Adding insult to injury neural sensitivity to social

exclusion is associated with internalizing symptoms

in chronically peer-victimized girlsKaren D Rudolph1 Michelle E Miernicki1 Wendy Troop-Gordon2

Megan M Davis1 and Eva H Telzer1

1Department of Psychology University of Illinois Urbana-Champaign 603 E Daniel St Champaign IL 61820USA and 2North Dakota State University

Correspondence should be addressed to Karen D Rudolph Department of Psychology University of Illinois 603 E Daniel St Champaign IL 61820 USAE-mail krudolphillinoisedu

Abstract

Despite evidence documenting activation of the social pain network in response to social rejection and its link to temporarydistress far less is known regarding its role in pervasive emotional difficulties Moreover research has not considered theintersection between neural activation to experimentally induced social exclusion and naturally occurring social adversityThis study examined an integrated social pain model of internalizing symptoms which posits that (i) neural sensitivity inthe social pain network is associated with internalizing symptoms (ii) this linkage is more robust in youth with thanwithout a history of social adversity and (iii) heightened avoidance motivation serves as one pathway linking neuralsensitivity and internalizing symptoms During a functional magnetic resonance imaging scan 47 adolescent girls(M agefrac141546 years SDfrac14 35) with well-characterized histories of peer victimization were exposed to social exclusionWhole-brain analyses revealed that activation to exclusion in the social pain network was associated with internalizingsymptoms As anticipated this linkage was stronger in chronically victimized than non-victimized girls and was partiallyaccounted for by avoidance motivation This research indicates the importance of integrating neural social andpsychological systems of development in efforts to elucidate risk for internalizing symptoms among adolescent girls

Key words neural sensitivity internalizing victimization avoidance

Introduction

Humans have a fundamental need for social connection asembodied in concepts such as need for affiliation (McClellandet al 1953) need to belong (Baumeister and Leary 1995) andneed for approval (Rudolph Caldwell amp Conley 2005)Relationships are particularly crucial for development duringadolescence a stage of acute sensitivity to social cues (Rudolph2009 Guyer et al 2012 Rudolph et al in press) Unfortunatelywhen social conditions threaten these core needs individualsexperience lsquosocial painrsquo with neural sequelae akin to those ofphysical pain (Eisenberger 2012) Although much researchestablishes activation of the social pain network in response to

social rejection and its association with temporary distress farless is known regarding its role in more pervasive emotional dif-ficulties such as internalizing symptoms Yet sensitivity in thisnetwork is a prime candidate for vulnerability to internalizingsymptoms and may even help to explain their dramatic rise inadolescent girls who show heightened reactivity to social stres-sors in the form of depression and anxiety (Davila et al 2010Rudolph et al in press)

The present research used functional magnetic resonanceimaging (fMRI) to elucidate the role of the social pain network ininternalizing symptoms with a focus on individual differencesin vulnerability to these effects and psychological processes

Received 6 May 2015 Revised 17 December 2015 Accepted 11 February 2016

VC The Author (2016) Published by Oxford University Press For Permissions please email journalspermissionsoupcom

829

Social Cognitive and Affective Neuroscience 2016 829ndash842

doi 101093scannsw021Advance Access Publication Date 17 February 2016Original article

that may explain this vulnerability In particular we testeda comprehensive model which posits that (i) neural sensitivityin the social pain network is associated with internalizingsymptoms (depression and social anxiety) (ii) this linkage ismore robust in youth with than without a history of social ad-versity (exposure to chronic peer victimization) and (c) height-ened avoidance motivation (psychological sensitivity to socialpunishment) serves as one pathway linking neural sensitivityand internalizing symptoms

Social pain network

A growing body of research suggests that exposure to acute so-cial exclusion in the laboratory triggers activation in the sameneural circuitry as that underlying the affective component ofphysical pain (Lieberman and Eisenberger 2006 Dewall et al2010 Eisenberger 2012) Several regions have been implicatedin the social pain network including the dorsal anterior cingu-late cortex (dACC) the subgenual anterior cingulate cortex(sgACC) and the anterior insula (Sebastian et al 2001 Mastenet al 2009 Masten et al 2011 for a review see Rotge et al 2014Eisenberger 2015) Eisenberger et al (2011) propose that theseregions serve as a neural alarm system or lsquosociometerrsquo alertingindividuals to a discrepancy between their desired social state(social acceptance) and current social conditions (social rejec-tion) This pattern of neural activation is exaggerated in youthexposed to chronic peer stress (Will et al 2016) and is associatedwith indicators of temporary distress (for a review see Rotgeet al 2014) including self-reported distress during exclusion(eg Eisenberger 2012) and greater threat to onersquos psychologicalneeds (eg Eisenberger et al 2003)

Neural sensitivity to rejection andinternalizing symptoms

Given the affective distress accompanying neural sensitivity tosocial rejection recent conceptualizations suggest that aheightened social pain response may constitute a risk factor forinternalizing symptoms particularly during adolescence(Masten et al 2011 Rotge et al 2014 Silk et al 2014) Althoughresearch linking activation of the social pain network and inter-nalizing symptoms is scarce two studies provide initial sup-portive evidence In one study adolescents diagnosed withmajor depressive disorder relative to healthy adolescents dem-onstrated heightened sgACC and left anterior insula activationand more sustained dACC activation in response to rejectionduring a chatroom task (Silk et al 2014) In another studysgACC activation to exclusion during Cyberball (Williams et al2000) predicted higher levels of depressive symptoms over a1-year period in a sample of adolescents (Masten et al 2011)Adolescents at risk for social anxiety disorder (ie those withhigh early levels of behavioral inhibition) and those with highlevels of general and social anxiety also demonstrate differ-ences in neural responses to peer rejection (for a review seeCaouette and Guyer 2014 Guyer et al 2015) with one studyfinding significantly stronger insula activation in anxious thannon-anxious youth (Lau et al 2012) To expand on this researchour first goal was to provide additional evidence that height-ened neural sensitivity to exclusion (social pain) is associatedwith internalizing symptoms as reflected in symptoms of de-pression and social anxiety (Hypothesis 1)

Targeted social rejection andinternalizing symptoms

Outside of the laboratory interpersonal theories of internalizingsymptoms also highlight social rejection as a mechanism of riskfor depression and social anxiety Slavich and colleaguesrsquo lsquoblacksheeprsquo theory of depression (Slavich et al 2009 2010) focuses onthe role of targeted social rejection or naturally occurring stres-sors involving a direct threat to onersquos sense of belongingTargeted rejection stressors involve three dimensions (i) an in-tent to reject or stressors characterized by an active and inten-tional severing of relational ties (ii) an isolated impact orstressors directed at and meant to affect a single person and(iii) social demotion or stressors involving a loss of social sta-tus This theory proposes that targeted rejection stressors play acentral role in depression Because these stressors triggerheightened social-evaluative threat devaluation of the self andsocial avoidance (Slavich et al 2009 2010) they also are likelyrisk factors for social anxiety Other models of social anxietyalso implicate peer rejection as a robust risk factor (Davila et al2010 La Greca et al 2011 Caouette and Guyer 2014)

Peer victimization is a relatively common form of social stressin childhood (Boivin et al 2010) that shares the three key featuresof targeted social rejection Specifically peer victimization in-volves purposeful and direct efforts to reject individuals throughphysical andor psychological means (eg exposure to hittingverbal assaults rumor-spreading social manipulation and activeexclusion Crick and Grotpeter 1996) moreover peer-victimizedyouth experience a loss of social status over time (Kochel et al2012) Consistent with theories of targeted rejection both earlyoccurring and chronic victimization are robust predictors of de-pressive symptoms (Rudolph et al 2011) and social anxietyavoidance (Siegel et al 2009 Rudolph et al 2014)

Integrating laboratory and naturalisticresearch on targeted rejection

Typically social rejection has been conceptualized and studiedeither in terms of the neural effects of acute social exclusion inthe laboratory or the psychological effects of naturally occurringtargeted rejection The present study represents one of the firstefforts to integrate these two lines of research to provide a com-prehensive picture regarding the joint role of neural sensitivityand targeted rejection in risk for internalizing symptomsSpecifically we anticipated that exposure to chronic peer vic-timization would impart a social lsquobruisersquo that intensifies the ef-fect of subsequent social insults thereby amplifying theassociation between acute neural sensitivity to rejection andinternalizing symptoms Thus our second goal was to investi-gate whether heightened neural sensitivity to exclusion (socialpain) is more strongly associated with internalizing symptomsamong youth with a history of chronic peer victimization thanamong youth with minimal exposure to peer victimization(Hypothesis 2)

Avoidance motivation as anexplanatory pathway

A third goal of this study was to better understand the psycho-logical pathway linking neural sensitivity to exclusion withinternalizing symptoms Although it is clear why heightenedactivation in the social pain network might lead to temporaryincreases in emotional distress why would this acute sensitiv-ity foster pervasive and persistent emotional difficulties such as

