sites, 1:12.eM The Effect of Music on the Human Stress Response PublanedAuqut', 2013+ noed:or/1.17 Moura one 0070156 Abstract "Music listening has been suggested to beneficially impact health via stress-reducing effects. However, the existing literature presents liself witha linted number of investigations and with discrepancies in reported findings tal may resut from methodological shortcomings (eg. small ample size, no val stressor. Itwas the aim ofthe curent study to address ths gap in knowledge and avercome previous shortcomings by thoroughly examining music effects across endocrine, aulonomic, cognitive, ‘and emotional domains ofthe human stress response Sixty healthy female volunteers (mean age = 25 years) were exposed to a standardized psychosocial stress test after having been randomly assigned to one of thre diferent conditions prior tothe stress test 1) relaxing music Miserer\ Alleg) (RM), 2) ound of rippling water (SW), and 3) rest without acoustic stimulaton (R). Salivary consol and salivary alpha-amylase (sAA), heart rate (HR), raspratory sinus arrhythmia (RSA), subjective str porcaption and anxiety were repeatedly assessed inal subjects. We hhypoines'zed that listening to RM prior othe stress test, compared to SW or R would result ina decreased stress response across allmeasured param ‘The three conditions significantly difered regarding cortisol response (p= 0.026) othe stressor, with highest concentrations inthe RM and lowestin the SW condition. After the stressor, SAA (p=0.026) baseline values were reached considerably faster in the RM group than in the R group. HR and psychological measures gid nat sign‘fcanllydifer betwoen groups. Our findings indicate that music tistening impacted the psychobiological stress system. Listening to music pio to a standarizea sitessor predominantly affected the autonomic nervous system (in tems ofa faster recovery), and fo a lesser degree the endocrine land psychological strss response. These findings may help Better understancing the beneficial effects of music onthe human body. Citation: Thoma MY, La Marca R, Brénnimann R, Finkel L, Ehlert U, Nater UM (2013) The Effect of Music an the Human Stress Response, PLoS ONE 88): 70156, htpsli6o.or9/10.137 ioumal pone.0070156 Ezltor: Robert L. Newton, Pennington Biomedical Research Center, United States of America Received: April 12, 2012; Accepted: June 20, 2013; Published: August §, 2013 Copyright: © 2013 Thoma et al. This i an open-access article distributed under the terms ofthe Creative Commons ‘Attnbution License, which permits unrestricted use, eistibuton, and production in any medium, provided the eriginal author and source are credit. Funding: This study was supported by a grant from the Young Investigator Grant of the University of Zurich: (teu cesearchers.uzn.chioromotonforschungskredi_ gn him), Grant no: 96233208 (MVT). The funders ha no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. ‘Competing interests: The authors have declared that no competing interests ex Introduction Prolonged experiences of stress are related to poor individual heath (1.2| and associated with substantial financial costs forthe sociey [3], Asa resll, the development af cost eflecive stress prevention or stress management approaches has become an important endeavor of current research efforts. Music has been shown to beneficially affect stress-related physiological [4-8], as. ‘well as cogniive [7], and emational processes [.9). Thus, the use of listening io musie as an economic, non-invasive, and Mighty Accepted intervention tol has received special interest in the management of stress and stress-related health issues. “The experiance of stress arises when an incvidual perceives the demands from the environment ..2s taxing or exceeding Wis or her resources and endangering his or her well-being’ [10]. Accordingly, physiologic stress effecis are regulated by top-down central ervous system processes (~cognidive slress component, 6.9. Tcan cope with the situation), as well a by sub-corical processes ‘within the imbic system (=emotional stress component, e.g, anxily). Both areas forward ther messages (e.g. 1am in danger!) via nouronal pathways 1 a central contol system, the hypothalamus [1]. The hypothalamus is closely intertwined with two major Stress systems, the Rypathalamus-pitutary-adrenal (HPA) axs and the sympathetic nervous systom (SNS) (=physiologc stress component, endocrine and autonomous responses). Together, tne HPA axis and the SNS orchestrate various psychological (e.g. emotional pracessing) and physiological (e.g. endocrine and cardiovascular activation) processes fo ensure ine maintenance ofthe Homeostasis ofthe organism thats challenged by the experience of tress [11-13], The main effector of the HPA axis i the so-called stress’ hormone cartsal is cancentraion is measured and evalua inorder to have an index for HPA axis activation nitpsiiourals.plos.org/plosonelarticle?id=10. 