Cerebral Cortex June 2007;17:1423--1432

doi:10.1093/cercor/bhl054
Advance Access publication August 21, 2006

Face Configuration Processing in the Chien-Chung Chen1,2, Kai-Ling C. Kao1 and

Christopher W. Tyler2
Human Brain: The Role of Symmetry
1
Department of Psychology, National Taiwan University,
Taipei 106, Taiwan and 2Smith-Kettlewell Eye Research
Institute, San Francisco, CA 94115, USA

Symmetry is an important cue in face perception. We manipulated mirror symmetry are important in face perception. There is

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

symmetry and other configurational variables to study their role in a well-established relationship between facial symmetry and
face processing in the human brain. We employed 2 types of perceived attractiveness (or beauty) of potential mates in
symmetry: image symmetry (where one part of the image is defined humans (Grammer and Thornhill 1994; Mealey and others
as the mirrored transform of the other part about an axis) and 1999; Rhodes and others 2001, 2002; Jacobsen and Hofel
object symmetry (where the spatial relationships among the image 2002). There is also a preference for facial asymmetry that is
components are interpreted as parts of a symmetric 3-dimensional related to emotion expressiveness (Swaddle and Cuthill 1995;
object). We compared blood oxygenation level dependent re- Zaidel and others 1995; Kowner 1996). Recently, Rhodes and
sponses in healthy human observers for upright front-view faces others (2005) showed that humans can detect symmetry better
with responses to different symmetry-controlled images. The in upright faces than in either inverted faces or contrast
cortical areas activated by the face images, relative to Fourier- reverted faces and argued that there should be a mechanism
matched scrambled images, were the fusiform (FFA) and occipital in the human visual system specific for the symmetry of faces.
(OFA) face areas, the middle occipital gyri (MOG), and areas around Brain areas supporting face processing have been reported
the superior temporal and intraoccipital sulci (IOS). Contrasting both in monkey (Perrett and others 1982) and in humans
faces and their image-symmetric scrambled versions showed (Kanwisher and others 1997; McCarthy and others 1997;
a similar activation pattern except in the right OFA, suggesting Halgren and others 1999; Haxby and others 2000). In particular,
an involvement in facial symmetry processing. The upright versus part of the fusiform gyrus and the inferior occipital gyrus have
inverted faces (with the same image symmetry but unfamiliar been reported as specific for face perception and are designated
object identity) showed robust differential activation in the FFA, the fusiform face area (FFA, Kanwisher and others 1997) and
OFA, MOG, IOS, and precuneus. The response to frontal-view occipital face area (OFA, Halgren and others 1999; Rossion and
versus 3/4-view faces (having the same object symmetry but others 2003), respectively (although there is disagreement on
disrupted image symmetry) showed little differential activation in the exact role of the FFA, see Gauthier and others 1999; Haxby
the FFA or the OFA but strong responses in the MOG and IOS, and others 2001). It is, however, not clear whether these brain
suggesting that face processing in the FFA and the OFA is holistic areas are involved in the analysis of local facial features, face-
and viewpoint invariant.
specific configurations, or other gestalt properties of faces. In
behavioral studies, the configuration-specific system is generally
Keywords: fMRI, fusiform, intraoccipital sulcus, inverted face,
isolated by contrasting performance for upright faces and
object perception, occipital face area
inverted faces (Yin 1969; Carey and Diamond 1977; Thompson
1980). A manipulation that causes performance degradation in
Introduction inverted relative to upright faces is considered to reflect
Face recognition and discrimination may be one of the most processing by mechanisms specific to the absolute facial con-
developed perceptual skills of visual object processing. An adult figuration because the relative configuration of the features is
can discriminate and recognize hundreds of faces (Bharick and undisturbed by rotation.
others 1975). Much evidence has shown that face perception There have been several neuroimaging studies contrasting
cannot be achieved by analyzing facial features alone and that brain activation to upright and inverted faces, but significant
facial configuration, or the spatial relationships among facial differential activations for upright face vs. inverted face have
features, is crucial to face perception (Fantz 1961; Valentine not been reported in either FFA or OFA (Kanwisher and others
1988; O’Toole and others 1994; Leder and Bruce 2000; Webster 1998; Aguirre and others 1999; Haxby and others 1999) until
and others 2004). recently (Yovel and Kanwisher 2004). Because faces are much
There are, however, many possible spatial relations among less recognizable when upside down, the implication of the
facial features. It is not clear what types of spatial relations are predominantly negative result is that these areas are responding
essential for a human observer to analyze the facial configura- to the basic aspects of facial stimuli that are equally present
tion. Much attention on facial configuration has focused on the when the face is inverted, such as the local features and the
relative distance between features (e. g., Fellous 1997), such as configuration aspects such as symmetry, rather than the higher-
the distance from the center of the face to several landmarks on level perceptual processes of facial recognition, which are
a face (e.g., Anderson and Wilson 2005), from the eyes to the substantially degraded by inversion.
eyebrows (e.g., Brown and Perrett 1993), from the eyes to the Our purpose here is to identify brain areas specific for the
nose and the mouth (Hosie and others 1988), and so on. On configuration properties of facial symmetry. Given recent
the other hand, there are reports that configuration cues such as reports on brain areas responding to image symmetry per se

Ó The Author 2006. Published by Oxford University Press. All rights reserved.
For permissions, please e-mail: journals.permissions@oxfordjournals.org
(Sasaki and others 2005; Tyler and others 2005a) and the 4. To examine whether the differential activation patterns
behavioral evidence for facial symmetry mechanisms (Rhodes from (3) are due to face-specific symmetry or to early image
and others 2005), we are interested in the role of symmetry in symmetry processing, we also contrasted front-view upright
brain areas responding to faces. We may conceptualize 2 types faces with their phase-scrambled versions that were verti-
of face symmetry. The first type, ‘‘image symmetry,’’ occurs cally symmetric. A contrast between faces and vertically
when one image component is a mirrored transform of another symmetric scrambled images should reveal all relevant areas
image component about some axis of transformation. This type supporting face processing that are insensitive to symmetry.
of symmetry specifies the direct spatial relations between the A comparison of the last 2 experiments should thus uncover
parts of the 2-dimensional (2D) projections of the faces and is whether areas that are attributed to face processing are
not concerned with the interpretation of the image as an object actually responding to the symmetry of the faces rather than
in space. The second type of symmetry is ‘‘object symmetry,’’ more informative cues.
which specifies the spatial relationships among the components 5. Finally, to study the role of object versus image symmetry
of the image interpreted as the representation of a 3D object. (under the assumption that faces are symmetric objects),