830 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

depressive symptoms and social anxiety To answer this ques-tion we focused on avoidance motivation as one explanatoryprocess Avoidance motivation can be conceptualized as a psy-chological sensitivity to aversive aspects of the social environ-ment In this study we incorporated three aspects of avoidancemotivation general avoidance motivation construed as a driveto avoid exposure to threat punishment and loss (Gray 1991Carver and White 1994) performance-avoidance goals con-strued as a drive to avoid negative judgments and loss of statusin the peer group (Rudolph et al 2013 Llewellyn and Rudolph2014) and avoidance-oriented need for approval construed as adepletion of onersquos self-worth in the face of social disapproval(Rudolph et al 2005 Rudolph and Bohn 2014)

An enhanced social pain response (ie greater neural activa-tion in response to rejection) may be reflected in an accompany-ing psychological sensitivity to aversive social cues in the form ofavoidance motivation Indeed individuals who show heighteneddACC activation in response to acute social exclusion also dem-onstrate pro-inflammatory responses to social evaluation(Slavich et al 2010) suggesting that social pain is associated withactivation of the stress response system in the context of socialevaluation Heightened social pain responses also are linked toself-reported rejection sensitivity (Burklund et al 2007 Mastenet al 2009) and an anxious attachment style marked by vigilanceto rejection cues (DeWall et al 2012) Thus youth who show aheightened social pain response may become sensitive to socialcues of evaluation and potential rejection may formulate socialgoals aimed at avoiding such adverse social judgments and maydevelop a sense of worth that is threatened by disapprovalChronically victimized youth with a heightened social pain re-sponse may be especially prone to developing these forms of per-sistent psychological sensitivity given their history of socialmaltreatment In turn heightened avoidance motivation predictsdepressive symptoms (Coplan et al 2006 Rudolph et al 2013Llewellyn and Rudolph 2014) and social anxietywithdrawal(Coplan et al 2006 Rudolph and Bohn 2014 for a review seeCaouette and Guyer 2014) We therefore hypothesized thatavoidance motivation would serve as one pathway linking neuralsensitivity to exclusion with internalizing symptoms among vic-timized youth (Hypothesis 3)

Study overview

To address these research questions we recruited adolescentgirls who had well-characterized histories of victimizationthrough the school years (2ndndash8th grades) We focused on thisgroup because adolescence is a stage of heightened neural sen-sitivity (ie heightened affective processing during peer evalu-ation) and psychological sensitivity (ie anxiety about peeracceptance and evaluation) to social threat particularly in girls(Nelson et al 2005 Guyer et al 2009 Guyer et al 2012)Moreover relative to boys girls show a dramatic rise in depres-sive symptoms (Hankin and Abramson 2001) and social anxiety(Nelemans et al 2014) during adolescence as well as a strongercontribution of interpersonal stress and social sensitivity tointernalizing symptoms (Gunnar et al 2009 Rudolph 2009Davila et al 2010)

Materials and methodsParticipants and procedures

Participants included 47 adolescent girls who were recruitedfrom a longitudinal study tracking youth from 2ndndash8th grade

(Mage second gradefrac14 791 years sdfrac14 095 years for detailsabout the longitudinal study see Rudolph Lansford et al 2014Rudolph Troop-Gordon et al 2014 Troop-Gordon et al 2015)Four additional girls were scanned but not included due to ei-ther a malfunction in the Cyberball program or missing data onkey measures Based on youthsrsquo annual reports ofvictimization across the 7 years we recruited 24 chronicallyvictimized girls (Magefrac14 1546 years sdfrac14 035) and 23 non-victimized girls (Magefrac14 1535 years sdfrac14 037) Chronically victi-mized girls scored 075 sd above the mean on victimizationfor at least 3 of 7 years (averagefrac14 433 years rangefrac14 3ndash7 years)with an average victimization score of 122 sd above the meanNon-victimized girls scored 075 sd below the mean on vic-timization for at least 3 of 7 years (averagefrac14 483 yearsrangefrac14 3ndash7 years) with an average victimization score of082 sd below the mean Parents provided written consent andadolescents provided written assent in accordance with theUniversity of Illinoisrsquo Institutional Review Board During thesummer following ninth grade participants completed a func-tional brain scan while playing Cyberball (Williams et al 2000)a well-established laboratory manipulation of acute social ex-clusion Following the scan they completed measures of de-pressive symptoms and social anxiety The participantsreceived a monetary incentive for their participation

Self-report measures

Peer victimization During the 2ndndash8th grades participants com-pleted a 21-item revised version (for details see Rudolph et al2014) of the Social Experiences Questionnaire (Crick andGrotpeter 1996) to assess exposure to peer victimization Thismeasure assesses overt victimization (being the target of behav-iors intended to harm others through physical damage threatof such damage or verbal aggression 11 items eg lsquoHow oftendo you get hit by another kidrsquo lsquoHow often does another kid in-sult you or put you downrsquo) and relational victimization (beingthe target of behaviors intended to harm others through ma-nipulation of relationships 10 items eg lsquoHow often does an-other kid say they wonrsquot like you unless you do what they wantyou to dorsquo) Youth checked a box indicating how often theyexperienced each type of victimization on a 5-point scaleScores were computed as the mean of the 21 items

Internalizing symptoms Two measures were used to assessinternalizing symptoms at the time of the scan First youthcompleted the Short Mood and Feelings Questionnaire (Angoldet al 1995) to assess depressive symptoms (eg lsquoI felt unhappyor miserablersquo) Youth indicated how much they experiencedeach symptom on a 4-point scale Scores were computed as themean of the 13 items Second youth completed the SocialAnxiety Scale for Adolescents (La Greca and Lopez 1998) to as-sess social anxiety (eg lsquoIrsquom afraid to invite peers to do thingswith me because they might say norsquo) Youth indicated howmuch they experienced each symptom on a 5-point scaleScores were computed as the mean of the 18 items Because wehad similar hypotheses for depressive symptoms and socialanxiety and the measures were strongly correlated (rfrac14 042Pfrac14 0004) a composite variable was formed by standardizingand averaging the two measures

Avoidance motivation Three measures were used to assessavoidance motivation at the time of the scan First youth com-pleted a slightly revised version (for details see Rudolph et al2013) of the Behavioral Inhibition Scale (BIS) (Carver and White

K D Rudolph et al | 831

1994 Muris et al 2005) The BIS subscale assesses general sensi-tivity to aversive stimuli although some items include a socialfocus (eg lsquoI feel hurt when people scold me or tell me that I didsomething wrongrsquo) Youth indicated how true each item was ona 4-point scale Scores were computed as the mean of the sevenitems Second youth completed the performance-avoidancesubscale of the Social Achievement Goals Survey (Rudolph andAbaied et al 2011) which assesses goals that focus on demon-strating competence by avoiding negative social judgments (eglsquoMy main goal is to make sure I donrsquot look like a loserrsquo) Youthreceived the prompt lsquoWhen I am around other kids rsquo and indi-cated how true each item was on a 5-point scale Scores werecomputed as the mean of the seven items Third youth com-pleted the avoidance subscale of the Need for ApprovalQuestionnaire (Rudolph et al 2005) which assesses sensitivityto peer disapproval (the extent to which peer disapproval weak-ens onersquos sense of self-worth eg lsquoI feel like I am a bad personwhen other kids donrsquot like mersquo) Youth indicated how true eachitem was on a 5-point scale Scores were computed as the meanof the four items Because the three indexes of avoidance motiv-ation assessed similar constructs and were strongly correlated(rrsquos ranged from 039 to 070 all Prsquoslt 0001) a composite variablewas formed by standardizing and averaging the three measuresSimilar composites have established validity (eg Rudolph et al2013 Llewellyn and Rudolph 2014)

fMRI task

While completing the scan participants were exposed to socialrejection using Cyberball (Williams et al 2000) which creates asubjective experience of being excluded Participants were toldthey would be playing an on-line ball-throwing game with twopeers (ostensibly in another room) also completing the samestudy and connected via the Internet Participants could see thephotographs of the other two players on a computer screen aswell as their own lsquohandrsquo that they controlled using a button-box Throughout the game the ball is thrown back and forthamong the three players When the participant received theball she returned it to either player by pushing one of two but-tons The throws of the other two lsquoplayersrsquo were determined bythe pre-set program Each participant completed two rounds Inthe inclusion round she was equally included in the tosses Inthe exclusion round she was excluded after 10 tosses Providingvalidity for the exclusion experience chronically victimizedgirls (Mfrac14 318 sdfrac14 085) reported feeling significantly morethreat to their need to belong following Cyberball than non-vic-timized girls [Mfrac14 263 sdfrac14 058 t(45)frac14 254 Pfrac14 002 dfrac14 074]measured by the Need-Threat Scale (Williams et al 2000) a 12-item self-report measure (higher scores reflect more threat toonersquos needs) assessing feelings of rejection (eg lsquoI felt rejectedrsquo)belongingness (eg lsquoI felt disconnectedrsquo) self-esteem (eg lsquoI feltgood about myselfrsquo) and social control (eg lsquoI felt powerfulrsquo)

