ORIGINAL RESEARCH ARTICLEpublished: 07 January 2015doi: 10.3389/fpsyg.2014.01341Music induces universal emotion-relatedpsychophysiological responses: comparing Canadianlisteners to Congolese PygmiesHauke Egermann1,2*, Nathalie Fernando3, Lorraine Chuen1and Stephen McAdams11Centre for Interdisciplinary Research in Music Media and Technology, Schulich School of Music, McGill University, Montreal, QC, Canada2Audio Communication Group, Technische Universitt Berlin, Berlin, Germany3Laboratoire de Musicologie Compare et Anthropologie de la Musique, Facult de Musique, Universit de Montral, Montral, QC, CanadaEdited by:Petri Laukka, Stockholm University,SwedenReviewed by:Emery Schubert, University of NewSouth Wales, AustraliaFrank A. Russo, Ryerson University,Canada*Correspondence:Hauke Egermann, AudioCommunication Group, TechnischeUniversitt Berlin, Sekr. EN-8,Einsteinufer 17c, 10587 Berlin,Germanye-mail: hauke@egermann.netSubjectiveandpsychophysiological emotional responsestomusicfromtwodifferentcultureswerecomparedwithinthesetwocultures. Twoidentical experimentswereconducted: the rst in the Congolese rainforest with an isolated population of MebenzlPygmies without any exposure to Western music and culture, the second with a groupofWesternmusiclisteners, withnoexperiencewithCongolesemusic. FortyPygmiesand 40 Canadians listened in pairs to 19 music excerpts of 2999s in duration in randomorder (eight from the Pygmy population and 11 Western instrumental excerpts). For bothgroups, emotioncomponentswerecontinuouslymeasured: subjectivefeeling(usingatwo- dimensional valence and arousal rating interface), peripheral physiological activation,and facial expression. While Pygmy music was rated as positive and arousing by Pygmies,ratingsof WesternmusicbyWesternerscoveredtherangefromarousingtocalmingandfrompositivetonegative. Comparingpsychophysiological responsestoemotionalqualitiesof Pygmymusicacrossparticipant groupsshowednosimilarities. However,Western stimuli, rated as high and low arousing by Canadians, created similar responsesin both participant groups (with high arousal associated with increases in subjective andphysiological activation). Several low-level acoustical featuresof themusicpresented(tempo, pitch, andtimbre) wereshowntoaffect subjectiveandphysiological arousalsimilarly in both cultures. Results suggest that while the subjective dimension of emotionalvalence might be mediated by cultural learning, changes in arousal might involve a morebasic, universal response to low-level acoustical characteristics of music.Keywords: emotion, affect, music, psychophysiology, universal, cross-culturalINTRODUCTIONAlthoughmusicoccursinallhumancultures,itsstructureandfunction are highly varied (examples include the use of differentpitchscalesanddifferencesinceremonialandemotionaluses).Despitethesedifferences, therearesomebasicperceptual andstructural universals in music that are observed across cultures.These universals can inform theories about the evolutionary ori-gins of music, as they might indicate innate properties underlyingmusical behaviors (Eibl-Eibesfeldt, 1979). Examples of perceptualuniversalsincludepitchperception, octavegeneralization, cate-gorical perception of discrete scale pitches, melodic stream seg-regation,andperceptionofmelodiccontour(Harwood,1976).Previous research also indicates that the communication of emo-tionalexpressioninmusic,amorecomplexphenomenon,maytranscend cultural boundaries. The recognition of expressedemotion in music has been described as being based on under-lyingpsychoacousticcuesthatareemployedinsimilarfashionacross cultures to convey emotion (Balkwill and Thompson, 1999;Balkwill etal., 2004; Laukkaetal., 2013; SieversandPolansky,2013). A controlled, cross-cultural investigation testing for suchinnateemotional universals inmusicrequires thecomparisonof participant responses fromdifferent cultures to the samemusic. Moreover, it wouldrequirethat participants fromoneculturebecompletelynavetothemusicof theotherculture.Itbecomesincreasinglydifculttondculturesthatfulllthisprerequisite due to the homogenizing effects of globalization, asmodern digital music distribution networks have caused Westernmusicandculturetospreadrapidlyacrosstheglobelikeneverbefore (Huron, 2008). In a cross-cultural study, Fritz et al. (2009)reported that members of the Mafa tribe, who live in the north-ernparts of Cameroon(without electricity or any access toWesternmedialikeradioandTV), areabletorecognizeemo-tional expressions of happiness, sadness or fear above chance levelin Western music. In addition to expressingemotion, music hasoften been shown to have an impact on several response compo-nentsofinducedemotion(Greweetal.,2007;Salimpooretal.,2011; Egermann et al., 2013), and emotional responses have beenreported in Western cultures as one of the primary motivations toengage in musical activities (Schfer et al., 2013). In the presentstudy, we explored whether universal effects of music could alsowww.frontiersin.org January 2015 | Volume 5 | Article 1341 | 1Egermann et al. Music induces universal psychophysiological responsesbe found for psychophysiological measures of emotion inductionto extend existing ndings on universal emotional recognition.Theories of emotioninductionthroughmusicsuggest thatresponsesaregeneratedwithinaframeworkofseveral mecha-nisms that independently generate emotional responses to music(Juslin and Vstfjll, 2008; Juslin et al., 2010): cognitive appraisalof music andthe listening situation, visual imagery inducedthroughsound, evaluativeconditioningfrompairingmusictoanotheremotion-inducingstimulus, emotional episodicmem-ory associated with the music, violation of musical expectations,emotional contagion through emotional expressions in the music,entrainmentofbodilyrhythms torecurringperiodicitiesinthemusic, and brainstem reexes to low-level acoustic characteristicsofthemusic. Althoughsomeofthesemechanismsstillrequireempirical testing, thevalidityofothershasbeenconrmedbypreviousresearch. Forexample, Egermannetal. (2013)reportthatunexpectedmusical