J Neurophysiol 103: 1467–1477, 2010.
First published December 23, 2009; doi:10.1152/jn.00637.2009.
fMRI–Adaptation Studies of Viewpoint Tuning in the Extrastriate and
Fusiform Body Areas
John C. Taylor, Alison J. Wiggett, and Paul E. Downing
Wales Institute of Cognitive Neuroscience, School of Psychology, Bangor University, Bangor, United Kingdom
Submitted 21 July 2009; accepted in final form 19 December 2009
INTRODUCTION
Human bodies are ubiquitous, salient, and easily recognized
from many viewpoints, but the structure of neural body representations is not well understood. The extrastriate and fusiform
body areas—EBA (Downing et al. 2001) and FBA (Peelen and
Downing 2005; Schwarzlose et al. 2005), respectively—are
two regions of the visual system that are highly sensitive to
visual depictions of the human body (for a review, see Peelen
and Downing 2007). EBA is located in the posterior inferior
temporal sulcus, and FBA is found in the fusiform gyrus,
where it overlaps with the fusiform face area (Peelen and
Downing 2005; Schwarzlose et al. 2005). Recent studies have
provided some clues regarding the respective contributions of
these regions to representing body form and action (Moro et al.
2008; Schwarzlose et al. 2005, 2008; Taylor et al. 2007; Urgesi
et al. 2007b). Furthermore, two studies have investigated the
viewpoint dependency of EBA (Chan et al. 2004; Saxe et al.
2006), showing a small right hemisphere bias in this region for
Address for reprint requests and other correspondence: J. Taylor, School of
Psychology, Bangor Univ., Adeilad Brigantia, Penrallt Rd., Bangor LL57 2AS,
UK (E-mail: [email protected]).
www.jn.org
allocentric over egocentric views of bodies and body parts.
However, to date, no comprehensive study of the view invariance of whole-body representations in EBA and FBA has been
reported.
One potential source of clues about the representations in
EBA and FBA is the evidence on nearby object representations
in the lateral occipital complex (LOC). The LOC is segregated
into discrete subregions—a dorsal-caudal region (LO) and a
ventral region (pFs), extending into the fusiform gyrus (GrillSpector et al. 1999). Anatomically, LO falls close to EBA
(Downing et al. 2007), and pFs is near FBA. There may be
some functional similarities between areas generally involved
in object perception and neighboring areas that are selective for
the human form. LO (relative to pFs) is sensitive to object size
(Grill-Spector et al. 1999) and shows greater sensitivity to
image scrambling (Lerner et al. 2001, 2002). Area pFs, but not
LO, responds to stimulus changes that alter 3D form but leave
the 2D outline intact (Kourtzi et al. 2003). Furthermore, LO
carries more information about local object parts than pFs
(Drucker and Aguirre 2009) and favors physical aspects of the
stimulus rather than the resulting subjective percept (Haushofer
et al. 2008).
Taken together, the above data imply a greater sensitivity to
local features in lateral object areas, compared with “higher
level,” more holistic representations in their ventral counterparts, a distinction that may extend to body-selective regions.
Indeed, EBA responds selectively to individual body parts, in
contrast to FBA, which shows selectivity for larger assemblies
of parts and whole bodies (Taylor et al. 2007). The claim for a
stronger representation of parts in EBA is supported by transcranial magnetic stimulation (TMS) evidence (Urgesi et al.
2007a). In this study, match to target performance for inverted
(but not upright bodies) was impaired by TMS over EBA. TMS
over ventral premotor cortex (vPMC) gave the opposite pattern. The authors concluded that performance in the upright
condition was driven by configural (template based) representations, whereas inversion disrupts template-based perception
and is reliant on local feature analysis. Disruption of EBA
therefore leads to a deficit in local feature processing.
Behavioral studies of the perception of human-like mannequins across changing viewpoints suggest that body representations are somewhat view-dependent. Kourtzi and Shiffrar
(1998) showed priming of unseen views of stimuli for rotations
of ⱕ30° from a previously seen view, but not for rotations of
60°, provided the views were not linked by motion. In subsequent functional MRI (fMRI) studies using synthetic objects,
Kourtzi et al. (2003) showed response invariance to rotations
of 30° in object-selective regions of LOC. Here we used a
similar approach to determine experimentally the degree of
0022-3077/10 $8.00 Copyright © 2010 The American Physiological Society
1467
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Taylor JC, Wiggett AJ, Downing PE. fMRI–adaptation studies of
viewpoint tuning in the extrastriate and fusiform body areas. J
Neurophysiol 103: 1467–1477, 2010. First published December 23,
2009; doi:10.1152/jn.00637.2009. People are easily able to perceive
the human body across different viewpoints, but the neural mechanisms underpinning this ability are currently unclear. In three experiments, we used functional MRI (fMRI) adaptation to study the
view-invariance of representations in two cortical regions that have
previously been shown to be sensitive to visual depictions of the
human body—the extrastriate and fusiform body areas (EBA and
FBA). The BOLD response to sequentially presented pairs of bodies
was treated as an index of view invariance. Specifically, we compared
trials in which the bodies in each image held identical poses (seen
from different views) to trials containing different poses. EBA and
FBA adapted to identical views of the same pose, and both showed a
progressive rebound from adaptation as a function of the angular
difference between views, up to ⬃30°. However, these adaptation
effects were eliminated when the body stimuli were followed by a
pattern mask. Delaying the mask onset increased the response (but not
the adaptation effect) in EBA, leaving FBA unaffected. We interpret
these masking effects as evidence that view-dependent fMRI adaptation is driven by later waves of neuronal responses in the regions of
interest. Finally, in a whole brain analysis, we identified an anterior
region of the left inferior temporal sulcus (l-aITS) that responded
linearly to stimulus rotation, but showed no selectivity for bodies. Our
results show that body-selective cortical areas exhibit a similar degree
of view-invariance as other object selective areas—such as the lateral
occipitotemporal area (LO) and posterior fusiform gyrus (pFs).
1468
J. C. TAYLOR, A. J. WIGGETT, AND P. E. DOWNING
METHODS
Experiment 1
Thirteen right-handed participants (6 female) were
recruited via the Bangor University Community Panel. All partici-
PARTICIPANTS.
J Neurophysiol • VOL
pants were briefed, and informed consent was obtained in compliance
with procedures set by the School of Psychology Ethics Panel, which
approved these studies. Participants were prescreened by questionnaire to identify any factors preventing inclusion, such as preexisting
medical or neurological conditions or claustrophobia. All participants
were right handed with normal (or corrected to normal) vision.
Data were acquired using a 3T Philips MRI scanner, fitted with a
SENSE parallel head coil (Philips, Best, The Netherlands). Stimulus
images were presented using a Sanyo LCD projector (Sanyo, Osaka,
Japan) focused onto a translucent, rear projection screen. Presentation
was controlled using MatLab Psychophysics Toolbox (Mathworks,
Natick, MA), running on an Apple iBook G3 laptop computer with
Macintosh OS9 operating system (Apple, Cupertino, CA). 1-back and
target stimulus responses were recorded using a nonferrous, fire optic
response pad (Current Designs, Philadelphia, PA). Preprocessing,
region of interest (ROI) definition and ROI/General Linear Model
(GLM) calculations were performed using BrainVoyager QX 1.9
software (Brain Innovation, Maastricht, The Netherlands).
Stimuli were images of headless male
and female human bodies (2 exemplars of each). Each body was
arranged into eight different poses and positioned on a minimally
defined floor plane (Poser, Curious Labs, Santa Cruz, CA). Each pose
was rotated through 24 successive 15° increments— giving a total of
768 images (24 views per pose, 8 poses per figure, 4 figures). All
images were scaled to 400 ⫻ 400 pixels, 72 dpi, and default Poser
image colors were retained. Seven poses were selected as experimental stimuli, whereas the eighth was a “target” pose for the detection
task. Masks were prepared from each of the four figures, to allow the
masking of each figure to be matched for low-level image properties.