137 journal pone.0070166Habstact0 anasites, 1:12.eM (14.5). Salivary alpha-amylase (AA) is a novel biochemical index for sympathetic nervous system (SNS) aciviy [16-19], Bath Parameters obtained particular interest in stress research as untke more traditional blood-derived stress markers (2.9. epinephrine fand norepinephrine), they can conveniently be assessed in saliva. Taken togather the experienca of stress fsa mull-faceleg Phenomenon that comprises cognitive and emotional components that are closely intertwined with physiological systems, whose ‘massengers / effectors found in saliva can be applied lo ebjecvely measure stress responses. Research on potentially beneficial effects of music listening on HPA axis functioning, ie. on stress induced cortisol release, has only cently been established. Signifeant postive changes in cortisol were reported when listening to music before ana / or during ‘matical interventions considered stressful (decreasas and lower increases in cortisol) [20-22] and after such interventons (graator Feduicons in cortisol 23,24]. The few laboratory.based studies show inconsistent findings, though some report that muse was fective in suppressing astress~elated increase in corisol [25], oF in decreasing cortsal levels flowing a stressor when ‘compared to a non-musie contol condition [5], However, some other investigations did not find a meaningful impact of music on ‘otial [26,27] As a consequence, no final conclusions can be drawn abou whether or how mus listening nuances strese- Induced eo'tso levels “The research on beneficial effects on SNS parameters has a longer tradition: series of clinical and laboratory-based studies revealed that listening to music can decrease sympatnetic actly [28-20]. However, postive SNS effects ofstening fo music are not consistently reported (20,31. Itis conceivable that knowledge achieved from the effects of music on an adetional SNS. parameter, such as the newly established SAA, would help fo increase understanding of inconsistent previous reports. However, to ate no laboratory study exists that has investigated the effects of music on stress-induced sAA levels ‘As listening to music has the capacity to inate a mulitude of cognitive processes inthe brain [32] it might be assumed that music also influences siress- related cognitive processes and, as @ consequence, physiological responses. Previous investigations found Feducions in perceived levels of psychological sess, increased coping ables, o altered lovee in perceived elaxaton ator Tstening to music nthe context of a stressful situation [7.33]. Another line of researcn has focused onthe effects of music on anxely, which may be considered an adapive response to the experience of stress. Given thal music listening can tigger actly in brain regions linked fo the experience of intense) emotions [824-36], Istening to music might also modulate anxiety levels induced by he experience of svess. Indeed, a decrease in anxiety afr lstening to music isthe most consstent findings roporlad in field stuties wih patients (22.37.28) anc lnboratory-nased studies (26.39). Nevertheless, not all investigations found ansiety Feductions through music listening [40-42]. Also here, no final conclusions can be drawn whether or how music is able lo influence Cognitive and emotional components of the stress response. Besides the insufficient quantity or pure lack of studies investigating the effects of music on stress-induced corisal or SAA levels, there are a number of methodological imitations that may account forthe wide discrepancies inthe already existing Merature. The ‘main reason for his dvergence inthe Iterature mignt be that many studies have been conducted ina clnial context, introducing heterogeneity by studying varous diferent medical seltngs and patient samples. Besides the valuable altempt of a small number of studies to investigate the effect of musi listening ina controled laboratory environment, these investigations suffered from methodological shoricamings, such as small sama size [S}andior the lack of a valid (e. HPA ax’ activating) stressor. From the perspective of biopsychologeal stress research, a major shortcoming 's the vast neglect of the control of confounding varables [25-271 Although acute stress responses occur rather uniformly across indlviduas (which makes ita good paradigm forthe investigation of acute stress), they may be modifed by previous individual experiences, such as chrono (affective) stress (43.44) What is mora, gven that stress [45], a8 well as music behavior [46,47] difers as a maior of how emotions are regulated in general, trails of emotion regulation should aiways be controlled, paricular in investigations examining the effct of music on stress Finally, the broad majority of previous work has used only ane (i any) contrel group (rest wilh no acoustic stimulation), and has not ‘examined whether postive effects of music are due the nature of music itself or du toa calming (non-music) acoustic stimulation, Consequently, the particular affacivaness of music listening on sass cannot yet be determined In sum, it appears that listing to music has the inherent ably to decrease the psychobiologial slress response. However, dus to the fact thatthe existing iterature isnot complete and often appears as inconsistent, definitive conclusions about the beneficial Slross-reducing effect of music may be loo premature. In light ofthese considerations, we set oul to examine the effect of fisting {to musie prior to a standaraized stressor across neuroendocrine, autonomic, cogntive, and emotional domains of the human sivess response in heallhy par‘paris in a laboratory seling. We pul @ spacial amphasis on he control of known influencing factors of the stress response and music effects, Le. depression, anxiety, chronic sess, and emotion regulation traits. To the best of our knowledge, such an endeavor has not been attempted thus far. We hypothesized that those participants wha listened to relaxing music pir tothe stress task would show a diferent stress responses in terms of corso, salivary alpha-amylase, heart rat, respiratory sinus arrhythmia, subjective perception of stress, and anxiety when compared to non-musie contol groups, ie. an Acoustic contol condtion (sound of ripping water) and a conirl condition resting without acoustic simulation, Methods. Participants were recruited by advertisement at the