In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information

(1)

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies are encouraged to visit:

http://www.elsevier.com/authorsrights

(2)

ContentslistsavailableatScienceDirect

Biological Psychology

j ou rn a l h o m epa g e :w w w . e l s e v i e r . c o m / l o c a t e / b i o p s y c h o

Emotion response coherence: A dual-process perspective

^夽

CatharineEvers^a,∗,HenrikHopp^b,JamesJ.Gross^c,AgnetaH.Fischer^d, AntonyS.R.Manstead^e,IrisB.Mauss^f

aClinicalandHealthPsychology,UtrechtUniversity,TheNetherlands

bDepartmentofPsychology,UniversityofDenver,USA

cDepartmentofPsychology,StanfordUniversity,USA

dFacultyofSocialandBehavioralSciences,UniversityofAmsterdam,TheNetherlands

eSchoolofPsychology,CardiffUniversity,UK

fDepartmentofPsychology,UniversityofCalifornia,Berkeley,USA

a r t i c l e i n f o

Articlehistory:

Received29April2013 Accepted5November2013 Availableonline14November2013

Keywords:

Emotion

Responsecoherence Automaticresponses Reﬂectiveresponses Dual-processtheories

a b s t r a c t

Emotionsarewidelythoughttoinvolvecoordinatedresponsesacrossmultipleresponses(e.g.,experi- ential,behavioral,andphysiological).However,empiricalsupportforthisgeneral“responsecoherence”

postulateisinconsistent.Thepresentresearchtakesadual-processperspective,suggestingthatresponse coherencemightbeconditionaluponresponsesystem(i.e.,automaticversusreflective).Inparticular, wetestedthehypothesisthatresponsecoherenceshouldbemaximalwithineachsystemandminimal acrossthetwosystems.Totestthisprediction,36participantsunderwentanangerprovocationwhile tworelativelyautomatic(angeraccessibilityandphysiology)andtworelativelyreflective(angerexpe- rienceandinstrumentalbehavior)responsesweremeasured.Aspredicted,coherencewasfoundwithin theautomaticandreflectivesystems,butnotacrossthem.Implicationsforemotionresponsecoherence, dual-processframeworks,andthefunctionsofemotionsarediscussed.

1. Introduction

Acentralpostulateofmanyemotiontheoriesisthatemotions involvecoordinatedchangesacrossexperiential,behavioral,and physiologicalresponses(e.g.,Averill,1980;Ekman,1992;Frijda, Ortony,Sonnemans,&Clore,1992;Lazarus,1991;Levenson,1994;

Scherer,1984;Tomkins,1962).Thisgeneralresponsecoherencepos- tulateisoftenassociatedwithanevolutionaryperspectiveonthe functionofemotions.Byimposing coherenceacrossthevarious componentsofanemotionalresponse,emotionsarethoughtto preparetheorganismfortheactionsrequiredtorespondoptimally toenvironmentaldemands(e.g.,Ekman, 1992;Levenson,1994;

Plutchik,1980).

Despitethecentralityofthispostulate,empiricalevidencefor responsesystemcoherenceissurprisinglyinconsistent,withsome

夽 Author Note:This researchwas supportedby aFulbright grant fromthe NetherlandsAmericanCommissionforEducationalExchange(NACEE)andatravel grantfromtheNetherlandsOrganizationforScientiﬁcResearch(NWO)assigned totheﬁrstauthorandbyafellowshipwithinthePostdoc-ProgramoftheGer- manAcademicExchangeService(DAAD)awardedtothesecondauthor.Wethank TaniaDavila,theStanfordPsychophysiologyLab,andYaeldeLiverfortheirhelpful comments.

∗ Correspondingauthorat:DepartmentofClinicalandHealthPsychology,Utrecht University,POBox80140,3508TCUtrecht,TheNetherlands.Tel.:+31302533301.

