A multimodal neural signature of face processing in autism within the fusiform gyrus

A multimodal neural signature of face processing in autism within the fusiform gyrus | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article A multimodal neural signature of face processing in autism within the fusiform gyrus Dorothea Floris, Alberto Llera, Mariam Zabihi, Emily Jones, Luke Mason, and 22 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3942971/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 02 Jan, 2025 Read the published version in Nature Mental Health → Version 1 posted You are reading this latest preprint version Abstract Differences in face processing are commonly reported in case/control studies of autism. Their neural correlates have been explored extensively across single neuroimaging modalities within key regions of the face processing network, such as the fusiform gyrus (FFG). Nonetheless, it is poorly understood how different variation(s) in brain anatomy and function combine to impact face processing and social functioning. Extracting the shared information across different modalities is essential to derive a more comprehensive understanding of the mechanisms underlying autism. Here, we leveraged a large multimodal sample to study the cross-modal signature of face processing within the FFG across four imaging modalities (structural MRI, resting-state fMRI [rs-fMRI], task-fMRI and EEG) in 204 individuals aged 7-30years comprising both autistic and non-autistic individuals (NAI). Combining two methodological innovations – normative modeling and linked independent component analysis – we integrated individual-level deviations across modalities to assess the efficacy of multimodal components in differentiating autistic from NAI and informing autism-associated social functioning. Autistic individuals differed significantly in a multimodal component, driven by bilateral rs-fMRI, bilateral structure, right task-fMRI, and left EEG loadings involving face-selective and retinotopic FFG regions. Multimodal components outperformed unimodal ones in differentiating autistic from NAI. Within the autism group, there was a significant multivariate association between multimodal components and a set of cognitive and clinical features associated with social functioning but not non-social features. These findings underscore the importance of elucidating individual-level, integrated neural associations of core social functioning in autism, offering potential for refined stratification and the identification of mechanistic and prognostic biomarkers. Biological sciences/Neuroscience/Diseases of the nervous system/Autism spectrum disorders Biological sciences/Neuroscience/Computational neuroscience Figures Figure 1 Figure 2 Figure 3 One sentence summary The fusiform gyrus is a central region differentially implicated in autistic compared to non-autistic individuals across inter-related structural and functional imaging modalities and category-selective regions. INTRODUCTION Autism is a lifelong neurodevelopmental condition with a prevalence of 1 in 36 children (1) . Social-communicative differences are among the most prominent features of autistic individuals (2) . Particularly, difficulties with processing social information and faces, such as perceiving and interpreting facial expressions of emotions and other mental states are thought to have a profound impact on their social functioning and daily living skills (3, 4) . While non-autistic individuals (NAI) appear to develop highly skilled strategies to discriminate facial cues at a very early age, autistic individuals have been reported to acquire less expertise with facial expression recognition (5) . This has, for example, been attributed to diminished social attention (6) and structural and functional differences in brain regions implicated in face processing (3, 5, 7) . While individual neuroimaging modalities have separately been used to characterize the neural correlates of face processing, multimodal studies of key regions associated with face processing remain scarce. Illuminating the rich multimodal information shared across different imaging modalities can unravel complex interactions and variations that may only be partially addressed by single modalities (8) . Specifically, elucidating cross-modal links with regards to face processing in autism will be crucial for understanding the biological mechanisms associated with core social difficulties and paving the way for the development of more personalised support. The fusiform gyrus (FFG) within the human ventral temporal cortex has been identified as a key neural region associated with higher-order processing of visual stimuli. The necessity for a thorough examination of the FFG in isolation is warranted by its detailed, functional heterogeneity exhibiting a fine-grained topographical organization with distinct category-selective patches (9, 10) that are differentially specialized for facial recognition (i.e., the fusiform face area [FFA]) (11) , body part discrimination (12) and object features recognition (13) . With regards to face processing, particularly the FFA has increased activation during face perception tasks in functional magnetic resonance imaging (fMRI) studies (14, 15) along with evidence from electroencephalography (EEG) studies showing an event-related potential of negative polarity that peaks at around 170ms when facial stimuli are presented (16, 17) . Furthermore, face processing is a lateralized cognitive function with right hemisphere dominance across these modalities (18, 19) . An exhaustive examination across different neural signatures of this fine-grained local and hemispheric heterogeneity of the FFG – beyond the FFA – has not been conducted in autistic individuals yet. This can offer valuable new insights in the light of reports of atypical functional specialisation in autism (20, 21) . Accumulating evidence suggests that there is atypical neural organization within the FFG in autistic individuals. Many studies show that the FFG is hypoactive during face processing fMRI tasks (22, 23) and, functionally atypically connected (24, 25) in autism. EEG studies show that the N170 latency is delayed in autistic individuals (26, 27) . Structurally, there are reports of volume increases in right FFG (28) , a reduction in mean FFG neuron density (29) and reversed leftward asymmetry (30) in autism. These atypical neural substrates are thought to be functionally relevant in autistic individuals. For example, they have been linked to differences in facial expression recognition (5) and face memory (31) , adaptive social functioning (27, 32) , and social symptom difficulties and severity (24, 31, 33, 34) . While these individual imaging modalities (i.e., structural MRI, task-fMRI, resting-state fMRI, EEG) converge to show atypical involvement of the FFG in face processing and related social functioning in autism, there is still little research into how these different neural substrates jointly inform fine-grained FFG organisation and social-communicative functioning in autism. Extracting common information from various modalities is crucial in gaining deeper insights into how brain structure and function reciprocally shape each other, and inform behaviour, cognition, and clinical conditions such as autism. To date, structure-function coupling has predominantly been addressed via univariate approaches where modalities are combined at the statistical or interpretation level (27, 35, 36) . However, only when employing multivariate multimodal approaches can we identify direct relationships between different neurobiological mechanisms and how they scale relative to each other. We can further penetrate across different biological spatial and temporal scales of variation leveraging the unique, complementary aspects covered by each individual imaging modality. Prior multimodal efforts are promising as they show that combining information from brain structure and function significantly increases accuracy in predictive frameworks (37–41) . Also, a recent study combining different neuroimaging measures of rs-fMRI, diffusion-weighted imaging and structural morphometry specifically showed that rs-connection topographies within the FFG were differentially implicated between autistic and NAI (42) . While such multimodal endeavours are still scarce in autism, this work specifically underscores the important role of the FFG in the neurobiology of autism. Still, the precise nature and a fine-grained topographical characterization of the multimodal neurobiological interactions within the FFG, and their relationship with the broader clinical phenotype related to social functioning in autism remain to be established. In the present study, our aim was to provide a more comprehensive understanding of the FFG in face processing in autism by elucidating the simultaneous involvement and multivariate interplay of different neural sources. Such analysis requires both large and deeply-phenotyped samples and given scarce availability, especially in clinical populations, this has previously limited its application. Hence, in this study, we leveraged the unique, large-scale and deeply-phenotyped EU-AIMS Longitudinal European Autism Project (43, 44) (LEAP) which is the largest European multi-centre initiative aimed at identifying biomarkers in autism. This dataset provides a rich set of different neuroimaging modalities, and cognitive, clinical measures as well as tasks related to face processing and social and non-social functioning. Differences in facial expression recognition in autistic individuals have been established in this dataset (5) . To further tap into their multimodal neural correlates, we combined two methodological innovations: (i) first, we employed normative modelling (45) on each imaging modality separately to derive individual-level deviations from a predicted age-related trajectory. Prior research shows that modelling cortical features as deviations from a normative neurodevelopmental trajectory provides more sensitive measure to map multimodal signatures in psychopathology (37) while also improving predictive performance (46) . (ii) Next, we conducted multi-modal fusion through Linked Independent Component Analysis (LICA) (47) across structural MRI, rs-fMRI, task-fMRI and EEG within the right and left FFG to simultaneously decompose the imaging data into underlying modes that characterise multi-modal signatures differentially in autistic and NAI. We further provided a fine-grained characterization of implicated regions shedding light on the topographic organisation within the FFG in autism. We hypothesized that multimodal components would be more sensitive to capturing subtle diagnostic effects cross-modally and would thus outperform unimodal components in discriminating autistic from NAI. Finally, we hypothesized that joint expression across modalities related to the FFG, and face processing would specifically inform social functioning in autism. Figure 1. Overview of the methodological approach . Features for each modality were extracted from the right and the left fusiform gyrus. These were: a) grey matter volume based on VBM for structural MRI; b) T-maps contrasting the faces condition to the shapes condition reflecting sensitivity to emotional faces from the Hariri paradigm for task-fMRI; c) seed-based (i.e., FFA) connectivity (SCA) for rs-fMRI; and d) the principal component of source reconstructed time series for EEG. Next, normative modelling was applied to each imaging modality using Bayesian Linear Regression. To model cross-subject individual-level variation, resulting Z-deviation maps per modality were statistically merged using linked independent component analysis resulting in measures of modality contributions and subject loadings. Next, we tested for group differences in ICs and group separability using either multi- or unimodal ICs. Finally, we computed multivariate associations (i.e., canonical correlation analysis) between subject loadings and clinical, cognitive measures related to either social-communicative or non-social features. RESULTS Sample For an overview of all methods see Fig. 1. The final sample of autistic (N = 99) and NAI (N = 105) did not differ significantly in sex ratio, age, measures of intellectual functioning, measures of structural image quality, number of EEG trials and head motion associated with task- and rs-fMRI (Table 1). Unimodal normative models First, unimodal normative models were estimated. Their accuracy was evaluated using the correlation between the true and the predicted voxel values (Rho), the explained variance, the mean standardized log-loss (MSLL) and standardized mean squared error (SMSE) (Fig. S1 ) and normative models per modality (Fig. S3). Evaluation metrics were largely within recommended ranges (48) and highly similar when modelling age linearly (Fig. S2). When testing for group differences in unimodal features, there were no significant differences in extreme Z-deviations between autistic and NAI for any of the eight features (Table S1 ). Linked independent component analysis Next, the Z-deviations (features) were merged using LICA (47) . Fifty independent components (IC) were derived across eight different brain feature maps per hemisphere (i.e., modalities) (Fig. S4). Overall, across these, the right hemisphere (51.7%) and the left hemisphere (48.3%) did not contribute differentially ( χ² =1.2, p = 0.72). Single modality contributions were as follows: EEG R (35.0%) > EEG L (33.2%) > rs-fMRI R (11.2%) > rs-fMRI L (9.6%) > task-fMRI R (3.5%) > task-fMRI L (3.4%) > structure L (2.1%) > structure R (2.1%). Fig. S5 shows the correlations between the 50-dimensional factorizations (y-axis) and alternative 40 (Fig.S5a) and 60 (Fig.S5b) dimensional factorizations. In line with previous reports (49) , most components were recovered with high accuracy independently of the order of the factorization. Group Differences Next, we compared the subject loadings of all (uni- and multimodal) ICs to test for differences between autistic and NAI. Among these, one multimodal IC (#44) showed a significant group difference with autistic individuals having higher contributions than NAI ( t = 3.5, p FDR =0.026) (Fig. 2b). There were no significant group differences in the remaining ICs (Table S2). The significant multimodal component was not differentially driven by the right (52.8%) or left hemisphere (47.2%) ( χ² =0.4, p = 0.51) and was associated with several functional modalities (rs-fMRI R [48.5%], rs-fMRI L [35.0%], EEG L [11.6%], task-fMRI R [3.3%]), and to a smaller extent with GM volume (structure R [1.0%], structure L [0.5%]). Figure 2c depicts the spatial and temporal patterns for each imaging modality within IC44. When characterizing these further in terms of their anatomical and functional overlaps with the Harvard-Oxford atlas (HOA) and the probabilistic functional atlas of human occipito-temporal visual cortex (9) (VIS), in the left hemisphere, autistic individuals showed more functional deviations than expected in rs-fMRI connectivity primarily in retinotopic regions of occipital FFG, while to a smaller extent also in lower-order face-selective regions (IOG) (Fig. 2d and 2i). In the right hemisphere, they showed linked increased deviations in rs-fMRI and structure primarily in higher-order face- (mFus, pFus) and bodies-selective (OTS) regions of temporal-occipital and occipital FFG (Fig. 2e and 2j). On the other hand, regions in the left hemisphere where autistic individuals showed linked decreased deviations compared to NAI, localized to both higher-order face-selective (mFus, pFus) and retinotopic regions of posterior, temporal-occipital and occipital FFG (Fig. 2f and 2k). In the right hemisphere, these were mostly in higher-order face face-selective regions (pFus) across rs-fMRI and task-fMRI and in retinotopic regions across structure in temporal-occipital FFG (Fig. 2g and 2l). Furthermore, autistic individuals showed more left EEG source activation than expected around 195-203ms and 417-426ms, whereas less source activation at 444–449ms than expected. For further details see Table S3. Multimodal components For further analyses, we focused on multimodal components only by excluding those which were primarily driven by one imaging modality, resulting in eleven multimodal ICs (Fig. 2a). Across these multimodal ICs, the right hemisphere (60.0%) contributed more than the left hemisphere (40.0%) ( χ² =7.2, p = 0.007). Single modality contributions across all multimodal ICs were as follows: EEG R (26.3%) > rs-fMRI L (19.7%) > EEG L (13.9%) > rs-fMRI R (12.1%) > task-fMRI L (9.8%) > task-fMRI R (9.6%) > structure R (4.4%) > structure L (4.2%). Autism classification Next, we applied a support vector machine (SVM) to compare unimodal and multimodal ICs’ efficacy in differentiating between the two diagnostic groups. Results showed that multimodal ICs performed significantly better at discriminating autistic from NAI (AUC unimodal = 0.48, AUC multimodal = 0.64, p < 0.001). This result was confirmed across a range of different multimodality thresholds (Fig. S6a) and was not influenced by varying amounts of features between multimodal and unimodal ICs (Fig. S6b). Figure 2 . Multimodal components and their spatial and temporal characterization . Among all ICs, eleven were considered multimodal (Fig. 2a), with a single modality contribution not more than 90%. IC44 showed a significant group difference with autistic individuals having higher contributions than NAI (Fig. 2b). Figure 2c shows spatial and temporal Z-maps thresholded at the 95th percentile of the different modalities associated with IC44. Positive values (yellow) depict positive loadings onto the IC where autistic have higher deviations than NAI; negative values (blue) depict negative loadings onto the IC where autistic individuals have lower deviations than NAI. Suprathreshold timepoints are depicted in red. Figure 2d-g depict the spatial overlap of suprathreshold voxels with a probabilistic functional atlas of the occipito-temporal cortex (i.e., VIS-atlas (9) ). Figure 2h depicts the VIS-atlas and its different category-selective subregions. Figure 2i-l show the spatial overlap of suprathreshold voxels with the structural Harvard-Oxford atlas and its four FFG subregions (i.e., anterior and posterior divisions of the temporal FFG, temporal occipital FFG and occipital FFG depicted in Fig. 2m). Figure 2d-e and 2i-j show the positive loadings (i.e., autism > NAI) and Fig. 2f-g and 2k-l the negative loadings (i.e., autism < NAI), whereas Fig. 2d/f and 2i/k depict the left and Fig. 2e/g and 2j/l the right hemisphere. Clinical, cognitive associations To test for brain-behaviour relations, we ran canonical correlation analysis (CCA). This revealed a significant multivariate association between the multimodal ICs and social-communicative features (i.e., ADOS-social affect, ADI-social, ADI-communication, Vineland Adaptive Behavior Scale (50) with Communication, Daily Living, Socialization subscales, the Reading the Mind in the Eyes test (51) , Hariri faces task (52) ) ( r = 0.65, p FDR =0.008; Fig. 3b). On the other hand, the relationship between the multimodal ICs and non-social features (i.e., ADOS-RRB, ADI-RRB, the Repetitive Behavior Scale (53) , the Short Sensory Profile (54) , the Systemizing Quotient (55–57) , Hariri shape matching condition) was not significant ( r = 0.49, p FDR =0.51; Fig. S7) pointing to specificity with social-related features of multimodal ICs. These associations remained stable when varying the multimodality threshold (Fig. S8). For the significant association, multimodal IC37 showed the largest contribution on the imaging side followed by IC38, IC44 and IC34 (Fig. 3a and 3c), whereas ADOS social-affect, RMET and Hariri face matching scores showed the largest contribution on the behavioural side (Fig. 3d). The ICs contributing most are depicted in Fig. 3c and Fig. S9-11. On average, the right (56.5%) and the left hemisphere (43.5%) did not contribute differentially to these four ICs ( χ² =2.9, p = 0.09) which were mostly driven by all functional modalities. Next, imaging patterns correlating with social-communication features were characterized in terms of their overlap with anatomical and functional overlaps with the HOA and VIS atlases (Fig. 3e and S12). Especially in higher-order face-selective regions (mFus and pFus) of posterior and temporal-occipital FFG, there were both linked increased deviations in bilateral rs-fMRI and task-fMRI and linked decreased deviations in bilateral structure and right rs-fMRI connectivity. At the same time, particularly in retinotopic regions of occipital FFG there was more bilateral GM volume along with less right task-activation than expected. There were more deviations in right EEG source activation at around 290ms, while left EEG did not reach significance. These joint imaging patterns were associated with more social difficulties as assessed by the ADOS, ADI and Vineland and more errors on the RMET, while also with greater accuracy on the Hariri faces task. For more details, see Tables S4-7. Figure 3 . The multivariate association (i.e., canonical correlation) was significant between the eleven multimodal ICs and the social-communicative features associated with autism. Figure 3a shows the loadings of each multimodal component contributing to the CCA mode, while Fig. 3d shows the loadings of each social-communicative feature contributing to the CCA mode; stars show the significant loadings. Figure 3b shows the canonical correlation scatterplot color-coded by the highest contributing clinical feature (ADOS social). The x-axis depicts the projected behavioural CCA variate and the y-axis the multimodal ICs CCA variates. Figure 3c shows the modality contributions of the four ICs that contribute significantly to the CCA. Figure 3e depicts the spatial and temporal patterns of each imaging modality that are significantly correlated with the social-communicative features. These are based on the significant correlation values between the Z-deviations of each imaging modality and the canonical imaging variate derived from the CCA. DISCUSSION In the present study, we aimed to characterize the multimodal neural signature of face processing in autism within the FFG, the core region of the face processing network. We identified several ICs that were differentially associated with the four modalities (structure, rs-fMRI, task-fMRI, and EEG), hemispheres, and functional subdivisions of the FFG. Autism-associated differences in FFG organization were more pronounced when penetrating across multiple than single modalities. Furthermore, a set of multimodal ICs was associated with core features related to social but not non-social functioning in autism. Taken together, these findings highlight the value of cross-modal analyses in characterizing a key structure in the multilevel neurobiology of autism and its implication in core cognitive and clinical features. Group differences Among all components, one multimodal component (i.e., IC44) showed a significant difference in subject loadings between autistic and NAI. Overall, the right and left hemispheres did not show differential contributions within this IC, and it was associated with all modalities fed into the analysis, with the functional modalities, especially rs-fMRI and EEG, contributing most (see Fig. 2). Particularly, the overlap with the VIS-atlas highlighted that face-selective and retinotopic regions of the FFG were most different between autistic and NAI. More specifically, in the right hemisphere, higher-order face-selective regions exhibited less task activation and FFA-connectivity than expected, primarily in occipital FFG areas (Fig. 2g and 2l). At the same time autistic individuals showed increased deviations in FFA-connectivity primarily in temporal-occipital FFG along with increased GM volume deviations in higher-order face-selective FFG regions (Fig. 2e and 2j). This strong right-hemisphere involvement of regions associated with FFA across several modalities is in line with reports of increased FFA volume (28) and decreased FFA task-activation (58, 59) and FFA-connectivity (24, 60) in autism. Similarly, temporally, autistic individuals showed more increased left deviations around 195ms potentially indicative of the consistently reported finding of a slower N170 in autistic individuals (26) . This has specifically also been shown and extensively characterised in