Early alterations of social brain networks in young children with … · alterations of social brain networks is a core component of atypical brain development at early stages of

Early alterations of social brain networks in youngchildren with autismHolger Franz Sperdin1,+,*, Ana Coito2,+, Nada Kojovic1, Tonia Rihs2, Reem Kais Jan2,Martina Franchini1, Gijs Plomp3, Serge Vulliemoz4, Stephan Eliez1, Christoph MartinMichel2, and Marie Schaer1

1Developmental Imaging and Psychopathology Laboratory, Department of Psychiatry, University of Geneva,Switzerland2Functional Brain Mapping Laboratory, Department of Fundamental Neurosciences, University of Geneva,Switzerland3Department of Psychology, University of Fribourg, Fribourg, Switzerland4EEG and Epilepsy Unit, Neurology, University Hospitals of Geneva*[email protected]+these authors contributed equally to this work

ABSTRACT

Social impairments are a hallmark of Autism Spectrum Disorders (ASD), but empirical evidence for early brain networkalterations in response to social stimuli is scant in ASD. Here, we recorded the gaze patterns and brain activity of toddlersand preschoolers with ASD and their typically developing (TD) peers while they explored dynamic social scenes. Source-space directed functional connectivity analyses revealed the presence of network alterations in the theta frequency band,manifesting as increased driving (hyper-activity) and stronger connections (hyper-connectivity) from key nodes of the socialbrain associated with autism. Further analyses of brain-behavioural relationships within the ASD group suggested thatcompensatory mechanisms from dorsomedial frontal, inferior temporal and insular cortical regions were associated with lowerclinical impairment and less atypical gaze patterns. Our results provide strong evidence that directed functional connectivityalterations of social brain networks is a core component of atypical brain development at early stages of ASD.

Keywords EEG; ASD; Granger causality; social brain; directed connectivity; toddlers; preschoolers

Early preferential attention to social cues is a fundamentalmechanism that facilitates interactions with other human be-ings. During the third trimester of gestation, the human fetusis already sensitive to both voices (Decasper & Spence, 1986)and face-like stimuli (Reid et al., 2017). Newborns orient to bi-ological motion (Simion et al., 2008) and prefer their mothers’voices to those of other females (Decasper & Fifer, 1980). In-fants as young as two weeks imitate faces and human gestures(Meltzoff & Moore, 1977). The orientation to and interactionwith social cues during infancy drives the later acquisitionof social communication skills during toddler and preschoolyears. As function of experience, the repeated exposure leadsto the progressive emergence of adaptive interactions withconspecifics. Alongside, the brain develops a network of cere-bral regions specialized in understanding the social behaviorsof others. This network includes the orbitofrontal and medialprefrontal cortices, the superior temporal cortex, the tempo-ral poles, the amygdala, the precuneus, the temporo-parietaljunction, the anterior cingulate cortex (ACC) and the insula(Brothers, 1990; Frith & Frith, 2010; Adolphs, 2009; Blake-more, 2008). Collectively, these areas form the social brain

and are all implicated to some extent in processing social cuesand encoding human social behaviors (Brothers, 1990; Frith& Frith, 2010; Adolphs, 2009; Blakemore, 2008).

Autism is a life-long lasting, highly prevalent neurodevelop-mental disorder that affects core areas of cognitive and adap-tive function, communication and social interactions (Chris-tensen et al., 2016). A common observation in infants laterdiagnosed with ASD is the presence of less sensitivity anddiminished preferential attention to social cues during thefirst year of life (Osterling & Dawson, 1994). Toddlers withASD orient preferentially to non-social contingencies (Klinet al., 2009). Indifference to voices (Sperdin & Schaer, 2016)and faces (Grelotti et al., 2002) in ASD leads to deficits inthe development of adapted social interactions with othersand to difficulties in understanding human behaviors. It isnot established why children with ASD show insensitivityto stimuli with social contingencies at early developmentalstages, but this apparent indifference to social cues ultimatelyhinders the normal development of the social brain networkor parts thereof (Pelphrey et al., 2011; Gotts et al., 2012).Some authors propose that deficits in the development of so-cial cognition (such as learning to attribute mental states toothers, “theory of mind” (Frith, 1989)) and/or in sensory pro-

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

cessing (Dinstein et al., 2012) prevent children with ASD toactively and appropriately engage with social stimuli. An-other hypothesis suggests that they have difficulties buildingup stimulus-reward contingencies for social stimuli, due to areduced motivation to attend and engage with them. Regard-less of the reasons behind reduced social orienting, diminishedinteraction and exposure to social stimuli may in turn impedethe development of the social brain at early developmentalstages in ASD (Chevallier et al., 2012; Dawson et al., 2004).

Evidence remains limited for brain network alterations inresponse to socially meaningful stimuli in ASD during theperiod spanning the toddler to preschool years, partly becausethe acquisition of data during that age period is extremelychallenging (Raschle et al., 2012). However, studying veryyoung children with ASD, closer to their diagnosis, is all themore important when recent findings suggest the presence ofmajor developmental changes in large-scale brain networksbetween adults and younger individuals with ASD (Nomi &Uddin, 2015). Currently, it remains unclear how autism affectsthe development of the functional brain networks implicatedin the processing of socially meaningful information at earlydevelopmental stages. A better delineation of the timing andnature of the neurodevelopmental alterations associated withcore social deficits in autism may in turn help to improvetherapeutic strategies.

Electroencephalography (EEG) is as a powerful non-invasive method to study atypical brain responses to socialstimuli in clinical pediatric populations with ASD. For exam-ple, surface-based experiments have reported aberrant evokedpotentials in response to dynamic eye gaze in infants at high-risk for ASD (Elsabbagh et al., 2012) or to speech stimuliin toddlers with ASD (Kuhl et al., 2013) with differences inresting EEG power in infants at high-risk for ASD (Tierneyet al., 2012). Whilst useful, most of the EEG experimentsperformed on very young children with ASD (younger thanfour years) have been done with few electrodes only and theanalysis restricted to the sensor space. Therefore, hypotheti-cal alterations in the functional brain networks underlying theprocessing of social stimuli remain to be determined for thatage period in ASD.

Here, we recorded high-density EEG and high resolutioneye-tracking in toddlers and preschoolers with ASD and theirTD peers as they watched naturalistic and ecologically validdynamic social movies. Using data-driven methods, we firstinvestigated whether the visual exploration behavior was atyp-ical in toddlers and preschoolers with ASD using kernel den-sity distribution estimations. Then, we explored whether theirongoing source-space directed functional connectivity was al-tered compared to their TD peers using Granger-causal model-ing applied to the EEG source signals. This method estimatesbrain connectivity in the frequency domain. It identifies whichbrain regions are the key drivers of information flow in a brainnetwork and directional relationships between brain regionsthat belong to a network (Coito et al., 2016b). Finally, welooked for relationships between directed functional connec-

tivity measures, visual exploration behavior and clinical phe-notype. As toddlers with ASD have less preferential attentionfor social cues, we hypothesized that they would show botha different visual exploration behavior of the dynamic socialimages and altered directed functional connectivity patternsin brain regions involved in processing social informationcompared to their TD peers.