E-mailaddress:C.Evers@uu.nl(C.Evers).

research providingsupportin favor of response coherence and otherresearchfailingtosupportit.Toreconciletheseinconsis- tentfindings,thepresentstudyproposesandtestsadual-process frameworkofemotionresponsecoherence,suggestingtwolargely independentsystems:anautomatic(relativelyunconscious,fast, andefficient)andareflective(relativelyconscious,deliberate,and effortful)system.Accordingtothisaccount,responsecoherence shouldbemaximalwithineachsystemandminimalacrossthetwo systems.

1.1. Responsecoherence

Studiesinvestigatingthedegreeofcoherenceamongemotion components are not only relatively rare, they “provide for the greaterpartatbestlimitedsupport”(Reisenzein,2000,p.2)for the assumptionof response coherence.Associationsamong the different emotion componentsare often weaker than expected (e.g.,Bonanno&Keltner,2004;Mauss,Wilhelm,&Gross,2004;

Reisenzein,2000;Ruch,1995),non-existent(e.g.,Jakobs,Fischer,

& Manstead, 2001; Mauss et al., 2004; Reisenzein, Bördgen, Holtbernd,&Matz,2006),orevennegative(e.g.,Buck,1977).These inconsistenciesacrossstudieshaveledsomepsychologiststoargue thatthecoherencepostulatemaybeoverstatedorevencompletely unfounded(Barrett,2006;Bradley&Lang,2000;Fridlund,1994;

Lang,1988;Reisenzein,2000).

http://dx.doi.org/10.1016/j.biopsycho.2013.11.003

(3)

44 C.Eversetal./BiologicalPsychology98(2014)43–49

Atthesametime,discardingthecoherencepostulateentirelyis notconsistentwiththeresearchthathasidentiﬁedatleastsome degreeofcoherenceamongresponses (e.g.,Bonanno&Keltner, 2004;Mauss,Levenson,McCarter,Wilhelm,&Gross,2005;Sze, Gyurak,Yuan,&Levenson,2010).Onesolutiontothistensionisthat coherencemaynotbeanall-or-nonefeatureofemotions.Rather, differentdegreesofcoherencemightbefoundfordifferenttypes ofresponses.Indeed,whendifferentiatingamongdifferenttypes ofresponses,somesystematicdifferencesincoherenceemerge.

Thestrongestassociationsaretypicallyfoundbetweenexperi- enceandbehavior(e.g.,Fischerand Roseman,2007;Zeelenberg

& Pieters, 2004). In contrast, associations between physiologi- cal responses, on the one hand, and experience and behavior, ontheother, are more modestor non-existent (e.g.,Borkovec, Stone,O’Brien,&Kaloupek,1974;Grossman,Wilhelm,Kawachi,&

Sparrow,2001;Mauss,Wilhelm,&Gross,2003;Maussetal.,2004, 2005;Weinstein, Averill,Opton,&Lazarus, 1968).Thus, coher- encehasbeentypicallyfoundbetweenexperienceandbehavior, whereaslesserornocoherencehasbeenfoundbetweenphysiolog- icalresponsesandotherresponses.Whatprinciplecouldaccount for these response-speciﬁc patterns of coherence? In the next section,weproposethatdual-processframeworksmightexplain systematicdifferencesincoherenceacrossdifferentresponses.

1.2. Dual-processframeworksandresponsecoherence

Dual-processframeworksassumethatpsychologicalresponses areajointfunctionoftwolargelyindependentsystems,oneauto- maticandtheotherreflective.Automaticresponsesarerelatively unconscious,fast,andefficient,whilereflectiveresponsesarerela- tivelyconscious,deliberate,andeffortful.Bothsystemsarethought toplayinconcerttopromoteadaptivebehavior,includingemo- tions(Bargh&Ferguson,2000;Baumeister,Vohs,DeWall,&Zhang, 2007;Kahneman &Frederick,2002; Lieberman,2007; Smith &

DeCoster,2000;Smith&Neumann,2005;Strack&Deutsch,2004).