the current sample (27) . Together these patterns converge to point towards autism-associated differences in face-selective areas of the FFG, both at the structural, functional, and temporal levels. Although these results align with earlier unimodal discoveries, previously it was uncertain whether disparate signals would be separate or coalesce to a joint multimodal expression. In this context, we provide evidence supporting the interconnected nature of distinct signals within a single, unified framework. In the left hemisphere, IC44-related increased deviations in EEG source activation at around 420ms may indicate reductions in the face-N400 which has been associated with familiar face recognition and semantic information (61) . While in NAI face processing becomes the most highly developed visual skill, in autistic individuals faces may convey greater novelty and thus decreased familiarity. Furthermore, occipital, retinotopic areas of the left FFG were most implicated as shown by increased functional connectivity deviations between the FFA and retinotopic and lower-order face-selective areas of the FFG in autistic individuals (Fig. 2d and 2i). This was echoed by less GM volume than expected in left retinotopic areas of FFG in autistic individuals (Fig. 2f and 2k). Retinotopic, early visual areas act as the first stage in a hierarchical network of face processing in which lower-level feature-based components are processed before more complex features in higher-order face-selective regions (62) . Neural deviations in early visual areas as seen here are in line with reports of autistic individuals showing differences in sensory processing at early perceptual stages and have been described at the cognitive level as weak central coherence (63) . Accordingly, studies show that autistic individuals exhibit a different strategy in processing facial and visual stimuli with a stronger focus on featural, local aspects at the expense of holistic, global information (64) . Similarly, fMRI studies converge to show greater feature-based perceptual strategies in autistic individuals who primarily tend to recruit object-related regions (65, 66) when viewing facial stimuli. Taken together, this suggests that differences we discovered in the left hemisphere point primarily to low-level, bottom-up processing differences, whereas in the right hemisphere they may indicate higher-level atypicalities in the FFA, with a differential involvement across the different structural and functional modalities. Clinical, cognitive associations Multimodal ICs showed a significant association with a set of clinical and cognitive features associated with social functioning in autism (Fig. 3). Group-differential IC44 was also among the significantly contributing ICs to this associations. Components loading significantly onto the CCA were mostly driven by functional modalities. Right EEG source activation deviations were at around 280-300ms, potentially indicative of the N250r generated in the FFG (66) and associated with repetition of familiar facial stimuli (68) . The amplitude of the N250r has been shown to decrease with increasing working memory (WM) load (69) . This would translate into increased deviations as seen in autistic individuals here and may imply differences in degrees of WM resources allocated to the processing of facial stimuli which in turn have a larger novelty character in autistic individuals requiring more attentional effort. With regards to the other modalities, increased deviations particularly in higher-order face-selective regions across brain function (task-fMRI and rs-fMRI) while also in lower-order early visual regions across brain structure were associated with more autistic features, such as more social difficulties as assessed by ADOS, and lower social sensitivity as assessed by the RMET. Previous unimodal studies showed that the delayed latency of the N170 predicts change in social adaptive behaviour in autistic individuals (27) (i.e., EEG), autistic individuals with low performance on facial emotion recognition have reduced bilateral FFG activation (i.e., task-fMRI) (5) and atypical FFA-connectivity is associated with increased social symptom severity in autism (i.e., rs-fMRI) (24) . Here, we extend unimodal results to a multivariate association across a range of social-communicative features that are related to cross-modal signatures within the FFG. Here, we provide evidence for an interrelated biological basis of core social functioning in autism and that appropriately modelled shared variance across different modalities increases sensitivity to clinical-cognitive features associated with autism (70) . Remarkably, at the same time, there was no association with a set of non-social features, such as repetitive behaviours or sensory processing, pointing to specificity of these multimodal ICs with regards to social functioning. Summary and implications Taken together, the multimodal neural signature within the FFG in autism presents differentially across hemispheres, modalities, and topography. Specifically, the picture emerges that (i) the functional modalities contribute more than the structural modalities and (ii) retinotopic, occipital regions are more implicated in the left hemisphere and higher-order regions more implicated in the right hemisphere within the FFG when it comes to group differences; but they do not contribute differentially with regards to social functioning. (i) Concurrent neural activity and functional co-expressions (task-, rs-fMRI, EEG) were strongly tied to social features observed in autistic individuals at present (such as current performance and ADOS assessment). On the other hand, more stable structural aspects of the brain established over time and historical symptoms reported through the ADI and Vineland – which provide insights into past behaviors – had a comparatively smaller impact on the observed association. These results highlight the dynamic nature of the relationship between neural activity and social functioning in autism and underscore the importance of considering the temporal dimension when investigating the neural correlates of social functioning in autism. Putative future neuroscientifically informed interventions targeting social features may thus benefit from a focus on concurrent neural functioning. (ii) Topographically, the FFG is known to exhibit an anterior to posterior gradient with more posterior regions related to lower-order, early visual processing, and more anterior regions related to higher-order processing (71) . Here, we see the involvement of both retinotopic and higher-order cognitive, particularly face-sensitive patches pointing to differences in both bottom-up perceptual processes and top-down cognitive information processing in face processing in autism which can amount to a difference in the face processing strategy employed (e.g., more feature-based). These different processing levels are not differentially implicated across hemispheres in the processing of social information in autism suggesting that the distinctive face processing strategy in autism transcends right hemisphere dominance of face processing. On the other hand, hemispheric differences are more apparent in the group-differential IC. Teasing apart hemispheric contributions is particularly important in the light of reports of atypical patterns of brain asymmetry in autistic individuals (21, 34, 72) . More extreme deviations from a normative model have for example been reported in right temporal-occipital fusiform cortex asymmetry in autistic females (21) , along with more left-lateralized volume in posterior temporal FFG in autistic individuals (30, 34) . Subsequent research should delve further into these more nuanced insights revealed by cross-modal analyses pointing to left-lateralized low-level and right-lateralized high-level differences between autistic and NAI. Strengths and limitations Integrating data from different modalities has the advantage of being biologically more informative and comprehensive in characterizing a complex, heterogenous condition like autism. Accordingly, when comparing unimodal deviations in each imaging modality, as well as comparing predominantly unimodal ICs between autistic and NAI, there were no significant group differences, despite employing a more sensitive individual-level measures derived from normative modelling. Also, multimodal features significantly outperformed unimodal features in differentiating autistic form NAI. These results together confirm our hypothesis and previous reports (70, 73) that appropriately modelling cross-modal variance increases sensitivity to detecting subtle effects that may otherwise be missed. Thus, integrating different structural and functional brain measures is the most promising and powerful method to achieve significant advances in our understanding of system-level atypicalities in autism and provides the basis for elucidating mechanisms through which interventions can most efficiently improve clinically relevant functioning (70) . Furthermore, we combine different innovative methods. LICA is particularly powerful when modelling modalities that are different in their numbers of features, spatial correlations, intensity distributions and units. This is, because LICA optimally weighs the contributions of each modality by the correction for the number of effective degrees of freedom and the use of automatic relevance determination priors on components (8, 47, 70) . Also, by combining normative modelling with LICA, we employ a previously validated approach that has been shown to increase sensitivity in detecting cross-modal effects in clinical populations (37) . At the same time, it needs to be pointed out that face processing involves an extended neural network across the whole brain including other structures such as the amygdala, superior temporal sulcus and occipital and frontal cortex (37, 66, 74, 75) . It may thus seem too simplistic to reduce face processing to a single brain region. Still, the FFG has been claimed the core node of a distributed face processing network, as also substantiated by FFG lesion studies (76, 77) , and its fine-grained functional heterogeneity warrants careful examination in isolation. Also, implementing cross-modal analyses presents with additional challenges, such as obtaining sufficiently large sample sizes with all participants having available data across all imaging modalities. Here, from a sample of over 600 individuals in the EU-AIMS LEAP dataset, we were able to conduct analysis in just over 200 individuals who had available imaging data across the four different modalities. Whole brain analyses based on multivariate techniques will ultimately require larger sample sizes. Conclusion Integrating information from multiple imaging modalities allows us to gain a more holistic and robust understanding of the complex neural processes underlying core clinical and cognitive features associated with autism. Present results suggest that the FFG is a central region differentially implicated across different neural signals and category-selective regions in autistic and NAI and that this informs cross-modally the mechanisms associated with core social functioning in autism. Eventually, elucidating more precise, integrated and individual-level neural associations of core cognitive and clinical features, will pave the way for further work identifying stratification, mechanistic and prognostic biomarkers, and the development of more personalised support, thereby eventually improving the quality of lives of autistic individuals. Materials and Methods Sample characterization Participants were part of the EU-AIMS/AIMS-2-TRIALS LEAP cohort (43, 44) . They underwent comprehensive clinical, cognitive and MRI assessment at one of six collaborating sites. All autistic participants had an existing clinical diagnosis of autism which was confirmed using the combined information of gold-standard diagnostic instruments, the Autism Diagnostic Interview-Revised (78) (ADI-R) and the Autism Diagnostic Observation Schedule (79) (ADOS). The study was approved by the respective research ethics committees at each site (IRAS, UK). Informed written consent was obtained from all participants, or—for minors or those unable to give informed consent—from a parent or legal guardian. For further details see Supplemental Information (SI) and our earlier papers (43, 44) . The final sample has both complete imaging data across four different imaging modalities and phenotypic information available consisting of 99 autistic and 105 NAI between 7–30 years (Table 1). Clinical and cognitive measures We split available autism-associated measures into two sets of feature sets based on the construct they measure 1) social-communicative features comprising measures of difficulties with social communication and daily living skills (i.e., ADOS-social affect, ADI-communication, ADI-social, Vineland Adaptive Behavior Scale (50) with Communication, Daily Living, Socialization subscales), emotional face matching performance (i.e., Hariri faces task (52) ), and social sensitivity to complex emotions (i.e., Reading the Mind in the Eyes test (51) [RMET]) and 2) non-social features comprising restricted, repetitive behaviours (RRBs) (i.e., ADOS-RRB, ADI-RRB, the Repetitive Behavior Scale (53) [RBS-R]), systemizing (i.e., the Systemizing Quotient (55–57) [SQ]), shape matching performance (i.e., Hariri shapes task, as the control condition to the Hariri emotional faces task) and sensory processing atypicalities (i.e., Short Sensory Profile (54) [SSP]) (see SI and Table S8). To tackle missing clinical data and to not further reduce sample size, we used imputed clinical data (80) , as in previous work with this dataset (41, 81) . Region of interest: fusiform gyrus All analyses were restricted to the right and left FFG based on the HOA (FMRIB, Oxford, UK) adjusted to have 100% coverage across all individuals for each imaging modality (see SI). Imaging modalities For MRI and EEG data acquisition parameters and detailed preprocessing steps per modality, see SI and Table S9-10. In summary, following were the features for subsequent normative modeling: a) structure: VBM-derived, voxel-wise GM volumes; b) rs-fMRI: seed-based correlation between the FFA and the remaining FFG; c) task-fMRI: T-contrast maps reflecting sensitivity to emotional faces; d) EEG: the principal component of source reconstructed activation obtained across different cortical parcels. Normative modelling Normative modelling is an emerging statistical technique that allows parsing heterogeneity by charting variation in brain-behaviour mappings relative to a normative range and provides statistical inference at the level of the individual (82) . The term ‘normative’ should not be seen as incompatible with the neurodiversity framework as it simply refers to statistical norms such as growth charts that vary by demographics such as age and gender. Here, we trained normative models using Bayesian Linear Regression (BLR) (83) for each brain imaging modality within the right and left FFG ROI independently using age, sex and scanning site as covariates. A B-spline basis expansion of the covariate vector was used to model non-linear effects of age. Normative models were derived in an unbiased manner across the entire sample under 10-fold cross-validation (37, 45, 84) . To estimate voxel-wise/time-point-wise deviations for each modality in each individual, we derived normative probability maps (NPM) that quantify the deviation from the normative model summarized in Z -scores. These subject-specific Z-score images provide a statistical estimate of how much each individual’s recorded value differs from the predicted value at each voxel/time-point. The accuracy of the normative model was evaluated using the correlation between the recorded and the predicted voxel values (Rho), MSLL, SMSE, and the EV (Fig. S1 ) as well as based on the forward models (Fig. S3). Furthermore, we compared model performance when modelling age linearly (without a B-spline basis expansion; Fig. S2). To assess whether autistic and NAI differed in their extreme deviations based on unimodal features, thresholded Z-scores (Z>|2.6| (21, 37, 85) , corresponding to the 99.5th percentile) were compared between the two groups using a two-sample t-test (see SI). Code is available at https://github.com/amarquand/PCNtoolkit . Linked Independent Component Analysis In order to gain more comprehensive insights into cross-modal signatures of face processing, we merged the different individual-level deviations from all imaging modalities (GM volume, FFA-connectivity, T-maps contrasting the faces to the shapes condition, and the principal components of source reconstructed time series) using LICA (47) (see SI). This is a Bayesian extension of the single modality ICA model which provides an automatic and simultaneous decomposition of the brain features into independent components (ICs) that characterize the inter-subject brain variability. These multiple decompositions share a mixing matrix (i.e., subject course) across individual feature factorizations that reflect the subject contributions to each IC. Here, LICA was used to merge the unthresholded Z-deviation maps derived from normative modeling across the four different imaging modalities within the right and left FFG ROIs. Each measure per hemisphere was treated separately resulting in eight input maps (i.e., modalities). Hemispheres were modelled separately given known brain asymmetric differences in autism (21, 34) and to study the hemispheric contributions and model the different noise characteristics individually. We estimated 50 independent components based on our sample size and following recommendations described in earlier papers (37, 41, 49, 86) (i.e., sample size ~ N / 4). To evaluate the robustness of our selected model order (N = 50), we re-ran LICA using different dimensional factorizations of subject loadings (N = 40 and N = 60) and computed correlations among them. Group Differences The subject loadings of all ICs were compared between autistic and NAI using a two-sample t-test. Multiple comparisons were corrected for using the False Discovery Rate (FDR) (87) . ICs showing significant group differences were further characterized by plotting each contributing modality’s spatial map and temporal profile (Z-thresholded at the 95th percentile). To further characterize the most implicated regions within the FFG per modality, we computed the overlap between supra-threshold voxels and a structural (i.e., the Harvard-Oxford atlas, which covers the entire FFG) and a functional atlas (i.e., a probabilistic functional atlas of the occipito-temporal cortex (9) which covers category-specific FFG patches) (see SI). Multimodal components Next, given the current work’s focus on multimodal neural sources, we tested the hypothesis that multimodal components performed superior to unimodal components in differentiating autistic from NAI. For this, we calculated a multimodal index (MMI) per IC to quantify the multimodal nature of modalities in each IC (86) (see SI). The MMI ranges from 0 (equating to 100% unimodal contribution) to 1 (equating to equal contributions from all modalities). Multimodal components were defined as each single imaging modality (i.e., regardless of hemisphere) not having more than a 90% contribution to each component and an MMI below 0.1 (Fig. S13). Components below this threshold were treated as unimodal. Autism classification Next, we implemented two support vector machine (SVM) classifiers with a linear kernel – one using unimodal and one using multimodal components as features to test for the added value of multimodal features. The SVM was trained and evaluated using 10-fold cross-validation. Class-weighting was used to account for group size imbalance. The area under the receiver operating characteristic curve (AUC) was used as the performance metric to assess the classifier's discrimination ability. To test for significant differences in AUC between multimodal and unimodal components, we generated a null distribution of AUC differences by shuffling the cross-validated scores 10,000 times and re-evaluating the classifier performance and computed the likelihood of observing the observed AUC difference under the null hypothesis. To test for robustness of results, we re-ran analyses across different thresholds resulting in slightly varying degrees of multimodality ranging between 85–99% of single modality contributions. Given that each threshold resulted in a different number of unimodal vs. multimodal components, we further checked whether results remained stable when forcing uni- and multimodal components to have the same number of features (see SI). Clinical-cognitive associations To test for the clinical relevance of multimodal ICs, we ran canonical correlation analyses (CCA) (88) modelling the multivariate relationship between multimodal ICs and cognitive, clinical features related to either a) social-communicative features related to social functioning and face processing in autism or b) non-social features associated with autism. The statistical significance of the CCA modes was assessed by a complete permutation inference algorithm (89) , where both brain and behaviour data were permuted separately across all participants with 10,000 iterations (see SI). To visualize the spatial and temporal patterns of each imaging modality associated with each clinical cognitive measure, we computed the correlations between the original imaging data (i.e., the Z-deviation maps) and the canonical imaging variate (V) derived from the CCA (90) . Significance of correlation maps was assessed with 1000 permutations and significant clusters/timepoints were next visualized and characterized in terms of their functional and anatomical overlap with the VIS-atlas (9) (Fig. 2h) and the HOA atlas (Fig. 2m) (see SI). To assess robustness of CCA results, we set a range of multimodal thresholds between 85–99% and selected ICs with modality contributions exceeding this threshold. We then re-ran the CCA for each threshold to assess stability of results across varying degrees of multimodality. Declarations Acknowledgements: We thank all participants and their families for participating in the studies that contribute to the datasets used in this research. We also gratefully acknowledge the contributions of all members of the EU-AIMS/AIMS-2-TRIALS LEAP group: Jumana Ahmad, Sara Ambrosino, Bonnie Auyeung, Sarah Baumeister, Sven Bölte, Carsten Bours, Michael Brammer, Daniel Brandeis, Claudia Brogna, Yvette de Bruijn, Bhismadev Chakrabarti, Ineke Cornelissen, Daisy Crawley, Guillaume Dumas, Jessica Faulkner, Vincent Frouin, Pilar Garcés, David Goyard, Lindsay Ham, Hannah Hayward, Joerg Hipp, Mark H. Johnson, Emily J.H. Jones, Xavier Liogier D’ardhuy, David J. Lythgoe, René Mandl, Luke Mason, Andreas Meyer-Lindenberg, Nico Mueller, Bethany Oakley, Laurence O’Dwyer, Bob Oranje, Gahan Pandina, Antonio M. Persico, Barbara Ruggeri, Amber Ruigrok, Jessica Sabet, Roberto Sacco, Antonia San José Cáceres, Emily Simonoff, Will Spooren, Roberto Toro, Heike Tost, Jack Waldman, Steve C.R. Williams, Caroline Wooldridge, and Marcel P. Zwiers. Funding : This project has received funding from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No 115300 (for EU-AIMS) and No 777394 (for AIMS-2-TRIALS). This Joint Undertaking receives support from the European Union's Horizon 2020 research and innovation programme and EFPIA and AUTISM SPEAKS, Autistica, SFARI. Any views expressed are those of the author(s) and not necessarily those of the funders (IHI-JU2). This work was also supported by the Netherlands Organization for Scientific Research through grants (Grant No. 864.12.003 [to CFB]; from the FP7 (Grant Nos. 602805) (AGGRESSOTYPE) (to JKB), 603016 (MATRICS), and 278948 (TACTICS); and from the European Community’s Horizon 2020 Programme (H2020/2014-2020) (Grant Nos. 643051 [MiND] and 642996 (BRAINVIEW). This work received funding from the Wellcome Trust UK Strategic Award (Award No. 098369/Z/12/Z) and from the National Institute for Health Research Maudsley Biomedical Research Centre (to DM). DLF is supported by the UZH postdoc grant and funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 101025785. EJHJ and RH received funding from SFARI GAIINS (grant number 10039678). SB-C is funded by the Autism Research Trust, the Wellcome Trust, the Templeton World Charitable Foundation and by the NIHR Biomedical Research Centre in Cambridge, during the period of this work. BHV is supported by the Swiss National Science Foundation [10001C_197480]. Author contributions : D.L.F.: study design, data preprocessing and analysis, wrote the paper. A.L., T.M., B.H.V. and N.F.: supervision and advise on linked independent component analysis and interpretation. M.Z., N.E.H. and C.E.: supervision and advise on normative modeling analysis and interpretation. C.M. and S.B.: fMRI processing. E.J.H.J., L.M. and R.H.: EEG data preprocessing. C.M.P: analysis direction and interpretation. T.C.: clinical/design of study/interpretation. F.D.A., S.D., T.B., R.J.H., S.B.