ResultsSummed outflowThe summed outflow (i.e.- the amount of information transfer)is a measure that reflects the importance (i.e.-the amount ofdriving) of a given region of interest (ROI) in the network (seemethod section). To understand the functional wiring and thedynamic flow underlying the processing of the dynamic socialstimuli, we used a data-driven method to explore in which fre-quency band the highest summed outflow occurred in 82 ROIsacross the whole brain. A ROI with a strong summed outflowhas a key role in directing the activity towards other ROIs inthe network. The strongest summed outflow across the wholebrain occurred in the theta band (4− 7Hz) in both groups.The summed outflow of the largest drivers across frequenciesis illustrated for each group in Figure 1a. As can be seen thelargest peaks of activity are present in the theta band range(4−7Hz) in both group. The global driving in the theta banddid not differ between the groups (t = 0.6201, p = 0.536). Inaddition, several regions common to both groups showed alarge driving (summed outflow) in this theta band, and no-tably the bilateral medial frontal and superior orbitofrontalregions, the bilateral hippocampi, the bilateral ACC and theright amygdala (Figure 1b).

Thereafter, we characterized the differences in the summedoutflow in the theta band across all brain regions betweenthe groups. Amongst all the ROIs, we identified sixROIs that showed a statistically higher driving (strongersummed outflow) in the ASD group in comparison to the TDgroup (Mann −Whitney −Wilcoxontest, two − tailed, p <0.05): the right orbital part of the superior frontal gyrus(Ws = 267,z = −2.088.p = 0.037,r = −0.348), the bilat-eral orbital parts of the middle frontal gyri (Le f t : Ws =259,z=−2.341, p= 0.019,r =−0.39;Right : Ws= 252,z=−2.563, p = 0.01,r = −0.427), the right middle cingulategyrus (Ws = 259,z = −2.341, p = 0.019,r = −0.390), theleft superior occipital gyrus (Ws = 270,z = −1.993, p =0.047,r = −0.332), and the left superior temporal gyrus(STG) (Ws = 255,z =−2.468, p = 0.013,r =−0.411) (Fig-ure 2). This indicates the presence of a stronger driving fromthese regions in the toddlers and preschoolers with ASD com-pared to their TD peers when viewing the dynamic socialimages. No other regions had a higher driving in the TDgroup compared to the ASD group.

Region-to-region directed functional connectivityWe looked for differences in the region-to region directedfunctional connectivity using Granger-causal modelling (see

2/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

method section) from each of the six nodes separatelyin both groups. The results indicated that all the con-nections in the toddlers and preschoolers with ASD werestronger than the strongest connections in the TD par-ticipants (Mann − Whitney − Wilcoxon, two − tailed, p <0.05,Ben jamini−Hochberg = 0.05). This suggests the pres-ence of hyper-connectivity in the toddlers and preschoolerswith ASD. In both groups, the strongest connections wasfrom a node within the frontal cortex: the right superiororbito-frontal region (OFC). The region-to-region directedfunctional connectivity from the right superior OFC is illus-trated in Figure 3. The estimation of the region-to-regiondirected connectivity (i.e. to which other ROIs the activitywas directed) revealed major differences between the groups.In the TD group, the right superior OFC had directed connec-tions towards the left middle frontal, the right precentral, theleft superior frontal, the left supramarginal, the left cuneus,left hippocampus, right paracentral, right opercular part ofthe inferior frontal cortex, left inferior parietal cortex and theright angular gyrus. In the group of toddlers and preschoolerswith ASD, the right superior OFC drove information flowtowards the left orbital part of the medial frontal cortex, leftamygdala, left supramarginal, right supplementary motor area,right lingual gyrus, right medial part of the medial frontalcortex, left paracentral, left inferior frontal triangularis region,right hippocampus and right orbital part of the inferior frontalcortex. Thus, the analysis reveals a different network patternin the toddlers and preschoolers with ASD compared to theirTD peers. We provide the directed connections results fromthe five remaining ROIs in the supplementary information.

Correlations with ADOS-2, PEP-3, VABS-II scoresand gaze Proximity IndexWe further explored associations between the summed out-flow from the ROIs and clinical and behavioural pheno-types. None of the correlations between the summed out-flow and ADOS-2 severity scores survived False discov-ery rate (FDR) correction (Ben jamini−Hochberg = 0.05).However, we found strong positive correlations betweenthe summed outflow in the right lingual area and scoresfrom the socialization domain (rs = 0.751, p = 0.0003, two−tailed,< 0.05;Ben jamini − Hochberg = 0.05) as well aswith scores from the leisure and play skills subdomainof the VABS-II (rs = 0.802, p = 0.0001, two − tailed,<0.05;Ben jamini−Hochberg = 0.05). Higher summed out-flow within the left Heschl area near the posterior convo-lutions of the insula and the left rolandic operculum nearthe circular sulcus of the insula rostrally were positively re-lated to better fine (rs = 0.745, p = 0.0004, two− tailed,<0.05;Ben jamini−Hochberg = 0.05) and gross motor skills(rs = 0.744, p = 0.0004, two − tailed,< 0.05;Ben jamini −Hochberg = 0.05) as measured by the PEP-3. Finally, tod-dlers and preschoolers with ASD with a gaze pattern similarto their TD peers showed an increased driving within theleft middle cingulate cortex (rs = 0.726, p = 0.0007, two−

tailed,< 0.05;Ben jamini−Hochberg = 0.05) and the rightparacentral lobule (rs = 0.738, p = 0.0005, two− tailed,<0.05;Ben jamini−Hochberg = 0.05). The correlations be-tween the summed outflows and the Proximity Index (Seemethod section), VABS-II scores and PEP-3 scores are dis-played in Figure 4.

DiscussionAbnormal processing of social cues is a hallmark of ASD(Chevallier et al., 2012; Dawson et al., 2004; Dichter et al.,2009; Elsabbagh et al., 2012; Gotts et al., 2012; Greene et al.,2011; Klin et al., 2009; Pelphrey et al., 2011). However, evi-dence for alterations of social brain networks at early stagesof ASD is scant. Using data-driven methods, we observedaberrant gaze patterns together with altered directed func-tional connectivity in toddlers and preschoolers with ASDwhen exploring dynamic social stimuli compared to their TDpeers. These differences manifested as increased driving andhyper-connectivity in the theta frequency band from nodesthat includes the right orbital part of the superior frontal gyrus,the bilateral orbital parts of the middle frontal gyri, the rightmiddle cingulate gyrus, the left superior occipital gyrus andthe left STG. To the best of our knowledge, this is the first ev-idence indicating concomitant alterations in the visual explo-ration of dynamic social images and in the directed functionalconnectivity involving key nodes of the social brain (Brothers,1990; Frith & Frith, 2010; Adolphs, 2009; Blakemore, 2008)at early stages of ASD.