One exemplar of each figure was selected and divided into an 11 ⫻ 11
matrix. The spatial arrangement of the resulting 121 tiles was then
scrambled. Examples of the four stimulus figures and mask images are
given in Fig. 1.
MAIN EXPERIMENT STIMULI.
LOCALIZER STIMULI. EBA and FBA localizer stimuli comprised 20,
400 ⫻ 400 pixel, 72 dpi grayscale images each of headless human
bodies (in a variety of postures) and chairs. Image positions were
jittered slightly on alternate presentations to prevent reliance on
low-level transient detection.
SCANNING PARAMETERS. Functional data were obtained using T2*weighted scans. A single shot echo planar (EPI) sequence was used
because this approach minimizes potential motion artifacts resulting
from participant head movement such as image ghosting and blurring.
To capture all cortical ROIs, 15 oblique-axial slices were scanned to
include, bilaterally, the inferior temporal and occipital lobes including
the fusiform gyrus. Acquisition parameters for all participants were
112 ⫻ 112 matrix, voxel dimensions ⫽ 2 mm isometric; echo time
(TE) ⫽ 35 ms; repetition time (TR) ⫽ 1,500 ms; flip angle ⫽ 65°.
Participants viewed rear-projected stimuli via a mirror positioned
above the SENSE head coil. Functional scans were recorded for the
duration of image presentation. Parameters for T1-weighted anatomical scans were: 256 ⫻ 256 matrix; voxel dimensions ⫽ 1 ⫻ 1 ⫻ 1
mm; TR ⫽ 12 ms, TE ⫽ 3 ms; flip angle ⫽ 8°.
For the main experiment, participants were
scanned on 3 of 32 possible sequences of a rapid event-related
adaptation design. Each image sequence comprised 128, 3-s trials,
including 16 fixation baseline trials. Nonbaseline trials comprised a
serial presentation of two frames, each depicting the same human
figure, for a duration of 300 ms each (Fig. 1). The first frame could be
any of the eight poses in any of the 24 possible views. The second
frame was either the same pose (5/6 trials) or a different pose (1/6
trials). Stimuli in frame 2 were either viewed from the same angle as
frame 1 (0° condition) or rotated by ⱕ60° clockwise or counterclockwise, in 15° increments (this applied to both same and different
poses). In this way, the absolute viewpoint of the frames was rendered
MAIN EXPERIMENT.
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
view invariance of human bodies in lateral and ventral bodyselective regions.
To do so, we used an fMRI adaptation (fMRI-A) design.
fMRI-A is a robust feature of cortical dynamics that manifests
as an attenuation of BOLD signal response to repeated presentations of the same stimulus. Although the precise mechanisms
behind fMRI-A are still debated (Epstein et al. 2008; for
reviews, see Grill-Spector et al. 2006; Henson 2003), there is a
consensus that the smaller BOLD signal response obtained for
repeated (compared with different) stimuli can be treated as an
index of stimulus similarity along a given dimension (GrillSpector and Malach 2001). The effect of varying any stimulus
parameter can be monitored in this way, providing researchers
with a powerful tool for characterizing the functional selectivity of brain areas.
Adaptation paradigms have been widely applied in neuroimaging studies. Considering face perception alone, this approach has been used to study: the role of the occipital and
fusiform face areas in holistic and part-wise analysis (Betts and
Wilson 2007; Harris and Aguirre 2008; Schlitz and Rossion
2006); dissociations between identity and expression (Andrews
and Ewbank 2004; Winston et al. 2004); the modulatory effects
of emotion (Rotshtein et al. 2001); contributions of nonface
cortical areas to face discrimination in prosopagnosic patients
(Dricot et al. 2008); scale invariance (Eger et al. 2004); and
viewpoint invariance (Fang et al. 2007). fMRI-A has proved to
be of particular use in the study of representational invariance
in the visual system, across a wide range of stimulus categories: e.g., sinusoidal gratings (Boynton and Finney 2003),
geometric structures (Weigelt et al. 2007), abstract shapes
(Kourtzi et al. 2003), object contours and perceived shape
(Kourtzi and Kanwisher 2001), and natural and man-made
objects (James et al. 2002). However, although fMRI-A has
previously been applied to study some body properties in
EBA— e.g., human action/identity pairings (Kable and Chatterjee 2006), self/other discrimination of body parts (Myers
and Sowden 2008)—to date, view invariance has not been
examined, and the adaptation characteristics of FBA remain
unknown.
We used an fMRI-A paradigm to examine BOLD response
changes to sequentially presented pairs (Buckner et al. 1998;
Kourtzi et al. 2003) of human bodies. These differed either in
the pose held by the figure or the angle at which they were
viewed. Identity was held constant within each trial and varied
across trials. For both EBA and FBA, we characterized the
modulation of adaptation by changes in viewpoint. The extent
of viewpoint change at which the response to a repeated pose
reaches that produced by two different poses is taken as an
index of the limit of view invariance for a given area. Our main
aim was to assess whether view invariance in these regions was
comparable to that seen in behavioral studies and in fMRI
studies of object representation in the lateral occipital complex.
Furthermore, on the basis of the evidence reviewed above, we
hypothesized that the body representation in FBA may be more
invariant to changes in view than EBA.
VIEWPOINT SENSITIVITY IN EBA AND FBA
1469
orthogonal to the extent of subsequent rotation. To disrupt an apparent
motion percept, successive frames were jittered by 100 pixels on the
x-axis— equivalent to 3.6° of visual angle, randomly from either right
to left or left to right (cf. Kourtzi and Shiffrar 1999). Each frame was
immediately followed by a stimulus specific mask (400 ms). Trials
concluded with a question mark task response cue (1,600 ms). Before
the start of the scan, participants were shown all 24 views of the target
pose and instructed that they should press a response button (when the
question mark appeared) if they had seen the target pose held by any
of the actors, in either trial frame, viewed from any angle. Each run
was bracketed with an additional five fixation-only epochs, giving a
total run time of 276 s (138 TR). Target RTs were recorded, allowing
independent modeling of both target trials and RTs. Trial orders were
counterbalanced such that every stimulus condition was preceded
equally often by every other condition, allowing an unbiased estimation of stimulus dependent HRFs (per subject, per run).
EBA and FBA were localized individually in two
separate runs, each comparing bodies to chairs. Each localizer run
consisted of 21 15-s blocks. Blocks 1, 6, 11, 16, and 21 were
fixation-only baseline epochs. In each of the remaining blocks, 20
images of either headless human bodies and images of chairs were
presented. Stimuli were each displayed for 300 ms, followed by a
blank screen for 450 ms.
LOCALIZERS.
DATA PREPROCESSING. To reduce the potential for T1 saturation,
seven dummy volumes (not processed) were acquired before each
experimental and localizer scan. Preprocessing entailed motion correction, with a maximum adjustment of ⫾2 mm (runs containing
corrections that exceeded these limits were rejected) and high-pass
filtering (0.006 Hz) to remove low-frequency and linear trends. The
analysis was performed without spatial smoothing. T1-weighted anatomical scans were transformed into Talairach space (Talairach and
Tournoux 1988). T2*-weighted, functional volumes were internally
J Neurophysiol • VOL
aligned to the first volume acquired for each participant manually
co-registered, within subjects, to T1-weighted anatomical scans. In
this way regional activation identified in functional scans could be
accurately attributed via spatial comparison with higher resolution
anatomical scans for individual participants.
ROI ANALYSIS. For each participant, GLMs were created for each
localizer and main experiment using a boxcar predictor, convolved to
a 2␥ hemodynamic response function (HRF) for each stimulus condition. EBA and FBA were localized in each participant by contrasting the BOLD signal response obtained for bodies with that obtained
for chairs [bodies – chairs]. ROIs were localized in both hemispheres
where present.
Objective definition of ROIs was ensured by a statistical attribution
method. The voxel exhibiting the strongest BOLD signal was identified within a relevant target region of the cortex suggested by current
literature (EBA: Downing et al. 2007; FBA: Peelen and Downing
2005). Each ROI was defined as the voxels exhibiting significant
activation (P ⬍ 0.0001, uncorrected), within a 9-mm cube surrounding the most strongly activated voxel. This approach further served to
demarcate ROIs from adjacent regions of selective activation and
limit them to only the most strongly activated voxels.
