The role of emotional inhibitory control in specific internet addiction – an fMRI study
Julia Dieter
a,∗, Sabine Hoffmann
a, Daniela Mier
b, Iris Reinhard
c, Martin Beutel
d, Sabine Vollstädt-Klein
a, Falk Kiefer
a, Karl Mann
a, Tagrid Leménager
aaDepartmentofAddictiveBehaviourandAddictionMedicine,CentralInstituteofMentalHealth,MedicalFacultyMannheim/HeidelbergUniversity,J5, D-68159,Mannheim,Germany
bDepartmentofClinicalPsychology,CentralInstituteofMentalHealth,MedicalFacultyMannheim/HeidelbergUniversity,J5,D-68159,Mannheim, Germany
cDepartmentofBiostatistics,CentralInstituteofMentalHealth,MedicalFacultyMannheim/HeidelbergUniversity,J5,D-68159,Mannheim,Germany
dKraichtalClinics,AmMühlberg,D-76703,Kraichtal,Germany
h i g h l i g h t s
•Highersocialanxiety&reducedemotionalcompetenceinspecificinternetaddicts.
•LongerreactiontimesinESTforsociallyanxiouswordsininternetgamingaddicts.
•Leftmiddleandsuperiortemporalgyrushypoactivationininternetgamingaddicts.
•Socialwordsmaybelessretrievableinsemanticstorageofinternetgamingaddicts.
Keywords:
Specificinternetaddiction EmotionalStroopTask AffectiveGo/No-Gotask emotionalinhibitorycontrol fMRI
a b s t r a c t
Background:Addictstospecificinternetapplicationsinvolvingcommunicationfeaturesshowedincreased socialanxiety,emotionalcompetencedeficitsandimpairedprefrontal-relatedinhibitorycontrol.The dorsalAnteriorCingulateCortex(dACC)likelyplaysanimportantroleincognitivecontrolandnegative affect(suchassocialexclusion,painoranxiety).
Aim:Toassess(social)anxiety-relatedinhibitorycontrolinspecificinternetaddiction(addicteduseof gamesandsocialnetworks)anditsrelationtoaltereddACCactivation.
Methods:N=44controlsandn=51specificinternetaddictscompletedananxiouswords-basedAffec- tiveGo/No-Gotask(AGN).Asubsampleofn=23healthycontrolsandn=25specificinternetaddicts underwentfunctionalMagneticResonanceImaging(fMRI)whilecompletinganEmotionalStroopTask (EST)withsociallyanxious,positive,negativeandneutralwords.Subgroupsofinternetgamingand socialnetworkaddictswereexplorativelyassessed.Psychometricmeasuresofsocialanxiety,emotional competenceandimpulsivitywereadditionallyexplored.
Results:Specificinternetaddictsshowedhigherimpulsivity,socialanxietyandreducedemotionalcom- petence.Between-groupdifferencesinAGNandESTbehavioralmeasureswerenotdetected.Nogroup differenceswerefoundinthedACC,butexplorativeanalysesrevealeddecreasedleftmiddleandsuperior temporalgyrusactivationduringinterferenceofsociallyanxiouswordsininternetgamingandrelative tosocialnetworkaddicts.
Abbreviations: AGN,AffectiveGo/No-Gotask;AICA,checklistfortheAssessmentofInternetandComputergameAddiction;AICA30,checklistfortheAssessmentof InternetandComputergameAddictionreferringtotheprevious30days;AICAlifetime,AssessmentofInternetandComputergameAddiction:Lifetimemaximum;EKF, emotionalcompetencequestionnaire(EKF;Germandesignation,Emotionale-Kompetenz-Fragebogen);EST,EmotionalStroopTask;IGA,InternetGamingAddiction;OSVe-S, scaleforonlineaddictivebehaviorinadultsself-report(Germandesignation:SkalazumOnlinesuchtverhaltenbeiErwachsenen);SASKO,questionnaireforSocialAnxiety andSocialCompetencedeficits(Germandesignation:FragebogenzuSozialerAngstundSozialenKompetenzdefiziten).
∗Correspondingauthorat:DepartmentofAddictiveBehaviourandAddictionMedicine,CentralInstituteofMentalHealth,J5,D-68159,Mannheim,Germany.
E-mailaddresses:Julia.Dieter@zi-mannheim.de(J.Dieter),Sabine.Hoffmann@zi-mannheim.de(S.Hoffmann),
Daniela.Mier@zi-mannheim.de(D.Mier),Iris.Reinhard@zi-mannheim.de(I.Reinhard),Martin.Beutel@kraichtal-kliniken.de(M.Beutel),S.Vollstaedt-Klein@zi-mannheim.de (S.Vollstädt-Klein),Falk.Kiefer@zi-mannheim.de(F.Kiefer),Karl.Mann@zi-mannheim.de(K.Mann),Tagrid.Lemenager@zi-mannheim.de(T.Leménager).
Konstanzer Online-Publikations-System (KOPS)
URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-146imnbluj1e81 Erschienen in: Behavioural Brain Research ; 324 (2017). - S. 1-14
https://dx.doi.org/10.1016/j.bbr.2017.01.046
Conclusion:Giventhefunctionoftheleftmiddletemporalgyrusintheretrievalofwordsorexpressions duringcommunication,ourfindingsgiveafirsthintthatsocialwordsmightbelessretrievableinthe semanticstorageofinternetgamingaddicts,possiblyindicatingdeficienciesinhandlingspeechinsocial situations.
