated, implying that posterior regions
are relatively insensitive to the ‘‘famous’’ factor. In accordance
with these studies, no difference between famous and
unknown faces was observed during the P160 in our study
in either the posterior or the middle FG. Indeed, the
intracranial face-specific P160 has been shown to reflect
a mandatory and invariant processing stage unaffected by
passive viewing of faces, face familiarity, face habituation, or
semantic priming (Puce et al. 1999). This led to the view that
the surface N170 or its intracerebral counterpart may reflect
operations carried out by a ‘‘structural encoding module’’
(Bruce and Young 1986) in charge of automatically processing
physiognomic aspects of the faces (Bentin et al. 1996).
However, recent findings using scalp recordings indicate that
other kinds of familiar faces such as personally known faces
(own face, family members’ faces, or friends’ faces) could
evoke larger N170s (Caharel et al. 2006). The N170 has also
been shown to be modulated by face repetition (Itier and
Taylor 2004a), and we further found in this study that the P160
was modulated by recently learned abstract patterns. These
results emphasize the fact that the mechanisms underlying
recognition could be different for famous faces than for other
types of faces or other types of visual stimuli. In particular,
recognition of repeated, newly learned, or personally known
faces can easily rely on visual clues because they have been
seen repeatedly and/or recently. On the other hand, the
photographs of famous faces are usually seen for the first time
during the experiments, implying that recognition may have
to rely more on person’s identification than on the visual
characteristics of the face.
Funding
Centre National de la Recherche Scientifique postdoctoral
fellowship to E.B.
Notes
EB was supported by a CNRS postdoctoral fellowship. The authors
wish to thank Dr Aileen McGonigal and Dr Fabrice Bartolomei for their
help with the manuscript as well as 2 anonymous reviewers for their
constructive remarks. Conflict of Interest : None declared.
Address correspondence to Emmanuel J. Barbeau, Centre de recherche
Cerveau et Cognition, UMR 5549, Centre National de la Recherche
Cerebral Cortex May 2008, V 18 N 5 1007
Scientifique, Universite Paul Sabatier Toulouse 3, Faculte de Medecine de
Rangueil, 31062 Toulouse Cedex 9, France. Email: emmanuel.barbeau@
cerco.ups-tlse.fr.
References
Adolphs R. 2002. Neural systems for recognizing emotion. Curr Opin
Neurobiol. 12:169--177.
Aggleton JP, Brown MW. 1999. Episodic memory, amnesia, and the
hippocampal-anterior thalamic axis. Behav Brain Sci. 22:425--444.
Allison T, Ginter H, McCarthy G, Nobre AC, Puce A, Luby M,
Spencer DD. 1994. Face recognition in human extrastriate cortex.
J Neurophysiol. 71:821--825.
Allison T, McCarthy G, Nobre A, Puce A, Belger A. 1994. Human
extrastriate visual cortex and the perception of faces, words,
numbers, and colors. Cereb Cortex. 4:544--554.
Allison T, Puce A, Spencer DD, McCarthy G. 1999. Electrophysiological
studies of human face perception. I: potentials generated in occipito-
temporal cortex by face and non-face stimuli. Cereb Cortex. 9:415--430.
Bar M. 2003. A cortical mechanism for triggering top-down facilitation
in visual object recognition. J Cogn Neurosci. 15:600--609.
Bar M, Kassam KS, Ghuman AS, Boshyan J, Schmidt AM, Dale AM,
HamalainenMS,MarinkovicK, SchacterDL,RosenBR. 2006. Top-down
facilitation of visual recognition. Proc Natl Acad Sci USA. 103:449--453.
Barbeau E, Wendling F, Regis J, Duncan R, Poncet M, Chauvel P,
Bartolomei F. 2005. Recollection of vivid memories after perirhinal
region stimulations: synchronization in the theta range of spatially
distributed brain areas. Neuropsychologia. 43:1329--1337.
Barbeau EJ, Felician O, Joubert S, Sontheimer A, Ceccaldi M, Poncet M.