Our results indicate that the highest information transfer(i.e. summed outflow) occurred at the global brain level in thetheta frequency band (4−7Hz). Slow waves are prominentbrain rhythms during infancy and toddlerhood. Throughoutdevelopment, they modulate attentional brain states, encodespecific information and ease communication between neu-ronal populations. Theta band activity is associated withsustained anticipatory attention, memory encoding, emotionalprocessing and cognitive performance during infancy, toddler-hood and preschool years (Saby & Marshall, 2012). EEGexperiments in individuals with ASD show a reduction or anincrease in the coherence patterns in the theta frequency bandcompared to their TD peers at various ages and under differ-ent experimental conditions with sensor-space based analysis(Schwartz et al., 2016). In very young children, theta bandactivity is linked to the development of the social brain. Forexample, surface-based EEG experiments in TD infants reporta greater theta power to social versus non-social stimuli at 12months (Jones et al., 2015). Theta increases during attentionto social stimulation in infants and preschool aged children(Orekhova et al., 2006). Hence, social contingencies modulatetheta band activity. Similarly, our results show high informa-tion transfer in the theta band when toddlers and preschoolersexplore dynamic social stimuli. However, most of the avail-able EEG experiments and analysis thereof are restricted tothe scalp surface. Therefore, information is limited regard-ing the involvement of specific brain regions when young

3/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

children are exposed to social stimuli. Here, our data-drivensource-space approach that estimates brain connectivity in thefrequency domain revealed not only high information trans-fer in the theta frequency band, but also, the involvement ofthe bilateral medial and the superior orbital frontal regions,the bilateral hippocampi, the bilateral ACC and right amyg-dala. These areas are implicated in processing social cues andencoding human social behaviours (Brothers, 1990; Frith &Frith, 2010; Adolphs, 2009; Blakemore, 2008).

Our results further indicate the presence of an altered thetaresponse (higher driving) in the toddlers and preschoolerswith ASD compared to their TD peers within key regions ofthe social brain. In TD individuals, theta generates withinthe frontal cortices or the ACC (Asada et al., 1999). In com-parison to their TD peers, both areas develop differently inyoung toddlers later diagnosed with ASD (Schumann et al.,2010). Accordingly, our results raise the possibility that thebrain regions generating theta also follow a different develop-ment as driving in the theta band from frontal and cingulateregions was altered in our participants with ASD comparedto their TD peers. A magnetoencephalography (MEG) studyusing a source-space approach reported lower coherence inthe theta band within parietal and occipital regions but theirsample included adolescents with ASD only (Ye et al., 2014).The differences between this specific study and our couldstem from either variations in the age groups (adolescentsversus toddlers and preschoolers), the stimuli used (grey crossinside a white circle versus dynamic biological visual stim-uli) or the methods. More generally, several factors explaindiscrepancies in brain connectivity results between studies.The type of connectivity measures applied, the approach (sur-face versus source based), the brain regions analysed andfrequency bands examined are variables that influence theresults (Mohammad-Rezazadeh et al., 2016). In ASD, hyper-connectivity is prevalent in younger populations (that is, in-fants at high-risk for ASD, toddlers and preschoolers withASD) while hypo-connectivity is more observed during ado-lescence and adulthood in ASD (Uddin et al., 2013b). Con-versely, a developmental shift occurs in brain growth withan initial period of early brain overgrowth followed by nor-malization sometime during adolescence (Courchesne et al.,2011). Accordingly, structural white matter connectivity stud-ies also highlight this shift from higher structural connectivityin very young children with ASD to lower connectivity inolder children with ASD (Hoppenbrouwers et al., 2014; Contiet al., 2015). Therefore, higher-driving and hyper-connectivityfrom key nodes of the social brain in the theta frequency bandin our ASD sample is consistent with reports in the litera-ture when considering the very young age of our participants(around 3 years of age on average).

We found increased activity from nodes of the orbital andmedial parts of the frontal cortex and cingulate cortex in thetoddlers and preschoolers with ASD. Both areas have been im-plicated in various complex aspects of social cognition, socialreward, social perception and social behaviour (Jonker et al.,

2015; Apps et al., 2013). Metabolic changes within the medialprefrontal cortex and the cingulate cortex are correlated withsocial interaction impairments in childhood ASD (Ohnishiet al., 2000). Several experiments report structural (Patriquinet al., 2016) and functional (Gotts et al., 2012; Patriquin et al.,2016; Greene et al., 2011; Cheng et al., 2015) alterationswithin these brain areas in school aged children, adolescentsand adults with ASD when compared to their TD peers. Arecent study described hyper-connectivity within the ACC andbilateral insular cortices in a sample including children agedbetween seven to 12 years (Uddin et al., 2013a). Some au-thors propose that the two together form the salience network,whose role is to direct attention to behaviourally-relevantstimuli (Menon & Uddin, 2010). Although we didn’t finddifferences in the driving from the insula compared to the TDpeers, there is an increasing number of evidence showing anabnormal development of the salience network or componentsthereof in ASD (Uddin, 2015), which may partially explainthe limited interest for and engagement with social stimuli thatis often observed in individuals with ASD and that constitutesa hallmark of the disorder (Klin et al., 2009; Pelphrey et al.,2011). Accordingly, the toddlers and preschoolers with ASDhad a different visual exploration behaviour of the dynamicsocial stimuli raising the possibility of a reduced interest to vi-sually engage with them. Alternatively, alterations in drivingfrom these regions could partially reflect a reduced motivationto attend and engage with the dynamic social stimuli (Cheval-lier et al., 2012; Dawson et al., 2004).

We also found higher driving from the superior tempo-ral and occipital gyri. Those brain areas are implicated inthe processing of biological motion, in analysing the inten-tions of other people’s actions and self-reflection (Pelphrey& Carter, 2008; Pelphrey et al., 2005; Pelphrey & Morris,2006; Pelphrey et al., 2004). Our result would suggest thatthe exploration of the dynamic social visual stimuli that con-tained biological movements led to altered driving from thesebrain areas in toddlers and preschoolers with ASD comparedto their TD peers. Beyond functional and structural brainalterations reported elsewhere in older children and adultswith ASD, our results suggest for the first time, the presenceof alterations in the driving of information flow from brainareas implicated in social information processing during theviewing of naturalistic dynamic social images in toddlers andpreschool with ASD.

We further explored associations between the driving inthe nodes of the network (that is, the summed outflow) andclinical and behavioural phenotypes. We didn’t observe anysignificant relationships between summed outflow and theADOS severity scores after FDR correction. Notwithstand-ing, we observed an increased driving from the median cin-gulate cortex and the paracentral lobule in the toddlers andpreschoolers with ASD who had a more similar visual explo-ration pattern to their TD peers. Thus, an improved visualexploration pattern of the dynamic social images was relatedto increased summed outflow from these regions. Likewise,

4/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

higher summed outflow from the right lingual area was relatedto better socialization behaviour and leisure and play skills.Higher summed outflow from the left Heschl’s area near theposterior convolutions of the insula and the left rolandic op-erculum near the circular sulcus of the insula rostrally werepositively related to better fine and gross motor skills. Assuch, increased activity within dorsomedial frontal, inferiortemporal and insular cortical regions were associated withlower clinical impairment and less atypical gaze patterns. Thepresence of hyperactivity within relevant brain region hasbeen interpreted as a possible compensatory mechanism whenperforming a social target detection task, in adults with ASDat least (Dichter et al., 2009). We speculate that the hyper-driving from these brain regions might be a mechanism tocompensate for atypical development of the brain’s circuitryover time as higher directed functional connectivity was re-lated to better visual exploration of dynamic social images,improved socialization and motor behaviours.