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

Hence, those components or facial features will be processed we compared the responses to upright frontal-view faces and
through a face gestalt or 3D template of the human head. This 3/4-view faces of the same persons. The 3/4-view faces
kind of object symmetry may not be consistent with image retain the same semantic and object-centered properties
symmetry due to an asymmetric viewpoint of the observer or as the frontal-view faces but have radically altered image
the rotation of the object’s axis of symmetry away from the line symmetry. Cortical areas encoding image symmetry should
of sight. Thus, a 3/4 view of a symmetric face forms an be differentially activated by this comparison, but those
asymmetric image, even though it is the image of a symmetric encoding object symmetry or semantic content should show
object and can be understood to have accurate object symme- little or no differential activation with the change in
try. Our experiments were designed to test the role of these 2 viewpoint of the same faces.
types of symmetry in the facial representations of the FFA and
the OFA. In detail, our goals were as follows:
Methods
1. To reveal all relevant areas supporting face processing
Stimuli
(except those for early visual processing), we first contrasted Figure 1 shows examples of the stimuli used in our experiment. We had
brain activations between faces and asymmetric scrambled 5 types of images: upright frontal-view faces, inverted faces, 3/4-view
versions of the same images with identical Fourier energy. faces, symmetric scrambled images, and asymmetric scrambled images.
These images should generate the same net activation for The face images were taken from The Facial Recognition Technology
arrays of neurons acting as local linear filters. (FERET) Database (Phillips and others 2000) provided by the National
Institute of Standards and Technology, USA. We computed the symme-
2. To identify the cortical areas responding to symmetry per se,
try index (Tyler and others, 2003) of the frontal-view faces of the
we compared the activation for the asymmetric random 1010 subjects in the FERET database and selected the 36 face images
images and symmetricized versions of the same images (as in with greatest symmetry index values. The symmetry index is computed
Tyler and others 2005a). based on the power spectrum of the Fourier transform of the face
3. To identify cortical regions involved in the processes of face images. Here we were only interested in the horizontal symmetry.
recognition, we then compared brain activation of the front- Hence, we computed the difference of the power at the points (kx, ky)
view faces and their inverted versions. Inverted faces are and (–kx, ky), where kx and ky are horizontal and vertical spatial
frequencies of the images (in the upper half-plane, excluding the
known to disrupt the facial configuration mechanisms and horizontal axis). The symmetry index is computed as a function of the
degrade the function of the face recognition mechanisms. root mean square difference of the power between corresponding
Thus, because inverted faces maintain the same local frequencies summed over the spectrum.
features, the same image symmetry, and the same object The inverted faces were simply the upside-down versions of the same
symmetry as the upright faces, the null hypothesis is that 36 images. The 3/4-view images were pictures of the same persons
they should produce no activation to regions containing available from the database with their head rotated 67.5° either to their
left (17 images) or to their right (19 images). The 2 types of scrambled
either local mechanisms of symmetry processing for facial
images were created by manipulating the phase spectrum of the frontal-
recognition. Hence, any significant differences in activation view face images. We first normalized the face images by subtracting
for the upright versus inverted faces should reveal the their mean luminance values to remove the DC component and then
operation of holistic facial configuration mechanisms. computed their Fourier transforms. The phase spectra were scrambled

Figure 1. Examples of the stimulus types used. (a) Frontal face, (b) inverted face, (c) 3/4-view face, (d) symmetricized phase-zeroed face image, (e) phase-scrambled face image.

1424 Face Symmetry d

Chen and others
randomly for the asymmetric scrambled images and set to zero phase for observer was then fed to a repeated measure analysis of variance
the symmetric scrambled images. We then generated the control images (ANOVA) that assessed the effect of conditions for the specified
by taking inverse Fourier transforms of the original amplitude spectra contrast. In addition, to get detailed information of the differential
combined with the new phase spectra of the 2 types. Note that this contrasts in specific ROIs defined by the localizers, we followed up
novel zero-phase manipulation introduces noticeable edge structures significant main effects in the ANOVA with pairwise comparison t-tests
throughout the images similar to that of the intact faces, so it tends to of differences in activation between conditions in the specified ROIs.
equate for nonlinear edge processing in the face images in addition to
the presence of bilateral symmetry. The luminance and contrast of the Localizers
inverse Fourier-transformed images were normalized to match the mean In addition to the main experiments, all observers participated in
luminance and contrast energy of the face images. The phase zeroed localizer experiments in a separate session, in order to identify relevant
scrambled images had point symmetry, so they were converted to brain areas reported in the literature. The lateral occipital complex
bilateral symmetry by inverting their left half relative to the right half. (LOC) was identified by the differential BOLD activation to pictures of
A fourth-power Gaussian aperture (1/e scale parameter = 3.2°) was common objects (objects that can be seen everyday life) versus block-
applied to all images to remove image features outside face areas (cf., scrambled versions of themselves, as developed by Kourtzi and
Tyler and Chen forthcoming). A fixation point of 6.759 3 6.759 size was Kanwisher (2001). The human motion sensitive complex (hMT+) was
inserted at the center of the face image.