Unversity of Zurch and the Swiss Federal Institute of Technology, Zurich (owe t). Ina telephone screening, eitera for elgibiliy of interested participants (female sex, BMI between 18-25 kg/m, 20-30 years of ago, [Swiss] German as naive language and a regular menstval clo) wore vorfied. Fomale sox was chosen to contol {or gender diferences, as sexual dimorphism in doth the HPA axis response to psychosocial sess [48 49] and in physiological and ‘emotional respenees to 9 music stoning [29.50] have been observed inthe pasl, Given thelr confounding elect onthe organism in general, and the HPA axis in particular, exclusion erteria of tre curent study were the following; current depression, self-reported acite and chronic somatic or peychiatre disorders, mediealon, uso of hormonal contraceplives, use of psychoactive substances, land excessive consumption of aloohol (> 2 aleohol beverages / day) or tooacco (> 5 cigarettes day). Adatonaly, selFreported hearing defies of innius wore exclusion cloria, Individuals with musica raining were not included in the study I elighlly Tequirements were met, ard ofal agreement was obtained, appointments were scheduled during the woman's follicular phase (days 410) ofthe menstrual cycle to contal for hormonal variation troughout tre menstrual eye nitpsiiourals.plos.org/plosonelarticle?id=10. 137 journal pone.0070166Habstract0 anasites, 1:12.eM Flow diagram ofthe process through the phases of enrollment, allocation and analysis. tpslldo1org/10.137 Vjoumal pane 0070156.0001, In advance ofthe appointment, patcigants were genta set o information and several questionnaires (see below). Inthe advance ‘material, partipan's were informed about the course ofthe study, but were nat given detaled information about the expermental Sress paradigm. Study language was (Swiss) Gorman. Pariciparis were instructed nat ta dink alcohol or caffeinated beverages ‘48 hours prot to the study, Ad tionall, they wore fold to retrain from any exercise acts 24 hours prior to the experiment Further, participants wore asked to refainffom brushing thir fet or eating al least 60 minutes belore the study. Fer their Participation inthe study, the participants were reimbursed with 50 Swiss Francs, ‘Ana prior power analysis was conducted to estimate the optimal sample size to answer the main hypothesis ofa decreased cotisal response inthe music group when compared Io the coriral groups. indicated thal 4 participants were required to reach {an 87% power for detecting an effect of 0.18 when employing an alpha eriterion of 0.05 of statistical significance, “The study was conducted in accordance with the Declaration of Helsinl, The study protocol was approved by the ethics committees of the University of Zurich and of tha Canton of Zurich. Oral and writen informed consent from al subjects was obtained suey ai ‘The experiment used a between subject design to compare the elect of acoustic stimulation (Independent variable) on cortisol, SAA, HR, RSA, mood, and anxity (dependent variables). There were three conaiions prior toa stress test (Trier Social Stress Tost, TSST, see description below) a music condition (elaxing music stoning prior to stress test, RM), a water sound condition (an acoustic control conation including stern to sound of rippling water, SW) and a contol condition (non-acoustc cairo condition, including resting without acousic simulation, R). Seventy-eight participants flfiled al study requirements and were randomly assigned fo one ofthe groups. Eightesn parcipants were not able to Keep their appointment (see Figure 1). Randomization was Accomplished tough the use of a computer generated randomization Ist ‘All participants underwent standardized psychosocial laboratory stess protocol. The TSST consists ofan introduction (In) that lasts 2 minutes in which participants are introduced tothe procedure of the TSST. Specicaly, they are old that the TST consists ofa puble speaking task folowed by a montal aritymetic ask in rant of an audience. In te publi speaking task (lasting 5 minutes), parbeipants are asked to apply for a job, in his simulated ob interview, they are asked to talk about their personal ‘qualifications for tre chasen job, e.g, why they area better Mfr the job than olher apalicants. Right after the jab interviow, Pariipants are explained the nature of ie mental arthmetic task, which lasts for another 5 minutes. The paricipants have to Calculate backwards in stops of 17 rom the numbor 2043. Aor each calculation error the participants are asked to re-start tleulating from 2043. The TSST has repeatedly been found tobe a reliable tool ‘o activate oth the HPA axis and the autonomous nervous system (ANS) [BI In the current study the standard TSST procadure as reported in the IMeralure was sightly modified: the Inv, the subjects were not told about the exact nature of the upcoming speaking task (Le. glving @ speech as part of a ‘simulated job interview) in order to prevent subjects trom mentally preparing for te task. For the current study all examinations were conducted between 1200 and 1700h ta minimize the confounding effect ofthe hormonal sural myinm, Circad:an fluctuations af hormone levels are pariculrly pronounced in the morning hours and faten lroughout the day (2.53), Participants arved atthe laboratory 60 min prior to the anset of the stress induction by the TSST (igure 2), Participants were then escorted to a non-ntervention room, where they spent thee wating me between the actual rental interventions. Immediately after arrival, participants were informed by the main experimenter about the course of the experiment, Oral and writen informed consent was obtained from all participants. Right afterwards, the LifeShit, an nitpsiiourals.plos.org/plosonelarticle?id=10. 