MAIN EXPERIMENT: GLM ANALYSIS. In the main experiment, predictors were defined as 300-ms events corresponding to the second
frame in each trial. Predictors were classified as follows: Different
Pose; Same/0°; Same/15°; Same/30°; Same/45°; Same/60°; Target
trial; and Button-press. The button-press predictor was modeled individually, by trial, from the button-press data and was defined as the
period spanning the offset of the second stimulus image in every
target image pair and the response time recorded for that target (i.e.,
the duration of this predictor varied with task performance). Within
each ROI, GLMs were calculated for all eight predictors, for each
participant.
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
FIG. 1. Trial structures for experiments
1–3: (A) experiment 1, masked; (B) experiment 2, immediate mask trials; (C) experiment 2, delayed mask trials; (D) experiment
3, unmasked. The figures show the 4 rendered actors used in all experiments and the
stimulus-specific pattern mask used in experiments 1 and 2. In all experiments, the
absolute viewpoint shown in frame 1 is chosen randomly, and the viewpoint shown in
frame 2 is defined relative to frame 1. The
pose held by the actor in frame 2 may be the
same or different from pose 1. The trial
examples above all show a “same pose” trial
with a 30° rotation.
1470
J. C. TAYLOR, A. J. WIGGETT, AND P. E. DOWNING
The time window used to define the button press predictor was
calculated on a per trial basis, varying with response time to the target.
Stimulus-dependent BOLD signal changes were expressed as standardized regression coefficients (␤) and subjected to repeated-measures ANOVA with planned contrasts.
Experiment 3
Twenty-two right-handed participants (12 female)
were recruited via the Bangor, School of Psychology Community
Panel. Ethics procedures were identical to experiments 1 and 2.
PARTICIPANTS.
Experiment 2
Ten right-handed participants (6 female) were recruited via the Bangor University Community Panel. Ethics procedures were identical to experiment 1.
PARTICIPANTS.
STIMULI AND SCANNING PARAMETERS. The stimulus set was identical to that used in experiment 1, with the exception that one of the
stimulus poses was replaced, as it was reported by participants to be
very similar to another pose in the set. All scanning and data
preprocessing procedures were the same as experiment 1. All instrument and stimulus presentation protocols, main experiment, and
EBA/FBA localizer stimuli were identical to experiment 1.
For the main experiment, participants were
scanned on 6 of 12 possible orders of a rapid event-related adaptation
design. Each image sequence comprised 91, 3-s trials, including 16
fixation-only, baseline trials. Nonbaseline trials comprised a serial
presentation of two frames, each depicting the same human figure
MAIN EXPERIMENT.
All apparatus and stimuli were
the same as experiment 1. The trial structure was identical to experiment 1, with the exception that the masks were replaced with blank
frames in the same color as the background. All scanning and data
preprocessing procedures were the same as experiments 1 and 2. The
trial structure is as shown in Fig. 2. Localization of EBA and FBA
used identical procedures to experiment 1. Data analysis procedures
were identical to experiment 1.
STIMULI, SCANNING, AND ANALYSIS.
RESULTS
Experiment 1
Because of a scanner malfunction, incomplete data sets were
obtained for two participants. Two further data sets were
FIG. 2. Example of functional regions of
interest (ROIs; left to right, sagittal, transverse, and coronal views in 1 subject, P ⬍
0.0001, uncorrected). The crosshairs indicate
extrastriate body areas (EBAs) bilaterally
(top) and extrastriate and fusiform body areas (FBAs) in the right hemisphere (bottom).
Talairach coordinates are given (top left) for
each slice.
J Neurophysiol • VOL
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
MAIN EXPERIMENT: TIME-COURSE MODELING. Raw MR signal was
extracted from each ROI (i.e., 1 value per TR) and assigned to
individual stimulus conditions. Alternating TRs—approximating to
the task response period in each trial—were coded separately. Mean
percent signal change (PSC) time-courses were calculated for every
stimulus condition in every trial: PSC (%) ⫽ (MRc – MRf)/MRf ⫻
100
Where MRc is the raw MR signal for an experimental condition and
MRf is the raw MR signal for the fixation condition. Average time
course data were used to confirm the presence of a stimulus dependent
HRF across a complete run for each subject. This was considered
necessary because of the possibility of both long-term adaptation and
participant fatigue in the final experimental run. Average time courses
for each run were therefore collapsed across trial conditions, and the
run was rejected if the time course showed no discernable stimulus
dependent profile.
(300 ms each). The first frame in each trial comprised any one of the
eight poses in any of the 24 possible views (randomized). The second
frame was either the same pose (2/3 trials) or a different pose (1/3
trials). The viewpoint of the first frame was randomized and the
orientation of the second frame was defined relative to the absolute
viewpoint of the first frame—rendering the extent of rotation orthogonal to the absolute viewpoints tested (as per experiment 1). Run
orders were (n ⫺ 1) counterbalanced such that all conditions were
preceded equally often by all other conditions.
To attain statistical power comparable to experiment 1, the number
of pose type conditions was reduced to three: different poses; same
pose (same angle: 0° condition); and same pose (different angle: 30°
condition). Each frame was followed by a stimulus specific mask (400
ms), either 0 (immediate condition) or 300 ms (delayed condition)
after the offset of each body image (Fig. 2). In the delayed condition,
the 300-ms poststimulus interval was filled with a blank frame in the
background color. Runs were bracketed with an additional five fixation-only epochs, giving a total run of 273 s (182 TR). In other
respects the procedure matched that of experiment 1.
VIEWPOINT SENSITIVITY IN EBA AND FBA
rejected on the basis of participant drowsiness. Only data from
the remaining nine participants was analyzed further.
ROI LOCALIZATION. The left and right EBA and right FBA
were successfully identified in all the remaining participants.
The mean spatial coordinates (with SE) of these regions, in
Talairach space, were as follows: left EBA: X: ⫺45(1.2), Y:
⫺73(0.6), Z: 1.5(1.9); right EBA: X: 47(0.6), Y: ⫺68(1.2), Z:
1.5(1.4); right FBA: X: 39(1.8), Y: ⫺46(1.7), Z: ⫺16(1.0).
Example ROI data are shown in Fig. 2.
Target detection was 84%, with all participants detecting a minimum of 40/48 targets. False alarm rate
was 2.5%, averaged across all trials.
BEHAVIORAL DATA.
ADAPTATION TO CHANGES IN BODY-VIEW: COMPARING DIFFERENT
AND PROGRESSIVELY ROTATED SAME POSES IN BILATERAL EBA
AND RIGHT-FBA. Comparing right versus left-EBA, a [2 (hemi-
sphere: L, R) ⫻ 6 (trial type: different, 0, 15, 30, 45, 60°)]
repeated-measures ANOVA showed no significant main effects or interactions: [hemisphere, F(1,8) ⫽ 0.04, P ⫽ 0.85;
trial type, F(5,40) ⫽ 5.45, P ⫽ 0.74; ROI ⫻ trial type,
F(5,40) ⫽ 1.12, P ⫽ 0.37]. Because no hemisphere-dependent
interaction was found, the EBA data were collapsed across
hemispheres for comparison with right-FBA (Fig. 3). Again,
no significant main effects [ROI, F(1,8) ⫽ 4.50, P ⫽ 0.067;
trial type, F(5,40) ⫽ 1.16, P ⫽ 0.35] or interactions [ROI ⫻
trial type, F(5,40) ⫽ 1.04, P ⫽ 0.41] were found. Within each
individual ROI, the effect of trial type was not significant and
all simple contrasts (0, 15, 30, 45, 60°, each vs. a reference
category of different pose) were not significant: F(1,8) ⫽ 0.22,
P ⫽ 0.65; F(1,8) ⫽ 2.62, P ⫽ 0.15; F(1,8) ⫽ 0.00, P ⫽ 1;
F(1,8) ⫽ 0.28, P ⫽ 0.61 and F(1,8) ⫽ 0.05, P ⫽ 0.84,
respectively.
