Cognitive and Emotional Processing of Music and its Effect on Pain

Transcription

1 Cognitive and Emotional Processing of Music and its Effect on Pain PhD Thesis by Eduardo Adrian Garza Villarreal, M.D. Faculty of Health Sciences Aarhus University Royal Academy of Music Aarhus/Aalborg 2011

2

3 Cognitive and emotional processing of music and its effect on pain Eduardo Adrian Garza Villarreal, M.D. Ph.D. Thesis Faculty of Health Sciences Aarhus University 2011

4 Cognitive and emotional processing of music and its effect on pain The thesis is based upon the following manuscripts: I. Garza-Villarreal, E.A., Brattico E., Leino, S., Østergaard, L., Vuust, P. (2010). Distinct neural generators of the MMN and the ERAN to chord violations: A multiple source analysis study. Brain Research. Accepted. II. Garza-Villarreal, E.A., Brattico E., Vase, L., Østergaard, L., Vuust, P. (2010). Superior analgesic effects of mental arithmetic versus unfamiliar music and Sounds: The role of emotional impact and personality traits. Journal of Pain. Under Review. III. Garza-Villarreal, E.A., Brattico E., Vase, L., Østergaard, L., Vuust, P. (2010). The placebo effect of music-induced analgesia. Pain. Submitted. 2

5 Cognitive and emotional processing of music and its effect on pain Para Laura y mis padres, por su amor y apoyo. 3

6 Cognitive and emotional processing of music and its effect on pain 4

7 Cognitive and emotional processing of music and its effect on pain PREFACE The current Ph.D thesis work developed from an interest in the cognitive and emotional responses to music listening as possible mechanisms behind its well-known analgesic effects. Music is one of the passions in my life and, as such, the influence it possesses on all human cultures fascinates me. When I first learned there was an area in Neuroscience that focused on the research of music, something inside me sparked. My passion for the workings of human body and for music could actually be combined and become my field of work. It was when I embarked into the territory of the cognitive and emotional processing of music with Study 1, where I realized how different Psychology is from my background, Medicine. Understanding neural processes from a psychological perspective gave me nightmares. However after some time of reconciling psychology and medicine, it all started to make sense and music influence in cognition and emotion became my chosen path. As my understanding escalated, so did my interest in the effects of music in sensory systems beyond auditory, and thus I started to become entangled in researching the complex, and still unknown, mechanisms that give rise to analgesia with music. With my experience in cognitive and emotional processing of music, it was clear to me that the way music affects pain should be closely related to these neural processes. Therefore, I focused Study 2 specifically on the analgesic mechanisms of music by manipulating the cognitive and emotional elements of the auditory stimuli researched. I believe that it is possible to understand how music reduces pain. With this knowledge we can enhance the therapeutic uses of music, and even further understand about how cognition and emotion themselves influence pain. 5

8 Cognitive and emotional processing of music and its effect on pain 6

9 Cognitive and emotional processing of music and its effect on pain TABLE OF CONTENTS PREFACE... 5 TABLE OF CONTENTS... 7 ABBREVIATIONS INTRODUCTION AIMS BACKGROUND MUSIC Auditory processing of music Cognitive processing of music Emotional processing of music PAIN Pain processing Cognitive modulation of pain Emotional modulation of pain Placebo analgesia MUSIC AND PAIN MATERIALS AND METHODS ETHICS CHOICE OF METHODS STUDY Participants Stimuli Procedure Data acquisition and analysis STUDY Participants Stimuli Procedure Data acquisition and analysis RESULTS EXPERIMENT Event-related potentials Discrete and distributed source analysis Behavioral analysis EXPERIMENT Pain, emotion and cognition Placebo analgesia DISCUSSION COGNITIVE RESPONSES TO MUSIC AND THEIR RELATION TO EMOTIONAL PERCEPTION

10 Cognitive and emotional processing of music and its effect on pain 6.2 INFLUENCE OF EMOTION AND COGNITION IN ANALGESIA Cognitive modulation of pain Emotional modulation of pain Placebo analgesia of music CONCLUSION PERSPECTIVES COGNITIVE MODELS OF SENSORY INPUT STRUCTURE LANGUAGE AND MUSIC PREDICTIVE CODING AND MUSIC MECHANISMS OF MUSIC ANALGESIA MUSIC COMPOSITION FOR ANALGESIA THERAPEUTIC INDICATION FOR PAIN MEDICINE FUTURE STUDIES SUMMARY ACKNOWLEDGEMENTS REFERENCES APPENDICES

11 Cognitive and emotional processing of music and its effect on pain ABBREVIATIONS AC1 ACC Am ANOVA BA BESA BPM CC DBS EEG ERAN ERP fmri HT IC IFG MMN OFC PAG PASAT PFC PPC NAcc racc RVM Primaryauditorycortex Anteriorcingulatecortex Amygdala Analysisofvariance Brodmannarea Brainelectricsourceanalysis beatsperminute Cingulatecortex Deepbrainstimulation Electroencephalogram Earlyrightanteriornegativity Eventrelatedpotential Functionalmagneticresonanceimaging Hypothalamus Insularcortex Inferiorfrontalgyrus Mismatchnegativity Orbitofrontalcortex Periacqueductalgray Pacedauditoryserialadditiontest Prefrontalcortex Posteriorparietalcortex Nucleusaccumbens Rostralanteriorcingulatecortex Rostralventromedialmedulla 9

12 Cognitive and emotional processing of music and its effect on pain SI SII SMA SPM Std STG SVC VAS VLPFC VMPF Somatosensorycortex Secondsomatosensorycortex Supplementarymotorarea Statisticalparametricmapping Standard superiortemporalgyrus Smallvolumecorrection Visualanalogscale Ventrolateralprefrontalcortex Ventralmedialprefrontalcortex 10

13 Cognitive and emotional processing of music and its effect on pain 1. INTRODUCTION The processing of the auditory elements that form music such as pitch, loudness, duration, etc., starts in the cochlea located in the inner ear, follows through the brainstem and continues in every level of the auditory pathway until the auditory stimulus reaches the auditory cortex, located in the temporal lobes, where it is first made conscious (Pickles 2008). However, this description barely scratches the surface of the complexity entailed in the processing of music. Music is processed by several regions of the brain associated with every aspect of our behavior, influencing them at the same time. These regions include the executive functions of the frontal cortex, to the impulsive and primitive emotions of the amygdala, the memories stored in the hippocampus, and even the somatosensory cortex that specializes in the processing of touch and physical pain. This wide range of neural influence by music seems to be universal as it is present in all human beings, with, however, clear cultural variants (Fritz, et al. 2009). When we listen to music, the pre-frontal cortex expects structural (syntax-analog) patterns in the music derived from long- and short-term memory (Koelsch 2005). Regardless of whether these expectations are fulfilled or violated, areas that process reward and emotions interact and are modulated by a multitude of internal and external factors related to the music (i.e. preference, personality, environment, etc) (Brattico and Jacobsen 2009). In other words, music listening involves complex cognitive and emotional processes engaging several different areas that influence behavior such as the limbic, language, motor and somatosensory systems. It is, therefore, not surprising that a complex stimulus such as music could affect the perception of one of the most complex sensations, pain. 11

14 Cognitive and emotional processing of music and its effect on pain Music has been used as a therapy for pain since ancient times (Munro and Mount 1978). However, even to this day the mechanisms, effects, and even dosage of music for pain are not yet fully understood. Current theories suggest the analgesia secondary to music works via cognitive and emotional processes (Garza-Villarreal, et al. submitted; Mitchell, et al. 2006; Roy, et al. 2008). Cognitive processes, such as expectation of pain relief, distraction from the pain, and reappraisal, contribute to analgesia via the descending pathway of pain (Wiech, et al. 2008b). Another well-known cognitive analgesic effect is placebo analgesia. Positive emotions attached to a stimulus, as well as the mood of the individual, have also been shown to reduce pain (Chung, et al. 2007; Roy, et al. 2009; Villemure and Bushnell 2009). From all this evidence, it appears that the main mechanisms working to reduce pain during music listening are mainly derived from cognition and emotion. Therefore, it is crucial to investigate music with pharmacology-like pain designs, by studying musical stimuli as active analgesic drugs and isolate these mechanisms. This will help deduce how and to which extent cognition and emotion influence pain and whether there are other unknown mechanisms. A profound knowledge of the mechanisms responsible for the analgesic effects of any drug will effectively enhance its therapeutic use, and music is not an exception. In this thesis I will discuss the cognitive processing of music by the auditory and frontal cortices. Then, I will discuss the implications of cognitive processing and its influence for emotional processing of music, and for that of other sensory systems. Finally, the focus of the rest of this work is on the effects of music listening on pain and the cognitive and emotional mechanisms producing analgesia. 12

15 Cognitive and emotional processing of music and its effect on pain 2. AIMS To understand the neural responses of music, the cognitive and emotional mechanisms and also their influence on the analgesic effects of music, we performed two experiments with two different aims: 1) To study the cognitive and emotional processing of musical structures, we examined the event-related brain responses recorded with electroencephalography (EEG) in nonmusicians. 2) To study the influence of cognitive and emotional responses to music on pain, we induced experimental pain to investigate the effect different auditory stimuli with similar emotional characteristics, the placebo effect, and the effect of suggestion. The results of these experiments were discussed in three articles, each of which corresponded to the following hypotheses: I) Cognitive processing of music follows a hierarchical structure acquired from longterm exposure to a music culture and stored in long-term memory. Violations of culturedependent harmony rules would elicit a specific event-related brain response (early right anterior negativity or ERAN) mainly generated in the frontal cortex, whose amplitude is related to the degree of violation of the hierarchical structure and reflected by reported emotional responses. Violations of simple hierarchical structures (music scale rules) would instead elicit a brain response (mismatch negativity or MMN) mainly localized in the temporal cortex. II) Active auditory distraction (mental arithmetic) reduces pain more than environmental Sounds and Mozart Music. Passive auditory stimuli (environmental Sounds and Mozart 13

16 Cognitive and emotional processing of music and its effect on pain Music) with similar emotional characteristics have the same analgesic effect, which is also influenced by cognitive style (such as being emotional or having a tendency to focus on systematic and analytic structures). III) An important part of the analgesic effect of music is mediated by suggestion and belief (placebo analgesia), and these are influenced by happy and sad emotions attributed to the auditory stimuli. 14

17 Cognitive and emotional processing of music and its effect on pain 3. BACKGROUND 3.1 Music Auditory processing of music Music begins to be processed inside the cochlea as soon as it enters the ear. The basilar membrane encodes the elements of the music by activating the hair cells along its length. The hair cells are arranged at equal distances along the membrane to encode the sensory information. This sensory information is then transferred through the auditory nerve to the brainstem where elements like direction of the sound, pitch, timing, loudness, and consonance are unconsciously determined (Langner 1992; Scott and Johnsrude 2003). From there, music reaches the auditory cortices, both located in the temporal lobes, after roughly 13 ms. The right hemisphere is thought to be specialized for the processing of music, particularly of its spectral feature, like pitch (Maess, et al. 2001; Platel, et al. 1997; Rauschecker 1998; Trehub, et al. 1993; Zatorre 2001). However, there is clear evidence of musical processing in the left hemisphere as well (Jentschke and Koelsch 2009; Koelsch, et al. 2005; Platel, et al. 1997; Schlaug, et al. 1995b; Tervaniemi and Hugdahl 2003; Tervaniemi, et al. 2000). The auditory cortex processes all sensory aspects and elements of the music such as perceived pitch, timbre, rhythm, loudness some aspects of melody (Koelsch and Sammler 2008; Rauschecker 1998; Rauschecker 2001; Zatorre, et al. 2002). 15

18 Cognitive and emotional processing of music and its effect on pain Figure 1: (adapted from Purves, D. et al. (eds.) Neuroscience (3 rd edition, 2004)) Simplification of the auditory pathway, from the cochlea to the auditory cortex. It is anatomically and functionally divided in primary and non-primary auditory cortex. The primary auditory cortex (AC1) is located in Heschl s gyrus and it is tonotopically organized. It is thought to be specialized for the processing of pitch and loudness. The non-primary auditory cortex is loosely located surrounding AC1 in the planum temporal and planum polare. Its function is less known, and it is thought to process broadband sounds (bandwidth), sound motion and location, and integration of other elements of the sounds (Hall, et al. 2003). 16

19 Cognitive and emotional processing of music and its effect on pain Cognitive processing of music The auditory and frontal cortices are able to extract regularities and form hierarchical structures from the music (Koelsch, et al. 2001; Koelsch, et al. 2006; Näätänen 1995). Some regularities may consist of repetitions of one or more features contained in the sounds or in rules of succession of particular sound features (Paavilainen, et al. 2007). Other kinds of regularities are hierarchically organized, meaning that events within those regularities have different weights and roles according to previous knowledge, e.g., music harmony (Koelsch and Sammler 2008). Some theories state that harmony processing is analog to syntax processing in language, as both processes occur in the frontal cortex in a region called Broca s area (although harmony processing occurs more frequently in the right hemisphere). These areas keep us attentive to the music and make us expect a coherent musical structure derived from repeated exposure and long-term memory. Any resolution or violation of these expectations influences other brain regions. Violations in successive sounds elicit a specific event-related potential (ERP), detected on electroencephalography (EEG) called the mismatch negativity (MMN), which originates in the auditory cortex and inferior frontal cortex, whereas violations of harmonic structure elicit the early right anterior negativity (ERAN), which is thought to be located in Broca s area and its right homologue (Maess, et al. 2001). 17

20 Cognitive and emotional processing of music and its effect on pain Figure 2: (adapted from Tramo (2001) Representation of the areas of the brain that might be involved in music perception and performance, and their suggested function in the network. Music processing also involves pre-motor and motor areas, as it is not strange to tap to a beat or even dance to music (Koelsch 2006; Schlaug, et al. 1995a; Zatorre, et al. 2007). It also involves memory areas like the hippocampus and the dorsal medial prefrontal cortex, as familiar and even unfamiliar music can evoke strong memories and nostalgia (Barrett, et al. 2010; Janata 2009; Miranda and Ullman 2007; Quoniam, et al. 2003). 18

21 Cognitive and emotional processing of music and its effect on pain Emotional processing of music Music is a strong inducer of emotions (Barrett, et al. 2010; Blood, et al. 1999; Fritz, et al. 2009), and the consequent emotional reactions involve endogenous opioids (Benedetti and Amanzio 1997; Goldstein 1980; Rhudy and Meagher 2001). These reactions correlate with activity in paralimbic brain regions, changes in serotonin levels, increased levels of growth hormone and decreased levels of IL-6 and epinephrine (Blood, et al. 1999; Conrad, et al. 2007; Evers and Suhr 2000). During unpleasant music listening, serotonin levels increase and there is activity in the parahippocampal gyrus and precuneus regions (regions related to memory), whereas pleasant music activates brain regions implicated in reward and emotion, such as the ventral striatum, midbrain, amygdala (Am), the orbitofrontal cortex (OFC), and ventral medial prefrontal cortex (VMPF) (Blood and Zatorre 2001; Green, et al. 2008; Koelsch, et al. 2008; Menon and Levitin 2005). In general, the processing of music involves an extensive active neural network connected to almost every cortical and sub-cortical structure in the brain. Several studies have shown that pleasant and unpleasant music modulates emotion and mood (Roy, et al. 2009; Vrana, et al. 1988; Walker, et al. 1997). Clinical studies with music have also shown that it is successful at reducing anxiety and pain in patients (Allred, et al. 2010; Conrad, et al. 2007; Klassen, et al. 2008; Nilsson 2008; Shabanloei, et al. 2010). One recent study by Juslin and Västfjäll (2008) proposes that there are many mechanisms of how music evokes emotions: brain stem reflexes, evaluate conditioning, emotional contagion, visual imagery, episodic memory, and musical expectancy. In conjunction, all of these mechanisms could explain the variation and wide extent of the musical experience, and could possibly explain the mechanisms behind the analgesic effect of music. 19

22 Cognitive and emotional processing of music and its effect on pain In paper I, our aim was to finally dissociate the ERAN from the MMN to further understand the cognitive processing of musical structure and the relation between the elicited eventrelated potentials and emotional perception. 3.2 Pain Pain processing Pain is an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage according to the International Association for the Study of Pain (IASP). It is a subjective experience created by a neural representation, influenced by psychological factors. The pathways that process pain are twofold: ascending and descending (Bingel and Tracey 2008; Jensen 1997; Kong, et al. 2006). The ascending pathway originates from the peripheral nerve endings, ascends through the spinal cord and the brainstem, and ends at different brain regions usually referred to as the pain matrix, which includes the cingulated cortex (CC), somatosensory cortex (SI), thalamus and insular cortex (IC). However, this concept oversimplifies the extent of the sensory, cognitive and affective regions normally involved in pain perception depending on the type of pain, its intensity and unpleasantness, such as the posterior parietal cortex (PPC), hypothalamus (HT), supplementary motor area (SMA), prefrontal cortex (PFC) and Am (Kong, et al. 2006; Porro, et al. 1998; Price 2000). Two dimensions of pain are studied: pain intensity and pain unpleasantness. 20

23 Cognitive and emotional processing of music and its effect on pain Figure 3: (adapted from Bingel and Tracey (2006)) fmri BOLD activity in response to thermal painful stimuli overlaid on a structural MRI to visualize the pain matrix. Pain intensity can be described as the physical or sensory dimension of pain. Perception of the intensity of pain can be divided into sensory encoding and cognitive evaluation. Sensory encoding is thought to be processed by the SI, secondary somatosensory cortex (SII), anterior cingulate cortex (ACC), thalamus, and IC (Coghill, et al. 1999; Derbyshire, et al. 1997; Kong, et al. 2006; Porro, et al. 1998). Pain unpleasantness is described as the emotional or affective dimension of pain, and the main neural regions related are the ACC, IC, and Am (Price 2000; Schon, et al. 2008) Cognitive modulation of pain The descending pain pathway is connected to the ascending sensory pathway, and it modulates and controls pain perception through various neurotransmitters including endogenous opioids (Bingel and Tracey 2008; Jensen 1997; Jensen and Sindrup 2002). These 21