137 journal pone.0070166Habstract0 anasities, 11:12.eM lectrophysiclagical measurement device (gee below), was atlached, Alteran adapiation period of 20 min, a basal salva sample 1, -30 min) was taken. Twenty minutes prior to the TSST, the participants were drought othe TSST room, where they were introduced by the main experimenter tothe procedure of the TSST (= introduction: Ino, 2 min). The subjacis were then brought to the intervention room, seated in a comfortable chai, and provided with headphones. Al participants had to adjust atest signal (ius tone, sound prassure = -70d8) tothe ind vidual hearing threshold lovl forth calibration of the volume. After this, the Parllpants were to undergo thelr assigned condition, le, RM, SW. or R fr ten minutes, No instruction was given for any ofthe Eondltans. Immadiataly after ths part a second saliva sampla was taken (2, -5 min). Following this, subjects were Laken back into the TSST room where they were undergoing the TSST. Aer the completion ofthe TSST, the subjects were then returned back the non-ntervention raom and a tid salva sample was taken (3, + 10 min) Further sampies were faken 1 min (T4, + 25 min), 30, mmin (TS, + 40 min), 25 min (T6, + 55 min). and 60 min (T7, + 70 min) afer the TSST. In adaton, the subjects completed various selFreport stress measures (see below) at T1, before and after T2, at T3 and Té, ‘Timeline ofthe testing procedure ifpsldo\.org/0.137 Vioural pone.0070156.0002, ‘Miseror’ by Aller (CD Gimell 454 939-2) isa soothing and calming music piece (Latin choral singing) that was chosen ta induce relaxation in our subjecs, The simulis was selected on the basis of provious research [S, We decided to use a single Standarcized music stimulus, as ths approach is though! fo have a greater effect on stress reduction than music stimuli selected by the subjects themselves (34). Further, we wanted to avoid possible influences of memory or subjective associatons wih sel chosen music simul by parteipants Wi included a nor-music acoustic control coniton, Le. stoning to sound of rippling water, in our study. This control condition has been chosen to conta or effects on psychological and physiological parameters, which might be caused by mere acoustic stimulation alone. The sound of ripping water's missing the (ypical characteristics of music, such as a stuctured mology and ‘rythm. SU itis an acouste stimulus witha certain perceptual quay for the listener. Whats mor, in comparison to arficially produced sounds (such as white or pink noise or single tones), the sound of rippling water may be presente for langer periods of time without exerting sess or boredom in the listener ($0). craton ata ee ees Heart rate (HR) and respratory sinus arthytnmia (RSA) were measured wih the LfeShint® System, an ambulatory detection system that allows the continuous monitoring of cardiorespiratory parameters [55], and exited manta to correct for ectopic beats ‘with the VivLogic 3-1 sofware (Vivomotrics, Ventura, CA, USA). RSA is a measure er varallons in HR within a breathing ‘Sequence: itis used as an indicator for parasympathetic eardie contra HR and RSA were determined for Smite segments, ranging fom a baseline interval prior tothe Ino unl! 30 minutes after completion af the TSST. For the analysis of cortisol (as an indicator of HPA axis activity) [15] and salivary alpha-amylase (SAA, as an indicator of autonomic ‘actvy) (16,17, salva was callectes using smal coon swabs (Salvettes, Sarstedt, Sevelen, Switzerland). Stimulated saliva was taken by having the patpants gently chewing the cotton roll for 1 min. Thereafter, the cation rll was placed into a smal pase ‘ube. Samples were stored at -20" C uni biochemical analysis took place. Salivary free cortisol was determined by using & commercial chomiuminesconce immunoassay (LIA) (BL, Hamburg, Gormany).Inir- and inaassay coofcients of variation wore below 10%. samples of one subject were analyzec inthe same run to reduce error variance caused by imprecssion ofthe intraassay. Activity in SAA was analyzed using the microplate reader Synergy HT Mult-Modo (BioTok) and adaptod assay kis ‘obtained from Roche, The assay isa kinetic colorimetric test, Inter- and intreassay variance was below 1%, Demographic information such as age, education, medication intake, nicotine use ang illesses were collected using a demographic questionnaire. Questonnaires were used lo investigalo the role of music proference and psychological factors ‘The Music Preference Questionnaie (MPA) [56] was used to assess participants’ general preference for Classical musi, alsa in relation with their general music preference forthe most common muse styles: Pop, Rap / Hip Hop, Latin, Soul Funk, Hard Rock, Electr, New Age, Country, and Jaze music. On a &pomnt Likert scale participants indicates how much they liked the particular music style (1 not ata to very much) ‘The Beck Depression Inventory (BD!) [57] was used to contol fora possible impact of depression onthe HPA axis response (58) Scores higher than 18 are suggestive of clinically relevant depression Depending onthe dispositional preferred emotion regulation strategy, diferent cognitions, emotions, and behavior may resull in and ater emotional sivatons, To contal forthe impact of how emations are regulates in goneral the validated Gorman version [58] of the Emotion Regulation Questionnaire (ERG) by Gross and John [60] was used. The ERG assesses two comman tat emotion regulation strategies, reappraisal and suppression. Higher values on each scale denote greater expressiveness of the respecive