FAST-TRACK REPORT Differential development of selectivity for faces and bodies in the fusiform gyrus

Transcription

1 Developmental Science (2009), pp F16 F25 DOI: /j x FAST-TRACK REPORT Differential development of selectivity for faces and bodies in the fusiform gyrus Marius V. Peelen, 1,2,3 Bronwyn Glaser, 4,5 Patrik Vuilleumier 1,2,6 and Stephan Eliez 4,7 1. Swiss Center for Affective Sciences, University of Geneva, Switzerland 2. Laboratory for Neurology and Imaging of Cognition, Department of Neurosciences, Geneva University Medical Center, Switzerland 3. Department of Psychology and Princeton Neuroscience Institute, Princeton University, USA 4. Service MØdico-PØdagogique, Department of Child Psychiatry, University of Geneva Medical School and University Hospitals, Switzerland 5. Department of Psychology and Education, University of Geneva, Switzerland 6. Department of Neurology, Geneva University Hospital, Switzerland 7. Department of Genetic Medicine and Development, University of Geneva Medical School and University Hospitals, Switzerland Abstract Viewing faces or bodies activates category-selective areas of visual cortex, including the fusiform face area (FFA), fusiform body area (FBA), and extrastriate body area (EBA). Here, using fmri, we investigate the development of these areas, focusing on the right FFA and FBA. Despite the overlap of functionally defined FFA and FBA (54% 75% overlap), we found that these regions developed along different trajectories. With age (7 32 years old), the FFA gradually increased in size and selectivity, and was significantly larger and more face-selective in adults than children. By contrast, the size and selectivity of the FBA did not correlate with age, and were equivalent in children and adults. Whereas in adults the FFA and FBA were comparable in size, in children the FBA was on average 70% larger than the FFA. These findings suggest that, in children, the fusiform gyrus is predominantly selective for bodies, with commensurate face-selective responses apparent later in development. Moreover, differences in the development of the FFA and FBA indicate that overlapping functional brain areas, supported by the same anatomical structure, can develop along different trajectories. Introduction Quick and accurate perception of the identities, actions, intentions, and emotional states of conspecifics is critical for survival. Accordingly, healthy human adults have developed remarkable proficiency and expertise in perceiving various aspects of other people s faces and bodies (Diamond & Carey, 1986; Kozlowski & Cutting, 1977; Mondloch, Geldart, Maurer & Le Grand, 2003; Reed, Stone, Bozova & Tanaka, 2003; Slaughter, Stone & Reed, 2004). Recent neuroimaging studies have identified several face- and body-selective visual brain areas that may support these perceptual skills (Downing, Jiang, Shuman & Kanwisher, 2001; Kanwisher, McDermott & Chun, 1997; Peelen & Downing, 2005a). For example, two areas in right fusiform gyrus have been identified that respond selectively to either faces (fusiform face area or FFA; Kanwisher et al., 1997) or bodies (fusiform body area or FBA; Peelen & Downing, 2005a). Across a group of subjects, the anatomical locations of the FFA and FBA are nearly identical, and even within individual subjects these areas largely overlap (Peelen & Downing, 2005a), so that they are typically identified by contrasting faces or bodies to a third control category (e.g. tools). However, these overlapping fusiform responses can be dissociated with high-resolution fmri (Schwarzlose, Baker & Kanwisher, 2005) or multi-voxel pattern analysis (Peelen, Wiggett & Downing, 2006), suggesting that, at a neuronal level, the selectivity for faces and bodies may be independent or non-overlapping. In the present study, we examined the development of the FFA and FBA during late childhood. Of particular interest was the direct comparison of the developmental trajectories of these two areas, to test whether their close proximity and overlap is translated into similar developmental trajectories, or, instead, whether the areas can be dissociated in this respect. Address for correspondence: Marius V. Peelen, Department of Psychology and Princeton Neuroscience Institute, Princeton University, Princeton, NJ, USA; mpeelen@princeton.edu Ó 2009 Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