24 Cognitive and emotional processing of music and its effect on pain modulatory pathways descend mainly from the prefrontal cortex, cingulated cortex, amygdala, and hypothalamus, to converge in an area in the midbrain called the periacqueductal gray (PAG), that controls pain transmission neurons through a relay in the rostral ventromedial medulla (RVM) (Bingel and Tracey 2008; Fields 2000; Wiech, et al. 2008b). Therefore, at least anatomically, there seems to be a relationship between cognition and pain. Figure4:(adapted from Bingel and Tracey (2006)) Schematic of the ascending and descending pathways of pain, showing cortical structures and pain modulation in the brainstem. 22

25 Cognitive and emotional processing of music and its effect on pain Cognitive modulation of pain is ruled by three mechanisms: attention, expectation and reappraisal (Wiech, et al. 2008b). First, the direction of attention affects the perceived intensity and unpleasantness of painful stimuli (i.e. attending the pain increases it) (Brooks, et al. 2002; Miron, et al. 1989; Mitchell, et al. 2006; Tracey, et al. 2002). Functional Magnetic Resonance Imaging (fmri) studies show that when we expect pain or pain relief, the brain areas related to pain itself increase or decrease activation respectively (Bingel and Tracey 2008; decharms, et al. 2005; Ploghaus, et al. 1999; Wager, et al. 2004). If we anticipate pain, the noxious stimulus is felt more strongly than it really is and several brain areas related to pain are activated (Ploghaus, et al. 1999), whereas if we expect analgesia, the pain will be reduced (Christopher, et al. 2008). The belief of an individual that she or he is in control (coping resources are believed to be sufficient) of the painful situation gives rise to reappraisal (reevaluating the meaning of the painful situation) Reappraisal has been correlated with activation of the ventrolateral prefrontal cortex (VLPFC), an area related to cognitive evaluation of pain that interacts with the amygdala (Blair 2004; decharms, et al. 2005; Goldin, et al. 2008; Kringelbach and Rolls 2003). The best example of cognitive modulation of pain is placebo analgesia, which we will discuss more ahead Emotional modulation of pain Studies on the relationship between emotion and pain have shown that there is a clear affective modulation of pain at spinal and supraspinal levels (Meagher, et al. 2001; Rhudy, et al. 2007; Rhudy, et al. 2005; Rhudy, et al. 2006; Rhudy, et al. 2008; Williams and Rhudy 2009a; Williams and Rhudy 2009b). Emotional stress, anxiety, and negative affect decrease the pain threshold i.e. pain is felt with less stimulation, whereas positive affect increases the 23

26 Cognitive and emotional processing of music and its effect on pain pain threshold. According to the evidence, pleasant stimuli reduce pain, whereas unpleasant stimuli increase pain perception (Berna, et al. 2010; de Tommaso, et al. 2008; Rhudy, et al. 2008; Wiech and Tracey 2009). In imaging studies, empathy toward pain and social exclusion (social pain) activate brain areas related to pain (Cheng, et al. 2010; Singer, et al. 2004). Also, with verbal suggestions and environmental cues, it is possible to create pain in an individual without physical stimulation (Bayer, et al. 1991). There also seems to be a difference in pain perception related to gender (Fillingim, et al. 2009; Rhudy and Williams 2005). In general, emotional people may, in theory, be more affected by emotional stimuli during pain (Ploner, et al. 2010; Salovey and Birnbaum 1989; Taenzer, et al. 1986; Wakabayashi, et al. 2007) Placebo analgesia The best example of cognitive modulation of pain is placebo analgesia. Placebo analgesia refers to pain relief by means of the placebo effect (Benedetti and Amanzio 1997; Wager, et al. 2007). The placebo effect is a treatment effect caused not by the physical properties of the treatment, but by the meaning ascribed to it. The person expects pain relief from a certain treatment due to verbal suggestion and/or previous exposure, and as a result the person could experience an analgesic effect as strong as morphine by ingesting a placebo substance (i.e. a sugar pill ) (Benedetti and Amanzio 1997; Petrovic, et al. 2005). As the person assumes the placebo will somehow reduce the pain, she or he expects pain relief (expectation) and feels in control of the pain (reappraisal) (Wiech, et al. 2008a; Wiech, et al. 2008b). Recent studies show that during placebo analgesia there is increased activity in the PFC, OFC, PAG, and the lower pons, areas in which cognitive modulation of analgesia takes place, whereas activity decreases in the thalamus, IC, and ACC, areas related to pain unpleasantness (Petrovic, et al. 24

27 Cognitive and emotional processing of music and its effect on pain 2005; Wager, et al. 2004). According to some theories, placebo treatment affects the cognitive interpretation of pain, and rostral anterior cingulate cortex (racc) activation triggers endogenous opioid release in the brainstem, in other words, a top-down pain regulation (Bingel and Tracey 2008; Petrovic, et al. 2002). Figure5:(adaptedfromWiechetal.(2008))Possibleneuralpathwaysforcognitivemodulationofpain.I includedareaswhichmayberelatedtoemotionalmodulationsuchasofc,nucleusaccumbensandam. 25

28 Cognitive and emotional processing of music and its effect on pain 3.3 Music and pain A considerable number of studies support the notion that music reduces pain intensity and unpleasantness and increases pain tolerance. (Huang, et al. 2010; Klassen, et al. 2008; McCaffrey and Freeman 2003; Mitchell and MacDonald 2006; Mitchell, et al. 2006; Mitchell, et al. 2007; Nilsson 2008; Podder 2007; Roy, et al. 2008). Listening to music has been shown to increase the expression of mu opiate receptors, the place where opioids bind to produce analgesia (Stefano, et al. 2004). A meta-analysis that included several pain studies involving music as a therapy showed that there was a positive analgesic effect secondary to music listening in 59% of those studies (Nilsson 2008). Another recent study showed that music reduces pain by 18%, comparable to the analgesic effect of ibuprofen (Roy, et al. 2008). Furthermore, listening to music reduces the dosage of sedatives and analgesic medication in institutionalized patients, and benefits their overall well-being (Allred, et al. 2010; Browning 2000; Huang, et al. 2010; Klassen, et al. 2008; Laopaiboon, et al. 2009; Mitchell, et al. 2007; Nichols and Humenick 2000; Nilsson 2008; Podder 2007; Spiby, et al. 2003). The main mechanisms behind the analgesic effect of music are believed to be cognitive (Kreutz, et al. 2008; Mitchell, et al. 2006; Wiech, et al. 2008b) and emotional (Koelsch, et al. 2008; Roy, et al. 2009; Roy, et al. 2008). Nevertheless, most studies about music and pain have not fully controlled for confounders such as distractibility of the auditory stimulus, emotional elements, familiarity with the music, and the personality of the participants. Furthermore, the potential placebo analgesia of music has not been taken into consideration as possibly one of the main underlying analgesic mechanisms. 26

29 Cognitive and emotional processing of music and its effect on pain In paper II, we investigated the analgesic effects of music, attempting to control for as many confounders as possible. We compared a distraction task to passive music listening to control for distractibility. We investigated environmental Sounds and Mozart Music characterized by similar valence, arousal and liking. We explored the individual variation in analgesic effects of primary tasks due to gender, and cognitive styles. In paper III, we investigated the placebo effect of music by the use of a placebo stimulus (Sounds) and an active stimulus (Music). We investigated the influence of suggestion and emotion on the placebo effect of music-induced analgesia. 27

30 Cognitive and emotional processing of music and its effect on pain 28

31 Cognitive and emotional processing of music and its effect on pain 4. MATERIALS AND METHODS 4.1 Ethics All studies described in this dissertation followed the instructions, rules, and restrictions determined by The Danish National Committee on Biomedical Research Ethics. Ethical agreement was received from The Health Committee for Region Mid-Jutland in Denmark. The experiments were in accordance with the Declaration of Helsinki and the recommendations from the American Pain Society, the World Medical Association, the World Health Organization and the Council for International Organizations of Medical Sciences, the American College of Physicians, and the American Psychological Association. The ethics of research involving normal human volunteers or patients have been clarified by several international and national organizations. The following principles taken from the American Pain Society were followed in our experiments: 1. All planned clinical studies should be reviewed by an independent committee on human research prior to implementation. The committee, which should include scientists, healthcare practitioners, and lay persons, should evaluate the risks inherent in the research and the extent to which the significance of the potential results justifies the risks involved, even if minimal. The committee also should ensure that potential subjects have the opportunity to provide informed consent prior to participation. 2. Informed consent implies that potential subjects are fully apprised of the goals, procedures, and risks of the study. Potential subjects must be able to decline, and those who consent must be able to withdraw prior to completion, without any risk of penalty. 29

32 Cognitive and emotional processing of music and its effect on pain 3. Those who may be incapable of giving fully informed consent, including children and adults who are not fully competent, should not be used in pain research unless it is essential to the goals of the experiment. 4. In pain studies, the minimal intensity of noxious stimulation necessary to the goals of the experiment should be established beforehand and not exceeded during the study. Subjects should be able to stop the painful stimulus at will. 4.2 Choice of Methods Electroencephalography (EEG) is a method that measures the electrical activity of the brain produced by the neurons. To study the cognitive and emotional processing of music, we used EEG for the following reasons: a) EEG is a non-invasive method that records brain electric potentials from electrodes attached to the scalp. Its high temporal resolution makes possible to record electrical changes in milliseconds, optimal for studying events at specific time points. b) The EEG is attached to the participant, who can move almost freely, and it is noise free, an important feature when studying the auditory cortex. c) The ERPs such as the MMN and ERAN have been studied and described extensively. 30

33 Cognitive and emotional processing of music and its effect on pain Figure 6: Example of an EEG experiment. The participant wears a cap with electrodes which measures the electrical activity from the brain, while performs the experiment in front of a computer. A contact thermode is a device used for pain research in which a small area (i.e. 3 x 3 cm), usually in contact with the skin, is heated using controlled electrical current. The resulting effect is pain secondary to heat. To study the analgesic effects of music we used a thermode for the following reasons: a) A thermode is a standardized device used in pain studies to induce pain using temperatures high enough to elicit pain but low enough to not burn the skin. b) The thermode we used was specially made for this type of experiments and can be controlled by an external computer to automate the study and give more reliable results. c) Pain elicited by heat was preferred because the experiment required acute pain that could also be sustained over a long period of time. d) The same thermode can be used in future fmri studies. 31

34 Cognitive and emotional processing of music and its effect on pain Figure 7: Image of a typical contact thermode attached to the forearm of a participant. To study the cognitive, affective and analgesic effects of music, as well as their underlying emotional mechanisms we used behavioral ratings for the following reasons: a) The Visual Analog Scale (VAS) and the Likert scale are the most studied and reliable self-report method for measuring pain intensity and pain unpleasantness, as well as affective and cognitive responses to music. b) Numerical ratings have been extensively studied in psychology as reliable measures of self-reported emotions and emotions attached to auditory stimuli like music. 4.3 Study 1 EEG study of the cognitive and emotional processing of music Participants 32

35 Cognitive and emotional processing of music and its effect on pain Twenty four right-handed (10 male, 14 female) Finnish speaking non-musicians participated in the experiment. Mean age of 24.6 years (18 30). The 24 were included in the behavioral study, but only 15 in the EEG analysis due to recording issues. All participants had normal hearing and no musical expertise or explicit knowledge of music theory. They also reported having no audiologic, cognitive, neurologic, or linguistic (word finding, writing, reading, and speech production and comprehension) deficits Stimuli We generated piano chords lasting 600 ms each arranged into melodies. Each melody had seven chords. There were three types of melodies, depending on the type of musical violation it included: Standard (Std), Mistuned (MMN) and Neapolitan (ERAN). The Std melodies had no musical violations. The MMN and ERAN melodies had violations in chords 3, 5, or 7. The MMN included chords that had a mistuned fifth (increased 50 cents), whereas the ERAN included chords that had a musically incongruent composition. In total, there was one experimental condition for the Std chords (Std3, Std5, Std7 counted as one), three experimental conditions for the MMN chords (MMN3, MMN5, MMN7), and three for the ERAN chords (ERAN3, ERAN5, ERAN7). The melodies were arranged in a way to resemble real flowing music. During the EEG recordings, tonic chords played with an organ timbre were used as target stimuli for the primary task performed by the participants. The deviant organ chords were uniformly distributed among conditions and matched with the piano chords in all aspects except for timbre. The organ chord was presented twelve times in each condition, each time in a different key (approximately 8% of all cadences), and always replaced a tonic chord of the cadence. 33

36 Cognitive and emotional processing of music and its effect on pain Figure 8: Stimuli. T=Tonic, S=Subdominant, Sn=Neapolitan Subdominant, D=Dominant, T3= Inverted tonic, Tm=Mistuned tonic, Sm=Mistuned Subdominant. Top box: Musical notation of the chords use for the stimulus, only in the C key for illustration purposes. The arrows point to the mistuned note in Tm and Sm respectively. Bottom box: Illustration of the musical cadence. The left column shows the type of condition. The right columns show the 7 positions of the cadence and the key deviant chords (3,5,7) Procedure Each experimental condition was presented in twelve different keys and 144 times in total. Each cadence was presented 5 ms after the preceding cadence, preserving the musical meter, in order to give an impression of real, flowing music. Stimulus cadences in different keys and with different experimental manipulations were presented in a random order, while participants, wearing headphones, were sitting in a comfortable chair inside a soundproof 34

37 Cognitive and emotional processing of music and its effect on pain room. The stimuli were presented using the software Presentation 9.30 (Neurobehavioral Systems, Ltd.) at a volume of 50 db above the individual hearing threshold of each participant, determined at the beginning of the experiment. We used EEG to record the ERPs while the participants listened to the 144 melodies in random order. A primary task was included, so that the participants did not fully attend to the auditory stimuli (semi-attended experimental paradigm). The purpose of the task was to ensure that participants attended to the stimuli without asking him or her to attend to the MMN or ERAN chords. After the EEG recordings, the participants were instructed to listen to 28 melodies taken randomly from each of the seven experimental conditions and rate them in a 5-point Likert scale according to how pleasant did they found each melody, and how well did the chords fitted the melody Data acquisition and analysis EEG pre-processing and data analysis were performed with Brain Electric Source Analysis (BESA version5.2) (Berg and Scherg 1994; Miltner, et al. 1994). The scalp EEG was recorded with 128 active scalp electrodes fitted into a stretching cap and following the BioSemi ABC position system. Additionally, three active electrodes were placed on the participant s nose and mastoid areas, respectively, and four further ones around the eyes to monitor eye muscle activity. 24-bit EEG data were recorded with BioSemi ActiView 5.32 using no reference, with a sampling rate of 2048 Hz, and a recording bandwidth of up to 417 Hz. The EEG data for each participant were re-sampled offline for ERP analysis using a sampling rate of 256 Hz. We averaged the ERPs for each condition and cadence. After averaging, the ERP data were filtered with a low-pass filter of 40 Hz and re-referenced to the average of the mastoids. The event-related (ERP) data were then statistically evaluated using 35

38 Cognitive and emotional processing of music and its effect on pain repeated measures analyses of variance (ANOVAs) with the mean ERP amplitude as the dependent variable. The mean ERP amplitudes were calculated for each participant and condition from 40-ms time windows surrounding the grand-averaged peak at the Fz electrode (channel 21 in the Biosemi system). Then the mean amplitude data were analyzed using repeated measures analyses of variance (ANOVAs) with ERP Component (levels: ERAN, MMN) and Position (levels: 3, 5, 7) as within-participant factors. Furthermore, the mean amplitude values were computed for four regions of interest (ROIs): left anterior (C25, C26, C32, D3, D4), right anterior (C3, C4, C10, C12, C13), left posterior (A5, A6, A7, A8, A18), and right posterior (A31, A32, B3, B4, B5). The data were then analyzed using four-way repeated measures ANOVA with ERP Component (ERAN-MMN), Position (3, 5, 7), Distribution (anteriorposterior), and Hemisphere (right-left) as within-participant factors. For all statistical analyses, type I errors were controlled for by using Mauchly s test and the Greenhouse Geisser epsilon when appropriate. The alpha-level for all statistical analysis was.05, unless stated otherwise. After statistical evaluation, the grand-average ERP waveforms were filtered with a 10-Hz low-pass filter for illustration purposes only. Discrete source localization was performed with the 4-shell spherical head model using regional sources that were individually acquired, and subsequently analyzed using ANOVA with the Brain Electrical Source Analysis (BESA version 5.2) software (Berg and Scherg 1994). The coordinates of the resulting RS s were registered according to the Talairach coordinate system (Talairach and Tournoux 1988). Distributed source localization was performed with Statistical Parametric Mapping (SPM8) (Kiebel and Friston 2004a; Kiebel and Friston 2004b). As the locations of the MMN main sources are suggested to be located in 36

39 Cognitive and emotional processing of music and its effect on pain the superior temporal gyrus (STG) (Heschl s gyrus), and the ERAN sources are in inferior frontal gyrus (IFG), we performed small volume corrections (SVCs) for these locations using IFG and STG maps, included in the MarsBaR Toolbox (Brett 2002) as bounding masks. The behavioral variables were compared using non-parametric Friedman s ANOVA with the Exact test method. Post hoc tests were performed using non-parametric Wilcoxon tests with Bonferroni correction (significance level =.0167). 4.4 Study 2 Behavioral study of the cognitive and emotional mechanisms behind the analgesic effects of music Participants Forty-eight native Danish speakers, right-handed and non-musician took part in the experiment. They had not consumed any analgesic medication in the 24 h prior to the experiment. To have a balance sample, the participants were chosen according to cognitive style by first answering the Baron-Cohen Systemizer Quotient (Baron-Cohen 2009; Wakabayashi, et al. 2007) and a self-report of Emotional-Rational cognitive type. The Baron- Cohen questionnaire divides the population into three cognitive styles: Empathizers, Systemizers and Balanced. Empathizers are more empathic, Systemizers are interested in the analysis of systems and Balanced do not have any of these extremes. The self-report was a simple binomial question: Do you believe you are more emotional or rational?. Therefore, 37