variable. nitpsiiourals.plos.org/plosonelarticle?id=10, 137 journal pone.0070186Habstract0 anasities, 11:12.eM Visual analog scales (VAS) were employed to repeatedly measure subjective perception of eress during the experiment. To contal ‘or the experience of chronic stress in our sample, we used the screening scale ofthe Ter Inventory forthe Assessment of Chronic ‘Stress (TICS) [5], which assesses the global perceived chronic strass laad af an inavidual with 12 tems (Screening Scale of Grvonie Stress, SSCS), Paricipants were required to rate how often they had experienced cerain stressful situations during the past three months on aS-point Likert scale. High values are indicative tha the individual soften woried, overburden bverstrained, and unacknowledged ‘The Stale and Trat Anxiety Inventory (STAN) [82] was used to assess anxiety. The STAI consists of two 20-tems questionnaires which assess state respectively trait levels of anxity in clinical and non-clinical populations. Scores for both scales range between 20 (ow anxiety) and 80 (high anxiety), The STAL-siate was Used as a continuous measurement for possible changes In anxiety during the experiment. The STAl-rat was used to contral forthe effec of anxiely as a personaly train our sample [63] ‘The simul questions were used to assess the subjective perception of ether music or sound of rippling water. Subjects were required to rate how much they iked the stimulus, and how relaxing they perceived the stimulus on a S-poin Likert scale immediately after te stimulus presentation. High values are indicate for increased liking and of an increased relaxing effect ofthe stimulus Data analyses wore performed using SPSS (17.0) software packages (SPSS, Chicago, IL, USA). Homogeneity of variance was tested using Levene’ test before statistical analyses were applied. ll reported results were corecled by the Greenhouse-Geisser procedure where appropriate (viola of sphericity assumption [84 6S]. n case of missing data, cases wore excluded ist wise ‘Analyses of variance (ANOVAS) for repeated measures were compuied fo analyze possible me, conition and interaction effec. Fr comparison of the scale means of the questionnaires with normative samples, Student's Lies's were computed. Cortisol (-30 mri fo * 70 min), alpha-amylase levels as well as heart rate measures (-30 min to + 40 min) were evaluated according to the area lnder the curve with respect fo increase (AUC). The AUC} is related to the sensitivity ofthe biaogical system: itis pronouncing changes overtime, and is characterized by accumulation of the eror ofthe baseline, as the formala ts based on the difference between the baseline and the subsequent measures [6]. To estimate the extent of stress react of cortisol, SAA, HR, and RSA, ‘we calculated the deta measures af the stross responses (peak values aftr stressor minus baseline values before stressor), and Fefer tot as peak dott For the estimation of a recavery valve, we subtracted the fist baseline value afer the stressor from the peak values after te stressor (delta), and rear fo tas recovery deta. Calculated measures of AUC, peak della and recovery dala ‘were analyzad using ANCOVAS. For all analyses, rasults were considered stalsticalysignifcanl at Ihe p $0.05 level, and were onsidared-a and at the p < 0.1 level. Al tests were two-tallad, Unless indicates atharwiso all results shown aro means # standard ‘deviations ($0), Results ‘Sint healthy female subjects participated inthe study (age mean = 25.3 years, SO = 3.21 years; BMI (calculated 3s weight in kelograms divided by the square of height in melers) mean = 21.63, SD = 2.34: years of education mean = 15.3, SD = 2.56). Mean BDI scores of§6 (SD = 3.0; range = 0- 17) indicate no clnicalysignfieant depressive symptom severy in our sample. Mean TICS summary scores of 17.22 (SD = 755; range = 0-36) indicate low levels of chronic stress load nour sample, Nisan STALrat scores of 37.47 (S0 = 8.8; range = 22-53) Indicate low alt anxiely levels n our sample [see 62), Regarding Walk emotion regulation strategies, tie mean scores of the two scales of the ERG (reappraisal mean 4.83, SD = 1.0, suppression mean 3.29, SO £4108) were in a comparable range tothe female norm sample described by Gross and John [6Q]. Music preference for classical ‘music was high in the current sample, only Pop music was preferred more (Classical music preference mean = 3.33, SD = 1-1: Pop music preference mean = 3.72, SD = 1.18), Randomizabon resulted in 20 partispanis undergoing the experimental condiion (RM, 20 participants undergoing the non-music acoustic contrl condition (SW), and 20 participants undergoing the cortrol condition ‘without acoustic simulation (R). The randomized assignment to groups was evalualed by comparng demiographic variables (age, BMI years of education), preference for Classical musi, and means ofall contol variables (BDI, FRQ, STAI-ral, TICS) between ‘10up8 (all p= ns.) (IASLE +), We found no significant differences between groups, chart m=) (20) Rema) psi ‘ee ee) 2288) 007) sar em) ore ‘ears education 32225 ase 2a 58 (05) am {MP Cesen mle sasita) ssi05) 45 (080) ow ow 28200) 72) rm @9) ian ior ee 958) 553621) sows (ons Ra respoaeal 475(089) 550098) arsi35) ian [Ra suppression sown 3.68088) satan, on staneat 358169") 3835(1087) 625,095) 86 nes 6750277 corr 2678) oe ‘Table 1. Sample characteristics (Means, standard deviations and group illerances among study vara ?. Patcipantsked both acoustical stimuli (music: mean=3.21; SD=1.96; sound of ripping Waler: mean=3.84; SD=1.17) ané both ‘wore perceived as relaxing (music: mean=4.0; SD=0.88; sound of ripping water: mean=4.0; SO=1.2). None of aur participants in the acoustic simulation groups expressed any negative comments about the stmull (music sound of ripping wate). (One participant ofthe RM corsdltion and two participants ofthe R condition showed levels of cortisol that were more than 3 standare deviations higher than the moan, As @ consequence, these participants were excluded from all futher analyses, The ‘Sess protocol induced signfieantineceases in salvary corso in all groups overtime (F(2.48/183.74)=18.46; peO.001% 12=0 255), nitpsiiourals.plos.org/plosonelarticle?id=10. 