Finally, to establish whether the raw effect of adaptation was
present, we used a paired samples t-test to compare the two
most extreme conditions (Same, 0° view change vs. Different
pose). All comparisons were not significant: in right-EBA:
t(8) ⫽ 0.996, P ⫽ 0.35; in left-EBA t(8) ⫽ ⫺0.027, P ⫽ 0.98;
and in FBA t(8) ⫽ 1.39, P ⫽ 0.20.
Experiment 2
Because of an instrument fault, no behavioral data were recorded for one participant. Target
detection was at ceiling for all remaining participants in
BEHAVIORAL DATA.
both masking conditions, with a total of five false alarms
(⬍1%) across all trials and subjects. However, a comparison
of RTs showed a 75-ms performance advantage in the
delayed masking condition, t(7) ⫽ 3.51, P ⬍ 0.01: delayed,
507 ms; immediate, 581 ms.
In the case of one participant, inconsistent shimming gave rise to unacceptable distortion in the L⬎R plane of
the functional images in the first localizer run. This participant
was excluded from further analyses. All subsequent analysis
was restricted to the eight participants satisfying motion correction and task performance criteria and showing right-FBA
and bilateral EBA.
fMRI DATA.
ROI LOCALIZATION. The left and right EBA and right FBA
were successfully identified in all of the remaining eight
participants at a threshold of P ⬍ 0.0001 (uncorrected). The
mean spatial coordinates (with SE) of these regions, in Talairach space, were as follows: left EBA: X: ⫺45(1.5), Y:
⫺70(2.5), Z: 2(1.9); right EBA: X: 45(1.2), Y: ⫺70(2.0), Z: ⫺2
(2.3); right FBA: X: 41(1.3), Y: ⫺47(2.7), Z: ⫺18(1.2).
Mean betas (Fig. 4) for the six predictors
of interest were entered into a [2 (hemisphere, left-EBA,
right-EBA) ⫻ 2 (trial type: delayed, immediate) ⫻ 3 (pose
type: same 0°; same 30°; different)] ANOVA. In line with
experiment 1, no significant interaction of hemisphere was
detected in EBA, and therefore the EBA data were collapsed
across hemispheres. Comparisons between collapsed EBA and
right-FBA showed significant main effects of trial type (delayed ⬎ immediate), F(1,7) ⫽ 8.16, P ⫽ 0.024 and ROI
(EBA ⬎ FBA) F(1,7) ⫽ 6.06, P ⫽ 0.043 and a significant
interaction of ROI ⫻ trial type: F(1,7) ⫽ 23.30, P ⫽ 0.001.
The effect of masking and pose type was examined within each
ROI. Comparisons showed a significant main effect of trial
type (delayed ⬎ immediate) in EBA: F(1,7) ⫽ 43.83, P ⬍
0.001. The main effect of trial type was marginally significant
in right-FBA: F(1,7) ⫽ 5.15, P ⫽ 0.058. The main effect of
pose type and the interaction of pose type ⫻ trial type were not
significant (P ⬎ 0.05) in all cases.
MAIN EXPERIMENT.
Experiment 3
The data for one participant was rejected on the basis of anomalous false alarm scores (16/42
targets identified vs. 173 false alarms). Data from one of three
scans by a further participant were also excluded because of
incorrect task switching (confirmed during debriefing). All
remaining subjects performed close to ceiling, correctly identifying ⱖ43/48 targets (⬍1% false alarms).
BEHAVIORAL MEASURES.
FIG. 3. Results of experiment 1. Graphs
show response magnitude (beta) vs. trial
condition for 6 trial categories in EBA (collapsed across hemispheres) and right-FBA
(n ⫽ 9). Error bars denote SE. Categories
denote pairs of different pose pairs (Diff)
and 5 magnitudes of rotation applied to same
pose pairs (0 – 60°).
J Neurophysiol • VOL
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Stimulus-dependent HRF curves were obtained in all ROIs for each
of the nine participants. On this basis, all data were included in
the subsequent GLM analysis.
QUALITATIVE ASSESSMENT OF FMRI MAIN EXPERIMENT DATA.
1471
1472
J. C. TAYLOR, A. J. WIGGETT, AND P. E. DOWNING
Excessive head motion (4 –9 mm) was found in four participant data sets, which
were dropped from further analyses. One further participant’s
data set was rejected on the basis that there was no stimulus
dependent HRF in right-FBA for two of the three experimental
runs.
QUALITATIVE ASSESSMENT OF fMRI DATA.
All analysis was restricted to the 15 participants satisfying motion correction and task performance
criteria and demonstrating right-FBA and bilateral EBA (a
threshold of P ⬍ 0.005 was required to obtain right FBA in 1
participant). The mean spatial coordinates (with SE) of these
regions, in Talairach space, were as follows: left EBA: X:
⫺46(1.1), Y: ⫺71(1.2), Z: 3.4(1.1); right EBA: X: 43(4.5), Y:
⫺71(0.8), Z: 1.7(1.2); right FBA: X: 33(5.5), Y: ⫺44(1.2), Z:
⫺17(0.8).
ROI LOCALIZATION.
ROI ANALYSES. The response pattern obtained for each ROI is
shown in Fig. 5. Comparing right versus left EBA, a [2
(hemisphere: L, R) ⫻ 6 (pose type: different, 0, 15, 30, 45,
60°)] repeated-measures ANOVA showed a significant main
effect of pose type only, F(5, 70) ⫽ 4.93, P ⫽ 0.001. The
interaction of pose type ⫻ hemisphere was not significant, so
the EBA response was collapsed across hemispheres for comparison with right-FBA. In this analysis, the main effect of
pose type only was significant, F(5,70) ⫽ 5.19, P ⬍ 0.001. The
interaction of pose type ⫻ ROI was not significant F(5,70) ⫽
0.44, P ⫽ 0.82. Separate analyses of pose type in each ROI
showed significant effects in both: EBA, F(5,70) ⫽ 4.93, P ⫽
0.001; FBA, F(5,70) ⫽ 3.71, P ⫽ 0.005.
Planned (simple) contrasts were undertaken in each ROI to
quantitatively assess the effect of viewpoint change on adaptation. In EBA, simple contrasts (0, 15, 30, 45, 60°, each vs. the
J Neurophysiol • VOL
reference category of different pose) were significant over the
range 0 –30°: F(1,14) ⫽ 12.77, P ⫽ 0.003; F(1,14) ⫽ 10.29,
P ⫽ 0.006; F(1,14) ⫽ 6.46, P ⫽ 0.024, marginally significant
at 45° F(1,14) ⫽ 4.60, P ⫽ 0.05, and not significant at 60°:
F(1,14) ⫽ 1.03, P ⫽ 0.33. In FBA, the same contrasts were
significant over the range 0 – 45°: F(1,14) ⫽ 14.45, P ⫽ 0.002;
F(1,14) ⫽ 12.70, P ⫽ 0.003; F(1,14) ⫽ 9.47, P ⫽ 0.008;
F(1,14) ⫽ 5.42, P ⫽ 0.035, respectively, and not significant at
60°: F(1,14) ⫽ 1.27, P ⫽ 0.28.
To further characterize the profile of viewpoint adaptation,
the same body views were considered apart from the (unadapted) different views. EBA response was collapsed across
hemispheres for comparison with right-FBA. Only the main
effect of angle was significant, F(4,56) ⫽ 2.75, P ⫽ 0.04. The
interaction of angle ⫻ ROI was not significant, F(4,56) ⫽ 0.55,
P ⫽ 0.70.