fMRI data acquisition and analysis

fMRI data acquisition Imaging data were collected using a 3Tesla Siemens Trio MRI scanner The Cyberball task includedT2-weighted echoplanar images (EPI) (slice thicknessfrac14 3 mm38 slices TR (temporal resolution) frac14 2 s TE (echo time) frac14 25 msmatrixfrac14 92 92 FOV (field of view) frac14 230 mm voxel size25 25 3 mm3) Structural scans consisted of a T2weightedmatched-bandwidth (MBW) high-resolution anatomical scan(TRfrac14 4 s TEfrac14 64 ms FOVfrac14 230 matrixfrac14 192 192 slice thick-nessfrac14 3 mm 38 slices) and a T1 magnetization-prepared rapid-

acquisition gradient echo (MPRAGE TRfrac14 19 s TEfrac14 23 msFOVfrac14 230 matrixfrac14 256 256 sagittal plane slice thick-nessfrac14 1 mm 192 slices) The orientation for the MBW and EPIscans was oblique axial to maximize brain coverage

fMRI data preprocessing and analysis Neuroimaging data werepreprocessed and analyzed using Statistical ParametricMapping (SPM8 Wellcome Department of Cognitive NeurologyInstitute of Neurology London UK) Preprocessing for each par-ticipantrsquos images included spatial realignment to correct forhead motion (no participant exceeded 2 mm of maximumimage-to-image motion in any direction) The realigned func-tional data were coregistered to the high resolution MPRAGEwhich was then segmented into cerebrospinal fluid gray matterand white matter The normalization transformation matrixfrom the segmentation step was then applied to the functionaland T2 structural images thus transforming them into standardstereotactic space as defined by the Montreal NeurologicalInstitute and the International Consortium for Brain MappingThe normalized functional data were smoothed using an 8 mmGaussian kernel full-width-at-half maximum to increase thesignal-to-noise ratio

Statistical analyses were performed using the general linearmodel (GLM) in SPM8 The task was modeled as a block designwith two blocks inclusion and exclusion High-pass temporalfiltering with a cutoff of 128 s was applied to remove low-frequency drift in the time series Serial autocorrelations wereestimated with a restricted maximum-likelihood algorithm withan autoregressive model order of 1 The parameter estimates re-sulting from the GLM were used to create linear contrast imagescomparing exclusion to inclusion Random effects group-levelanalyses were performed on all individual subject contrasts

To correct for multiple comparisons we conducted a MonteCarlo simulation implemented using 3dClustSim in the soft-ware package AFNI (Ward 2000) We used our group-level brainmask which included only gray matter Results of the simula-tion indicated a voxel-wise threshold of Plt 0005 combinedwith a minimum cluster size of 42 voxels for the whole braincorresponding to Plt 005 False Wise Error corrected

Overview of analyses

At the group level our primary analysis examined neural acti-vation for the contrast exclusionndashinclusion To examine the as-sociation between neural activation during exclusion andinternalizing symptoms (Hypothesis 1) we conducted whole-brain regression analyses in which we regressed internalizingsymptoms for the whole sample (Nfrac14 47) onto neural activationduring exclusionndashinclusion We then extracted parameter esti-mates of signal intensity from the clusters of activation fora priori regions of interest (ROIs ie dACC sgACC and bilateralinsula) that showed significant correlations with internalizingsymptoms These values were used in a series of path analysesto test whether (i) heightened neural activation to exclusion(specifically in regions linked to social pain) would be morestrongly associated with internalizing symptoms in victimizedthan non-victimized girls (Hypothesis 2) and (ii) higher levels ofavoidance motivation would account in part for the link be-tween neural activation and internalizing symptoms in victi-mized girls (Hypothesis 3) These analyses were conducted inMplus (Muthen and Muthen 1998ndash2007) using full informationmaximum likelihood (Enders and Bandalos 2001)

Figure 2 presents the integrated conceptual model For eachindex of neural sensitivity (dACC sgACC and insula activation)

832 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

a separate path model was estimated We conducted model-testing in two steps To test Hypothesis 2 Step 1 examined thetotal effect of neural sensitivity on internalizing symptoms andits moderation by victimization status by setting the paths toand from avoidance motivation to 0 This step allowed us to es-timate the total effect of neural sensitivity on internalizingsymptoms for victimized and non-victimized girls To testHypothesis 3 Step 2 examined the extent to which this total ef-fect could be accounted for by an indirect effect through avoid-ance motivation Specifically we freely estimated the path fromneural sensitivity to avoidance motivation (Path b) and the pathfrom avoidance motivation to internalizing symptoms (Path c)To determine whether moderation of the total effect of neuralsensitivity on internalizing symptoms could be accounted forby a difference in the neural sensitivityndashavoidance motivationlink among victimized vs non-victimized girls victimizationstatus served as moderator of Path b Path a and its moderationby victimization status also were estimated in Step 2 providingan estimate of the remaining direct effect of neural sensitivityon internalizing symptoms after taking into account the indir-ect effect (ie whether neural sensitivity was linked to internal-izing symptoms above and beyond the path through avoidancemotivation)

This two-step model-testing approach allowed us to test thesignificance of (i) the conditional total effect of neural sensitiv-ity on internalizing symptoms (ie the effect for victimized andnon-victimized girls Step 1) (ii) the conditional indirect effectof neural sensitivity on internalizing symptoms through avoid-ance motivation (ie Path b estimated separately for victimizedand non-victimized girls Path c Step 2) and (iii) the conditionaldirect effect of neural sensitivity on internalizing symptoms (iethe direct effect for victimized and non-victimized girls after ac-counting for the indirect effect in Step 2) In sum this analysisexamined the total effect (ie directthorn indirect effects) of neuralsensitivity on internalizing symptoms the indirect effect viaavoidance motivation and the remaining direct effect as well aswhether these effects were significant for victimized and non-victimized girls (Preacher et al 2007)

ResultsDescriptive statistics

Table 1 presents descriptive data and group comparisons for2ndndash8th grade peer victimization in the longitudinal study aswell as depressive symptoms social anxiety and the threeindexes of avoidance motivation at the time of the scanVictimized girls reported more 2ndndash8th grade peer victimiza-tion all drsquosgt 155 as well as higher levels of depressive symp-toms dfrac14 137 social anxiety dfrac14 089 behavioral inhibitiondfrac14 095 and avoidance-oriented need for approval dfrac14 062 atthe time of the scan than did non-victimized girls They did notreport more performance avoidance than did non-victimizedgirls dfrac14 012 As shown in Supplementary Figure S1 victimizedgirls showed greater activation in the dACC during exclusionrelative to inclusion than non-victimized girls as well as greateractivation in the amygdala and inferior fusiform gyrus(Supplementary Table S1)

Association between neural activationand internalizing symptoms

In whole-brain regression analyses we regressed internalizingsymptoms onto neural activation during exclusionndashinclusion

Consistent with Hypothesis 1 greater activation in the socialpain network including the dACC sgACC and anterior insulawas associated with heightened internalizing symptoms (Table2) For descriptive purposes we extracted parameter estimatesof signal intensity from these regions and plotted the associ-ation for neural activation and internalizing symptoms in thetotal sample as well as within victimized and non-victimizedgirls (Figure 1) As reflected in Table 2 regions outside of the so-cial pain network also showed heightened activation in relationto internalizing symptoms For parsimony follow-up analyseswere conducted only with the three a priori ROIs

Tests of the full model

To test the model depicted in Figure 2 (Hypotheses 2 and 3) weextracted parameter estimates of signal intensity from each ofthe three ROIs that showed a significant correlation with inter-nalizing symptoms for the whole sample For parsimony wetook an average of the standardized values for the two regionsof activation in the dACC and sgACC respectively creating asingle score for each region We then conducted a separate pathanalysis for each ROI In Step 1 only paths reflecting the mainand interactive effects of neural sensitivity and victimizationstatus on internalizing symptoms were estimated Moderationof the effect of neural sensitivity on internalizing symptoms byvictimization status was tested by including a NeuralSensitivityVictimization Status interaction term This stepprovided an estimate of the conditional total effect of neuralsensitivity on internalizing symptoms for victimized and non-victimized girls In Step 2 the paths reflecting the indirecteffects of neural sensitivity victimization status and theirinteraction on avoidance motivation also were estimatedModeration of the effect of neural sensitivity on avoidance mo-tivation by victimization status was tested by including aNeural SensitivityVictimization Status interaction term Thisstep provided an estimate of the extent to which the conditionaltotal effect was accounted for by a conditional indirect effectthrough avoidance motivation (for additional detail seeOverview of Analyses) Table 3 presents results of theseanalyses