Morespeciﬁcally,dual-processframeworksassumethattheauto- matic system activatesbehavioral schemata through spreading activation,whichoriginatesmainlyfromperceptualinput.Oneof thegreatestadvantagesoftheautomaticsystemisthatitisnot onlyfastbutalsorequires littleornocognitiveeffortandhasa lowthresholdforprocessingincominginformation.Thereﬂective system,incontrast,generatesdeclarativeknowledgebyassigning perceptualinputtoasemanticcategory(e.g.,EvansandStanovich, 2013;Strack&Deutsch,2004).Itisthoughttooperaterelatively slowlyandtoinvolverelativelygreatereffort.

Research – mainly fromthe ﬁeld of social cognition– sup- portstheideaoftwoindependentsystems.Forexample,thereis (a)psychometricevidencethatautomaticand reﬂectiveaspects ofthesameconstructaredistinct(e.g.,Cunningham,Preacher,&

Banaji,2001;Greenwald&Farnham,2000;Hofmann,Gawronski, Gschwendner,Le,&Schmitt,2005;Nosek&Smith,2007;Payne, Burkley,&Stokes,2008);(b)neurologicalevidencethatimplicitand explicitmeasurescorrespondtodistinctcognitiveprocessesanddo notsimplyconstitutedifferentmeasurementmodes(Cunningham, Johnson,Gatenby,Gore&Banaji,2003;Cunninghametal.,2004;

Phelps et al., 2000); and (c) empirical evidence showing that implicit and explicit measures both have different domains of predictivepotency (Dijksterhuis&Nordgren, 2006; Greenwald, Poehlman, Uhlmann, &Banaji, 2009).Taken together,evidence supportstwoindependentsystems:anautomaticandareﬂective system.

Weproposethatonecanapplythisdual-process framework tounderstandemotionresponsecoherence.Thisidealeadstothe predictionthatcoherenceshouldnotbeanall-or-nonefeatureof emotions.Rather,coherenceshouldbeconditionalonthesystem oftheinvolvedemotionalresponse.Relativelyautomaticresponses

shouldcoherewithoneanotherandrelativelyreﬂectiveresponses shouldcoherewithoneanother.However,responsesacrossthe twosystemsshouldcoheretoalesserdegreewithoneanother.

Althoughexistingresearch,asnotedabove,appearstobegen- erallyconsistentwiththenotionoftwoindependentsystemsof coherence,verylittleresearchtodatehasdirectlytestedthisidea utilizingmultiplemeasuresfromwithintheautomaticandmulti- plemeasuresfromwithinthereflectivesystem.Thatis,although previousresearchonresponsecoherencehastypicallyexamined severalindicatorsofthereflectivesystem(usuallyself-reported emotionalexperienceandbehavior),ithastypicallyonlyexam- inedphysiologicalrespondingasthesoleindicatoroftheautomatic system.Inaddition,forsomeemotionalresponsesitisnotentirely cleartowhatextenttheyarerelativelymoreautomaticorreflective (Smith&Neumann,2005).Forexample,inmoststudiestestingthe coherencepostulate,facialexpressivebehaviorwasmeasured(e.g., Bonanno&Keltner,2004;Fernández-Dols&Crivelli,2013;Mauss etal.,2005;Reisenzein,2000;Reisenzeinetal.,2006;Reisenzein, Studtmann,&Horstmann,2013).Facialbehaviorcanoccurinarel- ativelyautomaticorreflectivemode(e.g.,Baumeisteretal.,2007;

Ekman,1972).Therefore,andin lightofthegoalofthepresent study,weincludedbehaviorthatisevidentlymoreatthereﬂective endofthecontinuum(hostileevaluationsoftheexperimenter;see alsobelow).Totesttheideathatcoherenceis conditionalupon responsesystem(automaticversusreﬂective),multiplemeasures fromeach systemare necessary.Thepresentresearchprovided suchdata.