The collective results suggest that directed functional con-nectivity network alterations within regions of the social brainare present at early stages of ASD, justifying further investiga-tion into how early therapeutic interventions targeting socialorienting skills may help to remediate social brain develop-ment during this critical age period when plasticity is stillpossible.

MethodsParticipantsRecruitment of toddlers and preschoolers with ASD wasachieved via clinical centres specialized in ASD and French-speaking parent associations. TD toddlers and preschoolerswere recruited via announcements in the Geneva community.Prior to the experiments, all the procedures were approvedby the Ethics Committee of the Faculty of Medicine of theUniversity of Geneva Hospital in accordance with the ethi-cal standards proclaimed in the Declaration of Helsinki. Forall participants, an interview over the phone and a medicaldevelopmental history questionnaire were completed beforetheir initial visit. All participants’ parents gave their informedconsent prior to inclusion in the study. 120 participants wererecruited for the experiment. We did not manage to put theEEG cap on the head of 23 ASD and 7 TD participants. Wemanaged to put the cap on 90 participants. Out of those,we excluded 28 ASD and 26 TD participants because of toomany movement-related artefacts, unrepairable noisy signal,lack of interest, or insufficient amounts of epochs availablefor subsequent analysis. This was to be expected given theextremely sensitive population at study here. As a result,36 participants were included in the final sample: 18 youngchildren with ASD (2 females; mean age 3.1 years +/- 0.8,range 2.2-4.4) and 18 age matched (t = 2.72, p = 0.852) TDpeers (5 females; mean age 3.1 years +/- 0.9, range 2.0-4.8).All participants with ASD included in the study received aclinical diagnosis prior to their inclusion in the research proto-col. Diagnosis of ASD was rigorously verified and confirmed

with either the Autism Diagnosis Observation Schedule –Generic (Lord et al., 2000) or the Autism Diagnosis Observa-tion Schedule, second edition (ADOS-2)(Luyster et al., 2009).The latter contains a toddler module that defines concernfor ASD. ADOS assessments were administered and scoredby experienced clinicians working at the institution and spe-cialized in ASD identification. In order to compare scoresfrom different modules, we transformed the ADOS-G scoresinto Calibrated Severity Scores (ADOS-CSS) (Gotham et al.,2009). For the participants that underwent the ADOS-2- tod-dler module, we calibrated the scores into Severity Scores(Esler et al., 2015). Five children under 30 months of ageperformed the toddler module of the ADOS-2. All scoredin the moderate to severe range of concern for ASD. For allthe participants younger than 3 years of age (n=10) at theEEG acquisition, clinical diagnosis was confirmed after oneyear by a clinician specialized in ASD identification usingthe ADOS-G or ADOS-2. The mean global ADOS-CSS forthe entire group of patients with ASD was 7.9 (SD = 1.6).The assessment of the participants with ASD also includedthe administration of additional clinical standardized tests.Adaptive behaviour was assessed using the Vineland AdaptiveBehaviour Scale-II (VABS-II)(Sparrow et al., 2005), a stan-dardized parent report interview. Developmental level wasassessed with the Psycho-educational Profile Third Edition(PEP-3)(Schopler et al., 2005). See Table 1 for characteristicsof study participants. Prior to their inclusion in our researchprotocol, potential TD participants were initially screened forneurological/psychiatric problems and learning disabilities us-ing a medical and developmental history questionnaire beforetheir visit. Moreover, they underwent ADOS-G or ADOS-2evaluations to exclude any ASD symptomatology. Fourteencontrols were tested with Modules 1 or 2 and four underwentthe toddler module of the ADOS-2. All TD participants had aminimal severity score of 1, except one child who had a scoreof 3.

StimuliStimuli consisted of two video sequences of dynamic socialimages without audio information of approximatively twominutes each. These videos included ecologically valid andcomplex naturalistic dynamic images where young childrenpractised yoga alone, imitated animal-like behaviours (behav-ing like a monkey or jumping like a frog), waived their arms,struck a pose, jumped, made faces or whistled (Yoga Kids 3; Gaiam, Boulder, Colorado, http://www.gaiam.com, createdby Marsha Wenig, http://yogakids.com/). Presentation andtiming of stimuli were controlled by Tobii Studio software(Sweden, http://www.tobii.com).

Procedure and taskThe experiment was conducted in a lit room at the officeMedico-Pedagogique in Geneva. To familiarize the child withthe procedure, the families received a kit containing a custom-made EEG replica cap and pictures illustrating the protocolin order to familiarize the children with the experiment two

5/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

weeks prior to their first visit. Participants were seated on theirparents lap in order to make them feel as secure as possibleand to minimize head and body movements or alone. Onceseated, the experimenter measured the circumference of thehead and placed the corresponding cap on the participant’shead. A couple of minutes were taken in order to allow theparticipants to settle into the experiment’s environment andget used to the cap before starting the experiment. Followingthis, a five point eye-tracking calibration procedure was initi-ated using the Tobii system (Sweden, http://www.tobii.com).An attractive colourful object (either a kitten, a bus, a duck,a dog or a toy) was presented together with its correspond-ing sound on a white background and the participants had tofollow the object visually. The recording and presentation ofthe visual stimuli started when a minimum of four calibrationpoints were acquired for each eye. To best capture the child’sattention, we first showed them an age-appropriate animatedcartoon, followed by some fractals and another animated car-toon. The block ended with a film containing dynamic socialimages, the condition of interest in the present experiment.All participants were presented with the same visual stimuliin the same order. Following the first block, impedances wererechecked and electrodes were readjusted where needed tomaintain them below 40 kOhm. A second block was thenacquired (animated cartoon; animated fractals; animated car-toon; second condition of interest: dynamic social images).The experimenter continuously monitored the eye-tracking toensure children were looking at the screen. The whole experi-ment lasted about half an hour. We used stringent criteria andonly participants with the highest data quality were kept forsubsequent analysis.

Eye-tracking measurements

Eye-tracking data were recorded with the TX300 Tobii eye-tracking system (sampling rate resolution of 300 Hz). In orderto analyse and quantify differences in visual exploration in oursample, we developed a data-driven method to define dynamicnorms of the exploration of the visual scenes (Kojovic et al., inpreparation). First, we applied a kernel density distribution es-timation (Botev et al., 2010) on the eye-tracking data recordedfrom the TD group at each time frame of the films containingdynamic complex social images to compute a normative gazedistribution pattern. Then, for each of the participants withASD individually, we computed a deviation index from thisnormative gaze distribution, and this, for each single timeframe separately (Figure 5). We averaged these values acrossthe two films to obtain a mean Proximity Index (PI) value.This index describes for a given ASD participant, his distancefrom the normative gaze distribution pattern calculated on theTD group. A high index value indicates a visual behaviour ap-proaching the visual exploration of the TD participants (moresimilarity), while a low index indicates a visual behaviourdeviating from the TD group (more dissimilarity).