The demonstration of a systematic, view-dependent adaptation profile in the absence, but not in the presence, of pattern
masking (see experiment 1) suggests that pattern masking
significantly disrupts viewpoint dependent adaptation. However, a comparison between the findings of the two experiments is complicated by the differing sample sizes (experiment
1: n ⫽ 9; experiment 3, n ⫽ 15). It is therefore possible that the
significance of the results obtained in experiment 3 may be
partially attributable to an increase in statistical power rather
than solely to the removal of the mask. To address this
possibility, we undertook a simple Monte Carlo analysis of the
experiment 3 data set. We repeatedly (n ⫽ 100) randomly
sampled subsets of 9 data sets from the 15 available in
experiment 3 and compared the effect of raw adaptation (Same,
0° view change vs. Different pose) for each subsample. We
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
FIG. 4. Results of experiment 2. Mean response magnitudes (beta) for the 3 pose conditions (different, same/0°, same/30°) under each masking condition
(immediate, delayed). Error bars denote SE, n ⫽ 10.
VIEWPOINT SENSITIVITY IN EBA AND FBA
1473
found that 96% of EBA and 93% of FBA subsamples showed
a significant effect of adaptation.
If the pattern of response obtained in
the ROI analysis was caused by attentional artifacts or to
low-level retinotopic image changes, we would expect to see
effects of angle outside of the body-specific ROIs we examined. To test this, a random-effects whole brain group analysis
was performed, testing the linear effect of rotational angle in
the “same” conditions: 0, 15, 30, 45, 60°, [⫺2, ⫺1, 0, ⫹1, ⫹2].
A cluster threshold of four contiguous voxels and a statistical
threshold of P ⬍ 0.001 (uncorrected) were applied. For comparison, a second whole brain random effects analysis was
performed using localizer runs. Clusters were defined by the
contrast [bodies – objects] at a threshold of P ⬍ 0.05 (Bonferroni corrected). The more lenient threshold used for the
linear contrast reflected the relatively low power per-condition
of a five-level contrast derived from an event related design,
relative to the two condition blocked localizer scans.
Three clusters responded significantly to the linear effect of
viewpoint: Bilaterally, two areas of the lateral occipitotemporal
cortex, close to EBA, and an anterior portion of the left anterior
inferior temporal sulcus (aITS), at the boundary of BA20/21.
Peak Talairach coordinates and cluster sizes are summarized in
Table 1. To confirm that the lateral occipital activation was
similar to EBA, the linear contrast was superimposed on a map
WHOLE BRAIN ANALYSIS.
TABLE
1.
of the group-average localizer contrast (Fig. 6). Extremely
good agreement was found between the two areas, suggesting
that the viewpoint tuning effect is restricted to body-selective
regions. No significant clusters for the linear effect of viewpoint were observed in the fusiform gyrus and no bodyselective clusters were found in the vicinity of l-aITS. Furthermore, l-aITS was not body-selective: Mean beta values (n ⫽
15) for bodies and chairs: ⫺0.16 (SE ⫽ 0.03) and ⫺0.15
(SE ⫽ 0.03), respectively. The contrast [bodies ⫺ chairs] was
not significant, t(14) ⫽ ⫺0.46 (P ⫽ 0.65).
DISCUSSION
Experiment 1
No view-dependent adaptation was obtained in any of the
body-selective ROIs that were examined. Indeed, no significant
response differences were seen even for the comparison of an
identical repetition of a pose to two different poses. Task
performance was strong, suggesting that participants were
attending to the relevant stimuli features. A qualitative assessment of stimulus-dependent time-courses confirmed that, in all
participants, in all runs, the stimuli were effectively engaging
the ROIs. Furthermore, the magnitude of the t-values obtained
for the comparison of (Same, 0° view change vs. Different
pose) indicate that effect sizes are so small that no significant
Cortical regions responsive to the linear effect of rotational angle
Talairach Coordinates (Peak)
Region
x
y
z
Number of Voxels
t
Left occipitotemporal cortex
Left inferior temporal sulcus
Right occipitotemporal cortex
⫺42.00
⫺60.00
51.00
⫺67.00
⫺4.00
⫺64.00
7.00
⫺17.00
⫺2.00
144
244
377
5.10
4.60
5.10
J Neurophysiol • VOL
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
FIG. 5. Results of experiment 3. Mean
response magnitude (beta) in bilateral EBA
and right FBA for Same/0 – 60° and Different pose conditions. Parentheses denote SE
(n ⫽ 15). Asterisks denote significance of
planned simple contrasts relative to Different pose trials. *P ⬍ 0.05, **P ⬍ 0.01.
1474
J. C. TAYLOR, A. J. WIGGETT, AND P. E. DOWNING
effect of raw adaptation would be detectable at any practically
attainable sample size. Thus the absence of the predicted
stimulus adaptation seems unlikely to be attributable either to
statistical power, attentional factors (Murray and Wojciulik
2004) or to an inadequate response from the regions of interest.
Although the rapid fMRI-A paradigm has been used in many
previous studies, the stimuli are not routinely masked in
designs of this type (Fang et al. 2005; Kourtzi et al. 2003;
Schiltz and Rossion 2006; Weigelt et al. 2007). The purpose of
the mask in the current study was to assist in fully disrupting
any percept of apparent motion between the two frames of the
display, to prevent interpolation between successive body image frames in higher visual areas (Liu et al. 2004). This was
because such interpolation would interfere with the aim of the
study to examine the view specificity of body-selective neural
populations (Kourtzi and Shiffrar 1997, 1999).
However, in this study, it is likely that the mask also
interrupted ongoing visual analysis of the body images in EBA
and FBA. When a visual image containing a human body is
attended, the neural activity in body-selective areas presumably develops systematically over a brief time window, normally culminating in a full representation of the stimulus. In
experiment 1, the mask appeared 300 ms after the offset of
each body stimulus. Intracranial (McCarthy et al. 1999; Pourtois et al. 2007) and scalp (Gliga and Deheane-Lambertz 2005;
Stekelenburg and de Gelder 2004; Thierry et al. 2006) ERP
studies show a body-selective negativity peaking at around 200
ms poststimulus. While little further evidence is available for
body stimuli, a recent study of the MEG analog of the faceselective N170 (M170: Liu et al. 2002), showed this component to be significantly viewpoint dependent over small rotational angles (Ewbank et al. 2007). These findings raise the
possibility that the emergence of a complete representation in
EBA and FBA extends beyond 200 ms and may have been
disrupted by masking.
J Neurophysiol • VOL
Experiment 2 further examined the role of the mask in
eliminating adaptation effects in experiment 1 by manipulating
the timing of the mask onset. Masks were either presented at 0
ms relative to stimulus offset (immediate mask trials, comparable to experiment 1) or delayed by 300 ms (delayed mask
trials). Two specific predictions follow from this manipulation.
First, RTs should be faster in the delayed mask condition,
because more time is available after the second body stimulus
in which to select and prepare the motor response. Second, and
more important, if delaying the mask weakens or abolishes its
interfering effects on the creation of a body representation in
the regions of interest, we would expect a significant modulation of the BOLD responses in EBA/FBA as a function of the
change in viewpoint from the first to the second body image.
Experiment 2
The data from experiment 2 showed that immediate pattern
masking, relative to delayed pattern masking, produced a
decrement in task performance as indexed by RT. In EBA
bilaterally, but only marginally in FBA, delaying the mask
onset significantly increased the BOLD response to bodies.
However, in both ROIs, the absence of any view-dependent
adaptation, even in the delayed mask condition, indicates that
adaptation effects on body representations in these areas can
still be substantially disrupted as late as 600 ms after stimulus
onset. Notably the effect of masking differed between the two
ROIs, suggesting a different time course of developing representations in these areas.
In experiment 3, we tested whether removing the mask
altogether would show the sensitivity (or lack thereof) to
changes in body viewpoint in the EBA and FBA. We repeated
the design used in experiment 1, replacing the pattern mask
with a blank, grayscale frame in the same color as the background.
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
FIG. 6. Results of whole brain analysis of
experiment 3. Statistical maps (A and B) show the
high degree of spatial overlap between EBA
(green areas, bilaterally) and clusters defined by
the linear effect of rotational angle from experiment 3. Statistical maps (C and D) show the
absence of spatial overlap between body selectivity in FBA (purple areas) and the linear effect of
angle in left anterior inferior temporal sulcus (orange areas). EBA and FBA are defined by a
whole brain random effects analysis for the contrast [bodies – objects], P ⬍ 0.05 (Bonferroni
corrected). The linear effect of angle is defined by
the contrast: 0, 15, 30, 45, and 60° [ ⫺2, ⫺1, 0,
1, 2], P ⬍ 0.001 (uncorrected).