1. Introduction
Internetaddictioncurrentlyisthefastestgrowingaddictionand gainsworldwideattentionduetoits’risingprevalence[1].Inter- netgamesandsocialnetworks(subsumingtheexcessiveuseof chats,forumsandsocialnetworkssuchasFacebook)arethetwo mostcommonlyusedapplicationsbyinternetaddicts[2,3].Both involvecommunicationandsocialcognitions[4],buthavenever beenexploredinthisspecificcombinationbefore.
The5theditionoftheDiagnosticandStatisticalManualofMen- talDisorders(DSM-5;AmericanPsychiatricAssociation,[5])refers onlytotheconditionofproblematicinternetgameusetermedas
“InternetGamingDisorder”,whileitisoftendesignatedas“Inter- netGamingAddiction”(IGA)inthecurrentliterature.Forreasons ofuniformityaswellastounderlinethecondition’saddictivechar- acter,theterms“IGA”and“socialnetworkaddiction”areapplied inthismanuscriptandbotharesummarizedunderthedesignation of“specificinternetaddiction”.
Similartosubstanceaddictions[6,7],IGAhasbeenassociated withincreasedsocialandgeneralanxiety[8,9],impulsivity[10]and emotionalcompetencedeficits[11],thelatterpotentiallyimpair- ingaddictsintherecognition,control,regulationandexpression oftheirownemotionsaswellasintherecognitionofother’semo- tions.
Furthermore,thefewexistingpsychometricstudiesonsocial networkusereportedpathologicalFacebookusetobeassociated withpooremotionregulationskills,comprisingalackofacceptance ofownemotionalresponses,limitedaccesstoemotionregulation strategies,poorimpulsecontroland,similartoIGA,increasedsocial andgeneralanxiety[12–14].Gamesand social networksmight especiallycapturetheirusersbyfacilitatingtheexperienceofpos- itive social emotions(e.g.viasocial contacts ina game [15] or positivesocial feedbackin socialnetworks[12]),which socially anxiousindividualsmightnotbeabletogaininreallife.
Brandetal.[4]consideredtheseaspectsin their‘Interaction ofPerson-Affect-Cognition-Execution’(I-PACE)modelofinternet use.Theauthorssuggestthatgratification(e.g.socialrecognition), leadingtopositiveandnegativereinforcement,isoneofthecore motivesintheearlystagesofspecificinternetusedisorders.With repeateduse,gratificationdecreasesandthecompensationforneg- ativefeelings(e.g.loneliness)andcognitions(e.g.“Iamnobody withouttheinternet”)gainsincreasingimportance:Theaddicted useofspecificinternetapplicationswhichincludecommunication featuresmightcompensatefortheperceivedlackofsocialsupport, socialdistrust,feelingsofisolation,andlonelinessintherealworld [16–19].Thesenegativeemotionalconditionsmightbeinduced byincreasedsocialanxietyandimpairedemotionregulation,i.e.
emotionalinhibitorycontrol,inreallifesocialcontexts.Emotional inhibitorycontrolisdefinedastheabilitytocontrolonesimpulses, emotionsandthoughts[20]andcanbeassessedbymeansofAffec- tiveGo/No-Go(AGN;e.g.[21,22])aswellasEmotionalStroopTasks (EST;[23]).
TheAGNtaskassessesparticipants’emotionalresponseinhibi- tionabilitybymeasuringtheamountofresponseerrorsinterms ofcommissionerrors(definedasbuttonpressestoNo-Gostimuli and thereforemirroringfailedinhibition [22]).The ESTinvesti- gatesinterference(i.e.theabilitytosuppressirrelevantemotional
informationduringcognitiveperformance)bymeasuringtheReac- tionTime(RT)forthecolornamingofdifferentlyvalencedwords (amongotherparameters;[24]).
Neurobiologicalstudies suggesta complex network of brain regionsincludingtheprefrontalcortex(PFC),amygdala,hippocam- pus, insular cortex, ventral striatum and other interconnected regionstobeinvolvedinemotionalinhibitorycontrol[25,26].In particular,theroleofthePFCinthisdomainhasbeenreinforced byresearchin therecentyears [27].Thedorsolateralprefrontal cortex(DLPFC)issuggestedtobeinvolvedinthecognitiveselec- tionof sensoryinformation [26] andtheorbitofrontal cortex is relatedtorewardanticipationaswellasdecisionmakingunder ambiguity[28,29].The AnteriorCingulateCortex (ACC)is func- tionallyinvolvedinconflictmonitoringandininhibitionprocesses, regardlessofstimulustype[4,30].Structurally,severalrecentstud- iessuggest reduced graymattervolume inthe ACCof internet andinternetgamingaddictsrelativetohealthycontrols[31–33].
Moreover,Wanget al.(2015)observed anassociationbetween reducedgraymattervolumeoftheACCand impairedcognitive inhibitorycontrolprocessingininternetgamingaddictsduringthe performanceofaclassicalcolor-wordStrooptask[32].Thereare nocomparablestudiesforemotionalinhibitorycontrolprocessing relatedtopositive,negativeoranxiousaffectsininternetaddicts, althoughincreasedsocialandgeneralanxietyhavebeenassociated withinternetgamingand social network addiction[8,9,12–14].