2005. Preserved visual recognition memory in an amnesic patient
with hippocampal lesions. Hippocampus. 15:587--596.
Barense MD, Bussey TJ, Lee AC, Rogers TT, Davies RR, Saksida LM,
Murray EA, Graham KS. 2005. Functional specialization in the
human medial temporal lobe. J Neurosci. 25:10239--10246.
Bentin S, Allison T, Puce A, Perez E, McCarthy G. 1996. Electrophysiological
studies of face perception in humans. J Cogn Neurosci. 8:551--565.
Bentin S, Deouell LY. 2000. Structural encoding and identification in
face processing: ERP evidence for separate mechanisms. Cogn
Neuropsychol. 17:35--54.
Bentin S, Deouell LY, Soroker N. 1999. Selective visual streaming in face
recognition: evidence from developmental prosopagnosia. Neuro-
report. 10:823--827.
Bodamer J. 1947. Die Prosopagnosie. Arch Psychiatr Nervenkr. 179:6--53.
Bruce V, Young A. 1986. Understanding face recognition. Br J Med
Psychol. 77:305--327.
Bullier J. 2001. Integrated model of visual processing. Brain Res Brain
Res Rev. 36:96--107.
Bullier J, Hupe JM, James AC, Girard P. 2001. The role of feedback
connections in shaping the responses of visual cortical neurons.
Prog Brain Res. 134:193--204.
Caharel S, Fiori N, Bernard C, Lalonde R, Rebai M. 2006. The effects of
inversion and eye displacements of familiar and unknown faces on
early and late-stage ERPs. Int J Psychophysiol. 62:141--151.
Charcot J. 1883. Un cas de suppression brusque et isolee de la vision
mentaledes signeset desobjets formes et couleurs. ProgrMed. 11:568.
De Renzi E, Perani D, Carlesimo GA, Silveri MC, Fazio F. 1994.
Prosopagnosia can be associated with damage confined to the right
hemisphere-an MRI and PET study and a review of the literature.
Neuropsychologia. 32:893--902.
Dietl T, Trautner P, Staedtgen M, Vannuchi M, Mecklinger A,
Grunwald T, Clusmann H, Elger CE, Kurthen M. 2005. Processing
of famous faces and medial temporal lobe event-related potentials:
a depth electrode study. Neuroimage. 25:401--407.
Eimer M. 2000. Effects of face inversion on the structural encoding and
recognition of faces. Evidence from event-related brain potentials.
Brain Res Cogn Brain Res. 10:145--158.
Evans J, Heggs A, Antoun N, Hodges J. 1995. Progressive prosopagnosia
associated with selective right temporal lobe atrophy. A new
syndrome? Brain. 118:1--13.
Fairhall SL, Ishai A. 2006. Effective Connectivity within the Distributed
Cortical Network for Face Perception. Cereb Cortex. Advance Access
published December 26, 2006, doi:10.1093/cercor/bhl148.
Fernandez G, Effern A, Grunwald T, Pezer N, Lehnertz K,
Dumpelmann M, Van Roost D, Elger CE. 1999. Real-time tracking
of memory formation in the human rhinal cortex and hippocampus.
Science. 285:1582--1585.
Halgren E, Baudena P, Heit G, Clarke JM, Marinkovic K. 1994. Spatio-
temporal stages in face and word processing. I. Depth-recorded
potentials in the human occipital, temporal and parietal lobes. J
Physiol Paris. 88:1--50.
Halgren E, Baudena P, Heit G, Clarke JM, Marinkovic K, Chauvel P. 1994.
Spatio-temporal stages in face and word processing. 2. Depth-
recorded potentials in the human frontal and Rolandic cortices. J
Physiol Paris. 88:51--80.
Halgren E, Chauvel P. 1993. Experiential phenomena evoked by human
brain electrical stimulation. In: Devinsky O, Beric A, Dogali M,
editors. Electrical and magnetic stimulation of the brain and spinal
cord. Advances in Neurology. Raven Press, New York: Raven Press;
p. 123--140.