2 Development of selectivity for faces and bodies F17 Recent developmental fmri studies investigating the development of the FFA have provided evidence for developmental changes well into late childhood (Aylward, Park, Field, Parsons, Richards, Cramer & Meltzoff, 2005; Gathers, Bhatt, Corbly, Farley & Joseph, 2004; Golarai, Ghahremani, Whitfield- Gabrieli, Reiss, Eberhardt, Gabrieli & Grill-Spector, 2007; Passarotti, Paul, Bussiere, Buxton, Wong & Stiles, 2003; Scherf, Behrmann, Humphreys & Luna, 2007). For example, in two recent studies, the right FFA (functionally defined in each individual) was larger (Golarai et al., 2007; Scherf et al., 2007) and more face-selective (Scherf et al., 2007) in adults than children. Furthermore, the size of the right FFA correlated with performance on a face recognition memory task, indicating that the gradual increase in FFA size was linked to the increase in face perception skills (Golarai et al., 2007). To date, no study has investigated the development of areas involved in processing the rest of the body. Apart from the FBA discussed above, bodies and body parts also selectively activate a bilateral region in lateral occipito-temporal cortex, the extrastriate body area (EBA; Downing et al., 2001). This region responds strongly to human bodies and body parts, but weakly to faces and other objects. Here we investigate and compare the development of face- and body-selective areas. Because the FFA and FBA occupy nearly identical cortical territory, it is possible to directly compare the developmental trajectories of these areas, without confounds typically associated with developmental neuroimaging, such as regional and developmental differences in the shape and variance of the hemodynamic response function (Grill-Spector, Golarai & Gabrieli, 2008). Therefore, our main focus was on the development of the FFA and FBA, and to test whether these regions develop along similar or different developmental trajectories. Methods Subjects Fifty-one volunteers participated in the study. Seven subjects (all children) were excluded from the analysis because of a temporal lobe cyst (one subject), severe artifacts from dental bands (two subjects), history of ODD ADHD (one subject), and excessive head motion (three subjects), resulting in a total of 44 subjects in the analyses, ranging in age from 7 to 32 years old. Twentytwo of the volunteers were below the age of 18 (ten male; age range 7 17, mean age 11.8), and 22 were adults (nine male; age range 20 32, mean age 25.5). Throughout the rest of the article we will refer to the younger (< 18) group as children, although part of this group consisted of (young) adolescents. None of the 44 participants had a history of neurological or psychiatric disease. Participants all gave informed consent according to ethics regulations. Design and procedure Participants were scanned on an experiment previously shown to reliably localize body- and face-selective areas in visual cortex (Peelen & Downing, 2005a, 2005b). The experiment consisted of s blocks. Of these 21 blocks, five were fixation-only baseline conditions, occurring on blocks 1, 6, 11, 16, and 21. During the other 16 blocks, subjects were presented with pictures of faces, headless bodies, tools, or scenes. Forty full-color exemplars of each category were tested. Each image was presented for 300 ms, followed by a blank screen for 450 ms. Twice during each stimulus block the same image was presented twice in succession. Participants were instructed to detect these immediate repetitions and report them with a button press (one-back task). The position of the image was jittered slightly on alternate presentations, in order to disrupt attempts to perform the one-back task based on visual transients. This task was included to keep subjects attention focused on the stimuli throughout the experiment. It was not aimed at providing a behavioral measure of perceptual expertise for the different categories, as performance can rely on non-specific visual cues. The same experimental paradigm was used for all participants. Responses could not be recorded in one (adult) participant. Data acquisition Scanning was performed on a 3T Siemens Trio Tim MRI scanner at Geneva University Hospital, Center for Bio- Medical Imaging (CIBM). For functional imaging, a single shot EPI sequence was used (T2* weighted, gradient echo sequence). Scanning parameters were: TR = 2490 ms, TE = 30 ms, 36 off-axial slices, voxel dimensions: mm, 3.6-mm slice thickness (no gap). Anatomical images were acquired using a T1-weighted sequence. Scanning parameters were: TR TE: 2200 ms 3.45 ms; slice thickness = 1 mm; in-plane resolution: 1 1 mm. The same scanning protocol was used for all participants. Pre-processing Data were analyzed using the AFNI software package and MATLAB (The MathWorks, Natick, MA). All functional data were motion corrected, spatially smoothed with a Gaussian kernel (4 mm FWHM), and low-frequency drifts were removed with a temporal high-pass filter (cutoff of Hz). Functional data were co-registered with 3D anatomical T1 scans. The 3D anatomical scans were transformed into Talairach space, and the parameters from this transformation were subsequently applied to the co-registered functional data.