-C., T.B., E.L., D.G.M.M., J.K.B. and C.F.B.: secured funding and supervised the study overall and analysis direction and interpretation. N.L.: study design and supervision of overall analytic strategy. All authors revised the manuscript for intellectual content. Competing interests : JKB has been a consultant to, advisory board member of, and a speaker for Takeda/Shire, Medice, Roche, and Servier. He is not an employee of any of these companies and not a stock shareholder of any of these companies. He has no other financial or material support, including expert testimony, patents, or royalties. CFB is director and shareholder in SBGneuro Ltd. TC has received consultancy from Roche and Servier and received book royalties from Guildford Press and Sage. TB served in an advisory or consultancy role for ADHS digital, Infectopharm, Lundbeck, Medice, Neurim Pharmaceuticals, Oberberg GmbH, Roche, and Takeda. He received conference support or speaker’s fee by Medice and Takeda. He received royalities from Hogrefe, Kohlhammer, CIP Medien, Oxford University Press; the present work is unrelated to these relationships. The other authors report no biomedical financial interests or potential conflicts of interest. Data and materials availability : LEAP data are shared through an application process codesigned with autistic people that preserves security and privacy—contact the corresponding author for further details and application forms. Scripts used to implement the experimental task are covered by a material transfer agreement, which can be obtained through the corresponding author. All the data associated with this study are present in the paper or the Supplementary Materials. References M. J. Maenner, K. A. Shaw, A. V. Bakian, D. A. Bilder, M. S. Durkin, A. Esler, S. M. Furnier, L. Hallas, J. Hall-Lande, A. Hudson, M. M. Hughes, M. Patrick, K. Pierce, J. N. Poynter, A. Salinas, J. Shenouda, A. Vehorn, Z. Warren, J. N. Constantino, M. DiRienzo, R. T. Fitzgerald, A. Grzybowski, M. H. Spivey, S. Pettygrove, W. Zahorodny, A. Ali, J. G. Andrews, T. Baroud, J. Gutierrez, A. Hewitt, L.-C. Lee, M. Lopez, K. C. Mancilla, D. McArthur, Y. D. Schwenk, A. Washington, S. Williams, M. E. Cogswell, Prevalence and Characteristics of Autism Spectrum Disorder Among Children Aged 8 Years - Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2018. MMWR Surveill Summ 70 , 1–16 (2021). S. M. Kanne, A. J. Gerber, L. M. Quirmbach, S. S. Sparrow, D. V. Cicchetti, C. A. Saulnier, The role of adaptive behavior in autism spectrum disorders: Implications for functional outcome. Journal of Autism and Developmental Disorders 41 (2011), doi:10.1007/s10803-010-1126-4. G. Dawson, S. J. Webb, J. McPartland, Understanding the nature of face processing impairment in autism: insights from behavioral and electrophysiological studies. Dev Neuropsychol 27 , 403–424 (2005). N. J. Sasson, The development of face processing in autism. J Autism Dev Disord 36 , 381–394 (2006). H. Meyer-Lindenberg, C. Moessnang, B. Oakley, J. Ahmad, L. Mason, E. J. H. Jones, H. L. Hayward, J. Cooke, D. Crawley, R. Holt, J. Tillmann, T. Charman, S. Baron-Cohen, T. Banaschewski, C. Beckmann, H. Tost, A. Meyer-Lindenberg, J. K. Buitelaar, D. G. Murphy, M. J. Brammer, E. Loth, Facial expression recognition is linked to clinical and neurofunctional differences in autism. Molecular Autism 13 , 43 (2022). C. Chevallier, P. Huguet, F. Happé, N. George, L. Conty, Salient Social Cues are Prioritized in Autism Spectrum Disorders Despite Overall Decrease in Social Attention. J Autism Dev Disord 43 , 1642–1651 (2013). R. T. Schultz, Developmental deficits in social perception in autism: the role of the amygdala and fusiform face area. International Journal of Developmental Neuroscience 23 , 125–141 (2005). J. Sui, T. Adali, Q. Yu, J. Chen, V. D. Calhoun, A review of multivariate methods for multimodal fusion of brain imaging data. Journal of Neuroscience Methods 204 , 68–81 (2012). M. Rosenke, R. van Hoof, J. van den Hurk, K. Grill-Spector, R. Goebel, A Probabilistic Functional Atlas of Human Occipito-Temporal Visual Cortex. Cerebral Cortex 31 , 603–619 (2021). W. Zhang, J. Wang, L. Fan, Y. Zhang, P. T. Fox, S. B. Eickhoff, C. Yu, T. Jiang, Functional organization of the fusiform gyrus revealed with connectivity profiles. Hum Brain Mapp 37 , 3003–3016 (2016). I. Gauthier, M. J. Tarr, J. Moylan, P. Skudlarski, J. C. Gore, A. W. Anderson, The fusiform “face area” is part of a network that processes faces at the individual level. Journal of Cognitive Neuroscience (2000), doi:10.1162/089892900562165. J. C. Taylor, A. J. Wiggett, P. E. Downing, Functional MRI Analysis of Body and Body Part Representations in the Extrastriate and Fusiform Body Areas. Journal of Neurophysiology 98 , 1626–1633 (2007). X. Wang, Z. Han, Y. He, A. Caramazza, L. Song, Y. Bi, Where color rests: Spontaneous brain activity of bilateral fusiform and lingual regions predicts object color knowledge performance. NeuroImage 76 , 252–263 (2013). G. McCarthy, A. Puce, J. C. Gore, T. Allison, Face-Specific Processing in the Human Fusiform Gyrus. Journal of Cognitive Neuroscience 9 , 605–610 (1997). N. Kanwisher, G. Yovel, The fusiform face area: a cortical region specialized for the perception of faces. Philosophical Transactions of the Royal Society B: Biological Sciences 361 , 2109–2128 (2006). S. Bentin, T. Allison, A. Puce, E. Perez, G. McCarthy, Electrophysiological Studies of Face Perception in Humans. Journal of Cognitive Neuroscience 8 , 551–565 (1996). K. Bötzel, S. Schulze, S. R. Stodieck, Scalp topography and analysis of intracranial sources of face-evoked potentials. Exp Brain Res 104 , 135–143 (1995). B. Rossion, L. Dricot, A. Devolder, J.-M. Bodart, M. Crommelinck, B. de Gelder, R. Zoontjes, Hemispheric Asymmetries for Whole-Based and Part-Based Face Processing in the Human Fusiform Gyrus. Journal of Cognitive Neuroscience 12 , 793–802 (2000). G. Yovel, A. Tambini, T. Brandman, The asymmetry of the fusiform face area is a stable individual characteristic that underlies the left-visual-field superiority for faces. Neuropsychologia 46 , 3061–3068 (2008). R. C. Cardinale, P. Shih, I. Fishman, L. M. Ford, R. A. Müller, Pervasive rightward asymmetry shifts of functional networks in autism spectrum disorder. JAMA Psychiatry (2013), doi:10.1001/jamapsychiatry.2013.382. D. L. Floris, T. Wolfers, M. Zabihi, N. E. Holz, M. P. Zwiers, T. Charman, J. Tillmann, C. Ecker, F. Dell’Acqua, T. Banaschewski, C. Moessnang, S. Baron-Cohen, R. Holt, S. Durston, E. Loth, D. G. M. Murphy, A. Marquand, J. K. Buitelaar, C. F. Beckmann, Atypical Brain Asymmetry in Autism—A Candidate for Clinically Meaningful Stratification. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 6 , 802–812 (2021). T. Nickl-Jockschat, C. Rottschy, J. Thommes, F. Schneider, A. R. Laird, P. T. Fox, S. B. Eickhoff, Neural networks related to dysfunctional face processing in autism spectrum disorder. Brain Struct Funct 220 , 2355–2371 (2015). K. Pierce, F. Haist, F. Sedaghat, E. Courchesne, The brain response to personally familiar faces in autism: findings of fusiform activity and beyond. Brain 127 , 2703–2716 (2004). N. M. Kleinhans, T. Richards, L. Sterling, K. C. Stegbauer, R. Mahurin, L. C. Johnson, J. Greenson, G. Dawson, E. Aylward, Abnormal functional connectivity in autism spectrum disorders during face processing. Brain 131 , 1000–1012 (2008). H. Koshino, R. K. Kana, T. A. Keller, V. L. Cherkassky, N. J. Minshew, M. A. Just, fMRI Investigation of Working Memory for Faces in Autism: Visual Coding and Underconnectivity with Frontal Areas. Cerebral Cortex 18 , 289–300 (2008). E. Kang, C. M. Keifer, E. J. Levy, J. H. Foss-Feig, J. C. McPartland, M. D. Lerner, Atypicality of the N170 Event-Related Potential in Autism Spectrum Disorder: A Meta-analysis. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 3 , 657–666 (2018). L. Mason, C. Moessnang, C. Chatham, L. Ham, J. Tillmann, G. Dumas, C. Ellis, C. S. Leblond, F. Cliquet, T. Bourgeron, C. Beckmann, T. Charman, B. Oakley, T. Banaschewski, A. Meyer-Lindenberg, S. Baron-Cohen, S. Bölte, J. K. Buitelaar, S. Durston, E. Loth, B. Oranje, A. Persico, F. Dell’Acqua, C. Ecker, M. H. Johnson, D. Murphy, E. J. H. Jones, Stratifying the autistic phenotype using electrophysiological indices of social perception. Science Translational Medicine 14 , eabf8987 (2022). G. D. Waiter, J. H. G. Williams, A. D. Murray, A. Gilchrist, D. I. Perrett, A. Whiten, A voxel-based investigation of brain structure in male adolescents with autistic spectrum disorder. NeuroImage 22 (2004), doi:10.1016/j.neuroimage.2004.02.029. I. A. J. Van Kooten, S. J. M. C. Palmen, P. Von Cappeln, H. W. M. Steinbusch, H. Korr, H. Heinsen, P. R. Hof, H. Van Engeland, C. Schmitz, Neurons in the fusiform gyrus are fewer and smaller in autism. Brain 131 (2008), doi:10.1093/brain/awn033. M. R. Herbert, G. J. Harris, K. T. Adrien, D. a Ziegler, N. Makris, D. N. Kennedy, N. T. Lange, C. F. Chabris, A. Bakardjiev, J. Hodgson, M. Takeoka, H. Tager-Flusberg, V. S. Caviness, Abnormal asymmetry in language association cortex in autism. Annals of neurology 52 , 588–96 (2002). E. Neuhaus, E. J. H. Jones, K. Barnes, L. Sterling, A. Estes, J. Munson, G. Dawson, S. J. Webb, The Relationship Between Early Neural Responses to Emotional Faces at Age 3 and Later Autism and Anxiety Symptoms in Adolescents with Autism. J Autism Dev Disord 46 , 2450–2463 (2016). S. J. Webb, E. J. H. Jones, K. Merkle, K. Venema, J. Greenson, M. Murias, G. Dawson, Developmental change in the ERP responses to familiar faces in toddlers with autism spectrum disorders versus typical development. Child Dev 82 , 1868–1886 (2011). C. Gao, S. Conte, J. E. Richards, W. Xie, T. Hanayik, The neural sources of N170: Understanding timing of activation in face-selective areas. Psychophysiology 56 , e13336 (2019). C. C. Dougherty, D. W. Evans, G. J. Katuwal, A. M. Michael, Asymmetry of fusiform structure in autism spectrum disorder: Trajectory and association with symptom severity. Molecular Autism 7 (2016), doi:10.1186/s13229-016-0089-5. T. Iidaka, A. Matsumoto, K. Haneda, T. Okada, N. Sadato, Hemodynamic and electrophysiological relationship involved in human face processing: Evidence from a combined fMRI–ERP study. Brain and Cognition 60 , 176–186 (2006). A. Puce, A. Syngeniotis, J. C. Thompson, D. F. Abbott, K. J. Wheaton, U. Castiello, The human temporal lobe integrates facial form and motion: evidence from fMRI and ERP studies. NeuroImage 19 , 861–869 (2003). N. E. Holz, D. L. Floris, A. Llera, P. M. Aggensteiner, S. M. Kia, T. Wolfers, S. Baumeister, B. Böttinger, J. C. Glennon, P. J. Hoekstra, A. Dietrich, M. C. Saam, U. M. E. Schulze, D. J. Lythgoe, S. C. R. Williams, P. Santosh, M. Rosa-Justicia, N. Bargallo, J. Castro-Fornieles, C. Arango, M. J. Penzol, S. Walitza, A. Meyer-Lindenberg, M. Zwiers, B. Franke, J. Buitelaar, J. Naaijen, D. Brandeis, C. Beckmann, T. Banaschewski, A. F. Marquand, Age-related brain deviations and aggression. Psychol Med , 1–10 (2022). T. T. Brown, J. M. Kuperman, Y. Chung, M. Erhart, C. McCabe, D. J. Hagler, V. K. Venkatraman, N. Akshoomoff, D. G. Amaral, C. S. Bloss, B. J. Casey, L. Chang, T. M. Ernst, J. A. Frazier, J. R. Gruen, W. E. Kaufmann, T. Kenet, D. N. Kennedy, S. S. Murray, E. R. Sowell, T. L. Jernigan, A. M. Dale, Neuroanatomical assessment of biological maturity. Current Biology 22 (2012), doi:10.1016/j.cub.2012.07.002. G. Erus, H. Battapady, T. D. Satterthwaite, H. Hakonarson, R. E. Gur, C. Davatzikos, R. C. Gur, Imaging patterns of brain development and their relationship to cognition. Cerebral Cortex 25 (2015), doi:10.1093/cercor/bht425. F. Liem, G. Varoquaux, J. Kynast, F. Beyer, S. Kharabian Masouleh, J. M. Huntenburg, L. Lampe, M. Rahim, A. Abraham, R. C. Craddock, S. Riedel-Heller, T. Luck, M. Loeffler, M. L. Schroeter, A. V. Witte, A. Villringer, D. S. Margulies, Predicting brain-age from multimodal imaging data captures cognitive impairment. NeuroImage 148 (2017), doi:10.1016/j.neuroimage.2016.11.005. T. Mei, N. J. Forde, D. L. Floris, F. Dell’Acqua, R. Stones, I. Ilioska, S. Durston, C. Moessnang, T. Banaschewski, R. J. Holt, S. Baron-Cohen, A. Rausch, E. Loth, B. Oakley, T. Charman, C. Ecker, D. G. M. Murphy, EU-AIMS LEAP group, C. F. Beckmann, A. Llera, J. K. Buitelaar, Autism Is Associated With Interindividual Variations of Gray and White Matter Morphology. Biol Psychiatry Cogn Neurosci Neuroimaging , S2451-9022(22)00212–9 (2022). L. M. Oblong, A. Llera, T. Mei, K. Haak, C. Isakoglou, D. L. Floris, S. Durston, C. Moessnang, T. Banaschewski, S. Baron-Cohen, E. Loth, F. Dell’Acqua, T. Charman, D. G. M. Murphy, C. Ecker, J. K. Buitelaar, C. F. Beckmann, The EU-AIMS LEAP Group, Linking functional and structural brain organisation with behaviour in autism: a multimodal EU-AIMS Longitudinal European Autism Project (LEAP) study. Molecular Autism 14 , 32 (2023). T. Charman, E. Loth, J. Tillmann, D. Crawley, C. Wooldridge, D. Goyard, J. Ahmad, B. Auyeung, S. Ambrosino, T. Banaschewski, S. Baron-Cohen, S. Baumeister, C. Beckmann, S. Bölte, T. Bourgeron, C. Bours, M. Brammer, D. Brandeis, C. Brogna, Y. De Bruijn, B. Chakrabarti, I. Cornelissen, F. D. Acqua, G. Dumas, S. Durston, C. Ecker, J. Faulkner, V. Frouin, P. Garcés, L. Ham, H. Hayward, J. Hipp, R. J. Holt, J. Isaksson, M. H. Johnson, E. J. H. Jones, P. Kundu, M. C. Lai, X. L. D’Ardhuy, M. V. Lombardo, D. J. Lythgoe, R. Mandl, L. Mason, A. Meyer-Lindenberg, C. Moessnang, N. Mueller, L. O’Dwyer, M. Oldehinkel, B. Oranje, G. Pandina, A. M. Persico, B. Ruggeri, A. N. V. Ruigrok, J. Sabet, R. Sacco, A. S. J. Cáceres, E. Simonoff, R. Toro, H. Tost, J. Waldman, S. C. R. Williams, M. P. Zwiers, W. Spooren, D. G. M. Murphy, J. K. Buitelaar, The EU-AIMS Longitudinal European Autism Project (LEAP): Clinical characterisation. Molecular Autism 8 (2017), doi:10.1186/s13229-017-0145-9. E. Loth, T. Charman, L. Mason, J. Tillmann, E. J. H. Jones, C. Wooldridge, J. Ahmad, B. Auyeung, C. Brogna, S. Ambrosino, T. Banaschewski, S. Baron-Cohen, S. Baumeister, C. Beckmann, M. Brammer, D. Brandeis, S. Bölte, T. Bourgeron, C. Bours, Y. De Bruijn, B. Chakrabarti, D. Crawley, I. Cornelissen, F. D. Acqua, G. Dumas, S. Durston, C. Ecker, J. Faulkner, V. Frouin, P. Garces, D. Goyard, H. Hayward, L. M. Ham, J. Hipp, R. J. Holt, M. H. Johnson, J. Isaksson, P. Kundu, M. C. Lai, X. L. D’Ardhuy, M. V. Lombardo, D. J. Lythgoe, R. Mandl, A. Meyer-Lindenberg, C. Moessnang, N. Mueller, L. O’Dwyer, M. Oldehinkel, B. Oranje, G. Pandina, A. M. Persico, A. N. V. Ruigrok, B. Ruggeri, J. Sabet, R. Sacco, A. S. J. Cáceres, E. Simonoff, R. Toro, H. Tost, J. Waldman, S. C. R. Williams, M. P. Zwiers, W. Spooren, D. G. M. Murphy, J. K. Buitelaar, The EU-AIMS Longitudinal European Autism Project (LEAP): Design and methodologies to identify and validate stratification biomarkers for autism spectrum disorders. Molecular Autism 8 (2017), doi:10.1186/s13229-017-0146-8. A. F. Marquand, I. Rezek, J. Buitelaar, C. F. Beckmann, Understanding Heterogeneity in Clinical Cohorts Using Normative Models: Beyond Case-Control Studies. Biological Psychiatry 80 , 552–561 (2016). L. Parkes, T. M. Moore, M. E. Calkins, P. A. Cook, M. Cieslak, D. R. Roalf, D. H. Wolf, R. C. Gur, R. E. Gur, T. D. Satterthwaite, D. S. Bassett, Transdiagnostic dimensions of psychopathology explain individuals’ unique deviations from normative neurodevelopment in brain structure. Transl Psychiatry 11 , 232 (2021). A. R. Groves, C. F. Beckmann, S. M. Smith, M. W. Woolrich, Linked independent component analysis for multimodal data fusion. NeuroImage 54 , 2198–2217 (2011). S. Rutherford, P. Barkema, I. F. Tso, C. Sripada, C. F. Beckmann, H. G. Ruhe, A. F. Marquand, C. I. Baker, T. Constable, O. Esteban, Eds. Evidence for embracing normative modeling. eLife 12 , e85082 (2023). A. Llera, T. Wolfers, P. Mulders, C. F. Beckmann, Inter-individual differences in human brain structure and morphology link to variation in demographics and behavior. eLife 8 (2019), doi:10.7554/eLife.44443. S. S. Sparrow, D. V. Cicchetti, D. A. Balla, in Major psychological assessment instruments , (2005). S. Baron-Cohen, S. Wheelwright, J. Hill, Y. Raste, I. Plumb, The “Reading the Mind in the Eyes” Test revised version: A study with normal adults, and adults with Asperger syndrome or high-functioning autism. Journal of Child Psychology and Psychiatry and Allied Disciplines 42 (2001), doi:10.1017/S0021963001006643. A. R. Hariri, A. Tessitore, V. S. Mattay, F. Fera, D. R. Weinberger, The amygdala response to emotional stimuli: A comparison of faces and scenes. NeuroImage 17 (2002), doi:10.1006/nimg.2002.1179. J. W. Bodfish, F. J. Symons, D. E. Parker, M. H. Lewis, Repetitive Behavior Scale–Revised. PsycTESTS® (2000). S. D. Tomchek, W. Dunn, Sensory processing in children with and without autism: a comparative study using the short sensory profile. American Journal of Occupational Therapy 61 (2007), doi:10.5014/ajot.61.2.190. B. Auyeung, S. Wheelwright, C. Allison, M. Atkinson, N. Samarawickrema, S. Baron-Cohen, The children’s empathy quotient and systemizing quotient: Sex differences in typical development and in autism spectrum conditions. Journal of Autism and Developmental Disorders 39 (2009), doi:10.1007/s10803-009-0772-x. B. Auyeung, C. Allison, S. Wheelwright, S. Baron-Cohen, Brief report: Development of the adolescent empathy and systemizing quotients. Journal of Autism and Developmental Disorders 42 (2012), doi:10.1007/s10803-012-1454-7. S. Baron-Cohen, J. Richler, D. Bisarya, N. Gurunathan, S. Wheelwright, The systemizing quotient: An investigation of adults with Asperger syndrome or high-functioning autism, and normal sex differences. Philosophical Transactions of the Royal Society B: Biological Sciences 358 (2003), doi:10.1098/rstb.2002.1206. H. D. Critchley, E. M. Daly, E. T. Bullmore, S. C. Williams, T. Van Amelsvoort, D. M. Robertson, A. Rowe, M. Phillips, G. McAlonan, P. Howlin, D. G. Murphy, The functional neuroanatomy of social behaviour: changes in cerebral blood flow when people with autistic disorder process facial expressions. Brain 123 ( Pt 11) , 2203–2212 (2000). B. A. Corbett, V. Carmean, S. Ravizza, C. Wendelken, M. L. Henry, C. Carter, S. M. Rivera, A functional and structural study of emotion and face processing in children with autism. Psychiatry Research: Neuroimaging 173 , 196–205 (2009). A. C. Lynn, A. Padmanabhan, D. Simmonds, W. Foran, M. N. Hallquist, B. Luna, K. O’Hearn, Functional connectivity differences in autism during face and car recognition: underconnectivity and atypical age-related changes. Developmental Science 21 , e12508 (2018). S. Bentin, L. Y. Deouell, Structural Encoding and Identification in Face Processing: Erp Evidence for Separate Mechanisms. Cognitive Neuropsychology 17 , 35–55 (2000). D. Pitcher, V. Walsh, B. Duchaine, The role of the occipital face area in the cortical face perception network. Exp Brain Res 209 , 481–493 (2011). F. Happé, U. Frith, The weak coherence account: detail-focused cognitive style in autism spectrum disorders. Journal of autism and developmental disorders 36 , 5–25 (2006). T. L. Watson, Implications of holistic face processing in autism and schizophrenia. Front Psychol 4 , 414 (2013). R. T. Schultz, I. Gauthier, A. Klin, R. K. Fulbright, A. W. Anderson, F. Volkmar, P. Skudlarski, C. Lacadie, D. J. Cohen, J. C. Gore, Abnormal Ventral Temporal Cortical Activity During Face Discrimination Among Individuals With Autism and Asperger Syndrome. Archives of General Psychiatry 57 , 331–340 (2000). K. Pierce, R. A. Müller, J. Ambrose, G. Allen, E. Courchesne, Face processing occurs outside the fusiform “face area” in autism: Evidence from functional MRI. Brain 124 (2001), doi:10.1093/brain/124.10.2059. S. R. Schweinberger, E. C. Pickering, I. Jentzsch, A. M. Burton, J. M. Kaufmann, Event-related brain potential evidence for a response of inferior temporal cortex to familiar face repetitions. Cognitive Brain Research 14 , 398–409 (2002). S. R. Schweinberger, V. Huddy, A. M. Burton, N250r: a face-selective brain response to stimulus repetitions. NeuroReport 15 , 1501 (2004). H. M. Morgan, C. Klein, S. G. Boehm, K. L. Shapiro, D. E. J. Linden, Working memory load for faces modulates P300, N170, and N250r. J Cogn Neurosci 20 , 989–1002 (2008). A. R. Groves, S. M. Smith, A. M. Fjell, C. K. Tamnes, K. B. Walhovd, G. Douaud, M. W. Woolrich, L. T. Westlye, Benefits of multi-modal fusion analysis on a large-scale dataset: Life-span patterns of inter-subject variability in cortical morphometry and white matter microstructure. NeuroImage 63 (2012), doi:10.1016/j.neuroimage.2012.06.038. J. Caspers, K. Zilles, K. Amunts, A. R. Laird, P. T. Fox, S. B. Eickhoff, Functional characterization and differential coactivation patterns of two cytoarchitectonic visual areas on the human posterior fusiform gyrus. Human Brain Mapping 35 , 2754–2767 (2014). D. L. Floris, M.-C. Lai, T. Auer, M. V. Lombardo, C. Ecker, B. Chakrabarti, S. J. Wheelwright, E. T. Bullmore, D. G. M. Murphy, S. Baron‐Cohen, J. Suckling, Atypically rightward cerebral asymmetry in male adults with autism stratifies individuals with and without language delay. Human Brain Mapping 37 , 230–253 (2016). V. D. Calhoun, J. Sui, Multimodal Fusion of Brain Imaging Data: A Key to Finding the Missing Link(s) in Complex Mental Illness. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 1 , 230–244 (2016). Y. Aoki, S. Cortese, M. Tansella, Neural bases of atypical emotional face processing in autism: A meta-analysis of fMRI studies. World Journal of Biological Psychiatry 16 (2015), doi:10.3109/15622975.2014.957719. J. V. Haxby, E. A. Hoffman, M. I. Gobbini, The distributed human neural system for face perception. Trends Cogn Sci 4 , 223–233 (2000). B. Rossion, R. Caldara, M. Seghier, A. M. Schuller, F. Lazeyras, E. Mayer, A network of occipito-temporal face-sensitive areas besides the right middle fusiform gyrus is necessary for normal face processing. Brain (2003), doi:10.1093/brain/awg241. J. J. S. Barton, D. Z. Press, J. P. Keenan, M. O’Connor, Lesions of the fusiform face area impair perception of facial configuration in prosopagnosia. Neurology 58 , 71–78 (2002). C. Lord, M. Rutter, A. Le Couteur, Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. Journal of autism and developmental disorders 24 , 659–685 (1994). C. Lord, S. Risi, L. Lambrecht, E. H. Cook, B. L. Leventhal, P. C. DiLavore, A. Pickles, M. Rutter, The autism diagnostic observation schedule-generic: a standard measure of social and communication deficits associated with the spectrum of autism. Journal of autism and developmental disorders 30 , 205–223 (2000). A. Llera, M. Brammer, B. Oakley, J. Tillmann, M. Zabihi, J. S. Amelink, T. Mei, T. Charman, C. Ecker, F. Dell’Acqua, T. Banaschewski, C. Moessnang, S. Baron-Cohen, R. Holt, S. Durston, D. Murphy, E. Loth, J. K. Buitelaar, D. L. Floris, C. F. Beckmann, Evaluation of data imputation strategies in complex, deeply-phenotyped data sets: the case of the EU-AIMS Longitudinal European Autism Project. BMC Medical Research Methodology 22 , 229 (2022). T. Looden, D. L. Floris, A. Llera, R. J. Chauvin, T. Charman, T. Banaschewski, D. Murphy, Andre. F. Marquand, J. K. Buitelaar, C. F. Beckmann, the AIMS-2-TRIALS group, Patterns of connectome variability in autism across five functional activation tasks: findings from the LEAP project. Molecular Autism 13 , 53 (2022). A. F. Marquand, S. M. Kia, M. Zabihi, T. Wolfers, J. K. Buitelaar, C. F. Beckmann, Conceptualizing mental disorders as deviations from normative functioning. Molecular Psychiatry 24 , 1415–1424 (2019). C. J. Fraza, R. Dinga, C. F. Beckmann, A. F. Marquand, Warped Bayesian linear regression for normative modelling of big data. NeuroImage 245 , 118715 (2021). N. E. Holz, M. Zabihi, S. M. Kia, M. Monninger, P.-M. Aggensteiner, S. Siehl, D. L. Floris, A. L. W. Bokde, S. Desrivières, H. Flor, A. Grigis, H. Garavan, P. Gowland, A. Heinz, R. Brühl, J.