VIEWPOINT SENSITIVITY IN EBA AND FBA
Experiment 3
General Discussion
The extent of viewpoint sensitivity in EBA and FBA is in
close agreement with behavioral data examining the perception
of synthetic objects (Kourtzi and Shiffrar 1999) and mannequins (Kourtzi and Shiffrar 1998) across changing viewpoints
J Neurophysiol • VOL
and with fMRI data examining object representations in the
LOC (Kourtzi et al. 2003). The studies present new evidence
that the viewpoint invariance of cortical body representations
extends to ⱖ30° but to ⬍60°. Similarly, a recent imaging study
examining sensitivity to changes in viewpoint for vehicles and
animals (Andresen et al. 2009) found that the rotational limit
for adaptation effects in these categories was also ⬍60° in both
LO and pFs. Thus our data on body representations in posterior
and anterior body-selective regions appear broadly consistent
with current data on general object-viewpoint selectivity.
One aim of the study was to compare the properties of the
two body-selective regions of extrastriate cortex. The contrast
analysis of the linear effect of angle was significant in EBA but
not FBA, and this observation was supported by whole brain
analysis. However, no interaction was found between EBA and
FBA—indeed, the overall patterns of response were visually
similar—and the patterns of adaptation suggest broadly similar
tuning properties in these areas. Furthermore, in experiment 2,
delaying the mask produced an increase in the response to the
body stimuli in bilateral EBA but not FBA. Thus the additional
300 ms facilitated a measurable strengthening of the representation in EBA but not FBA. Taken together, these two findings
hint at distinctions between EBA and FBA but do not provide
strong evidence for a functional dissociation (cf. Hodzic et al.
2009; Taylor et al. 2007).
In addition to the effects seen in our a priori regions of
interest, we also found a significant linear effect of viewpoint
in a left-anterior portion of ITS (corresponding to Brodmann’s
area 21). This region did not coincide with body-selective
areas. One previous study (Vanrie et al. 2001) implicated a
similar region in a subset of participants, in a task requiring the
detection of local feature differences when making decisions
about the similarity of objects viewed from different angles
(our peak coordinates: x, y, z: ⫺60, ⫺4, ⫺17; Vanrie et al. x,
y, z: ⫺67, ⫺6, ⫺16). Although a significant body of literature
posits a primary role for nearby areas in the functional organization of semantic information about person identity and
naming (Damasio et al. 1996; Tsukiura et al. 2002), the
Talairach coordinates reported are typically more lateral
(Tsukiura et al. 2006, x, y, z: ⫺42, ⫺11 ⫺22; Grabowski et al.
2001: ⫺37, ⫺14, ⫺20) compared with those identified here.
Although the evidence remains tentative, our findings and
those of Vanrie et al. (2001) are suggestive of a possible
category-general role in the decoding of spatial rotation.
The unexpected effects of pattern masking in experiments 1
and 2 allow us to make some speculative inferences about the
time course of neural processes of the body-selective regions
we tested here. The pair of body images jointly produced a
strong BOLD response in the ROIs, and the ability to detect a
prespecified target in a single frame was intact. Therefore—to
the extent that neural activity in these regions supports body
perception (Moro et al. 2008; Peelen and Downing 2007;
Pitcher et al. 2009; Urgesi et al. 2007a)— masking did not
seem to abolish that activity. However, there was no fMRI
adaptation effect in either region, even for an exact repetition
(in different spatial locations) of a body image. It is likely,
however, that the masks interrupted the later stages of neural
activity in these regions, which perhaps are influenced by
reciprocal interactions with higher level cortical areas (Fahrenfort et al. 2007). Therefore a working hypothesis is that the
BOLD adaptation effect observed in experiment 3 was driven
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
In experiment 3, we showed that bilateral EBA and right
FBA undergo view-dependent adaptation to bodies and that the
pattern of view invariance is significantly linear in EBA but not
in FBA (although the patterns in these regions are highly
similar in both shape and extent). Comparison of these results
to the first two experiments shows that masking the stimuli
substantially disrupts the analysis of bodies in EBA and FBA
and specifically prevents these areas from establishing a representation that is partially invariant to viewing angle. Furthermore, both the Monte-Carlo analysis of experiment 3 data and
the extremely small effect sizes indicated by the t-test of raw
adaptation in experiment 1 indicate that the different patterns of
results from the two experiments cannot be attributed simply to
differences in sample size.
The results of the functional ROI analyses were confirmed in
a whole brain analysis, which showed that the linear response
to increasingly large changes in viewpoint did not extend
widely throughout visual cortex but was restricted to lateral
occipitotemporal cortex. The absence of significant linear effects in the early visual cortex indicates that the origin of the
effect is not retinotopic. Furthermore, the highly focal responses in lateral OTC, consistent with the location of EBA,
suggest that the monotonic pattern is not a global attentional
artifact, which would be expected to have widespread effects.
We also considered the close proximity of EBA to the
motion sensitive hMT⫹. Because there is a potential confound
between the angular difference between two images and the
magnitude of any perceived motion attributable to that difference, could a motion percept account for the pattern of results
shown in experiment 3? This possibility can be rejected on two
counts: first previous studies (Kourtzi and Shiffrar 1997, 1999)
have shown that apparent motion is reliably disrupted under
similar interstimulus interval (ISI) and jitter conditions to those
used here—indeed we used a larger spatial jitter than has been
shown to be necessary. Furthermore, in this study (and during
piloting), no motion percept was visible and none was reported.
Second, we also obtained strong view-dependent adaptation in
FBA, which is not spatially adjacent to hMT⫹ and is distant
from other classical motion-selective areas, and hence the
effect in this area is unlikely to be an artifact of apparent
motion.
In this study, as in the previous two, behavioral performance
was near ceiling on the target-detection task. This may seem at
first to be paradoxical—performance might be expected to drop
when masking impairs the formation of neural body representations. However, given the extensive practice participants had
with the target images, it is likely that they performed the task
on the basis of detecting simple local features that co-vary with
the particular target pose used. Also, note that the target trials
were not included in the fMRI data analyses so performance on
these trials does not contribute to the patterns of adaptation
observed.
1475
1476
J. C. TAYLOR, A. J. WIGGETT, AND P. E. DOWNING
J Neurophysiol • VOL
FFA) have been shown to decrease as lag-time increases. This
effect is stable across multiple stimulus categories, suggesting
that such temporal contingencies may be a general property of
object sensitive regions (Henson 2003; Henson et al. 2000,
2002). This study adds to this picture, indicating that examining adaptation effects, coupled with systematic manipulation of
stimulus masking, may prove a useful tool to further investigate category-selective regions of the extrastriate cortex. This
combination has the potential to add information about how
representations evolve over time and how this relates to neurophysiological evidence and to the temporal demands of real
world viewing tasks.
In summary, we showed viewpoint selective tuning to bodies in EBA and FBA and found some provisional evidence for
a category-independent effect in the left aITS. The close
agreement between the view-dependent tuning in EBA and
FBA reported here and that seen for other object categories in
adjacent regions of the LOC suggests analogous representational processes, operating across stimulus classes. We interpret the abolition of view dependent fMRI adaptation by
pattern masking as evidence that the view-dependent fMRI
adaptation effect seen in experiment 3 is driven by later waves
of neuronal responses in the regions of interest.
GRANTS
This research was supported by a Biotechnology and Biological Sciences
Research Council grant to P. E. Downing, an Economic and Social Research
Council (ESRC) studentship to J. C. Taylor, and a Wales Institute of Cognitive
Neuroscience grant to P. E. Downing and J. C. Taylor.
REFERENCES
Andresen DR, Vinberg J, Grill-Spector K. The representation of object
viewpoint in human visual cortex. Neuroimage 45: 522–536, 2009.