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

identified by the differential BOLD activation to low-contrast moving
dots and their stationary versions. The moving dots (4.59 by 4.59 square)
Procedure oscillated between expansion and contraction once per second with
We used the blood oxygenation level dependent (BOLD) signal from a spatial density of 2% and a speed of 4.5°/s. The retinotopic areas were
a block-design functional Magnetic Resonance Imaging (fMRI) paradigm identified with checkerboard rotating wedges that spanned 180° and
to measure the cortical response to the stimulus contrasts. Each of the 4 rotated 30° every 3 s, taking 36 s to go around the display, with 6 periods
experimental runs had 18-s epochs of upright frontal-view faces in one retinotopic run. The symmetry areas were identified by the
alternating with 18-s epochs of one of the other image types in 6 cycles differential BOLD activation to the symmetric versus the asymmetric
(36-s period) of the 2-epoch block alternation. Each 18-s epoch scrambled images.
consisted of 18 presentations of the images for 850 ms followed by a
150-ms blank period. To keep the observers’ attention on the presented
images regardless of block content, the observers were instructed to Result and Discussion
press a response key when the currently presented image physically
matched the previous one (one-back task). Localizers
Stimuli were delivered with MRVision 2000 goggles from Resonance The symmetry localizer consisted of two types of nonsense
Technology, Inc. (Northridge, CA). At 800 (H) by 600 (V) resolution, the
scrambled images computed from the Fourier spectrum of the
pixel size of the display was specified as 2.259.
Six observers participated in this study, including 2 of the authors and face stimuli: one vertically symmetric and the other nonsym-
4 paid observers naı̈ve to the purpose of the study. All observers had metric. Figure 2(a) shows the activation map averaged across
normal or corrected-to-normal (by the focusing lenses included with observers on 3 views of the inflated brain. The gray shading
the goggles) visual acuity. All observers gave informed consent for their denotes the gyral (light) and sulcal (dark) layout. The pseudo-
participation. color denotes the average coherence between the response
waveform and the experimental sequence. The predominant
Data Acquisition and Analysis symmetry response occurred in the inferior occipital gyrus, the
The images were collected with a Bruker 3T scanner located at National middle occipital gyrus (MOG) and the areas around intra-
Taiwan University. Within each scanning session, both functional (T2*
weighted, BOLD) responses and anatomical (T1 weighted) images were
occipital sulcus (IOS). The MOG activation is posterior to
acquired in identical planes. The images were collected in 20 transverse hMT+ (red contour) and partially overlaps with LOC (magenta
planes parallel to the anterior commissure--posterior commissure line. contour). The dorsal predominance of the symmetry response is
An echo-planar imaging (EPI) sequence (Stehling and others 1991) was used consistent with that observed by Tyler and others (2005a) and
to acquire the functional data (time repetition = 3000 ms, time echo = Sasaki and others (2005).
40 ms, flip angle = 90°, voxel resolution = 2.34 3 2.34 3 3 mm). A high- Figure 2(b) shows areas localized by localizers on a flatmap.
resolution anatomical (T1 weighted) MRI volume scan of the entire head
The flatmaps here and in the remainder of this paper had their
was run once on each observer (voxel size = 1 3 1 3 1 mm). The main
experiment lasted 225 s (75 images). The first 9 s (3 images) were center near the occipital poles and extended 80 mm in radius
excluded from further analyses to avoid the start-up transient. Thus, the around this point. The areas delineated by yellow borders are
data analyzed for each scan spanned 216 s (72 images). To correct head- the first-tier retinotopic areas (V1--V3), identified with a rotating
motion artifacts, we used SPM (Friston and others 1995) to realign the wedge. The criterion for delineating visual areas is discussed
EPI images acquired. The realigned images, as well as the anatomic elsewhere (Tyler and others 2005b). The magenta borders
images, were then normalized to a standard template with SPM. The
denote the LOC as identified by contrasting BOLD activation to
normalized images were fed to the mrVista software (Wandell and
others 2000) for coregistration, data analysis, and 3D visualization. pictures of objects and their scrambled versions. The red
Statistical analysis of the BOLD activation was based on the spectral borders denote hMT+ as identified by the motion localizer.
correlation between the BOLD activation time series and the experi- The blue contours delimit the extent of the symmetry response.
mental sequence (Engel and others 1997). For each run, the Fourier The outline of these areas will be shown for reference in the
transform was calculated for each BOLD time series. The spectral subsequent flat maps.
correlation, or coherence, was then computed as the power at the
frequency of block alternation divided by the square root of the energy
of the spectrum for each voxel. The signal amplitude is given by the Brain Areas for Face Processing
square root of the signal power. Figure 3 (row 1) shows the averaged activation map across
There were situations where we needed to test differential contrasts observers for a face localizer contrasting the upright frontal-
across 2 or more block design runs. For this purpose, we followed the
procedure proposed by Joseph and others (2002). That is, we first
view faces and the asymmetric scrambled images. We show
identified a region of interest (ROI) defined by a specific parametric selected slices of all conditions contrasting upright front-
contrast from coherence maps for individual runs. The signal amplitude view faces with other images in this figure for ready compar-
of the averaged waveform over the voxels in this ROI from each ison. The pseudocolored voxels in the figures denote voxels

Cerebral Cortex June 2007, V 17 N 6 1425

 Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024
Figure 2. (a) Activation maps for the symmetry localizer (vs. Fourier-equated random
images) averaged across observers on 3 views of the inflated brain. Gyral (light) and
sulcal (dark) denoted by gray shading. Orange color denotes the average activation
pattern. Anterior (A) and posterior (P) directions indicated on underside view. (b) The
Figure 3. Average group activations shown in standardized brain slices for (a) face
symmetry-specific activation (yellow-orange coloration) depicted in flatmaps centered
localizer, (b) symmetry localizer, (c) face versus inverted face, and (d) frontal versus
on the occipital poles. Yellow borders: V1--V3; magenta borders: LOC; red borders:
rotated faces. Colored symbols indicated mean locations of activations from other
hMT+. Yellow-orange coloration codes amplitude of significant activation at corrected
studies (see key). Activation colors as in Figure 2.
P < 0.001 (color bar). The dark blue border denotes the extent of the activated areas.

with coherence above 0.475 or Bonferroni corrected P-value

Table 1
(considering all cortical voxels) less than 0.001. Brain areas responding to face information
Compared with the scrambled images, the face images
bilaterally activated the fusiform, the inferior occipital, and Activated areas Talairach coordinates
the middle occipital gyri areas around the superior temporal Fusiform Left (ÿ32, ÿ46, ÿ17)
sulcus (STS) and areas around the IOS. The Talairach coordi- Right (28, ÿ57, ÿ14)
Inferior occipital gyrus Left (ÿ28, ÿ70, ÿ15)
nates (Talairach and Tournoux, 1988) of those activated areas Right (29, ÿ77, ÿ15)
are given in Table 1. Notice that the first-tier retinotopic areas IOS Left (ÿ31, ÿ75, 16)
did not show significant differential activation between the two Right (24, ÿ86, 13)
MOG Left (ÿ30, ÿ84, 1)
types of images. On average, the BOLD signal modulations in V1 Right (27, ÿ88, 1)
of the left and the right hemispheres were 0.24% and 0.20%
(with confidence intervals of 0.29% and 0.24%), respectively.
These values are not statistically significant and are much
(significantly) smaller than typical V1 responses to contrast developed a form of scrambling that reduces the residual
energy change, which is about 1--2% modulation, with a confi- activation in early areas below the typical level of fMRI
dence interval of about 10% of this value (Engel and others detectability.
1997; Tootell and others 1998; Boynton and others 1999). This The IOS activation extended to a more dorsal region along the
null result demonstrates that our phase spectrum manipulation sulcus than did the symmetry localizer. The green symbols
equated all the relevant activations in early vision within denote the mean locations of the FFA reported in the literature
experimental error. This is an important result because many (triangles: averaged from individuals in Kanwisher and others
previous approaches to equating images with their scrambled 1997; diamonds: average reported in Rossion and others 2003;
versions have resulted in significant activation in the early circle: reported value in Halgren and others 1999). The cyan
retinotopic areas (Murray and others 2002, 2004; Altmann and symbols denote the OFA reported in Rossion and others (2003).
others 2003, 2004). It is therefore a worthwhile advance to have The fusiform form gyrus activation is consistent with the FFA