Thus,thepsychologicalandneuralbasisofemotionalinhibitory control in relationto social and generalanxiety remains tobe explored and compared between healthy controlsand specific internetaddicts(usinginternetgamesorsocialnetworks).Specif- icallythedorsalACC(dACC)hasbeenfunctionallyassociatedwith bothnegativeaffect(suchassocialexclusion,painoranxiety)and cognitivecontrol[26,34–36].Byconductinga ‘coordinate-based’
meta-analysisof939studiesinordertoassesscommonordiffer- entactivationclustersintheACCduringnegativeaffect,painand cognitivecontrol,Shackmanetal.[36]observedthatallthesethree domainsactivateacommonregionwithinthedACC[36].
BasedonthetheoreticalmodelofBrandetal.[4]andempirical findings suggesting increasedsocial and general anxiety, emo- tionregulationdeficitsandimpairedinhibitorycontrolininternet addictionofspecificapplicationslinkedtosocialcommunication, thepresentstudyaimstoassessneuropsychologicalandneuro- biologicaldifferencesin emotionalinhibitorycontrolprocessing relatedtosociallyandgenerallyanxiousstimulibetweenspecific internetaddicts(i.e.internetgamingandsocialnetworkaddicts) andhealthycontrols.
While we neuropsychologically assessed social and general anxiety-related emotional inhibitory control by means of the AGN(responseinhibition)andEST(interference),interferencein relationtosociallyanxiousstimuliwasadditionallyassessedby functionalMagneticResonanceImaging(fMRI).
Firstly, we hypothesized specific internet addicts to show impairedemotionalinhibitorycontrol,especiallyrelatedtoanxious stimuli,asreflectedinworseemotionalresponseinhibition(i.e.
morecommissionerrors;[22,37])duringanxiety-relatedblocksof theAGNandastrongerinterference(i.e.longerRTs;[24])tosocially anxiouswordsrelativetohealthycontrolsintheEST.Furthermore, onthebackgroundofthedACC’sroleinanxiety-relatedaffect(e.g.
socialexclusion)andcognitivecontrolaswellasincreasedsocial anxietyinspecificinternetaddiction,wesecondlyhypothesized specificinternetaddictstohavestrongerinhibitorydemandsthan controlsinresponsetosocialanxiety-relatedstimuli(relativeto positive, negativeandneutral words),reflected inaltered dACC activationduringESTperformance.
Wefurtherexplorativelyassessedwhetherthesemeasuresdif- ferbetweenaddictedinternetgamersandsocialnetworkusersand whethertheyarerelatedtotheseverityofaddictionaswellasto anxietyandsocialanxiety-relatedmeasures.
2. Methods
2.1. Participants
Thestudywasapprovedbythelocalethicscommittee(applica- tionnumber2013-528N-MA).Participantswererecruitedatthe local day clinicof the Department of Addictive Behaviour and AddictionMedicineat theCentralInstituteof MentalHealth in Mannheim,aswellasbymeansofadvertisementsonthewebpages oftheCentralInstituteofMentalHealthandMannheimUniver- sity.Allparticipantswereinformedaboutthestudyprocedures andgavewritteninformedconsentaccordingtotheDeclarationof Helsinkipriortostudyparticipation.Forthisstudy,weincluded n=44healthycontrolsandn=51specificinternetaddicts(com- prisingn=30internetgamingandn=21socialnetworkaddicts).
DuetofMRIincapability(e.g.lefthandedness,metalpartsinthe body),theAGN samplewasdecimated toanEST subsample of n=23healthycontrolsandn=25specificinternetaddicts(subdi- videdinton=13internetgamingandn=12socialnetworkaddicts).
Decisivefor groupassignmentwasa clinicalinterviewwithan experiencedpsychologistandtheobtainedscoresonthesubscale ofthechecklistfortheAssessmentofInternetandComputergame Addictionwithintheprevious30days(AICA30,see“Psychometric instruments”;[38]).
Asan additional measure for addictionseverity, weapplied thescaleforonlineaddictivebehavior inadultsasaself-report questionnaire(OSVe-S; Germandesignation:Skala zumOnline- suchtverhaltenbeiErwachsenen;[39];seebelow).Participantsdid nothaveanyfurtheraxis-Ipsychiatricdisordersorsubstanceuse disorders(exceptnicotinedependence),werenottreatedwithany psychotropicmedicationandhadnormalorcorrected-to-normal vision.Potential psychiatriccomorbiditieswereassessedonthe basisoftheStructuredClinicalInterviewforDSM-IV(SCIDIand II;[40]).Therewasnosignificantdifferencebetweenthegroups regardingageorgender(althoughtherewereslightlymoremales intheaddictedgroupthaninthecontrolgroup).Specificinter- netaddictsoftheESTsubsamplewerelesseducatedthanhealthy controls(seeTable1).Regardingthethreesubgroupsofinternet gamingand social network addicts aswellas healthy controls, thecorrectedBonferronipost-hoctestsrevealednosignificance betweenthegroups(Tables2and3).
Thehoursofcomputerandinternetuseperdayaswellasthe addictionseverity(AICA30,AICAlifetimeandOSVe-S)weresig- nificantlydifferentbetweenspecificinternetaddictsandcontrols (seeTable1).