Model 1 DACC activation To determine whether dACC activationwas significantly associated with internalizing symptoms andwhether victimization status moderated this association (Step1) we first tested the model setting Path b Path c and moder-ation of Path b by Victimization Status equal to 0 (Figure 2) Thisprovided a test of the conditional total effect (ie the conditionaleffect of dACC activation on internalizing symptoms withoutconsidering avoidance) The analysis yielded a significant maineffect for victimization status bfrac14 077 SEfrac14 014 Plt 0001 and asignificant dACCVictimization Status interaction bfrac14 054SEfrac14 017 Pfrac14 0002 The total effect of dACC activation on inter-nalizing symptoms was significant for victimized girls bfrac14 070SEfrac14 012 Plt 0001 but not for non-victimized girls bfrac14 016SEfrac14 012 Pfrac14 020 (Figure 3A)

In Step 2 we tested the extent to which the conditional totaleffect of dACC activation on internalizing symptoms could beaccounted for by a conditional indirect effect through avoid-ance The paths from dACC activation to avoidance (Path b) andfrom avoidance to internalizing symptoms (Path c) were esti-mated as was the path from the dACCVictimization Statusinteraction to avoidance The left column in Table 3 presentsthe results of this analysis A significant dACCVictimizationStatus interaction emerged in the prediction of avoidance

K D Rudolph et al | 833

bfrac14 055 SEfrac14 019 Pfrac14 0004 showing that the effect of dACC ac-tivation on avoidance (Path b) was significantly moderated byvictimization status The effect of dACC activation on avoidancewas significant for victimized girls bfrac14 075 SEfrac14 014 Plt 0001but not for non-victimized girls bfrac14 019 SEfrac14 014 Pfrac14 017Moreover avoidance significantly predicted internalizing symp-toms bfrac14 034 SEfrac14 012 Pfrac14 0005 This resulted in a significantconditional indirect effect of dACC activation on internalizingsymptoms through avoidance for victimized girls bfrac14 025SEfrac14 010 Pfrac14 001 but not for non-victimized girls bfrac14 006SEfrac14 005 Pfrac14 022 The dACCVictimization Status interactionpredicting internalizing symptoms was smaller after accountingfor the conditional indirect effect but remained significantbfrac14 036 SEfrac14 017 Pfrac14 004 The direct effect of dACC activationon internalizing symptoms was significant for victimized girlsbfrac14 045 SEfrac14 014 Plt 0001 but not for non-victimized girlsbfrac14 009 SEfrac14 012 Pfrac14 042

In sum these results confirm the hypothesis that dACC acti-vation predicts internalizing symptoms in victimized but notnon-victimized girls As expected the effect of dACC activationon internalizing symptoms was partly explained by a link be-tween dACC activation and avoidance motivation for victimizedbut not non-victimized girls However there remained a directeffect of dACC activation on internalizing symptoms for victi-mized girls that was not accounted for by avoidancemotivation

Model 2 sgACC activation To determine whether sgACC was sig-nificantly associated with internalizing symptoms and whethervictimization status moderated this association (Step 1) we firsttested the model setting Path b Path c and moderation of Path bby Victimization Status equal to 0 (Figure 2) This provided atest of the conditional total effect The analysis yielded a signifi-cant main effect for victimization status bfrac14 077 SEfrac14 016Plt 0001 and a significant sgACCVictimization Status inter-action bfrac14 048 SEfrac14 020 Pfrac14 002 The total effect of sgACC acti-vation on internalizing symptoms was significant for victimizedgirls bfrac14 051 SEfrac14 011 Plt 0001 but not for non-victimizedgirls bfrac14 003 SEfrac14 017 Pfrac14 084 (Figure 3B)

In Step 2 we tested the extent to which the conditional totaleffect of sgACC activation on internalizing symptoms could be

accounted for by a conditional indirect effect through avoid-ance The paths from sgACC activation to avoidance (Path b)and from avoidance to internalizing symptoms (Path c) wereestimated as was the path from the sgACCVictimizationStatus interaction to avoidance The middle column in Table 3presents the results of this analysis Although thesgACCVictimization Status interaction did not significantlypredict avoidance bfrac14 033 SEfrac14 025 Pfrac14 020 the conditional ef-fect of sgACC activation on avoidance (Path b) was significantfor victimized girls bfrac14 038 SEfrac14 013 Pfrac14 0004 but not for non-victimized girls bfrac14 006 SEfrac14 022 Pfrac14 079 Moreover avoidancesignificantly predicted internalizing symptoms bfrac14 044SEfrac14 010 Plt 0001 This resulted in a significant conditional in-direct effect of sgACC activation on internalizing symptomsthrough avoidance for victimized girls bfrac14 017 SEfrac14 007Pfrac14 001 but not for non-victimized girls bfrac14 003 SEfrac14 010Pfrac14 079 The sgACCVictimization Status interaction predict-ing internalizing symptoms was smaller after accounting forthe conditional indirect effect but remained significantbfrac14 033 SEfrac14 017 Pfrac14 005 The direct effect of sgACC activationon internalizing symptoms was significant for victimized girls034 SEfrac14 010 Plt 0001 but not for non-victimized girls 001SEfrac14 014 Pfrac14 095

In sum these results confirm the hypothesis that sgACC ac-tivation predicts internalizing symptoms in victimized but notnon-victimized girls As expected the effect of sgACC activationon internalizing symptoms was partly explained by a link be-tween sgACC activation and avoidance motivation for victi-mized but not non-victimized girls However there remained adirect effect of sgACC activation on internalizing symptoms forvictimized girls that was not accounted for by avoidancemotivation

Model 3 Insula activation To determine whether insula activa-tion was significantly associated with internalizing symptomsand whether victimization status moderated this association(Step 1) we first tested the model setting Path b Path c andmoderation of Path b by Victimization Status equal to 0 (Figure2) This provided a test of the conditional total effect The ana-lysis yielded a significant main effect for victimization statusbfrac14 067 SEfrac14 019 Plt 0001 and a marginally significant

Table 1 Descriptive statistics and psychometrics for victimized and non-victimized girls

Victimized girls Non-victimized girls

Variable M sd M sd t-test (df) a

Second grade peer victimization 298 092 187 048 507 (42) 093Third grade peer victimization 294 087 147 045 715 (44) 096Fourth grade peer victimization 267 071 131 030 828 (42) 095Fifth grade peer victimization 254 068 122 028 866 (45) 095Sixth grade peer victimization 245 068 117 021 859 (45) 095Seventh grade peer victimization 245 055 107 008 1180 (45) 096Eighth grade peer victimization 244 058 110 014 1095 (45) 096Depressive symptoms 203 071 121 051 454 (44) 095Social anxiety 253 091 184 064 299 (45) 096Behavioral inhibition 263 072 203 054 319 (45) 083Performance avoidance 208 110 198 060 39 (44) 090Avoidance-oriented NFA 186 109 135 049 207 (45) 089

Note NFA need for approval

Plt005 Plt001 Plt0001

834 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

Fig 1 Greater activation in the dACC sgACC and anterior insula predicts higher levels of internalizing symptoms (A) dACC activation during the exclusion condition

compared with the inclusion condition that was positively correlated with internalizing symptoms (B) sgACC activation during the exclusion condition compared with

the inclusion condition that was positively correlated with internalizing symptoms (C) anterior insula activation during the exclusion condition compared with the in-

clusion condition that was positively correlated with internalizing symptoms Note In the scatterplots the solid black line indicates the trend line for the entire sample

the dashed black line indicates the trend line for victims and the dashed gray line indicates the trend line for non-victims

K D Rudolph et al | 835

InsulaVictimization Status interaction bfrac14 036 SEfrac14 020Pfrac14 007 As expected the total effect of insula activation oninternalizing symptoms was significant for victimized girlsbfrac14 043 SEfrac14 015 Plt 0004 but not for non-victimized girlsbfrac14 007 SEfrac14 013 Pfrac14 057 (Figure 3C)

In Step 2 we tested the extent to which the conditional totaleffect of insula activation on internalizing symptoms could be ac-counted for by a conditional indirect effect through avoidanceThe paths from insula activation to avoidance (Path b) and fromavoidance to internalizing symptoms (Path c) were estimated aswas the path from the InsulaVictimization Status interaction toavoidance The right column in Table 3 presents the results ofthis analysis Although the InsulaVictimization Status inter-action did not significantly predict avoidance bfrac14 029 SEfrac14 021ns the conditional effect of insula activation on avoidance (Pathb) was significant for victimized girls bfrac14 054 SEfrac14 016 Plt 0001and marginally significant for non-victimized girls bfrac14 025SEfrac14 013 Pfrac14 006 Moreover avoidance significantly predictedinternalizing symptoms bfrac14 055 SEfrac14 012 Plt 0001 This resultedin a significant conditional indirect effect of insula activation oninternalizing symptoms through avoidance for victimized girls

bfrac14 030 SEfrac14 011 Pfrac14 0005 and a marginally significant condi-tional indirect effect for non-victimized girls bfrac14 014 SEfrac14 008Pfrac14 008 After accounting for the conditional indirect effect theInsulaVictimization Status interaction did not significantly pre-dict internalizing symptoms bfrac14 020 SEfrac14 017 Pfrac14 022 and thedirect effect of insula activation on internalizing symptoms wasnonsignificant for victimized girls bfrac14 014 SEfrac14 014 Pfrac14 032 andfor non-victimized girls bfrac14006 SEfrac14 011 Pfrac14 055