40 Cognitive and emotional processing of music and its effect on pain we ended up with a balanced sample in terms of cognitive style and gender (16 males/16 females in each Baron-Cohen group and 24 males/ 24 f es in each self-report group) Stimuli The thermal stimuli were produced by a 3 x 3 cm contact thermode (Pathway model ATS from Medoc Ltd. Advanced Medical System, Israel) on the forearms. We investigated at which temperature each participant reported pain between mm (moderate to high) in the VAS, and this temperature was kept constant during the entire experiment to avoid a high variability of the scores. Each painful stimulation consisted of a plateau of 16 s with a rise/fall time of 2 s. The baseline temperature was 35 o C. Participants rated the thermal stimulus for pain intensity and unpleasantness by using the VAS. The auditory stimuli were six: four passive listening (two environmental Sounds (Rain and Water) and two Mozart string pieces ( String Quartet No. 1 in G major, K. 80/73f (1770) Adagio and Divertimento in E flat, K. 563 Adagio ) named Music1 and Music2 for simplicity), one active listening/distraction (the PASAT (Paced Auditory Serial Addition Test) (Gronwall 1977)), and one control (pink Noise). We made sure the Music was unfamiliar to the participants by choosing less popular compositions and by asking them at the end of the experiment if they knew the musical piece. Sounds and Music stimuli varied only slightly in affective valence, liking, and arousal, as determined with a previous pilot study. Each auditory stimulus lasted 300 s (5 min). 38

41 Cognitive and emotional processing of music and its effect on pain CONTROL - Noise DISTRACTION - PASAT SOUNDS { Rain Water MUSIC { Music1 Music2 Figure 9: Representation of the different auditory stimuli used in Study Procedure Each condition was named after the auditory stimuli they listened (Noise, PASAT, Rain, Water, Music1, Music2). The participants listened to each of the auditory stimuli while they received acute pain for a total run of six conditions in 30 minutes. This was repeated another time for a total of two runs. After each painful stimulus, they rated the pain for intensity and unpleasantness. At the end of the experiment, they were asked to listen to the auditory stimuli one more time and rate them for valence, liking, arousal, and fill out a questionnaire called 39

42 Cognitive and emotional processing of music and its effect on pain Perceived Emotional Intentions to rate different emotions perceived from the music like happiness or sadness. To investigate the placebo effect and the influence of verbal suggestion and emotional feelings for paper III, the participants were given certain suggestions about the effect of the Sounds and the Music before the experiment. The suggestions were different for each participant to balance the experiment, but it was always to suggest: a) one of the Sounds is better than the other to reduce pain, when this is a lie, and b) that one of the Music pieces was composed by Mozart, whereas the other was composed by Salieri. We also suggested them that Mozart Music had analgesic effects. In this way, we aimed to elicit placebo analgesia in Music by exploiting the popular belief that Mozart music is beneficial for the health and the brain, and at the same time, if one of the participants was familiar with the movie Amadeus they would recognize the name Salieri as the antagonist of Mozart. After the experiment, we asked the participants if they believed the suggestions. a pain rest pain rest pain rest pain 20 s 20 s 20 s 20 s 140 s s b AUDITORY STIMULUS AUDITORY + THERMAL STIMULUS x2 0 5 min 10 min 15 min 20 min 25 min 30 min Figure 10: Paradigm used in Study 2. Each condition (a) was formed by the auditory stimuli and the four painful stimuli. First the participants listened to the auditory stimulus passively for 140s. Then they received pain for 160s. The auditory stimulus was played for the whole 300s (5 min) of duration of the condition. The conditions were played randomly for the 30 minutes of one run (b). The experiment consisted of two runs with one minute of rest in between. 40

43 Cognitive and emotional processing of music and its effect on pain Data acquisition and analysis The participants rated pain intensity and unpleasantness using the Visual Analog Scale (VAS), a standardized subjective measure of pain. For the second paper (II), we analyzed the pain ratings, the emotional ratings and the cognitive styles in a Repeated Measures design to determine the effect and interaction of each of these last two on pain. We also analyzed the differences between each of the conditions (PASAT, Sounds and Music) to determine if there was an analgesic effect when compared to the control (Noise) and which of the conditions reduced pain the most. For the third paper (III) we analyzed only the Noise, Sounds and Music. To emulate a design analog to standard placebo studies, the Noise was the control condition, whereas the Sounds were the placebo condition (conditions that should not have any effect on the pain). Because Music was being studied as the analgesic drug, it was named the active condition (not to be confused with the PASAT, which was not analyzed for this paper). First, we investigated if there was placebo analgesia in the active condition by comparing the control condition versus the placebo condition. Second, we investigated if the active condition reduced pain more than the placebo condition by comparing them. Third, we analyzed the possible effect of suggestion and belief for the placebo and active condition. Fourth, we investigated if the placebo condition with the highest ratings of positive emotions differed from the condition with highest ratings of negative emotions. 41

44 Cognitive and emotional processing of music and its effect on pain 42

45 Cognitive and emotional processing of music and its effect on pain 5. RESULTS 5.1 Experiment Event-related potentials The Neapolitan chords violating the rules of harmony elicited the frontally distributed early right anterior negativity (ERAN) peaking on average at 228 ms post-stimulus. Mistuned chords violating the rules of musical-scale tuning elicited the fronto-centrally distributed MMN peaking on average at 270 ms post-stimulus. The ERAN was present in both hemispheres, whereas the MMN was lateralized to the right hemisphere. The contrast showed that the ERAN and the MMN differed in scalp distribution. Figure 11: Plots of the ERP amplitude. The ERAN shows amplitudes according to chord position, whereas the MMN does not Discrete and distributed source analysis 43

46 Cognitive and emotional processing of music and its effect on pain The ERAN was localized in the inferior frontal gyrus, and specifically in BA 44 and BA 45, corresponding to Broca s area and its right homolog, and the MMN was localized in Heschl s gyrus, and specifically in BA 41, corresponding to the primary auditory cortex. The statistical comparison of the ERAN and MMN localizations of the individual participants with repeated measures ANOVA revealed the ERAN was anterior to the MMN. In the distributed source localization, there were activations in the Heschl s gyrus, and specifically in BA 41 in both hemispheres for the MMN. For the ERAN, there were activations in the inferior frontal gyrus (IFG), and specifically in BA 44 and BA

47 Cognitive and emotional processing of music and its effect on pain Figure 12: The left column corresponds to the ERAN and the right column to the MMN. Inside each column there are two sub-columns. Left sub-columns: discrete source analysis of the ERAN and MMN responses for all experimental conditions. Right sub-columns: distributed (SPM) analysis of the ERAN and MMN responses for all experimental conditions. The first row shows the sagittal view, the second row shows the coronal view, and the third row shows the transversal view (A=Anterior, P=Posterior, L=Left, R=Right, IRS=Individual Regional Sources, MRS=Mean Regional Sources). The color bar represents the T-statistic. 45

48 Cognitive and emotional processing of music and its effect on pain Behavioral analysis The results show that the ERAN5 was consistently the most pleasant and fitting position in the Neapolitan melodies and the ERAN7 the least pleasant and fitting. The MMN3 was the most pleasant and fitting position amongst the mistuned melodies and the MMN7 the least pleasant and fitting. a 5 4 Mean Scores b Mean Scores Std E3 E5 E7 M3 M5 M7 Figure 13: Mean scores of Pleasantness (a) and Fittingness (b) behavioral ratings in the y axis, and each experimental condition in the x axis. (Std= Standard condition). 46

49 Cognitive and emotional processing of music and its effect on pain 5.2 Experiment Pain, emotion and cognition The PASAT (mental arithmetic), Sounds, and Music reduced pain more than Noise, however the PASAT reduced pain more than Sounds and Music. On the other hand, the Sounds and Music had the same analgesic effect, and the reason seems to be that they also had similar ratings of valence, liking, and arousal. a Intensity Unpleasantness c Intensity Unpleasantness Pain score (VAS) Pain score (VAS) Noise PASAT Sounds Music 0 Noise PASAT Rain Water Music1 Music2 b 10 d Valence Liking Arousal 9 8 Valence Liking Arousal 7 7 Emotion score Emotion score Noise PASAT Sounds Music 0 Noise PASAT Rain Water Music1 Music2 Figure 14: Mean pain (top row) and emotional (bottom row) ratings for each condition. Plots a and b represent the analysis of the averaged sounds and music pieces. Plots c and d represent the analysis of all the conditions. Error bars (SE). In the psychometric results, the Systemizers (both male and female) perceived less pain intensity during PASAT than the Balanced and Empathizers. Male Systemizers had lower 47

50 Cognitive and emotional processing of music and its effect on pain pain intensity ratings in the PASAT than the other cognitive styles, however they had higher pain unpleasantness ratings of all conditions except for PASAT. Female Empathizers reported higher pain intensity and pain unpleasantness ratings for all conditions except PASAT Pain intensity score Balanced Empathizers Systemizers Noise PASAT Rain Water Music1 Music2 100 Pain unpleasantness score Balanced Empathizers Systemizers Noise PASAT Rain Water Music1 Music2 Figure 15: Plots of the pain ratings of each condition by Systemizers, Empathizers and Balanced cognitive styles. 48

51 Cognitive and emotional processing of music and its effect on pain Placebo analgesia We found Music and Sounds induced a significant reduction in pain intensity and pain unpleasantness as compared to noise but music and sound did not differ in their pain relieving effect. These results suggest that the analgesic effect of music is partly due to placebo analgesia. Verbal suggestions for pain relief did not increase the magnitude of the placebo analgesia effect, not even for the participants who did believe the verbal suggestions for pain relief. However, positive emotions perceived in the placebo conditions influenced the analgesic effect by reducing pain. This finding suggests that it may not be the music per se that gives rise to the pain relief seen in music induced analgesia, and it suggests that placebo analgesia may be involved. 49

52 Cognitive and emotional processing of music and its effect on pain 50

53 Cognitive and emotional processing of music and its effect on pain 6. DISCUSSION 6.1 Cognitive responses to music and their relation to emotional perception In the first paper (I), we showed the neural representation of musical-scale and harmony rule violations, their distinct underlying neural generators, and the subjective perceptual attributes of those rule violations. We found that violations from harmonic regularities (i.e. Neapolitan chords) elicited an ERAN whose amplitude was modulated by the degree of violation of the rules of harmony. Mistuned chords violating the rules of the musical scale elicited an MMN whose amplitude did not differ according to chord context. The source localization based on both discrete and distributed algorithms gave comparable results by consistently locating the MMN around the auditory cortices and the ERAN around Broca s area and its right hemisphere homolog, even though the violations occurred in identical contexts. We also found that the melodies with the harmonic violation that showed the least error signal on EEG were rated the least unpleasant, whereas the violations that showed the highest error signal on EEG were rated very unpleasant. This suggests a close connection between cognitive and emotional perception in music. Several studies have showed that ERAN and MMN components are elicited by different musical properties of the chord cadence, i.e., scale vs. harmonic hierarchy (Koelsch, et al. 2001; Leino, et al. 2007; Ruiz, et al. 2009). Nevertheless, there are no previous studies that directly compare the MMN and ERAN localizations against each other within the same musical context and compare the cerebral sources. Hence, our results further support this distinction between ERAN and MMN components and contribute to knowledge on their cortical generators. 51

54 Cognitive and emotional processing of music and its effect on pain In this study, the findings of interest for us are the confirmation of distinct cognitive processing of different musical violations embedded in the same musical context, implying an influence of music on cognition. Also, there seems to be an influence of cognitive musical processing on emotion which is reflected by the perception of violations of harmony as unpleasant. We propose that the influence of music on cognitive and emotional processes goes beyond mere musical appreciation, and it probably affects the perception of other sensory systems. For example, Baumgartner et al. (2006) showed that listening to music congruent with an emotional visual stimuli enhances the emotional perception and experience of the visual stimuli. This could help us understand the mechanisms that induce analgesia by listening to music. 6.2 Influence of emotion and cognition in analgesia Cognitive modulation of pain In our second paper (II), we found that Mozart Music, environmental Sounds and mental arithmetic reduced pain. Moreover, the mental arithmetic reduced pain better than did Music and Sounds. These results are discrepant from those reported in other studies (Mitchell, et al. 2006), in which music had superior analgesic effects than did the mental arithmetic. However, the music in our study was not familiar to the participants, which could explain the results. Familiarity with the music could be more engaging, producing more distraction and perceived control, hence relieving from pain in a greater extent (Mitchell and MacDonald 52

55 Cognitive and emotional processing of music and its effect on pain 2006). The results also showed that individuals whose cognitive style was marked by stronger focus on patterns and systems (Systemizers), perceived less pain during mental arithmetic (Baron-Cohen 2009; Baron-Cohen, et al. 2003; Wakabayashi, et al. 2007). Thus, these individuals might find the mental arithmetic more interesting and distracting than others. The analgesic effect of the mental arithmetic is considered to reflect how distraction reduces pain (Tracey, et al. 2002; Villemure and Bushnell 2002). However, alternatively, mental arithmetic could reflect how distraction and stress induce analgesia (Butler and Finn 2009). Performing mental arithmetic while receiving and rating pain may provide enough stress to elicit analgesia as a survival mechanism. In sum, we suggest that the mechanisms responsible of the analgesic effects of mental arithmetic are distraction and stress, which in the present study outperformed the effects of Music and Sound Emotional modulation of pain In the second paper (II), we further showed that, although passive listening to Sounds and unfamiliar Music reduced pain compared to the control condition, the amount of pain reduction was not significantly different between the two listening conditions. Sounds and Music shared comparable levels of valence, liking and arousal. It may be that the perceived emotions of a stimulus play a bigger role than whether the stimulus consists of Music or Sounds. By themselves, feeling emotions influence the autonomic nervous system (Barrett, et al. 2007; Rainville, et al. 2005; Rhudy, et al. 2007) and modulate pain perception (Rhudy and Meagher 2001; Rhudy, et al. 2005; Wiech, et al. 2008a; Wiech and Tracey 2009; Williams and Rhudy 2009b). Several studies report that pleasant stimuli reduce pain, whereas 53

56 Cognitive and emotional processing of music and its effect on pain unpleasant stimuli increase pain (Berna, et al. 2010; de Tommaso, et al. 2008; Rhudy, et al. 2005; Rhudy, et al. 2006). Music elicits strong emotions (Blood and Zatorre 2001; Goldstein 1980) which can be pleasant or unpleasant depending on different factors such as the listener s personal preference, sound intensity, external context, etc (Juslin and Vastfjall 2008; Roy, et al. 2009). Besides the connection between valence and liking, music with low arousal is generally perceived as more pleasant (Green, et al. 2008). Several studies show that relaxing music reduces pain perception (Good, et al. 2005; Nilsson 2008). However, this similarity between the analgesic effects of environmental sounds and Mozart music could also be explained by the comparable unfamiliarity of the stimuli. Some studies suggest that self-chosen music has a significant impact on the amount of pain relief, possibly due to familiarity (Mitchell and MacDonald 2006; Mitchell, et al. 2006). Also, previous results showed that females with a more emotional cognitive style reported higher pain ratings than females with a more analytic cognitive style and males (Baron-Cohen 2009; Baron-Cohen, et al. 2003; Wakabayashi, et al. 2007). The most arousing condition had the highest analgesic effect. In the case of the mental arithmetic, the high ratings of arousal may be a result of the combination between the distraction it offers from the pain and the stress from the task. In the case of the passive stimuli (sounds and music), the low ratings of arousal may refer to the slow relaxing tempo. Although high ratings of arousal are present in the mental arithmetic condition, which resulted on higher pain relief, the analgesic mechanisms under active and passive conditions are different. Therefore, it cannot be stated that high ratings of arousal are related to better analgesic effect than low ratings of arousal. 54

57 Cognitive and emotional processing of music and its effect on pain Overall, these results suggest that emotional perception of the auditory stimuli greatly influences pain perception, rather than stimuli themselves. In other words, it is not the music, but the emotional value we attach to it. In summary, Mozart Music is no better at relieving pain than environmental Sounds when they are balanced in emotional connotations, familiarity, and when the cognitive style of the participants is also taken into account. The results also show that gender slightly influences the analgesic effect of music. Hence, in designing music for analgesic purposes these factors should be taken into consideration. However, it is noteworthy that an active distraction with mental arithmetic had superior analgesic effect than Music and Sounds. Future studies should aim at using neuroimaging methods such as fmri to investigate to further understand the neural mechanisms behind the analgesic effects of auditory stimuli Placebo analgesia of music The third paper (III) is the first experiment to investigate the possibility of a placebo effect in auditory stimuli and music. The results show that there might be placebo analgesia present in auditory stimuli and music. Moreover, we did not find an effect of suggestion and belief. We did find, however, an influence of positive emotions in the suggestion. It is not surprising that placebo analgesia could be present in the analgesic effect of music as it is present in every pain therapy. However, unlike other analgesic drugs, suggestion and belief did not seem to influence the participants. The reason could be that placebo analgesia with music is not mediated by cognitive processes, but mainly by emotional mechanisms 55