3 F18 Marius V. Peelen et al. An analysis was performed on the amount of head motion in children and adults. The maximum absolute value of each of the six motion correction parameters (three translation and three rotation parameters) was determined for each subject, reflecting the maximum amount of head motion throughout the experiment. On average, for both age groups all six maximum motion values were below 1 mm 1 deg, indicating very little overall head motion. The average amount of head motion (the average of the six maximum motion values) was similar for children (0.35) and adults (0.31; p =.55). ROI definition For each subject, we defined face- and body-selective ROIs by contrasting responses to Faces and Bodies with responses to Tools, at three statistical thresholds (p <.05, p <.01, p <.001; all uncorrected). Fusiform ROIs (rffa and rfba) were restricted to the part of ventral visual cortex corresponding to mid-fusiform gyrus (Talairach coordinates: 30 < x < 50; )55 < y < )30; )25 < z < )10; Peelen & Downing, 2005a). Analyses of the FFA and FBA were restricted to right hemisphere ROIs, because these regions were often weaker or non-existent in the left hemisphere, as previously observed (Kanwisher et al., 1997; Peelen & Downing, 2005a). EBA ROIs (leba and reba) were restricted to lateral occipitotemporal cortex (Talairach coordinates: 35 < x < 60; )80 < y < )55; )10 < z < 15; Peelen & Downing, 2005b). There was no overlap between reba and rfba. Calculation of overlap between ROIs The overlap between rffa and rfba was expressed by an index, which was calculated by dividing the number of overlapping voxels (i.e. voxels shared by rffa and rfba) by the number of voxels of the smaller of the two ROIs (which varies across subjects). This measure was preferred over other overlap measures (e.g. Kung, Peissig & Tarr, 2007) because it is less influenced by relative size differences between the ROIs, which were different for children compared to adults (see Results). An index of 1 indicates that the smaller of the two ROIs falls completely within the other ROI, whereas an index of 0 indicates no overlap between the two ROIs. subjects. Three rotation and three translation parameters from the motion correction were also added as regressors-of-no-interest. Results Behavioral results An Age (children, adults) Category (faces, bodies) ANOVA on percent correct scores yielded a significant main effect of Age (p <.01), with adults (98.3%) being overall more accurate than children (95.0%). There was no main effect of Category (p =.2), indicating that the task was equally difficult for faces (96.3%) and bodies (96.9%). Importantly, there was also no interaction between Age and Category (p >.6). Adults were also more accurate than children for the other categories, tools and scenes (average: 98.4% vs. 96.4%; p =.05), suggesting a general difference in task performance not specific to bodies or faces. An analysis of reaction times (RT) showed significantly (p <.01) faster responses to repetitions in face blocks (514 ms) than in body blocks (548 ms). There was no main effect of Age (p >.3), and no interaction between Age and Category (p >.1). Mean RTs for tools and scenes were 511 ms and 556 ms, respectively, and did not differ between children and adults (p >.5, for both). These results confirm that both children and adults paid attention to the presented images, and adequately detected image repetitions. Although the task was easier for adults than children, this difference was comparable for all conditions. Region of interest analysis Face- and body-selective ROIs (rffa, rfba, leba, reba) were defined in each subject individually, at three statistical thresholds (p <.05, p <.01, p <.001), to allow for a detailed assessment of the development of these regions. For each of the three thresholds, statistical analyses were performed on the size of the ROIs and their peak T-values. The fusiform ROIs (rffa, rfba) were analyzed together in Age ROI ANOVAs to directly compare the development of these overlapping regions. Whole-brain analyses Whole-brain, random-effects group-average analyses were conducted for the contrasts Faces > Tools and Bodies > Tools, separately for the two age groups: children (< 18 years old, N = 22), adults ( > 18 years old, N = 22). Contrasts were performed at uncorrected thresholds of p <.001. A general linear model was created with one predictor for each condition of interest. Regressors-of-no-interest were included to account for differences in the mean MR signal across scans and rffa rfba Talairach coordinates The average Talairach coordinates (x [SD], y [SD], z [SD]) of the most selective (peak) voxel of the rffa and rfba were nearly identical in both children (rffa: 38 [5], )43 [8], )14 [4]; rfba: 39 [6], )43 [6], )14 [4]) and adults (rffa: 37 [4], )47 [6], )14 [4]; rfba: 39 [4], )46 [5], )14 [5]), confirming that these regions occupy the same part of visual cortex. Within subjects, the Euclidian

4 Development of selectivity for faces and bodies F19 distance between the peaks of the rffa and rfba was, on average, 10.2 mm in children and 8.2 mm in adults, a difference that was not significant (p =.21). Overlap The amount of overlap (expressed by an index, see Methods) between the rffa and rfba was calculated for each subject for the three thresholds. There was no difference in the overlap index between children and adults at any of the thresholds (p >.4, for all thresholds). With ROIs defined at p <.05, the average overlap index was 0.75, indicating that 75% of the voxels of the smaller of the two ROIs were contained within the other ROI. At p <.01 and p <.001, the average overlap was 0.58 and 0.54, respectively. Size Figure 1a shows the size (in mm 3 ) of the rffa and rfba as a function of Age, separately for the three thresholds. At all thresholds, the size of the rffa gradually increased with Age, as indicated by a positive correlation with Age. This correlation was strongest for ROIs defined at p <.001 (r = 0.30; p <.05), and approached significance for the other two thresholds (r > 0.25 and p <.10, for both thresholds). By contrast, for the rfba no positive correlations with Age were found (r < 0 and p >.8, for all thresholds). Furthermore, the difference between the size of the rffa and the rfba correlated significantly with Age. That is, there was a significant correlation between [size_rffa size_rfba] and Age, for all three thresholds (r > 0.31 and p <.05, for all thresholds). These findings provide evidence for differential development of the two ROIs. Figure 1b shows the average size of the rffa and rfba for children (< 18 years old) and adults ( > 18 years old). For each of the three thresholds a separate Age (children, adult) ROI (rffa, rfba) ANOVA was performed on the size of the ROIs (in mm 3 ). Age interacted significantly with ROI for all three thresholds (p <.05, for all thresholds), providing further evidence for different developmental trajectories of the rffa and rfba. The rffa was larger in adults than children for all three thresholds (50% increase at p <.05, 68% at p <.01, 90% at p <.001), which was significant when ROIs were defined at p <.001 or p <.01 (p <.05, for both thresholds), and approached significance when ROIs were defined at p <.05 (p <.10). By contrast, the size of the rfba did not differ between children and adults at any of the three thresholds ()8% increase at p <.05, )8% at p <.01, )4% at p <.001; p >.6, for all (a) (b) Figure 1 (a) Size (in mm 3 ) of rffa and rfba in all 44 participants, at three different thresholds (p <.05 (top), p <.01 (middle), p <.001 (bottom)) as a function of age. (b) Mean size (in mm 3 ) of rffa and rfba, defined at three different thresholds, as a function of age group (children, adults). Error bars indicate SEM.