2.2. Psychometricinstruments
Participantswerescreenedforexistenceand degreeof com- puterandinternetaddictionbymeansofaninterviewbasedonthe AICA[38].Thelatterisanestablisheddiagnosticclinicalinterview assessing theseverityof participants’ computerand/orinternet addictionduringtheprevious30days (AICA30)aswellasover thelifetime(AICAlifetime)byrecordingparticipant’scomputeror
internetuse(e.g.“Isthereanyimpairmentinthepersonalareaof lifeduetotheusageofcomputergames/internetoffers?”).Besides, itwasmadeuseoftheOSVe-S[39],asitadditionallyenablesthe differentiationbetweenabusive(cut-off <13.5and>7)andnor- malinternetuse(cut-off<7);e.g.“Howstrongareyouoccupied withthoughtsaboutonlineoffers/activitiesperday?”).Participants scoring≥13ontheAICA30or≥13.5ontheOSVe-Swereassigned totheaddictedgroup.Participantsscoring<7ontheOSVe-Swere assignedtothecontrolgroup[38].TheBarrattImpulsivenessScale (BIS-11;[41])wasusedtorecordimpulsivityandfeelingsofstress wereassessedbymeansofthePerceivedStressScale(PSS,[42]).
Theneuroticismscaleof theNEO-FFIservedasanindicatorfor unpleasantemotionssuchasanxiety[43].Participant’sdegreeof social anxiety wasexploredby means of thequestionnaire for SocialAnxietyandSocialCompetencedeficits(SASKO;[44]).The latter investigatesthe anxietyofspeaking in frontof others or beinginthefocusofsocialattention(subscale‘speaking’),ofbeing sociallyrejected(‘rejection’),ofinteractingsocially(‘interaction’), deficitsinsocialperception(‘information’)aswellasfeelingsof loneliness(‘loneliness’).Theemotionalcompetencequestionnaire (EKF; German designation: Emotionale-Kompetenz-Fragebogen) wasappliedfortheassessmentoftheself-ratedemotionalcompe- tence[45].Theinstrumentcomprises62items,whicharedivided intofoursubscalesmeasuringtheconstructsofa)recognizingand understandingone’sownemotions(EKF-EE),b)recognizingand understandingothers’ emotions(EKF-EA),c)theabilitytoregu- lateandcontrolone’sownemotions(EKF-RE)andd)emotional expressiveness(EKF-EX).
2.3. AffectiveGo/No-Gotask(AGN)
TheAGNtaskwasappliedinavalidatedGermanversion[46]as partoftheCambridgeNeuropsychologicalTestAutomatedBattery (CANTAB)to assessparticipant’s emotional response inhibition.
TheAGNcomprised13blocksofeither18positiveandneutral,neg- ativeandneutral,oranxiousandneutralwords,asalreadygiven intheGermanversion.Thevalencedwords(i.e.positive,negative andanxious)weretargetstimuli,whiletheneutralwordsrepre- senteddistractorstimuli.Therecognitionoftargetstimuliwasto beacknowledgedbymeansofabuttonpressasfastaspossible, whilethereactiontoneutralstimuliwastobeinhibited.Theorder oftheblock’svalence(i.e.positive,negativeoranxiousblocks)was constantacrossparticipants,whilethepresentationofvalencedvs.
neutralwordswithineachblockwasrandomized.Singlewords appearedonthetestingscreenafterintervalsof900msandwere presented for300ms. Thecommission errors intherespective blocks ofanxiousvalence(i.e.amountofbuttonpresses todis- tractor stimuli,here: neutral words) served astest outcome. A lowernumberofcommissionerrorsreflectstheabilitytoinhibit unwantedbehaviorinanemotionalcontext(i.e.valencedwords) andthereforeservesasanindicatorforanindividuals’abilityof emotionalresponseinhibition.
2.4. EmotionalStroopTask(EST)
TheEST wascarried out duringfMRI. Thetaskwasadapted fromWitthöft etal.[24] toincludethefourcategories ofposi- tive,negative,neutralaswellassociallyanxiouswords([24];see Table4).Throughouttheparadigm,eachcategorywaspresented fourtimes,resultinginatotalof16blocks.Eachofthewordstim- uliwaspresentedonceinoneofthecolorsred,green,blueand yellow,distributedacrosstheblocks.Participantswereaskedto indicatethecolorofeachpresentedwordasfastaspossibleby pressingoneofthefourbuttonscorrespondingtothefourcolors (indexfinger=red, middlefinger=green,ring finger=blue,little finger=yellow).Wordswerepresented oneafteranotherinthe
Table1
Sampledescriptionforhealthycontrolsandspecificinternetaddicts.Thestudywasperformedwithalarge(AGN)sampleaswellaswithasubsample(EST)forfMRI explorations.