EEG acquisition and preprocessingThe EEG was acquired with a Hydrocel Geodesic Sen-sor Net (HCGSN, Electrical Geodesics, USA) with 129scalp electrodes at a sampling frequency of 1000Hz.On-line recording was band-pass filtered at 0–100Hzusing the vertex as reference. Data pre-processingwas done using Matlab (Natick, MA) and Cartool(http://sites.google.com/site/cartoolcommunity/). We down-sampled the montage to a 111-channel electrode array toexclude electrodes on the cheek and the neck since those areoften contaminated with artefacts. Data were filtered between1 and 40Hz (using non-causal filtering) and a 50Hz notch fil-ter was applied. Each file was then visually inspected by oneof the three EEG experts (HFS, TR, and RKJ) to exclude peri-ods of movements artefacts. Periods where subjects were notlooking at the screen were excluded. Independent componentanalysis (ICA) was performed on the data to identify and re-move the components related to eye movement artefacts (eyeblinks, saccades). Subsequently, channels with substantialnoise were interpolated using spherical spline interpolationfor each recording. Finally, the cleaned data were down-sampled to 125Hz, recalculated against the average referenceand inspected by two EEG experts (HFS and AC) to ensurethat no artefacts had been missed. One hundred and twentyartefact-free epochs of 1 second per participant were includedfor further analysis and were considered as a minimum toensure high enough data quality.

Electrical Source Imaging and selection of Regionsof InterestThe general analysis strategy is summarized in Figure 6. Elec-trical source imaging (ESI) was performed to reconstruct thesources of brain activity that gave rise to the scalp EEG field.For this, we used an infant template head model (33-44 month)(using the Montreal Neurological Institute (MNI) brain) withconsideration of skull thickness (Locally Spherical Modelwith Anatomical Constraints, LSMAC). 4159 solution pointswere equally distributed in the grey matter. We used a dis-tributed linear inverse solution (Low Resolution Electromag-netic Tomography, LORETA (Pascual-Marqui et al., 1994)) tocompute the 3-dimensional (3D) current source densities. Wethen projected this 3D dipole time-series, onto the predomi-nant dipole direction of each region of interest (ROI) acrosstime and epochs, therefore obtaining a scalar time-series se-ries (Coito et al., 2016a, 2015; Plomp et al., 2015; Coito et al.,2016b). We parcelled the grey matter in 82 ROIs based on theautomated anatomical labelling (AAL) digital atlas (Tzourio-Mazoyer et al., 2002), after normalization to the MNI spaceusing SPM8 (Wellcome Trust Centre for Neuroimaging, Uni-versity College London, UK, www.fil.ion.ucl.ac.uk/spm). Inorder to reduce the dimensionality of the solution space, weconsidered the solution point closest to the centroid of eachROI as representative of the source activity in that ROI forfurther analysis. This allowed us to obtain the source activityacross time of 82 solution points, representative of 82 ROIs

6/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

(Coito et al., 2016b).

Directed Functional Connectivity using Granger-causalityDirected functional connectivity estimates the causal influencethat one signal exerts onto another, facilitating the study of di-rectional relationships between brain regions. It is commonlyassessed using the concept of Granger-causality: given twosignals in a process, if the knowledge of the past of one allowsa better prediction of the presence of the other signal in the pro-cess, then the former signal is said to Granger-cause the lattersignal (Granger, 1969). In order to estimate these directionalrelationships, we computed the weighted Partial Directed Co-herence (wPDC) (Baccala & Sameshima, 2001; Astolfi et al.,2006; Plomp et al., 2014) using the 82 source signals. PDC isa multivariate approach, which considers all signals simulta-neously in the same model and estimates brain connectivityin the frequency domain. It is computed using multivariateautoregressive models of a certain model order. Here, we useda model order of 5, corresponding to 40ms. The wPDC wascomputed for each subject and epoch and then the average ofthe PDC values within subjects was taken (Coito et al., 2016b).The average PDC was then scaled (0− 1) across ROIs andfrequencies (1− 40Hz) by subtracting the minimum powerand dividing by the range. In order to weight the PDC bythe spectral power (SP) of each source signal, while avoidingfrequency doubling, we computed the Fast Fourier Transform(FFT ) for each electrode, applied ESI to the real and imag-inary part of the FFT separately and then combined them(Coito et al., 2016a, 2015; Plomp et al., 2015; Yuan et al.,2008). The mean SP was obtained for each subject and scaled(that is 0-1, in the same way as PDC) for further details on themethodological approach to compute directed functional con-nectivity from electrical-source imaging signals, we refer thereader to (Coito et al., 2016b). For each subject, we obtained a3D connectivity matrix (ROIs x ROIs x frequency), represent-ing the outflow from one ROI to another for each frequency.For further analysis, we reduced the connectivity matrix to 3frequency bands: theta (4−7Hz), alpha (7−12Hz) and beta(12− 30Hz), by calculating the mean connectivity value ineach band. For each subject and frequency band, we com-puted the summed outflow, which is the sum of the outflows(wPDC values) from a given ROI to all the others and re-flects the driving importance of this ROI in the network: ROIswith high summed outflow strongly drive the activity of otherROIs. We identified the highest information transfer (summedoutflow) in the theta band. Therefore, we focused our sub-sequent analysis on this frequency band. We carried out sta-tistical comparisons of the summed outflows between sub-ject groups using a non-parametrical statistical test (Mann−Whitney−Wilcoxon, two− tailed, p < 0.05). We then inves-tigated the outflows from the ROIs that showed statisticallysignificant summed outflow between groups to the wholebrain (remaining 81 ROIs) and carried out a statistical com-parison of these outflows between groups(Mann−Whitney−

Wilcoxon, two− tailed, p < 0.05,Ben jamini−Hochberg =0.05). We correlated (Spearman − rho, two − tailed, p <0.05) the summed outflow results obtained in each of the82 ROIs with ADOS-CSS scores, with developmental scoresobtained from the PEP-3, with adaptive scores obtained fromVABS-II and with the PI values obtained from the eye-trackingdata. In all cases, correlation p-values were Benjamini-Hochberg–corrected for multiple testing with p = 0.05. Con-nectivity computations were performed in Matlab. Figures1,2,3 and 4 were produced using the BrainNet Viewer toolbox(Xia et al., 2013).

Data availabilityThe data and codes that support the findings ofthe present work are available upon reasonable re-quest to the corresponding authors Holger FranzSperdin ([email protected]) or Marie Schaer([email protected]).

ReferencesAdolphs, R. (2009). The social brain: Neural basis of social

knowledge. Annual Review of Psychology, 60, 693–716.Apps, M. A., Lockwood, P. L., & Balsters, J. H. (2013). The

role of the midcingulate cortex in monitoring others’ deci-sions. Front Neurosci, 7, 251.