events, identiedthroughsimulationsof auditory statistical learning, induced reactions in the emotionalresponse components of physiological arousal and subjective feel-ing (Scherer, 2005). This was indicatedby increases inskinconductance, decreases in heart rate, and increases (decreases) insubjective arousal (valence) ratings. Induction mechanisms thatarebasedonmemoryarethought tobehighlyinuencedbyindividual andcultural learning(suchasevaluativecondition-ing, episodicmemory, ormusical expectancy). However, othermechanisms of emotion induction have been described as beingbased on culture-independent universal response patterns (Juslinand Vstfjll, 2008). Potential candidates for such universal mech-anisms include emotional contagion(see alsoEgermannandMcAdams, 2013), rhythmic entrainment, and brainstem reexes.Theuniversalityof brainstemreexes was partiallyshownbyFritzetal.(2009),whoreportedthatspectralmanipulationsofmusicrecordingsthatincreasedsensorydissonanceuniversallyled to decreased ratings of pleasantness in both Mafa and Westerngroups. However, the cultural independence of emotional conta-gionandrhythmicentrainmenthasyettobeproven. Allthreeculture-independent mechanisms are based on several low-levelacoustical features of music such as changes in loudness, tempo,timbre, andpitch, whichhavebeenshowntobesignicantlyrelatedtopsychophysiologicalindicatorsofemotionalresponsein Westernlisteners (Gomez and Danuser, 2007). Furthermore,previous research indicates that these effects might generalize toothercultures, as similaremotion-specicautonomicnervoussystem activity occurs in participants with different cultural back-grounds(Levenson, 1992;Breugelmansetal., 2005;Sotoetal.,2005).We ran two identical experiments within two cultural settings:one during a eld study of the Mbenzl Pygmies in the north-ern Congo and another one within a Western group of Canadianmusic listeners. The Pygmies live isolated in small hunter-gatherercommunities in the rainforest. Because of the lack of electricity,they do not have access to radio, television, or electricity, and thusare completely unfamiliar with Western music.We measured activity in the three response componentsof emotion(feeling, physiological activation, andexpression,Scherer, 2005) when presenting Mbenzel and Western music toboth cultural groups. After exploring their subjective emotionalresponses tothe music, we testedfor response similarity bycomparing physiological responses to Western and Pygmy stim-uli, which were rated with respect to their relative arousaland valence by the respective culturally familiar participantgroup. Furthermore, we tested whether responses across groupswouldbesimilarlyrelatedtolow-level acousticfeaturesof themusic, which could suggest the operation of culture-independentresponse mechanisms.METHODSPARTICIPANTSForty Mbenzl Pygmies (22 males, M = 35 years, SD = 14) and40 Canadians (22 males, M = 22 years, SD = 6) participated inthe two experiments. We recruited as many Mbenzl Pygmy par-ticipants as there were villagers available that were interested inparticipating. Subsequently, the number of Canadian participantswasmatchedtothat of Pygmyparticipants. ThePygmiesreg-ularlyengageinlocalmusicalactivityforceremonialpurposes,andeveryoneparticipatesactivelyinmusicmaking. Therefore,Canadian participants were recruited as amateur or professionalmusicianswhowerealsoregularlyengaginginmusicalactivity.No participants reported any hearing impairments. Both partic-ipantgroupsreceivednancialcompensationforparticipation.ItisimportanttonotethatthePygmieshavenocontactwithWesternmusicandminimal contactwithpopularmusicfromZare when traveling to nearby towns to trade. They do have alsocontact with the music of the Bantu people in the region. Theirmusic is similar in some ways but the Pygmies make a point ofhonor in distinguishing their music from Bantu music. In partic-ular, the Bantus do not produce the complex polyphonies uniqueto the Pygmy music in all of Africa.MUSICAL STIMULIThestimulusmaterialsconsistedof 19musical excerptstakenfromboth Western and Congolese Pygmy repertoires plus one testWestern excerpt (see Table S1). Eight Western orchestral musicalexcerptsweretakenfromapreviousstudythatgroupedpiecesaccordingtosimilaremotional meaning(Bigandet al., 2005).Weselectedtwostimuli peremotionquadrant fromthetwo-dimensional emotionmodel (Russell, 1980), whichrepresentsemotionalongthedimensionsofvalence(negativetopositive)andarousal(calmtoexcited).Additionally,threeexcerptswereexperimenter-selectedfrompopular Westernlms toexpressthreebasicemotions: happiness, fear, andsadness. ThePygmymusic was recorded in the eld. It is well known for its polyphoniccomplexity, oftensung onpentatonic scales (Shebasta, 1952;Turnbull, 1961; Arom, 1991). APygmymusicexpert selectedeight typical vocal pieces that are usually performed in differentceremonial contexts that often had emotional connotations, suchas being sung in order to calm down anger or fear or to alleviatebereavement. Stimulus duration ranged from 30 s to 2 min, andall were presented in random order.MEASUREMENTSAn iPod Touch (Apple Inc., Cupertino, CA) was used to contin-uously rate subjective feeling during stimulus presentation on aFrontiers in Psychology | Emotion Science January 2015 | Volume 5 | Article 1341 | 2Egermann et al. Music induces universal psychophysiological responsestwo-dimensional emotion space. The horizontal axis was a rep-resentation of valence, the relative negative or positive nature ofanemotion,andtheverticalaxiswasarepresentationofemo-tional arousal, aspectrumfromcalmtoexcited. Theheuristicvalue of the two-dimensional emotion space was previously con-rmed in numerous other studies measuring emotion in music(Schubert, 1999; Nagel et al., 2007; Egermann et al., 2009, 2013).Participantswereinstructedtostartratingforeveryexcerptinthecenteroftheemotionspaceandthentomovetheirngeraccordingtotheircurrentemotionalstate.Todenotetheemo-tionalqualitiesoftheemotionspace,illustratedfacesdepictinganegative(upside-downsmile)andpositive(smile)facewereplaced