Halgren E, Raij T, Marinkovic K, Jousmaki V, Hari R. 2000. Cognitive
response profile of the human fusiform face area as determined by
MEG. Cereb Cortex. 10:69--81.
Haxby JV, Hoffman EA, Gobbini MI. 2000. The distributed human neural
system for face perception. Trends Cogn Sci. 4:223--233.
Haxby JV, Ungerleider LG, Clark VP, Schouten JL, Hoffman EA, Martin A.
1999. The effect of face inversion on activity in human neural
systems for face and object perception. Neuron. 22:189--199.
Humphreys GW, Riddoch MJ, Price CJ. 1997. Top-down processes in
object identification: evidence from experimental psychology,
neuropsychology and functional anatomy. Philos Trans R Soc Lond
B Biol Sci. 352:1275--1282.
Ishai A, Bikle PC, Ungerleider LG. 2006. Temporal dynamics of face
repetition suppression. Brain Res Bull. 70:289--295.
Ishai A, Haxby JV, Ungerleider LG. 2002. Visual imagery of famous faces:
effects of memory and attention revealed by fMRI. Neuroimage.
17:1729--1741.
Ishai A, Pessoa L, Bikle PC, Ungerleider LG. 2004. Repetition
suppression of faces is modulated by emotion. Proc Natl Acad Sci
USA. 101:9827--9832.
Ishai A, Schmidt CF, Boesiger P. 2005. Face perception is mediated by
a distributed cortical network. Brain Res Bull. 67:87--93.
Ishai A, Yago E. 2006. Recognition memory of newly learned faces. Brain
Res Bull. 71:167--173.
Itier RJ, Taylor MJ. 2004a. Effects of repetition learning on upright,
inverted and contrast-reversed face processing using ERPs. Neuro-
image. 21:1518--1532.
Itier RJ, Taylor MJ. 2004b. N170 or N1? Spatiotemporal differences
between object and face processing using ERPs. Cereb Cortex.
14:132--142.
Jemel B, Pisani M, Calabria M, Crommelinck M, Bruyer R. 2003. Is the
N170 for faces cognitively penetrable? Evidence from repetition
priming of Mooney faces of familiar and unfamiliar persons. Brain
Res Cogn Brain Res. 17:431--446.
Joubert S, Felician O, Barbeau E, Ranjeva JP, Christophe M, Didic M,
Poncet M, Ceccaldi M. 2006. The right temporal lobe variant of
frontotemporal dementia: cognitive and neuroanatomical profile of
three patients. J Neurol. 253:1447--1458.
Kanwisher N, McDermott J, Chun MM. 1997. The fusiform face area:
a module in human extrastriate cortex specialized for face
perception. J Neurosci. 17:4302--4311.
Klopp J, Marinkovic K, Chauvel P, Nenov V, Halgren E. 2000. Early
widespread cortical distribution of coherent fusiform face selective
activity. Hum Brain Mapp. 11:286--293.
Leveroni CL, Seidenberg M, Mayer AR, Mead LA, Binder JR, Rao SM.
2000. Neural systems underlying the recognition of familiar and
newly learned faces. J Neurosci. 20:878--886.
Marinkovic K, Trebon P, Chauvel P, Halgren E. 2000. Localised face pro-
cessing by the human prefrontal cortex: face-selective intracerebral
potentials and post-lesion deficits. Cogn Neuropsychol. 17:187--199.
McCarthy G, Puce A, Belger A, Allison T. 1999. Electrophysiological
studies of human face perception. II: response properties of face-
specific potentials generated in occipitotemporal cortex. Cereb
Cortex. 9:431--444.
1008 Spatio temporal Dynamics of Face Recognition d Barbeau et al.
Molholm S, Sehatpour P, Mehta AD, Shpaner M, Gomez-Ramirez M,
Ortigue S, Dyke JP, Schwartz TH, Foxe JJ. 2006. Audio-
visual multisensory integration in superior parietal lobule
revealed by human intracranial recordings. J Neurophysiol.