5 F20 Marius V. Peelen et al. tests). Furthermore, in children the rfba was significantly larger than the rffa (72% larger at p <.05, 77% at p <.01, 64% at p <.001; p <.05, for all tests), while in adults there were no significant differences between the size of the rfba and rffa (6% larger at p <.05, )4% at p <.01, )17% at p <.001; p >.3, for all tests). Confirming the correlation analysis, the consistent Age ROI interactions indicated that the rffa and rfba developed differently, with only the rffa being larger in adults than children. Furthermore, in children the body-selective rfba was approximately 70% larger than the face-selective rffa, whereas in adults the two regions were comparable in size. In the above analyses, significant differences were found between the sizes of the rffa in subjects younger versus older than 18 years old. In a follow-up analysis, we tested whether this developmental increase was gradual, by dividing the children group into two subgroups: a child group with ages between 7 and 11 years old (N = 11) and an adolescent group with ages between 12 and 17 years old (N = 11). For all three thresholds, rffa size was smallest for the child group (mean = 246 mm 3 ) and largest for the adult group (647 mm 3 ), with intermediate sizes for the adolescent group (547 mm 3 ), suggesting a gradual development of the rffa that continues during adolescence. At all three thresholds, the rffa in the child group was significantly smaller than the rffa in adults (p <.05, for all tests), but was significantly different (p <.05) from the adolescent group only when defined at a threshold of p <.05. No significant differences were found between the adolescent group and the adult group. Overall, these results suggest that the developmental increases in the size of the rffa were gradual, but reached significance only when comparing the younger children (or all children) with the adult group. Finally, we tested whether the age-related increase in the size of the rffa was related to the increase in performance on the one-back repetition detection task (adults were generally more accurate than children in detecting image repetitions; see above). No significant correlations were found between the size of the rffa and accuracy or RT for any of the three thresholds; all correlations were between )0.03 (p =.85; between RT and the size of the rffa defined at p <.05) and (p =.17; between accuracy and the size of the rffa defined at p <.001). This suggests that performance in our task, involving a simple detection of image repetition, did not depend on the size of the rffa. Peak T-value Figure 2a shows the peak T-value (the T-value of the most selective voxel) of the rffa and rfba as a function of Age. The peak T-value of the rffa gradually increased with Age, as indicated by a significant positive correlation with Age (r = 0.37, p <.05). For the rfba, a non-significant positive correlation with Age was found (r = 0.21, p =.16). Figure 2b shows the average peak T-value of the rffa and rfba for children and adults. An Age ROI ANOVA revealed a nearly significant interaction between Age and ROI (p =.07), a main effect of Age (p <.05) and a main effect of ROI (p <.05). There was no difference between the peak T-value of the rfba in children (T = 4.3) and adults (T = 4.9; p =.12), whereas the peak T-value of the rffa was significantly higher in adults (T = 5.9) than children (T = 4.4; p <.01). Furthermore, in adults, the average peak T-value of the rffa was significantly higher than the rfba (p <.05), whereas in children the peak T-values of rffa and rfba were comparable (p >.6). The results of the analyses on the peak T-value of rffa and rfba generally followed the results of the analyses on the size of the ROIs. Again, a significant development (i.e. increase in peak T-value) was found for rffa but not rfba. When dividing the children group into a child group (7 11 years old) and an adolescent group (12 17 years old), we found that the peak T-value of rffa was lowest for the child group (T = 4.0) and highest for the adult group (T = 5.9), with an intermediate value for the adolescent group (T = 4.8), again suggesting a gradual development of the rffa that continues during adolescence. The peak T-value of the rffa in the child group was significantly lower than the peak T-value in adults (p <.05), but was not significantly different from the adolescent group (p =.26). The difference between adolescents and adults did not reach significance (p =.12). Overall, these results suggest that the developmental increases in the peak T-value of the rffa were gradual, but reached significance only when comparing the younger children (or all children) with the adult group. Finally, we also tested whether the age-related increase in the selectivity (peak T-value) of the rffa was related to better performance on the one-back repetition detection task, as was done for the size of the rffa. No significant correlations were found between the selectivity of the rffa and accuracy (r = 0.17, p =.27) or RT (r = 0.09, p =.56). EBA Talairach coordinates The average Talairach coordinates (x [SD], y [SD], z [SD]) were: leba: children )46 [5], )72 [8], 6 [7]; adults )47 [6], )72 [6], 7 [4]; reba: children 47 [6], )70 [5], 1 [6]; adults 49 [7], )67 [7], 5 [6]. Size The size of the leba showed a non-significant positive correlation with Age, for all three thresholds (r > 0.14