on the extreme ends of the horizontal axis outside the iPod.Aroused (open eyes) and calm (closed eyes) faces were placed onthe extreme ends of the vertical axis. Participants were instructedto rate their own felt emotions caused by the music, and not torate what they thought they recognized as an emotional expres-sion in the music. At the beginning of the eld trip in the Congo,all Pygmies who were interested in participating were invited toanextensivecollectivepresentationdescribingtheexperiments,andinstructionsweregiven. Additionally, all instructionswererepeatedright beforeeachexperiment. Physiological measure-ments were recorded through ProComp Inniti units (ThoughtTechnology Ltd, Montreal, QC). Respiration was measured usinga respiration belt attached around the chest. Blood volume pulse(BVP)wasmeasuredusingaphotoplethysmographonthepal-mar side of the distal phalange of the middle nger of thenon-dominant hand. Electrodermal activity was measured usingelectrodesonthedistalphalangesoftheindexandringngersofthenon-dominanthand. Expressivemuscleactivationsweremeasured using two electromyography electrodes (MyoScan-Prosurfacesensors)placedonthecorrugatorsupercilii (frowning)and zygomaticus major (smiling) muscles (Cacioppo et al., 1986).PROCEDUREThe research reported in this manuscript was carried out accord-ing to six the principles expressed in the Declaration of Helsinkiand the Research Ethics Board of McGill University has reviewedand approved this study (certicate #156-0107).In order to create identical testing settings, the procedure waskept constant between experiments with both subpopulations. ToensurecomfortforthePygmiesintheunfamiliartestingsitua-tion, all participants had to be tested in pairs of close friends orfamilymembers. Thetwoparticipantswereseatedonoppositesides of a table facing each other, with the experimenter on thesideinbetweenthem. ExperimentsinCanadawereconductedinEnglish. For the Congolese testing session, aninterpreterwasseatedoppositetheexperimenter(translatingbetweentheMbenzl language of the Pygmies and French, the rst languageof the experimenter). There was a small barrier between partic-ipantspreventingthemfromseeingeachothersratings. Afterbiosensorswereattached, baselinephysiological activityduring2 min of silence was recorded, followed by the practice trial. Thestimuli were presented through a pair of MDR-NC7 headphones(Sony Corporation, Tokyo, Japan) at a comfortable listening levelfor one third of participants of both groups. For the rest of theparticipants, stimuli werepresentedthroughbuilt-inMacbookPro speakers at maximal volume, because of technical difcultiesencountered during testing in the eld in the Congo rainforest,whereapairofheadphonesbrokeafterthefourteenthpartici-pant. After every stimulus, participants in both experiments wereasked, if they knew the music previously presented to them. Here,allCanadianparticipantsrespondedthattheydidnotknoworhad not heard the Pygmy excerpts before, whereas all Pygmy par-ticipants indicated that they did not know or had not heard theWestern excerpts before.DATA ANALYSISAfterremovalofvisuallyerroneousdatasetsduetomovement-relatedsensor displacement (per excerpt andparticipant), weextracted averaged response scores per music excerpt, signalandparticipant.First,BVP(lowpass2 Hz),respirationactivity(low pass 1 Hz), and skin conductance (low pass 0.3 Hz) signalswerelteredusingMatlab(Version8.14.0.604, TheMathworksInc., Natick, MA)inordertoremoveextraneous informationusingalinear-phaselterbasedontheconvolutionof a4th-orderButterworthlterimpulseresponse(alsoconvolvedwithitself inreverse time toavoidphase shifting). The MyoScan-Pro EMG sensors automatically converted their signal to a rootmeansquare(RMS) signal (after aninternal analogrectica-tion), whichwas thereforenot lteredanyfurther (capturingEMGactivityatfrequenciesupto500 Hz). Asthereweresev-eral errorsintheEMGcorrugatorsupercilii recordingsintheCongolese sessions, we are not presenting any results from thesemeasurements. We performed linear detrending on the skin con-ductance signal inorder toremove any negative trends overtimewithbreakpointsevery60 s. [Thesetrendsarecausedbyanaccumulationof charge over time betweenthe skinandsensor (Salimpoor et al., 2009)]. Skinconductance responseevents were subsequently computationally identied at onsets ofskinconductanceincreasesofanysize. Inordertoremoveanybetweenparticipant differencesinphysiological baselineactiv-ity, we subtracted the corresponding baseline response score foreveryparticipant fromall of her/hisstimulusresponsescores.Furthermore, wecalculatedthemeanarousal andvalencerat-ing per participant and excerpt excluding a 7-s orientation phaseafter stimulus onset, whichwas identiedafter visual inspec-tionofcontinuousratingdata(Bachoriketal.,2009;Schubert,2012).This procedure resultedinsevendifferent meanresponsescoresperparticipantandmusicexcerpt: heartrate, skincon-ductance level (SCL), number of skinconductance responsesper minute(SCR), respirationrate, EMGzygomaticus activa-tion, and arousal and valence ratings. However, before using theseresulting response score vectors in subsequent analyses, they werez-standardized across excerpts and participants.For testing of signicance on response scores, we employed ahierarchicallinearmodelingapproach(Westetal.,2006)usingthe MIXEDprocedure inSPSS Statistics (IBM, Version21).Estimationof parameters was basedonrestrictedmaximumlikelihood. Beside xed effects coefcients, the models includedan intercept and a rst-order autoregressive residual covari-ance structure modeling carryover effects from previous stimulustrials (AR1).www.frontiersin.org January 2015 | Volume 5 | Article 1341 | 3Egermann et al. Music induces universal psychophysiological responsesRESULTSEFFECTS OF MUSICAL EXCERPTS ON SUBJECTIVE FEELINGFigure 1 presents the mean arousal and valence rating per excerpt,separatedbystimuluscultural originandparticipantgroupinthetwo-dimensional