Andrews TJ, Ewbank M. Distinct representations for facial identity and
changeable aspects of faces in the human temporal lobe. Neuroimage 23:
905–913, 2004.
Betts LR, Wilson HR. fMRI adaptation in fusiform face area and occipital
face area. Perception 36: 145, 2007.
Boehm SG, Paller KA. Do I know you? Memory for faces from brain
potentials. Clin EEG Neurosci 37: 322–329, 2006.
Boynton GM, Finney EM. Orientation-specific adaptation in human visual
cortex. J Neurosci 23: 8781– 8787, 2003.
Buckner RL, Goodman J, Burrock M, Rotte M, Koutstaal W, Schacter D,
Rosen B, Dale AM. Functional-anatomic correlates of object priming in humans
revealed by rapid presentation event related fMRI. Neuron 20: 285–296, 1998.
Chan AW-Y, Peelen M, Downing P. The effect of viewpoint on body representation in the extrastriate body area. Neuroreport 15: 2407–2410, 2004.
Damasio H, Grabowski TJ, Tranal D, Hichwa, RD, Damasio AR. A neural
basis for lexical retrieval. Nature 380: 499 –505, 1996.
Downing PE, Jiang YH, Shuman M, Kanwisher N. A cortical area selective
for visual processing of the human body. Science 293: 2470 –2473, 2001.
Downing PE, Wiggett AJ, Peelen MV. Functional magnetic resonance
imaging investigation of overlapping lateral occipitotemporal activations
using multi-voxel pattern analysis. J Neurosci 27: 226 –233, 2007.
Dricot L, Sorger B, Schiltz C, Goebel R, Rossion B. Evidence for individual
face discrimination in non-face selective areas of the visual cortex in
acquired prosopagnosia. Behav Neurol 19: 75–79, 2008.
Drucker DM, Aguirre GK. Different spatial scales of shape similarity representation in lateral and ventral LOC. Cereb Cortex 19: 2269 –2280, 2009.
Eger E, Schyns PG, Kleinschmidt A. Scale invariant adaptation in fusiform
face-responsive regions. Neuroimage 22: 232–242, 2004.
Epstein RA, Parker WE, Feiler AM. Two kinds of fMRI repetition suppression? Evidence for dissociable neural mechanisms. J Neurophysiol 99:
2877–2886, 2008.
Ewbank MP, Smith WAP, Hancock ER, Andrews TJ. The M170 reflects a
viewpoint-dependent representation for both familiar and unfamiliar faces.
Cerebral Cortex 18: 364 –370, 2007.
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
primarily by these later aspects of the neural response, which
were interrupted in experiments 1 and 2. One (indirect) line of
supporting evidence is that event-related potential responses to
faces show repetition effects on later but not earlier faceselective components (Boehm and Paller 2006). (Of course a
further possibility is that neuronal and BOLD dynamics are
coupled in a complex way such that a preserved neuronal
adaptation effect is not transmitted via concomitant changes in
the hemodynamic response; cf. Sawamura et al. 2006). An
additional useful test would be to present masks at still further
poststimulus delays, with the prediction that fMRI adaptation
effects would return when later neural responses were no
longer influenced by masking. However, this would require
significant changes to the trial design used here, confounding
comparisons with the present experiments.
One additional consideration arises from the trial structure in
experiments 1 and 2: Although both frames were backward
masked, frame 2 was also effectively forward masked with a
⫺400-ms SOA (i.e., by the mask following frame 1). Because
the design of experiment 3 did not directly address this issue,
it is necessary to consider the possible implications of forward
masking on the abolition of adaptation. Perrett et al. (2009)
showed the temporally extended effects of forward masking on
the firing rate of category selective neurons in macaque STS.
Masking was shown to attenuate firing rates evoked by target
stimuli for as long as 400 ms following 125-ms mask presentations. However, the effectiveness of forward masking of this
type was directly related to the similarity (determined independently by stand-alone firing rate) between the mask and the
target. In this case, the finely scrambled patterns used in the
masks carry little object structure and so probably do not
effectively drive activity in the body-selective regions tested
here (see also Downing et al. 2001). This suggests that scrambled images may not be very effective forward masks on neural
activity in EBA and FBA; however, more direct manipulations
would be required to separately assess the contributions of
forward and backward masking on adaptation in these areas.
Finally, task performance (as indexed by hit rate) was similarly
high in experiments 1 and 2, suggesting that participants were
easily able to identify the pose (structure) of the second frame
in the majority of target trials. This also suggests that the effect
of forward masking on explicit perception was very weak.
Our understanding of high-level visual areas benefits from
considering category-selective response properties not as fixed,
but as evolving over time. Single-unit recording studies in the
macaque IT cortex suggest a progressive emergence of category selectivity, with greater selectivity during later spiking
(⬃240 ms following stimulus onset) compared with early
transient responses at ⬃130 ms (Tamura and Tanaka 2001).
Convergent findings are reported by Sugase et al. (1999).
These authors reported a faster emergence of global compared
with fine category information, with former being maximal
during the spiking phase and the latter maximal during the
decay phase. Significant information transmission was sustained to ⱖ500 ms following onset. In human neuroimaging
studies, comparisons of short-term and long-term adaptation
indicate that the adaptation properties of face selective regions
of visual cortex (i.e., OFA/FFA) may differ depending on the
duration of the adaptation period (Fang et al. 2007; Kovács et
al. 2008). Furthermore, the magnitude of both repetition suppression (in LO) and suppression and enhancement (in pFs/
VIEWPOINT SENSITIVITY IN EBA AND FBA
J Neurophysiol • VOL
Peelen MV, Downing PE. The neural basis of visual body perception. Nat Rev
Neurosci 8: 637– 648, 2007.
Perrett DI, Xiao D, Keysers C, Oram MW. Seeing the future: neural image
sequences produce “anticipatory” neuronal activity and bias perceptual
report. Q J Exp Psychol B 62: 2081–2104, 2009.
Peuskens H, Claeys KG, Todd JT, Norman JF, Van Hecke P, Orban GA.
Attention to 3-D shape, 3-D motion, and texture in 3-D structure from
motion displays. J Cogn Neurosci 16: 665– 682, 2004.
Pitcher D, Charles L, Devlin JT, Walsh V, Duchaine B. Triple dissociation
of faces, bodies, and objects in extrastriate cortex. Current Biology 19:
319 –324, 2009.
Pourtois G, Peelen M, Spinelli L, Seeck M, Vuilleumier P. Direct intracranial recording of body-selective responses in human extrastriate visual
cortex. Neuropsychologia 45: 2621–2625, 2007.
Rotshtein P, Malach R, Hadar U, Graif M, Hendler T. Feeling or features:
different sensitivity to emotion in high-order visual cortex and amygdala.
Neuron 32: 747–757, 2001.
Sawamura H, Orban GA, Vogels R. Selectivity of neuronal adaptation does
not match response selectivity: a single-cell study of the FMRI adaptation
paradigm. Neuron 49: 307–318, 2006.
Saxe R, Jamal N, Powell L. My body or yours? The effect of visual perspective
on cortical body representations. Cereb Cortex 16: 178 –182, 2006.
Schiltz C, Rossion B. Faces are represented holistically in the human occipitotemporal cortex. Neuroimage 32: 1385–1394, 2006.
Schwarzlose RF, Baker CI, Kanwisher N. Separate face and body selectivity
on the fusiform gyrus. J Neurosci 25: 11055–11059, 2005.
Schwarzlose RF, Swisher JD, Dang SB, Kanwisher N. The distribution of
category and location information across object-selective regions in human
visual cortex. Proc Natl Acad Sci USA 105: 4447– 4452, 2008.
Stekelenburg J, de Gelder B. The neural correlates of perceiving human
bodies: an ERP study on the body-inversion effect. Neuroreport 15: 777–
780, 2004.
Sugase Y, Yamane S, Ueno S, Kawano K. Global and fine information coded
by single neurons in the temporal visual cortex. Nature 400: 869 – 873, 1999.