6 Development of selectivity for faces and bodies F21 (a) (b) Figure 2 (a) Peak T-value of rffa and rfba as a function of age. (b) Mean peak T-value of rffa and rfba as a function of age group (children, adults). Error bars indicate SEM. and p >.1, for all thresholds). The size of the reba was negatively correlated with Age, which was significant for thresholds p <.05 (r = )0.34; p <.05) and p <.01 (r = )0.30; p <.05), but not for p <.001 (r = )0.23; p =.13). As the reba could still be reliably defined in all but one participant at the most stringent threshold used (p <.001; Table 1), we also tested for developmental effects in this region when defined at a threshold of p < At this threshold, reba could be defined in (100%) children and (91%) adults. The size of the reba was again negatively correlated with Age, but this correlation did not reach significance at this threshold (r = )0.18; p =.24). Table 2 gives the size of the left and right EBA for the different thresholds. No significant differences were found between children and adults in leba (p >.4, for all thresholds). The reba was significantly larger in children than adults when it was defined at p <.05 and p <.01 (p <.05, for both thresholds). No significant differences were found when reba was defined at p <.001 (p =.14) or p <.0001 (p =.28). Peak T-value The peak T-values of the left and right EBA showed nonsignificant positive correlations with Age (r > 0.07 and p >.1, for both ROIs). No significant differences were found between children and adults in the peak T-values of the left EBA (children: T = 4.5, adults: T = 5.1; p >.2) or right EBA (children: T = 6.2, adults: T = 6.6; p >.5). Summary of EBA results The developmental trajectory of the reba contrasted with that of the rffa, as the reba showed trends Table 1 Percentage of subjects with significant categoryselective activation in ROIs, at three statistical thresholds rffa rfba leba reba p <.05 Children Adults p <.01 Children Adults p <.001 Children Adults Table 2 Size of EBA (mm 3 ) for different age groups and statistical thresholds leba reba p <.05 p <.01 p <.001 p <.05 p <.01 p <.001 Children Adults Figure 3 Results of the whole-brain group analysis (at p <.001) for the children (left panels) and adults (right panels) groups. The top panels show axial slices (z = )18) with activation in right fusiform gyrus for the contrasts Faces > Tools (rffa) and Bodies > Tools (rfba). The bottom panels show coronal slices (y = )64) with activation in lateral occipito-temporal cortex for the contrast Bodies > Tools (EBA).

7 F22 Marius V. Peelen et al. towards size reduction as a function of age. The reba tended to be smaller (but no less selective) in adults compared to children, particularly when defined at more lenient thresholds. No significant developmental changes were found for leba in any of the analyses. Whole-brain group analysis Figure 3 shows the results of whole-brain group analyses (at p <.001, uncorrected) for the contrasts Faces > Tools and Bodies > Tools, separately for the children and adult groups. In both age groups, significant activation was observed in rffa for the contrast Faces > Tools, and in rfba, leba, and reba for the contrast Bodies > Tools. Similar to the individual ROI analyses, the group analysis revealed relatively weak face-selective fusiform responses in children (T = 5.3, volume = 135 mm 3 ; xyz: 42, )44, )19) compared to adults (T = 6.0, volume = 810 mm 3 ; xyz: 41, )42, )16), but strong body-selective fusiform responses in both children (T = 6.9, volume = 2106 mm 3 ; xyz: 42, )43, )16) and adults (T = 7.1, volume = 1809 mm 3 ; xyz: 41, )42, )15). Discussion The present study investigated the developmental trajectory of face- and body-selective visual areas. Despite their close overlap, significant differences were found in the development of rffa and rfba. With age, the rffa increased in both size and selectivity, replicating previous findings of developmental increases in this region (Aylward et al., 2005; Golarai et al., 2007; Scherf et al., 2007). By contrast, the overlapping rfba did not show such development, and was comparable in children and adults. Importantly, this difference in developmental trajectory cannot be explained by possible confounding factors typically associated with developmental neuroimaging studies, such as differences in the amount of head motion during the experiment, variations in cortical anatomy or normalization pre-processing, or differences in the shape and variance of the hemodynamic response function (Grill-Spector et al., 2008). In the present study, these factors would equally affect the measured development of rffa and rfba, particularly because these regions were located in the same part of visual cortex. Other potentially confounding factors, such as task difficulty and visual attention (Casey, Tottenham, Liston & Durston, 2005), are also unlikely to account for the present results. To control for attentional differences between conditions, subjects performed a one-back repetition detection task that required subjects to pay attention to all stimuli. Analyses on the behavioral performance showed that, although adults were generally more accurate than children, there was no interaction between object