emotionspace. TheWesternmusicratedby Canadian participants covered all four quadrants: pieces wererated with both high and lowsubjective arousal, and someinduced a positive affect whereas others induced a negative affect.In contrast, Pygmies rated most Pygmy music excerpts on aver-age as both positive and arousing (except excerpt 14). However,within this quadrant, pieces can still be categorized relatively asmoreorlessarousingthanothers,andasmoreorlesspositivethan others.EFFECTS OF WESTERN MUSICAL EXCERPTSWe subsequently tested whether subjective ratings of the Westernstimuli by the Canadian participant group were similarly relatedtothesubjectiveandphysiologicalresponsesignalsrecordedinbothgroups. Therefore, we estimatedtwohierarchical linearmodelsforeveryresponsescore, oneforeachoftheCanadianand Pygmy participant groups (Equation 1):response score = b0+b1A +b2V +b3A V (1)As xed effects (b0 to b3), we added an intercept [Int], the Westerngroups mean arousal [A] and valence [V] ratings (per excerpt),and an interaction between the two, modeling an additional effectof the co-occurrence of positive and arousing feelings [A V].Subsequently, we also tested if the effect coefcients estimated forthe two participant groups signicantly differed from each otherthrough estimating a third model with interaction effects betweenthexedeffects[A,V,A V]ontheonesideandbeinginthePygmy participant group (dummy variable with being Canadianas areferencegroup) ontheother. Inthesemodels, wealsoaddedbeinginthePygmyparticipant groupasaxedeffect,modelingsignicant differencesbetweenresponsescoregroupmeans that were independent of any music excerpt. Figure 2 visu-allypresentstheestimatedxed-effectscoefcientsaserror-bargraphs (see Table S2 for statistical details). The effects displayedFIGURE 1 | Mean arousal and valence ratings separated by musicexcerpt and participant group. High valence was dened as positive, lowvalence as negative.arethetwoIntercepts for bothgroups [Int], followedbythearousal [A], valence [V], and interaction [A V] effects. A signif-icant effect (greater or smaller than zero) is indicated by the factthat the condence interval does not overlap with the zero line.Signicant differences of xed effects between participant groupsare indicated with asterisks.Subjective Arousal Scores: In general, Canadians rated theWestern music as more arousing than the Pygmy group, indicatedby a signicant difference between group means [Int]. However,Canadian arousal ratings [A] are a signicant predictor of Pygmyarousal ratings: for excerpts that Canadians rate as more arous-ingthanotherexcerpts, Pygmiesalsoratedthemasarousing.Noothereffectsaresignicant. SubjectiveValenceScores: OnlyFIGURE 2 | Error bar graphs of xed effect coefcients estimated forWestern music excerpts arousal and valence ratings separated byparticipant group and response score types. CI, condence interval;[Int], Intercept; [A], Mean arousal rating of excerpt by Canadian participantgroup; [V], Mean valence rating of excerpt by Canadian participant group; [V A], Interaction effect between [A] and [V]. The solid line represents effectestimates for Pygmies, the dashed line represents effects for Canadians.Asterisks indicate signicance of xed effect differences between Pygmyand Canadian groups: p< 0.05, p< 0.01 (two sided t-tests).Frontiers in Psychology | Emotion Science January 2015 | Volume 5 | Article 1341 | 4Egermann et al. Music induces universal psychophysiological responsesintercepts [Int] in both groups are signicantly lower than zero.HeartRateScores: MusicratedasmorearousingbyCanadianselicitedincreasedheart ratesinbothgroups[A], althoughforPygmiesthiswasonlyanon-signicanttrend(p < 0.10). SCLScores:BothgroupsrespondedsimilarlytoincreasesinArousal[A]: higher arousal ratingsby Canadiansfor a givenpiecewererelated to higher skin conductance levels in both groups. Whenthe Westernmusic was ratedas positive andarousingat thesametime[A V], bothgroupssurprisinglyrespondedwithdecreased skin conductance. SCR Scores: Only Canadian partic-ipants responded with an increased number of skin conductanceresponses when they rated the music as arousing [A]. RespirationScores: Bothgroupsrespondedwithincreasedrespirationrateswhenthe Westernmusic was arousing comparedtocalming[A]. EMGZygomaticusScores: OnlyCanadiansrespondedwithincreasedzygomaticusactivationwhentheWesternmusicwasarousing [A] and positive [V].Summarizedtogether, these results indicate that there wasa positive correlation between subjective arousal ratings ofCanadiansandPygmiesthatwasaccompaniedinbothgroupsbyincreasesinphysiological arousal (heart rate, skinconduc-tance, and respiration rate). There were no similarities betweenthe two participant groups in responses to stimuli that were ratedas having different degrees of valence.EFFECTS OF PYGMY MUSIC EXCERPTSWe subsequently tested for effects of Pygmy music excerpts usingmodel congurations that were identical to the Western excerptanalyses, except that now Pygmies subjective excerpt ratings wereused as xed effects (Figure 3 and Table S3).Subjective Arousal Scores: In general, Pygmies rated their ownmusicassignicantlymorearousingthandidCanadians[Int].Arousing Pygmy excerpts (vs. less arousing excerpts as ratedbyPygmies, [A])werenegativelyrelatedtoarousal ratings inCanadians, as were excerpts that were rated as arousing and pos-itive [A V]. Subjective Valence Scores: No effect was signicantin the valence model, except the two intercepts [Int], which werebothpositive. HeartRateScores: OnlyPygmiesrespondedwithdecreasedheart rateswhentheylistenedtoexcerptsthat werearousing and positive to them [A V]. SCL Scores: Only Pygmiesresponded with increased skin conductance when they rated someexcerpts as more arousing and positive than other excerpts [A, V,A V]. SCR Scores: No signicant effect could be observed here.RespirationScores:Canadiansrespondedwithincreasedbreath-ingrates tomusicthePygmiesratedas morearousing. EMGZygomaticusScores: AlthoughPygmiesdidntshowanysigni-cant effects here, Canadian