137 journal pone.0070166Habstract0 sinasites, 1:12.eM Without including the contra variables, repegted-measures ANOVA revealed no significant group diferences (group-by-time interaction. F(4.95!733.74)=1.67, p=0.146, n?=0.058) With the inclusion othe contol variables (BOI, ERA, STALrat, TICS), cotisl concentrations difered significant between groups, wth highest values in the RM and lowest values inthe SW (group-by- time interaction: F(5.05/116.13)=2.68, p=0.025; 4°=0:104) (Figure 3). Single contralle group compargons revealed a signicant difference between the RM and SW groups (group-by-ime interaction: F(2,33/67 6)=4,72; p=0,009; 7°=0, 140), but not between RM ‘or SW when compared to R (alin. Final, univariate analyses revealed a significant cfference between RM and SW in their pak deta (F(1/29)=5, 18; 0.03; 77=0.152), and in the AUG, again between the groups RM and SW (F(1/29)=4.7; p=0.088; 177=0.139). No significant differences were found when RM and SW were compared to R regarding peak delta and AUC, Salivary cortisol levels in response tothe TSST (means + SEM), Inthe experimental group istening to relaxing music (RM), the contral group isting to sound of ripling water (SW), and the contol group resting without acoustle simulation (R} ifps ldo. org/0.137 Vjoumal pone.0070158. 0003. SAA activity increased significantly over the course ofthe stress task (A2.62/146.82)+15.60; p<0.001; 120.218). Without the inclusion ofthe contol variables (.e. BDI, ERO, STAL-rait, TICS), repeated-measures ANOVA revealed no significant group differences (group-by-time interaction: F(5.24/146.82)=1.19; p=0.318; 1°=0.041). Also with the inclusion of the control variables, we fgund were no significant diferences in sAA actly between groups (group-by-time interaction: F(4.96/119.01 1.4; p=0.23; 170,055) (Figure 4). Univariate analyses however revealed a significant difference in the recovery delta between groups (F(248)=4.18; p=0.022; n?=0.147). Single group comparisons revealed a signficant ference between RM and R (F(1/31)=0.547; 026; 1°=0.15) and between RM and SW (F(1/29)=4.7; p=0.039; 17°=0.139) in the recovery delta; sAA activity in the RM Condon is back at baseline at T (+ 25 min), compared to R or SW at TS (+ 40 min, 8 ¢ etry Ah ne Ut be Figute 4 Savery alph-amylase activin response to the TSST. Salivary alpha-amylase actly in response to the TSST (means SEM) In the experimental group listening to relaxing music (RM), the contol group listening to sound of rippling wate (SW), andthe control group resting without acoustic stimulation ® ‘ifpsido‘.org/10.137 Vfoumal pone.0070156.9004 Cardiac measures changed signifcanty aver the course ofthe exporimant over ime (HR. 16/151.77) = 122.05; p<0.001 10.027. RSA: F(3.3/188.48)=20 41, p<0.00'; nf=0.288), HR and RSA showed mifrored stress responses (FIGURES § and 8) Without the incision of the control variables (ie. BDI, ERG, STAL-talt TICS), repeated-measyres ANOVA revealed no significant group diferances conceming HR (group-by-ime interaction: F(6.32/151.77)=0.68; p=0.692; 7°=0.027) or RSA (group-bytime interaction: F(6,6/158.49)=0,86; p=0.533; 1°=0.035), Also with the inclusion of the control variables, groups did not significantly dior over the course of the experiment concerning HR (graup-by-time interaction: (6.7311 03.2)=0.6:p=0.72; 7°=0.082) or RSA (group-by-time Interaction: F(S.76/103.7)-0.98; p-0.456; 10.08), However, groups signfcarlydifered in the recovery delta of RSA (5 min after cessation of TSST) (F(2/40)=4.06; 2=0.025; n2=0.169). Single group comparisons revealed a sgnifcant diference betwaen SW ang R (F(1/27)=6.70; p=0.015; 7?=0. 199), suggesting a faster recovery of SW afer he TSST in RSA when compared wR nitpsiiourals.plos.org/plosonelarticle?id=10. 137 journal pne.0070166Habstract0 enasities, 1:12.eM : ayy Heart rate in respanse tothe TSST (means + SEM) inthe experimental group Istening to relaxing music (RM), ‘0up listening to sound of rippling water (SW), and the contol group resting without acoustic stmalation (R. ips ido. org10.137 Vjoumal pone.0070156.6005, i a i 2 * eciy iat ft mat peers E ZEEE Figure 6 Respiratory sinus arrhythmia in response tothe TSST, Respiratory sinus armhytmia in response to the TSST (means « SEM) in the experimental group listening to relaxing music (RM), the contol group stoning te sound of rippling water (SW), and the control group resting without acoustéc stimulation ® iipsldoiorg/10,137'/joumal pone 0070156.0006, Peet ‘The subjective perception of stress during the experiment (VAS), as well asthe anxity (STAL-state) significantly changed overtime YAS and STAI-stae increased from baseline to ater the Inio of TST, deceases in response to tne experimental and contol conadltons (72, -§ min), inereased in response to tne TSST (T3, + 70 min) and decreased again thereafter (T4, + 25 min) (VAS. F{2.87/141.43}-18.79, p<0.001, STAlstate’ F[2,9/50)-22.55; p<0.001), For all further analyses, we controlled forthe influence of chronic sts (TICS) when analyzing the VAS scale ofthe porception of stress, and we controled forthe influence of rat arwiaty (STALArait when analyzing the STAFsate, Nether the VAS (group-by-time