category (faces, bodies) and age group. Therefore, task difficulty cannot account for the differential developmental trajectory of face- and body-selective areas observed here. What could be the reason for the prolonged development of the rffa? One possibility is that it may be directly related to an increase in perceptual expertise and configural processing of faces, which have been shown to continue to develop through late childhood (Carey & Diamond, 1977; Chance, Turner & Goldstein, 1982; Diamond & Carey, 1977; Mondloch, Dobson, Parsons & Maurer, 2004; but see Crookes & McKone, in press). For example, previous studies have shown that perceptual expertise can substantially increase, and even create, category-selective responses in fusiform gyrus (Baker, Liu, Wald, Kwong, Benner & Kanwisher, 2007; de Heering & Rossion, 2008; Gauthier & Nelson, 2001; Gauthier, Skudlarski, Gore & Anderson, 2000; Gauthier, Tarr, Anderson, Skudlarski & Gore, 1999; van der Linden, Murre & van Turennout, 2008). It is therefore plausible that perceptual expertise is also important in the development of face and body selectivity in this region, perhaps in interaction with initial connection and architectural biases (Cohen Kadosh & Johnson, 2007; Johnson, 2001, 2005). In this framework, the present finding of an absence of a significant development of the rfba beyond age 7 predicts that perceptual expertise for bodies may similarly develop little beyond early childhood, a hypothesis that future studies could directly test. It should be noted that our task (detecting image repetitions) was not aimed at providing a behavioral measure of perceptual expertise, as it could be performed using relatively simple featural cues in the image and does not necessarily require a fine subordinate discrimination of exemplars as tested in studies of visual expertise (e.g. Gauthier et al., 2000). As such, our behavioral finding that adults were generally more accurate is unlikely to reflect differences in perceptual expertise with the stimuli. More likely, adults performed better than children because of more generic reasons, related for example to memory of the task instructions, sustained attention, etc. Indeed, subsequent analyses showed that the developmental increases in the rffa were uncorrelated with the superior performance of adults in detecting face repetitions. Our behavioral data further showed that both children and adults were faster at detecting repetitions during the face blocks compared to the body blocks (but not compared to the tool blocks). Again, these differences are likely to be related to nonspecific visual differences between the stimuli such as the variability of image features or contours from one image to the next. Clearly, future behavioral studies are needed to compare the development of face and body processing, using tasks that are specifically designed to test for perceptual expertise. There may be several reasons why face perception mechanisms continue to develop while body perception mechanisms do not. Most people will be exposed to

8 Development of selectivity for faces and bodies F23 many new faces throughout their lives, particularly during the first 20 years or so when major life changes (e.g. attending different schools) result in an enormous relative increase in the number of encountered faces. An increase in the size and selectivity of face-selective cortex may be necessary to discriminate and remember these new faces, including those of other races (Golby, Gabrieli, Chiao & Eberhardt, 2001). Even though individual differences in body shape may in principle be equally informative as differences in faces for determining the identity of other people, clothes typically obscure such information. Therefore, subtle differences in the body shape of newly met individuals are unlikely to be discriminated and remembered to the same extent as differences in faces. Second, increases in human height as a function of age may also play an important role in the development of body and face selectivity. Specifically, young children, when not looking up, will typically observe the bodies (and their movements) rather than the faces of adults and older children, whereas adults will mostly observe the faces of other people. Indeed, it is likely that children, more often than adults, use information (e.g. regarding identity, emotional state, or action intentions) provided by the bodies around them. Such differences may contribute to the differential development of body and face selectivity observed in the present study. Another finding of the present study was that the rfba was substantially (70%) larger than the rffa in children (but not adults). This suggests that the rfba develops earlier and or faster than the rffa, reaching adult size before the rffa does. Although it is possible that the rfba is a precursor of the rffa, in the present study most of the younger children showed significant selectivity for both faces and bodies, at least at more lenient thresholds (see Figure 1a), which suggests that these areas may instead develop in parallel. Indeed, behavioral studies have found that infants preferentially look at faces (Johnson, Dziurawiec, Ellis & Morton, 1991; Valenza, Simion, Macchi Cassia & Umilta, 1996), as well as body movements (Bertenthal, Proffitt & Kramer, 1987; Fox & McDaniel, 1982). Although these attentional biases are unlikely to be mediated only by cortical mechanisms, they may be important for the eventual development of category selectivity in the fusiform gyrus (Cohen Kadosh & Johnson, 2007; Johnson, 2001, 2005). The finding of stronger body-selective than faceselective fusiform responses in children is relevant for our understanding of the functional organization of this region. Previous studies investigating face and body selectivity in fusiform cortex in adult subjects found somewhat stronger selectivity for faces than for bodies (Peelen & Downing, 2005a), as also found here in the adult group. This raised the possibility that fusiform body selectivity may be explained in terms of indirect activation of face-selective mechanisms (e.g. through mental imagery or semantic association). The present finding that the fusiform gyrus in children was more strongly activated by bodies than faces indicates that body selectivity in this region is highly unlikely to reflect (indirect) activation of face-selective mechanisms, and provides further evidence for the existence of independent face- and body-selective neuronal populations in the fusiform gyrus (Peelen & Downing, 2005a; Schwarzlose et al., 2005). These populations appear to be interleaved at a relatively fine spatial scale below the standard resolution of fmri, but can be partly separated at high spatial resolution (Schwarzlose et al., 2005). A possible reason for the close proximity of faceand body-selective neurons in fusiform cortex may be to support the efficient integration of social cues conveyed by both faces and bodies, for example regarding the identities and emotions of other people (Peelen & Downing, 2007). Future developmental fmri studies at high spatial resolution are needed to test the development of face and body selectivity at a finer spatial scale. One possible outcome of such a study may be that in young children a large proportion of fusiform voxels respond uniquely to body stimuli, with a smaller proportion responding uniquely to face stimuli. With increasing age, the proportion of uniquely face- and body-selective voxels may become more balanced, with many of the previously body-selective voxels starting to respond selectively to both faces and bodies. Another finding of the present study was that right EBA tended to be smaller (but equally selective) in adults compared to children. This may reflect a developmental shift towards a more focal representation of bodies in reba (e.g. Passarotti et al., 2003). However, it should be noted that this effect was significant only when reba was defined at more lenient thresholds. Future studies are therefore necessary to confirm this finding. The different developmental trajectory of overlapping face- and body-selective activations has implications for the interpretation of developmental neuroimaging findings more generally. Previous studies have found that the development of certain cognitive functions (e.g. inhibitory control) is often linked to neuroanatomical maturation of particular brain regions (e.g. prefrontal cortex) underlying these functions (for a review, see Johnson, 2001). The strong link between structural development and functional development may suggest that multiple functions of a particular brain region would typically develop along similar trajectories. Importantly, our results indicate that this is not necessarily the case, at least not at the spatial resolution used here, by showing that multiple functions supported by the same anatomical structure can develop along different trajectories. Acknowledgements Data collection was supported by a Swiss National Fund for Research to SE (PP00B ), as well as grants

9 F24 Marius V. Peelen et al. from the Eagle Foundation, the NARSAD, and the Fondation Handicap Mental & SociØtØ. The authors would like to thank FranÅois Lazeyras, Frank Henry, and Pascal Challande, from the Center for Biomedical Imaging (CIBM) of the Geneva and Lausanne Universities, the EPFL and the Geneva and Lausanne University Hospitals, for help with data collection. References Aylward, E.H., Park, J.E., Field, K.M., Parsons, A.C., Richards, T.L., Cramer, S.C., & Meltzoff, A.N. (2005). Brain activation during face perception: evidence of a developmental change. Journal of Cognitive Neuroscience, 17 (2), Baker, C.I., Liu, J., Wald, L.L., Kwong, K.K., Benner, T., & Kanwisher, N. (2007). Visual word processing and experiential origins of functional selectivity in human extrastriate cortex. Proceedings of the National Academy of Sciences, USA, 104 (21), Bertenthal, B.I., Proffitt, D.R., & Kramer, S.J. (1987). Perception of biomechanical motions by infants: implementation of various processing constraints. Journal of Experimental Psychology: Human Perception and Performance, 13 (4), Carey, S., & Diamond, R. (1977). From piecemeal to configurational representation of faces. Science, 195 (4275), Casey, B.J., Tottenham, N., Liston, C., & Durston, S. (2005). Imaging the developing brain: what have we learned about cognitive development? Trends in Cognitive Sciences, 9 (3), Chance, J.E., Turner, A.L., & Goldstein, A.G. (1982). Development of differential recognition for own- and other-race faces. Journal of Psychology, 112, Cohen Kadosh, K., & Johnson, M.H. (2007). Developing a cortex