AGNsample ESTsample
Overall sample (N=95)
Healthy controls (n=44)
Specific internet addicts (n=51)
Test-statistic p-value Overall sample (N=48)
Healthy controls (n=23)
Specific internet addicts (n=25)
Test-statistic p-value
Gender (n=male)
52 (54.74%)
20 (45.45%)
32 (62.75%)
2.850x2(CT) 0.091 27 (56.25%)
10 (43.48%)
17 (68.00%)
2.927x2(CT) 0.087 Age
(SD)
27.15 (8.21)
28.59 (9.80)
25.90 (6.39)
−1.198z 0.231 25.85 (5.67)
26.13 (5.49)
25.60 (5.93)
−0.591z 0.554 Yearsof
education (SD)
14.86 (2.39)
15.30 (2.30)
14.49 (2.42)
−1.626z 0.104 15.06 (2.43)
15.87 (2.26)
14.32 (2.38)
−2.216z 0.027*
Computer- /internet-use (Ø,h/day;SD)
2.73 (1.36)
1.89 (.90)
3.45 (1.27)
−5.576z <0.001** 2.63 (1.36)
1.91 (.90)
3.28 (1.40)
−3.348z 0.001**
AICA30 (SD)
9.07 (7.90)
2.25 (1.95)
14.96 (6.13)
−7.818z <0.001** 8.38 (7.53)
2.35 (1.87)
13.92 (6.41)
−5.455z <0.001**
AICAlifetime (SD)
14.75 (10.67)
4.16 (2.60)
23.88 (4.96)
−8.386z <0.001** 14.46 (10.08)
4.87 (2.58)
23.28 (4.86)
−5.945z <0.001**
OSVe-S (SD)
8.25 (5.83)
2.93 (1.55)
12.84 (3.92)
−8.386z <0.001** 7.64 (5.39)
3.09 (1.57)
11.82 (4.06)
−5.892z <0.001**
SD=StandardDeviation,AICA30=AssessmentofInternetandComputergameAddictionoftheprevious30days,AICAlifetime=LifetimemaximumofInternetandComputer gameAddiction,OSVe-S=Self-reportquestionnaireforinternetaddiction-relatedbehavior(SkalazumOnlinesuchtverhaltenbeiErwachsenen),z=Mann-WhitneyUtest statistic,x2(CT)=Chi2Crosstabteststatistic;*p≤0.05,**p≤0.01.
Table2
Sampledescriptionforthesubgroups.
AGNsample ESTsample
Healthy controls (n=44)
Internet gamingaddicts (n=30)
Socialnetwork addicts (n=21)
Test-statistic p-value Healthy controls (n=23)
Internet gamingaddicts (n=13)
Socialnetwork addicts (n=12)
Test-statistic p-value
Gender (n=male)
20 (45.45%)
22 (73.33%)
10 (47.62%)
6.147x2(CT) 0.046* 10 (43.48%)
11 (84.62%)
6 (50.00%)
5.965x2(CT) 0.051 Age
(SD)
28.59 (9.80)
27.07 (7.73)
24.24 (3.24)
2.387x2(KW) 0.303 26.13 (5.49)
27.15 (7.39)
23.92 (3.34)
1.938x2(KW) 0.380 Yearsof
education(SD) 15.30 (2.30)
14.30 (2.38)
14.76 (2.51)
3.139x2(KW) 0.208 15.87 (2.26)
14.69 (2.36)
13.92 (2.43)
5.420x2(KW) 0.067 Computer-
/internetuse (Ø,h/day;SD)
1.89 (.90)
3.67 (1.12)
3.14 (1.42)
32.975x2(KW) <0.001** 1.91 (.90)
3.69 (1.32)
2.83 (1.40)
13.434x2(KW) 0.001**
AICA30 (SD)
2.25 (1.95)
15.33 (6.60)
14.43 (5.51)
61.346x2(KW) <0.001** 2.35 (1.87)
14.69 (7.49)
13.08 (5.20)
30.024x2(KW) <0.001**
AICAlifetime (SD)
4.16 (2.60)
25.77 (3.58)
21.19 (5.47)
73.254x2(KW) <0.001** 4.87 (2.58)
25.85 (3.56)
20.50 (4.64)
37.075x2(KW) <0.001**
OSVe-S (SD)
2.93 (1.55)
13.28 (3.93)
12.21 (3.92)
70.230x2(KW) <0.001** 3.09 (1.57)
12.54 (4.40)
11.04 (3.68)
34.899x2(KW) <0.001**
SD=StandardDeviation,AICA30=AssessmentofInternetandComputergameAddictionoftheprevious30days,AICAlifetime=LifetimemaximumofInternetandComputer gameAddiction,OSVe-S=Self-reportquestionnaireforinternetaddiction-relatedbehavior(SkalazumOnlinesuchtverhaltenbeiErwachsenen),x2(CT)=Chi2Crosstabtest statistic,x2(KW)=Chi2Kruskal-Wallisteststatistic;*p≤0.05,**p≤0.01.
middleofthescreen,whiletheassignmentofthefourbuttonstothe respectivecolorwasdisplayedthroughoutthetaskatthebottomof thescreeninsmallerwriting.Thesefourcolorwordswerewritten intherespectivecolor.Forthebehavioralanalyses,RTdifferences ofcorrectresponsesforsociallyanxious,positiveandnegativerel- ativetoneutralstimuliwereusedasthemainbehavioraloutcome measureforemotionalinterference.
Thewordandcolororderwithineachblockwasrandomized andtheblocksofthedifferentcategorieswerepresentedinthe orderofaLatinsquare.The10wordspercategorywerematched category-wiseforwordlength,frequency,valenceandarousalin writtenGermanlanguage.Theywerepresentedfor1.5s,followed byafixationcrossasinter-stimulusintervalwithameanduration of0.3s[24].Betweentheblocks,afixationcrossand theinfor- mationthatanextblockwillfollowwerepresentedfor9sand 1s,respectively.Thedurationofeachblockamountedto18sand thecompleteparadigmtookabout7min.Priortotheactualtask
performance,participantscompleteda shorttraining session in thescanner,butwithoutscanning.Thetrainingparadigminvolved arraysoflettersinsteadofrealwordsinordertogetusedtothe responsebuttons.Theexperimentwasimplementedwiththepro- gramPresentationVersion16.3(NeurobehavioralSystems,Albany, Calif., USA)and taskpresentationin thescannerwasmediated bydigitalgoggles(ResonanceTechnology,Inc.,LosAngeles,Cali- fornia).Afterthescanning,allwordspresentedasstimuliinthe ESTwereratedviapaper-pencilbyeachparticipantregardingeach word’svalence(i.e.emotionalresonance)andarousalona9-point scalewiththehelpoftheSelf-AssessmentManikin(SAM)affective ratingsystem[47].