-L. Martinot, M.-L. P. Martinot, D. P. Orfanos, T. Paus, L. Poustka, J. H. Fröhner, M. N. Smolka, N. Vaidya, H. Walter, R. Whelan, G. Schumann, A. Meyer-Lindenberg, D. Brandeis, J. K. Buitelaar, F. Nees, C. Beckmann, T. Banaschewski, A. F. Marquand, A stable and replicable neural signature of lifespan adversity in the adult brain. Nat Neurosci , 1–10 (2023). T. Wolfers, N. T. Doan, T. Kaufmann, D. Alnæs, T. Moberget, I. Agartz, J. K. Buitelaar, T. Ueland, I. Melle, B. Franke, O. A. Andreassen, C. F. Beckmann, L. T. Westlye, A. F. Marquand, Mapping the Heterogeneous Phenotype of Schizophrenia and Bipolar Disorder Using Normative Models. JAMA Psychiatry 75 , 1146–1155 (2018). W. Francx, A. Llera, M. Mennes, M. P. Zwiers, S. V. Faraone, J. Oosterlaan, D. Heslenfeld, P. J. Hoekstra, C. A. Hartman, B. Franke, J. K. Buitelaar, C. F. Beckmann, Integrated analysis of gray and white matter alterations in attention-deficit/hyperactivity disorder. NeuroImage: Clinical 11 , 357–367 (2016). Y. Benjamini, Y. Hochberg, On the adaptive control of the false discovery rate in multiple testing with independent statistics. Journal of Educational and Behavioral Statistics (2000), doi:10.3102/10769986025001060. H. Hotelling, Relations Between Two Sets of Variates. Biometrika 28 , 321–377 (1936). A. M. Winkler, O. Renaud, S. M. Smith, T. E. Nichols, Permutation inference for canonical correlation analysis. NeuroImage 220 (2020), doi:10.1016/j.neuroimage.2020.117065. G. Ball, P. Aljabar, P. Nongena, N. Kennea, N. Gonzalez-Cinca, S. Falconer, A. T. M. Chew, N. Harper, J. Wurie, M. A. Rutherford, S. J. Counsell, A. D. Edwards, Multimodal image analysis of clinical influences on preterm brain development. Annals of Neurology 82 , 233–246 (2017). Table Table 1 Demographic, clinical and imaging-related information of the sample. Abbreviations: NAI = non-autistic individuals; FIQ = full-scale IQ; VIQ = verbal IQ; PIQ = performance IQ; ADI = Autism Diagnostic Interview; ADOS = Autism Diagnosis Observation Schedule; QC = quality control; FD = framewise displacement. Variable Autism NAI post hoc N 99 105 sex 71M : 28F 71M : 34F χ2 = 0.23, p = 0.63 autism = NAI mean std range mean std range age 18.62 5.32 7.58-30.0 18.3 4.87 10.37–30.9 t = 0.44, p = 066 autism = NAI FIQ 107 14.3 63–148 106 12.3 77–142 t = 0.08, p = 0.94 autism = NAI VIQ 106 15.62 51–160 107 14.1 74–142 t = 0.09, p = 0.93 autism = NAI PIQ 107 16.52 57–145 106 14.5 70–147 t = 0.51, p = 0.61 autism = NAI ADI social 14.54 6.66 1.0–28.0 ADI communication 11.68 5.66 0.0–26.0 ADI RRB 3.63 2.41 0.0–12.0 ADOS CSS 5.06 2.67 1.0–10.0 ADOS SA CSS 5.79 2.7 1.0–10.0 ADOS RRB CSS 4.5 2.49 1.0–10.0 median std range median std range QC structure 2.15 0.11 2.04-3.0 2.13 0.19 1.96–3.42 W = 4926, p = 0.52 autism = NAI mean FD task fMRI 0.09 0.06 0.03–0.33 0.08 0.07 0.03–0.39 W = 4833, p = 0.39 autism = NAI mean FD rsfMRI 0.06 0.05 0.03–0.27 0.06 0.06 0.02–0.4 W = 4915, p = 0.5 autism = NAI number of EEG trials 90.6 24.8 29–148 93.1 29.7 27–149 t = 0.65, p = 0.52 autism = NAI Additional Declarations Yes there is potential Competing Interest. JKB has been a consultant to, advisory board member of, and a speaker for Takeda/Shire, Medice, Roche, and Servier. He is not an employee of any of these companies and not a stock shareholder of any of these companies. He has no other financial or material support, including expert testimony, patents, or royalties. CFB is director and shareholder in SBGneuro Ltd. TC has received consultancy from Roche and Servier and received book royalties from Guildford Press and Sage. TB served in an advisory or consultancy role for ADHS digital, Infectopharm, Lundbeck, Medice, Neurim Pharmaceuticals, Oberberg GmbH, Roche, and Takeda. He received conference support or speaker’s fee by Medice and Takeda. He received royalities from Hogrefe, Kohlhammer, CIP Medien, Oxford University Press; the present work is unrelated to these relationships. The other authors report no biomedical financial interests or potential conflicts of interest. Supplementary Files SupplementalMaterial.docx Supplemental Material Cite Share Download PDF Status: Published Journal Publication published 02 Jan, 2025 Read the published version in Nature Mental Health → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3942971","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":273008233,"identity":"91eb0309-7b36-4156-9868-a5c969ae5573","order_by":0,"name":"Dorothea Floris","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4klEQVRIie3PPQrCMBTA8RcCmUpnu9grvOLqx1UUQSfBUZcSETJZnL1Fx44JAbtEXR0VT+AoCJqKiIutbg75Qx4k5EcIgMv1j0m7CLeDajsR6gDe86SMqOIC6z1I41cCYGcF8fPkeDpnzRAYmZ0m43iY7jYSLtlnEpi8gcoMIs7IPDKoR6ncdkliPhPcD1hNCU14OBMBRzlK1QIpEZXk1rGvFCQeova+IrL3JLSL6woSmDXFjej3RfEXjjpaGQ9VUkL8XJDDVLRbS0bVkV/j0N+Z6HApIa/Y+0Z+AVwul8tV0h1jrFXzzn6MuwAAAABJRU5ErkJggg==","orcid":"","institution":"Methods of Plasticity Research, Department of Psychology, University of Zurich","correspondingAuthor":true,"prefix":"","firstName":"Dorothea","middleName":"","lastName":"Floris","suffix":""},{"id":273008234,"identity":"54405e5d-28ec-4660-ac4e-420b882817b4","order_by":1,"name":"Alberto Llera","email":"","orcid":"","institution":"Donders Institute for Brain, Cognition, and Behavior, Radboud University Nijmegen; Department for Cognitive Neuroscience, Radboud University Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Alberto","middleName":"","lastName":"Llera","suffix":""},{"id":273008235,"identity":"d6985990-bb2c-410d-baf3-a24eba2b3b23","order_by":2,"name":"Mariam Zabihi","email":"","orcid":"","institution":"Donders Institute for Brain, Cognition, and Behavior, Radboud University Nijmegen; MRC Unit Lifelong Health and Aging, University College London","correspondingAuthor":false,"prefix":"","firstName":"Mariam","middleName":"","lastName":"Zabihi","suffix":""},{"id":273008236,"identity":"9629cb5c-5281-490f-8213-92f028d87bd5","order_by":3,"name":"Emily Jones","email":"","orcid":"","institution":"University of London","correspondingAuthor":false,"prefix":"","firstName":"Emily","middleName":"","lastName":"Jones","suffix":""},{"id":273008237,"identity":"02040077-82c1-46a2-b6a7-7bc4425834ff","order_by":4,"name":"Luke Mason","email":"","orcid":"","institution":"Centre for Brain and Cognitive Development, Birkbeck, University of London; Department of Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and Neuroscience,","correspondingAuthor":false,"prefix":"","firstName":"Luke","middleName":"","lastName":"Mason","suffix":""},{"id":273008238,"identity":"fb8125f5-5b27-4886-bbf5-01a744e59a2c","order_by":5,"name":"Rianne Haartsen","email":"","orcid":"","institution":"Centre for Brain and Cognitive Development, Birkbeck, University of London","correspondingAuthor":false,"prefix":"","firstName":"Rianne","middleName":"","lastName":"Haartsen","suffix":""},{"id":273008239,"identity":"f6bd1a00-5148-4f42-93b3-e75a2ef1aa23","order_by":6,"name":"Nathalie Holz","email":"","orcid":"","institution":"ZI Mannheim","correspondingAuthor":false,"prefix":"","firstName":"Nathalie","middleName":"","lastName":"Holz","suffix":""},{"id":273008240,"identity":"a53b4158-7b39-43c9-9e20-5fc6ca0b3423","order_by":7,"name":"Ting Mei","email":"","orcid":"https://orcid.org/0000-0003-3029-4460","institution":"Donders Institute for Brain, Cognition, and Behavior, Radboud University Nijmegen; Department for Cognitive Neuroscience, Radboud University Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Ting","middleName":"","lastName":"Mei","suffix":""},{"id":273008241,"identity":"b701e52b-322d-4cfc-92e7-a2ddd0114223","order_by":8,"name":"Camille Elleaume","email":"","orcid":"","institution":"Methods of Plasticity Research, Department of Psychology, University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Camille","middleName":"","lastName":"Elleaume","suffix":""},{"id":273008242,"identity":"db55c0d1-0e34-4a33-8f1b-1cf96ae01013","order_by":9,"name":"Bruno Hebling Vieira","email":"","orcid":"","institution":"Methods of Plasticity Research, Department of Psychology, University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Bruno","middleName":"Hebling","lastName":"Vieira","suffix":""},{"id":273008243,"identity":"1b532c61-b06c-4cde-bca5-564263d86890","order_by":10,"name":"Charlotte Pretzsch","email":"","orcid":"https://orcid.org/0000-0002-2761-6628","institution":"Institute of Psychiatry, Psychology and Neuroscience, King's College London","correspondingAuthor":false,"prefix":"","firstName":"Charlotte","middleName":"","lastName":"Pretzsch","suffix":""},{"id":273008244,"identity":"7ef46898-49c1-4397-8940-9e660ffe8427","order_by":11,"name":"Natalie Forde","email":"","orcid":"","institution":"Donders Institute for Brain, Cognition and Behaviour","correspondingAuthor":false,"prefix":"","firstName":"Natalie","middleName":"","lastName":"Forde","suffix":""},{"id":273008245,"identity":"6ea20f84-e779-47f1-bb32-d5bcc2cfb0bb","order_by":12,"name":"Sarah Baumeister","email":"","orcid":"https://orcid.org/0000-0001-9005-0084","institution":"Central Institute of Mental Health","correspondingAuthor":false,"prefix":"","firstName":"Sarah","middleName":"","lastName":"Baumeister","suffix":""},{"id":273008246,"identity":"61ed24dc-0577-435c-956b-0b5efa2abe12","order_by":13,"name":"Flavio Dell’Acqua","email":"","orcid":"","institution":"Department of Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and Neuroscience, King’s College London; Institute for Translational Neurodevelopment, King’","correspondingAuthor":false,"prefix":"","firstName":"Flavio","middleName":"","lastName":"Dell’Acqua","suffix":""},{"id":273008247,"identity":"3caaaa1e-9fd8-44b9-939d-cca422447acd","order_by":14,"name":"Sarah Durston","email":"","orcid":"","institution":"Department of Psychiatry, Brain Center Rudolf Magnus, University Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Sarah","middleName":"","lastName":"Durston","suffix":""},{"id":273008248,"identity":"e4de124f-108a-4173-bf87-f41762514d45","order_by":15,"name":"Tobias Banaschewski","email":"","orcid":"https://orcid.org/0000-0003-4595-1144","institution":"Central Institute of Mental Health, Mannheim","correspondingAuthor":false,"prefix":"","firstName":"Tobias","middleName":"","lastName":"Banaschewski","suffix":""},{"id":273008249,"identity":"fcceee59-a37c-486d-96e2-bee6faaaccb3","order_by":16,"name":"Christine Ecker","email":"","orcid":"","institution":"Department of Child and Adolescent Psychiatry, University Hospital, Goethe University","correspondingAuthor":false,"prefix":"","firstName":"Christine","middleName":"","lastName":"Ecker","suffix":""},{"id":273008250,"identity":"62d85d26-824f-4e5c-a382-3e3071b3b579","order_by":17,"name":"Rosemary Holt","email":"","orcid":"","institution":"Autism Research Centre, Department of Psychiatry, University of Cambridge","correspondingAuthor":false,"prefix":"","firstName":"Rosemary","middleName":"","lastName":"Holt","suffix":""},{"id":273008251,"identity":"acb8d34b-dc5e-47bc-826b-c4425a6b787d","order_by":18,"name":"Simon Baron-Cohen","email":"","orcid":"https://orcid.org/0000-0001-9217-2544","institution":"University of Cambridge","correspondingAuthor":false,"prefix":"","firstName":"Simon","middleName":"","lastName":"Baron-Cohen","suffix":""},{"id":273008252,"identity":"f164f284-bdf8-4e11-b549-00a3504a5ec5","order_by":19,"name":"Thomas Bougeron","email":"","orcid":"","institution":"Institut Pasteur","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Bougeron","suffix":""},{"id":273008253,"identity":"9d300110-ddbb-4bed-9bab-b32719526d09","order_by":20,"name":"Tony Charman","email":"","orcid":"https://orcid.org/0000-0003-1993-6549","institution":"King's College London","correspondingAuthor":false,"prefix":"","firstName":"Tony","middleName":"","lastName":"Charman","suffix":""},{"id":273008254,"identity":"443ba27c-6087-440d-805c-7292a61c8cb0","order_by":21,"name":"Eva Loth","email":"","orcid":"https://orcid.org/0000-0001-9458-9167","institution":"King's College London","correspondingAuthor":false,"prefix":"","firstName":"Eva","middleName":"","lastName":"Loth","suffix":""},{"id":273008255,"identity":"698e495f-a5a5-4234-b8ad-12b3699db97c","order_by":22,"name":"Declan Murphy","email":"","orcid":"https://orcid.org/0000-0002-6664-7451","institution":"King's College London","correspondingAuthor":false,"prefix":"","firstName":"Declan","middleName":"","lastName":"Murphy","suffix":""},{"id":273008256,"identity":"7c0ab932-944f-479e-b907-0a65d011ed0e","order_by":23,"name":"Jan Buitelaar","email":"","orcid":"https://orcid.org/0000-0001-8288-7757","institution":"Radboud University Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Jan","middleName":"","lastName":"Buitelaar","suffix":""},{"id":273008257,"identity":"1d26d698-ca31-488d-8bc9-07fcc4bed097","order_by":24,"name":"Christian Beckmann","email":"","orcid":"","institution":"Donders Institute for Brain, Cognition, and Behavior, Radboud University Nijmegen; Department for Cognitive Neuroscience, Radboud University Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Christian","middleName":"","lastName":"Beckmann","suffix":""},{"id":273008258,"identity":"b0af0c79-d7d4-46c0-808f-a882e320cd24","order_by":25,"name":"EU-AIMS LEAP group","email":"","orcid":"","institution":"Donders Institute for Brain, Cognition, and Behavior, Radboud University Nijmegen; Department for Cognitive Neuroscience, Radboud University Medical Center","correspondingAuthor":false,"prefix":"","firstName":"EU-AIMS","middleName":"LEAP","lastName":"group","suffix":""},{"id":273008259,"identity":"11120a24-70b3-400c-b3b4-66e8accfe822","order_by":26,"name":"Nicolas Langer","email":"","orcid":"","institution":"Methods of Plasticity Research, Department of Psychology, University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Nicolas","middleName":"","lastName":"Langer","suffix":""}],"badges":[],"createdAt":"2024-02-09 12:26:05","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3942971/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3942971/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s44220-024-00349-4","type":"published","date":"2025-01-02T05:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":51238935,"identity":"38da4478-0795-455c-b06b-2b862b19ca76","added_by":"auto","created_at":"2024-02-16 16:57:53","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":506187,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOverview of the methodological approach\u003c/strong\u003e. Features for each modality were extracted from the right and the left fusiform gyrus. These were: a) grey matter volume based on VBM for structural MRI; b) T-maps contrasting the faces condition to the shapes condition reflecting sensitivity to emotional faces from the Hariri paradigm for task-fMRI; c) seed-based (i.e., FFA) connectivity (SCA) for rs-fMRI; and d) the principal component of source reconstructed time series for EEG. Next, normative modelling was applied to each imaging modality using Bayesian Linear Regression. To model cross-subject individual-level variation, resulting Z-deviation maps per modality were statistically merged using linked independent component analysis resulting in measures of modality contributions and subject loadings. Next, we tested for group differences in ICs and group separability using either multi- or unimodal ICs. Finally, we computed multivariate associations (i.e., canonical correlation analysis) between subject loadings and clinical, cognitive measures related to either social-communicative or non-social features.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-3942971/v1/57d43b80687818f52d0c32cb.png"},{"id":51238411,"identity":"faa70504-7718-4f6e-b54c-d1f6581ce229","added_by":"auto","created_at":"2024-02-16 16:49:53","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":645603,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMultimodal components and their spatial and temporal characterization\u003c/strong\u003e. Among all ICs, eleven were considered multimodal (Fig. 2a), with a single modality contribution not more than 90%. IC44 showed a significant group difference with autistic individuals having higher contributions than NAI (Fig. 2b). Fig. 2c shows spatial and temporal Z-maps thresholded at the 95\u003csup\u003eth\u003c/sup\u003e percentile of the different modalities associated with IC44. Positive values (yellow) depict positive loadings onto the IC where autistic have higher deviations than NAI; negative values (blue) depict negative loadings onto the IC where autistic individuals have lower deviations than NAI. Suprathreshold timepoints are depicted in red. Fig. 2d-g depict the spatial overlap of suprathreshold voxels with a probabilistic functional atlas of the occipito-temporal cortex (i.e., VIS-atlas \u003cem\u003e(9)\u003c/em\u003e). Fig. 2h depicts the VIS-atlas and its different category-selective subregions. Fig. 2i-l show the spatial overlap of suprathreshold voxels with the structural Harvard-Oxford atlas and its four FFG subregions (i.e., anterior and posterior divisions of the temporal FFG, temporal occipital FFG and occipital FFG depicted in Fig. 2m). Fig. 2d-e and 2i-j show the positive loadings (i.e., autism \u0026gt; NAI) and Fig. 2f-g and 2k-l the negative loadings (i.e., autism \u0026lt; NAI), whereas Fig. 2d/f and 2i/k depict the left and Fig. 2e/g and 2j/l the right hemisphere.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-3942971/v1/7741368fd4ed3e0286c46855.png"},{"id":51239271,"identity":"80424236-81d5-4c15-9095-ad4d52a777cf","added_by":"auto","created_at":"2024-02-16 17:05:53","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":424872,"visible":true,"origin":"","legend":"\u003cp\u003eThe multivariate association (i.e., canonical correlation) was significant between the eleven multimodal ICs and the social-communicative features associated with autism. Fig. 3a shows the loadings of each multimodal component contributing to the CCA mode, while Fig. 3d shows the loadings of each social-communicative feature contributing to the CCA mode; stars show the significant loadings. Fig. 3b shows the canonical correlation scatterplot color-coded by the highest contributing clinical feature (ADOS social). The x-axis depicts the projected behavioural CCA variate and the y-axis the multimodal ICs CCA variates. Fig. 3c shows the modality contributions of the four ICs that contribute significantly to the CCA. Fig. 3e depicts the spatial and temporal patterns of each imaging modality that are significantly correlated with the social-communicative features. These are based on the significant correlation values between the Z-deviations of each imaging modality and the canonical imaging variate derived from the CCA.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-3942971/v1/df6f5daafcb38578020825b4.png"},{"id":72875694,"identity":"0babee12-01e1-43ac-8d67-4d2f89666599","added_by":"auto","created_at":"2025-01-03 08:09:53","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2443089,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3942971/v1/b46be1bc-9169-4211-9563-63e162c01c0e.pdf"},{"id":51238414,"identity":"3c49fb42-c6ce-4998-a48d-8ae1663fcfab","added_by":"auto","created_at":"2024-02-16 16:49:53","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":7690030,"visible":true,"origin":"","legend":"\u003cp\u003eSupplemental Material\u003c/p\u003e","description":"","filename":"SupplementalMaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-3942971/v1/67fcaa0ae90d53c34e50fcaa.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nJKB has been a consultant to, advisory board member of, and a speaker for Takeda/Shire, Medice, Roche, and Servier. He is not an employee of any of these companies and not a stock shareholder of any of these companies. He has no other financial or material support, including expert testimony, patents, or royalties. CFB is director and shareholder in SBGneuro Ltd. TC has received consultancy from Roche and Servier and received book royalties from Guildford Press and Sage. TB served in an advisory or consultancy role for ADHS digital, Infectopharm, Lundbeck, Medice, Neurim Pharmaceuticals, Oberberg GmbH, Roche, and Takeda. He received conference support or speaker’s fee by Medice and Takeda. He received royalities from Hogrefe, Kohlhammer, CIP Medien, Oxford University Press; the present work is unrelated to these relationships. The other authors report no biomedical financial interests or potential conflicts of interest.","formattedTitle":"A multimodal neural signature of face processing in autism within the fusiform gyrus","fulltext":[{"header":"One sentence summary","content":"\u003cp\u003eThe fusiform gyrus is a central region differentially implicated in autistic compared to non-autistic individuals across inter-related structural and functional imaging modalities and category-selective regions.\u003c/p\u003e"},{"header":"INTRODUCTION","content":"\u003cp\u003eAutism is a lifelong neurodevelopmental condition with a prevalence of 1 in 36 children \u003cem\u003e(1)\u003c/em\u003e. Social-communicative differences are among the most prominent features of autistic individuals \u003cem\u003e(2)\u003c/em\u003e. Particularly, difficulties with processing social information and faces, such as perceiving and interpreting facial expressions of emotions and other mental states are thought to have a profound impact on their social functioning and daily living skills \u003cem\u003e(3, 4)\u003c/em\u003e. While non-autistic individuals (NAI) appear to develop highly skilled strategies to discriminate facial cues at a very early age, autistic individuals have been reported to acquire less expertise with facial expression recognition \u003cem\u003e(5)\u003c/em\u003e. This has, for example, been attributed to diminished social attention \u003cem\u003e(6)\u003c/em\u003e and structural and functional differences in brain regions implicated in face processing \u003cem\u003e(3, 5, 7)\u003c/em\u003e. While individual neuroimaging modalities have separately been used to characterize the neural correlates of face processing, multimodal studies of key regions associated with face processing remain scarce. Illuminating the rich multimodal information shared across different imaging modalities can unravel complex interactions and variations that may only be partially addressed by single modalities \u003cem\u003e(8)\u003c/em\u003e. Specifically, elucidating cross-modal links with regards to face processing in autism will be crucial for understanding the biological mechanisms associated with core social difficulties and paving the way for the development of more personalised support.\u003c/p\u003e \u003cp\u003eThe fusiform gyrus (FFG) within the human ventral temporal cortex has been identified as a key neural region associated with higher-order processing of visual stimuli. The necessity for a thorough examination of the FFG in isolation is warranted by its detailed, functional heterogeneity exhibiting a fine-grained topographical organization with distinct category-selective patches \u003cem\u003e(9, 10)\u003c/em\u003e that are differentially specialized for facial recognition (i.e., the fusiform face area [FFA]) \u003cem\u003e(11)\u003c/em\u003e, body part discrimination \u003cem\u003e(12)\u003c/em\u003e and object features recognition \u003cem\u003e(13)\u003c/em\u003e. With regards to face processing, particularly the FFA has increased activation during face perception tasks in functional magnetic resonance imaging (fMRI) studies \u003cem\u003e(14, 15)\u003c/em\u003e along with evidence from electroencephalography (EEG) studies showing an event-related potential of negative polarity that peaks at around 170ms when facial stimuli are presented \u003cem\u003e(16, 17)\u003c/em\u003e. Furthermore, face processing is a lateralized cognitive function with right hemisphere dominance across these modalities \u003cem\u003e(18, 19)\u003c/em\u003e. An exhaustive examination across different neural signatures of this fine-grained local and hemispheric heterogeneity of the FFG \u0026ndash; beyond the FFA \u0026ndash; has not been conducted in autistic individuals yet. This can offer valuable new insights in the light of reports of atypical functional specialisation in autism \u003cem\u003e(20, 21)\u003c/em\u003e.