Asada, H., Fukuda, Y., Tsunoda, S., Yamaguchi, M., &Tonoike, M. (1999). Frontal midline theta rhythms re-flect alternative activation of prefrontal cortex and anteriorcingulate cortex in humans. Neuroscience Letters, 274,29–32.

Astolfi, L., Cincotti, F., Mattia, D., Marciani, M. G., Baccala,L. A., de Vico Fallani, F., Salinari, S., Ursino, M., Zavaglia,M., & Babiloni, F. (2006). Assessing cortical functionalconnectivity by partial directed coherence: simulations andapplication to real data. IEEE Trans Biomed Eng, 53, 1802–12.

Baccala, L. A. & Sameshima, K. (2001). Partial directedcoherence: a new concept in neural structure determination.Biol Cybern, 84, 463–74.

Blakemore, S. J. (2008). The social brain in adolescence. NatRev Neurosci, 9, 267–77.

Botev, Z. I., Grotowski, J. F., & Kroese, D. P. (2010). Kerneldensity estimation via diffusion. Annals of Statistics, 38,2916–2957.

Brothers, L. (1990). The social brain: a project for integratingprimate behavior and neurophysiology in a new domain.Concepts Neurosci, 1, 27–61.

Cheng, W., Rolls, E. T., Gu, H., Zhang, J., & Feng, J. (2015).Autism: reduced connectivity between cortical areas in-volved in face expression, theory of mind, and the sense ofself. Brain, 138, 1382–93.

Chevallier, C., Kohls, G., Troiani, V., Brodkin, E. S., &Schultz, R. T. (2012). The social motivation theory ofautism. Trends Cogn Sci, 16, 231–9.

7/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Christensen, D. L., Baio, J., Braun, K. V., Bilder, D., Charles,J., Constantino, J. N., Daniels, J., Durkin, M. S., Fitzger-ald, R. T., Kurzius-Spencer, M., Lee, L. C., Pettygrove,S., Robinson, C., Schulz, E., Wells, C., Wingate, M. S.,Zahorodny, W., & Yeargin-Allsopp, M. (2016). Prevalenceand characteristics of autism spectrum disorder among chil-dren aged 8 years - autism and developmental disabilitiesmonitoring network, 11 sites, united states, 2012. MmwrSurveillance Summaries, 65, 1–23. Dl0pm Times Cited:100Cited References Count:42.

Coito, A., Genetti, M., Pittau, F., Iannotti, G. R., Thom-schewski, A., Holler, Y., Trinka, E., Wiest, R., Seeck, M.,Michel, C. M., Plomp, G., & Vulliemoz, S. (2016a). Altereddirected functional connectivity in temporal lobe epilepsyin the absence of interictal spikes: A high density eeg study.Epilepsia, 57, 402–11.

Coito, A., Michel, C. M., van Mierlo, P., Vulliemoz, S., &Plomp, G. (2016b). Directed functional brain connectivitybased on eeg source imaging: Methodology and applicationto temporal lobe epilepsy. IEEE Trans Biomed Eng.

Coito, A., Plomp, G., Genetti, M., Abela, E., Wiest, R., Seeck,M., Michel, C. M., & Vulliemoz, S. (2015). Dynamicdirected interictal connectivity in left and right temporallobe epilepsy. Epilepsia, 56, 207–17.

Conti, E., Calderoni, S., Marchi, V., Muratori, F., Cioni, G.,& Guzzetta, A. (2015). The first 1000 days of the autisticbrain: a systematic review of diffusion imaging studies.Frontiers in Human Neuroscience, 9.

Courchesne, E., Campbell, K., & Solso, S. (2011). Braingrowth across the life span in autism: age-specific changesin anatomical pathology. Brain Res, 1380, 138–45.

Dawson, G., Toth, K., Abbott, R., Osterling, J., Munson,J., Estes, A., & Liaw, J. (2004). Early social attentionimpairments in autism: social orienting, joint attention, andattention to distress. Dev Psychol, 40, 271–83.

Decasper, A. J. & Fifer, W. P. (1980). Of human bonding -newborns prefer their mothers voices. Science, 208, 1174–1176. Jt663 Times Cited:863 Cited References Count:24.

Decasper, A. J. & Spence, M. J. (1986). Prenatal maternalspeech influences newborns perception of speech sounds.Infant Behavior and Development, 9, 133–150. D6996Times Cited:352 Cited References Count:34.

Dichter, G. S., Felder, J. N., & Bodfish, J. W. (2009). Autismis characterized by dorsal anterior cingulate hyperactiva-tion during social target detection. Social Cognitive andAffective Neuroscience, 4, 215–226.

Dinstein, I., Heeger, D. J., Lorenzi, L., Minshew, N. J.,Malach, R., & Behrmann, M. (2012). Unreliable evokedresponses in autism. Neuron, 75, 981–91.

Elsabbagh, M., Mercure, E., Hudry, K., Chandler, S., Pasco,G., Charman, T., Pickles, A., Baron-Cohen, S., Bolton,P., Johnson, M. H., & Team, B. (2012). Infant neuralsensitivity to dynamic eye gaze is associated with lateremerging autism. Current Biology, 22, 338–342. 896twTimes Cited:130 Cited References Count:31.

Esler, A. N., Bal, V. H., Guthrie, W., Wetherby, A., Ellis Weis-mer, S., & Lord, C. (2015). The autism diagnostic observa-tion schedule, toddler module: Standardized severity scores.J Autism Dev Disord, 45, 2704–20.

Frith, U. (1989). A new look at language and communicationin autism. British Journal of Disorders of Communication,24, 123–150.

Frith, U. & Frith, C. (2010). The social brain: allowing hu-mans to boldly go where no other species has been. Philo-sophical Transactions of the Royal Society B-BiologicalSciences, 365, 165–175.

Gotham, K., Pickles, A., & Lord, C. (2009). Standardizingados scores for a measure of severity in autism spectrumdisorders. J Autism Dev Disord, 39, 693–705.

Gotts, S. J., Simmons, W. K., Milbury, L. A., Wallace, G. L.,Cox, R. W., & Martin, A. (2012). Fractionation of socialbrain circuits in autism spectrum disorders. Brain, 135,2711–25.

Granger, C. W. J. (1969). Investigating causal relations byeconometric models and cross-spectral methods. Econo-metrica, 37, 414–38.

Greene, D. J., Colich, N., Iacoboni, M., Zaidel, E.,Bookheimer, S. Y., & Dapretto, M. (2011). Atypical neuralnetworks for social orienting in autism spectrum disorders.Neuroimage, 56, 354–62.

Grelotti, D. J., Gauthier, I., & Schultz, R. T. (2002). Socialinterest and the development of cortical face specializa-tion: what autism teaches us about face processing. DevPsychobiol, 40, 213–25.

Hoppenbrouwers, M., Vandermosten, M., & Boets, B. (2014).Autism as a disconnection syndrome: A qualitative andquantitative review of diffusion tensor imaging studies. Re-search in Autism Spectrum Disorders, 8, 387–412.