1426 Face Symmetry d

Chen and others
and the inferior occipital gyrus activation is consistent with the Image Symmetry Effect
OFA reported in the literature. Regions around the STS have The contrast between the upright frontal-view faces and the
been reported as relating to face processing both in humans symmetric scrambled images (Fig. 3, row 2) was designed to
(Puce and others 1998) and in monkeys (Perrett and others remove the factor of symmetry from the cortical face response
1982). The IOS activation partially overlaps with the area in the previous condition. This contrast showed significant
identified by the symmetry localizer. activation in the fusiform, the inferior occipital and the middle
For a better understanding of the spatial relationships occipital gyri, and also the area around the IOS. Figure 5
between the activated areas, Figure 4 shows an occipital flatmap compares the differential activation (pseudocolor patches) in
and inflated version of the same activation. As in Figure 2, the relation to the symmetry (blue contour) and the face areas (FFA:
flatmap is centered at the occipital pole and displays the spatial green contours, OFA: cyan contours, and IOS: black contours)
location of the activation across the cortical surface. The first- identified above. Qualitatively speaking, the major difference
tier retinotopic areas and the human motion area, identified in between this and the previous condition contrasting with
separate scans, are retained for reference. It may be seen that asymmetric scrambled was in the right OFA. The number of

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

the face-specific activation runs in a strip lying between these activated voxels dropped from 82 to 10 in the right OFA when
two reference regions. The concentrations corresponding to the symmetry information was available in the scrambled
the FFA and OFA from other studies are identified by triangles images. The differential contrast (Joseph and others 2002)
and diamonds, respectively. In addition, there is a pronounced between the face versus symmetric scrambled condition and
dorsal activation in the IOS centered at Talairach (–37, 75, 16) the face versus asymmetric scrambled was also significant (F5,10 =
for the left and (24, –86, 13) for the right hemisphere. This 9.98, P = 0.0251 < 0.05) in the right OFA. The average response
region has rarely been mentioned in previous studies of face- amplitude drops from 0.33% to 0.19% (t (5) = 3.87, P = 0.0059 <
specific activation, but appears to be the strongest site of 0.05). On the other hand, the left OFA and the bilateral FFA did
activation in the present dataset. The subsequent conditions are not show significant activation response change when symme-
designed to disentangle their particular roles in the agglomer- try was present in the scrambled images. The activation
ation of cues present in the face localizer. changes were only from 0.49% to 0.46% (t (5) = 0.58, P = 0.29
> 0.05, not significant [NS]) for the left OFA, from 0.49% to
0.41% (t (5) = 1.56, P = 0.09 > 0.05, NS) for the right fusiform,
and 0.44% to 0.35% (t (5) = 1.21, P = 0.14 > 0.05, NS) for the left.

Figure 4. Activation maps for the face localizer (vs. Fourier-equated random images)
averaged across observers (a) on 3 views of the inflated brain and (b) occipital-pole
flatmaps. Yellow borders: V1--V3; magenta borders: LOC; red borders: hMT+; green Figure 5. Activation maps for the face responses versus symmetry-equated
borders: FFA; cyan borders: OFA; black borders: IOS; blue borders: symmetry scrambled averaged across observers (a) on 3 views of the inflated brain and
activation—Colored symbols indicated locations of activations in individual subjects (b) occipital-pole flatmaps. FFA: green borders; OFA: cyan borders; IOS: black borders;
from other studies (see key). Activation colors as in Figure 2. blue borders: symmetry activation. Other details as in Figure 2.

Cerebral Cortex June 2007, V 17 N 6 1427

Thus, the data for the latter 3 ROIs are consistent with the null As shown in Figure 3 (row 3), upright versus inverted faces
hypothesis that the face activation in these areas was insensitive showed robust differential activation in the areas in the
to the symmetry information in the faces, whereas the right OFA fusiform, the inferior occipital, and the middle occipital gyri,
showed a significant symmetry component in its response. and also the areas around IOS and the precuneus. Figure 6
Because the difference between the asymmetric and symmet- shows the flatmap version of the same activation. The differen-
ric scrambled images lies in their image symmetry, the activation tial activation in the FFA and the OFA was no less than that
difference between the 2 conditions should reveal the areas between faces and asymmetric scrambled images (F5,10 =
processing 2D image symmetry information. Therefore, the OFA 0.0277, P = 0.88 > 0.05). These results confirm that the
activation is apparently influenced by image symmetry. How- symmetry processing in the OFA found in the previous
ever, the OFA is not one of the areas identified by our symmetry condition is indeed face specific, or at least orientation specific
localizer but lies just next to it. The plausible explanation is that with respect to face stimuli.
the OFA processes face-specific symmetry information (Rhodes BOLD activation for face inversion has been in reported in
and others 2005). The adjacent relation between the OFA and many areas throughout the brain (Servos and others 1999;