96:721--729.
Murray EA, Bussey TJ. 1999. Perceptual-mnemonic functions of the
perirhinal cortex. Trends Cogn Sci. 3:142--151.
Nowak LG, Bullier J. 1997. The timing of information transfer in the
visual system. In: Rockland KS, Kaas JH, Peters A, editors.
Extrastriate visual cortex in primates. New York: Plenum Press;
205--241.
Paller KA, Gonsalves B, Grabowecky M, Bozic VS, Yamada S. 2000.
Electrophysiological correlates of recollecting faces of known and
unknown individuals. Neuroimage. 11:98--110.
Puce A, Allison T, Gore JC, McCarthy G. 1995. Face-sensitive regions in
human extrastriate cortex studied by functional MRI. J Neuro-
physiol. 74:1192--1199.
Puce A, Allison T, Asgari M, Gore JC, McCarthy G. 1996. Differential
sensitivity of human visual cortex to faces, letterstrings, and textures:
a functional magnetic resonance imaging study. J Neurosci.
16:5205--5215.
Puce A, Allison T, McCarthy G. 1999. Electrophysiological studies of
human face perception. III: effects of top-down processing on face-
specific potentials. Cereb Cortex. 9:445--458.
Quian Quiroga R, Reddy L, Kreiman G, Koch C, Fried I. 2005. Invariant
visual representation by single neurons in the human brain. Nature.
435:1102--1107.
Rossion B, Dricot L, Devolder A, Bodart JM, Crommelinck M, De
Gelder B, Zoontjes R. 2000. Hemispheric asymmetries for whole-
based and part-based face processing in the human fusiform gyrus.
J Cogn Neurosci. 12:793--802.
Schweinberger SR, PickeringEC, Jentzsch I, BurtonAM,Kaufmann JM.2002.
Event-related brain potential evidence for a response of inferior
temporal cortex to familiar face repetitions.CognBrainRes. 14:398--409.
Sergent J, Ohta S, MacDonald B. 1992. Functional neuroanatomy of face
and object processing. A positron emission tomography study. Brain.
115:15--36.
Talairach J, Bancaud J. 1973. Stereotaxic approach to epilepsy:
methodology of anatomo-functional stereotactic investigations.
Prog Neurol Surg. 5:297--354.
Taylor MJ, George N, Ducorps A. 2001. Magnetoencephalographic
evidence of early processing of direction of gaze in humans.
Neurosci Lett. 316:173--177.
Tomita H, Ohbayashi M, Nakahara K, Hasegawa I, Miyashita Y. 1999.
Top-down signal from prefrontal cortex in executive control of
memory retrieval. Nature. 401:699--703.
Trautner P, Dietl T, Staedtgen M, Mecklinger A, Grunwald T, Elger CE,
Kurthen M. 2004. Recognition of famous faces in the medial
temporal lobe: an invasive ERP study. Neurology. 63:1203--1208.
Ungerleider L, Mishkin M. 1982. Two cortical visual systems. In: Ingle D,
Goodale M, Mansfield R, editors. Analysis of visual behaviour. MIT
Press, p. 549--586.
Ungerleider LG, Haxby JV. 1994. ‘What’ and ‘where’ in the human brain.
Curr Opin Neurobiol. 4:157--165.
Vignal JP, Chauvel P, Halgren E. 2000. Localised face processing by the
human prefrontal cortex: stimulation-evoked hallucinations of faces.
Cogn Neuropsychol. 17:281--291.
Wilbrand H. 1892. Ein Fall von Seelenblindheit und Heminopsie mit
Sectionsbefund. Z Nervenheilkd. 2:361 Deutsch.
Wilson FA, Scalaidhe SP, Goldman-Rakic PS. 1993. Dissociation of object
and spatial processing domains in primate prefrontal cortex.
Science. 260:1955--1958.
Cerebral Cortex May 2008, V 18 N 5 1009