Jones, E. J. H., Venema, K., Lowy, R., Earl, R. K., & Webb,S. J. (2015). Developmental changes in infant brain activ-ity during naturalistic social experiences. DevelopmentalPsychobiology, 57, 842–853.

Jonker, F. A., Jonker, C., Scheltens, P., & Scherder, E. J.(2015). The role of the orbitofrontal cortex in cognitionand behavior. Rev Neurosci, 26, 1–11.

Klin, A., Lin, D. J., Gorrindo, P., Ramsay, G., & Jones, W.(2009). Two-year-olds with autism orient to non-socialcontingencies rather than biological motion. Nature, 459,257–61.

Kuhl, P. K., Coffey-Corina, S., Padden, D., Munson, J., Estes,A., & Dawson, G. (2013). Brain responses to words in2-year-olds with autism predict developmental outcomes atage 6. PLoS One, 8, e64967.

Lord, C., Risi, S., Lambrecht, L., Cook, E. H., J., Leventhal,B. L., DiLavore, P. C., Pickles, A., & Rutter, M. (2000). Theautism diagnostic observation schedule-generic: a standardmeasure of social and communication deficits associatedwith the spectrum of autism. J Autism Dev Disord, 30, 205–23. Lord, C Risi, S Lambrecht, L Cook, E H Jr Leventhal,B L DiLavore, P C Pickles, A Rutter, M eng 2000/10/31

8/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

11:00 J Autism Dev Disord. 2000 Jun;30(3):205-23.Luyster, R., Gotham, K., Guthrie, W., Coffing, M., Petrak, R.,

Pierce, K., Bishop, S., Esler, A., Hus, V., Oti, R., Richler,J., Risi, S., & Lord, C. (2009). The autism diagnosticobservation schedule-toddler module: a new module ofa standardized diagnostic measure for autism spectrumdisorders. J Autism Dev Disord, 39, 1305–20.

Meltzoff, A. N. & Moore, M. K. (1977). Imitation of facialand manual gestures by human neonates. Science, 198,74–8.

Menon, V. & Uddin, L. Q. (2010). Saliency, switching, atten-tion and control: a network model of insula function. BrainStruct Funct, 214, 655–67.

Mohammad-Rezazadeh, I., Frohlich, J., Loo, S. K., & Jeste,S. S. (2016). Brain connectivity in autism spectrum disorder.Curr Opin Neurol, 29, 137–47.

Nomi, J. S. & Uddin, L. Q. (2015). Developmental changesin large-scale network connectivity in autism. NeuroimageClin, 7, 732–41.

Ohnishi, T., Matsuda, H., Hashimoto, T., Kunihiro, T.,Nishikawa, M., Uema, T., & Sasaki, M. (2000). Abnormalregional cerebral blood flow in childhood autism. Brain,123 ( Pt 9), 1838–44.

Orekhova, E., Stroganova, T. A., Posikera, I. N., & Elam, M.(2006). Eeg theta rhythm in infants and preschool children.Clinical Neurophysiology, 117, 1047–1062.

Osterling, J. & Dawson, G. (1994). Early recognition ofchildren with autism: A study of first birthday home video-tapes. Journal of Autism and Developmental Disorders, 24,247–257.

Pascual-Marqui, R. D., Michel, C. M., & Lehmann, D. (1994).Low resolution electromagnetic tomography: a new methodfor localizing electrical activity in the brain. Int J Psy-chophysiol, 18, 49–65.

Patriquin, M. A., DeRamus, T., Libero, L. E., Laird, A., &Kana, R. K. (2016). Neuroanatomical and neurofunctionalmarkers of social cognition in autism spectrum disorder.Hum Brain Mapp, 37, 3957–3978.

Pelphrey, K. A. & Carter, E. J. (2008). Charting the typicaland atypical development of the social brain. Developmentand Psychopathology, 20, 1081–1102.

Pelphrey, K. A. & Morris, J. P. (2006). Brain mechanisms forinterpreting the actions of others from biological-motioncues. Curr Dir Psychol Sci, 15, 136–140.

Pelphrey, K. A., Morris, J. P., & McCarthy, G. (2004). Grasp-ing the intentions of others: the perceived intentionality ofan action influences activity in the superior temporal sulcusduring social perception. J Cogn Neurosci, 16, 1706–16.

Pelphrey, K. A., Morris, J. P., Michelich, C. R., Allison, T.,& McCarthy, G. (2005). Functional anatomy of biologicalmotion perception in posterior temporal cortex: an fmristudy of eye, mouth and hand movements. Cereb Cortex,15, 1866–76.

Pelphrey, K. A., Shultz, S., Hudac, C. M., & Vander Wyk,B. C. (2011). Research review: Constraining heterogeneity:

the social brain and its development in autism spectrumdisorder. J Child Psychol Psychiatry, 52, 631–44.

Plomp, G., Astolfi, L., Coito, A., & Michel, C. M. (2015).Spectrally weighted granger-causal modeling: Motivationand applications to data from animal models and epilepticpatients. Conf Proc IEEE Eng Med Biol Soc, 2015, 5392–5.

Plomp, G., Quairiaux, C., Michel, C. M., & Astolfi, L. (2014).The physiological plausibility of time-varying granger-causal modeling: normalization and weighting by spectralpower. Neuroimage, 97, 206–16.

Raschle, N., Zuk, J., Ortiz-Mantilla, S., Sliva, D. D.,Franceschi, A., Grant, P. E., Benasich, A. A., & Gaab,N. (2012). Pediatric neuroimaging in early childhood andinfancy: challenges and practical guidelines. Ann N YAcad Sci, 1252, 43–50.

Reid, V. M., Dunn, K., Young, R. J., Amu, J., Donovan, T.,& Reissland, N. (2017). The human fetus preferentiallyengages with face-like visual stimuli. Curr Biol. Reid,Vincent M Dunn, Kirsty Young, Robert J Amu, JohnsonDonovan, Tim Reissland, Nadja eng England 2017/06/1306:00 Curr Biol. 2017 Jun 2. pii: S0960-9822(17)30580-8.doi: 10.1016/j.cub.2017.05.044.

Saby, J. N. & Marshall, P. J. (2012). The utility of eeg bandpower analysis in the study of infancy and early childhood.Developmental Neuropsychology, 37, 253–273. Sp. Iss. SI936es Times Cited:42 Cited References Count:135.

Schopler, E., Lansing, M. D., Reichler, R. J., & Marcus, L. M.(2005). PEP-3 psychoeducational profile. (Austin: Pro-Ed),3rd edn.

Schumann, C. M., Bloss, C. S., Barnes, C. C., Wideman,G. M., Carper, R. A., Akshoomoff, N., Pierce, K., Ha-gler, D., Schork, N., Lord, C., & Courchesne, E. (2010).Longitudinal magnetic resonance imaging study of corticaldevelopment through early childhood in autism. Journal ofNeuroscience, 30, 4419–4427.

Schwartz, S., Kessler, R., Gaughan, T., & Buckley, A. W.(2016). Electroencephalogram coherence patterns in autism:An updated review. Pediatr Neurol.