58 Cognitive and emotional processing of music and its effect on pain (Amanzio and Benedetti 1999; Chung, et al. 2007). So far we only mentioned the cognitive mechanisms of placebo analgesia, however, it can also be mediated by emotion. Imaging studies have described dopaminergic responses and activation of regions related to emotion, like the OFC, IC and racc, secondary to placebo analgesia (Petrovic, et al. 2005; Petrovic, et al. 2002; Scott, et al. 2007; Wager 2005; Wager, et al. 2007). Our results also suggests that the happy placebo had lower pain ratings based on suggestion than the sad placebo. Future imaging studies would help determine the neural correlates of the placebo mechanisms. Overall, we propose that one of the main mechanisms behind music-induced analgesia is placebo analgesia mediated by emotional processes. To sum up the results from Study 2, analgesia induced by music is a result of complex cognitive and emotional mechanisms, and individual cognitive styles. This, however, does not diminish the importance of music therapy for pain. On the contrary, placebo analgesia can have effects comparable to the strongest analgesic drug, morphine (Benedetti and Amanzio 1997). Therefore, it is imperative to determine the exact mechanisms of the analgesic effect of music to potentiate it and enhance its use for pain management. 56

59 Cognitive and emotional processing of music and its effect on pain 7. CONCLUSION Individuals without musical training are able to unconsciously detect simple and complex violations of musical rules. This is reflected by specific neural responses that are measurable with EEG. We found that the MMN is elicited by simple sensory violations located in auditory cortex, whereas the ERAN is elicited by complex violations of harmony rules and is located in the frontal cortex. These distinct localizations suggest cognitive processing of sensory input depending on the complexity of the stimulus structure, since the ERAN reflects violations of greater hierarchical complexity than the MMN. Moreover, we found that ratings of pleasantness for each melody representing different types of violations followed a similar pattern of the electrical activity elicited by each type of hierarchical violation. This suggests a relation between the cognitive processing of music and emotion, and the extent of the influence of music on other modules (Koelsch 2009). We further investigated the relation between cognitive and emotional processing of music in a study of the influence of music on pain perception. Cognitive processes such as mental arithmetic provided better pain relief than music. We found that, although environmental sounds and classical music by Mozart had an analgesic effect compared to a control noise condition, there was no difference on the amount of pain reduced by unfamiliar music when compared to sounds that share similar ratings of valence, liking and arousal. Although individual cognitive styles did have some influence on the pain ratings, it seemed to be minimal. We found that placebo analgesia is present in music induced analgesia. There was no influence of suggestion of analgesia even in people who believed it. However, when the placebo condition was rated with high positive emotions, the suggestion worked. These findings suggest that the analgesia effect of music may to a large extent be to complex 57

60 Cognitive and emotional processing of music and its effect on pain cognitive and emotional mechanisms. We propose that the main mechanism by which music induces analgesia are emotion, distraction and placebo analgesia. 58

61 Cognitive and emotional processing of music and its effect on pain 8. PERSPECTIVES Music research has become a growing field in Neuroscience, with topics ranging from the study of music itself, to the effects of music in other systems such as pain. The present studies have implications for notably different directions in music research. 8.1 Cognitive models of sensory input structure Simple local relationships, such as mistuning of a single chord in a chord sequence are mainly processed by the auditory cortex. However, with increases in complexity of hierarchical musical structures we see activation of the frontal lobe, a process reflected by the ERAN. Chord sequences follow certain patterns that are stored in long-term memory, and when the pattern is violated, a frontal error-signal in the brain reflected by the ERAN, is elicited. Thus, by studying musical structures with a variety of harmonical complexities it may be possible to understand how sensory input is cataloged. 8.2 Language and Music In music research, there is still an ongoing debate on the distinction between perceptual and cognitive processes related to music vs. language ignited by the discovery of harmony processing in Broca s area, a region thought to be specialized for syntax processing of language. On one side, the modularity view postulates the independence and separation of the 59

62 Cognitive and emotional processing of music and its effect on pain two processes (Peretz and Coltheart 2003), whereas on the other side, scholars argue for the commonality of neural mechanisms underlying the cognitive processes in the two domains (Koelsch and Sammler 2008; Patel 2003). This discovery raised many questions about the evolutionary origins of language and music, as well as the origins of the conceptual distinction between them. Harmonic structures have been suggested to be analog representations of language syntax, and vice versa. However, there is also the possibility of a shared neural network specialized not for the type of sensory input, but to patterns within the sensory input. Following this direction, research on music processing could expand the knowledge about language processing, and perhaps help determine the existence and workings of multimodal neural networks. Our finding contributes to the debate by highlighting once again the role of the inferior frontal gyrus specifically for hierarchical processing and not merely for any musical rule. Hence, it is likely that the nature of the rule processed rather than its domain (being it musical or linguistic) that determines the involvement of this brain structure in neural processing. 8.3 Predictive coding and music Sensory input is thought to be processed by specialized areas or modules which have been mapped using neuroimaging methods. However, brain mapping fails to show the overall brain function. The predictive coding theory was formulated by Karl Friston as a general theory of brain function as an attempt to understand how modules communicate with each other. In short, this theory states that specialized brain networks identify and categorize the 60

63 Cognitive and emotional processing of music and its effect on pain causes of their sensory inputs, integrates the information with other networks, and adapts to new stimuli by learning predictive patterns. Music research involving predictive sequences, such as demonstrated in our MMN /ERAN study, could be a way of testing the predictive coding theory for several reasons. Prediction or anticipation is fundamental in music, inducing positive or negative emotions depending on the perceived patterns derived from long- and short-term memory systems. Neural networks ruling musical expectation are thus shaped by culture, as well as by personal listening history, musical training and biology. Another reason for choosing music when testing the predictive coding theory is that music is primarily self-referential, unlike e.g. language, as it is possible to produce music without concrete meaning and connotations. Thus, it would be possible to study anticipation in a simpler form and study the related neural network. 8.4 Mechanisms of music analgesia Analgesic drugs must pass rigorous and long tests to determine their action mechanisms, their quantified analgesic effect and way of administration (e.g. oral). This aims to determine a therapeutic dosage and duration of treatment to avoid the unwanted secondary effects. Unlike analgesic drugs, music for analgesia has not been rigorously tested to determine its analgesic effect when compared to analgesic drugs, nor have its action mechanisms been completely described to determine a therapeutic dosage and avoid possible secondary effects. I believe the reasons for this are three-fold: 1) Music is thought to lack secondary effects, 2) finding 61

64 Cognitive and emotional processing of music and its effect on pain the correct drug-analog method is difficult, and 3) skepticism about the analgesic effects of music. Although it could be true that music used for analgesic purposes lacks secondary effects, it is naive to blindly believe this assertion, for it is as important to determine the dosage of music for pain, as it is for any other analgesic treatment. A better understanding of the underlying mechanisms for the analgesic effects of music will greatly improve its therapeutic use. To find well-suited experimental designs analog to that of analgesic drugs, it is necessary to explore the different types of standard designs and merge them with new ideas derived from music and affective neuroscience research experience, and cross-modal cooperation. The skepticism about the analgesic effects of music comes from the idea that is an alternative type of treatment, whose effect may be mediate purely by placebo analgesia. Although further research is necessary to answer this question, placebo analgesia itself reduces pain and should not be implied to be a lack of effect. The effect of every analgesic drug is produced by active and placebo analgesia. Therefore, even though it is imperative to determine the active analgesic effect of music, it is also crucial to determine its placebo effect. By systematically investigating the precise mechanisms by which listening to music reduces pain, we could greatly enhance the therapeutic use of music for pain from a clinical perspective. 8.5 Music composition for analgesia 62

65 Cognitive and emotional processing of music and its effect on pain From the research literature and our study, it is clear that certain elements of the music and the emotions the music evokes are important to elicit analgesia. Music that is highly pleasant, liked and that has low arousal has been shown to have analgesic effects. This means that music composed with a slow tempo, roughly bpm, in a cultural context such as the major key for most of the Western tonal culture, with a few added twists and thrilling musical surprises, could in theory have analgesic effects even if such a piece is not familiar to the listener. Therefore, I would venture to say that that further research of the neural mechanisms and specific musical structure that produces the analgesic effect of music will allow the creation of music specifically composed for analgesia. 8.6 Therapeutic indication for pain medicine Music has been studied under acute and chronic pain. Although the results have been positive, the neural pathways and mechanisms that produce acute and chronic pain are different. Acute pain is seen as a short sensory condition necessary for survival, whereas chronic pain is seen as a highly debilitating long-term illness. Hence, differentiating the characteristics and elements of the music that reduce pain in both acute and chronic conditions is important. It may not be possible to use the same type of music to reduce pain and the use of analgesic drugs during a tooth extraction or a bone marrow aspiration, than for patients with terminal bone cancer or phantom limb pain. Whether this holds or not, it may only be determined by clinical research on music analgesia once the basic foundations have been exhaustively tested and understood. 63

66 Cognitive and emotional processing of music and its effect on pain 64

67 Cognitive and emotional processing of music and its effect on pain 9. FUTURE STUDIES The present research raises several questions that we are currently studying and others that will be investigated, such as: These studies raised more questions about the underlying mechanisms of analgesia induced by music. It would be of interest to study Music and Sounds using imaging techniques such as fmri to determine the neural correlates of the analgesic effect of the Sounds compared to Music. In our Study 2, we found that active distraction (mental arithmetic) reduced pain more than all the other conditions investigated. Therefore, it is important to study distraction induced specifically by music to determine if and how it influences pain. This mechanism could be responsible for higher analgesic effects than we think. According to our results, there seems to be a placebo analgesia in music-induced analgesia. Never before has placebo analgesia being investigated in auditory stimuli until now. More studies using pharmacological and neuroimaging methods in patients with chronic pain and healthy participants are necessary to determine the mechanisms and areas related to placebo analgesia induced by music and other auditory stimuli in these particular pathological conditions, using pharmacological and imaging methods in healthy participants and patients with chronic pain To determine the effectiveness of analgesia induced by music, we could also compare the effects of music with standard analgesic drugs (e.g. ibuprofen) and different types of placebo 65

68 Cognitive and emotional processing of music and its effect on pain (placebo drug vs. placebo auditory stimuli). This would greatly increase knowledge about the analgesic effects of music and other auditory stimuli by a direct comparison with the standardize methods and designs in pain research. Furthermore, to investigate the electrophysiological activity of analgesia in the PAG of the brainstem associated to music listening, we have started experiments with patients suffering from chronic pain treated with a brain pacemaker (Deep Brain Stimulation). In these experiments we can measure local field potentials of the PAG while the patients listen to music and feel pain, in order to determine whether music analgesia acts at the brainstem level and how it influences the PAG activity. In a similar approach, we would like to investigate the electrophysiological activity of violations of complex harmonical structures in the brainstem. These neural responses to violations, reflected by the ERAN described in our study, could arise before the cortex suggesting a sub-cortical error-detection and musical syntax processing mechanism. This could potentially change many of the current assumptions that exist about music and language processing. 66

69 Cognitive and emotional processing of music and its effect on pain SUMMARY The Ph.D. is based on a Study 1 about the cognitive and emotional processing of musical structures, and Study 2 about the influence of cognitive and emotional responses to music on pain. The experiments were performed at the Cognitive Brain Research Unit, University of Helsinki, Helsinki, Finland, and the Center of Functionally Integrative Neuroscience, University of Aarhus, Aarhus, Denmark. In Study 1 we examined the event-related potentials recorded with electroencephalography in non-musicians. We hypothesized that cognitive processing of music follows a hierarchical structure derived from long-term memory, and violations of culture-dependent musical rules elicit specific event-related potentials whose amplitude is related to the hierarchical structure and reflected by reported emotional responses. Violations of simple hierarchical structures elicited an MMN localized in the temporal cortex, whereas violations of more complex hierarchical structures elicited an ERAN localized in the frontal cortex. We confirmed distinct cognitive processing of different musical violations embedded in the same musical context, implying an influence of music on cognition. In particular, mistuned chords elicited an MMN located in the temporal lobe whereas syntactically incongruous chords elicited an ERAN located in the frontal lobe. Also, we showed an influence of cognitive musical processing on emotion which is reflected by the perception of violations of harmony as unpleasant. In Study 2 investigated the analgesic effect of different auditory stimuli with similar emotional characteristics in participants with different cognitive styles, and the possibility of placebo analgesia in music. We hypothesized that active auditory distraction reduces pain 67

70 Cognitive and emotional processing of music and its effect on pain more than environmental Sounds and Mozart Music. Passive auditory stimuli with similar emotional characteristics have the same analgesic effect, which is also influenced by cognitive styles. Further, we hypothesize that an important part of the analgesic effect of music may be mediated by placebo analgesia, and these are influenced by happy and sad emotions attributed to the auditory stimuli. Our study showed that mental arithmetic is better than unfamiliar music to reduce pain. We confirmed that emotional perception of the auditory stimuli greatly influences pain perception, rather than stimuli themselves. We also found that there is placebo analgesia present in auditory stimuli and music. Suggestion and belief did not fully affect the pain ratings. However, placebo analgesia seems to be mediated mainly by emotion, as the placebo condition with perceived positive emotions influenced the effect of suggestion. Taken together, these studies show a close relation between cognitive and emotional processing of auditory stimuli. We propose that understanding this relation as a shared neural network could help explain why music has analgesic effects. We also suggest that the main mechanism by which music induces analgesia are emotion, distraction and placebo effect. Individual differences as deriving from cognitive style have, instead, a minor role in determining music-induced analgesia. This could mean that, regardless of cognitive style, everyone could potentially benefit from the analgesic effects of music. 68

71 Cognitive and emotional processing of music and its effect on pain ACKNOWLEDGEMENTS This thesis represents my work as a Ph.D. student at the Center of Functionally Integrative Neuroscience (CFIN), University of Aarhus, Denmark, which is also associated with the Royal Academy of Music, Aarhus, Denmark (DJM). The EEG experiment and measurements were conducted at the Cognitive Brain Research Unit, University of Helsinki, Finland. Part of this work was also done at the Hedonia Research Group, Department of Psychiatry, University of Oxford, UK. At the CFIN, I wish to thank Peter Vuust for his trust, guidance, and for giving me the necessary freedom to work on my projects which made this possible. Thanks to Elvira Brattico for her invaluable intelligence, supervision and support. To Leif Østergaard for his outstanding kindness and help throughout my Ph.D. To Mai Dustrup for her problem-solving skills and kindness. To Arne Møller for his willingness to help me at the slightest need. To Bjørn Petersen for being there when I needed him. To my co-workers on the experiments and articles, and the people that somehow had some impact on them Lene Vase, Kristjana Yr Jonsdottir, Kim Mouridsen, Joshua Skewes, Daniel Campbell-Meiklejohn, Mallar Chakravarty, Else-Marie Jegindø, Nana Brix Finnerup, and Chris Bailey. Finally, to all my colleagues at the MIB group and the CFIN for making this a very enjoyable experience. At the CBRU, I wish to thank Teija Kujala and Mari Tervaniemi for allowing me to use the EEG facilities, and to the rest of the CBRU staff. At the Hedonia Research Group, thanks to: Morten L. Kringelbach for his invaluable support, brilliant insight, and for sharing his football skills. To Christine Parsons, Katie Young, and Morten Jønsson for their selfless help and friendship that made my stay at the lab fun. 69

72 Cognitive and emotional processing of music and its effect on pain Mostly, however, I would like to thank the love of my life, my wife Laura Helena, for being there whenever I needed her, and for helping me through this and all my other somewhat crazy endeavors. I am very lucky to have someone by my side that is so loving, as well as stronger and more intelligent than me. You truly are a gift from life. 70

73 Cognitive and emotional processing of music and its effect on pain REFERENCES Allred KD, Byers JF, Sole ML. (2010): The effect of music on postoperative pain and anxiety. Pain ManagNurs11(1):1525. AmanzioM,BenedettiF.(1999):Neuropharmacologicaldissectionofplaceboanalgesia:expectation activatedopioidsystemsversusconditioningactivatedspecificsubsystems.jneurosci19(1): BaronCohenS.(2009):Autism:theempathizingsystemizing(ES)theory.AnnNYAcadSci1156: BaronCohenS,RichlerJ,BisaryaD,GurunathanN,WheelwrightS.(2003):Thesystemizingquotient: an investigation of adults with Asperger syndrome or highfunctioning autism, and normal sex differences.philostransrsoclondbbiolsci358(1430): BarrettFS,GrimmKJ,RobinsRW,WildschutT,SedikidesC,JanataP.(2010):Musicevokednostalgia: affect,memory,andpersonality.emotion10(3): BarrettLF,MesquitaB,OchsnerKN,GrossJJ.(2007):TheExperienceofEmotion.AnnualReviewof Psychology58(1): Baumgartner T, Lutz K, Schmidt CF, Jancke L. (2006): The emotional power of music: how music enhancesthefeelingofaffectivepictures.brainres1075(1): Bayer TL, Baer PE, Early C. (1991): Situational and psychophysiological factors in psychologically inducedpain.pain44(1):4550. BenedettiF,AmanzioM.(1997):Theneurobiologyofplaceboanalgesia:fromendogenousopioidsto cholecystokinin.progneurobiol52(2): BergP,SchergM. (1994):Afastmethodfor forwardcomputationofmultipleshellsphericalhead models.electroencephalogrclinneurophysiol90(1):5864. BernaC,LeknesS,HolmesEA,EdwardsRR,GoodwinGM,TraceyI.(2010):Inductionofdepressed mooddisruptsemotionregulationneurocircuitryandenhancespainunpleasantness.biolpsychiatry 67(11): BingelU,TraceyI.(2008):ImagingCNSmodulationofpaininhumans.Physiology(Bethesda)23: Blair RJ. (2004): The roles of orbital frontal cortex in the modulation of antisocial behavior. Brain Cogn55(1): BloodAJ,ZatorreRJ.(2001):Intenselypleasurableresponsestomusiccorrelatewithactivityinbrain regionsimplicatedinrewardandemotion.procnatlacadsciusa98(20): BloodAJ,ZatorreRJ,BermudezP,EvansAC.(1999):Emotionalresponsestopleasantandunpleasant musiccorrelatewithactivityinparalimbicbrainregions.natneurosci2(4):