interaction: F(5-14/136.12)=0,78; p=0.562),nor the STAL-stale (gcoup-by-ime inleracton: F(5.791141.87)=1.42; p=0.213) did signiicanly flr between groups. Discussion “The purpose ofthe curent study was to examine the effects of Istenng fo relaxing music prior toa laboratory stressor on endocrine, aulonomic, cognitive, and emational responses in heallhy women. Wa hypathesized that thase participants who listened {relaxing music prior toa stress task would show ifferentsiress responses in comparison to non-musi control groups, Le. an Acoustic contol condtion (sound of ripping waler) and a contra candivon, resting without acaustie stimulation, With regard to endocrine responses, we found significant ferences between the music andthe acoustic control condition in cortisol: nghest Cotisol concentrations were absorved in incviduals who were Istening to music prior the stress tas, lowest concertvatans were ‘ound in those who were istening tothe sound of oping water. Although there was no signficant effect of music regarding Autonomic responses, we observed a fond towards faster racovery in SAA activity and in RSA\n the musie group. As for psychological measures, we didnot find signficant diferences between the three groups. ‘The lack of a decreased cortisol response in the music group corroborates findings ofa previous study by Knight and Rickard (26) “To the best of our knowledge, that investigation isthe only laboratory-based study in which the influence of music listening price to @ sttessor was investigatee, However, the explanatory power af the dings was somew/at limited asthe authors did not apaly a Stressor that was strong enough fo elt a signtfcant cortisol strass response and that participants were examined in small groups (6-12 partepanis per conction). As forthe latter, recent research has suggested that music mterventions might be more effective fn an individual compared tothe group level (Sd. Nevertheless, evan though we appliec a signicant stressor and tested our Parlipants one-by-one, we didnot find an alonvation of sess-induces corti! levels aller music listening Wie di not expect the observed relative inevease of cortisol cancentrations inthe music group in comparison to the contol groups. Cur findings seem to be the opposite rom what the majony of previous studies have reporied when investigating the effect of music on baseline HPA axis funcloning, Le. 2 sgreant decrease in cortisol concenatone [82-77] Our findings also seen) to die rom tn effects of music, wen music is presenteg during (suppressed cortisol response) [73] or ater siressor (decreased Catisal response) [S.Inlerpreting our findings in the light of these studies, we may deduce that the beneficial effect of music on HPA axis functioning depends on situational context (rest vs. stress) and chronology of events (rir vs. during vs. after stress). A possible explanation forthe context-dependent influence may bs the involvement of certain brain areas and ther subdivisions, such 85 for instance the hypothalamus, the amygdala, the hippocampus and the prefntal cortex, In both listening to music and stress nitpsiiouals.plos.org/plosonelarticle?id=10. 137 journal pone.0070166Habstract0 masities, 1:12.2M processing [8,34.38.78-80]). It appears that music Iistening prior to the experience of stress may add fo, faciltate, or increase Subsequent HPA axis activation by a staggered actvation of a parly shared neuronal network. Brain-imaging studies are needed to investigate whather hare is indeed a spectic'cambinationa’ affect of music and stress tis of great interest hat the lowest concentrations of cortisol were abserved in the acoustic carol condition (listening to the sound of rpping water). Given that, to the best of our knowledge, no study 60 far has investigated the effects of natural sounds on Siross-rlalad HPA axis rasponses, wo ci not anticipate this outcom. The sound of rippling water was equally preferred and perceived as relaxing as music by ourpartcipants. This may be a nt to differental effects on the HPA axis of music and ron-music Acoustic simulation. Due to the lack of stud'es on neuronal activation pattorns of natural sounds ts diffcu to determine the exact mechanisms for ths effect, though, Although unexpectes, the fnding of deceased corbsol oncentratons inthe acoustic contol Conalion is of polentialy great relevance and may ierease our unce'standing of mechanisms of beneficial interventions in natural envronments, based an the concept of biopilia for instance (B18), According to this view, humans, who have lve in natural tenvronments throughout evaition, are equipped with brains mental funetioning that ~.subcansciously seek conection with all thats alive and vita (p. 4860) [81], As a consequence, humans indeed feel more comforted in natural than n urban environments. Lingering in natural environments, such as for instance ina forest, has been foun to have signficant beneficial physiolog cal effects (83). Accoreingly. a detachment from natural environments might lead to decreased physical and psychological wel being Tad}. I might therefore be reasoned thatthe saund of ppling water in our study had a relaxation eect stronger than that of music, ddue tots inherent characterstes asa sound of nature, OF course, ths nation hghly speculate, More studies are needed to fexamine the diferenal endocrine effects of natural and non-natural acoustic sound stimulation, We observed a diferentialinfuence of music listening on autonomic actviy: musi resulted ina faster autonomic recovery afler sires compared to the contral groups. Ths pally corresponds wih ndings from an investigation by Arai etal [85] who found signficanly decreased