In sum these results confirm the hypothesis that insula ac-tivation predicts internalizing symptoms and this effect wasexplained by avoidance motivation Although the difference be-tween the effects for the victimization groups (ie the inter-actions) tended not to be significant the within-group effectssuggested a more robust effect of insula activation on internal-izing symptoms (via avoidance motivation) for victimized thannon-victimized girls

Discussion

Theory and research implicate challenges to the human need tobelong as a risk factor for emotional distress (Rudolph et al 2005Slavich et al 2010) Previous support for this idea has emergedfrom two distinct lines of investigation One set of studies focuseson neural sensitivity to experimentally induced acute social exclu-sion (eg Masten et al 2011 Eisenberger 2012 for a review seeRotge et al 2014) whereas a second set of studies focuses on emo-tional sensitivity to naturally occurring social rejection (eg Slavichet al 2009 Rudolph et al 2011) The present study makes a novelcontribution by integrating these two lines of theory and researchto examine (i) whether neural sensitivity to exclusion is associatedwith internalizing symptoms (ii) whether this link is contingenton adolescent girlsrsquo naturally occurring social experiences and(iii) what psychological processes are involved in this process

Association between neural sensitivity andinternalizing symptoms

Social pain theory suggests that threats to social bonds activateneural regions associated with physical pain including thedACC sgACC and insula Although this theory highlights theemotional distress associated with exposure to social painmost research has examined distress immediately following anexperimental manipulation of social rejection (for exceptionssee Masten et al 2011 Lau et al 2012 Silk et al 2014) Our firstgoal was to examine whether neural activation in the socialpain network is associated with indexes of more pervasive andenduring distress as reflected in internalizing symptoms in asample of adolescent girls Consistent with our hypothesis anda few prior studies whole-brain regression analyses revealedthat activation in the dACC sgACC and insula was significantlyassociated with higher levels of internalizing symptoms

These findings suggest that challenges to social bonds notonly result in temporary emotional perturbations but also maycreate pervasive emotional difficulties It is reasonable that atthe time of a social rejection many youth feel a sense of socialpain and accordingly report more emotional distressMoreover enhanced social pain responses co-occur withincreasing age (Guyer et al 2009) and puberty (Silk et al 2014)suggesting that adolescence may be a time of particular neuralsensitivity to rejection But not all girls develop internalizingsymptoms during adolescencemdashfor whom might this height-ened neural sensitivity foster more pervasive internalizingsymptoms and how does this process unfold We sought to ad-dress these two questions by testing an integrated model of

Fig 2 Conceptual model showing conditional direct effects of neural sensitivity

on internalizing symptoms and conditional indirect effects of neural sensitivity

on internalizing symptoms through avoidance motivation Dotted lines repre-

sent moderation of Path a and Path b

Table 2 Regions of activation to exclusion condition vs inclusioncondition that correlated significantly with internalizing symptoms

Region name x y z t k

Fusiform gyrus 36 34 14 598 779a

Hippocampus 21 16 20 392 a

Subgenual ACC 15 11 14 318 a

Amygdala 15 4 14 334 a

Fusiform gyrus 30 67 11 438 858Posterior insula 51 1 1 355 419IFG 36 23 28 427 200b

Middle frontal gyrus 27 44 34 355 b

Superior medial gyrus 0 44 40 298 b

Subgenual ACC 12 17 11 426 194c

Anterior insula 42 14 14 390 c

Amygdala 17 3 17 414 c

TPJ 42 52 19 424 106Dorsolateral ACC 3 35 31 346 642d

Dorsolateral ACC 3 20 37 381 d

Cerebellar vermis 6 46 1 368 51Precentral gyrus 45 10 49 338 49

Notes x y and z refer to MNI (Montreal Neurological Institute) coordinates t

refers to the t-score at those coordinates (local maxima) k refers to the number

of voxels in each significant cluster Clusters that share the same superscript are

part of the same cluster of activation IFG inferior fusiform gyrus TPJ temporal

parietal junction

836 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

InsulaVictimization Status interaction bfrac14 036 SEfrac14 020Pfrac14 007 As expected the total effect of insula activation oninternalizing symptoms was significant for victimized girlsbfrac14 043 SEfrac14 015 Plt 0004 but not for non-victimized girlsbfrac14 007 SEfrac14 013 Pfrac14 057 (Figure 3C)

In Step 2 we tested the extent to which the conditional totaleffect of insula activation on internalizing symptoms could be ac-counted for by a conditional indirect effect through avoidanceThe paths from insula activation to avoidance (Path b) and fromavoidance to internalizing symptoms (Path c) were estimated aswas the path from the InsulaVictimization Status interaction toavoidance The right column in Table 3 presents the results ofthis analysis Although the InsulaVictimization Status inter-action did not significantly predict avoidance bfrac14 029 SEfrac14 021ns the conditional effect of insula activation on avoidance (Pathb) was significant for victimized girls bfrac14 054 SEfrac14 016 Plt 0001and marginally significant for non-victimized girls bfrac14 025SEfrac14 013 Pfrac14 006 Moreover avoidance significantly predictedinternalizing symptoms bfrac14 055 SEfrac14 012 Plt 0001 This resultedin a significant conditional indirect effect of insula activation oninternalizing symptoms through avoidance for victimized girls

bfrac14 030 SEfrac14 011 Pfrac14 0005 and a marginally significant condi-tional indirect effect for non-victimized girls bfrac14 014 SEfrac14 008Pfrac14 008 After accounting for the conditional indirect effect theInsulaVictimization Status interaction did not significantly pre-dict internalizing symptoms bfrac14 020 SEfrac14 017 Pfrac14 022 and thedirect effect of insula activation on internalizing symptoms wasnonsignificant for victimized girls bfrac14 014 SEfrac14 014 Pfrac14 032 andfor non-victimized girls bfrac14006 SEfrac14 011 Pfrac14 055

In sum these results confirm the hypothesis that insula ac-tivation predicts internalizing symptoms and this effect wasexplained by avoidance motivation Although the difference be-tween the effects for the victimization groups (ie the inter-actions) tended not to be significant the within-group effectssuggested a more robust effect of insula activation on internal-izing symptoms (via avoidance motivation) for victimized thannon-victimized girls

Discussion

Theory and research implicate challenges to the human need tobelong as a risk factor for emotional distress (Rudolph et al 2005Slavich et al 2010) Previous support for this idea has emergedfrom two distinct lines of investigation One set of studies focuseson neural sensitivity to experimentally induced acute social exclu-sion (eg Masten et al 2011 Eisenberger 2012 for a review seeRotge et al 2014) whereas a second set of studies focuses on emo-tional sensitivity to naturally occurring social rejection (eg Slavichet al 2009 Rudolph et al 2011) The present study makes a novelcontribution by integrating these two lines of theory and researchto examine (i) whether neural sensitivity to exclusion is associatedwith internalizing symptoms (ii) whether this link is contingenton adolescent girlsrsquo naturally occurring social experiences and(iii) what psychological processes are involved in this process

Association between neural sensitivity andinternalizing symptoms

Social pain theory suggests that threats to social bonds activateneural regions associated with physical pain including thedACC sgACC and insula Although this theory highlights theemotional distress associated with exposure to social painmost research has examined distress immediately following anexperimental manipulation of social rejection (for exceptionssee Masten et al 2011 Lau et al 2012 Silk et al 2014) Our firstgoal was to examine whether neural activation in the socialpain network is associated with indexes of more pervasive andenduring distress as reflected in internalizing symptoms in asample of adolescent girls Consistent with our hypothesis anda few prior studies whole-brain regression analyses revealedthat activation in the dACC sgACC and insula was significantlyassociated with higher levels of internalizing symptoms

These findings suggest that challenges to social bonds notonly result in temporary emotional perturbations but also maycreate pervasive emotional difficulties It is reasonable that atthe time of a social rejection many youth feel a sense of socialpain and accordingly report more emotional distressMoreover enhanced social pain responses co-occur withincreasing age (Guyer et al 2009) and puberty (Silk et al 2014)suggesting that adolescence may be a time of particular neuralsensitivity to rejection But not all girls develop internalizingsymptoms during adolescencemdashfor whom might this height-ened neural sensitivity foster more pervasive internalizingsymptoms and how does this process unfold We sought to ad-dress these two questions by testing an integrated model of

Fig 2 Conceptual model showing conditional direct effects of neural sensitivity

on internalizing symptoms and conditional indirect effects of neural sensitivity

on internalizing symptoms through avoidance motivation Dotted lines repre-

sent moderation of Path a and Path b

Table 2 Regions of activation to exclusion condition vs inclusioncondition that correlated significantly with internalizing symptoms