specialized for face perception. Trends in Cognitive Sciences, 11 (9), Crookes, K., & McKone, E. (in press). Early maturity of face recognition: no childhood development of holistic processing, novel face encoding, or face space. Cognition. de Heering, A., & Rossion, B. (2008). Prolonged visual experience in adulthood modulates holistic face perception. PLoS ONE, 3 (5), e2317. Diamond, R., & Carey, S. (1977). Developmental changes in the representation of faces. Journal of Experimental Child Psychology, 23 (1), Diamond, R., & Carey, S. (1986). Why faces are and are not special: an effect of expertise. Journal of Experimental Psychology: General, 115 (2), Downing, P.E., Jiang, Y., Shuman, M., & Kanwisher, N. (2001). A cortical area selective for visual processing of the human body. Science, 293 (5539), Fox, R., & McDaniel, C. (1982). The perception of biological motion by human infants. Science, 218 (4571), Gathers, A.D., Bhatt, R., Corbly, C.R., Farley, A.B., & Joseph, J.E. (2004). Developmental shifts in cortical loci for face and object recognition. NeuroReport, 15 (10), Gauthier, I., & Nelson, C.A. (2001). The development of face expertise. Current Opinion in Neurobiology, 11 (2), Gauthier, I., Skudlarski, P., Gore, J.C., & Anderson, A.W. (2000). Expertise for cars and birds recruits brain areas involved in face recognition. Nature Neuroscience, 3 (2), Gauthier, I., Tarr, M.J., Anderson, A.W., Skudlarski, P., & Gore, J.C. (1999). Activation of the middle fusiform face area increases with expertise in recognizing novel objects. Nature Neuroscience, 2 (6), Golarai, G., Ghahremani, D.G., Whitfield-Gabrieli, S., Reiss, A., Eberhardt, J.L., Gabrieli, J.D., & Grill-Spector, K. (2007). Differential development of high-level visual cortex correlates with category-specific recognition memory. Nature Neuroscience, 10 (4), Golby, A.J., Gabrieli, J.D., Chiao, J.Y., & Eberhardt, J.L. (2001). Differential responses in the fusiform region to samerace and other-race faces. Nature Neuroscience, 4 (8), Grill-Spector, K., Golarai, G., & Gabrieli, J. (2008). Developmental neuroimaging of the human ventral visual cortex. Trends in Cognitive Sciences, 12 (4), Johnson, M.H. (2001). Functional brain development in humans. Nature Reviews Neuroscience, 2 (7), Johnson, M.H. (2005). Subcortical face processing. Nature Reviews Neuroscience, 6 (10), Johnson, M.H., Dziurawiec, S., Ellis, H., & Morton, J. (1991). Newborns preferential tracking of face-like stimuli and its subsequent decline. Cognition, 40 (1 2), Kanwisher, N., McDermott, J., & Chun, M.M. (1997). The fusiform face area: a module in human extrastriate cortex specialized for face perception. Journal of Neuroscience, 17 (11), Kozlowski, L.T., & Cutting, J.E. (1977). Recognizing the sex of a walker from a dynamic point-light display. Perception and Psychophysics, 21 (6), Kung, C.C., Peissig, J.J., & Tarr, M.J. (2007). Is regionof-interest overlap comparison a reliable measure of category specificity? Journal of Cognitive Neuroscience, 19 (12), Mondloch, C.J., Dobson, K.S., Parsons, J., & Maurer, D. (2004). Why 8-year-olds cannot tell the difference between Steve Martin and Paul Newman: factors contributing to the slow development of sensitivity to the spacing of facial features. Journal of Experimental Child Psychology, 89 (2), Mondloch, C.J., Geldart, S., Maurer, D., & Le Grand, R. (2003). Developmental changes in face processing skills. Journal of Experimental Child Psychology, 86 (1), Passarotti, A.M., Paul, B.M., Bussiere, J.R., Buxton, R.B., Wong, E.C., & Stiles, J. (2003). The development of face and location processing: an fmri study. Developmental Science, 6 (1), Peelen, M.V., & Downing, P.E. (2005a). Selectivity for the human body in the fusiform gyrus. Journal of Neurophysiology, 93 (1), Peelen, M.V., & Downing, P.E. (2005b). Within-subject reproducibility of category-specific visual activation with functional MRI. Human Brain Mapping, 25 (4), Peelen, M.V., & Downing, P.E. (2007). The neural basis of visual body perception. Nature Reviews Neuroscience, 8 (8),

10 Development of selectivity for faces and bodies F25 Peelen, M.V., Wiggett, A.J., & Downing, P.E. (2006). Patterns of fmri activity dissociate overlapping functional brain areas that respond to biological motion. Neuron, 49 (6), Reed, C.L., Stone, V.E., Bozova, S., & Tanaka, J. (2003). The body-inversion effect. Psychological Science, 14 (4), Scherf, K.S., Behrmann, M., Humphreys, K., & Luna, B. (2007). Visual category-selectivity for faces, places and objects emerges along different developmental trajectories. Developmental Science, 10 (4), F15 F30. Schwarzlose, R.F., Baker, C.I., & Kanwisher, N. (2005). Separate face and body selectivity on the fusiform gyrus. Journal of Neuroscience, 25 (47), Slaughter, V., Stone, V.E., & Reed, C. (2004). Perception of faces and bodies: similar or different? Current Directions in Psychological Science, 13 (6), Valenza, E., Simion, F., Macchi Cassia, V., & Umilta, C. (1996). Face preference at birth. Journal of Experimental Psychology: Human Perception and Performance, 22 (4), van der Linden, M., Murre, J.M., & van Turennout, M. (2008). Birds of a feather flock together: experience-driven formation of visual object categories in human ventral temporal cortex. PLoS ONE, 3 (12), e3995. Received: 21 March 2009 Accepted: 20 May 2009