SAA evels at wound closure in patents who listened to intra- operative music when compared to @ non- music control canditon. Music might thus faciltate autonomic recovery from a stressor in comparison to listening fo non-musical Sounds or no acoustic stimulation. The fact that aur finding only showed a statistical rend narrows ts relevance, however. Other investigations assessing the effects of music on the ANS (e.g. via epinephrine and norepinephrine) have found no beneficial elects [27.6 As for cardiac mensures, we found a decrease in Ht and an inerease in RSA in response to RM, SW, and R. After stress fexposure, we found an increase in HR and a decrease in RSA. On the one hand, these Indings correspond lo investigations that ‘ound an increase of parasympathetic actviy in response fo sedative music listening (87-89) On the ofr hang, our results cotroborale findings from studies reporing decreased parasympathetic acy in response to stoss [20 81). As wilh SAA, we found 8 trend fora faster recovery ofthe RSA n the music group when compared othe resting contol group. Ht appears that music Istering might be effective In accelerating the recovery process ofthe parasympathetic branch of te ANS. It's interesting, however, thatthe sound of ppling water was even more effective than musi in etuming RSA levels back to baseline. Clearly, further studies are warranted for further elcking the diferential physiological elfects of music and non-music acoustic simulation Music tstoning had no diferential eect on psychological measures (stress perception or anxety) in comgarson tothe two contol conditions. Ths isnot in ine with investigations that report stening fo musio to be eectve in reducing psycholog cal stress [33] or ancety [26.37-36]. One explanation might be thal music listening may only reduce psychological stress / anxely inthe presence of a relatvely mid stressor It might be the case that the stressor in our study (Le. the TSST] was too strong. Knight and Rickard [26], ‘who were using a (mild) cognitive stressor inthe laboratory, found anxely-reducing effects of music Istenng prior to stress. MacDonald and colleagues foune simiar effects only in those patients who had a minor surgery (mild stressor) and notin tose ‘who had a major surgery [92]. Evans (40), nally, systematically reviewed sludies ofthe effeciveness of music interventions for hospital patients. He found that musicstening was effective forthe reduction of anxiety during normal care delivery (which may be Considared as mid stressors), but not fr palionts undergoing invasive or unpleasant procedures (srong stesso). In contrast to those findings, however, patents in the study by Alen etal were experiencing ..a high level of stress and anxiety.” [33] related to surgery, so thal ane may assume thal this was a stong stressor. Sil, music was ellectve in decreasing perceived stress levels in that study, However, paionis were allowed to Isten to their own choice of music. It might be argued that nat the musi ise, ut the positve memories associated wih it caused this effec. What is mara, control patients didnot wear headphones and ware therafore exposed to the sounds of surgery, thus further inducing stress inthe contol group. Future studies are needed to test for the assumption hat music listening might only reduce stress related psychological processes and anxiety n the context of mil Stressors, ‘Taken together, our results seem fo indicate that pre-sress musiclistening might not be effective in reducing the biopsychologcal slress response, but might, in contrast, add fof faciate a stress response. However, or results may also be inferpreted inthe light of another explanation: it may be that the participants In the musie group were actualy so relaxed thatthe subsequent stress induction was incompatile with ths state of relaxation, and that they produced an increased strass response as a consequence. We might have therefore measured the effect ofthe convast between a relaxing and a stvessul state rather than the preparatory fects of relaxing music on the subsequent stress response. This notion is supporie by the greatest increase in stess percapton in the relaxing music group. Future studies should folow-up on this explanation and futher dissect the effects of preparatory music Istening on stass responses. ‘Although this isthe fist stugy in which the eect of pre-stress music Istening on a mulitude of stress response domains was examined inthe context ofa rigorously conzolled laboratory setting, our findings need to be considered in the ght ofthe following Iimitatons Standardized music stmul, selected by the researchers, might have diferent effects than those chosen by the participants thomeeles. In our sludy, however, we used a music sumulus which hag already been evaluated as relaxing in previous research IB], 80 we were confident that this stmulus had stress-attenuating capacity independent af individual preferences. Also, using Fesearcher-selected music stmull has been shown lo have greator effects on siess reduction than music stiull selected by the Subjects themselves [54 The focus on healthy young female partcipants, which were non-smoking, not taking ary oral contraceptives, and being inthe follcutar phase of the menstual cycle, restrets the generalization ofthe results beyond this particular sample, While the exclusion of potential confounding varables certainly improves intemal vai, further research should investigate men also, as well as @ mixed sample of man and women with more lberal inclusion criteria in order to reach more general conelusion. nitpsiiourals.plos.org/plosonelarticle?id=10. 137 journal pane.0070186Habstract0 ana