\u003c/p\u003e \u003cp\u003eAccumulating evidence suggests that there is atypical neural organization within the FFG in autistic individuals. Many studies show that the FFG is hypoactive during face processing fMRI tasks \u003cem\u003e(22, 23)\u003c/em\u003e and, functionally atypically connected \u003cem\u003e(24, 25)\u003c/em\u003e in autism. EEG studies show that the N170 latency is delayed in autistic individuals \u003cem\u003e(26, 27)\u003c/em\u003e. Structurally, there are reports of volume increases in right FFG \u003cem\u003e(28)\u003c/em\u003e, a reduction in mean FFG neuron density \u003cem\u003e(29)\u003c/em\u003e and reversed leftward asymmetry \u003cem\u003e(30)\u003c/em\u003e in autism. These atypical neural substrates are thought to be functionally relevant in autistic individuals. For example, they have been linked to differences in facial expression recognition \u003cem\u003e(5)\u003c/em\u003e and face memory \u003cem\u003e(31)\u003c/em\u003e, adaptive social functioning \u003cem\u003e(27, 32)\u003c/em\u003e, and social symptom difficulties and severity \u003cem\u003e(24, 31, 33, 34)\u003c/em\u003e.\u003c/p\u003e \u003cp\u003eWhile these individual imaging modalities (i.e., structural MRI, task-fMRI, resting-state fMRI, EEG) converge to show atypical involvement of the FFG in face processing and related social functioning in autism, there is still little research into how these different neural substrates jointly inform fine-grained FFG organisation and social-communicative functioning in autism. Extracting common information from various modalities is crucial in gaining deeper insights into how brain structure and function reciprocally shape each other, and inform behaviour, cognition, and clinical conditions such as autism. To date, structure-function coupling has predominantly been addressed via univariate approaches where modalities are combined at the statistical or interpretation level \u003cem\u003e(27, 35, 36)\u003c/em\u003e. However, only when employing multivariate multimodal approaches can we identify direct relationships between different neurobiological mechanisms and how they scale relative to each other. We can further penetrate across different biological spatial and temporal scales of variation leveraging the unique, complementary aspects covered by each individual imaging modality. Prior multimodal efforts are promising as they show that combining information from brain structure and function significantly increases accuracy in predictive frameworks \u003cem\u003e(37\u0026ndash;41)\u003c/em\u003e. Also, a recent study combining different neuroimaging measures of rs-fMRI, diffusion-weighted imaging and structural morphometry specifically showed that rs-connection topographies within the FFG were differentially implicated between autistic and NAI \u003cem\u003e(42)\u003c/em\u003e. While such multimodal endeavours are still scarce in autism, this work specifically underscores the important role of the FFG in the neurobiology of autism. Still, the precise nature and a fine-grained topographical characterization of the multimodal neurobiological interactions within the FFG, and their relationship with the broader clinical phenotype related to social functioning in autism remain to be established.\u003c/p\u003e \u003cp\u003eIn the present study, our aim was to provide a more comprehensive understanding of the FFG in face processing in autism by elucidating the simultaneous involvement and multivariate interplay of different neural sources. Such analysis requires both large and deeply-phenotyped samples and given scarce availability, especially in clinical populations, this has previously limited its application. Hence, in this study, we leveraged the unique, large-scale and deeply-phenotyped EU-AIMS Longitudinal European Autism Project \u003cem\u003e(43, 44)\u003c/em\u003e (LEAP) which is the largest European multi-centre initiative aimed at identifying biomarkers in autism. This dataset provides a rich set of different neuroimaging modalities, and cognitive, clinical measures as well as tasks related to face processing and social and non-social functioning. Differences in facial expression recognition in autistic individuals have been established in this dataset \u003cem\u003e(5)\u003c/em\u003e. To further tap into their multimodal neural correlates, we combined two methodological innovations: (i) first, we employed normative modelling \u003cem\u003e(45)\u003c/em\u003e on each imaging modality separately to derive individual-level deviations from a predicted age-related trajectory. Prior research shows that modelling cortical features as deviations from a normative neurodevelopmental trajectory provides more sensitive measure to map multimodal signatures in psychopathology \u003cem\u003e(37)\u003c/em\u003e while also improving predictive performance \u003cem\u003e(46)\u003c/em\u003e. (ii) Next, we conducted multi-modal fusion through Linked Independent Component Analysis (LICA) \u003cem\u003e(47)\u003c/em\u003e across structural MRI, rs-fMRI, task-fMRI and EEG within the right and left FFG to simultaneously decompose the imaging data into underlying modes that characterise multi-modal signatures differentially in autistic and NAI. We further provided a fine-grained characterization of implicated regions shedding light on the topographic organisation within the FFG in autism. We hypothesized that multimodal components would be more sensitive to capturing subtle diagnostic effects cross-modally and would thus outperform unimodal components in discriminating autistic from NAI. Finally, we hypothesized that joint expression across modalities related to the FFG, and face processing would specifically inform social functioning in autism.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure\u0026nbsp;1. Overview of the methodological approach\u003c/b\u003e. Features for each modality were extracted from the right and the left fusiform gyrus. These were: a) grey matter volume based on VBM for structural MRI; b) T-maps contrasting the faces condition to the shapes condition reflecting sensitivity to emotional faces from the Hariri paradigm for task-fMRI; c) seed-based (i.e., FFA) connectivity (SCA) for rs-fMRI; and d) the principal component of source reconstructed time series for EEG. Next, normative modelling was applied to each imaging modality using Bayesian Linear Regression. To model cross-subject individual-level variation, resulting Z-deviation maps per modality were statistically merged using linked independent component analysis resulting in measures of modality contributions and subject loadings. Next, we tested for group differences in ICs and group separability using either multi- or unimodal ICs. Finally, we computed multivariate associations (i.e., canonical correlation analysis) between subject loadings and clinical, cognitive measures related to either social-communicative or non-social features.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003eSample\u003c/p\u003e \u003cp\u003eFor an overview of all methods see Fig.\u0026nbsp;1. The final sample of autistic (N\u0026thinsp;=\u0026thinsp;99) and NAI (N\u0026thinsp;=\u0026thinsp;105) did not differ significantly in sex ratio, age, measures of intellectual functioning, measures of structural image quality, number of EEG trials and head motion associated with task- and rs-fMRI (Table\u0026nbsp;1).\u003c/p\u003e \u003cp\u003eUnimodal normative models\u003c/p\u003e \u003cp\u003eFirst, unimodal normative models were estimated. Their accuracy was evaluated using the correlation between the true and the predicted voxel values (Rho), the explained variance, the mean standardized log-loss (MSLL) and standardized mean squared error (SMSE) (Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) and normative models per modality (Fig. S3). Evaluation metrics were largely within recommended ranges \u003cem\u003e(48)\u003c/em\u003e and highly similar when modelling age linearly (Fig. S2). When testing for group differences in \u003cem\u003eunimodal\u003c/em\u003e features, there were no significant differences in extreme Z-deviations between autistic and NAI for any of the eight features (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLinked independent component analysis\u003c/p\u003e \u003cp\u003eNext, the Z-deviations (features) were merged using LICA \u003cem\u003e(47)\u003c/em\u003e. Fifty independent components (IC) were derived across eight different brain feature maps per hemisphere (i.e., modalities) (Fig. S4). Overall, across these, the right hemisphere (51.7%) and the left hemisphere (48.3%) did not contribute differentially (\u003cem\u003eχ\u0026sup2;\u003c/em\u003e=1.2, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.72). Single modality contributions were as follows: EEG R (35.0%)\u0026thinsp;\u0026gt;\u0026thinsp;EEG L (33.2%)\u0026thinsp;\u0026gt;\u0026thinsp;rs-fMRI R (11.2%)\u0026thinsp;\u0026gt;\u0026thinsp;rs-fMRI L (9.6%)\u0026thinsp;\u0026gt;\u0026thinsp;task-fMRI R (3.5%)\u0026thinsp;\u0026gt;\u0026thinsp;task-fMRI L (3.4%)\u0026thinsp;\u0026gt;\u0026thinsp;structure L (2.1%)\u0026thinsp;\u0026gt;\u0026thinsp;structure R (2.1%). Fig. S5 shows the correlations between the 50-dimensional factorizations (y-axis) and alternative 40 (Fig.S5a) and 60 (Fig.S5b) dimensional factorizations. In line with previous reports \u003cem\u003e(49)\u003c/em\u003e, most components were recovered with high accuracy independently of the order of the factorization.\u003c/p\u003e \u003cp\u003eGroup Differences\u003c/p\u003e \u003cp\u003eNext, we compared the subject loadings of all (uni- and multimodal) ICs to test for differences between autistic and NAI. Among these, one multimodal IC (#44) showed a significant group difference with autistic individuals having higher contributions than NAI (\u003cem\u003et\u003c/em\u003e\u0026thinsp;=\u0026thinsp;3.5, \u003cem\u003ep\u003c/em\u003e\u003csub\u003eFDR\u003c/sub\u003e=0.026) (Fig.\u0026nbsp;2b). There were no significant group differences in the remaining ICs (Table S2). The significant multimodal component was not differentially driven by the right (52.8%) or left hemisphere (47.2%) (\u003cem\u003eχ\u0026sup2;\u003c/em\u003e=0.4, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.51) and was associated with several functional modalities (rs-fMRI R [48.5%], rs-fMRI L [35.0%], EEG L [11.6%], task-fMRI R [3.3%]), and to a smaller extent with GM volume (structure R [1.0%], structure L [0.5%]). Figure\u0026nbsp;2c depicts the spatial and temporal patterns for each imaging modality within IC44. When characterizing these further in terms of their anatomical and functional overlaps with the Harvard-Oxford atlas (HOA) and the probabilistic functional atlas of human occipito-temporal visual cortex\u003cem\u003e(9)\u003c/em\u003e (VIS), in the left hemisphere, autistic individuals showed more functional deviations than expected in rs-fMRI connectivity primarily in retinotopic regions of occipital FFG, while to a smaller extent also in lower-order face-selective regions (IOG) (Fig.\u0026nbsp;2d and 2i). In the right hemisphere, they showed linked increased deviations in rs-fMRI and structure primarily in higher-order face- (mFus, pFus) and bodies-selective (OTS) regions of temporal-occipital and occipital FFG (Fig.\u0026nbsp;2e and 2j). On the other hand, regions in the left hemisphere where autistic individuals showed linked decreased deviations compared to NAI, localized to both higher-order face-selective (mFus, pFus) and retinotopic regions of posterior, temporal-occipital and occipital FFG (Fig.\u0026nbsp;2f and 2k). In the right hemisphere, these were mostly in higher-order face face-selective regions (pFus) across rs-fMRI and task-fMRI and in retinotopic regions across structure in temporal-occipital FFG (Fig.\u0026nbsp;2g and 2l). Furthermore, autistic individuals showed more left EEG source activation than expected around 195-203ms and 417-426ms, whereas less source activation at 444\u0026ndash;449ms than expected. For further details see Table S3.\u003c/p\u003e \u003cp\u003eMultimodal components\u003c/p\u003e \u003cp\u003eFor further analyses, we focused on multimodal components only by excluding those which were primarily driven by one imaging modality, resulting in eleven multimodal ICs (Fig.\u0026nbsp;2a). Across these multimodal ICs, the right hemisphere (60.0%) contributed more than the left hemisphere (40.0%) (\u003cem\u003eχ\u0026sup2;\u003c/em\u003e=7.2, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.007). Single modality contributions across all multimodal ICs were as follows: EEG R (26.3%)\u0026thinsp;\u0026gt;\u0026thinsp;rs-fMRI L (19.7%)\u0026thinsp;\u0026gt;\u0026thinsp;EEG L (13.9%)\u0026thinsp;\u0026gt;\u0026thinsp;rs-fMRI R (12.1%)\u0026thinsp;\u0026gt;\u0026thinsp;task-fMRI L (9.8%)\u0026thinsp;\u0026gt;\u0026thinsp;task-fMRI R (9.6%)\u0026thinsp;\u0026gt;\u0026thinsp;structure R (4.4%)\u0026thinsp;\u0026gt;\u0026thinsp;structure L (4.2%).\u003c/p\u003e \u003cp\u003eAutism classification\u003c/p\u003e \u003cp\u003eNext, we applied a support vector machine (SVM) to compare unimodal and multimodal ICs\u0026rsquo; efficacy in differentiating between the two diagnostic groups. Results showed that multimodal ICs performed significantly better at discriminating autistic from NAI (AUC unimodal\u0026thinsp;=\u0026thinsp;0.48, AUC multimodal\u0026thinsp;=\u0026thinsp;0.64, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This result was confirmed across a range of different multimodality thresholds (Fig. S6a) and was not influenced by varying amounts of features between multimodal and unimodal ICs (Fig. S6b).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure\u0026nbsp;2\u003c/b\u003e. \u003cb\u003eMultimodal components and their spatial and temporal characterization\u003c/b\u003e. Among all ICs, eleven were considered multimodal (Fig.\u0026nbsp;2a), with a single modality contribution not more than 90%. IC44 showed a significant group difference with autistic individuals having higher contributions than NAI (Fig.\u0026nbsp;2b). Figure\u0026nbsp;2c shows spatial and temporal Z-maps thresholded at the 95th percentile of the different modalities associated with IC44. Positive values (yellow) depict positive loadings onto the IC where autistic have higher deviations than NAI; negative values (blue) depict negative loadings onto the IC where autistic individuals have lower deviations than NAI. Suprathreshold timepoints are depicted in red. Figure\u0026nbsp;2d-g depict the spatial overlap of suprathreshold voxels with a probabilistic functional atlas of the occipito-temporal cortex (i.e., VIS-atlas \u003cem\u003e(9)\u003c/em\u003e). Figure\u0026nbsp;2h depicts the VIS-atlas and its different category-selective subregions. Figure\u0026nbsp;2i-l show the spatial overlap of suprathreshold voxels with the structural Harvard-Oxford atlas and its four FFG subregions (i.e., anterior and posterior divisions of the temporal FFG, temporal occipital FFG and occipital FFG depicted in Fig.\u0026nbsp;2m). Figure\u0026nbsp;2d-e and 2i-j show the positive loadings (i.e., autism\u0026thinsp;\u0026gt;\u0026thinsp;NAI) and Fig.\u0026nbsp;2f-g and 2k-l the negative loadings (i.e., autism\u0026thinsp;\u0026lt;\u0026thinsp;NAI), whereas Fig.\u0026nbsp;2d/f and 2i/k depict the left and Fig.\u0026nbsp;2e/g and 2j/l the right hemisphere.\u003c/p\u003e \u003cp\u003eClinical, cognitive associations\u003c/p\u003e \u003cp\u003eTo test for brain-behaviour relations, we ran canonical correlation analysis (CCA). This revealed a significant multivariate association between the multimodal ICs and social-communicative features (i.e., ADOS-social affect, ADI-social, ADI-communication, Vineland Adaptive Behavior Scale \u003cem\u003e(50)\u003c/em\u003e with Communication, Daily Living, Socialization subscales, the Reading the Mind in the Eyes test \u003cem\u003e(51)\u003c/em\u003e, Hariri faces task \u003cem\u003e(52)\u003c/em\u003e) (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.65, \u003cem\u003ep\u003c/em\u003e\u003csub\u003eFDR\u003c/sub\u003e=0.008; Fig.\u0026nbsp;3b). On the other hand, the relationship between the multimodal ICs and non-social features (i.e., ADOS-RRB, ADI-RRB, the Repetitive Behavior Scale \u003cem\u003e(53)\u003c/em\u003e, the Short Sensory Profile \u003cem\u003e(54)\u003c/em\u003e, the Systemizing Quotient \u003cem\u003e(55\u0026ndash;57)\u003c/em\u003e, Hariri shape matching condition) was not significant (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.49, \u003cem\u003ep\u003c/em\u003e\u003csub\u003eFDR\u003c/sub\u003e=0.51; Fig. S7) pointing to specificity with social-related features of multimodal ICs. These associations remained stable when varying the multimodality threshold (Fig. S8). For the significant association, multimodal IC37 showed the largest contribution on the imaging side followed by IC38, IC44 and IC34 (Fig.\u0026nbsp;3a and 3c), whereas ADOS social-affect, RMET and Hariri face matching scores showed the largest contribution on the behavioural side (Fig.\u0026nbsp;3d). The ICs contributing most are depicted in Fig.\u0026nbsp;3c and Fig. S9-11. On average, the right (56.5%) and the left hemisphere (43.5%) did not contribute differentially to these four ICs (\u003cem\u003eχ\u0026sup2;\u003c/em\u003e=2.9, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.09) which were mostly driven by all functional modalities. Next, imaging patterns correlating with social-communication features were characterized in terms of their overlap with anatomical and functional overlaps with the HOA and VIS atlases (Fig.\u0026nbsp;3e and S12). Especially in higher-order face-selective regions (mFus and pFus) of posterior and temporal-occipital FFG, there were both linked \u003cem\u003eincreased\u003c/em\u003e deviations in bilateral rs-fMRI and task-fMRI and linked \u003cem\u003edecreased\u003c/em\u003e deviations in bilateral structure and right rs-fMRI connectivity. At the same time, particularly in retinotopic regions of occipital FFG there was more bilateral GM volume along with less right task-activation than expected. There were more deviations in right EEG source activation at around 290ms, while left EEG did not reach significance. These joint imaging patterns were associated with more social difficulties as assessed by the ADOS, ADI and Vineland and more errors on the RMET, while also with greater accuracy on the Hariri faces task. For more details, see Tables S4-7.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure\u0026nbsp;3\u003c/b\u003e. The multivariate association (i.e., canonical correlation) was significant between the eleven multimodal ICs and the social-communicative features associated with autism. Figure\u0026nbsp;3a shows the loadings of each multimodal component contributing to the CCA mode, while Fig.\u0026nbsp;3d shows the loadings of each social-communicative feature contributing to the CCA mode; stars show the significant loadings. Figure\u0026nbsp;3b shows the canonical correlation scatterplot color-coded by the highest contributing clinical feature (ADOS social). The x-axis depicts the projected behavioural CCA variate and the y-axis the multimodal ICs CCA variates. Figure\u0026nbsp;3c shows the modality contributions of the four ICs that contribute significantly to the CCA. Figure\u0026nbsp;3e depicts the spatial and temporal patterns of each imaging modality that are significantly correlated with the social-communicative features. These are based on the significant correlation values between the Z-deviations of each imaging modality and the canonical imaging variate derived from the CCA.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn the present study, we aimed to characterize the multimodal neural signature of face processing in autism within the FFG, the core region of the face processing network. We identified several ICs that were differentially associated with the four modalities (structure, rs-fMRI, task-fMRI, and EEG), hemispheres, and functional subdivisions of the FFG. Autism-associated differences in FFG organization were more pronounced when penetrating across multiple than single modalities. Furthermore, a set of multimodal ICs was associated with core features related to social but not non-social functioning in autism. Taken together, these findings highlight the value of cross-modal analyses in characterizing a key structure in the multilevel neurobiology of autism and its implication in core cognitive and clinical features.\u003c/p\u003e \u003cp\u003eGroup differences\u003c/p\u003e \u003cp\u003eAmong all components, one multimodal component (i.e., IC44) showed a significant difference in subject loadings between autistic and NAI. Overall, the right and left hemispheres did not show differential contributions within this IC, and it was associated with all modalities fed into the analysis, with the functional modalities, especially rs-fMRI and EEG, contributing most (see Fig.\u0026nbsp;2). Particularly, the overlap with the VIS-atlas highlighted that face-selective and retinotopic regions of the FFG were most different between autistic and NAI. More specifically, in the right hemisphere, higher-order face-selective regions exhibited less task activation and FFA-connectivity than expected, primarily in occipital FFG areas (Fig.\u0026nbsp;2g and 2l). At the same time autistic individuals showed increased deviations in FFA-connectivity primarily in temporal-occipital FFG along with increased GM volume deviations in higher-order face-selective FFG regions (Fig.\u0026nbsp;2e and 2j). This strong right-hemisphere involvement of regions associated with FFA across several modalities is in line with reports of increased FFA volume \u003cem\u003e(28)\u003c/em\u003e and decreased FFA task-activation \u003cem\u003e(58, 59)\u003c/em\u003e and FFA-connectivity \u003cem\u003e(24, 60)\u003c/em\u003e in autism. Similarly, temporally, autistic individuals showed more increased left deviations around 195ms potentially indicative of the consistently reported finding of a slower N170 in autistic individuals \u003cem\u003e(26)\u003c/em\u003e. This has specifically also been shown and extensively characterised in the current sample \u003cem\u003e(27)\u003c/em\u003e. Together these patterns converge to point towards autism-associated differences in face-selective areas of the FFG, both at the structural, functional, and temporal levels. Although these results align with earlier unimodal discoveries, previously it was uncertain whether disparate signals would be separate or coalesce to a joint multimodal expression. In this context, we provide evidence supporting the interconnected nature of distinct signals within a single, unified framework.\u003c/p\u003e \u003cp\u003eIn the left hemisphere, IC44-related increased deviations in EEG source activation at around 420ms may indicate reductions in the face-N400 which has been associated with familiar face recognition and semantic information \u003cem\u003e(61)\u003c/em\u003e. While in NAI face processing becomes the most highly developed visual skill, in autistic individuals faces may convey greater novelty and thus decreased familiarity. Furthermore, occipital, retinotopic areas of the left FFG were most implicated as shown by increased functional connectivity deviations between the FFA and retinotopic and lower-order face-selective areas of the FFG in autistic individuals (Fig.\u0026nbsp;2d and 2i). This was echoed by less GM volume than expected in left retinotopic areas of FFG in autistic individuals (Fig.\u0026nbsp;2f and 2k). Retinotopic, early visual areas act as the first stage in a hierarchical network of face processing in which lower-level feature-based components are processed before more complex features in higher-order face-selective regions \u003cem\u003e(62)\u003c/em\u003e. Neural deviations in early visual areas as seen here are in line with reports of autistic individuals showing differences in sensory processing at early perceptual stages and have been described at the cognitive level as weak central coherence \u003cem\u003e(63)\u003c/em\u003e. Accordingly, studies show that autistic individuals exhibit a different strategy in processing facial and visual stimuli with a stronger focus on featural, local aspects at the expense of holistic, global information \u003cem\u003e(64)\u003c/em\u003e. Similarly, fMRI studies converge to show greater feature-based perceptual strategies in autistic individuals who primarily tend to recruit object-related regions \u003cem\u003e(65, 66)\u003c/em\u003e when viewing facial stimuli. Taken together, this suggests that differences we discovered in the left hemisphere point primarily to low-level, bottom-up processing differences, whereas in the right hemisphere they may indicate higher-level atypicalities in the FFA, with a differential involvement across the different structural and functional modalities.