1.3. Thepresentstudy

Thepresentstudytestedthehypothesisthatresponsecoher- enceshouldbemaximalwithineach systemandminimalacross thetwosystems.Weexaminedemotionalresponsesinthecon- textofangerbecauseangerinvolvespronouncedresponsesinall emotionalcomponents(Mauss,Cook,&Gross,2007;Reisenzein, 2000).Toinduceanger,weusedawell-validatedangerprovoca- tionprocedure(Mauss,Cook,&Gross,2007;Stemmler,Heldmann, Pauls,&Scherer,2001).Weassessedtwoautomatic(angeraccessi- bilityandphysiologicalresponses)andtworeﬂective(self-reported experienceandinstrumentalbehavior)angerresponses.

First, concerning the automatic system, we assessed anger accessibility by means of a lexical decision task (cf. Bargh &

Ferguson,2000;Niedenthal&Setterlund,1994).Inthistaskpar- ticipantshavetodecideasfastaspossiblewhetheragivenletter stringisorisnotaword,withsomewordsbeinganger-related.

These lexicaldecisions occurrelatively fastand do not depend ontheindividualhavingtheintentionorawareness toevaluate thecontentofthewords,andthusconstituterelativeautomatic responses.Additionally, weassessed physiologicalresponsesby measuringmeanarterialbloodpressure,akeyresponseinthecon- textofanger(Stemmleretal.,2001).Peoplearegenerallyrelatively unawareofthesebodilyresponses,andtheyarerelativelydifﬁ- culttocontrol(Edelmann&Baker,2002;Katkin,1985;Pennebaker, 1982).Physiologicalresponsesthusconstituterelativelyautomatic responses.

Second, concerning the reflectivesystem, we assessed sub- jective experience of an emotion, which was measured with participants’self-reportedangerexperience.Theseemotionexpe- riences wereconsidered tobereflectiveresponses, asthey are controllable (e.g., Mauss, Cook, Cheng, & Gross, 2007) and by definition subject to introspection. Our second measure of the reflective emotional system was instrumental anger behavior, whichwasmeasuredbyobtainingparticipants’anonymouseval- uationsofthepersonwhoangeredthem.Participantswereledto believethattheirevaluationscouldhavedisadvantageouseffects forthecareeroftheirobjectofanger.Thisbehaviorwastherefore

(4)

Table1

Means,standarddeviations,andintercorrelationsofstudyvariables(N=36).

M SD 1 2 3 4 5 6 7 8

1.Sex^a 1.45 .51 –

2.Age 21.03 1.52 −.09 –

3.Angeraccessibilitycontrol −.01 .06 −.17 .07 –

4.Angeraccessibilityprovocation −.04 .05 .22 −.01 −.15 –

5.Bloodpressurebaseline 93.74 17.04 −.18 .34 .15 .25 –

6.Bloodpressureprovocation 112.75 22.44 −.26 .31 .04 .43^* .93^** –

7.Self-reportedangerbaseline 1.05 1.45 −.16 −.20 .00 .05 −.19 −.15 –

8.Self-reportedangerprovocation 4.21 2.81 .31 −.19 −.27 −.07 −.19 −.15 .42^* –

9.Angerbehavior 5.95 1.24 .51^** −.22 −.27 −.01 −.24 −.15 .19 .39^*

a1,male;2,female.

*Signiﬁcantat˛<.05(two-tailed).

**Signiﬁcantat˛<.01(two-tailed).

relativelycontrollableandslower,opentointrospection,andthus constitutedarelativelyreflectiveresponse.Insum,weobtained measuresofexperienceandbehavioralresponsesasindicesofthe reflectiveemotionalsystem.Whilethetwoautomaticresponses mayarguablyhavesomereflectivecharacteristics,andviceversa, theimportantpointisthatthetwoautomaticresponsesaremore automaticthanthereflectiveonesonan“automatic–reflective”

continuum.