Simion, F., Regolin, L., & Bulf, H. (2008). A predisposi-tion for biological motion in the newborn baby. Proc NatlAcad Sci U S A, 105, 809–13. Simion, Francesca Re-golin, Lucia Bulf, Hermann eng 2008/01/05 09:00 ProcNatl Acad Sci U S A. 2008 Jan 15;105(2):809-13. doi:10.1073/pnas.0707021105. Epub 2008 Jan 3.

Sparrow, S. S., Cicchetti, V. D., & Balla, A. D. (2005).Vineland adaptive behavior scales. American GuidanceService.

Sperdin, H. F. & Schaer, M. (2016). Aberrant developmentof speech processing in young children with autism: Newinsights from neuroimaging biomarkers. Front Neurosci,10, 393.

Tierney, A. L., Gabard-Durnam, L., Vogel-Farley, V., Tager-Flusberg, H., & Nelson, C. A. (2012). Developmentaltrajectories of resting eeg power: an endophenotype ofautism spectrum disorder. PLoS One, 7, e39127.

9/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Criv-ello, F., Etard, O., Delcroix, N., Mazoyer, B., & Joliot, M.(2002). Automated anatomical labeling of activations inspm using a macroscopic anatomical parcellation of themni mri single-subject brain. Neuroimage, 15, 273–89.

Uddin, L. Q. (2015). Salience processing and insular corticalfunction and dysfunction. Nat Rev Neurosci, 16, 55–61.

Uddin, L. Q., Supekar, K., Lynch, C. J., Khouzam, A., Phillips,J., Feinstein, C., Ryali, S., & Menon, V. (2013a). Saliencenetwork-based classification and prediction of symptomseverity in children with autism. JAMA Psychiatry, 70,869–79.

Uddin, L. Q., Supekar, K., & Menon, V. (2013b). Recon-ceptualizing functional brain connectivity in autism from adevelopmental perspective. Front Hum Neurosci, 7, 458.

Xia, M., Wang, J., & He, Y. (2013). Brainnet viewer: anetwork visualization tool for human brain connectomics.PLoS One, 8, e68910.

Ye, A. X., Leung, R. C., Schafer, C. B., Taylor, M. J., &Doesburg, S. M. (2014). Atypical resting synchrony inautism spectrum disorder. Human Brain Mapping, 35, 6049–6066.

Yuan, H., Doud, A., Gururajan, A., & He, B. (2008). Corticalimaging of event-related (de)synchronization during onlinecontrol of brain-computer interface using minimum-normestimates in frequency domain. IEEE Trans Neural SystRehabil Eng, 16, 425–31.

AcknowledgementsThe authors would like to express their gratitude to all thefamilies who took part in this research. This research is sup-ported by a grant from the National Centre of Competence inResearch (NCCR) ”SYNAPSY-The Synaptic Bases of MentalDiseases” financed by the Swiss National Science Foundation(number: 51AU40-125759) and by private funding by theFondation Pole Autisme (http://www.pole-autisme.ch). Thiswork was further supported by a SNSF grant to M.S. (num-ber: 163859), as well as SNSF grants number 320030-159705to C.M.M and number 169198 and CRSII5170873 to S.V..R. K. J received individual support from a Marie Curie fel-lowship, which received funding from the European UnionSeventh Framework Programme (FP7:2007-2013) under grantagreement number 267171.

Author contributionsConception and design of the experiment: H.F.S, A.C., S.E.,C.M.M, G.P., T.R and M.S..Acquisition of data H.F.S., T.R,N.K, R.J, and M.F.. Analysis and/or interpretation of data:H.F.S, A.C. and M.S.. Drafting of the manuscript: H.F.S.and A. C.. All authors revised the manuscript critically forimportant intellectual content.

Competing financial interests.All authors declare that the research was conducted in theabsence of any financial or commercial relationships thatcould be construed as a potential conflict of interest.

10/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Figure 1. a. The summed outflow of the largest drivers across frequencies is illustrated for each group (TD, Left; ASD, Right).b. Regions consistently showing large driving in both groups (TD, Left; ASD, Right): the bilateral superior frontal gyri, medialorbital (ORBsupmed.L, ORBsupmed.R), the bilateral superior frontal gyri, orbital (ORBsup.L, ORBsup.R), the bilateral rectusgyri (REC.L, REC.R), the bilateral olfactory cortices (OLF.L,OLF.R), the bilateral hippocampi (HIP.L, HIP.R), the bilateralparahippocampi (PHG.L, PHG,R), the bilateral anterior parts of the cingulate gyri (ACG.L, ACG.R) and the right amygdala(AMYG.R). Summed outflows are represented as spheres: the larger the sphere, the higher the summed outflow.

11/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Figure 2. The 6 ROIs with a statistically significant increased summed outflow in the ASD group compared to their TD peers:the right orbital part of the superior frontal gyrus (ORBsup.R), the bilateral orbital parts of the middle frontal gyri (ORBmid.L,ORBmid.R), the right middle cingulate gyrus (DCG.R), the left superior occipital gyrus (SOG.L), and the left superior temporalgyrus (TPOsup.L).

12/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Figure 3. Region-to-region directed functional connectivity in TD group (left) and in ASD group (right). Connections strengthand networks are different between the groups. Exemplary node of the right orbital part of the superior frontal gyrus (ORBsup.R)(in large red). For the TD group, the ORBsup.R drove activity towards the left Superior frontal gyrus, dorsolateral (SFGdor.L),the left middle frontal gyrus (MFG.L), the right inferior frontal gyrus, opercular (IFGoperc.R), the left hippocampus (HIP.L),the left cuneus (CUN.L), the left inferior parietal gyrus (IPL.L), the left supramarginal gyrus (SMG.L), the right angular gyrus(ANG.R), the right precuneus (PCUN.R) and the right paracentral lobule (PCL.R). In the ASD group, the node is driving to theleft inferior frontal gyrus, triangular (IFGtriang.L), the right inferior frontal gyrus, orbital (ORBinf.R), the right supplementarymotor area (SMA.R), the bilateral superior frontal gyrus, medial orbital (ORBsupmed.L ORBsupmed.R), the right hippocampus(HIP.R), the left amygdala (AMYG.L), the right lingual gyrus (LING.R), the supramarginal gyrus (SMG.L) and the paracentrallobule (PCL.L).

13/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Figure 4. Correlations between the summed outflows and (a.) Proximity Index (b.) VABS-II scores and (c.) PEP-3 scores.

14/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Figure 5. Exemplar single time frame of the normative gaze pattern for each group on one random time frame. Each dotrepresents the gaze position for an individual participant.

15/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Figure 6. The general analysis strategy.

16/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Figure 7. The directed connections from the five remaining ROIs : the left orbital part of the middle frontal gyrus, the rightmiddle cingulate gyrus, the left superior occipital gyrus and the left superior temporal gyrus

17/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint

Table 1. Characteristics of Study Participants

18/18

.CC-BY-NC-ND 4.0 International licensecertified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprint (which was notthis version posted August 30, 2017. . https://doi.org/10.1101/180703doi: bioRxiv preprint