\u003c/p\u003e \u003cp\u003eClinical, cognitive associations\u003c/p\u003e \u003cp\u003eMultimodal ICs showed a significant association with a set of clinical and cognitive features associated with social functioning in autism (Fig.\u0026nbsp;3). Group-differential IC44 was also among the significantly contributing ICs to this associations. Components loading significantly onto the CCA were mostly driven by functional modalities. Right EEG source activation deviations were at around 280-300ms, potentially indicative of the N250r generated in the FFG \u003cem\u003e(66)\u003c/em\u003e and associated with repetition of familiar facial stimuli \u003cem\u003e(68)\u003c/em\u003e. The amplitude of the N250r has been shown to decrease with increasing working memory (WM) load \u003cem\u003e(69)\u003c/em\u003e. This would translate into increased deviations as seen in autistic individuals here and may imply differences in degrees of WM resources allocated to the processing of facial stimuli which in turn have a larger novelty character in autistic individuals requiring more attentional effort. With regards to the other modalities, increased deviations particularly in higher-order face-selective regions across brain function (task-fMRI and rs-fMRI) while also in lower-order early visual regions across brain structure were associated with more autistic features, such as more social difficulties as assessed by ADOS, and lower social sensitivity as assessed by the RMET. Previous unimodal studies showed that the delayed latency of the N170 predicts change in social adaptive behaviour in autistic individuals \u003cem\u003e(27)\u003c/em\u003e (i.e., EEG), autistic individuals with low performance on facial emotion recognition have reduced bilateral FFG activation (i.e., task-fMRI) \u003cem\u003e(5)\u003c/em\u003e and atypical FFA-connectivity is associated with increased social symptom severity in autism (i.e., rs-fMRI) \u003cem\u003e(24)\u003c/em\u003e. Here, we extend unimodal results to a multivariate association across a range of social-communicative features that are related to cross-modal signatures within the FFG. Here, we provide evidence for an interrelated biological basis of core social functioning in autism and that appropriately modelled shared variance across different modalities increases sensitivity to clinical-cognitive features associated with autism \u003cem\u003e(70)\u003c/em\u003e. Remarkably, at the same time, there was no association with a set of non-social features, such as repetitive behaviours or sensory processing, pointing to specificity of these multimodal ICs with regards to social functioning.\u003c/p\u003e \u003cp\u003eSummary and implications\u003c/p\u003e \u003cp\u003eTaken together, the multimodal neural signature within the FFG in autism presents differentially across hemispheres, modalities, and topography. Specifically, the picture emerges that (i) the functional modalities contribute more than the structural modalities and (ii) retinotopic, occipital regions are more implicated in the left hemisphere and higher-order regions more implicated in the right hemisphere within the FFG when it comes to group differences; but they do not contribute differentially with regards to social functioning. (i) Concurrent neural activity and functional co-expressions (task-, rs-fMRI, EEG) were strongly tied to social features observed in autistic individuals at present (such as current performance and ADOS assessment). On the other hand, more stable structural aspects of the brain established over time and historical symptoms reported through the ADI and Vineland \u0026ndash; which provide insights into past behaviors \u0026ndash; had a comparatively smaller impact on the observed association. These results highlight the dynamic nature of the relationship between neural activity and social functioning in autism and underscore the importance of considering the temporal dimension when investigating the neural correlates of social functioning in autism. Putative future neuroscientifically informed interventions targeting social features may thus benefit from a focus on concurrent neural functioning. (ii) Topographically, the FFG is known to exhibit an anterior to posterior gradient with more posterior regions related to lower-order, early visual processing, and more anterior regions related to higher-order processing \u003cem\u003e(71)\u003c/em\u003e. Here, we see the involvement of both retinotopic and higher-order cognitive, particularly face-sensitive patches pointing to differences in both bottom-up perceptual processes and top-down cognitive information processing in face processing in autism which can amount to a difference in the face processing strategy employed (e.g., more feature-based). These different processing levels are not differentially implicated across hemispheres in the processing of social information in autism suggesting that the distinctive face processing strategy in autism transcends right hemisphere dominance of face processing. On the other hand, hemispheric differences are more apparent in the group-differential IC. Teasing apart hemispheric contributions is particularly important in the light of reports of atypical patterns of brain asymmetry in autistic individuals \u003cem\u003e(21, 34, 72)\u003c/em\u003e. More extreme deviations from a normative model have for example been reported in right temporal-occipital fusiform cortex asymmetry in autistic females \u003cem\u003e(21)\u003c/em\u003e, along with more left-lateralized volume in posterior temporal FFG in autistic individuals \u003cem\u003e(30, 34)\u003c/em\u003e. Subsequent research should delve further into these more nuanced insights revealed by cross-modal analyses pointing to left-lateralized low-level and right-lateralized high-level differences between autistic and NAI.\u003c/p\u003e \u003cp\u003eStrengths and limitations\u003c/p\u003e \u003cp\u003eIntegrating data from different modalities has the advantage of being biologically more informative and comprehensive in characterizing a complex, heterogenous condition like autism. Accordingly, when comparing unimodal deviations in each imaging modality, as well as comparing predominantly unimodal ICs between autistic and NAI, there were no significant group differences, despite employing a more sensitive individual-level measures derived from normative modelling. Also, multimodal features significantly outperformed unimodal features in differentiating autistic form NAI. These results together confirm our hypothesis and previous reports \u003cem\u003e(70, 73)\u003c/em\u003e that appropriately modelling cross-modal variance increases sensitivity to detecting subtle effects that may otherwise be missed. Thus, integrating different structural and functional brain measures is the most promising and powerful method to achieve significant advances in our understanding of system-level atypicalities in autism and provides the basis for elucidating mechanisms through which interventions can most efficiently improve clinically relevant functioning \u003cem\u003e(70)\u003c/em\u003e. Furthermore, we combine different innovative methods. LICA is particularly powerful when modelling modalities that are different in their numbers of features, spatial correlations, intensity distributions and units. This is, because LICA optimally weighs the contributions of each modality by the correction for the number of effective degrees of freedom and the use of automatic relevance determination priors on components \u003cem\u003e(8, 47, 70)\u003c/em\u003e. Also, by combining normative modelling with LICA, we employ a previously validated approach that has been shown to increase sensitivity in detecting cross-modal effects in clinical populations \u003cem\u003e(37)\u003c/em\u003e.\u003c/p\u003e \u003cp\u003eAt the same time, it needs to be pointed out that face processing involves an extended neural network across the whole brain including other structures such as the amygdala, superior temporal sulcus and occipital and frontal cortex \u003cem\u003e(37, 66, 74, 75)\u003c/em\u003e. It may thus seem too simplistic to reduce face processing to a single brain region. Still, the FFG has been claimed the core node of a distributed face processing network, as also substantiated by FFG lesion studies \u003cem\u003e(76, 77)\u003c/em\u003e, and its fine-grained functional heterogeneity warrants careful examination in isolation. Also, implementing cross-modal analyses presents with additional challenges, such as obtaining sufficiently large sample sizes with all participants having available data across all imaging modalities. Here, from a sample of over 600 individuals in the EU-AIMS LEAP dataset, we were able to conduct analysis in just over 200 individuals who had available imaging data across the four different modalities. Whole brain analyses based on multivariate techniques will ultimately require larger sample sizes.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIntegrating information from multiple imaging modalities allows us to gain a more holistic and robust understanding of the complex neural processes underlying core clinical and cognitive features associated with autism. Present results suggest that the FFG is a central region differentially implicated across different neural signals and category-selective regions in autistic and NAI and that this informs cross-modally the mechanisms associated with core social functioning in autism. Eventually, elucidating more precise, integrated and individual-level neural associations of core cognitive and clinical features, will pave the way for further work identifying stratification, mechanistic and prognostic biomarkers, and the development of more personalised support, thereby eventually improving the quality of lives of autistic individuals.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003eSample characterization\u003c/p\u003e\n\u003cp\u003eParticipants were part of the EU-AIMS/AIMS-2-TRIALS LEAP cohort \u003cem\u003e(43, 44)\u003c/em\u003e. They underwent comprehensive clinical, cognitive and MRI assessment at one of six collaborating sites. All autistic participants had an existing clinical diagnosis of autism which was confirmed using the combined information of gold-standard diagnostic instruments, the Autism Diagnostic Interview-Revised \u003cem\u003e(78)\u003c/em\u003e (ADI-R) and the Autism Diagnostic Observation Schedule \u003cem\u003e(79)\u003c/em\u003e (ADOS). The study was approved by the respective research ethics committees at each site (IRAS, UK). Informed written consent was obtained from all participants, or\u0026mdash;for minors or those unable to give informed consent\u0026mdash;from a parent or legal guardian. For further details see Supplemental Information (SI) and our earlier papers \u003cem\u003e(43, 44)\u003c/em\u003e. The final sample has both complete imaging data across four different imaging modalities and phenotypic information available consisting of 99 autistic and 105 NAI between 7\u0026ndash;30 years (Table\u0026nbsp;1).\u003c/p\u003e\n\u003cp\u003eClinical and cognitive measures\u003c/p\u003e\n\u003cp\u003eWe split available autism-associated measures into two sets of feature sets based on the construct they measure 1) \u003cem\u003esocial-communicative features\u003c/em\u003e comprising measures of difficulties with social communication and daily living skills (i.e., ADOS-social affect, ADI-communication, ADI-social, Vineland Adaptive Behavior Scale \u003cem\u003e(50)\u003c/em\u003e with Communication, Daily Living, Socialization subscales), emotional face matching performance (i.e., Hariri faces task \u003cem\u003e(52)\u003c/em\u003e), and social sensitivity to complex emotions (i.e., Reading the Mind in the Eyes test \u003cem\u003e(51)\u003c/em\u003e [RMET]) and 2) \u003cem\u003enon-social features\u003c/em\u003e comprising restricted, repetitive behaviours (RRBs) (i.e., ADOS-RRB, ADI-RRB, the Repetitive Behavior Scale \u003cem\u003e(53)\u003c/em\u003e [RBS-R]), systemizing (i.e., the Systemizing Quotient \u003cem\u003e(55\u0026ndash;57)\u003c/em\u003e [SQ]), shape matching performance (i.e., Hariri shapes task, as the control condition to the Hariri emotional faces task) and sensory processing atypicalities (i.e., Short Sensory Profile \u003cem\u003e(54)\u003c/em\u003e [SSP]) (see SI and Table S8). To tackle missing clinical data and to not further reduce sample size, we used imputed clinical data \u003cem\u003e(80)\u003c/em\u003e, as in previous work with this dataset \u003cem\u003e(41, 81)\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eRegion of interest: fusiform gyrus\u003c/p\u003e\n\u003cp\u003eAll analyses were restricted to the right and left FFG based on the HOA (FMRIB, Oxford, UK) adjusted to have 100% coverage across all individuals for each imaging modality (see SI).\u003c/p\u003e\n\u003cp\u003eImaging modalities\u003c/p\u003e\n\u003cp\u003eFor MRI and EEG data acquisition parameters and detailed preprocessing steps per modality, see SI and Table S9-10. In summary, following were the features for subsequent normative modeling: a) structure: VBM-derived, voxel-wise GM volumes; b) rs-fMRI: seed-based correlation between the FFA and the remaining FFG; c) task-fMRI: T-contrast maps reflecting sensitivity to emotional faces; d) EEG: the principal component of source reconstructed activation obtained across different cortical parcels.\u003c/p\u003e\n\u003cp\u003eNormative modelling\u003c/p\u003e\n\u003cp\u003eNormative modelling is an emerging statistical technique that allows parsing heterogeneity by charting variation in brain-behaviour mappings relative to a normative range and provides statistical inference at the level of the individual \u003cem\u003e(82)\u003c/em\u003e. The term \u0026lsquo;normative\u0026rsquo; should not be seen as incompatible with the neurodiversity framework as it simply refers to statistical norms such as growth charts that vary by demographics such as age and gender. Here, we trained normative models using Bayesian Linear Regression (BLR) \u003cem\u003e(83)\u003c/em\u003e for each brain imaging modality within the right and left FFG ROI independently using age, sex and scanning site as covariates. A B-spline basis expansion of the covariate vector was used to model non-linear effects of age. Normative models were derived in an unbiased manner across the entire sample under 10-fold cross-validation \u003cem\u003e(37, 45, 84)\u003c/em\u003e. To estimate voxel-wise/time-point-wise deviations for each modality in each individual, we derived normative probability maps (NPM) that quantify the deviation from the normative model summarized in \u003cem\u003eZ\u003c/em\u003e-scores. These subject-specific Z-score images provide a statistical estimate of how much each individual\u0026rsquo;s recorded value differs from the predicted value at each voxel/time-point. The accuracy of the normative model was evaluated using the correlation between the recorded and the predicted voxel values (Rho), MSLL, SMSE, and the EV (Fig. \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e) as well as based on the forward models (Fig. S3). Furthermore, we compared model performance when modelling age linearly (without a B-spline basis expansion; Fig. S2). To assess whether autistic and NAI differed in their extreme deviations based on \u003cem\u003eunimodal\u003c/em\u003e features, thresholded Z-scores (Z\u0026gt;|2.6| \u003cem\u003e(21, 37, 85)\u003c/em\u003e, corresponding to the 99.5th percentile) were compared between the two groups using a two-sample t-test (see SI). Code is available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/amarquand/PCNtoolkit\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eLinked Independent Component Analysis\u003c/p\u003e\n\u003cp\u003eIn order to gain more comprehensive insights into cross-modal signatures of face processing, we merged the different individual-level deviations from all imaging modalities (GM volume, FFA-connectivity, T-maps contrasting the faces to the shapes condition, and the principal components of source reconstructed time series) using LICA \u003cem\u003e(47)\u003c/em\u003e (see SI). This is a Bayesian extension of the single modality ICA model which provides an automatic and simultaneous decomposition of the brain features into independent components (ICs) that characterize the inter-subject brain variability. These multiple decompositions share a mixing matrix (i.e., subject course) across individual feature factorizations that reflect the subject contributions to each IC. Here, LICA was used to merge the unthresholded Z-deviation maps derived from normative modeling across the four different imaging modalities within the right and left FFG ROIs. Each measure per hemisphere was treated separately resulting in eight input maps (i.e., modalities). Hemispheres were modelled separately given known brain asymmetric differences in autism \u003cem\u003e(21, 34)\u003c/em\u003e and to study the hemispheric contributions and model the different noise characteristics individually. We estimated 50 independent components based on our sample size and following recommendations described in earlier papers \u003cem\u003e(37, 41, 49, 86)\u003c/em\u003e (i.e., sample size\u0026thinsp;~\u0026thinsp;N / 4). To evaluate the robustness of our selected model order (N\u0026thinsp;=\u0026thinsp;50), we re-ran LICA using different dimensional factorizations of subject loadings (N\u0026thinsp;=\u0026thinsp;40 and N\u0026thinsp;=\u0026thinsp;60) and computed correlations among them.\u003c/p\u003e\n\u003cp\u003eGroup Differences\u003c/p\u003e\n\u003cp\u003eThe subject loadings of all ICs were compared between autistic and NAI using a two-sample t-test. Multiple comparisons were corrected for using the False Discovery Rate (FDR) \u003cem\u003e(87)\u003c/em\u003e. ICs showing significant group differences were further characterized by plotting each contributing modality\u0026rsquo;s spatial map and temporal profile (Z-thresholded at the 95th percentile). To further characterize the most implicated regions within the FFG per modality, we computed the overlap between supra-threshold voxels and a structural (i.e., the Harvard-Oxford atlas, which covers the entire FFG) and a functional atlas (i.e., a probabilistic functional atlas of the occipito-temporal cortex \u003cem\u003e(9)\u003c/em\u003e which covers category-specific FFG patches) (see SI).\u003c/p\u003e\n\u003cp\u003eMultimodal components\u003c/p\u003e\n\u003cp\u003eNext, given the current work\u0026rsquo;s focus on multimodal neural sources, we tested the hypothesis that multimodal components performed superior to unimodal components in differentiating autistic from NAI. For this, we calculated a multimodal index (MMI) per IC to quantify the multimodal nature of modalities in each IC \u003cem\u003e(86)\u003c/em\u003e (see SI). The MMI ranges from 0 (equating to 100% unimodal contribution) to 1 (equating to equal contributions from all modalities). Multimodal components were defined as each single imaging modality (i.e., regardless of hemisphere) not having more than a 90% contribution to each component and an MMI below 0.1 (Fig. S13). Components below this threshold were treated as unimodal.\u003c/p\u003e\n\u003cp\u003eAutism classification\u003c/p\u003e\n\u003cp\u003eNext, we implemented two support vector machine (SVM) classifiers with a linear kernel \u0026ndash; one using unimodal and one using multimodal components as features to test for the added value of multimodal features. The SVM was trained and evaluated using 10-fold cross-validation. Class-weighting was used to account for group size imbalance. The area under the receiver operating characteristic curve (AUC) was used as the performance metric to assess the classifier's discrimination ability. To test for significant differences in AUC between multimodal and unimodal components, we generated a null distribution of AUC differences by shuffling the cross-validated scores 10,000 times and re-evaluating the classifier performance and computed the likelihood of observing the observed AUC difference under the null hypothesis. To test for robustness of results, we re-ran analyses across different thresholds resulting in slightly varying degrees of multimodality ranging between 85\u0026ndash;99% of single modality contributions. Given that each threshold resulted in a different number of unimodal vs. multimodal components, we further checked whether results remained stable when forcing uni- and multimodal components to have the same number of features (see SI).\u003c/p\u003e\n\u003cp\u003eClinical-cognitive associations\u003c/p\u003e\n\u003cp\u003eTo test for the clinical relevance of multimodal ICs, we ran canonical correlation analyses (CCA) \u003cem\u003e(88)\u003c/em\u003e modelling the multivariate relationship between multimodal ICs and cognitive, clinical features related to either \u003cem\u003ea)\u003c/em\u003e social-communicative features related to social functioning and face processing in autism or \u003cem\u003eb)\u003c/em\u003e non-social features associated with autism. The statistical significance of the CCA modes was assessed by a complete permutation inference algorithm \u003cem\u003e(89)\u003c/em\u003e, where both brain and behaviour data were permuted separately across all participants with 10,000 iterations (see SI).\u003c/p\u003e\n\u003cp\u003eTo visualize the spatial and temporal patterns of each imaging modality associated with each clinical cognitive measure, we computed the correlations between the original imaging data (i.e., the Z-deviation maps) and the canonical imaging variate (V) derived from the CCA \u003cem\u003e(90)\u003c/em\u003e. Significance of correlation maps was assessed with 1000 permutations and significant clusters/timepoints were next visualized and characterized in terms of their functional and anatomical overlap with the VIS-atlas \u003cem\u003e(9)\u003c/em\u003e (Fig.\u0026nbsp;2h) and the HOA atlas (Fig.\u0026nbsp;2m) (see SI).\u003c/p\u003e\n\u003cp\u003eTo assess robustness of CCA results, we set a range of multimodal thresholds between 85\u0026ndash;99% and selected ICs with modality contributions exceeding this threshold. We then re-ran the CCA for each threshold to assess stability of results across varying degrees of multimodality.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u003c/strong\u003e We thank all participants and their families for participating in the studies that contribute to the datasets used in this research. We also gratefully acknowledge the contributions of all members of the EU-AIMS/AIMS-2-TRIALS LEAP group: Jumana Ahmad, Sara Ambrosino, Bonnie Auyeung, Sarah Baumeister, Sven B\u0026ouml;lte, Carsten Bours, Michael Brammer, Daniel Brandeis, Claudia Brogna, Yvette de Bruijn, Bhismadev Chakrabarti, Ineke Cornelissen, Daisy Crawley, Guillaume Dumas, Jessica Faulkner, Vincent Frouin, Pilar Garc\u0026eacute;s, David Goyard, Lindsay Ham, Hannah Hayward, Joerg Hipp, Mark H. Johnson, Emily J.H. Jones, Xavier Liogier D\u0026rsquo;ardhuy, David J. Lythgoe, Ren\u0026eacute; Mandl, Luke Mason, Andreas Meyer-Lindenberg, Nico Mueller, Bethany Oakley, Laurence O\u0026rsquo;Dwyer, Bob Oranje, Gahan Pandina, Antonio M. Persico, Barbara Ruggeri, Amber Ruigrok, Jessica Sabet, Roberto Sacco, Antonia San Jos\u0026eacute; C\u0026aacute;ceres, Emily Simonoff, Will Spooren, Roberto Toro, Heike Tost, Jack Waldman, Steve C.R. Williams, Caroline Wooldridge, and Marcel P. Zwiers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e: This project has received funding from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No 115300 (for EU-AIMS) and No 777394 (for AIMS-2-TRIALS). This Joint Undertaking receives support from the European Union\u0026apos;s Horizon 2020 research and innovation programme and EFPIA and AUTISM SPEAKS, Autistica, SFARI. Any views expressed are those of the author(s) and not necessarily those of the funders (IHI-JU2). This work was also supported by the Netherlands Organization for Scientific Research through grants (Grant No. 864.12.003 [to CFB]; from the FP7 (Grant Nos. 602805) (AGGRESSOTYPE) (to JKB), 603016 (MATRICS), and 278948 (TACTICS); and from the European Community\u0026rsquo;s Horizon 2020 Programme (H2020/2014-2020) (Grant Nos. 643051 [MiND] and 642996 (BRAINVIEW). This work received funding from the Wellcome Trust UK Strategic Award (Award No. 098369/Z/12/Z) and from the National Institute for Health Research Maudsley Biomedical Research Centre (to DM). DLF is supported by the UZH postdoc grant and funding from the European Union\u0026rsquo;s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 101025785. EJHJ and RH received funding from SFARI GAIINS (grant number 10039678). SB-C is funded by the Autism Research Trust, the Wellcome Trust, the Templeton World Charitable Foundation and by the NIHR Biomedical Research Centre in Cambridge, during the period of this work. BHV is supported by the Swiss National Science Foundation [10001C_197480].