Region name x y z t k

Fusiform gyrus 36 34 14 598 779a

Hippocampus 21 16 20 392 a

Subgenual ACC 15 11 14 318 a

Amygdala 15 4 14 334 a

Fusiform gyrus 30 67 11 438 858Posterior insula 51 1 1 355 419IFG 36 23 28 427 200b

Middle frontal gyrus 27 44 34 355 b

Superior medial gyrus 0 44 40 298 b

Subgenual ACC 12 17 11 426 194c

Anterior insula 42 14 14 390 c

Amygdala 17 3 17 414 c

TPJ 42 52 19 424 106Dorsolateral ACC 3 35 31 346 642d

Dorsolateral ACC 3 20 37 381 d

Cerebellar vermis 6 46 1 368 51Precentral gyrus 45 10 49 338 49

Notes x y and z refer to MNI (Montreal Neurological Institute) coordinates t

refers to the t-score at those coordinates (local maxima) k refers to the number

of voxels in each significant cluster Clusters that share the same superscript are

part of the same cluster of activation IFG inferior fusiform gyrus TPJ temporal

parietal junction

836 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

internalizing symptoms that places neural sensitivity withinthe context of girlsrsquo everyday social lives and considers one pos-sible explanatory pathway through which neural sensitivity isassociated with internalizing symptoms

Individual differences in the neural sensitivityndashinternalizing symptoms link

To better understand whether some adolescents are more vul-nerable than others to the pervasive emotional effects of aheightened social pain response we examined whether neuralsensitivity to social exclusion has particularly robust implica-tions for emotional well-being among adolescents with a his-tory of chronic social rejection in the form of peer victimizationSupporting this hypothesis heightened activation in each ofthe three social pain regions was associated with internalizingsymptoms among adolescent girls with a history of peer victim-ization but not among those without a history of peer victimiza-tion with a particularly strong interactive effect for the dACCBeing exposed to negative social feedback across the schoolyears may leave a social bruise that intensifies the meaning ofsubsequent painful social experiences such that a heightenedsocial pain response is linked to more pervasive and lastingemotional difficulties such as depressive symptoms and socialanxiety in victimized than non-victimized girls Consistentwith the idea that exclusion has more aversive implications forvictimized girls this group reported higher levels of threat totheir social needs after exposure to acute social exclusion thandid non-victimized girls This research suggests the need toconsider how individual differences in youthsrsquo neural responsesto social cues of rejection help to determine their emotional vul-nerability in the face of victimization or similar social stressors

In a supplementary analysis (Supplementary Data) we alsoexamined whether victimized and non-victimized girls differed

in their level of neural activation (rather than the link betweenactivation and internalizing symptoms) In one prior study Willet al (2016) found heightened dACC activation to exclusion rela-tive to inclusion in chronically rejected youth compared withstably accepted youth The present study yielded a similar pat-tern chronically victimized girls compared with non-victimizedgirls showed greater dACC activation to exclusion relative to in-clusion It is noteworthy that this pattern replicated acrossgroups with a different gender and age composition as well asacross different operationalizations of social rejection The Willet al (2016) study classified children according to peer sociomet-ric nominations of social preference from this perspective so-cial rejection is operationalized as an attitude of the peer group(feelings of dislike toward others) In contrast our study classi-fied youth according to self-reports of peer victimization fromthis perspective social rejection is operationalized as exposureto specific threatening behaviors from peers (eg physical harmverbal abuse manipulation of relationships) The similar pat-tern of results may reflect the co-occurrence of attitudes of re-jection and behavioral manifestations of these attitudesmdashthatis youth who are rejected by their peers often are exposed tovictimization andor peers may develop negative attitudes to-ward victimized youth (Kochel et al 2014) Collectively the Willet al (2016) findings along with the present research suggest a2-fold risk in youth exposed to early social adversity theseyouth show more neural sensitivity to exclusion and this sensi-tivity is more strongly associated with internalizing symptomscompared with youth exposed to low levels of social adversity

Psychological pathway from neural sensitivity tointernalizing symptoms

To better understand why neural sensitivity in the social painnetwork might be associated with internalizing symptoms we

Table 3 Path analyses testing the indirect direct and total effects of neural sensitivity (dACC sgACC and insula activation) on internalizingsymptoms

Model 1 dACC Model 2 sgACC Model 3 InsulaUnstandardized path coefficients

b (SE) b (SE) b (SE)

Neural sensitivity to internalizing symptoms (Path a) 009 (012) 001 (014) 006 (011)Victimization status to internalizing symptoms 065 (014) 060 (014) 060 (016)Neural SensitivityVictimization Status to Internalizing

symptoms (ie moderation of Path a by victimization status)036 (017) 033dagger (017) 020 (017)

Neural sensitivity to avoidance (Path b) 019 (014) 006 (022) 025dagger (013)Victimization status to avoidance 035 (016) 039 (020) 015 (020)Neural SensitivityVictimization Status to Avoidance

(ie moderation of Path b by victimization status)055 (019) 033 (025) 029 (021)

Avoidance to internalizing symptoms (Path c) 034 (012) 044 (010) 055 (012)

Conditional Effects on Avoidance for Victimized and Non-Victimized GirlsVicGirls

Non-VicGirls

VicGirls

Non-VicGirls

VicGirls

Non-VicGirls

Neural sensitivity to avoidance 075 (014) 019 (014) 038 (013) 006 (022) 054 (016) 025dagger (013)

Decomposition of Effect of Neural Sensitivity on Internalizing SymptomsVicGirls

Non-VicGirls

VicGirls

Non-VicGirls

VicGirls

Non-VicGirls

Total effect (estimated at Step 1) 070 (012) 016 (012) 051 (011) 003 (017) 043 (015) 007 (013)Indirect effect (estimated at Step 2) 025 (010) 006 (005) 017 (007) 003 (010) 030 (011) 014dagger (008)Direct effect (estimated at Step 2) 045 (014) 009 (012) 034 (009) 001 (014) 014 (014) 006 (011)

Note Non-vic non-victimized Vic victimizeddaggerPlt010 Plt005 Plt001 Plt0001

K D Rudolph et al | 837

examined avoidance motivation as one possible underlyingpsychological process Across all three social pain regions wefound a significant indirect effect from neural sensitivitythrough avoidance motivation to internalizing symptoms invictimized girls For girls who have been exposed to chronic vic-timization heightened neural sensitivity to exclusion trans-lated into a generalized psychological sensitivity to aversivesocial cues as reflected in a drive to avoid negative judgmentspeer disapproval and loss of social status Thus chronically vic-timized adolescent girls with heightened neural sensitivity toexclusion not only may experience more adverse reactions toactual rejection experiences but also may show constant

vigilance to potential social threats and a tendency to avoid thepossibility of future rejection Having a sense of self and a set ofsocial goals that is contingent on the judgments and approval ofpeers may set these youth up for the development of criticalself-appraisals helplessness and negative emotions character-istic of depression and social anxiety These findings are con-sistent with conceptualizations of targeted rejection whichemphasize its role in triggering social-evaluative threat nega-tive self-appraisals social withdrawal and depression (Slavichet al 2009 2010)

Future research will need to continue exploring why neuralsensitivity to exclusion serves as a more robust predictor of

Fig 3 (A) dACC activation (B) sgACC activation and (C) insula activation interact with victimization status to predict internalizing symptoms

838 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

avoidance motivation and internalizing symptoms among victi-mized than non-victimized girls Prior research shows that vic-timization predicts negative self-appraisals (Cole et al 2010)emotion dysregulation (McLaughlin et al 2009 Rudolph et al2009) and maladaptive responses to social stressors (Troop-Gordon et al 2015) suggesting that perhaps non-victimizedgirls can recover more quickly from social rejection experiencesby engaging in effective regulation of their cognitive and emo-tional reactions Thus it would be beneficial to explore variousexplanations for why neural sensitivity to exclusion has feweradverse psychological and emotional effects on non-victimizedthan victimized youth

Study strengths limitations and future directions

This study is among the first to establish an association be-tween neural sensitivity in the social pain network and inter-nalizing symptoms (see also Masten et al 2011 Silk et al 2014)and is the first to reveal individual differences in this associ-ation that are contingent on naturally occurring exposure to so-cial stress In particular these findings suggest that sensitivityto social exclusion may serve as a neural marker of vulnerabilityfor internalizing symptoms only in youth who have a history ofchronic rejection (eg frequent exclusion from the peer grouprejection by friends or romantic break-ups) Moreover we iden-tify one pathway explaining the association between neuralsensitivity and internalizing symptoms thereby addressing theneed to better understand psychological and emotional proc-esses linked to the social pain response (Eisenberger 2015)More broadly this research unites social pain theory with inter-personal theories of depression and social anxiety providing amulti-level perspective on the increasing risk for internalizingsymptoms that emerges in girls over the course of adolescence