74 Cognitive and emotional processing of music and its effect on pain BratticoE,JacobsenT.(2009):Subjectiveappraisalofmusic:neuroimagingevidence.AnnNYAcad Sci1169: Brett M, Anton, JL., Valabregue, R., Poline, JB. (2002): Region of interest analysis using an SPM toolbox [abstract]. Presented at the 8th International Conference on Functional Mapping of the HumanBrain,Sendai,Japan.16(2). BrooksJC,NurmikkoTJ,BimsonWE,SinghKD,RobertsN.(2002):fMRIofthermalpain:effectsof stimuluslateralityandattention.neuroimage15(2): BrowningCA.(2000):Usingmusicduringchildbirth.Birth27(4):2726. ButlerRK,FinnDP.(2009):Stressinducedanalgesia.ProgNeurobiol88(3): Cheng Y, Chen C, Lin CP, Chou KH, Decety J. (2010): Love hurts: an fmri study. Neuroimage 51(2):9239. Christopher AB, Ben S, Wael ED, Anthony KPJ. (2008): Confidence in beliefs about pain predicts expectancy effects on pain perception and anticipatory processing in right anterior insula. Pain 139(2): ChungSK,DonaldDP,VerneGN,MichaelER.(2007):Revelationofapersonalplaceboresponse:Its effectsonmood,attitudesandfutureplaceboresponding.pain132(3): Coghill RC, Sang CN, Maisog JM, Iadarola MJ. (1999): Pain intensity processing within the human brain:abilateral,distributedmechanism.jneurophysiol82(4): ConradC,NiessH,JauchKW,BrunsCJ,HartlWH,WelkerL.(2007):Overtureforgrowthhormone: Requiemforinterleukin6?CritCareMed. de Tommaso M, Sardaro M, Livrea P. (2008): Aesthetic value of paintings affects pain thresholds. ConsciousCogn17(4): decharmsrc, MaedaF,GloverGH,Ludlow D, PaulyJM, Soneji D,Gabrieli JD, Mackey SC.(2005): ControloverbrainactivationandpainlearnedbyusingrealtimefunctionalMRI.ProcNatlAcadSci USA102(51): DerbyshireSW,JonesAK,GyulaiF,ClarkS,TownsendD,FirestoneLL.(1997):Painprocessingduring threelevelsofnoxiousstimulationproducesdifferentialpatternsofcentralactivity.pain73(3): EversS,SuhrB.(2000):Changesoftheneurotransmitterserotoninbutnotofhormonesduringshort timemusicperception.eurarchpsychiatryclinneurosci250(3):1447. Fields HL. (2000): Pain modulation: expectation, opioid analgesia and virtual pain. Prog Brain Res 122: FillingimRB,KingCD,RibeiroDasilvaMC,RahimWilliamsB,RileyJL,3rd.(2009):Sex,gender,and pain:areviewofrecentclinicalandexperimentalfindings.jpain10(5):

75 Cognitive and emotional processing of music and its effect on pain Fritz T, Jentschke S, Gosselin N, Sammler D, Peretz I, Turner R, Friederici AD, Koelsch S. (2009): Universalrecognitionofthreebasicemotionsinmusic.CurrBiol19(7):5736. GarzaVillarrealE,BratticoE,VaseL,OstergaardL,VuustP.(submitted). GoldinPR,McRaeK,RamelW,GrossJJ.(2008):Theneuralbasesofemotionregulation:reappraisal andsuppressionofnegativeemotion.biolpsychiatry63(6): GoldsteinA.(1980):Thrillsinresponsetomusicandotherstimuli.PhysiologicalPsychology8(1): GoodM,AndersonGC,AhnS,CongX,StantonHicksM.(2005):Relaxationandmusicreducepain followingintestinalsurgery.researchinnursing&health28(3): GreenAC,BaerentsenKB,StodkildeJorgensenH,WallentinM,RoepstorffA,VuustP.(2008):Music inminoractivateslimbicstructures:arelationshipwithdissonance?neuroreport19(7):7115. GronwallDM.(1977):Pacedauditoryserialadditiontask:ameasureofrecoveryfromconcussion. PerceptMotSkills44(2): HallDA,HartHC,JohnsrudeIS.(2003):Relationshipsbetweenhumanauditorycorticalstructureand function.audiolneurootol8(1):118. HuangST,Good M,ZauszniewskiJA. (2010): The effectiveness ofmusicinrelieving pain incancer patients:arandomizedcontrolledtrial.intjnursstud47(11): JanataP.(2009):Theneuralarchitectureofmusicevokedautobiographicalmemories.CerebCortex 19(11): JensenTS.(1997):Opioidsinthebrain:supraspinalmechanismsinpaincontrol.ActaAnaesthesiol Scand41(1Pt2): JensenTS,SindrupSH.(2002):Opioids:awaytocontrolcentralpain?Neurology58(4):5178. JentschkeS,KoelschS.(2009):Musicaltrainingmodulatesthedevelopmentofsyntaxprocessingin children.neuroimage47(2): Juslin PN, Vastfjall D. (2008): Emotional responses to music: the need to consider underlying mechanisms.behavbrainsci31(5):55975;discussion Kiebel SJ, Friston KJ. (2004a): Statistical parametric mapping for eventrelated potentials (II): a hierarchicaltemporalmodel.neuroimage22(2): KiebelSJ,FristonKJ.(2004b):Statisticalparametricmappingforeventrelatedpotentials:I.Generic considerations.neuroimage22(2): KlassenJA,LiangY,TjosvoldL,KlassenTP,HartlingL.(2008):Musicforpainandanxietyinchildren undergoingmedicalprocedures:asystematicreviewofrandomizedcontrolledtrials.ambulpediatr 8(2):

76 Cognitive and emotional processing of music and its effect on pain Koelsch S. (2005): Neural substrates of processing syntax and semantics in music. Curr Opin Neurobiol15(2): Koelsch S. (2006): Significance of Broca's area and ventral premotor cortex for musicsyntactic processing.cortex42(4): Koelsch S. (2009): Musicsyntactic processing and auditory memory: similarities and differences betweeneranandmmn.psychophysiology46(1): Koelsch S, Fritz T, Schlaug G. (2008): Amygdala activity can be modulated by unexpected chord functionsduringmusiclistening.neuroreport19(18): Koelsch S, Gunter TC, Schroger E, Tervaniemi M, Sammler D, Friederici AD. (2001): Differentiating ERANandMMN:anERPstudy.Neuroreport12(7): Koelsch S, Gunter TC, Wittfoth M, Sammler D. (2005): Interaction between syntax processing in languageandinmusic:anerpstudy.jcognneurosci17(10): Koelsch S, Heinke W, Sammler D, Olthoff D. (2006): Auditory processing during deep propofol sedationandrecoveryfromunconsciousness.clinneurophysiol117(8): KoelschS,SammlerD.(2008):Cognitivecomponentsofregularityprocessingintheauditorydomain. PLoSONE3(7):e2650. KongJ,WhiteNS,KwongKK,VangelMG,RosmanIS,GracelyRH,GollubRL.(2006):UsingfMRIto dissociate sensory encoding from cognitive evaluation of heat pain intensity. Hum Brain Mapp 27(9): Kreutz G, Schubert E, Mitchell LA. (2008): Cognitive Styles of Music Listening. Music Perception 26(1):5773. Kringelbach ML,RollsET.(2003): Neural correlates ofrapidreversal learningin asimple modelof humansocialinteraction.neuroimage20(2): LangnerG.(1992):Periodicitycodingintheauditorysystem.HearRes60(2): LaopaiboonM,LumbiganonP,MartisR,VatanasaptP,SomjaivongB.(2009):Musicduringcaesarean sectionunderregionalanaesthesiaforimprovingmaternalandinfantoutcomes.cochranedatabase SystRev(2):CD LeinoS,BratticoE,TervaniemiM,VuustP.(2007):Representationofharmonyrulesinthehuman brain:furtherevidencefromeventrelatedpotentials.brainres1142: MaessB,KoelschS,GunterTC,FriedericiAD.(2001):MusicalsyntaxisprocessedinBroca'sarea:an MEGstudy.NatNeurosci4(5):5405. McCaffreyR,FreemanE.(2003):Effectofmusiconchronicosteoarthritispaininolderpeople.JAdv Nurs44(5): MeagherMW,ArnauRC,RhudyJL.(2001):Painandemotion:effectsofaffectivepicturemodulation. PsychosomMed63(1):

77 Cognitive and emotional processing of music and its effect on pain MenonV,LevitinDJ.(2005):Therewardsofmusiclistening:responseandphysiologicalconnectivity ofthemesolimbicsystem.neuroimage28(1): MiltnerW,BraunC,JohnsonR,Jr.,SimpsonGV,RuchkinDS.(1994):Atestofbrainelectricalsource analysis(besa):asimulationstudy.electroencephalogrclinneurophysiol91(4): MirandaRA,UllmanMT.(2007):Doubledissociationbetweenrulesandmemoryinmusic:anevent relatedpotentialstudy.neuroimage38(2): MironD,DuncanGH,BushnellMC.(1989):Effectsofattentionontheintensityandunpleasantness ofthermalpain.pain39(3): Mitchell LA, MacDonald RA. (2006): An experimental investigation of the effects of preferred and relaxingmusiclisteningonpainperception.jmusicther43(4): Mitchell LA, MacDonald RA, Brodie EE. (2006): A comparison of the effects of preferred music, arithmeticandhumouroncoldpressorpain.eurjpain10(4): MitchellLA,MacDonaldRAR,KnussenC,SerpellMG.(2007):Asurveyinvestigationoftheeffectsof musiclisteningonchronicpain.psychologyofmusic35(1):3757. MunroS,MountB.(1978):Musictherapyinpalliativecare.CanMedAssocJ119(9): NäätänenR.(1995):Themismatchnegativity:apowerfultoolforcognitiveneuroscience.EarHear 16(1):618. NicholsFH,HumenickSS.2000.Childbirtheducation:practice,researchandtheory.Philadelphia; London:Saunders.xxviii,731p.p. NilssonU.(2008):Theanxietyandpainreducingeffectsofmusicinterventions:asystematicreview. AORNJ87(4): Paavilainen P, Arajarvi P, Takegata R. (2007): Preattentive detection of nonsalient contingencies betweenauditoryfeatures.neuroreport18(2): PatelAD.(2003):Language,music,syntaxandthebrain.NatNeurosci6(7): PeretzI,ColtheartM.(2003):Modularityofmusicprocessing.NatNeurosci6(7): PetrovicP,DietrichT,FranssonP,AnderssonJ,CarlssonK,IngvarM.(2005):Placeboinemotional processinginduced expectations of anxiety relief activate a generalized modulatory network. Neuron46(6): PetrovicP,KalsoE,PeterssonKM,IngvarM.(2002):Placeboandopioidanalgesiaimagingashared neuronalnetwork.science295(5560): PicklesJO.2008.Anintroductiontothephysiologyofhearing.UnitedKingdom:Emerald.xxi,410p. p. 75

78 Cognitive and emotional processing of music and its effect on pain PlatelH,PriceC,BaronJC,WiseR,LambertJ,FrackowiakRS,LechevalierB,EustacheF.(1997):The structuralcomponentsofmusicperception.afunctionalanatomicalstudy.brain120(pt2): PloghausA,TraceyI,GatiJS,ClareS,MenonRS,MatthewsPM,RawlinsJN.(1999):Dissociatingpain fromitsanticipationinthehumanbrain.science284(5422): Ploner M, Lee MC, Wiech K, Bingel U, Tracey I. (2010): Prestimulus functional connectivity determinespainperceptioninhumans.procnatlacadsciusa107(1): Podder L. (2007): Effects of music therapy on anxiety levels and pain perception. Nurs J India 98(7):161. Porro CA, Cettolo V, Francescato MP, Baraldi P. (1998): Temporal and intensity coding of pain in humancortex.jneurophysiol80(6): PriceDD.(2000):Psychologicalandneuralmechanismsoftheaffectivedimensionofpain.Science 288(5472): Quoniam N, Ergis AM, Fossati P, Peretz I, Samson S, Sarazin M, Allilaire JF. (2003): Implicit and explicit emotional memory for melodies in Alzheimer's disease and depression. Ann N Y Acad Sci 999:3814. Rainville P, Bao QV, Chretien P. (2005): Painrelated emotions modulate experimental pain perceptionandautonomicresponses.pain118(3): RauscheckerJP.(1998):Corticalprocessingofcomplexsounds.CurrOpinNeurobiol8(4): RauscheckerJP.(2001):Corticalplasticityandmusic.AnnNYAcadSci930:3306. RhudyJL,McCabeKM,WilliamsAE.(2007):Affectivemodulationofautonomicreactionstonoxious stimulation.intjpsychophysiol63(1):1059. Rhudy JL, Meagher MW. (2001): The role of emotion in pain modulation. Current Opinion in Psychiatry14(3): Rhudy JL, Williams AE. (2005): Gender differences in pain: do emotions play a role? Gend Med 2(4): Rhudy JL, Williams AE, McCabe KM, Nguyen MA, Rambo P. (2005): Affective modulation of nociceptionatspinalandsupraspinallevels.psychophysiology42(5): Rhudy JL, Williams AE, McCabe KM, Rambo PL, Russell JL. (2006): Emotional modulation of spinal nociceptionandpain:theimpactofpredictablenoxiousstimulation.pain126(13): RhudyJL,WilliamsAE,McCabeKM,RussellJL,MaynardLJ.(2008):Emotionalcontrolofnociceptive reactions(econ):doaffectivevalenceandarousalplayarole?pain136(3): Roy M, Mailhot JP, Gosselin N, Paquette S, Peretz I. (2009): Modulation of the startle reflex by pleasantandunpleasantmusic.intjpsychophysiol71(1):

79 Cognitive and emotional processing of music and its effect on pain RoyM,PeretzI,RainvilleP.(2008):Emotionalvalencecontributestomusicinducedanalgesia.Pain 134(12):1407. RuizMH,KoelschS,BhattacharyaJ.(2009):Decreaseinearlyrightalphabandphasesynchronization andlategammabandoscillationsinprocessingsyntaxinmusic.humbrainmapp30(4): SaloveyP,BirnbaumD.(1989):Influenceofmoodonhealthrelevantcognitions.JPersSocPsychol 57(3): Schlaug G, Jancke L, Huang Y, Staiger JF, Steinmetz H. (1995a): Increased corpus callosum size in musicians.neuropsychologia33(8): SchlaugG,JanckeL,HuangY,SteinmetzH.(1995b):Invivoevidenceofstructuralbrainasymmetryin musicians.science267(5198): Schon D, Rosenkranz M, Regelsberger J, Dahme B, Buchel C, von Leupoldt A. (2008): Reduced perception of dyspnea and pain after right insular cortex lesions. Am J Respir Crit Care Med 178(11): ScottDJ,StohlerCS,EgnatukCM,WangH,KoeppeRA,ZubietaJK.(2007):Individualdifferencesin rewardrespondingexplainplaceboinducedexpectationsandeffects.neuron55(2): Scott SK, Johnsrude IS. (2003): The neuroanatomical and functional organization of speech perception.trendsneurosci26(2):1007. ShabanloeiR,GolchinM,EsfahaniA,DolatkhahR,RasoulianM.(2010):Effectsofmusictherapyon painandanxietyinpatientsundergoingbonemarrowbiopsyandaspiration.aornj91(6): SingerT,SeymourB,O'DohertyJ,KaubeH,DolanRJ,FrithCD.(2004):Empathyforpaininvolvesthe affectivebutnotsensorycomponentsofpain.science303(5661): SpibyH, SladeP,EscottD,HendersonB,FraserRB.(2003):Selectedcopingstrategiesinlabor:an investigationofwomen'sexperiences.birth30(3): StefanoGB,ZhuW,CadetP,SalamonE,MantioneKJ.(2004):Musicaltersconstitutivelyexpressed opiateandcytokineprocessesinlisteners.medscimonit10(6):ms1827. Taenzer P, Melzack R, Jeans ME. (1986): Influence of psychological factors on postoperative pain, moodandanalgesicrequirements.pain24(3): Talairach J, Tournoux P Coplanar stereotaxic atlas of the human brain : 3dimensional proportionalsystem:anapproachtomedicalcerebralimaging.stuttgart NewYork:Thieme; ThiemeMedical.viii,122p.p. TervaniemiM,HugdahlK.(2003):Lateralizationofauditorycortexfunctions.BrainResBrainResRev 43(3): TervaniemiM,MedvedevSV,AlhoK,PakhomovSV,RoudasMS,VanZuijenTL,NaatanenR.(2000): Lateralizedautomaticauditoryprocessingofphoneticversusmusicalinformation:aPETstudy.Hum BrainMapp10(2):

80 Cognitive and emotional processing of music and its effect on pain TraceyI,PloghausA,GatiJS,ClareS,SmithS,MenonRS,MatthewsPM.(2002):Imagingattentional modulationofpainintheperiaqueductalgrayinhumans.jneurosci22(7): Trehub SE,Trainor LJ,UnykAM. (1993): Music andspeech processing inthe first year of life.adv ChildDevBehav24:135. Villemure C, Bushnell MC. (2002): Cognitive modulation of pain: how do attention and emotion influencepainprocessing?pain95(3):1959. Villemure C, Bushnell MC. (2009): Mood influences supraspinal pain processing separately from attention.jneurosci29(3): Vrana SR, Spence EL, Lang PJ. (1988): The startle probe response: a new measure of emotion? J AbnormPsychol97(4): WagerTD.(2005):Expectationsandanxietyasmediatorsofplaceboeffectsinpain.Pain115(3): Wager TD,Rilling JK,SmithEE, SokolikA,CaseyKL, DavidsonRJ,KosslynSM,Rose RM, Cohen JD. (2004): Placeboinduced changes in FMRI in the anticipation and experience of pain. Science 303(5661): Wager TD, Scott DJ, Zubieta JK. (2007): Placebo effects on human muopioid activity during pain. ProcNatlAcadSciUSA104(26): Wakabayashi A, BaronCohen S, Uchiyama T, Yoshida Y, Kuroda M, Wheelwright S. (2007): Empathizingandsystemizinginadultswithandwithoutautismspectrumconditions:crosscultural stability.jautismdevdisord37(10): Walker DL, Cassella JV, Lee Y, De Lima TC, Davis M. (1997): Opposing roles of the amygdala and dorsolateralperiaqueductalgrayinfearpotentiatedstartle.neuroscibiobehavrev21(6): Wiech K, Farias M, Kahane G, Shackel N, Tiede W, Tracey I. (2008a): An fmri study measuring analgesiaenhancedbyreligionasabeliefsystem.pain139(2): WiechK,PlonerM,TraceyI.(2008b):Neurocognitiveaspectsofpainperception.TrendsinCognitive Sciences12(8): WiechK,TraceyI.(2009):Theinfluenceofnegativeemotionsonpain:behavioraleffectsandneural mechanisms.neuroimage47(3): WilliamsAE,RhudyJL.(2009a):Emotionalmodulationofautonomicresponsestopainfultrigeminal stimulation.intjpsychophysiol71(3):2427. Williams AE, Rhudy JL. (2009b): Supraspinal modulation of trigeminal nociception and pain. Headache49(5): ZatorreRJ.(2001):Neuralspecializationsfortonalprocessing.AnnNYAcadSci930:

81 Cognitive and emotional processing of music and its effect on pain ZatorreRJ,BelinP,PenhuneVB.(2002):Structureandfunctionofauditorycortex:musicandspeech. TrendsCognSci6(1):3746. ZatorreRJ,ChenJL,PenhuneVB.(2007):Whenthebrainplaysmusic:auditorymotorinteractionsin musicperceptionandproduction.natrevneurosci8(7):

82 Cognitive and emotional processing of music and its effect on pain 80

83 Cognitive and emotional processing of music and its effect on pain APPENDICES Paper I Paper II Paper III 81

84

85 Elsevier Editorial System(tm) for Brain Research Manuscript Draft Manuscript Number: BRES-D R2 Title: Distinct neural responses to chord violations: A multiple source analysis study. Article Type: Research Report Section/Category: Cognitive and Behavioral Neuroscience Keywords: ERP; EEG; Music; Source Analysis; MMN; ERAN Corresponding Author: Dr. Eduardo Adrian Garza Villarreal, M.D. Corresponding Author's Institution: University of Aarhus First Author: Eduardo Adrian Garza Villarreal, M.D. Order of Authors: Eduardo Adrian Garza Villarreal, M.D.; Elvira Brattico, PhD; Sakari Leino, MSc; Leif Østergaard, MD, MSc, PhD, DMSc; Peter Vuust, PhD Manuscript Region of Origin: DENMARK Abstract: The human brain is constantly predicting the auditory environment by representing sequential similarities and extracting temporal regularities. It has been proposed that simple auditory regularities are extracted at lower stations of the auditory cortex and more complex ones at other brain regions, such as the prefrontal cortex. Deviations from auditory regularities elicit a family of early negative electric potentials distributed over the frontal regions of the scalp. In this study we wished to disentangle the brain processes associated with sequential vs. hierarchical auditory regularities in a musical context by studying the event-related potentials (ERPs), the behavioral responses to violations of these regularities, and the localization of the underlying ERP generators using two different source analysis algorithms. To this aim, participants listened to musical cadences constituted by seven chords, each containing either harmonically congruous chords, harmonically incongruous chords, or harmonically congruous but mistuned chords. EEG was recorded and multiple source analysis was performed. Incongruous chords violating the rules of harmony elicited a bilateral ERAN, whereas mistuned chords within chord sequences elicited a right-lateralized MMN. We found that the dominant cortical sources for the ERAN were localized around Broca's area and its right homologue, whereas the MMN generators were localized around primary auditory cortex. These findings suggest a predominant role of the auditory cortices in detecting sequential scale regularities and the posterior prefrontal cortex in parsing hierarchical regularities in music.

86 *Manuscript Click here to view linked References 1 Distinct neural responses to chord violations: A multiple source analysis study. Eduardo A. Garza Villarreal a,b,*, Elvira Brattico c, Sakari Leino c, Leif Østergaard a,d, Peter Vuust a,b a Center of Functionally Integrative Neuroscience, University of Aarhus, Denmark. (Danish Neuroscience Center, Aarhus University Hospital, Nørrebrogade 44 Building 10G-6, 8000) b Royal Academy of Music, Aarhus, Denmark. (Skovgaardsgade 2C, DK-8000) c Cognitive Brain Research Unit, Institute of Behavioral Sciences, University of Helsinki and Center of Excellence in Interdisciplinary Music Research, University of Jyväskylä, Finland. (PO Box 9, Siltavuorenpenger 1 B, FI-00014). d Department of Neuroradiology, Aarhus University Hospital, Denmark. ( Nørrebrogade 44 Building 10, 8000) *Corresponding Author: Eduardo Adrian Garza Villarreal Center of Functionally Integrative Neuroscience, Danish Neuroscience Center, Aarhus University Hospital, Nørrebrogade 44 Building 10G , Aarhus C, Denmark Mobile: Work: Fax: URL: [email protected], [email protected] 7

87 2 Abstract The human brain is constantly predicting the auditory environment by representing sequential similarities and extracting temporal regularities. It has been proposed that simple auditory regularities are extracted at lower stations of the auditory cortex and more complex ones at other brain regions, such as the prefrontal cortex. Deviations from auditory regularities elicit a family of early negative electric potentials distributed over the frontal regions of the scalp. In this study we wished to disentangle the brain processes associated with sequential vs. hierarchical auditory regularities in a musical context by studying the event-related potentials (ERPs), the behavioral responses to violations of these regularities, and the localization of the underlying ERP generators using two different source analysis algorithms. To this aim, participants listened to musical cadences constituted by seven chords, each containing either harmonically congruous chords, harmonically incongruous chords, or harmonically congruous but mistuned chords. EEG was recorded and multiple source analysis was performed. Incongruous chords violating the rules of harmony elicited a bilateral ERAN, whereas mistuned chords within chord sequences elicited a right-lateralized MMN. We found that the dominant cortical sources for the ERAN were localized around Broca s area and its right homologue, whereas the MMN generators were localized around primary auditory cortex. These findings suggest a predominant role of the auditory cortices in detecting sequential scale regularities and the posterior prefrontal cortex in parsing hierarchical regularities in music. Keywords: ERP; EEG; Music; Source Analysis; MMN; Harmony. Abbreviations: ERAN = Early right anterior negativity; Sn = Neapolitan subdominant; DD = Double dominant chord; T = Tonic position; BESA = Brain electric source analysis; RS = Regional sources; SVC = Small volume correction; ROI = Region of interest; SPM = Statistical Parametric Mapping.

88 3 1. Introduction The human brain is able to extract regularities and form hierarchic structures from the auditory environment. These regularities may consist of repetitions of one or more features contained in the sounds, or in the rules of succession of particular sound features, e.g., the higher the pitch, the louder the sound intensity (Paavilainen et al., 2007). Sometimes the detection of sequential regularities (and the deviations from them) during the course of an experiment may be reinforced by their long-term neural representations existing in the auditory cortex, as in the case of phonemes or musical-scale pitches or (Brattico et al., 2006; Näätänen et al., 1997; for a review see Näätänen et al., 2001). Other kinds of regularities are hierarchically organized, meaning events within those regularities have different weights and roles according to previous knowledge, e.g., music harmony or language syntax (Koelsch and Sammler, 2008). The brain processing of regularities can be studied with EEG by eliciting their corresponding event-related potentials (ERP): mismatch negativity (MMN) and early right anterior negativity (ERAN). According to Koelsch et al. (2009), the MMN is elicited by a deviant stimulus introduced in a sequence in which an acoustic or local abstract feature is repeated and a local rule is established in the course of the experimental session. The ERAN on the other hand, is elicited by culturally-biased errors in harmony within a melody, depending on long-term knowledge of a musical culture. However, the MMN is similarly modulated when sounds correspond to previous culturally-biased regularities established by exposure to a certain music or language. Hence, the main difference between the phenomena associated with ERAN and MMN resides in the locality vs. the hierarchical structure of the sound regularities. The way in which the brain deals with these two processes in the musical domain, has so far gone largely unrevealed. We believe that music is ideal for the study of the neural processing of these types of regularities. In this study, we aimed at investigating the neural mechanisms responsible for detecting musical-scale pitch and harmony errors within a balanced musical context, to determine their brain sources, and the relationship between the brain activity and the perception of these errors.

89 4 In the past decade, extensive research has focused on the extraction of simple regularities from auditory stimulation by using the mismatch negativity (MMN), a type of even-related potential (ERP) component elicited by a deviant stimulus randomly introduced in a train of repetitive stimuli (Näätänen, 1995; Picton et al., 2000). The MMN is a fronto-central negative potential peaking at around ms, occurring at early stages of auditory processing regardless of attention and hypothesized to be important for survival by detecting unexpected events (Näätänen, 1995; Näätänen et al., 2007). It has main sources in the superior temporal gyrus, and sometimes with weak additional sources in the right inferior frontal gyrus (Opitz et al., 2002; Rinne et al., 2005; Tse and Penney, 2008), and the inferior parietal lobule (Park et al., 2002). The MMN is suggested to reflect the automatic formation of brief neural models of sequential regularities in the auditory environment that are typically formed in the course of the experimental session by repeating acoustic or abstract features of one or several sounds (Näätänen et al., 2007; Winkler et al., 1996). For instance, an abstract feature MMN has been elicited by the violation of the rule of ascending sound pairs coupled with increasing loudness (Näätänen et al., 2001; Paavilainen et al., 2001) Music follows culture-dependent rules that govern the way we perceive and conceive it, with pitch as a central dimension. From the set of all the possible sound pitches, Western tonal music is founded on a small subset of 12 pitches included in the chromatic equal-tempered scale, where the interval between consecutive pitches is a semitone (100 cents). With very few exceptions such as the blue notes of the blues scale, this discrete selection of pitches and their relationship to each other determines the most basic and fundamental rules of Western tonal music, be it classical, jazz or pop musical genres. The rules of the equal-tempered scale concern sequential aspects of sound pitches, as they can be extracted by comparing the pitch of the incoming sound with that of the immediately preceding one, with a human sensitivity to pitch deviations of 10 to 30 cents (Krumhansl, 2000; Lehmann, 2008). Violations of these scale rules in unfamiliar melodies also elicit an MMN-like response with main generators in the non-primary auditory cortex (Brattico et al., 2006). These findings suggest that the MMN is also elicited

90 5 by violations of musical regularities that are stored in long-term memory rather than only those formed during the course of the experimental session. Most pieces in Western tonal music are based on smaller subsets from the chromatic scale, e.g., the diatonic scales, which define the hierarchical relations between sounds by means of the rules of tonality. However, when several sounds are played simultaneously creating musical chords, harmony rules determine their relationship instead, and are simultaneously influenced by the rules of voice leading (Grove and Colles, 1944). The rules of harmony also determine the order and structural importance of the harmonic events within a musical sequence, thus determining a hierarchical structure within the musical piece. Chord progressions (harmonic progressions) such as the authentic cadence (Tonic-Subdominant- Dominant-Tonic), are common representative examples of a hierarchical structure and have been recognized as main carriers of harmony in Western tonal music theory (Piston, 1941). Violations of the harmonic structure of a musical sequence, elicit an ERP component called early right anterior negativity (ERAN), which occurs at ms after stimulus onset over anterior regions of the scalp, and tends to be right lateralized (Koelsch et al., 1999; Koelsch et al., 2000; Leino et al., 2007). The amplitude of the ERAN is modulated by attention, decreasing as the attention increases and vice versa. Therefore most studies present a semi-attentive paradigm (Koelsch et al., 2000; Koelsch et al., 2002b; Loui et al., 2005), where the participant listens to the music while performing a task unrelated to the harmony manipulation of interest. In musical experts the ERAN amplitude is higher than in novices (Koelsch et al., 2002a). The ERAN is elicited specifically by the incongruity of harmonically unexpected events in musical sequences e.g., by Neapolitan subdominant (Sn) or double dominant chords (DD), at the tonic (T) position (Koelsch and Sammler, 2008; Koelsch, 2009; Leino et al., 2007; Loui et al., 2005; Maess et al., 2001). The similarities and differences of the ERAN and the MMN have been thoroughly reviewed in Koelsch, et al. (2009). These two electrophysiological brain responses share similar temporal and scalp distributions, with amplitudes increasing with the degree of (acoustic or harmony) violation and both are linked to behavioral discrimination performance (Koelsch et al., 2001). Although the MMN seems to be

91 6 more strictly automatic as it can be elicited in participants under deep sedation by anesthesia, both the MMN and the ERAN can be elicited pre-attentively (Heinke et al., 2004; Koelsch et al., 2006). In contrast to the MMN, the ERAN has peak latency, and amplitude that specifically depends on the degree of harmonic appropriateness (Koelsch et al., 2001; Leino et al., 2007; Steinbeis et al., 2006), which, in turn, is related to prior (implicit or explicit) knowledge of Western tonal harmony. Koelsch et al. (2009) suggested that the MMN relies on short-term memory to compare an event with an immediate prior event, whereas the ERAN relies on long-term memory to compare an event with knowledge derived from prior long-term exposure. However, the MMN is elicited even by violations of the sequential rules of the equaltempered musical scale, such as mistuning of a pitch in a melody or in a chord (Brattico et al., 2001; Brattico et al., 2006; Brattico et al., 2009; Koelsch et al., 1999), and by changes in the local regularities of phonology (Shestakova et al., 2002). These findings suggest that the MMN is not simply a measure of formation and violation of sensory memory traces, but can also be modulated by long-term representations of certain kinds of musical rules already at the level of the auditory cortex. Previous findings tentatively indicate that the processing of the auditory violations tackled by MMN and ERAN takes place in different scalp locations. Using magnetoencephalography (MEG), Maess et al. (2001) localized the sources of the magnetic ERAN response in pars opercularis of the Broca s area (BA44) and its right hemispheric homologue, with a non-significant tendency towards a righthemispheric superiority. While some studies support this bilateral localization, calling the component simply early anterior negativity (EAN) (Leino et al., 2007; Loui et al., 2005), others provide evidence for a right-lateralized ERAN (Koelsch et al., 2000; Koelsch et al., 2001; Koelsch et al., 2002b; Koelsch, 2005). Leino et al. (2007) made a direct comparison between the ERAN and the MMN using deviant chords embedded in the same harmonic context. Three types of sequences consisting of seven chords each were presented: a harmonically congruous sequence adapted from the authentic cadence, a harmonically incongruous sequence that had a Neapolitan subdominant (Sn) chord in the 3, 5, or 7 positions, and a harmonically congruous sequence that had a mistuned fifth note within the chords in positions 3, 5, or 7.

92 7 Their results suggested that the harmonically incongruous chords elicited an ERAN, with an amplitude that was affected by the harmonic hierarchy of the sequence, as it was larger when the chord succession was less expected (when the Sn was placed instead of the Tonic chord, following a Dominant chord) than when it was more expected (as when the Sn was placed instead of the Subdominant chord, following a Tonic chord). On the other hand, mistuned chords elicited an MMN, with an amplitude that was not affected by the position of the mistuned chord in the sequence. What remains unclear from other studies is how harmony and musical-scale pitch violations are consciously perceived, their relation to the neural activity and if their underlying processing differs when presented in a similar musical context. Koelsch et al. (2009) do not explicitly inform about the relationship between serial and hierarchical processing, an important relationship described in language literature (Bahlmann et al., 2008). Furthermore, this review uses data of paradigms in which the two brain responses were elicited by unmatched sound contexts, which alone could explain the differences in the neural findings. Furthermore, the similarities and differences between language and music processing are still being debated. Therefore, there is a need to study these types of violation in a similar musical context, and to determine their corresponding neural correlates when elicited in a similar context. In this study, we hypothesize that the MMN indexes sequential regularities and the ERAN indexes processing of hierarchical sound structure. For this, we wished to determine the relationship between the ERAN and MMN with subjective perception of tuning and harmonic violations. We also wanted to further investigate the results from Leino et al (2007), by identifying the brain processes associated with the representation of the two types of musical regularities presented in a comparable context, such as chord progressions, using EEG. Furthermore, we wished to determine the cortical sources generating those brain responses by means of discrete source analysis using the software BESA (Brain Electric Source analysis, (Grandori et al., 1990; Scherg, 1984; Scherg and Von Cramon, 1986; Scherg and Berg, 1991; Scherg and Picton, 1991; Scherg, 1994), and distributed source analysis with the software SMP 8 (Statistical Parametric Mapping software, (Kiebel and Friston, 2004a; Kiebel and Friston, 2004b). To these aims, we adopted the Leino et al. (2007) paradigm in which different kinds of musical

93 8 violations are embedded in the same context, presenting 7-chord cadences with harmonically incongruous Neapolitan chords as well as mistuned chords (see Fig. 1). For the behavioral part of the study, we asked the participants to rate the chord cadences for Valence (1=unpleasant, to 5=pleasant) and Fittingness (1=does not fit, to 5=fits). Based on behavioral ratings of emotionality and tension obtained by (Steinbeis et al., 2006) and of relatedness obtained by (Tillmann and Lebrun-Guillaud, 2006) with similar chord sequence material, we expected higher pleasantness and fittingness for musical sequences with a harmonically incongruous chord in the 5 th position than the 3 rd or 7 th positions, and similar pleasantness and fittingness for the mistuned chord in all three positions within the chord cadences, related to the ERPs amplitude. Furthermore, we hypothesized that the mistuned chord, representing a violation of the chromatic musical scale, would elicit an MMN with predominant neural sources in the superior temporal gyrus, whereas the Neapolitan chord, violating the hierarchical rules of chord successions, was expected to elicit an ERAN originating in the inferior frontal gyrus (Brattico et al., 2006; Leino et al., 2007; Näätänen, 1995; Näätänen et al., 2004; Winkler et al., 1996). 2. Results 2.1. ERP Data Neapolitan subdominants violating the rules of harmony elicited the frontally distributed ERAN component peaking on average at 228 ms post-stimulus (see difference waves, Fig. 2). Mistuned chords, violating the rules of the musical-scale tuning, elicited the fronto-centrally distributed MMN peaking on average at 270 ms post-stimulus (see difference waves, Fig. 3). The ERAN was most prominent over anterior regions, as reflected by the significant interaction ERP Component x Distribution: F (1, 14) = 6.20, p =.02, and the contrast comparing the ERAN to the MMN for Anterior compared to Posterior Distribution: F (1, 6) = 9.80, p =.02. We observed overall significant differences in the mean response amplitude between the left and right hemispheres which depended on the type of ERP present, reflected by the interaction ERP Component x Hemisphere: F (1, 14) = 8.01, p =.01). This resulted from the bilateral distribution of the ERAN and the right-lateralized