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e: D.L.F.: study design, data preprocessing and analysis, wrote the paper. A.L., T.M., B.H.V. and N.F.: supervision and advise on linked independent component analysis and interpretation. M.Z., N.E.H. and C.E.: supervision and advise on normative modeling analysis and interpretation. C.M. and S.B.: fMRI processing. E.J.H.J., L.M. and R.H.: EEG data preprocessing. C.M.P: analysis direction and interpretation. T.C.: clinical/design of study/interpretation. F.D.A., S.D., T.B., R.J.H., S.B.-C., T.B., E.L., D.G.M.M., J.K.B. and C.F.B.: secured funding and supervised the study overall and analysis direction and interpretation. N.L.: study design and supervision of overall analytic strategy. All authors revised the manuscript for intellectual content.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e: JKB has been a consultant to, advisory board member of, and a speaker for Takeda/Shire, Medice, Roche, and Servier. He is not an employee of any of these companies and not a stock shareholder of any of these companies. He has no other financial or material support, including expert testimony, patents, or royalties. CFB is director and shareholder in SBGneuro Ltd. TC has received consultancy from Roche and Servier and received book royalties from Guildford Press and Sage. TB served in an advisory or consultancy role for ADHS digital, Infectopharm, Lundbeck, Medice, Neurim Pharmaceuticals, Oberberg GmbH, Roche, and Takeda. He received conference support or speaker\u0026rsquo;s fee by Medice and Takeda. He received royalities from Hogrefe, Kohlhammer, CIP Medien, Oxford University Press; the present work is unrelated to these relationships. The other authors report no biomedical financial interests or potential conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData and materials availability\u003c/strong\u003e: LEAP data are shared through an application process codesigned with autistic people that preserves security and privacy\u0026mdash;contact the corresponding author for further details and application forms. Scripts used to implement the experimental task are covered by a material transfer agreement, which can be obtained through the corresponding author. All the data associated with this study are present in the paper or the Supplementary Materials.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eM. J. Maenner, K. A. Shaw, A. V. Bakian, D. A. Bilder, M. S. Durkin, A. Esler, S. M. Furnier, L. Hallas, J. Hall-Lande, A. Hudson, M. M. Hughes, M. Patrick, K. Pierce, J. N. Poynter, A. Salinas, J. Shenouda, A. Vehorn, Z. Warren, J. N. Constantino, M. DiRienzo, R. T. Fitzgerald, A. Grzybowski, M. H. Spivey, S. Pettygrove, W. Zahorodny, A. Ali, J. G. Andrews, T. Baroud, J. Gutierrez, A. Hewitt, L.-C. Lee, M. Lopez, K. C. Mancilla, D. McArthur, Y. D. Schwenk, A. Washington, S. Williams, M. E. Cogswell, Prevalence and Characteristics of Autism Spectrum Disorder Among Children Aged 8 Years - Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2018. \u003cem\u003eMMWR Surveill Summ\u003c/em\u003e \u003cstrong\u003e70\u003c/strong\u003e, 1\u0026ndash;16 (2021).\u003c/li\u003e\n\u003cli\u003eS. M. Kanne, A. J. Gerber, L. M. Quirmbach, S. S. Sparrow, D. V. Cicchetti, C. A. Saulnier, The role of adaptive behavior in autism spectrum disorders: Implications for functional outcome. \u003cem\u003eJournal of Autism and Developmental Disorders\u003c/em\u003e \u003cstrong\u003e41\u003c/strong\u003e (2011), doi:10.1007/s10803-010-1126-4.\u003c/li\u003e\n\u003cli\u003eG. Dawson, S. J. Webb, J. McPartland, Understanding the nature of face processing impairment in autism: insights from behavioral and electrophysiological studies. \u003cem\u003eDev Neuropsychol\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 403\u0026ndash;424 (2005).\u003c/li\u003e\n\u003cli\u003eN. J. Sasson, The development of face processing in autism. \u003cem\u003eJ Autism Dev Disord\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 381\u0026ndash;394 (2006).\u003c/li\u003e\n\u003cli\u003eH. Meyer-Lindenberg, C. Moessnang, B. Oakley, J. Ahmad, L. Mason, E. J. H. Jones, H. L. Hayward, J. Cooke, D. Crawley, R. Holt, J. Tillmann, T. Charman, S. Baron-Cohen, T. Banaschewski, C. Beckmann, H. Tost, A. Meyer-Lindenberg, J. K. Buitelaar, D. G. Murphy, M. J. Brammer, E. Loth, Facial expression recognition is linked to clinical and neurofunctional differences in autism. \u003cem\u003eMolecular Autism\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 43 (2022).\u003c/li\u003e\n\u003cli\u003eC. Chevallier, P. Huguet, F. Happ\u0026eacute;, N. George, L. Conty, Salient Social Cues are Prioritized in Autism Spectrum Disorders Despite Overall Decrease in Social Attention. \u003cem\u003eJ Autism Dev Disord\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 1642\u0026ndash;1651 (2013).\u003c/li\u003e\n\u003cli\u003eR. T. Schultz, Developmental deficits in social perception in autism: the role of the amygdala and fusiform face area. \u003cem\u003eInternational Journal of Developmental Neuroscience\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 125\u0026ndash;141 (2005).\u003c/li\u003e\n\u003cli\u003eJ. Sui, T. Adali, Q. Yu, J. Chen, V. D. Calhoun, A review of multivariate methods for multimodal fusion of brain imaging data. \u003cem\u003eJournal of Neuroscience Methods\u003c/em\u003e \u003cstrong\u003e204\u003c/strong\u003e, 68\u0026ndash;81 (2012).\u003c/li\u003e\n\u003cli\u003eM. Rosenke, R. van Hoof, J. van den Hurk, K. Grill-Spector, R. Goebel, A Probabilistic Functional Atlas of Human Occipito-Temporal Visual Cortex. \u003cem\u003eCerebral Cortex\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 603\u0026ndash;619 (2021).\u003c/li\u003e\n\u003cli\u003eW. Zhang, J. Wang, L. Fan, Y. Zhang, P. T. Fox, S. B. Eickhoff, C. Yu, T. Jiang, Functional organization of the fusiform gyrus revealed with connectivity profiles. \u003cem\u003eHum Brain Mapp\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 3003\u0026ndash;3016 (2016).\u003c/li\u003e\n\u003cli\u003eI. Gauthier, M. J. Tarr, J. Moylan, P. Skudlarski, J. C. Gore, A. W. Anderson, The fusiform \u0026ldquo;face area\u0026rdquo; is part of a network that processes faces at the individual level. \u003cem\u003eJournal of Cognitive Neuroscience\u003c/em\u003e (2000), doi:10.1162/089892900562165.\u003c/li\u003e\n\u003cli\u003eJ. C. Taylor, A. J. Wiggett, P. E. Downing, Functional MRI Analysis of Body and Body Part Representations in the Extrastriate and Fusiform Body Areas. \u003cem\u003eJournal of Neurophysiology\u003c/em\u003e \u003cstrong\u003e98\u003c/strong\u003e, 1626\u0026ndash;1633 (2007).\u003c/li\u003e\n\u003cli\u003eX. Wang, Z. Han, Y. He, A. Caramazza, L. Song, Y. Bi, Where color rests: Spontaneous brain activity of bilateral fusiform and lingual regions predicts object color knowledge performance. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e76\u003c/strong\u003e, 252\u0026ndash;263 (2013).\u003c/li\u003e\n\u003cli\u003eG. McCarthy, A. Puce, J. C. Gore, T. Allison, Face-Specific Processing in the Human Fusiform Gyrus. \u003cem\u003eJournal of Cognitive Neuroscience\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 605\u0026ndash;610 (1997).\u003c/li\u003e\n\u003cli\u003eN. Kanwisher, G. Yovel, The fusiform face area: a cortical region specialized for the perception of faces. \u003cem\u003ePhilosophical Transactions of the Royal Society B: Biological Sciences\u003c/em\u003e \u003cstrong\u003e361\u003c/strong\u003e, 2109\u0026ndash;2128 (2006).\u003c/li\u003e\n\u003cli\u003eS. Bentin, T. Allison, A. Puce, E. Perez, G. McCarthy, Electrophysiological Studies of Face Perception in Humans. \u003cem\u003eJournal of Cognitive Neuroscience\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 551\u0026ndash;565 (1996).\u003c/li\u003e\n\u003cli\u003eK. B\u0026ouml;tzel, S. Schulze, S. R. Stodieck, Scalp topography and analysis of intracranial sources of face-evoked potentials. \u003cem\u003eExp Brain Res\u003c/em\u003e \u003cstrong\u003e104\u003c/strong\u003e, 135\u0026ndash;143 (1995).\u003c/li\u003e\n\u003cli\u003eB. Rossion, L. Dricot, A. Devolder, J.-M. Bodart, M. Crommelinck, B. de Gelder, R. Zoontjes, Hemispheric Asymmetries for Whole-Based and Part-Based Face Processing in the Human Fusiform Gyrus. \u003cem\u003eJournal of Cognitive Neuroscience\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 793\u0026ndash;802 (2000).\u003c/li\u003e\n\u003cli\u003eG. Yovel, A. Tambini, T. Brandman, The asymmetry of the fusiform face area is a stable individual characteristic that underlies the left-visual-field superiority for faces. \u003cem\u003eNeuropsychologia\u003c/em\u003e \u003cstrong\u003e46\u003c/strong\u003e, 3061\u0026ndash;3068 (2008).\u003c/li\u003e\n\u003cli\u003eR. C. Cardinale, P. Shih, I. Fishman, L. M. Ford, R. A. M\u0026uuml;ller, Pervasive rightward asymmetry shifts of functional networks in autism spectrum disorder. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e (2013), doi:10.1001/jamapsychiatry.2013.382.\u003c/li\u003e\n\u003cli\u003eD. L. Floris, T. Wolfers, M. Zabihi, N. E. Holz, M. P. Zwiers, T. Charman, J. Tillmann, C. Ecker, F. Dell\u0026rsquo;Acqua, T. Banaschewski, C. Moessnang, S. Baron-Cohen, R. Holt, S. Durston, E. Loth, D. G. M. Murphy, A. Marquand, J. K. Buitelaar, C. F. Beckmann, Atypical Brain Asymmetry in Autism\u0026mdash;A Candidate for Clinically Meaningful Stratification. \u003cem\u003eBiological Psychiatry: Cognitive Neuroscience and Neuroimaging\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 802\u0026ndash;812 (2021).\u003c/li\u003e\n\u003cli\u003eT. Nickl-Jockschat, C. Rottschy, J. Thommes, F. Schneider, A. R. Laird, P. T. Fox, S. B. Eickhoff, Neural networks related to dysfunctional face processing in autism spectrum disorder. \u003cem\u003eBrain Struct Funct\u003c/em\u003e \u003cstrong\u003e220\u003c/strong\u003e, 2355\u0026ndash;2371 (2015).\u003c/li\u003e\n\u003cli\u003eK. Pierce, F. Haist, F. Sedaghat, E. Courchesne, The brain response to personally familiar faces in autism: findings of fusiform activity and beyond. \u003cem\u003eBrain\u003c/em\u003e \u003cstrong\u003e127\u003c/strong\u003e, 2703\u0026ndash;2716 (2004).\u003c/li\u003e\n\u003cli\u003eN. M. Kleinhans, T. Richards, L. Sterling, K. C. Stegbauer, R. Mahurin, L. C. Johnson, J. Greenson, G. Dawson, E. Aylward, Abnormal functional connectivity in autism spectrum disorders during face processing. \u003cem\u003eBrain\u003c/em\u003e \u003cstrong\u003e131\u003c/strong\u003e, 1000\u0026ndash;1012 (2008).\u003c/li\u003e\n\u003cli\u003eH. Koshino, R. K. Kana, T. A. Keller, V. L. Cherkassky, N. J. Minshew, M. A. Just, fMRI Investigation of Working Memory for Faces in Autism: Visual Coding and Underconnectivity with Frontal Areas. \u003cem\u003eCerebral Cortex\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 289\u0026ndash;300 (2008).\u003c/li\u003e\n\u003cli\u003eE. Kang, C. M. Keifer, E. J. Levy, J. H. Foss-Feig, J. C. McPartland, M. D. Lerner, Atypicality of the N170 Event-Related Potential in Autism Spectrum Disorder: A Meta-analysis. \u003cem\u003eBiological Psychiatry: Cognitive Neuroscience and Neuroimaging\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 657\u0026ndash;666 (2018).\u003c/li\u003e\n\u003cli\u003eL. Mason, C. Moessnang, C. Chatham, L. Ham, J. Tillmann, G. Dumas, C. Ellis, C. S. Leblond, F. Cliquet, T. Bourgeron, C. Beckmann, T. Charman, B. Oakley, T. Banaschewski, A. Meyer-Lindenberg, S. Baron-Cohen, S. B\u0026ouml;lte, J. K. Buitelaar, S. Durston, E. Loth, B. Oranje, A. Persico, F. Dell\u0026rsquo;Acqua, C. Ecker, M. H. Johnson, D. Murphy, E. J. H. Jones, Stratifying the autistic phenotype using electrophysiological indices of social perception. \u003cem\u003eScience Translational Medicine\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, eabf8987 (2022).\u003c/li\u003e\n\u003cli\u003eG. D. Waiter, J. H. G. Williams, A. D. Murray, A. Gilchrist, D. I. Perrett, A. Whiten, A voxel-based investigation of brain structure in male adolescents with autistic spectrum disorder. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e (2004), doi:10.1016/j.neuroimage.2004.02.029.\u003c/li\u003e\n\u003cli\u003eI. A. J. Van Kooten, S. J. M. C. Palmen, P. Von Cappeln, H. W. M. Steinbusch, H. Korr, H. Heinsen, P. R. Hof, H. Van Engeland, C. Schmitz, Neurons in the fusiform gyrus are fewer and smaller in autism. \u003cem\u003eBrain\u003c/em\u003e \u003cstrong\u003e131\u003c/strong\u003e (2008), doi:10.1093/brain/awn033.\u003c/li\u003e\n\u003cli\u003eM. R. Herbert, G. J. Harris, K. T. Adrien, D. a Ziegler, N. Makris, D. N. Kennedy, N. T. Lange, C. F. Chabris, A. Bakardjiev, J. Hodgson, M. Takeoka, H. Tager-Flusberg, V. S. Caviness, Abnormal asymmetry in language association cortex in autism. \u003cem\u003eAnnals of neurology\u003c/em\u003e \u003cstrong\u003e52\u003c/strong\u003e, 588\u0026ndash;96 (2002).\u003c/li\u003e\n\u003cli\u003eE. Neuhaus, E. J. H. Jones, K. Barnes, L. Sterling, A. Estes, J. Munson, G. Dawson, S. J. Webb, The Relationship Between Early Neural Responses to Emotional Faces at Age 3 and Later Autism and Anxiety Symptoms in Adolescents with Autism. \u003cem\u003eJ Autism Dev Disord\u003c/em\u003e \u003cstrong\u003e46\u003c/strong\u003e, 2450\u0026ndash;2463 (2016).\u003c/li\u003e\n\u003cli\u003eS. J. Webb, E. J. H. Jones, K. Merkle, K. Venema, J. Greenson, M. Murias, G. Dawson, Developmental change in the ERP responses to familiar faces in toddlers with autism spectrum disorders versus typical development. \u003cem\u003eChild Dev\u003c/em\u003e \u003cstrong\u003e82\u003c/strong\u003e, 1868\u0026ndash;1886 (2011).\u003c/li\u003e\n\u003cli\u003eC. Gao, S. Conte, J. E. Richards, W. Xie, T. Hanayik, The neural sources of N170: Understanding timing of activation in face-selective areas. \u003cem\u003ePsychophysiology\u003c/em\u003e \u003cstrong\u003e56\u003c/strong\u003e, e13336 (2019).\u003c/li\u003e\n\u003cli\u003eC. C. Dougherty, D. W. Evans, G. J. Katuwal, A. M. Michael, Asymmetry of fusiform structure in autism spectrum disorder: Trajectory and association with symptom severity. \u003cem\u003eMolecular Autism\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e (2016), doi:10.1186/s13229-016-0089-5.\u003c/li\u003e\n\u003cli\u003eT. Iidaka, A. Matsumoto, K. Haneda, T. Okada, N. Sadato, Hemodynamic and electrophysiological relationship involved in human face processing: Evidence from a combined fMRI\u0026ndash;ERP study. \u003cem\u003eBrain and Cognition\u003c/em\u003e \u003cstrong\u003e60\u003c/strong\u003e, 176\u0026ndash;186 (2006).\u003c/li\u003e\n\u003cli\u003eA. Puce, A. Syngeniotis, J. C. Thompson, D. F. Abbott, K. J. Wheaton, U. Castiello, The human temporal lobe integrates facial form and motion: evidence from fMRI and ERP studies. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 861\u0026ndash;869 (2003).\u003c/li\u003e\n\u003cli\u003eN. E. Holz, D. L. Floris, A. Llera, P. M. Aggensteiner, S. M. Kia, T. Wolfers, S. Baumeister, B. B\u0026ouml;ttinger, J. C. Glennon, P. J. Hoekstra, A. Dietrich, M. C. Saam, U. M. E. Schulze, D. J. Lythgoe, S. C. R. Williams, P. Santosh, M. Rosa-Justicia, N. Bargallo, J. Castro-Fornieles, C. Arango, M. J. Penzol, S. Walitza, A. Meyer-Lindenberg, M. Zwiers, B. Franke, J. Buitelaar, J. Naaijen, D. Brandeis, C. Beckmann, T. Banaschewski, A. F. Marquand, Age-related brain deviations and aggression. \u003cem\u003ePsychol Med\u003c/em\u003e , 1\u0026ndash;10 (2022).\u003c/li\u003e\n\u003cli\u003eT. T. Brown, J. M. Kuperman, Y. Chung, M. Erhart, C. McCabe, D. J. Hagler, V. K. Venkatraman, N. Akshoomoff, D. G. Amaral, C. S. Bloss, B. J. Casey, L. Chang, T. M. Ernst, J. A. Frazier, J. R. Gruen, W. E. Kaufmann, T. Kenet, D. N. Kennedy, S. S. Murray, E. R. Sowell, T. L. Jernigan, A. M. Dale, Neuroanatomical assessment of biological maturity. \u003cem\u003eCurrent Biology\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e (2012), doi:10.1016/j.cub.2012.07.002.\u003c/li\u003e\n\u003cli\u003eG. Erus, H. Battapady, T. D. Satterthwaite, H. Hakonarson, R. E. Gur, C. Davatzikos, R. C. Gur, Imaging patterns of brain development and their relationship to cognition. \u003cem\u003eCerebral Cortex\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e (2015), doi:10.1093/cercor/bht425.\u003c/li\u003e\n\u003cli\u003eF. Liem, G. Varoquaux, J. Kynast, F. Beyer, S. Kharabian Masouleh, J. M. Huntenburg, L. Lampe, M. Rahim, A. Abraham, R. C. Craddock, S. Riedel-Heller, T. Luck, M. Loeffler, M. L. Schroeter, A. V. Witte, A. Villringer, D. S. Margulies, Predicting brain-age from multimodal imaging data captures cognitive impairment. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e148\u003c/strong\u003e (2017), doi:10.1016/j.neuroimage.2016.11.005.\u003c/li\u003e\n\u003cli\u003eT. Mei, N. J. Forde, D. L. Floris, F. Dell\u0026rsquo;Acqua, R. Stones, I. Ilioska, S. Durston, C. Moessnang, T. Banaschewski, R. J. Holt, S. Baron-Cohen, A. Rausch, E. Loth, B. Oakley, T. Charman, C. Ecker, D. G. M. Murphy, EU-AIMS LEAP group, C. F. Beckmann, A. Llera, J. K. Buitelaar, Autism Is Associated With Interindividual Variations of Gray and White Matter Morphology. \u003cem\u003eBiol Psychiatry Cogn Neurosci Neuroimaging\u003c/em\u003e , S2451-9022(22)00212\u0026ndash;9 (2022).\u003c/li\u003e\n\u003cli\u003eL. M. Oblong, A. Llera, T. Mei, K. Haak, C. Isakoglou, D. L. Floris, S. Durston, C. Moessnang, T. Banaschewski, S. Baron-Cohen, E. Loth, F. Dell\u0026rsquo;Acqua, T. Charman, D. G. M. Murphy, C. Ecker, J. K. Buitelaar, C. F. Beckmann, The EU-AIMS LEAP Group, Linking functional and structural brain organisation with behaviour in autism: a multimodal EU-AIMS Longitudinal European Autism Project (LEAP) study. \u003cem\u003eMolecular Autism\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 32 (2023).\u003c/li\u003e\n\u003cli\u003eT. Charman, E. Loth, J. Tillmann, D. Crawley, C. Wooldridge, D. Goyard, J. Ahmad, B. Auyeung, S. Ambrosino, T. Banaschewski, S. Baron-Cohen, S. Baumeister, C. Beckmann, S. B\u0026ouml;lte, T. Bourgeron, C. Bours, M. Brammer, D. Brandeis, C. Brogna, Y. De Bruijn, B. Chakrabarti, I. Cornelissen, F. D. Acqua, G. Dumas, S. Durston, C. Ecker, J. Faulkner, V. Frouin, P. Garc\u0026eacute;s, L. Ham, H. Hayward, J. Hipp, R. J. Holt, J. Isaksson, M. H. Johnson, E. J. H. Jones, P. Kundu, M. C. Lai, X. L. D\u0026rsquo;Ardhuy, M. V. Lombardo, D. J. Lythgoe, R. Mandl, L. Mason, A. Meyer-Lindenberg, C. Moessnang, N. Mueller, L. O\u0026rsquo;Dwyer, M. Oldehinkel, B. Oranje, G. Pandina, A. M. Persico, B. Ruggeri, A. N. V. Ruigrok, J. Sabet, R. Sacco, A. S. J. C\u0026aacute;ceres, E. Simonoff, R. Toro, H. Tost, J. Waldman, S. C. R. Williams, M. P. Zwiers, W. Spooren, D. G. M. Murphy, J. K. Buitelaar, The EU-AIMS Longitudinal European Autism Project (LEAP): Clinical characterisation. \u003cem\u003eMolecular Autism\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2017), doi:10.1186/s13229-017-0145-9.\u003c/li\u003e\n\u003cli\u003eE. Loth, T. Charman, L. Mason, J. Tillmann, E. J. H. Jones, C. Wooldridge, J. Ahmad, B. Auyeung, C. Brogna, S. Ambrosino, T. Banaschewski, S. Baron-Cohen, S. Baumeister, C. Beckmann, M. Brammer, D. Brandeis, S. B\u0026ouml;lte, T. Bourgeron, C. Bours, Y. De Bruijn, B. Chakrabarti, D. Crawley, I. Cornelissen, F. D. Acqua, G. Dumas, S. Durston, C. Ecker, J. Faulkner, V. Frouin, P. Garces, D. Goyard, H. Hayward, L. M. Ham, J. Hipp, R. J. Holt, M. H. Johnson, J. Isaksson, P. Kundu, M. C. Lai, X. L. D\u0026rsquo;Ardhuy, M. V. Lombardo, D. J. Lythgoe, R. Mandl, A. Meyer-Lindenberg, C. Moessnang, N. Mueller, L. O\u0026rsquo;Dwyer, M. Oldehinkel, B. Oranje, G. Pandina, A. M. Persico, A. N. V. Ruigrok, B. Ruggeri, J. Sabet, R. Sacco, A. S. J. C\u0026aacute;ceres, E. Simonoff, R. Toro, H. Tost, J. Waldman, S. C. R. Williams, M. P. Zwiers, W. Spooren, D. G. M. Murphy, J. K. Buitelaar, The EU-AIMS Longitudinal European Autism Project (LEAP): Design and methodologies to identify and validate stratification biomarkers for autism spectrum disorders. \u003cem\u003eMolecular Autism\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2017), doi:10.1186/s13229-017-0146-8.\u003c/li\u003e\n\u003cli\u003eA. F. Marquand, I. Rezek, J. Buitelaar, C. F. Beckmann, Understanding Heterogeneity in Clinical Cohorts Using Normative Models: Beyond Case-Control Studies. \u003cem\u003eBiological Psychiatry\u003c/em\u003e \u003cstrong\u003e80\u003c/strong\u003e, 552\u0026ndash;561 (2016).\u003c/li\u003e\n\u003cli\u003eL. Parkes, T. M. Moore, M. E. Calkins, P. A. Cook, M. Cieslak, D. R. Roalf, D. H. Wolf, R. C. Gur, R. E. Gur, T. D. Satterthwaite, D. S. Bassett, Transdiagnostic dimensions of psychopathology explain individuals\u0026rsquo; unique deviations from normative neurodevelopment in brain structure. \u003cem\u003eTransl Psychiatry\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 232 (2021).\u003c/li\u003e\n\u003cli\u003eA. R. Groves, C. F. Beckmann, S. M. Smith, M. W. Woolrich, Linked independent component analysis for multimodal data fusion. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e54\u003c/strong\u003e, 2198\u0026ndash;2217 (2011).\u003c/li\u003e\n\u003cli\u003eS. Rutherford, P. Barkema, I. F. Tso, C. Sripada, C. F. Beckmann, H. G. Ruhe, A. F. Marquand, C. I. Baker, T. Constable, O. Esteban, Eds. Evidence for embracing normative modeling. \u003cem\u003eeLife\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e85082 (2023).\u003c/li\u003e\n\u003cli\u003eA. Llera, T. Wolfers, P. Mulders, C. F. Beckmann, Inter-individual differences in human brain structure and morphology link to variation in demographics and behavior. \u003cem\u003eeLife\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2019), doi:10.7554/eLife.44443.\u003c/li\u003e\n\u003cli\u003eS. S. Sparrow, D. V. Cicchetti, D. A. Balla, in \u003cem\u003eMajor psychological assessment instruments\u003c/em\u003e, (2005).\u003c/li\u003e\n\u003cli\u003eS. Baron-Cohen, S. Wheelwright, J. Hill, Y. Raste, I. Plumb, The \u0026ldquo;Reading the Mind in the Eyes\u0026rdquo; Test revised version: A study with normal adults, and adults with Asperger syndrome or high-functioning autism. \u003cem\u003eJournal of Child Psychology and Psychiatry and Allied Disciplines\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e (2001), doi:10.1017/S0021963001006643.\u003c/li\u003e\n\u003cli\u003eA. R. Hariri, A. Tessitore, V. S. Mattay, F. Fera, D. R. Weinberger, The amygdala response to emotional stimuli: A comparison of faces and scenes. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e (2002), doi:10.1006/nimg.2002.1179.\u003c/li\u003e\n\u003cli\u003eJ. W. Bodfish, F. J. Symons, D. E. Parker, M. H. Lewis, Repetitive Behavior Scale\u0026ndash;Revised. \u003cem\u003ePsycTESTS\u0026reg;\u003c/em\u003e (2000).\u003c/li\u003e\n\u003cli\u003eS. D. Tomchek, W. Dunn, Sensory processing in children with and without autism: a comparative study using the short sensory profile. \u003cem\u003eAmerican Journal of Occupational Therapy\u003c/em\u003e \u003cstrong\u003e61\u003c/strong\u003e (2007), doi:10.5014/ajot.61.2.190.\u003c/li\u003e\n\u003cli\u003eB. Auyeung, S. Wheelwright, C. Allison, M. Atkinson, N. Samarawickrema, S. Baron-Cohen, The children\u0026rsquo;s empathy quotient and systemizing quotient: Sex differences in typical development and in autism spectrum conditions. \u003cem\u003eJournal of Autism and Developmental Disorders\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e (2009), doi:10.1007/s10803-009-0772-x.\u003c/li\u003e\n\u003cli\u003eB. Auyeung, C. Allison, S. Wheelwright, S. Baron-Cohen, Brief report: Development of the adolescent empathy and systemizing quotients. \u003cem\u003eJournal of Autism and Developmental Disorders\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e (2012), doi:10.1007/s10803-012-1454-7.