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

the symmetry areas hints that the OFA may receive input from Leube and others 2003; Yovel and Kanwisher 2004, 2005).
nearby symmetry area for facial processing. Reports of the face inversion effect in the literature for the FFA,
The IOS activations, on the other hand, are significantly however, are mixed. Early fMRI studies showed either no
stronger for the symmetry-equated face/scrambled contrast (Aguirre and others 1999; Haxby and others 1999) or a weak
than for the symmetric/asymmetric noise contrast (F5,10 = (Kanwisher and others 1998) face inversion effect in the FFA.
17.85, P = 0.0083 < 0.05, for the right hemisphere, F5,10 = More recent fMRI studies have shown a robust face inversion
10.39, P = 0.032 < 0.05 for the left; activation change increased effect in the FFA (Yovel and Kanwisher 2004, 2005). The Yovel
from 0.89% to 1.18%, t (5) = 2.86, P = 0.017 < 0.05 for the right and Kanwisher (2005) face inversion effect extended beyond
and from 0.46% to 0.62%, t (5) = 3.73, P = 0.007 < 0.05 for the the FFA to an extensive area on the ventral and the lateral
left). Thus, although it is the major locus of activation by occipital surfaces, which is quite similar to ours. Although the
symmetry, the IOS is even more strongly activated by the face cause of this discrepancy in the literature is not resolved, it is
configuration with symmetry equated. The implication is that possible that the face stimuli used in those studies played a role.
this is a cortical region specialized for processing configuration The papers reporting no face inversion effect in FFA (Aguirre
effects, of which symmetry and face arrangements are 2 and others 1999; Haxby and others 1999) used stimuli that had
separate forms. the external contours of the face removed and the one that

Face Configuration/Inverted Face Effect

We next compared mean brain activation for the upright versus
inverted faces. This is an interesting manipulation because it
distinguishes between absolute and relative configuration
effects. An absolute configuration effect is one that is specific
to the particular orientation and scale of the configuration.
For instance, the mouth is located 3° below the eyes with
a horizontal position midway between the eyes. This type of
configuration is also called the first-order configuration in some
face perception literature (Carey and Diamond 1977). A relative
configuration effect is one that maintains all the geometric relations
among a set of features, regardless of their absolute orientation
or scale. For instance, the mouth and the 2 eyes form a triangle
regardless whether the faces are upright, tilted, or inverted. This
type of configuration should not be confused with the second-
order configuration specified in some face perception litera-
ture (Carey and Diamond 1977), which refers to the relative
distances between facial features. The relative configuration
here emphasizes the orientation of the configuration, which is
not generally discussed in the second-order configuration
literature. In terms of the particular property of symmetry,
inverted faces are equated to the upright faces with respect to
both the relative and absolute configurations of ‘‘image’’
symmetry. However, the absolute face configuration (or face
template) response would be disrupted by the face inversion.
Hence, if a brain area such as the OFA processes symmetry in
faces in the same way as generic image symmetry, this area
should not show differential activation for upright versus
inverted face stimuli. On the other hand, if the symmetry is
processed as a part of an orientation-specific face gestalt (as
Figure 6. Activation maps for upright faces versus inverted faces averaged across
would be implied by our constant environmental exposure to observers (a) on 3 views of the inflated brain and (b) occipital-pole flatmaps. FFA:
upright faces), this area should still show differential activation green borders; OFA: cyan borders; IOS: black borders; blue borders: symmetry
for these 2 types of faces. activation. Other details as in Figure 2.

1428 Face Symmetry d

Chen and others
reported weak face inversion effect (Kanwisher and others
1998) had part of the face covered. On the other hand, studies
reporting robust face inversion effects (Yovel and Kanwisher
2004, 2005; and the current study) showed pictures of whole
faces as stimuli, including the hair and the neck. It has been
suggested that the external contour of a face is also an
important part of the face template (Sinha and Poggio 1996;
Maurer and others 2002). An incomplete face stimulus may not
optimally activate the neurons responsible for the full-face
template and may in turn produce insufficient differential ac-
tivation between upright face and inverted face.
The precuneus activation has a temporal phase opposite to
that of the other significant activations. It has been reported that

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

the precuneus is involved in mental rotation (Bonda and others
1995; Cohen and others 1996; Jordan and others 2001). Hence,
one likely explanation for the greater activation in the precu-
neus is that the observers were rotating the inverted face
information to enhance recognition. To clarify the exact role of
the precuneus in inverted face processing would, however,
require further study.

Object Structure
The upright frontal-view faces and the 3/4-view faces share the
same 3D object structure but not the same image symmetry.
Hence, a brain area for image symmetry would show differential
activation for these 2 types of images, whereas an area
processing symmetry based on the 3D object structure inferred
from the image information should respond equally to these 2
types of faces. Perceptually, we understand that the face is still
symmetric although it is 3D rotated, so there must be some
brain circuitry that carries this perceptual equivalence. This
analysis is supported by the finding that sparsely sampled
information about the positions of object features relies on an
exclusively 3D coding of object properties (Likova and Tyler
2003). The fact that the face retains the same object structure
when 3D rotated relative to the plane of projection can be
appreciated only by a neural circuit that has encoded the 3D
structure of the projected images.
The upright frontal-view faces and the 3/4-view faces showed
less differential activation (Fig. 3, row 4, also see Fig. 7 for
inflated and flat versions) than the face template areas Figure 7. Activation maps for frontal-view faces versus 3/4-view faces averaged
identified by faces versus inverted faces. The pseudocolor across observers (a) on 3 views of the inflated brain and (b) occipital-pole flatmaps. (c)
patches in Figure 7(c) identify the regions showing significant Flatmaps showing the regions of differential activation between faces and inverted
faces but not between faces and 3/4-view faces (yellow-orange coloration). FFA:
differential activation (F5,10 = 8.95, P = 0.03 < 0.05) between green borders; OFA: cyan borders; IOS: black borders; blue borders: symmetry
faces and inverted faces but not between faces and 3/4-view activation. Other details as in Figure 2.
faces. This reduction is quite pronounced in the FFA, as more
than 61% of voxels showing significant differential activation for the FFA and OFA is based on the 3D representation of the faces,
faces versus inverted faces (52% for the right and 87% for the which ‘‘look the same’’ to these neural circuits despite the
left) did not show significant activation in this condition. The radical change in image symmetry. The areas near the IOS and
activation modulation reduces from 0.49% for faces versus the MOG, however, still showed robust activation to the 3D
inverted faces to 0.20% for faces versus 3/4-view faces in the rotation of the faces as 65% (right) to 90% (left) of IOS voxels
left FFA (t (5) = 2.43, P = 0.029 < 0.05). A similar reduction was showed significant activation in this condition. This strong
observed in the right FFA as well (from 0.43% to 0.25%) though response implies that they are coding image symmetry rather
the difference is not statistically significant (t (5) = 0.79, P = 0.23 than object symmetry.
> 0.05, NS). To understand this activation, we may consider the factors
The activated area in the OFA was also substantially reduced that change versus those that are invariant in the image of a 3D-
(see Figs 6 vs. 7), although there were some spots still showing rotated face (Fig. 1a vs. 1c). The invariant factors are low-level
activation. The activation modulation was reduced from 0.49% ones such as the local edge structure and feature properties,
to 0.20% (t (5) = 2.43, P = 0.0296 < 0.05) for the left OFA and midlevel ones such as the 3D representation of the face
0.47% to 0.26% for the right OFA (t (5) = 2.28, P = 0.036 < 0.05) structure and high-level ones such as the emotional expression,
The implication of these results is that much of the coding in social position, and personal identity of the individual depicted.