Specificinternetaddictsdidnotshowsignificantdifferencesin arousalandvalencecomparedtohealthycontrols[Mann-Whitney U(MWU)tests].Within-groupanalyses(Friedmantests)compar- ingthevalenceandarousalbetweenthefourcategoriesrevealed significantdifferencesinallcomparisons.Onlythevalencebetween
Table3
Post-hoctestsforthesubgroups.P-valuesBonferronicorrected.
AGNsample ESTsample
Healthy controls-internet gamingaddicts
Healthy controls-social networkaddicts
Internetgaming addicts-social networkaddicts
Healthy controls-internet gamingaddicts
Healthy controls-social networkaddicts
Internetgaming addicts-social networkaddicts Gender
(n=male)
5.649x2(CT) 0.017 0.027x2(CT) 0.870 3.494x2(CT) 0.062 Age(SD)
Yearsofeducation (SD)
Computer- /internetuse (Ø,h/day;SD)
−5.601z <0.001** −3.411z 0.001** −1.259 0.208 −3.551z <0.001** −1.932z 0.053 −1.592z 0.111
AICA 30 (SD)
−6.756z <0.001** −6.128z <0.001** −0.969 0.333 −4.654z <0.001** −4.304z <0.001** −0.791z 0.429 AICAlifetime(SD) −7.283z <0.001** −6.498z <0.001** −3.181z 0.001** −4.940z <0.001** −4.810z <0.001** −2.517z 0.012*
OSVe-S (SD)
−7.278z <0.001** −6.548z <0.001** −0.872z 0.383 −4.879z <0.001** −4.816z <0.001** −0.872z 0.383
SD=StandardDeviation,AICA30=AssessmentofInternetandComputergameAddictionoftheprevious30days,AICAlifetime=LifetimemaximumofInternetandComputer gameAddiction,OSVe-S=Self-reportquestionnaireforinternetaddiction-relatedbehavior(SkalazumOnlinesuchtverhaltenbeiErwachsenen),z=Mann-WhitneyUtest statistic,m=MeandifferenceintheTukeyHSDpost-hoctest;*p≤0.05,**p≤0.01,x2(CT)=Chi2Crosstabteststatistic.
Fig.1.RatingsforvalenceandarousalofESTwords.
Table4
OriginalandEnglishESTstimuliwords.
Social anxiety-related words
Positivewords Negativewords Neutralwords
OriginalGermanwords
ängstlich anziehend verendet förmlich
hilflos super habgierig belanglos
einsam lebenslustig unangenehm neutral
unterwürfig euphorisch zerfressen zusammenhanglos
blamiert großartig böse bürgerlich
sitzengelassen interessant verfault solide
panisch vergnügt impotent nichtssagend
gehemmt glücklich asozial typisch
abstoßend leidenschaftlich herablassend anatomisch
dämlich witzig verwahrlost eckig
Translatedwords
anxious attractive perished formal
helpless super avaricious insignificant
lonely fun-loving awkward neutral
submissive euphoric eroded incoherent
disgraced great bad civil
abandoned interesting rotten solid
panic-fueled cheery impotent expressionless
inhibited happy antisocial typical
repellent passionate dismissive anatomical
dimwitted funny shabby angled
sociallyanxiousandnegativewordsdidnotsignificantlydifferin specificinternetaddictsaswellasinhealthycontrols(seeFig.1).
AKruskal-Wallistestfortheexplorationofbetween-subgroup differences revealed that internet gaming addicts rated the valenceofpositivewordssignificantlylowerthansocialnetwork addicts(x2=8.687;p=0.013;post-hoczpositive=−2.807,p=0.005).
Within-group analyses(Friedman tests) showedsignificant dif- ferencesin valence and arousal betweeneach category, except betweennegativeandsociallyanxiouswordblocksforvalence(in everygroup)aswellasbetweensociallyanxiousandnegativeones inthegroupofsocialnetworkaddictsforarousal(seeFig.1).
2.5. fMRI
The scanning sessions were conducted at a 3T whole- bodytomograph(Trio;Siemens,Erlangen, Germany).Functional whole-brain images were collected under the application of a T2*-weighted Echo-Planar Imaging (EPI) sequence [Repeti- tion Time (TR)=2000 ms, Echo Time (TE)=30 ms, Flip Angle (FA)=80◦,FieldofView (FOV)=192mmx192mm,matrix size 64×64, 32 slices, slice thickness=3.00mm, distancefactor=33
%,voxelsize=3×3×3mm]. Thenumberof acquiredfunctional volumesamountedto219. Structuralimageswere acquiredby means of a T1-weighted Magnetization Prepared Rapid Gradi- entEcho(MPRAGE)sequence(TR=2300ms,TE=3.03ms,FA=9◦, FOV=256mm×256mm,192slices,slicethickness=1.00mm,dis- tancefactor=50%,voxelsize=1×1×1mm)inordertoexclude participants exhibiting brain abnormalities (see [48]). The cor- rection of magnetic field inhomogeneity was mediated by the automatedSiemensmulti-angleprojection(MAP)-Shim.Thereg- istrationofscannertriggersaswellastherecordingofbehavioral responseswasaccomplishedbythePresentation®software(Ver- sion16.3,NeurobehavioralSystems,Inc.,Albany,CA,USA).