\u003c/li\u003e\n\u003cli\u003eS. Baron-Cohen, J. Richler, D. Bisarya, N. Gurunathan, S. Wheelwright, The systemizing quotient: An investigation of adults with Asperger syndrome or high-functioning autism, and normal sex differences. \u003cem\u003ePhilosophical Transactions of the Royal Society B: Biological Sciences\u003c/em\u003e \u003cstrong\u003e358\u003c/strong\u003e (2003), doi:10.1098/rstb.2002.1206.\u003c/li\u003e\n\u003cli\u003eH. D. Critchley, E. M. Daly, E. T. Bullmore, S. C. Williams, T. Van Amelsvoort, D. M. Robertson, A. Rowe, M. Phillips, G. McAlonan, P. Howlin, D. G. Murphy, The functional neuroanatomy of social behaviour: changes in cerebral blood flow when people with autistic disorder process facial expressions. \u003cem\u003eBrain\u003c/em\u003e \u003cstrong\u003e123 ( Pt 11)\u003c/strong\u003e, 2203\u0026ndash;2212 (2000).\u003c/li\u003e\n\u003cli\u003eB. A. Corbett, V. Carmean, S. Ravizza, C. Wendelken, M. L. Henry, C. Carter, S. M. Rivera, A functional and structural study of emotion and face processing in children with autism. \u003cem\u003ePsychiatry Research: Neuroimaging\u003c/em\u003e \u003cstrong\u003e173\u003c/strong\u003e, 196\u0026ndash;205 (2009).\u003c/li\u003e\n\u003cli\u003eA. C. Lynn, A. Padmanabhan, D. Simmonds, W. Foran, M. N. Hallquist, B. Luna, K. O\u0026rsquo;Hearn, Functional connectivity differences in autism during face and car recognition: underconnectivity and atypical age-related changes. \u003cem\u003eDevelopmental Science\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, e12508 (2018).\u003c/li\u003e\n\u003cli\u003eS. Bentin, L. Y. Deouell, Structural Encoding and Identification in Face Processing: Erp Evidence for Separate Mechanisms. \u003cem\u003eCognitive Neuropsychology\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 35\u0026ndash;55 (2000).\u003c/li\u003e\n\u003cli\u003eD. Pitcher, V. Walsh, B. Duchaine, The role of the occipital face area in the cortical face perception network. \u003cem\u003eExp Brain Res\u003c/em\u003e \u003cstrong\u003e209\u003c/strong\u003e, 481\u0026ndash;493 (2011).\u003c/li\u003e\n\u003cli\u003eF. Happ\u0026eacute;, U. Frith, The weak coherence account: detail-focused cognitive style in autism spectrum disorders. \u003cem\u003eJournal of autism and developmental disorders\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 5\u0026ndash;25 (2006).\u003c/li\u003e\n\u003cli\u003eT. L. Watson, Implications of holistic face processing in autism and schizophrenia. \u003cem\u003eFront Psychol\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 414 (2013).\u003c/li\u003e\n\u003cli\u003eR. T. Schultz, I. Gauthier, A. Klin, R. K. Fulbright, A. W. Anderson, F. Volkmar, P. Skudlarski, C. Lacadie, D. J. Cohen, J. C. Gore, Abnormal Ventral Temporal Cortical Activity During Face Discrimination Among Individuals With Autism and Asperger Syndrome. \u003cem\u003eArchives of General Psychiatry\u003c/em\u003e \u003cstrong\u003e57\u003c/strong\u003e, 331\u0026ndash;340 (2000).\u003c/li\u003e\n\u003cli\u003eK. Pierce, R. A. M\u0026uuml;ller, J. Ambrose, G. Allen, E. Courchesne, Face processing occurs outside the fusiform \u0026ldquo;face area\u0026rdquo; in autism: Evidence from functional MRI. \u003cem\u003eBrain\u003c/em\u003e \u003cstrong\u003e124\u003c/strong\u003e (2001), doi:10.1093/brain/124.10.2059.\u003c/li\u003e\n\u003cli\u003eS. R. Schweinberger, E. C. Pickering, I. Jentzsch, A. M. Burton, J. M. Kaufmann, Event-related brain potential evidence for a response of inferior temporal cortex to familiar face repetitions. \u003cem\u003eCognitive Brain Research\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 398\u0026ndash;409 (2002).\u003c/li\u003e\n\u003cli\u003eS. R. Schweinberger, V. Huddy, A. M. Burton, N250r: a face-selective brain response to stimulus repetitions. \u003cem\u003eNeuroReport\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 1501 (2004).\u003c/li\u003e\n\u003cli\u003eH. M. Morgan, C. Klein, S. G. Boehm, K. L. Shapiro, D. E. J. Linden, Working memory load for faces modulates P300, N170, and N250r. \u003cem\u003eJ Cogn Neurosci\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 989\u0026ndash;1002 (2008).\u003c/li\u003e\n\u003cli\u003eA. R. Groves, S. M. Smith, A. M. Fjell, C. K. Tamnes, K. B. Walhovd, G. Douaud, M. W. Woolrich, L. T. Westlye, Benefits of multi-modal fusion analysis on a large-scale dataset: Life-span patterns of inter-subject variability in cortical morphometry and white matter microstructure. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e63\u003c/strong\u003e (2012), doi:10.1016/j.neuroimage.2012.06.038.\u003c/li\u003e\n\u003cli\u003eJ. Caspers, K. Zilles, K. Amunts, A. R. Laird, P. T. Fox, S. B. Eickhoff, Functional characterization and differential coactivation patterns of two cytoarchitectonic visual areas on the human posterior fusiform gyrus. \u003cem\u003eHuman Brain Mapping\u003c/em\u003e \u003cstrong\u003e35\u003c/strong\u003e, 2754\u0026ndash;2767 (2014).\u003c/li\u003e\n\u003cli\u003eD. L. Floris, M.-C. Lai, T. Auer, M. V. Lombardo, C. Ecker, B. Chakrabarti, S. J. Wheelwright, E. T. Bullmore, D. G. M. Murphy, S. Baron‐Cohen, J. Suckling, Atypically rightward cerebral asymmetry in male adults with autism stratifies individuals with and without language delay. \u003cem\u003eHuman Brain Mapping\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 230\u0026ndash;253 (2016).\u003c/li\u003e\n\u003cli\u003eV. D. Calhoun, J. Sui, Multimodal Fusion of Brain Imaging Data: A Key to Finding the Missing Link(s) in Complex Mental Illness. \u003cem\u003eBiological Psychiatry: Cognitive Neuroscience and Neuroimaging\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 230\u0026ndash;244 (2016).\u003c/li\u003e\n\u003cli\u003eY. Aoki, S. Cortese, M. Tansella, Neural bases of atypical emotional face processing in autism: A meta-analysis of fMRI studies. \u003cem\u003eWorld Journal of Biological Psychiatry\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e (2015), doi:10.3109/15622975.2014.957719.\u003c/li\u003e\n\u003cli\u003eJ. V. Haxby, E. A. Hoffman, M. I. Gobbini, The distributed human neural system for face perception. \u003cem\u003eTrends Cogn Sci\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 223\u0026ndash;233 (2000).\u003c/li\u003e\n\u003cli\u003eB. Rossion, R. Caldara, M. Seghier, A. M. Schuller, F. Lazeyras, E. Mayer, A network of occipito-temporal face-sensitive areas besides the right middle fusiform gyrus is necessary for normal face processing. \u003cem\u003eBrain\u003c/em\u003e (2003), doi:10.1093/brain/awg241.\u003c/li\u003e\n\u003cli\u003eJ. J. S. Barton, D. Z. Press, J. P. Keenan, M. O\u0026rsquo;Connor, Lesions of the fusiform face area impair perception of facial configuration in prosopagnosia. \u003cem\u003eNeurology\u003c/em\u003e \u003cstrong\u003e58\u003c/strong\u003e, 71\u0026ndash;78 (2002).\u003c/li\u003e\n\u003cli\u003eC. Lord, M. Rutter, A. Le Couteur, Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. \u003cem\u003eJournal of autism and developmental disorders\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 659\u0026ndash;685 (1994).\u003c/li\u003e\n\u003cli\u003eC. Lord, S. Risi, L. Lambrecht, E. H. Cook, B. L. Leventhal, P. C. DiLavore, A. Pickles, M. Rutter, The autism diagnostic observation schedule-generic: a standard measure of social and communication deficits associated with the spectrum of autism. \u003cem\u003eJournal of autism and developmental disorders\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 205\u0026ndash;223 (2000).\u003c/li\u003e\n\u003cli\u003eA. Llera, M. Brammer, B. Oakley, J. Tillmann, M. Zabihi, J. S. Amelink, T. Mei, T. Charman, C. Ecker, F. Dell\u0026rsquo;Acqua, T. Banaschewski, C. Moessnang, S. Baron-Cohen, R. Holt, S. Durston, D. Murphy, E. Loth, J. K. Buitelaar, D. L. Floris, C. F. Beckmann, Evaluation of data imputation strategies in complex, deeply-phenotyped data sets: the case of the EU-AIMS Longitudinal European Autism Project. \u003cem\u003eBMC Medical Research Methodology\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 229 (2022).\u003c/li\u003e\n\u003cli\u003eT. Looden, D. L. Floris, A. Llera, R. J. Chauvin, T. Charman, T. Banaschewski, D. Murphy, Andre. F. Marquand, J. K. Buitelaar, C. F. Beckmann, the AIMS-2-TRIALS group, Patterns of connectome variability in autism across five functional activation tasks: findings from the LEAP project. \u003cem\u003eMolecular Autism\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 53 (2022).\u003c/li\u003e\n\u003cli\u003eA. F. Marquand, S. M. Kia, M. Zabihi, T. Wolfers, J. K. Buitelaar, C. F. Beckmann, Conceptualizing mental disorders as deviations from normative functioning. \u003cem\u003eMolecular Psychiatry\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 1415\u0026ndash;1424 (2019).\u003c/li\u003e\n\u003cli\u003eC. J. Fraza, R. Dinga, C. F. Beckmann, A. F. Marquand, Warped Bayesian linear regression for normative modelling of big data. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e245\u003c/strong\u003e, 118715 (2021).\u003c/li\u003e\n\u003cli\u003eN. E. Holz, M. Zabihi, S. M. Kia, M. Monninger, P.-M. Aggensteiner, S. Siehl, D. L. Floris, A. L. W. Bokde, S. Desrivi\u0026egrave;res, H. Flor, A. Grigis, H. Garavan, P. Gowland, A. Heinz, R. Br\u0026uuml;hl, J.-L. Martinot, M.-L. P. Martinot, D. P. Orfanos, T. Paus, L. Poustka, J. H. Fr\u0026ouml;hner, M. N. Smolka, N. Vaidya, H. Walter, R. Whelan, G. Schumann, A. Meyer-Lindenberg, D. Brandeis, J. K. Buitelaar, F. Nees, C. Beckmann, T. Banaschewski, A. F. Marquand, A stable and replicable neural signature of lifespan adversity in the adult brain. \u003cem\u003eNat Neurosci\u003c/em\u003e , 1\u0026ndash;10 (2023).\u003c/li\u003e\n\u003cli\u003eT. Wolfers, N. T. Doan, T. Kaufmann, D. Aln\u0026aelig;s, T. Moberget, I. Agartz, J. K. Buitelaar, T. Ueland, I. Melle, B. Franke, O. A. Andreassen, C. F. Beckmann, L. T. Westlye, A. F. Marquand, Mapping the Heterogeneous Phenotype of Schizophrenia and Bipolar Disorder Using Normative Models. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e \u003cstrong\u003e75\u003c/strong\u003e, 1146\u0026ndash;1155 (2018).\u003c/li\u003e\n\u003cli\u003eW. Francx, A. Llera, M. Mennes, M. P. Zwiers, S. V. Faraone, J. Oosterlaan, D. Heslenfeld, P. J. Hoekstra, C. A. Hartman, B. Franke, J. K. Buitelaar, C. F. Beckmann, Integrated analysis of gray and white matter alterations in attention-deficit/hyperactivity disorder. \u003cem\u003eNeuroImage: Clinical\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 357\u0026ndash;367 (2016).\u003c/li\u003e\n\u003cli\u003eY. Benjamini, Y. Hochberg, On the adaptive control of the false discovery rate in multiple testing with independent statistics. \u003cem\u003eJournal of Educational and Behavioral Statistics\u003c/em\u003e (2000), doi:10.3102/10769986025001060.\u003c/li\u003e\n\u003cli\u003eH. Hotelling, Relations Between Two Sets of Variates. \u003cem\u003eBiometrika\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 321\u0026ndash;377 (1936).\u003c/li\u003e\n\u003cli\u003eA. M. Winkler, O. Renaud, S. M. Smith, T. E. Nichols, Permutation inference for canonical correlation analysis. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e220\u003c/strong\u003e (2020), doi:10.1016/j.neuroimage.2020.117065.\u003c/li\u003e\n\u003cli\u003eG. Ball, P. Aljabar, P. Nongena, N. Kennea, N. Gonzalez-Cinca, S. Falconer, A. T. M. Chew, N. Harper, J. Wurie, M. A. Rutherford, S. J. Counsell, A. D. Edwards, Multimodal image analysis of clinical influences on preterm brain development. \u003cem\u003eAnnals of Neurology\u003c/em\u003e \u003cstrong\u003e82\u003c/strong\u003e, 233\u0026ndash;246 (2017).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Table","content":" \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eDemographic, clinical and imaging-related information of the sample.\u003c/span\u003e Abbreviations: NAI\u0026thinsp;=\u0026thinsp;non-autistic individuals; FIQ\u0026thinsp;=\u0026thinsp;full-scale IQ; VIQ\u0026thinsp;=\u0026thinsp;verbal IQ; PIQ\u0026thinsp;=\u0026thinsp;performance IQ; ADI\u0026thinsp;=\u0026thinsp;Autism Diagnostic Interview; ADOS\u0026thinsp;=\u0026thinsp;Autism Diagnosis Observation Schedule; QC\u0026thinsp;=\u0026thinsp;quality control; FD\u0026thinsp;=\u0026thinsp;framewise displacement.\u003c/div\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003eVariable\u003c/div\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003eAutism\u003c/div\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003eNAI\u003c/div\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003epost hoc\u003c/div\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eN\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e99\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e105\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003esex\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e71M : 28F\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e71M : 34F\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003eχ2\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.23, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.63\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003emean\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003estd\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003erange\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003emean\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003estd\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003erange\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eage\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e18.62\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e5.32\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e7.58-30.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e18.3\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e4.87\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e10.37\u0026ndash;30.9\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003et\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.44, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;066\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eFIQ\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e107\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e14.3\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e63\u0026ndash;148\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e106\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e12.3\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e77\u0026ndash;142\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003et\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.08, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.94\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eVIQ\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e106\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e15.62\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e51\u0026ndash;160\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e107\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e14.1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e74\u0026ndash;142\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003et\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.09, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.93\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003ePIQ\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e107\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e16.52\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e57\u0026ndash;145\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e106\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e14.5\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e70\u0026ndash;147\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003et\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.51, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.61\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eADI social\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e14.54\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e6.66\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.0\u0026ndash;28.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eADI communication\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e11.68\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e5.66\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.0\u0026ndash;26.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eADI RRB\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e3.63\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e2.41\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.0\u0026ndash;12.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eADOS CSS\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e5.06\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e2.67\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.0\u0026ndash;10.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eADOS SA CSS\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e5.79\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e2.7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.0\u0026ndash;10.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eADOS RRB CSS\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e4.5\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e2.49\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.0\u0026ndash;10.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003emedian\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003estd\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003erange\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003emedian\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003estd\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003erange\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003eQC structure\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e2.15\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.11\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e2.04-3.0\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e2.13\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.19\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e1.96\u0026ndash;3.42\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003eW\u003c/span\u003e\u0026thinsp;=\u0026thinsp;4926, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.52\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003emean FD task fMRI\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.09\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.06\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.03\u0026ndash;0.33\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.08\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.07\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.03\u0026ndash;0.39\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003eW\u003c/span\u003e\u0026thinsp;=\u0026thinsp;4833, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.39\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003emean FD rsfMRI\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.06\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.05\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.03\u0026ndash;0.27\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.06\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.06\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e0.02\u0026ndash;0.4\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003eW\u003c/span\u003e\u0026thinsp;=\u0026thinsp;4915, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.5\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Bold\" class=\"Bold\" name=\"Emphasis\"\u003enumber of EEG trials\u003c/span\u003e\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cdiv class=\"SimplePara\"\u003e90.6\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cdiv class=\"SimplePara\"\u003e24.8\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cdiv class=\"SimplePara\"\u003e29\u0026ndash;148\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cdiv class=\"SimplePara\"\u003e93.1\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cdiv class=\"SimplePara\"\u003e29.7\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cdiv class=\"SimplePara\"\u003e27\u0026ndash;149\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cdiv class=\"SimplePara\"\u003e\u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003et\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.65, \u003cspan type=\"Italic\" class=\"Italic\" name=\"Emphasis\"\u003ep\u003c/span\u003e\u0026thinsp;=\u0026thinsp;0.52\u003c/div\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cdiv class=\"SimplePara\"\u003eautism\u0026thinsp;=\u0026thinsp;NAI\u003c/div\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003cbr/\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3942971/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3942971/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eDifferences in face processing are commonly reported in case/control studies of autism. Their neural correlates have been explored extensively across single neuroimaging modalities within key regions of the face processing network, such as the fusiform gyrus (FFG). Nonetheless, it is poorly understood how different variation(s) in brain anatomy and function \u003cem\u003ecombine\u003c/em\u003e to impact face processing and social functioning. Extracting the shared information across different modalities is essential to derive a more comprehensive understanding of the mechanisms underlying autism. Here, we leveraged a large multimodal sample to study the cross-modal signature of face processing within the FFG across four imaging modalities (structural MRI, resting-state fMRI [rs-fMRI], task-fMRI and EEG) in 204 individuals aged 7-30years comprising both autistic and non-autistic individuals (NAI). Combining two methodological innovations – normative modeling and linked independent component analysis – we integrated individual-level deviations across modalities to assess the efficacy of multimodal components in differentiating autistic from NAI and informing autism-associated social functioning. Autistic individuals differed significantly in a multimodal component, driven by bilateral rs-fMRI, bilateral structure, right task-fMRI, and left EEG loadings involving face-selective and retinotopic FFG regions. Multimodal components outperformed unimodal ones in differentiating autistic from NAI. Within the autism group, there was a significant multivariate association between multimodal components and a set of cognitive and clinical features associated with social functioning but not non-social features. These findings underscore the importance of elucidating individual-level, integrated neural associations of core social functioning in autism, offering potential for refined stratification and the identification of mechanistic and prognostic biomarkers.\u003c/p\u003e","manuscriptTitle":"A multimodal neural signature of face processing in autism within the fusiform gyrus","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-02-16 16:49:48","doi":"10.21203/rs.3.rs-3942971/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-mental-health","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"natmentalhealth","sideBox":"Learn more about [Nature Mental Health](https://www.nature.com/natmentalhealth/)","snPcode":"44220","submissionUrl":"https://mts-natmentalhealth.nature.com/cgi-bin/main.plex","title":"Nature Mental Health","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"69b3cd66-43a3-46c1-981f-ed14dfafed71","owner":[],"postedDate":"February 16th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":28778558,"name":"Biological sciences/Neuroscience/Diseases of the nervous system/Autism spectrum disorders"},{"id":28778559,"name":"Biological sciences/Neuroscience/Computational neuroscience"}],"tags":[{"value":"featured","date":"2024-02-16 20:16:35"}],"updatedAt":"2025-01-03T08:09:48+00:00","versionOfRecord":{"articleIdentity":"rs-3942971","link":"https://doi.org/10.1038/s44220-024-00349-4","journal":{"identity":"nature-mental-health","isVorOnly":false,"title":"Nature Mental Health"},"publishedOn":"2025-01-02 05:00:00","publishedOnDateReadable":"January 2nd, 2025"},"versionCreatedAt":"2024-02-16 16:49:48","video":"","vorDoi":"10.1038/s44220-024-00349-4","vorDoiUrl":"https://doi.org/10.1038/s44220-024-00349-4","workflowStages":[]},"version":"v1","identity":"rs-3942971","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3942971","identity":"rs-3942971","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}