Cerebral Cortex June 2007, V 17 N 6 1429

The factors that do change with 3D rotation are the local view and 3/4-view faces. Hence, the FFA is likely to incorporate
relations between local feature properties (such as angles an orientation-specific face gestalt (that is, however, viewpoint
between the edges) and 2D configurational relations among invariant with respect to lateral rotation).
the features as a whole, including the symmetry relations that The roles of the dorsal areas may be understood in terms of
activated these regions in the first experiment. We therefore relative versus absolute configuration analysis. The symmetry
conclude that the MOG and IOS regions are more involved in 2D localizer activated part of the areas around the IOS. Those dorsal
configural processing, including symmetry relations, than the symmetry areas, at least in the left hemisphere, did show
FFA and OFA, which are mostly concerned with properties that differential activation for the symmetry/asymmetry manipula-
are invariant under rotation. tion and for front-view versus 3/4-view faces, and also between
upright and inverted faces, which are equally symmetric. Hence,
the IOS area may be specialized for processing spatial config-
Viewpoint Invariance
urations, either purely of the symmetry type or of the face-
Our result on viewpoint invariance is consistent with those of
inversion type (with symmetry equated). Because this area
Grill-Spector and others (1999), which showed that the fusiform

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

responds well to configuration changes whether or not faces
activation in both hemispheres had greater viewpoint invariance
are present, whether or not symmetry was manipulated, and
than did the dorsal occiput. Vuilleumier and others (2002) used
whether or not the relative configuration was varied, the
the repetition priming technique to show that left (but not
implication is that the IOS region is involved in a wide variety
right) fusiform activation had viewpoint-invariant behavior for
of configuration analyses.
viewing objects, although size invariance was exhibited bi-
The only manipulation that reduced the response in the
laterally. It is interesting to note that both our stimulus sets
symmetry activation area was 3D face rotation, which elimi-
and those of Grill-Spector and others (1999) contained faces,
nated significant response in much of the MOG region. This
whereas those of Vuilleumier and others (2002) contained only
region has been particularly associated with the processing of
nonbiological objects. Hence, the discrepancy with the latter
3D structure (Tyler and others 2006), so the lack of response
results may be due to the sampling of stimuli.
may be understood as a similar response to the 2 forms of 3D
Behavioral studies of face recognition showed a viewpoint-
structure carried by both the frontal and rotated faces. The
dependent effect. When asked whether an image of a face was
strong response to faces versus scrambled textures makes sense
from the same person as a previously shown face image, the
in this context as a response to the depth structure in the faces,
observers‘ recognition performance depends on the viewpoint
but the response to symmetric versus asymmetric textures and
of the learned face (Troje and Bulthoff 1996; Hill and others
to face inversion is harder to understand. It may be that there is
1997) but not the test face (Troje and Bulthoff 1996). This
a stronger perception of depth structure in upright than in
evidence seems contradictory to our fMRI results, which show
inverted faces due to the match to the 3D gestalt of the upright
little differential activation between 3/4-view and front-view
face, and a stronger sense of 3D structure from the luminance
faces in the FFA and OFA. Such viewpoint invariance is com-
cues of the symmetric versus asymmetric noise (especially in
patible with models suggesting that viewpoint invariance can
the case of noise with the 1/f-type spectrum of the faces). These
be achieved from the concatenation of view-based processes
are suggestions that would require further study to pin down.
(for reviews, see Riesenhuber and Poggio 2000; Rolls 2000).
However, it should be noted that an invariance property in
a brain area as a whole does not necessary mean that cells in this Conclusions
area all have the same invariance property. Given the spatial
We draw 5 main conclusions from the pattern of results to
resolution of the fMRI, it is likely that each voxel covers
the symmetry manipulations: 1) The complex of cortical areas
a network of view-based units. Hence, brain areas with such
activated by the face images, relative to Fourier-matched scram-
units would appear as viewpoint invariant in their fMRI
bled images, were the fusiform, inferior, and middle occipital gyri,
responses, even though neural recognition responses have
and areas around the STS and IOS. This is a larger range of dorsal
access to the individual components of such networks.
cortical areas than has been previously reported for face-specific
activation. 2) The areas responding to symmetry per se were
Ventral and Dorsal Face Processing predominantly in the MOG and the IOS. Conversely, the OFA and
By this point, the roles of the ventral areas in face configuration the FFA did not show significant correlation changes for symmetry
processing are clear. The right OFA, located just next to in the scrambled images. 3) To identify cortical regions involved in
common image symmetry areas, does process face-related the processes of face recognition, we compared brain activation
image symmetry. Hence, it shows differential activation be- of the front-view faces and their inverted versions. The upright
tween faces and asymmetric scrambled images but not between versus inverted faces showed robust differential activation in
faces and symmetric scrambled images. the areas in the fusiform, inferior occipital and middle occipital
The symmetry information processed in the right OFA is either gyri, around the IOS, and in the precuneus. 4) Contrasting the
orientation specific or a part of face gestalt. After all, the OFA responses to faces and to their vertically symmetric phase-
does show differential activation between upright and inverted scrambled versions showed a response pattern similar to that
faces, along with the reduction of differential activation produced by contrasting the responses to faces and asymmetric
between front view and 3/4 view. The FFA (at least its anterior scrambled images (except in the right OFA). This result suggests
part) on the other hand, has little interest in image symmetry. that the OFA may be involved in processing symmetry specific to
Alternation between symmetric- and asymmetric scrambled face images (in addition to other aspects of the facial configura-
images produces no detectable effect in this area. In terms of tion). 5) Finally, to study the role of object versus image symmetry,
face-specific stimuli, the FFA shows strong differential activa- we compare the responses to frontal and 3/4-view faces of the
tion between upright and inverted faces but not between front- same individuals. This contrast showed little differential activation