94 9 distribution of the MMN in all the chord positions. Finally, we found a significant interaction ERP Component x Position x Distribution: F (2, 28) = 5.17, p=.01. The contrast showed that the ERAN and the MMN differed in scalp distribution when comparing position 5 to 7: F (1, 5) = 23.36, p =.02). This means the Distribution varied according to the type of ERP and Position of the violation. For further testing of these scalp topography differences between ERP components, we conducted discrete and distributed source analyses. Insert Figure 2 and Figure 3 about here 2.2. Discrete Source Analysis (BESA) The MMN in the three chord positions was modeled with 2 main symmetrical regional sources (RS) originating in the temporal lobe whereas the ERAN in the three chord positions was modeled with 2 main symmetrical RS s localized anterior to the MMN sources. Other RS s were mainly localized to the occipital lobe, and most likely caused by alpha waves and non-encephalic noise. The average coordinates of the individually-computed RS models are shown in Table 1. The ERAN RS were localized in BA 44 and BA 45 (Broca s area and right homologue). The MMN RS were localized in BA 41, corresponding to auditory cortex. These two sources accounted, on average, for > 90-80% of the variance in the data. The rest of the individually-computed RS s fitted were not considered for statistical analysis, as they explained a negligible amount of variance in the waveforms (< 1-3%). The statistical comparison of the ERAN and MMN individual RS localizations of each participant with repeated-measures ANOVA revealed a significant main effect of the ERP component factor at the y axis, F (1, 14) = 59.94, p <.001, due to the ERAN RS being anterior to the MMN RS as showed in Fig. 4 and Fig. 5. Insert Figure 4 and Table 1 about here 2.3. Distributed Source Analysis (SPM8)

95 10 In the small volume correction (SVC) of the ROIs, the MMN showed significant activation in the STG (left hemisphere (L): BA 22; right hemisphere (R): BA 41 and BA 42), whereas the ERAN showed significant activation in the IFG (L: BA 44, BA 45, BA 47; R: BA 44, BA 45, BA 47) with p <.001 (see Table 2 and Fig. 5). Insert Figure 5 and Table 2 about here Behavioral Data As visible in Figure 6, the ERAN5 is consistently the most pleasant and fitting position amongst the ERAN conditions, whereas MMN3 is the most pleasant and fitting position amongst the MMN conditions. Statistical non-parametric tests confirmed that the ratings of Pleasantness of the chords differed significantly according to their positions in the cadence. For the harmonically incongruous chords: X 2 (2) = 6.506, p=.03; for the mistuned chords: X 2 (2) = 8.222, p=.01. In the pairwise comparisons, we found significant differences between ERAN5 vs. ERAN7: T= 45.5, r= -.38 (T= smaller sum of ranks, r = effect size ), meaning ERAN5 was rated more pleasant than the ERAN7, and between MMN7 vs., both, MMN3 and MMN5: T= 23.5, r= -.42; T= 38, r= -.36, respectively, meaning MMN7 was rated as the least pleasant as compared with MMN3 and MMN5. Importantly, ERAN3 did not significantly differ in unpleasantness from ERAN7 (T = 66.5, r = -.31). As for the ratings of Fittingness, the position in the cadence had near significant effect on harmonically incongruous chords (X 2 (2)= 5.302, p=.07), whereas it was significant for mistuned chords (X 2 (2)= 9.791, p<001). However, there was only one significant difference in the pairwise comparisons between MMN3 and MMN7 (T= 26, r= -.47), meaning MMN7 was rated as the least fitting. Insert Figure 6 about here 3. Discussion

96 11 The present study was conducted to determine the relationship between the ERAN and the MMN with subjective perception of tuning and harmonic violations, the brain processes related to the neural representation of two different kinds of musical regularities and their underlying cortical generators. We found that violations from harmonic regularities, i.e., Neapolitan subdominants replacing the expected tonal chords at the ending of a cadence, elicited an ERAN at around 228 ms with an amplitude that was modulated by the degree of violation of the rules of harmony. Mistuned chords violating the rules of the musical scale elicited an MMN at around 270 ms with an amplitude that did not differ according to chord context. The reason for the early peak latency of the ERAN is that it reflects long-term memory detection of the auditory violations. In contrast, the detection of local auditory violations reflected by the MMN is more computationally demanding and, therefore, slower (Koelsch, 2009). Although, the MMN is also elicited by violations of musical regularities that are stored in long-term memory, it is important to note that the sound regularities that can be formed within an experimental session or that are stored in longterm memory representations, possibly in auditory cortical loci, are local and sequential as they are characterized by specific ratios between neighboring sounds (Brattico et al., 2001; Brattico et al., 2006; Shestakova et al., 2002). In our study, the source analyses based on both discrete and distributed algorithms gave comparable localization results, with the MMN consistently localized around the auditory cortices and the ERAN generated from Broca s area and its right hemisphere homologue, even though the chord incongruities giving rise to the ERAN and the MMN occurred in identical contexts. Moreover, the ERAN was bilaterally distributed whereas the MMN was slightly lateralized to the right hemisphere. Koelsch et al. (2001) wished to determine whether the ERAN is an abstract feature MMN or a completely distinct ERP. While playing a videogame, participants were presented with musical sequences that included either tone pairs raising (standard) or falling (deviant) in pitch. Their results indicated that the ERAN amplitude was specifically dependent on the degree of harmonic appropriateness, whereas the MMN one was not. However, due to the specific paradigm used in that study, the authors did not decisively determine whether the elicited ERAN amplitude variability was a consequence of the tonality

97 12 change or of the harmonic relationships in the chord sequences. Moreover, the paradigm did not balance the conditions in terms of the spectral complexity of the context, since the ERAN was elicited by chords in a chord context, which have a higher sound density compared to the isolated tone pairs eliciting the MMN. This was later clarified by Leino et al. (2007), who found that the ERAN amplitude modulation resulted from the harmonic relationships within the chord cadence. These two studies suggested that the components are elicited by different musical properties of the chord cadence, i.e., scale vs. harmonic hierarchy. Nevertheless, there are no previous studies that directly compare the MMN and ERAN localizations and cerebral sources against each other within the same musical context. Hence, our results further support this distinction between ERAN and MMN components and contribute to knowledge on their cortical generators. Finally, in our paradigm, mistuning can be detected as a violation of musicalscale sequential rules based on long-term memory. However, it should be noted that the mistuned chord could in principle be detected in isolation, as it contains other pitches simultaneously presented that could hint for the pitch classes according to the Western tonal scale system. Thus, the MMN here may be, strictly speaking, a "local" process. Importantly, though, mistuning detection is not a mere sensory process based on the roughness and beating in the chord, but rather relies on the presence of memory representations of scale pitch relations (Brattico et al., 2006; Brattico et al., 2009). Maess et al. s (2001) MEG study and two fmri studies by Koelsch et al. (2005) and Tillmann et al. (2003) localized the ERAN neural activity in the frontal lobe (Broca s area and right homologue), and other studies have localized the MMN to pitch changes in the auditory cortices at the supratemporal lobe (e.g., ; (Brattico et al., 2006; Tervaniemi et al., 2006a)), with additional sources in the right inferior frontal gyrus and inferior parietal lobule (Opitz et al., 2002; Park et al., 2002). In our discrete source analysis, the individual and grand average sources of the ERAN were located significantly more anterior than the MMN sources (see Fig. 4). The average sources of the ERAN were localized around BA 44 and BA 45 in IFG, similar to the Maess et al. (2001) study. The MMN was instead localized in the STG, around BA 41 and BA 42, also consistent with previous ERP and fmri literature (see Fig. 5). However, we did not find sources in IFG. This may be because our stimuli were not attention catching like the

98 13 simple acoustic changes used in previous studies, as the frontal activation of the MMN has been connected to stimulus salience and triggering of involuntary attention (Deouell, 2007; Naatanen et al., 2010). The SPM source analysis results correspond to the findings in the discrete analysis, strongly suggesting that these ERPs represent different types of processes, and that they are likely generated by separated cortical sources. In the ERP analysis, the MMN was significant in both hemispheres, but stronger in the right hemisphere, whereas the ERAN was significant in both hemispheres. This bilateral distribution disagrees with the fundamental nomenclature of the ERAN that identifies it as right lateralized, with similar temporal and mirrored spatial properties as the early left anterior negativity (LAN) (Friederici et al., 1996; Hahne and Friederici, 2002; Neville et al., 1991). On the other hand, in the present study the MMN was found to be right-lateralized. Previous studies also obtained a right-hemispheric predominance of the MMN, especially during the detection of non-phonetic auditory stimuli (Brattico et al., 2006; Opitz et al., 2002; Tervaniemi et al., 2000). However, left-lateralization has also been described depending on the non-phonetic feature (Grimm et al., 2006; Tervaniemi et al., 2006b). Furthermore, several studies failed to replicate the right lateralization of the ERAN (Loui et al., 2005; Leino et al., 2007), rather suggesting that this brain response has bilateral generators. In the behavioral study, we found that the participants consistently rated each cadence differently, indicating that the paradigm successfully examined different musical properties in the same context. Participants expressed higher pleasantness and fittingness for ERAN5 than ERAN3 and ERAN7, whereas MMN3 was rated as the most pleasant and fitting and MMN7 as the least pleasant and fitting. Furthermore, we obtained significant differences of pleasantness between ERAN5 vs ERAN7 but not between ERAN3 and ERAN7, contrasting with the MMN results where MMN7 was significantly different than MMN3. These findings are in accordance to harmony rules that state that a Neapolitan subdominant in the 5 th position is harmonically acceptable. On the other hand, the ratings of the cadences containing mistuned chords are in accordance to sequence rules, which state that errors in a sequence increase as the

99 14 tonality gets more established. However, this is not reflected in increasing amplitude of the MMN according to position. Notably, the pleasantness and fittingness ratings were obtained after the EEG recordings with full attention to the stimuli, whereas the observed MMNs reflect pre-attentive processing, which may account for the slight discrepancy between the behavioral and ERP measures Conclusion In a direct comparison between the cortical sources of the MMN and the ERAN using a contextbalanced paradigm, the present study strongly suggests that the MMN and the ERAN are components reflecting the representation of sequential and hierarchical musical regularities respectively, which have distinct cortical sources. The cognitive processes eliciting these brain responses are hence most likely different as well (Koelsch et al., 2001; Koelsch et al., 2005; Koelsch et al., 2006; Leino et al., 2007). In particular, the extraction of rules dictating a hierarchical structure of musical events seem to require generators in the prefrontal cortex, whereas for the extraction of sequential music rules determining the frequency ratio between two consecutive sounds the auditory cortex plays a primary role. 4. Materials and Methods 4.1. Participants Twenty four right-handed participants (10 male, 14 female; age range years, mean age= 24.6 years) participated in the experiment. Of those twenty-four, only five were new participants from whom we recorded new EEG data sets, whereas nineteen were from the study Leino et al. (2007). The original data set was complemented with the ERP data of the new subjects. In addition to redoing the ERP and statistical analysis, we performed source analysis to corroborate the difference between ERPs. Behavioral ratings of the musical sequences were further analyzed (not reported in Leino et al., 2007). Nine participants were rejected from the EEG analysis due to problems with the data acquisition (i.e. excessive noise in channels, technical problems, time constraints), hence they were only included in the behavioral part of the study to increase its statistical power (typically lower for behavioral tests as compared with

100 15 electrophysiological tests). In total, data from 15 participants (5 male, 10 female; age range 18 26, mean age= 24.3 years) were included in the ERP analysis and data from the 24 in the behavioral analysis. This study is an extension of the Leino et al. (2007) study, with an increased sample size, behavioral ratings and source analyses. All participants had normal hearing and no musical expertise or explicit knowledge of music theory. They also reported having no audiologic, cognitive, neurologic, or linguistic (word finding, writing, reading, and speech production and comprehension) deficits. Written informed consent was received from all participants. Participants received compensation for taking part in the experiment. Ethical permission was obtained from the ethical committee of the Faculty of Behavioral Sciences, University of Helsinki Stimuli The stimuli used in the experiment were digitally generated piano and organ chords organized into cadences. The stimulus chords were prepared according to the rules of voice leading of Western functional harmony and further edited to have equal duration and intensity. Each stimulus cadence consisted of seven 600 ms long chords. The last 50 ms of each chord was gradually faded out, and each chord in the cadence was separated from the next chord by a 5 ms silent period. Cadences consisting of chords played with a piano timbre were created for 7 different experimental conditions and in 12 different keys. In the standard condition (control), each of the 7 chords in the cadence belonged to the same key and together they composed a simple chord sequence following the rules of the Western functional harmony. In the three Neapolitan conditions (ERAN3, ERAN5, ERAN7), one in the chords of the standard cadence was replaced with a Neapolitan subdominant (Sn). The Sn replaced the tonic chord of the cadence (position 3), the subdominant chord of the cadence (position 5), or the ending tonic chord of the cadence (position 7). In the three mistuned conditions (MMN3, MMN5, MMN7), one of the chords of the standard cadence was replaced by a mistuned major triad (Mn), in which the fifth of the chord (at pitch distance of a fifth, or seven semitones, from the lowest note of the chord; see Fig. 1) was increased by 50 cents. Mirroring the Neapolitan conditions, the mistuned chord replaced either the third tonic chord

101 16 of the cadence (position 3), the fifth subdominant chord of the cadence (position 5), or the seventh tonic chord of the cadence (position 7). During the EEG recordings, tonic chords played with an organ timbre were used as target stimuli for the primary task performed by the participants. The deviant organ chords were uniformly distributed among conditions and matched with the piano chords in all aspects except for timbre. The organ chord was presented twelve times in each condition, each time in a different key (approximately 8% of all cadences), and always replaced a tonic chord of the cadence. Insert Figure 1 about here 4.3. Procedure Each experimental condition was presented in twelve different keys, 144 times in total. Each cadence was presented 5 ms after the preceding cadence, preserving the musical meter in order to give an impression of real flowing music. The cadences in different keys and with different experimental manipulations were presented in a random order, while participants were wearing headphones and sitting in a comfortable chair inside a soundproof room. The stimuli were presented using the software Presentation 9.30 (Neurobehavioral Systems, Ltd.) at a volume of 50 db above the individual hearing threshold of each participant, determined at the beginning of the experiment. A semi-attended experimental paradigm was used, in which participants were instructed to attend to the musical sequences and press the response button after hearing the chord played by an organ. The purpose of the task was to ensure that participants were attending to the stimuli without asking him or her to attend to the Neapolitan or mistuned chords (Koelsch et al., 2002a). The stimuli were presented in eight separate stimulus blocks of approximately 10 min each in duration, and the entire experiment was approximately 3 hours including preparation. The EEG was measured using the BioSemi measuring system (BioSemi, Inc., Netherlands; The scalp EEG was recorded with 128 active scalp electrodes fitted into a stretching cap and following the BioSemi ABC position system. Additionally, three active electrodes were placed on the participant s nose and mastoid areas respectively, and four more electrodes were

102 17 placed around the eyes to monitor eye muscle activity. 24-bit EEG data were recorded with BioSemi ActiView 5.32 using no reference (standard Biosemi acquisition method), with a sampling rate of 2048 Hz, and a recording bandwidth of up to 417 Hz. After the EEG measurements, the participants were instructed to listen to 28 cadences in different keys taken randomly from each of the 7 experimental conditions. Each experimental condition was repeated 4 times in different random keys. Participants rated each cadence according to valence and fittingness on 5-point scales (with 1 being the lowest score and 5 the highest) by answering to the following questions: how pleasant did you find the melody?, and how well did you think the chords of the melody fitted together? ERP Analysis Pre-processing and data analysis were performed with Brain Electric Source Analysis (BESA version5.2) (Berg and Scherg, 1994; Miltner et al., 1994). The EEG data for each participant were re-sampled offline for ERP analysis to a sampling rate of 256 Hz. We averaged the ERPs for each condition and cadence. The epoch was 600 ms starting from target-chord onset (the third, fifth, or seventh chord of the cadence). As a pre-stimulus baseline, we used a 100 ms period preceding the onset of the target chord. Channels with excessively noisy data (skin potentials and high frequency noise) were rejected, and artifact rejection was performed in BESA. Before averaging, the EEG data were filtered with a 0.5 Hz high-pass filter. After averaging, the ERP data were filtered with a low-pass filter of 40 Hz and re-referenced to the average of the mastoids. Difference waveforms were calculated by subtracting the responses to the standard chords from those to the Neapolitan or mistuned chords. The ERP data were then statistically evaluated using repeated-measures analyses of variance (ANOVA) with the mean ERP amplitude as the dependent variable. First, the mean ERP amplitudes were calculated for each participant and condition from a 40-ms time windows surrounding the grand-averaged peak at the Fz electrode (channel 21 in the Biosemi system). Then the mean amplitude data were analyzed using repeated-measures ANOVA with ERP