Despite these contributions further investigation is warrantedto clarify the precise role of various regions of the social pain net-work in risk for internalizing symptoms The most robust findingsin the present study involve the dACC although similar patterns ofeffects were observed in the sgACC and insula Despite the existingdata base linking dACC and sgACC activation to social pain(Eisenberger et al 2003 Dewall et al 2010 Masten et al 2011Eisenberger 2012 for a review see Rotge et al 2014) the anteriorcingulate cortex (ACC) may be involved in a variety of processes ofpotential relevance to Cyberball including violation of expect-ations (Somerville et al 2006 Bolling et al 2011) and conflict moni-toring (Botvinick et al 2004) Thus it is possible that ACCactivation during Cyberball also is linked to processes other thanthe social pain response One study using Cyberball to distinguishsocial pain from neural activation in response to violation of ex-pectations found evidence linking dACC activation specifically tosocial exclusion (but not overinclusion Kawamoto et al 2012)Moreover recent findings from a large-scale quantitative reverseinference analysis (Lieberman and Eisenberger 2015) indicate pref-erential activation of certain parts of the dACC in response to painHowever consistent with the conceptualization of ACC activationas a lsquoneural alarm systemrsquo (Eisenberger et al 2011) it is possiblethat heightened ACC activation in the context of Cyberball reflectsin part greater conflict monitoring as youth react to the discrep-ancy between their desired social state and current social condi-tions Of note recent evidence also suggests distinctions betweenthe neural representation of physical and social pain despite someoverlap (Woo et al 2014) Additional research is therefore neededto clarify the particular role of these regions in pain processing aswell as the extent of overlap vs distinctiveness between the neuralnetworks involved in physical and social pain

Future research also will need to determine whether thispattern of findings is specific to neural sensitivity to social re-jection or whether it would extend to other types of stress re-activity Because of our focus on victimized youth weanticipated that social exclusion would be a particularly salientstressor However we did not include a task measuring neuralactivation to other types of interpersonal stressors (eg familyconflict) or noninterpersonal stressors (eg physical threat aca-demic failure) It will be important to directly examine whetherneural reactivity to other forms of stress also is linked to inter-nalizing symptoms in victimized youth or whether they are par-ticularly sensitized to exclusion-related stressors

Finally although our study involved a prospective assess-ment of victimization providing unique data about girlsrsquo long-term history of social adversity we used concurrent assess-ments of neural processing avoidance motivation and internal-izing symptoms Thus we cannot draw firm conclusions aboutthe direction of effects Elucidating the interactive contributionof early social adversity and neural processing to girlsrsquo emo-tional development will require longitudinal designs that trackchanges in neural activation exposure to social stress andinternalizing symptoms over the course of adolescence Resultsfrom the present study therefore can serve as a basis for design-ing prospective studies that examine the dynamic interconnec-tions among these processes over time

Conclusions and implications

In conclusion this research makes a novel contribution both tosocial pain theory (Eisenberger et al 2003) and to interpersonaltheories of internalizing symptoms (Rudolph 2009 Davila et al2010 Rudolph et al in press) by providing evidence for theinteractive influence of neural sensitivity and social context oninternalizing symptoms highlighting the need for integrativemulti-level theoretical models that consider the joint influenceof biological social and psychological systems of developmentwhen elucidating the processes underlying heightened risk forinternalizing symptoms in adolescent girls Given the debilitat-ing and persistent burden associated with adolescent internal-izing symptoms (Rudolph and Flynn 2014) it is critical toidentify effective targets of prevention These results suggestthat in addition to the development and implementation of ef-fective anti-bullying programs (Williford et al 2012) alteringvictimized girlsrsquo neural reactivity to social exclusion in waysthat reduce their heightened focus on peer judgments and ap-proval may help prevent the onset of internalizing symptomsduring adolescence in girls with a history of social vulnerability

Acknowledgements

We would like to thank the families and schools who par-ticipated in this study We are grateful to Jamie AbaiedMonica Agoston Samirah Ali Suravi Changlani MeganFlynn Inge Karosevica Nicole Llewellyn Jennifer MontiHeather Ross and Niwako Sugimura for their assistance indata collection and management

Funding

This work was supported by a University of Illinois ResearchBoard Award and a National Institute of Mental HealthGrant (MH68444) awarded to KDR and a National Institute

K D Rudolph et al | 839

of Mental Health Grant (MH105655) awarded to KDR andEHT)

Supplementary data

Supplementary data are available at SCAN online

Conflict of interest None declared

ReferencesAngold A Costello EJ Messer SC Pickles A Winder F

Silver D (1995) Development of a short questionnaire for usein epidemiological studies of depression in children and ado-lescents International Journal of Methods in Psychiatric Research 5237ndash49

Baumeister RF Leary MR (1995) The need to belong desirefor interpersonal attachments as a fundamental human mo-tivation Psychological Bulletin 117(3) 497ndash529

Boivin M Petitclerc A Feng B Barker E (2010) The develop-mental trajectories of peer victimization in middle to latechildhood and the changing nature of their behavioral correl-ates Merrill-Palmer Quarterly Journal of DevelopmentalPsychology 56(3) 231ndash60

Bolling DZ Pitskel NB Deen B et al (2011) Dissociable brainmechanisms for processing social exclusion and rule violationNeuroimage 54(3) 2462ndash71

Botvinick MM Cohen JD Carter CS (2004) Conflict monitor-ing and anterior cingulate cortex an update Trends in CognitiveSciences 8(12) 539ndash46

Burklund LJ Eisenberger NI Lieberman MD (2007) The faceof rejection rejection sensitivity moderates dorsal anteriorcingulate activity to disapproving facial expressions Social

Neuroscience 2(3ndash4) 238ndash53Caouette JD Guyer AE (2014) Gaining insight into adolescent

vulnerability for social anxiety from developmental cognitiveneuroscience Developmental Cognitive Neuroscience 8 65ndash76

Carver CS White TL (1994) Behavioral inhibition behavioralactivation and affective responses to impending reward andpunishment the BISBAS scales Journal of Personality and SocialPsychology 67(2) 319ndash33

Cole DA Maxwell MA Dukewich TL Yosick R (2010)Targeted peer victimization and the construction of positiveand negative self-cognitions connections to depressive symp-toms in children Journal of Clinical Child and AdolescentPsychology 39(3) 421ndash35

Coplan RJ Wilson J Frohlick SL Zelenski J (2006) A person-oriented analysis of behavioral inhibition and behavioral acti-vation in children Personality and Individual Differences 41(5)917ndash27

Crick N R Grotpeter JK (1996) Childrenrsquos treatment by peersvictims of relational and overt aggression Development andPsychopathology 8(2) 367ndash80

Davila J La Greca AM Starr LR Landoll RR (2010) Anxietydisorders in adolescence In Beck JG editor InterpersonalProcesses in the Anxiety Disorders Implications for Understanding

Psychopathology and Treatment (pp 97ndash124) Washington DCAmerican Psychological Association

DeWall CN MacDonald G Webster GD et al (2010)Acetaminophen reduces social pain behavioral and neural evi-dence Psychological Science 21(7) 931ndash7

DeWall CN Masten CL Powell C Combs D Schurtz DREisenberger NI (2012) Do neural responses to rejection

depend on attachment style An fMRI study Social Cognitiveand Affective Neuroscience 7(2) 184ndash92

Eisenberger NI (2012) The pain of social disconnection exam-ining the shared neural underpinnings of physical and socialpain Nature Reviews Neuroscience 13(6) 421ndash34

Eisenberger NI (2015) Meta-analytic evidence for the role ofthe anterior cingulate cortex in social pain Social Cognitive andAffective Neuroscience 10 1ndash2

Eisenberger NI Inagaki TK Muscatell KA Haltom KEBLeary MR (2011) The neural sociometer Brain mechanismsunderlying state self-esteem Journal of Cognitive Neuroscience23(11) 3448ndash55

Eisenberger NI Lieberman MD Williams KD (2003) Does re-jection hurt An fMRI study of social exclusion Science302(5643) 290ndash2

Enders CK Bandalos DL (2001) The relative performance offull information maximum likelihood estimation for missingdata in structural equation models Structural EquationModeling 8(3) 430ndash57

Gray JA (1991) The neuropsychology of temperament InStrelau J Angleitner A editors Explorations in TemperamentInternational Perspectives on Theory and Measurement (pp105ndash128) New York Plenum

Gunnar MR Wewerka S Frenn K Long JD Griggs C (2009)Developmental changes in hypothalamusndashpituitaryndashadrenalactivity over the transition to adolescence normative changesand associations with puberty Development andPsychopathology 21(1) 69ndash85

Guyer AE Choate VR Pine DS Nelson EE (2012) Neuralcircuitry underlying affective response to peer feedback inadolescence Social Cognitive and Affective Neuroscience 7(1)81ndash92

Guyer AE McClure-Tone EB Shiffrin ND Pine DSNelson EE (2009) Probing the neural correlates of antici-pated peer evaluation in adolescence Child Development80(4) 1000ndash15

Guyer AE Jarcho JM Perez-Edgar K et al (2015)Temperament and parenting styles in early childhood dif-ferentially influence neural response to peer evaluation inadolescence Journal of Abnormal Child Psychology 43(5)863ndash74