1430 Face Symmetry d

Chen and others
in the FFA or the OFA but strong responses in areas near the MOG Haxby JV, Ungerleider LG, Clark VP, Schouten JL, Hoffman EA, Martin A.
and the IOS, indicating that viewpoint dependence is measurable 1999. The effect of face inversion on activity in human neural
in the BOLD responses of some cortical areas. Taken together, systems for face and object perception. Neuron 22:189--199.
Hill H, Schyns PG, Akamatsu S. 1997. Information and viewpoint
these results suggest that face processing in the FFA and the OFA
dependence in face recognition. Cognition 62:201--222.
is holistic and relatively invariant to the viewpoint from which the Hosie JA, Ellis HD, Haig ND. 1988. The effect of feature displacement on
face is seen. the perception of well-known faces. Perception 17:461--474.
Jacobsen T, Hofel L. 2002. Aesthetic judgments of novel graphic patterns
Notes analyses of individual judgments. Percept Mot Skills 95:755--766.
Jordan K, Heinze HJ, Lutz K, Kanowski M, Jancke L. 2001. Cortical
Supported by National Science council (Taiwan) 94-2752-H-002-007- activations during the mental rotation of different visual objects.
PAE to CCC and National Institutes of Health/National Eye Institute Neuroimage 13:143--152.
(USA) EY 13025 to CWT. Conflict of Interest: None declared. Joseph JE, Partin DJ, Jones M. 2002. Hypothesis testing for selective,
Address correspondence to Chien-Chung Chen, Department of differential, and conjoined brain activation. J Neurosci Methods
Psychology, National Taiwan University, Taipei 106, Taiwan. Email: 118:129--140.
c3chen@ntu.edu.tw.

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

Kanwisher N, McDermott J, Chun M. 1997. The fusiform face area:
a module in human extrastriate cortex specialized for face percep-
References tion. J Neurosci 17:4302--4311.
Kanwisher N, Tong F, Nakayama K. 1998. The effect of face inversion on
Aguirre GK, Singh R, D’Esposito M. 1999. Stimulus inversion and the
the human fusiform face area. Cognition 68:B1--B11.
responses of face and object-sensitive cortical areas. Neuroreport
Kourtzi Z, Kanwisher N. 2001. Representation of perceived object shape
10:189--194.
by the human lateral occipital complex. Science 293:1506--1509.
Altmann CF, Bulthoff HH, Kourtzi Z. 2003. Perceptual organization of
Kowner R. 1996. Facial asymmetry and attractiveness judgment in
local elements into global shapes in the human visual cortex. Curr
developmental perspective. J Exp Psychol Hum Percept Perform
Biol 13:342--349.
22:662--675.
Altmann CF, Deubelius A, Kourtzi Z. 2004. Shape saliency modulates
Leder H, Bruce V. 2000. When inverted faces are recognized: the role of
contextual processing in the human lateral occipital complex.
configuration information in face recognition. Q J Exp Psychol
J Cogn Neurosci 16:794--804.
53A:513--536.
Anderson ND, Wilson HR. 2005. The nature of synthetic face adaptation.
Leube DT, Yoon HW, Rapp A, Erb M, Grodd W, Bartels M, Kircher TT.
Vis Res 45:1815--1828.
Bharick HP, Bharick PO, Wittlinger RP. 1975. Fifty years of memory for 2003. Brain regions sensitive to the face inversion effect: a functional
names and faces: a cross-section approach. J Exp Psychol Gen magnetic resonance imaging study in humans. Neurosci Lett
104:54--75. 342:143--146.
Bonda E, Petrides M, Frey S, Evans A. 1995. Neural correlates of mental Likova LT, Tyler CW. 2003. Peak localization of sparsely sampled
transformations of the body-in-space. Proc Natl Acad Sci USA luminance patterns is based on interpolated 3D surface representa-
92:11180--11184. tion. Vis Res 43:2649--2957.
Boynton GM, Demb JB, Glover GH, Heeger DJ. 1999. Neuronal basis of Maurer S, Le Grand R, Mondloch, CJ. 2002. The many faces of configural
contrast discrimination. Vis Res 39:257--269. processing. Trends Cogn Sci 6:255--260.
Brown E, Perrett DI. 1993. What gives a face its gender? Perception McCarthy G, Puce A, Gore JC, Allison T. 1997. Face-specific processing
22:829--840. in the human fusiform gyrus. J Cogn Neurosci 9:605--610.
Carey S, Diamond R. 1977. From piecemeal to configural representation Mealey L, Bridgstock R, Townsend GC. 1999. Symmetry and perceived
of faces. Science 195:312--314. facial attractiveness a monozygotic co-twin comparison. J Pers Soc
Cohen MS, Kosslyn SM, Breiter HC, DiGirolamo GJ, Thompson WL, Psychol 76:151--158.
Anderson AK, Brookheimer SY, Rosen BR, Belliveau JW. 1996. Murray SO, Kersten D, Olshausen BA, Schrater P, Woods DL. 2002. Shape
Changes in cortical activity during mental rotation: a mapping study perception reduces activity in human primary visual cortex. Proc
using functional MRI. Brain 119:89--100. Natl Acad Sci USA 99:15164--15169.
Engel SA, Glover GH, Wandell BA. 1997. Retinotopic organization in Murray SO, Schrater P, Kersten D. 2004. Perceptual grouping and the
human visual cortex and the spatial precision of functional MRI. interactions between visual cortical areas. Neural Netw 17:695--705.
Cereb Cortex 7:181--192. O’Toole AJ, Deffenbacher KA, Valentin D, Abdi H. 1994. Structural
Fantz RL. 1961. The origin of form perception. Sci Am 204:66--72. aspects of face recognition and the other-race effect. Mem Cogn
Fellous J. 1997. Gender discrimination and prediction on the basis of 22:208--224.
facial metric information. Vis Res 37:1961--1973. Perrett DI, Rolls ET, Cann W. 1982. Visual neurons responsive to faces in
Friston KJ, Holmes AP, Worsley KJ, Poline JP, Frith CD, Frackowiak RSJ. the monkey temporal cortex. Exp Brain Res 47:329--342.
1995. Statistical parametric maps in functional imaging: a general Phillips P J, Moon H, Rauss PJ, Rizvi S. 2000. The FERET evaluation
linear approach. Hum Brain Mapp 2:189--210. methodology for face recognition algorithms. IEEE Trans Pattern
Gauthier I, Tarr MJ, Anderson AW, Skudlarski P, Gore JC. 1999. Anal Mach Intell 22:1090--1104.
Activation of the middle fusiform ‘‘face area’’ increases with Puce A, Allison T, Bentin S, Gore JC, McCarthy G. 1998. Temporal cortex
expertise in recognizing novel objects. Nat Neurosci 2:568--573. activation of human viewing eye and mouth movement. J Neurosci
Grammer K, Thornhill R. 1994. Human (Homo sapiens) facial attrac- 18:2188--2199.
tiveness and sexual selection the role of symmetry and averageness. Rhodes G, Geddes K, Jeffery L, Dziurawiec S, Clark A. 2002. Are average
J Comp Psychol 108:233--242. and symmetric faces attractive to infants? Discrimination and looking
Grill-Spector K, Kushnir T, Edelman S, Avidan G, Itzchak Y, Malach R. preferences. Perception 31:315--321.
1999. Differential processing of objects under various viewing condi- Rhodes G, Peters M, Lee K, Morrone MC, Burr D. 2005. Higher-level
tions in the human lateral occipital complex. Neuron 24:187--203. mechanisms detect facial symmetry. Proc Biol Sci 272:1379--1384.
Halgren E, Dale AM, Sereno MI, Tootell RBH, Marinkovic K, Rosen B. Rhodes G, Yoshikawa S, Clark A, Lee K, McKay R, Akamatsu S. 2001.
1999. Location of human face-selective cortex with respect to Attractiveness of facial averageness and symmetry in non-western
retinotopic areas. Hum Brain Mapp 7:29--37. cultures in search of biologically based standards of beauty.
Haxby JV, Gobbini MI, Furey ML, Ishai A, Schouten JL, Pietrini P. 2001. Perception 30:611--625.
Distributed and overlapping representations of faces and objects in Riesenhuber M, Poggio T. 2000. Models of object recognition. Nat
ventral temporal cortex. Science 293:2425--2430. Neurosci Suppl 3:1199--1204.
Haxby JV, Hoffman, EA, Gobbini MI. 2000. The distributed human neural Rolls ET. 2000. Functions of the primate temporal lobe cortical visual areas
system for face perception. Trends Cogn Sci 4:223--233. in invariant visual object and face recognition. Neuron 27:205--218.