participants responded with increasedactivationformusicratedbyPygmiesasarousingandpositive[A, V, A V].Summarized together, the responses to Pygmy excerpts indi-catethat participant groups respondedverydifferentlytothemusic. Changes in arousal and valence [A, V] were only associ-ated with changes in heart rate and skin conductance for Pygmies.However, Canadians responded with increases in respiration rateto increases in excerpt arousal [A]. Interestingly, Canadians alsoresponded with increases in zygomaticus activity to increases inexcerpt arousal [A] and valence [V, A V]. Thus, there were littleFIGURE 3 | Error bar graphs of xed effect coefcients estimated forPygmy music excerpts arousal and valence ratings separated byparticipant group and response score types. CI, condence interval;[Int], Intercept; [A], Mean arousal rating of excerpt by Pygmy participantgroup; [V], Mean valence rating of excerpt by Pygmy participant group; [V A], Interaction effect between [A] and [V]. The solid line represents effectestimates for Pygmies, the dashed line represents effects for Canadians.Asterisks indicate signicance of xed effect differences between Pygmyand Canadian groups: p< 0.05, p< 0.01 (two sided t-tests).similaritiesbetweenthetwoparticipantgroupsinresponsestothe emotional qualities of Pygmy music.ACOUSTICAL DESCRIPTOR EFFECTS OF MUSIC EXCERPTSGiventhatweobservedsomesimilaritiesbetweenthetwopar-ticipantgroupsinrespondingtothemusical excerpts, wesub-sequentlytestedwhether thesesimilarities couldbeexplainedbysimilarresponsestounderlyinglow-level acoustical featuresofthestimuli. Weextractedsixacoustical descriptorsaveragedacross each excerpt using the MIR Toolbox v1.4.1 (Lartillot et al.,2008):ThisincludedroughnessbasedontheSethares-method,the centroid of the spectral distribution, mode as the relative tbetween a major or minor chromagram (which may be difcultwww.frontiersin.org January 2015 | Volume 5 | Article 1341 | 5Egermann et al. Music induces universal psychophysiological responsesTable 1 | Component Loadings from Principal Component Analyses ofAcoustical Descriptors of Music Excerpts (n = 19).Principal ComponentAcoustical Descriptor11 2 3 4 5 6RMS Energy10.97 0.05 0.03 0.06 0.07 0.19Roughness10.91 0.02 0.08 0.02 0.34 0.17Spectral Centroid10.07 0.94 0.05 0.15 0.21 0.19Mode10.03 0.04 0.99 0.16 0.02 0.00Pitch10.03 0.14 0.16 0.97 0.06 0.01Event Density10.31 0.25 0.03 0.08 0.90 0.16Beats Per Minute 0.30 0.21 0.00 0.01 0.16 0.92Rotation Method: Varimax with Kaiser Normalization.1Mean across excerpt.to extract from the Pygmy music), dominant pitch based on anautocorrelation function of the audio waveform, and event den-sity as the number of note onsets per seconds. Tempo (in beatsper minute, BPM) was measured by tapping the dominant beatofttheexcerptintoanweb-basedBPM-Tracker.Alldescriptorswere subsequently entered into a Principle Component Analysis(Table 1).Theresultingsixprinciplecomponents[PC16]werethenusedasindependent predictorsinsubsequent linearmodelinganalyses. We also added an additional dummy variable that codedwhether the excerpt was from the Pygmy music repertoire or not(with Western music as reference group) [PM]. This was done toaccountforanyotheruncontrolledmusicfeaturesthatdifferedbetween the Western and Pygmy music. Similar to previous anal-yses, we also tested for signicant group differences in responsesto acoustical principal components and PM by estimating a thirdmodel for every response score with interaction effects betweenxed effects and being in the Pymgy group. Figure 4 presents theestimated xed-effects coefcients (separated by response scores),signicant group differences are indicated through asterisks (seeTable S4 for statistical details).Subjective Arousal Scores: The Canadian group responded withincreased arousal ratings to increases in all PCs, except for mode[PC 3] and PM. The Pygmy group respondedonly to increasesintempo[PC6]. AlthoughCanadianparticipants weremoreresponsive totempo(indicatedby the signicantly increasedeffect size), for bothparticipant groups a positive effect wasestimated. ThiscouldexplainwhyPygmiesreportedincreasedarousal toWesternmusicthat Canadians alsoratedas arous-ing. SubjectiveValenceScores: AlthoughCanadianparticipantsrespondedtoincreasesinroughness/RMSenergy, spectralcen-troid, andpitch[PCs1, 2, and4]withdecreasedvalenceandtoincreases inevent density [PC5] withincreasedvalence,Pygmies did not respond to any acoustical descriptor. They con-sistently rated their own music as signicantly more positive thandid Canadians. Heart Rate Scores: Only Canadians responded toincreasedtempo[PC6]withincreasedheart rate. SCLScores:Bothgroupsrespondedwithincreasedskinconductancewhenspectral centroid, pitch, and tempo increased [PC 2, 4, 6]. OnlyPygmiesrespondedwithincreasedSCLwhenRMSenergyandFIGURE 4 | Error bar graphs of xed effect coefcients estimated forsignicant effects of the Acoustical PCs, separated by participantgroup and response score types. The solid line represents effectestimates for Pygmies, the dashed line represents effects for Canadians.Asterisks indicate signicance of xed effect differences between Pygmyand Canadian groups: p< 0.05, p< 0.01 (two sided t-tests).roughness[PC1]increased, andonlyCanadiansrespondedtoincreasesinmode (SCL decrease),eventdensity (SCL increase)and the Pygmy music (SCL decrease) [PCs 3, 5, PM]. SCR Scores:Only Canadians responded to increased roughness/RMS energy,spectral centroid, and event density [PCs 1, 2, 5] with increases inSCR. Both groups responded with decreased SCR during listen-ing to the Pygmy music. Respiration Scores: Canadians respondedwithincreasedrespirationrates whenroughness/RMSenergy,spectral centroid, andevent density [PCs 1, 2, 5] increased.Notethat therearenosignicant differences for anyacousti-cal PC effect coefcients between the groups for SCL, SCR, andRespiration Scores. The shape of the error graph line is very sim-ilarinbothgroups. However, someeffectcoefcientsareonlysignicant in the Canadian participant group indicating