2. Method 2.1. Participants

Forty-threeundergraduatestudents(23malesand20females)participatedin thestudy,forwhichtheyreceived$12.Datafromsevenparticipantswhodidnot becomeangrybecausetheybecameawareofthegoaloftheangerprovocation(as determinedduringdebriefing)werenotincludedintheanalyses.Thefinalsample consistedof36participants(19males,17females).Theirmeanagewas21.0years (SD=1.5).Sixty-onepercentoftheparticipantsidentifiedthemselvesasCaucasian American,14%asAsianAmerican,6%asAfricanAmerican,6%asLatino,and11%as

‘other’.

2.2. Procedure

Onarrivalatthelaboratory,participantsweretoldthattheywouldbeparticipat- inginastudyoncognitiveperformanceandmood.Afterphysiologicalsensorswere attached,participantswatchedaneutral5-minﬁlmwhilebaselineresponseswere collected,andthenreportedtheirfrustration,annoyance,andangerexperience (alongwith13distractorterms).

FollowingStemmlerandcolleagues(e.g.,2001),participantsthenperformed tediouscountingtasksdesignedtoinduceanger.Thesetasksrequiredthemto countbackwardsfor1mininstepsof7or13fromlargenumbers,e.g.18,652.The experimenterinterruptedtheparticipantsseveraltimesviaintercomwithscripted andpre-recordedremarksontheirperformanceandcooperation,deliveredinan increasinglyunfriendlyandimpatienttoneofvoice.Theywereblamedfor“produc- ingartifacts”by“movingtheirhand;”infact,suchmovementwashardlypossible.

Additionally,therewasanunexplaineddelayfor30safterwhichparticipantswere blamedforthefactthattheexperimenter“couldnotusethedatatheywerepro- ducinglikethis.”Attheendoftheangerprovocation,theexperimentersaid,“Let’s juststophere.Justﬁlloutthenextsectioninyourquestionnairepacket,”inanirri- tatedtonethatimpliedthatthesessionhadnotgonesmoothly(seeMauss,Evers, Wilhelm,&Gross,2006forasimilarinduction).

Aftertheangerprovocation,participantsreportedontheiranger,afterwhich thenext“cognitiveperformancetask”wasadministeredonalaptopcomputer.Par- ticipantshadtodecideasquicklyaspossiblewhetherword-likestimulipresented onthescreenwerewordsornon-words.Thislexical-decisiontaskwasdesignedto provideameasureofangeraccessibility(seeSection2.3formoredetail).

Aftercompletingthistask,participantsweretoldthattheDepartmentofPsy- chologywishedtoevaluateexperimenterstoensurethatallexperimentswere conductedinaprofessionalmanner.Participantswereassuredthatalltheiranswers wouldbecompletelyconﬁdentialandanonymousandwereinstructedtosealthe completedsurveysinenvelopesprovided.Onthesurvey-formitwasstatedthat thesurveywasmeanttogaininsightintothefunctioningofnewexperimenters andthattheirresponseswouldgivevaluableinsightsforevaluatingtheexperi- menterandhelpindeterminingwhetherthispersonwasﬁttingtoconductfurther experimentsinfuture.Infact,thissurveywasmeanttobeanassessmentofinstru- mentalangerbehavior,inthatitprovidedparticipantswithanopportunitytoexpress angerinaresentful/vengeful/consequentialwaytowardtheirobjectofanger(the

experimenter).Next,participantswereprobedaboutanysuspicionstheyhadabout theangerprovocationanddebriefed.

Oneweeklater,onaverage,participantsreturnedforasecondsessionand, togetherwithquestionnairesnotrelevanttothepresentstudy,completedthelexi- caldecisiontasktoobtainmeasurementsofangeraccessibilitywhentheywerenot angered.¹Finally,participantswerethanked,debriefed,andpaid.