Hankin BL Abramson LY (2001) Development of gender dif-ferences in depression an elaborated cognitive vulnerability-transactional stress theory Psychological Bulletin 127 773ndash96

Kawamoto T Onoda K Nakashima KI Nittono HYamaguchi S Ura M (2012) Is dorsal anterior cingulate cor-tex activation in response to social exclusion due to expect-ancy violation An fMRI study Frontiers in EvolutionaryNeuroscience 4(11) 1ndash10

Kochel KP Ladd GW Rudolph KD (2012) Longitudinal asso-ciations among youth depressive symptoms peer victimiza-tion and low peer acceptance an interpersonal processperspective Child Development 83(2) 637ndash50

La Greca A Davila J Landoll RR Siegel R (2011) Dating ro-mantic relationships and social anxiety in young people InAlfano CA Beidel DC editors Social Anxiety in Adolescentsand Young Adults Translating Developmental Science Into Practice(pp 93ndash105) Washington DC American PsychologicalAssociation

La Greca AM Lopez N (1998) Social anxiety among adoles-cents linkages with peer relations and friendships Journal ofAbnormal Child Psychology 26(2) 83ndash94

Lau JY Guyer A Tone EB et al (2011) Neural responses topeer rejection in anxious adolescents contributions from the

840 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

amygdala-hippocampal complex International Journal ofBehavioral Development 36(1) 36ndash44

Lieberman MD Eisenberger NI (2006) A pain by any othername (rejection exclusion ostracism) still hurts the same therole of dorsal anterior cingulate cortex in social and physicalpain In Cacioppo J Visser P Pickett C editors SocialNeuroscience People Thinking About People (169ndash187) CambridgeMA MIT Press

Llewellyn N Rudolph KD (2014) Individual and sex differ-ences in the consequences of victimization Moderation by ap-proach and avoidance motivation Developmental Psychology50(9) 2210ndash20

Masten CL Eisenberger NI Borofsky LA et al (2009) Neuralcorrelates of social exclusion during adolescence understand-ing the distress of peer rejection Social Cognitive and AffectiveNeuroscience 4(2) 143ndash57

Masten CL Eisenberger NI Borofsky LA McNealy KPfeifer JH Dapretto M (2011) Subgenual anterior cingulateresponses to peer rejection a marker of adolescentsrsquo risk fordepression Development and Psychopathology 23(1) 283ndash92

McClelland DC Atkinson JW Clark RA Lowell EL (1953)The achievement motive New York Appleton-Century-Crofts

McLaughlin KA Hatzenbuehler ML Hilt LM (2009) Emotiondysregulation as a mechanism linking peer victimization tointernalizing symptoms in adolescents Journal of Consultingand Clinical Psychology 77(5) 894ndash904

Muris P Meesters C de Kanter E Timmerman PE (2005)Behavioural inhibition and behavioural activation systemscales for children relationships with Eysenckrsquos personalitytraits and psychopathological symptoms Personality andIndividual Differences 38 831ndash41

Muthen LK Muthen BO (1998ndash2007) Mplus Userrsquos GuideVersion 5

Nelemans SA Hale WW Branje SJ et al (2014)Heterogeneity in development of adolescent anxiety disordersymptoms in an 8-year longitudinal community studyDevelopment and Psychopathology 26(01) 181ndash202

Nelson EE Leibenluft E McClure E Pine DS (2005) The so-cial re-orientation of adolescence a neuroscience perspectiveon the process and its relation to psychopathologyPsychological Medicine 35(2) 163ndash74

Preacher KJ Rucker DD Hayes AF (2007) Addressing mod-erated mediation hypotheses theory methods and prescrip-tions Multivariate Behavioral Research 42(1) 185ndash227

Rotge JY Lemogne C Hinfray S et al (2014) A meta-analysisof the anterior cingulate contribution to social pain SocialCognitive and Affective Neuroscience 10(1) 19ndash27

Rudolph KD (2009) The interpersonal context of adolescent de-pression In Nolen-Hoeksema S Hilt L M editors Handbookof Depression in Adolescents (pp 377ndash418) New York Routledge

Rudolph KD Abaied JL Flynn M Sugimura N AgostonAM (2011) Developing relationships being cool and not look-ing like a loser social goal orientation predicts childrenrsquos re-sponses to peer aggression Child Development 82(5) 1518ndash30

Rudolph KD Bohn LE (2014) Translating social motivationinto action Contributions of need for approval to childrenrsquos so-cial engagement Social Development 23(2) 376ndash94

Rudolph KD Caldwell MS Conley CS (2005) Need for ap-proval and childrenrsquos well-being Child Development 76(2)309ndash23

Rudolph KD Flynn M (2014) Adolescent depression InGotlib I H Hammen C L editors Handbook of Depression 3rdedn 391ndash409 New York NY Guilford

Rudolph KD Lansford JE Agoston AM et al (2014) Peer vic-timization and social alienation predicting deviant peer affili-ation in middle school Child Development 85 124ndash39

Rudolph KD Lansford JE Rodkin PC (in press)Interpersonal theories of psychopathology In Cicchetti Deditor Developmental Psychopathology 3rd edn LondonEngland Wiley

Rudolph KD Troop-Gordon W Flynn M (2009) Relationalvictimization predicts childrenrsquos social-cognitive and self-regulatory responses in a challenging peer contextDevelopmental Psychology 45(5) 1444ndash54

Rudolph KD Troop-Gordon W Hessel ET Schmidt J D(2011) A latent growth curve analysis of early and increasingpeer victimization as predictors of mental health across elem-entary school Journal of Clinical Child and Adolescent Psychology

40 111ndash22Rudolph KD Troop-Gordon W Llewellyn N (2013)

Interactive contributions of self-regulation deficits and socialmotivation to psychopathology unraveling divergent path-ways to aggressive behavior and depressive symptomsDevelopment and Psychopathology 25(02) 407ndash18

Rudolph KD Troop-Gordon W Monti JD Miernicki ME(2014) Moving against and away from the world the adoles-cent legacy of peer victimization Development and

Psychopathology 26 721ndash34Sebastian CL Tan GCY Roiser JP Viding E Dumontheil I

Blakemore S (2011) Developmental influences on the neuralbases of responses to social rejection implications of socialneuroscience for education NeuroImage 57 686ndash94

Siegel RS La Greca AM Harrison HM (2009) Peer victimiza-tion and social anxiety in adolescents prospective and recip-rocal relationships Journal of Youth and Adolescence 38(8)1096ndash109

Silk JS Nelson E Dahl RE Stroud L Lee KH Siegle GJ(2014) Increased neural response to peer rejection associatedwith adolescent depression and pubertal development Social

Cognitive and Affective Neuroscience 9(11) 1798ndash807Slavich GM OrsquoDonovan A Epel ES Kemeny ME (2010)

Black sheep get the blues a psychobiological model of socialrejection and depression Neuroscience and Biobehavioral

Reviews 35(1) 39ndash45Slavich GM Thornton T Torres LD Monroe SM Gotlib IH

(2009) Targeted rejection predicts hastened onset of major de-pression Journal of Social and Clinical Psychology 28(2) 223ndash43

Slavich GM Way BM Eisenberger NI Taylor SE (2010)Neural sensitivity to social rejection is associated with inflam-matory responses to social stress Proceedings of the National

Academy of Sciences 107(33) 14817ndash22Somerville LH Heatherton TF Kelley WM (2006) Anterior

cingulate cortex responds differentially to expectancy viola-tion and social rejection Nature Neuroscience 9(8) 1007ndash8

Troop-Gordon W Rudolph KD Sugimura N Little T (2015)Peer victimization in middle childhood impedes adaptive re-sponses to stress a pathway to depressive symptoms Journal

of Clinical Child and Adolescent Psychology 44(3) 432ndash45Ward BD (2000) Simultaneous inference for fMRI data Retrieved

November 23 2015 from httpafninimhnihgovafnidocmanualAlphaSim

Will G-J van Lier PAC Crone EA Geurouroglu B (2016) Chronicchildhood peer rejection is associated with heightened neuralresponses to social exclusion during adolescence Journal of

Abnormal Child Psychology 44(1) 43ndash55

K D Rudolph et al | 841

Williams KD Cheung CK Choi W (2000) CyberostracismEffects of being ignored over the Internet Journal of Personalityand Social Psychology 79(5) 748ndash62

Williford A Boulton A Noland B Little T D Keuroarneuroa ASalmivalli C (2012) Effects of the KiVa anti-bullying program

on adolescentsrsquo depression anxiety and perception of peersJournal of Abnormal Child Psychology 40 289ndash300

Woo C Koban L Kross E et al (2014) Separate neural repre-sentations for physical pain and social rejection NatureCommunications 5 5380ndash91

842 | Social Cognitive and Affective Neuroscience 2016 Vol 11 No 5

• nsw021-TF1
• nsw021-TF2
• nsw021-TF3
• nsw021-TF4
• nsw021-TF5