that forFrontiers in Psychology | Emotion Science January 2015 | Volume 5 | Article 1341 | 6Egermann et al. Music induces universal psychophysiological responsesPygmies estimated effects were too small to become signicant,potentiallyduetogeneralincreasedphysiologicalmeasurementerrorintherainforest eldsetting. EMGZygomaticus Scores:Only Canadians responded with increased zygomaticus activationto increases spectral centroid, event density, and tempo [PC 2, 5,6] and only Canadians responded with decreased activation whenmode [PC 3] increased (more major) or when listening to Pygmymusic [PM].DISCUSSIONWe tested for effects of group membership (i.e., cultural familiar-ity), musicexcerptqualities(indexedbysubjectivearousal andvalenceratings of excerpts fromwithincultures), andunder-lyinglow-level acoustical descriptorsonemotional responsetomusic. The results demonstrate both similarities and differencesin participants responses to different music excerpts, which arelikely mediated through different response mechanisms.The following differences between both groups occurred:Except for the effects of Pygmy music in general, effects in mod-els estimated for SCR scores were never signicant in the Pygmyparticipant group, indicating that this score did not reect theirsympathetic arousal very well. Furthermore, although there weremany signicant effects on zygomaticus recordings in Canadians,inPygmiestheywerenotsignicantlyinuencedbyanyinde-pendentvariable, indicatinga)thattheydidnotrespondwithexpressive facial reactions in general (maybe because of increasedseriousness due to the unfamiliar experimental setting) or b) thatalsoforthisfacialmuscle,measurementerrorsoccurredintheeld setting.CanadiansratedWesternmusicasmoresubjectivelyarous-ingthandidPygmies, whereas Pygmies ratedtheir music asmorearousingthanCanadians(alsoillustratedbycorrespond-ingeffectsinSCLscores). Thiscouldreectageneral increaseinattentionalfocusformusic thatisculturallymeaningfulandfamiliar. Subjectivevalenceratingswerealsodifferent betweengroups (e.g., Pygmies rated their own music as more positive thanCanadiansanddidnotrespondwithchangesinvalencetoanyacoustical descriptor of the music). Only Pygmies responded withincreasesinphysiological arousal (SCL)whentheyratedtheirownmusicasmorearousingandpositive. Thesendingssug-gest that such response differences were mediated by mechanismsbased on cultural learning (cognitive appraisal, episodic memory,or musical expectancy).There were several similarities in both groups responses to thedifferent musical stimuli. When Canadians rated Western musicassubjectivelyarousing,bothgroupsrespondedwithincreasedphysiological arousal (heart rate, skinconductance, andres-piration). Additionally, inPygmies, subjective arousal ratingsincreased for arousing Western music. Taken together, these sim-ilarities observedinemotional respondingcouldbeexplainedby universal reaction patterns to several low-level features. Bothgroups respondedwithincreasedsubjective andphysiologicalarousal (SCL) when the music had a higher tempo. Furthermore,increases in spectral centroid and pitch lead to increased skin con-ductance for both participant groups. Although respiration mighthave inuenced skin conductance (Rittweger et al., 1997), theserelationships might reect increases in sympathetic arousal thatcould be caused by the brainstemreex, rhythmic entrainment, oremotional contagion mechanisms, which are thought to be ratherindependent from cultural learning (Juslin and Vstfjll, 2008).Increases in tempo may have led to synchronization of internalbodyrhythms(Juslinetal., 2010),leadingtoincreasedarousal(rhythmicentrainment). However, temporepresents afeaturethat hasbeenshowntoco-varywithemotional expressioninmusic (Juslin and Laukka, 2003), which might lead to emotionalcontagion, a mechanism that is also thought to be weakly inu-enced by cultural learning. Thus, the universal responses to tempocould also be explained by internal mimicking of the emotionalexpressions heard in the music, as emotion recognition has beenpreviously been shown to be based to some degree on universalfeatures (Fritz et al., 2009; Laukka et al., 2013).Increasesinpitchandspectralcentroid(timbralbrightness)arealsoassociatedwithemotionexpressionandcouldthusbeassociated with arousal because of emotional contagion. However,they might also have a direct inuence on arousal because of thebrainstemreexesthatreacttourgentandimportanteventsasdescribed by Juslin and Vstfjll (2008). Excerpt 17 in particular,from the soundtrack of the lm Psycho, features unusually highand bright violin sounds that could be inuential here.One might add that a brainstem-reex-mediated arousalresponse could also inuence emotion recognition, as some the-ories of emotion recognition suggest that it is based on the self-perception of simulated emotional resonance (Cochrane, 2010).Tosummarize, bothroutes of emotioninductionpreviouslydescribedremainplausible: (a)brainstemreexandrhythmicentrainmentcreatephysiologicalarousalthatisthenintegratedintoaconsciousfeelingor(b)expressionsareinternallymim-ickedandleadtoinducedemotionwithassociatedresponses(emotional contagion).Contrary to Fritz et al.s (2009) ndings of universal reactionpatterns inpleasantness/valence tochanges insensory disso-nance/roughness, the universal response patterns observed in thisstudywere mostlyrelatedtoemotional arousal. This mayofcourse be due to less extreme variability of roughness in our nat-ural stimuli thatwerenotarticiallymanipulated. However, itmight also be that the low-level emotional processing observed inthis study is only inuential in creating underlying sympatheti-cally mediated arousal responses that still require conscious inter-pretation into emotional qualities (Schachter and Singer, 1962).Physiological measures of activation indicate much fewer differ-ences between the participant groups. In only one linear model,the interaction term testing for group differences was signicantlydifferent from zero (increases in SCR due to increases in Pygmymusic