2.3. Measures

2.3.1. Automaticmeasures

Inthelexicaldecisiontaskreactiontimesweremeasuredtoassesshowquickly participantsrecognizedangerwordsrelativetoneutralwords.Thetaskconsisted of30anger(e.g.,angry,furious)and30neutralwords(e.g.,margin,total),aswellas ﬁllerwords.Allwordswerematchedforlengthandfrequencyandwerepresented foramaximumof3s.A1300-msﬁxationpoint(XXX)precededeveryword.The taskwasprogrammedinPsyScope1.1andwaspresentedonanApplecomputer.

Thewordswerepresentedinfourﬁxedrandomorders.Halftheparticipantsused therighthandforwords(keyL)andthelefthandfornon-words(keyA);theother halfusedthereversedkey-handassignments.Fortheassessmentofangeracces- sibilityallwordswithanerrorratehigherthan20%(N=5)andalllatencieslower than300msandhigherthan3000mswereexcluded.Reactiontimesforincorrect responsesweresettomissing,andthedatawerethenlogtransformed(Fazio,1990).

Alllatenciesforangerandneutralwordswerecombinedintomeansforangerand neutralwords.Inlinewithpreviousresearch(e.g.,Loney,Frick,Clements,Ellis,&

Kerlin,2003),wecreateddifferencescoressubtractingeachparticipant’saverage responsetimetoneutralwordsfromhis/heraverageresponsetimetoangerwords.

Werefertoangeraccessibilityassessedaftertheprovocationastheprovocation measurement.Werefertoangeraccessibilityassessedoneweeklaterasthecon- trolmeasurement.Thiscontrolmeasurementwasdeemednecessaryinorderto showthattheangerprovocationresultedinhigherangeraccessibility(asreﬂected inshorterresponselatencies)comparedtothecontrolmeasurement.Fortheﬁnal indexofangeraccessibilitywesubtracteddifferencescoresobtainedatthecontrol measurementfromdifferencescoresobtainedattheprovocationmeasurement.

Physiologicalrespondingwasassessedduringbaselineandangerprovocation bymeasuringmeanarterialbloodpressure(MAP)thatwassampledat400Hzusing laboratorysoftware.Thiscardiovascularmeasurewaschosenbecauseitisakey physiologicalindicatorofanger(Stemmleretal.,2001).Customizedanalysissoft- ware(Wilhelm,Grossman,&Roth,1999)wasappliedofﬂineforphysiologicaldata reduction,artifactcontrol,andcomputationofaveragemeanarterialbloodpres- surescoresforeachparticipant.MAP(mmHg)wasobtainedfromthethirdﬁngerof thenon-dominanthandbymeansoftheFinapres^TM2300(Ohmeda,Madison,WI) system.Fromthissignal,beat-to-beatstrokevolumewasmeasuredusingWessel- ing’spulse-contouranalysismethod(BEATFAST,TNO-BiomedicalInstrumentation, Amsterdam).Averageswereobtainedacrossthebaselineandacrosstheanger provocationperiod.Foranalyzingbloodpressure,changescores(angerprovocation minusbaseline)wereusedtocontrolforindividualdifferencesatbaseline.

2.3.2. Reﬂectivemeasures

Participantsratedtheirexperienceofanger,frustration,andannoyanceon11- pointscales(seealsoMaussetal.,2006),rangingfrom0(noneatall)to10(extremely).

Thethreeitemswerecombinedintoasingleself-reportedangerscale(beforeversus

1Weobtainedthismeasureafter(ratherthanbefore)theprovocationinorder tominimizeinsightintothegoalofthisstudy.Thepossibilitythatparticipants wouldstillbeangryafteroneweekseemedtobeminimal.Toadditionallyrule outthispossibility,self-reportedangerexperiencewasassessedinthefollow-up session.Apairedt-testshowedthatparticipantsreportedlessangerinthefollow- upsession(M=1.62,SD=1.47)thanaftertheangerprovocation(M=4.07,SD=2.83), t(32)=5.05,p<.001.