Talairach J, Tournoux P. Co-Planar Stereotaxic Atlas of the Human Brain.
New York: Thieme, 1988.
Tamura H, Tanaka K. Visual response properties of cells in the ventral and
dorsal parts of the macaque inferotemporal cortex. Cereb Cortex 11:
384 –399, 2001.
Taylor JC, Wiggett AJ, Downing PE. Functional MRI analysis of body and
body part representations in the extrastriate and fusiform body areas.
J Neurophysiol 98: 1626 –1633, 2007.
Thierry G, Martin CD, Downing P, Pegna AJ. Controlling for interstimulus
perceptual variance abolishes N170 face selectivity. Nat Neurosci 10:
505–511, 2007.
Thierry G, Pegna AJ, Dodds C, Roberts M, Basan S, Downing P. An
event-related potential component sensitive to images of the human body.
Neuroimage 32: 871– 879, 2006.
Tsukiura T, Fujii T, Fukatsu R, Otsuki T, Okudu J, Ometsu A, Suzuki K,
Tabuchi M, Yanagawa J, Nagasaka T, Kawashima R, Fukuda H,
Takahashi S, Yamadori A. Neural basis of retrieval of people’s names:
evidence from brain damaged patients and fMRI. J Cogn Neurosci 14:
922–937, 2002.
Tsukiura T, Mochizuki-Kawai H, Fujii T. Dissociable roles of the bilateral
anterior temporal lobe in face-name associations: an event related fMRI
study. Neuroimage 30: 617– 626, 2006.
Ullman S. Object recognition and segmentation by a fragment-based hierarchy. Trends Cogn Sci 11: 58 – 64, 2007.
Urgesi C, Calvo-Merino B, Haggard P, Aglioti SM. Transcranial magnetic
stimulation reveals two cortical pathways for visual body processing. J Neurosci 27: 8023– 8030, 2007a.
Urgesi C, Candidi M, Ionta S, Aglioti SM. Representation of body identity
and body actions in extrastriate body area and ventral premotor cortex. Nat
Neurosci 10: 30 –31, 2007b.
Vanrie J, Beatse E, Wagemans J, Sunaert S, Van Hecke P. Mental rotation
versus invariant features in object perception from different viewpoints: an
fMRI study. Neuropsychologia 40: 917–930, 2001.
Weigelt S, Kourtzi Z, Kohler A, Singer W, Muckli L. The cortical representation of objects rotating in depth. J Neurosci 27: 3864 –3874, 2007.
Winston JS, Henson RNA, Fine-Goulden MR, Dolan RJ. fMRI-adaptation
reveals dissociable neural representations of identity and expression in face
perception. J Neurophysiol 92: 1830 –1839, 2004.
103 • MARCH 2010 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.1 on June 18, 2017
Fahrenfort JJ, Scholte HS, Lamme VAF. Masking disrupts reentrant processing in human visual cortex. J Cogn Neurosci 19: 1488 –1497, 2007.
Fang F, Murray SO, He S. Duration-dependent fMRI adaptation and distributed viewer-centered face representation in human visual cortex. Cereb
Cortex 17: 1402–1411, 2007.
Fang F, Murray SO, Kersten D, He S. Orientation tuned fMRI adaptation in
human visual cortex. J Neurophysiol 94: 4188 – 4195, 2005.
Gliga T, Deheane-Lambertz G. Structural encoding of body and face in
human infants and adults. J Cogn Neurosci 17: 1328 –1340, 2005.
Grabowski TJ, Damasio H, Tranel D, Ponto LL, Hichwa RD, Damasio
AR. A role for the left tempral pole in the retrieval of words for unique
entities. Hum Brain Mapp 13: 199 –212, 2001.
Grill-Spector K, Henson R, Martin A. Repetition and the brain: neural
models of stimulus-specific effects. Trends Cogn Sci 10: 14 –23, 2006.
Grill-Spector K, Kushnir T, Edelman S, Avidan G, Itzchak Y, Malach R.
Differential processing of objects under various viewing conditions in the
human lateral occipital complex. Neuron 24: 187–203, 1999.
Grill-Spector K, Malach R. fMR-adaptation: a tool for studying the functional
properties of human cortical neurons. Acta Pyschol 107: 293–321, 2001.
Grill-Spector K, Malach R. The human visual cortex. Annu Rev Neurosci 27:
649 – 677, 2004.
Harris A, Aguirre GK. The representation of parts and wholes in faceselective cortex. J Cogn Neurosci 20: 863– 878, 2008.
Haushofer J, Livingstone MS, Kanwisher N. Multivariate patterns in objectselective cortex dissociate perceptual and physical shape similarity. PLoS
Biol 6: 1459 –1467, 2008.
Henson RN. Neuroimaging studies of priming. Prog Neurobiol 70: 53– 81,
2003.
Henson R, Shallice T, Dolan R. Neuroimaging evidence for dissociable forms
of repetition priming. Science 287: 1269 –1272, 2000.
Henson RNA, Shallice T, Gorno-Tempini ML, Dolan RJ. Face repetition
effects in implicit and explicit memory tests as measured by fMRI. Cereb
Cortex 12: 178 –186, 2002.
Hodzic A, Muckli L, Singer W, Stirn A. Cortical responses to self and others.
Hum Brain Mapp 30: 951–962, 2009.
James TW, Humphrey GK, Gati JS, Menon RS, Goodale MA. Differential
effects of viewpoint on object-driven activation in dorsal and ventral
streams. Neuron 35: 793– 801, 2002.
Kable JW, Chatterjee A. Specificity of action representations in the lateral
occipitotemporal cortex. J Cogn Neurosci 18: 1498 –1517, 2006.
Kourtzi Z, Erb M, Grodd W, Bulthoff HH. Representation of the perceived
3-D object shape in the human lateral occipital complex. Cereb Cortex 13:
911–920, 2003.
Kourtzi Z, Kanwisher N. Representation of perceived object shape by the
human lateral occipital complex. Science 293: 1506 –1509, 2001.
Kourtzi Z, Shiffrar M. One-shot view invariance in a moving world. Psychol
Sci 8: 461– 466, 1997.
Kourtzi Z, Shiffrar M. Dynamic representations of biological movement.
Perception 28: 49 – 62, 1998.
Kourtzi Z, Shiffrar M. The visual representation of three-dimensional,
rotating objects. Acta Psychol 102: 265–292, 1999.
Kovács G, Cziraki C, Vidnyánszky Z, Schweinberger S, Greenlee M.
Position-specific and position-invariant face aftereffects reflect the adaptation of different cortical areas. Neuroimage 43: 156 –164, 2008.
Lerner Y, Hendler T, Ben-Bashat D, Harel M, Malach R. A hierarchical
axis of object processing stages in the human visual cortex. Cereb Cortex
15: 287–297, 2001.
Lerner Y, Hendler T, Malach R. Object-completion effects in the human
lateral occipital complex. Cereb Cortex 12: 163–177, 2002.
Liu J, Harris A, Kanwisher N. Stages of processing in face perception: an
MEG study. Nat Neurosci 5: 910 –916, 2002.
Liu T, Slotnick SD, Yantis S. Human MT⫹ mediates perceptual filling-in
during apparent motion. Neuroimage 21: 1772–1780, 2004.
McCarthy G, Puce A, Belger A, Allison T. Electrophysiological studies of
human face perception. II. Response properties of face-specific potentials
generated in occipitotemporal cortex. Cereb Cortex 9: 431– 444, 1999.
Moro V, Urgesi C, Pernigo S, Lanteri P, Pazzaglia M, Aglioti SM. The neural
basis of body form and body action agnosia. Neuron 60: 235–246, 2008.
Murray SO, Wojciulik E. Attention increases neural selectivity in the human
lateral occipital complex. Nat Neurosci 7: 70 –74, 2004.
Myers A, Sowden PT. Your hand or mine? The extrastriate body area.
Neuroimage 42: 1669 –1677, 2008.
Peelen MV, Downing PE. Selectivity for the human body in the fusiform
gyrus. J Neurophysiol 93: 603– 608, 2005.
1477