valence rating). This suggests that subjective emotional rat-ingsmighthavebeenmoresubjecttocultural inuencesthanphysiological responses to the stimuli, a nding that is similar tothat of Soto et al. (2005): Chinese American participants reportedless extreme emotions thandidMexicanAmericans, whereasemotional behavior and physiology were less differentiated.The ndings from this study are subject to limitations. Firstly,in order to test for effects of natural stimuli, several of their acous-tical descriptors co-vary and their relative effects can only be sep-arated in a limited way. However, even being able to differentiatebetweenacoustical descriptorswouldnot allowustoseparatewww.frontiersin.org January 2015 | Volume 5 | Article 1341 | 7Egermann et al. Music induces universal psychophysiological responsespsychological emotion-inductionmechanismsfromeachother(JuslinandVstfjll, 2008), as many acoustical features con-tribute as inputs to several mechanisms at the same time. Thus,many conclusions onthe operationof response mechanismshavetoremainspeculative. Differencesbetweenmeasurementsfrom the two groups might stem from different responding, butalsofromother co-varyingfeatures of the twomeasurementsituations. Eventhoughwetriedtokeepasmanyvariablesasconstant as possible, Pygmies were still very unfamiliar with theexperimental procedure, astheywerenotusedtousingratinginterfaces or listening to recorded music. Additionally, emotionalself-reports might be biased by a different understanding of emo-tionalqualities(MatsumotoandYooHee, 2007). However, theexperimenterassuredthroughseveral lengthyinterviewsthatasimilar understanding of basic emotional qualities was present inthePygmypopulation. WhenaskediftheyhadeverheardtheWesternmusic,theyallansweredwithacategoricalno.Theyalways expressed appreciation for the Western music (Its goodmusic!) but when asked what it expresses, they answered I dontknow, its your music, you should know.The present ndings indicate that although valence responseswere often different between the two participant groups, music-inducedarousalresponsesappearedtobebasedonratheruni-versal, culturally independent response mechanisms. These maybe basedonlow-level acoustical characteristics of music liketempo, pitch, or timbre. There were more similarities in partic-ipants responsesinWesternthaninPygmymusic, indicatingthatWesternmusicmaymakestrongeruseofthesecharacter-istics than Pygmy music. Five out of the six acoustical principalcomponents extracted from stimuli showed a higher variance inthe selected Western music compared to the Pygmy music. It ispossiblethat thecommunicationof emotioninPygmymusicis insteadbasedmore onsymbolic andassociative meaning,mediated through learning. Summarized together, these ndingsmay help to understand the ubiquitously experienced emotionalresponses tomusicthat canbesometimes verypersonal andindividual, but at other times to some extent also universal andcollective.ACKNOWLEDGMENTSHaukeEgermannandStephenMcAdamss workwas partiallyfunded by the Canadian Social Sciences and Humanities ResearchCouncil through a grant to Stephen McAdams (#410-2009-2201), as well as StephenMcAdamss CanadaResearchChair.Author contributions: Hauke Egermann, Stephen McAdams, andNathalie Fernando designed experiments, Nathalie Fernando andLorraine Chuen conducted experiments, Hauke Egermann ana-lyzeddata, HaukeEgermann, StephenMcAdams, andLorraineChuen wrote the paper.SUPPLEMENTARY MATERIALThe Supplementary Material for this article can be foundonline at: http://www.frontiersin.org/journal/10.3389/fpsyg.2014.01341/abstract.REFERENCESArom, S. (1991). AfricanPolyphony andPolyrhythm: Musical Structure andMethodology. Cambridge, MA: Cambridge University Press.Bachorik, J. P., Bangert, M., Loui, P., Larke, K., Berger, J., Rowe, R., et al. (2009).Emotion in motion: investigating the time-course of emotional judgments ofmusical stimuli. Music Percept. 26, 355364. doi: 10.1525/mp.2009.26.4.355Balkwill, L.-L., and Thompson, W. F. (1999). A cross-cultural investigation of theperception of emotion in music: psychological and cultural cues. Music Percept.17, 4364. doi: 10.2307/40285811Balkwill, L.-L., Thompson, W. F., and Matsunaga, R. (2004). Recognition of emo-tioninJapanese, Western, andHindustani musicbyJapaneselisteners. Jpn.Psychol. Res. 46, 337349. doi: 10.1111/j.1468-5584.2004.00265.xBigand, E., Vieillard, S., Madurell, F., Marozeau, J., andDacquet, A. (2005).Multidimensional scaling of emotional responses to music: the effect of musicalexpertise and of the duration of the excerpts. Cogn. Emot. 19, 11131139. doi:10.1080/02699930500204250Breugelmans, S. M., Poortinga, Y. H., Ambadar, Z., Setiadi, B., Vaca, J. B.,Widiyanto, P., et al. (2005). Body sensations associatedwithemotions inRarmuri Indians, rural Javanese, and three student samples. Emotion 5,166174. doi: 10.1037/1528-3542.5.2.166Cacioppo, J. T., Petty, R. E., Losch, M. E., and Kim, H. S. (1986). Electromyographicactivityoverfacial muscleregionscandifferentiatethevalenceandintensityof affectivereactions. J. Pers. Soc. Psychol. 50, 260268. doi: 10.1037/0022-3514.50.2.260Cochrane, T. (2010). Music, emotions and the inuence of the cognitive sciences.Philos. Compass 11, 978988. doi: 10.1111/j.1747-9991.2010.00337.xEgermann, H., and McAdams, S. (2013). Empathy and emotional contagion as alink between recognized and felt emotions in music listening. Music Percept. 31,137153. doi: 10.1525/mp.2013.31.2.139Egermann, H., Nagel, F., Altenmller, E., andKopiez, R. (2009). ContinuousMeasurement of musically-induced emotion: a web experiment. Int. J. InternetSci. 4, 420.Egermann, H., Pearce, M. T., Wiggins, G. A., and McAdams, S. (2013).Probabilistic models of expectation violation predict psychophysiological emo-tional responses to live concert music. Cogn. Affect. Behav. Neurosci. 