3. Statisticalanalyses
The statistical analyses were conducted for the comparison betweenhealthycontrolsandspecificinternetaddicts.Thecom- parisonsbetweenhealthycontrolsandaddictsofgamingaswell
associalnetworkswereconductedexploratively.Wealsoincluded yearsofeducationasacovariateinouranalyses.
3.1. Psychometryandneuropsychology
Psychometric dataanalyseswereaccomplishedby meansof SPSSStatistics20(StatisticalPackagefortheSocialSciences,SPSS Inc.,Chicago,IL;Release20.0.0).Afterhavingcheckedthevariables fornormality,withinandbetween-groupdifferencesinthepsycho- metricaswellasinthebehavioraloutcomemeasuresoftheESTand theAGNwereassessedwithparametricpairedandunpairedt-tests (normallydistributedvariables)orvianonparametricWilcoxon’s signedrankandMWUtests(non-normallydistributedvariables), respectively.Fortheexplorativethree-groupcomparisons,Anal- yses of Variance (ANOVA) and Kruskal-Wallis tests as well as Friedmantestsforthecomparisonacrosscategorieswereapplied.
Forpost-hoctests,aBonferronia-correctionwasdone.
Furthermore,Spearman’srhocorrelationswereappliedinorder toexploreassociationsbetweenaddictionseverityandtheout- come measuresfor emotional response inhibition (commission errorsintheAGNtask)andinterference(RTofthecorrectresponses intheESTtask).
ThedifferencesinRToftheESTbetweenemotionalandneutral words (i.e. positive-neutral, negative-neutral, socially anxious- neutral),betweensociallyanxiousvs.positiveandnegativewords aswellasbetweensociallyanxiousvs.positive,negativeandneu- tral wordswere used for statistical withinand between-group analyses. As described by Witthöft et al. [24], extreme RTs of
<300msand>1000mswereremovedfromthedata[24].
3.2. Neurobiology
Forneurobiologicalevaluations,StatisticalParametricMapping (SPM,versionSPM8,WellcomeTrustCentreforNeuroimaging, UniversityCollegeLondon,London,UK)wasutilizedatthebasisof MatlabR2012b(TheMathWorks,Inc.,Natick,MA,USA).Thedata preprocessingcomprisedmotioncorrection,realignment,spatial normalizationtotheMontrealNeurologicalInstitute(MNI)tem- platewithanewvoxelsizeof3×3×3mmandsmoothingwithan 8mmfullwidthathalfmaximumkernel.
First-levelanalysesincludedtheevaluation ofeffectestima- tions forthefourcategories (i.e.positive, negative,neutral and sociallyanxious)byapplyingthegenerallinearmodelwithregres- sorsfortheonsetsanddurationofeveryconditionaswellasthesix movementparametersobtainedfromtherealignmentpreprocess- ingsteponavoxel-by-voxelbasis.Inasucceedingstep,individual contrastimagesofcondition-specificmeanbrainactivationforour contrastofinterest,i.e.socialanxietyvs.positive+negative+neutral (rest), were calculated. By means of second-level analyses, the individualcontrastimageswerethencomparedfordifferencesin BloodOxygenationLevelDependent(BOLD)responseswithinand betweenthegroupsforourcontrastofinterestbasedononeand two-samplet-testsaswellasfullfactorialANOVAs.Athresholdof p<0.001aswellasanextentthresholdof10voxelswereapplied andresultswerereportedataclusterthresholdofpFWE≤0.05.The AutomaticAnatomicalLabeling(AAL)-toolboxofSPMwasusedfor theassignmentofsignificantlyactivatedclusterstotheirrespec- tivebrainregions. The detectionof BrodmannAreas(BAs) was conductedviathexjview-toolbox(xjViewToolboxforSPM;http://
www.alivelearn.net/xjview8/)rununderMatlab.Activationmaps weredisplayedbymeansof ChrisRorden’sMRIcrobrainimage viewerinneurologicalconvention(i.e.left=left,right=right)on thech2bet-template(ChrisRorden,1999–2005,version1.40build 1).
Table5
PsychometricmeasuresoftheAGNandESTsample.Givenarethemeanswithstandarddeviationsinbrackets.