Cerebral Cortex June 2007, V 17 N 6 1431

Rossion B, Caldara R, Seghier M, Schuller AM, Lazeyras F, Mayer E. 2003. Tyler CW, Baseler HA, Kontsevich LL, Likova LT, Wade AR, Wandell BA.
A network of occipito-temporal face-sensitive areas besides the right 2005a. Predominantly extra-retinotopic cortical response to pattern
middle fusiform gyrus is necessary for normal face processing. Brain symmetry. Neuroimage 24:306--314.
126:2381--2395. Tyler CW, Chen CC. Spatial summation of face information. J Vis.
Sasaki Y, Vanduffel W, Knutsen T, Tyler CW, Tootell RH. 2005. Symmetry Forthcoming.
activates extrastriate visual cortex in human and nonhuman pri- Tyler CW, Likova LT, Chen CC, Kontsevich LL, Schira MM, Wade AR.
mates. Proc Natl Acad Sci USA 102:3159--3163. 2005b. Extended concepts of occipital retinotopy. Curr Med Imaging
Servos P, Engel SA, Gati J, Menon R. 1999. fMRI evidence for an inverted Rev 1:319--329.
face representation in human somatosensory cortex. Neuroreport Tyler CW, Likova LT, Kontsevich LL, Wade AR. 2006. The specificity of
10:1393--1395. cortical area KO to depth structure. Neuroimage 30:228--238.
Sinha P, Poggio T. 1996. I think I know that face. Nature 384:404. Valentine T. 1988. Upside-down faces: a review of the effect of inversion
Stehling MK, Turner R, Mansfield P. 1991. Echo-planar imaging: mag- upon face recognition. Br J Psychol 79:471--491.
netic resonance imaging in a fraction of a second. Science 254: Vuilleumier P, Henson RN, Driver J, Dolan RJ. 2002. Multiple levels of
43--50. visual object constancy revealed by event-related fMRI of repetition
Swaddle JP, Cuthill IC. 1995. Asymmetry and human facial attractiveness priming. Nat Neurosci 5:491--499.
symmetry may not always be beautiful. Proc R Soc Lond B Biol Sci Wandell BA, Chial S, Backus BT. 2000. Visualization and measurement of

Downloaded from https://academic.oup.com/cercor/article/17/6/1423/421531 by guest on 02 February 2024

261:111--116. the cortical surface. J Cogn Neurosci 12:739--752.
Talairach J, Tournoux P. 1988. Co-planar stereotaxic atlas of the human Webster MA, Kaping D, Mizokami Y, Duhamel P. 2004. Adaptation to
brain. New York: Thieme Medical Publishers. natural facial categories. Nature 428:557--561.
Thompson, P. 1980. Margaret Thatcher: a new illusion. Perception 9:483--484. Yin RK. 1969. Looking at upside-down faces. J Exp Psychol 81:141--145.
Tootell RB, Hadjikhani NK, Vanduffel W, Liu AK, Mendola JD, Sereno MI, Yovel G, Kanwisher N. 2004. Face perception: domain specific, not
Dale AM. 1998. Functional analysis of primary visual cortex (V1) in process specific. Neuron 44:889--898.
humans. Proc Natl Acad Sci USA 95:811--817. Yovel G, Kanwisher N. 2005. The neural basis of the behavioral face-
Troje NF, Bulthoff HH. 1996. Face recognition under varying poses: the inversion effect. Curr Biol 15:2256--2262.
role of texture and shape. Vis Res 36:1761--1771. Zaidel DW, Chen AC, German C. 1995. She is not a beauty even when she
Tyler CW, Baseler HA, Kontsevich LL. 2001. A global symmetry model smiles: possible evolutionary basis for a relationship between facial
predicts both fMRI time-series and psychophysical responses to attractiveness and hemispheric specialization. Neuropsychologia
symmetry patterns. Soc Neurosci Abs 11.11. 33:649--655.

1432 Face Symmetry d

Chen and others