13,533553. doi: 10.3758/s13415-013-0161-yEibl-Eibesfeldt, I. (1979). Humanethology: concepts andimplications for thesciences of man. Behav. Brain Sci. 2, 157. doi: 10.1017/S0140525X00060416Fritz, T., Jentschke, S., Gosselin, N., Sammler, D., Peretz, I., Turner, R., et al. (2009).Universal recognition of three basic emotions in music. Curr. Biol. 19, 573576.doi: 10.1016/j.cub.2009.02.058Gomez, P., andDanuser, B. (2007). Relationships betweenmusical structureand psychophysiological measures of emotion. Emotion 7, 377387. doi:10.1037/1528-3542.7.2.377Grewe, O., Nagel, F., Kopiez, R., and Altenmller, E. (2007). Emotions over time:synchronicity and development of subjective, physiological, and facial affectivereactions to music. Emotion 7, 774788. doi: 10.1037/1528-3542.7.4.774Harwood, D. (1976). Universals in music: a perspective from cognitive psychology.Ethnomusicology 20, 521533. doi: 10.2307/851047Huron, D. (2008). Lost in music. Nature 453, 456457. doi: 10.1038/453456aJuslin, P. N., and Laukka, P. (2003). Communication of emotions in vocal expres-sion and music performance: different channels, same code? Psychol. Bull. 129,770814. doi: 10.1037/0033-2909.129.5.770Juslin, P. N., Liljestrm, S., Vstfjll, D., and Lundqvist, L.-O. (2010). How doesmusicevokeemotions?Exploringtheunderlyingmechanisms,inHandbookof Music and Emotion: Theory, Research, Applications, eds P. N. Juslin and J. A.Sloboda (Oxford: Oxford University Press), 605643.Juslin, P. N., and Vstfjll, D. (2008). Emotional responses to music: the need toconsider underlying mechanisms. Behav. Brain Sci. 31, 55975; discussion 575621. doi: 10.1017/S0140525X08005293Lartillot, O., Toiviainen, P., andEerola, T. (2008). Amatlabtoolboxformusicinformationretrieval,inDataAnalysis, MachineLearningandApplications,Studies inClassication, DataAnalysis, andKnowledgeOrganization, eds C.Preisach, H. Burkhardt, L. Schmidt-Thieme, and R. Decker (Springer-Verlag).Laukka, P., Eerola, T., Thingujam, N. S., Yamasaki, T., and Beller, G.(2013). Universal andculture-specic factors inthe recognitionandper-formanceof musical affect expressions. Emotion13, 434449. doi: 10.1037/a0031388Levenson, R. W. (1992). Autonomic nervous system differences among emotions.Psychol. Sci. 3, 2327. doi: 10.1111/j.1467-9280.1992.tb00251.xMatsumoto, D., andYooHee, S. (2007). Methodological considerationsinthestudyof emotionacross cultures,inHandbookof EmotionElicitationandFrontiers in Psychology | Emotion Science January 2015 | Volume 5 | Article 1341 | 8Egermann et al. Music induces universal psychophysiological responsesAssessment, edsJ. A. CoanandJ. J. Allen(Oxford: OxfordUniversityPress),332348.Nagel, F., Kopiez, R., Grewe, O., and Altenmller, E. (2007). EMuJoy: software forcontinuous measurement of perceived emotions in music. Behav. Res. Methods39, 283290. doi: 10.3758/BF03193159Rittweger, J., Lambertz, M., andLanghorst, P. (1997). Inuencesof mandatorybreathingonrhythmical componentsofelectrodermal activity. Clin. Physiol.17, 609618. doi: 10.1046/j.1365-2281.1997.00058.xRussell, J. A. (1980). Acircumplex model of affect. J. Pers. Soc. Psychol. 39,11611178. doi: 10.1037/h0077714Salimpoor, V. N., Benovoy, M., Larcher, K., Dagher, A., and Zatorre, R. J.(2011). Anatomically distinct dopamine release during anticipation and expe-rienceof peakemotiontomusic. Nat. Neurosci. 14, 257262. doi: 10.1038/nn.2726Salimpoor, V. N., Benovoy, M., Longo, G., Cooperstock, J. R., andZatorre, R.J. (2009). Therewardingaspectsof musiclisteningarerelatedtodegreeofemotional arousal. PLoS ONE, 4:e7487. doi: 10.1371/journal.pone.0007487Schachter,S.,andSinger,J.(1962).Cognitive,social,andphysiologicaldetermi-nants of emotional state. Psychol. Revi. 69, 379399. doi: 10.1037/h0046234Schfer, T., Sedlmeier, P., Stdtler, C., and Huron, D. (2013). The psy-chological functions of music listening. Front. Psychol. 4:511. doi:10.3389/fpsyg.2013.00511Scherer, K. R. (2005). What are Emotions? And how can they be measured? Soc. Sci.Inf. 44, 695729. doi: 10.1177/0539018405058216Schubert, E. (1999). Measuring emotioncontinuously: validity andreliabilityof the two-dimensional emotion-space. Aust. J. Psychol. 51, 154165. doi:10.1080/00049539908255353Schubert, E. (2012). Reliabilityissuesregardingthebeginning, middleandendof continuous emotionratings tomusic. Psychol. Music, 41, 350371. doi:10.1177/0305735611430079Shebasta, P. (1952). Les Pygmes du Congo Belge. Brussels: Institut Royal ColonialBelge.Sievers, B., and Polansky, L. (2013). Music and movement share a dynamic struc-ture that supports universal expressions of emotion. Proc. Natl. Acad. Sci. U.S.A.110, 7075. doi: 10.1073/pnas.1209023110Soto, J. A., Levenson, R. W., andEbling, R. (2005). Cultures of moderationandexpression: emotional experience, behavior, andphysiologyinChineseAmericansandMexicanAmericans. Emotion5, 154165. doi: 10.1037/1528-3542.5.2.154Turnbull, C. M. (1961). The Forest People. New York, NY: Simon and Schuster.West, B. T., Welch, K. B., and Galecki, A. T. (2006). Linear Mixed Models: a PracticalGuide Using Statistical Software. Chapman and Hall/CRC Press.ConictofInterestStatement:Theauthorsdeclarethat theresearchwascon-ducted in the absence of any commercial or nancial relationships that could beconstrued as a potential conict of interest.Received: 20 August 2014; accepted: 03 November 2014; published online: 07 January2015.Citation: Egermann H, Fernando N, Chuen L and McAdams S (2015) Music inducesuniversal emotion-related psychophysiological responses: comparing Canadian listen-ers to Congolese Pygmies. Front. Psychol. 5:1341. doi: 10.3389/fpsyg.2014.01341This articlewas submittedto Emotion Science, a section of the journalFrontiersinPsychology.Copyright 2015 Egermann, Fernando, ChuenandMcAdams. Thisisanopen-access article distributed under the terms of the Creative Commons Attribution License(CC BY). The use, distribution or reproduction in other forums is permitted, providedthe original author(s) or licensor are credited and that the original publication in thisjournal is cited, in accordance with accepted academic practice. No use, distribution orreproduction is permitted which does not comply with these terms.www.frontiersin.org January 2015 | Volume 5 | Article 1341 | 9