AGNsample ESTsample
Healthy controls (n=44)
Specificinternet addicts (n=51)
Test-statistic p-value Healthy controls (n=23)
Specificinternet addicts (n=25)
Test-Statistic p-value
Socialanxiety(sum SASKO)
21.95 (13.53)
49.94 (24.01)
−5.816z <0.001** 22.96 (13.98)
46.52 (24.33)
−3.768z <0.001**
SASKO:Speaking 6.41(4.27) 14.00(7.92) −4.905z <0.001** 7.09 (3.85)
13.40 (7.65)
−3.066z 0.002**
SASKO:Rejection 5.68(3.67) 12.73(6.95) −5.223z <0.001** 6.00 (3.90)
11.44 (6.39)
−3.197z 0.001**
SASKO:Interaction 3.86(3.49) 9.76(5.34) −5.616z <0.001** 4.26 (3.73)
8.84 (5.59)
−3.175z 0.001**
SASKO:
Information
4.70(3.07) 8.90(4.76) −4.281z <0.001** 4.65 (2.96)
8.72 (5.23)
−2.773z 0.006**
SASKO:Loneliness 1.30(1.79) 4.55(3.72) −4.525z <0.001** 0.96 (1.46)
4.12 (3.41)
−3.721z <0.001**
Emotional competence(mean EKF)
60.59 (6.37)
52.19 (7.62)
5.773t <0.001** 60.90 (5.38)
51.21 (7.99)
4.885t <0.001**
EKF-EE 60.80
(8.15)
51.20 (10.52)
4.911t <0.001** 61.43 (7.27)
51.16 (10.31)
3.959t <0.001**
EKF-EA 68.36
(9.19)
63.27 (12.07)
2.283t 0.025* 68.83
(8.24)
60.68 (12.57)
2.630t 0.012*
EKF-RE 51.89
(6.49)
44.55 (9.47)
4.332t <0.001** 52.35 (6.45)
45.92 (8.42)
2.950t 0.005**
EKF-EX 61.30
(11.69)
49.75 (14.09)
4.306t <0.001** 61.00 (11.26)
47.08 (15.13)
3.590t 0.001**
Neuroticism (NEO-FFI)
1.25 (0.67)
2.01 (0.79)
−4.434z <0.001** 1.20 (.68)
1.87 (.71)
−3.067z 0.002**
Perceivedstress (PSS)
12.98(5.80) 19.96(7.09) −5.199t <0.001** 13.43 (6.13)
19.16 (7.51)
−2.879t 0.006**
Impulsivity(sum BIS-11)
56.09 (8.76)
63.57 (10.99)
−3.399z 0.001** 54.65 (9.05)
65.20 (11.66)
−3.348z 0.001**
SASKO=SocialAnxietyandSocialCompetencedeficits-scale,EKF=Emotionalcompetencequestionnaire,EKF-EE=Recognizingandunderstandingtheownemotions,EKF- EA=Recognizingandunderstandingothers’emotions,EKF-RE=Regulationandcontroloftheownemotions,EKF-EX=Emotionalexpressiveness,NEO-FFI=NEOFiveFactor Inventory,PSS=PerceivedStressScale,BIS–11=BarrattImpulsivenessScale,z=Mann-WhitneyUteststatistic,t=tstatistic,*p≤0.05,**p≤0.01.
4. Results
Psychometrically,weobservedsignificantdeficitsinimpulsiv- ityandperceivedstressaswellasonallscalesofsocialanxiety andemotionalcompetenceinspecificinternetaddictscompared tohealthycontrols(Table5).
Explorative analysesinthe subgroupsshowedthat between healthycontrolsandinternetgamingaddictsmostofthepsycho- metricvariablesdifferedsignificantly(seeTables6and7).Healthy controlsandsocialnetworkaddictsshowedaheterogeneouspat- ternofsignificantresultsacrossthevariables.However,regarding theresultsin theAGN andESTsamples,social networkaddicts showedsignificantly higher scoresin the overall social anxiety score(sumSASKO)aswellasintheSASKOsubscalesof‘speaking’
and‘rejection’.Furthermore,thegeneralemotionalcompetencies were significantly lower in this group. Between both addicted groups(internetgamingandsocialnetworkaddicts),onlytheinter- actionanxiety(AGNsample)andfeelingsofloneliness(AGNand ESTsample)weresignificantlydifferent.
Inallpsychometricanalyses,includingtheyearsofeducationas acovariatedidnotchangetheresults.
4.1. AGN
Regarding the main outcome measure of emotional response inhibition, specific internet addicts and healthy controls showed significantly more commission errors at positive words than at negative or anxious ones (Friedman: x2healthycontrols=32.224, p<0.001; post-hoc zpositive-negative=−4.118, p<0.001; zpositive-anxiety=-4.432, p<0.001; x2specificinternetaddicts=40.923, p <0.001, post-hoc
zpositive-negative=−4.525, p=<0.001; zpositive-anxiety=-4.931, p=<0.001). The same holds for the subgroups of specific internet addicts (x2internetgamingaddicts=18.893, p<0.001; post- hoc zpositive-negative=−3.052, p=0.002; zpositive-anxiety=−3.172, p=0.002; x2socialnetworkaddicts=23.570, p<0.001; post-hoc zpositive-negative=−3.494, p<0.001; zpositive-anxiety=-3.839, p<0.001;seeFig.2).
In contrast to our first hypothesis, we did not find any significant differences in commission errors on anxious stimuli between specific internet addicts and healthy con- trols (zspecificinternetaddictsvs.healthycontrols;anxiousstimuli=−0.494, p=0.621).Wealsodidnotfindanybetween-groupdifferencesin otherwordcategories.Spearmancorrelationsbetweenaddiction severityandAGNanxiety-relatedcommissionerrorsintheoverall sample revealed that the emotional response inhibition ability in anxiety-blocksdecreased (i.e. morecommission errors) with increasing addiction severity(i.e. higher OSVe-S scores; Spear- man’s rho correlation coefficient rcommissionsanxiety-OSVe-S=0.220, p=0.032).
Regardingdifferencesincommissionerrorsforeachcategory betweenthesubgroups(internetgamingaddicts,socialnetwork addictsandcontrols),explorativeKruskalWallistestsrevealedno significantresults.Includingyearsofeducationasacovariateinall neuropsychologicalanalysesdidnotchangethefindings.
4.2. EST 4.2.1. Behavior
Ourdatarevealed97.85%correctresponsesintheoverallsam- ple.