103 18 Component (levels: ERAN, MMN) and Position (levels: 3, 5, 7) as within-subject factors. Second, the mean amplitude values were computed for four regions of interest (ROIs): left anterior (C25, C26, C32, D3, D4), right anterior (C3, C4, C10, C12, C13), left posterior (A5, A6, A7, A8, A18), and right posterior (A31, A32, B3, B4, B5). The data were then analyzed again using a four-way repeated-measures ANOVA with ERP Component (ERAN-MMN), Position (3, 5, 7), Distribution (Anterior-Posterior), and Hemisphere (Right-Left) as within-subject factors. For all statistical analyses, type I errors were controlled for by using Mauchly s test and the Greenhouse Geisser epsilon when appropriate. The alphalevel for all statistical analysis was.05, unless stated otherwise. After statistical evaluation, the grandaverage ERP waveforms were filtered with a 10-Hz low-pass filter for illustration purposes only Source Analysis Discrete source localization was performed with the Brain Electrical Source Analysis (BESA version 5.2) software (Berg and Scherg, 1994). The 4-shell spherical head model was used. In a first exploratory analysis aimed at determining the number and the starting locations of the source models, we performed the principal component analysis (PCA) for each participant and condition. We then studied the distribution of the resulting PCs and considered their physiological feasibility. The number and locations of the acceptable sources were fed into the subsequent analysis step. A regional source analysis was performed for the time window of 40 ms around the individual peaks of the global field power (GFP) curves for each experimental condition (ERAN3, ERAN5, ERAN7, MMN3, MMN5, MMN7). Regional sources (RS) are more stable than dipoles in the presence of noise because they model the dipolar current flow in the three Cartesian directions. The source analysis was performed in each participant with a residual variance (RV) < 20%. The first two RS fitted were symmetrically constrained because in the individual data stability is low and the noise is high, hence this procedure was applied in order to avoid noise from moving the encephalic regional sources to the middle and outside of the brain. Also, this was done because of the physiological assumption that the sources of the ERAN and MMN are typically located in similar structures of the two cerebral hemispheres (e.g., (Brattico et al., 2008; Maess et al.,

104 )). Next, we fitted more RS taking the PCA with an explanatory variance percentage above 1% as a reference for the number of possible sources. The first two symmetric RS fitted in each participant explained > 80% of the observed signal in all cases, with corresponding activity shown in the waveforms. Therefore, those two RS were considered as the main sources of MMN and ERAN activity respectively, and used for further analysis and plotting. The coordinates of the resulting RS were registered according to the Talairach coordinate system (Talairach and Tournoux, 1988) for each condition (ERAN3, ERAN5, ERAN7, MMN3, MMN5, MMN7) and participant (N=15). In this coordinate system, the x-axis of the brain passes through the two preauricular points with positive to the right, the y-axis passes through the nasion and is perpendicular to the x-axis (positive to the front), and the z-axis points up and is perpendicular to the xy-plane (positive upwards). The differences in the coordinate locations between the ERAN and MMN regional sources were separately studied for each axis (hemispheres were not taken into account because the RS were symmetric) with repeated-measures ANOVA with ERP Component (levels: ERAN, MMN), and Position (levels: 3, 5, 7) as within-subject factors. Finally, the mean coordinates of the RS were computed for the x, y and z axes in all conditions and plotted. To show the localization of the individual and mean RS localization, we superimposed them on MRI images using BrainVoyager Brain Tutor software ( This software also provides localization of cortical regions according to the Brodmann areas (BA). Considering the uncertainty related to a single source analysis solution due to the inverse problem, we decided to perform an additional source analysis with Statistical Parametric Mapping (SPM 8 (Kiebel and Friston, 2004a; Kiebel and Friston, 2004b). SPM 8 uses Multiple Sparse Priors (MSP), a L2-norm-like approach to the inverse problem in which hundreds of patches of cortex are treated as spatial priors within a parametric empirical Bayesian framework (Friston et al., 2008). The Bayesian approach is a statistical method used to incorporate a priori information into the estimation of the sources and can result in linear or non-linear estimators. The types of a priori information include information on the neural current, the sparse focal nature of the sources, combined

105 20 spatial and temporal constraints as well as strategies to penalize ghost sources, amongst others (multiple constrains) (Friston, 2007). After this, it performs a Group Inversion where subjects are pooled over to optimize the MSP before inverting any single subject (Litvak and Friston, 2008). It uses canonical cortical mesh (inverse-normalized), obviating the need for complex manual creation of individual cortical meshes from MRI, providing a mapping with a template in MNI space and allowing statistics across subjects. Then it creates statistic parametric maps to test for topographical differences in ERP amplitude across the scalp and time. Overall, it provides several constrains to minimize human manipulation. The data from each subject is transformed into two-dimensional sensor-space (interpolated from 128 channels) and time window as a third dimension. This transformation produces three-dimensional spatiotemporal maps of the ERP. These maps can be then compared to localize condition effects using standard mass univariate statistical parametric maps in the fmri way, by relying on t or F statistics to test for effects that are localized in peristimulus time. For this analysis the ERP data were first converted to SPM format, and subsequently we assigned sensor positions, performed co-registration using BEM (boundary element method) head model and Group Inversion. After the inversion, we created statistical maps (images) for the subsequent general linear model (GLM) analysis. We created baseline images within the window of -100 to 0 ms to contrast with the conditions. Subsequently, we individually created images for the ERAN and MMN conditions using average windows of 100 ms around the individual peak location (between latencies 100 and 300 ms). Then, we performed the group analysis with a paired t-test (condition > baseline) for each condition and an uncorrected p-value of As the locations of the MMN main sources are suggested to be located in Superior Temporal Gyrus (STG), and the ERAN sources are in Inferior Frontal Gyrus (IFG), we performed small volume correction (SVC) for these locations using IFG and STG maps, respectively, included in the MarsBaR Toolbox (Brett, 2002) as bounding masks, and a constrain of 4 mm for minimum spatial separation between peaks. For illustration purposes, we used the MRI template included in SPM 8 (single subject T1).

106 Behavioral Analysis We conducted two different analyses of the behavioral ratings, one for the ratings of pleasantness and another for ratings of fittingness of chords. Participants rated each cadence according to valence and fittingness on 5-point scales (with 1 being the lowest score and 5 the highest) by answering to the following questions: how pleasant did you find the melody?, and how well did you think the chords of the melody fitted together? The variables were then compared using the non-parametric Friedman s ANOVA with the Exact test method. We compared harmonically incongruous chords (ERAN3, ERAN5, ERAN7) vs. mistuned chords (MMN3, MMN5, MMN7), and the three positions within the cadence (3, 5, 7). Post hoc tests were performed using non-parametric Wilcoxon tests with Bonferroni correction (significance level =.0167), contrasting positions of harmonically incongruous chords (ERAN3-ERAN5, ERAN5-ERAN7, ERAN3- ERAN7), and positions of pairs of mistuned chords (MMN3-MMN5, MMN5-MMN7, MMN3-MMN7). We then plotted the data using mean scores as the y value in a bar graph (T= smaller sum of ranks, r = effect size). Acknowledgements This work was supported by the Danish National Research Foundation s Center for Functionally Integrative Neuroscience, University of Aarhus in Denmark; the Cognitive Brain Research Unit, University of Helsinki, in Finland, and Ulla & Mogens Folmer Andersens Fond. We would like to thank Prof. Teija Kujala for allowing the use of the Cognitive Brain Research Unit facilities for our experiment, and thank the rest of the CBRU staff for their help. We would also like to thank M.Sc., Chris Bailey, M.Sc., Prof. Risto Näatänen, M.Sc., PhD, Mallar Chakravarty and M.Sc., PhD Kristjana Yr Jonsdottir for their helpful insights and invaluable comments. References

107 Bahlmann, J., Schubotz, R.I., Friederici, A.D., Hierarchical artificial grammar processing engages Broca's area. Neuroimage. 42, Berg, P., Scherg, M., A fast method for forward computation of multiple-shell spherical head models. Electroencephalogr Clin Neurophysiol. 90, Brattico, E., Näätänen, R., Tervaniemi, M., Context Effects on Pitch Perception in Musicians and Nonmusicians: Evidence from Event-Related-Potential Recordings. Music Perception. 19, Brattico, E., Tervaniemi, M., Naatanen, R., Peretz, I., Musical scale properties are automatically processed in the human auditory cortex. Brain Res. 1117, Brattico, E., Pallesen, K.J., Varyagina, O., Bailey, C., Anourova, I., Jarvenpaa, M., Eerola, T., Tervaniemi, M., Neural Discrimination of Nonprototypical Chords in Music Experts and Laymen: An MEG Study. J Cogn Neurosci. Brattico, E., Pallesen, K.J., Varyagina, O., Bailey, C., Anourova, I., Jarvenpaa, M., Eerola, T., Tervaniemi, M., Neural discrimination of nonprototypical chords in music experts and laymen: an MEG study. J Cogn Neurosci. 21, Brett, M., Anton, JL., Valabregue, R., Poline, JB., Region of interest analysis using an SPM toolbox [abstract]. Presented at the 8th International Conference on Functional Mapping of the Human Brain, Sendai, Japan. 16. Deouell, L.Y., The Frontal Generator of the Mismatch Negativity Revisited. Journal of Psychophysiology. 21, Friederici, A.D., Hahne, A., Mecklinger, A., Temporal structure of syntactic parsing: early and late event-related brain potential effects. J Exp Psychol Learn Mem Cogn. 22, Friston, K., Harrison, L., Daunizeau, J., Kiebel, S., Phillips, C., Trujillo-Barreto, N., Henson, R., Flandin, G., Mattout, J., Multiple sparse priors for the M/EEG inverse problem. Neuroimage. 39, Friston, K.J., Statistical parametric mapping : the analysis of functional brain images. Vol., Elsevier/Academic Press, Amsterdam ; London. Grandori, F., Hoke, M., Romani, G.L., Auditory evoked magnetic fields and electric potentials. Advances in audiology, Vol., Karger, Basel ; London. Grimm, S., Roeber, U., Trujillo-Barreto, N.J., Schroger, E., Mechanisms for detecting auditory temporal and spectral deviations operate over similar time windows but are divided differently between the two hemispheres. Neuroimage. 32, Grove, G., Colles, H.C., Grove's Dictionary of music and musicians. Vol., Macmillan, New York. Hahne, A., Friederici, A.D., Differential task effects on semantic and syntactic processes as revealed by ERPs. Brain Res Cogn Brain Res. 13, Heinke, W., Kenntner, R., Gunter, T.C., Sammler, D., Olthoff, D., Koelsch, S., Sequential effects of increasing propofol sedation on frontal and temporal cortices as indexed by auditory eventrelated potentials. Anesthesiology. 100, Kiebel, S.J., Friston, K.J., 2004a. Statistical parametric mapping for event-related potentials: I. Generic considerations. Neuroimage. 22, Kiebel, S.J., Friston, K.J., 2004b. Statistical parametric mapping for event-related potentials (II): a hierarchical temporal model. Neuroimage. 22, Koelsch, S., Schroger, E., Tervaniemi, M., Superior pre-attentive auditory processing in musicians. Neuroreport. 10, Koelsch, S., Gunter, T., Friederici, A.D., Schroger, E., Brain indices of music processing: "nonmusicians" are musical. J Cogn Neurosci. 12, Koelsch, S., Gunter, T.C., Schroger, E., Tervaniemi, M., Sammler, D., Friederici, A.D., Differentiating ERAN and MMN: an ERP study. Neuroreport. 12,

108 Koelsch, S., Schmidt, B.H., Kansok, J., 2002a. Effects of musical expertise on the early right anterior negativity: an event-related brain potential study. Psychophysiology. 39, Koelsch, S., Schroger, E., Gunter, T.C., 2002b. Music matters: preattentive musicality of the human brain. Psychophysiology. 39, Koelsch, S., Neural substrates of processing syntax and semantics in music. Curr Opin Neurobiol. 15, Koelsch, S., Gunter, T.C., Wittfoth, M., Sammler, D., Interaction between syntax processing in language and in music: an ERP Study. J Cogn Neurosci. 17, Koelsch, S., Heinke, W., Sammler, D., Olthoff, D., Auditory processing during deep propofol sedation and recovery from unconsciousness. Clin Neurophysiol. 117, Koelsch, S., Sammler, D., Cognitive components of regularity processing in the auditory domain. PLoS ONE. 3, e2650. Koelsch, S., Music-syntactic processing and auditory memory: similarities and differences between ERAN and MMN. Psychophysiology. 46, Krumhansl, C.L., Rhythm and pitch in music cognition. Psychol Bull. 126, Lehmann, M., The impact of uniqueness in musical scales on mistuning detection. In International Conference on Music Perception and Cognition. Vol., G.I.f.R.i.M.E. Hanover University of Music and Drama, ed.^eds. Hanover University of Music and Drama, Germany. Institute for Research in Music Education Sapporo, Japan. Leino, S., Brattico, E., Tervaniemi, M., Vuust, P., Representation of harmony rules in the human brain: further evidence from event-related potentials. Brain Res. 1142, Litvak, V., Friston, K., Electromagnetic source reconstruction for group studies. Neuroimage. 42, Loui, P., Grent-'t-Jong, T., Torpey, D., Woldorff, M., Effects of attention on the neural processing of harmonic syntax in Western music. Brain Res Cogn Brain Res. 25, Maess, B., Koelsch, S., Gunter, T.C., Friederici, A.D., Musical syntax is processed in Broca's area: an MEG study. Nat Neurosci. 4, Miltner, W., Braun, C., Johnson, R., Jr., Simpson, G.V., Ruchkin, D.S., A test of brain electrical source analysis (BESA): a simulation study. Electroencephalogr Clin Neurophysiol. 91, Naatanen, R., Kujala, T., Winkler, I., Auditory processing leading to conscious perception: A unique window to central auditory processing opened by the mismatch negativity (MMN) and related responses. International Journal of Psychophysiology. 77, Näätänen, R., The mismatch negativity: a powerful tool for cognitive neuroscience. Ear Hear. 16, Näätänen, R., Lehtokoski, A., Lennes, M., Cheour, M., Huotilainen, M., Iivonen, A., Vainio, M., Alku, P., Ilmoniemi, R.J., Luuk, A., Allik, J., Sinkkonen, J., Alho, K., Language-specific phoneme representations revealed by electric and magnetic brain responses. Nature. 385, Näätänen, R., Tervaniemi, M., Sussman, E., Paavilainen, P., Winkler, I., "Primitive intelligence" in the auditory cortex. Trends Neurosci. 24, Näätänen, R., Pakarinen, S., Rinne, T., Takegata, R., The mismatch negativity (MMN): towards the optimal paradigm. Clin Neurophysiol. 115, Näätänen, R., Paavilainen, P., Rinne, T., Alho, K., The mismatch negativity (MMN) in basic research of central auditory processing: A review. Clinical Neurophysiology. 118, Neville, H., Nicol, J.L., Barss, A., Forster, K.I., Garrett, M.F., Syntactically Based Sentence Processing Classes - Evidence from Event-Related Brain Potentials. Journal of Cognitive Neuroscience. 3,

109 Opitz, B., Rinne, T., Mecklinger, A., von Cramon, D.Y., Schroger, E., Differential contribution of frontal and temporal cortices to auditory change detection: fmri and ERP results. Neuroimage. 15, Paavilainen, P., Simola, J., Jaramillo, M., Naatanen, R., Winkler, I., Preattentive extraction of abstract feature conjunctions from auditory stimulation as reflected by the mismatch negativity (MMN). Psychophysiology. 38, Paavilainen, P., Arajarvi, P., Takegata, R., Preattentive detection of nonsalient contingencies between auditory features. Neuroreport. 18, Park, H.J., Kwon, J.S., Youn, T., Pae, J.S., Kim, J.J., Kim, M.S., Ha, K.S., Statistical parametric mapping of LORETA using high density EEG and individual MRI: application to mismatch negativities in schizophrenia. Hum Brain Mapp. 17, Picton, T.W., Alain, C., Otten, L., Ritter, W., Achim, A., Mismatch negativity: different water in the same river. Audiol Neurootol. 5, Piston, W., Harmony. vi, 310 p. Rinne, T., Degerman, A., Alho, K., Superior temporal and inferior frontal cortices are activated by infrequent sound duration decrements: an fmri study. Neuroimage. 26, Scherg, M., Spatio-temporal modelling of early auditory evoked potentials. Rev Laryngol Otol Rhinol (Bord). 105, Scherg, M., Von Cramon, D., Evoked dipole source potentials of the human auditory cortex. Electroencephalogr Clin Neurophysiol. 65, Scherg, M., Berg, P., Use of prior knowledge in brain electromagnetic source analysis. Brain Topogr. 4, Scherg, M., Picton, T.W., Separation and identification of event-related potential components by brain electric source analysis. Electroencephalogr Clin Neurophysiol Suppl. 42, Scherg, M., From EEG source localization to source imaging. Acta Neurol Scand Suppl. 152, Shestakova, A., Brattico, E., Huotilainen, M., Galunov, V., Soloviev, A., Sams, M., Ilmoniemi, R.J., Naatanen, R., Abstract phoneme representations in the left temporal cortex: magnetic mismatch negativity study. Neuroreport. 13, Steinbeis, N., Koelsch, S., Sloboda, J.A., The role of harmonic expectancy violations in musical emotions: evidence from subjective, physiological, and neural responses. J Cogn Neurosci. 18, Talairach, J., Tournoux, P., Co-planar stereotaxic atlas of the human brain : 3-dimensional proportional system : an approach to medical cerebral imaging. Vol., Thieme Medical, Stuttgart, New York. Tervaniemi, M., Medvedev, S.V., Alho, K., Pakhomov, S.V., Roudas, M.S., Van Zuijen, T.L., Naatanen, R., Lateralized automatic auditory processing of phonetic versus musical information: a PET study. Hum Brain Mapp. 10, Tervaniemi, M., Castaneda, A., Knoll, M., Uther, M., 2006a. Sound processing in amateur musicians and nonmusicians: event-related potential and behavioral indices. Neuroreport. 17, Tervaniemi, M., Szameitat, A.J., Kruck, S., Schroger, E., Alter, K., De Baene, W., Friederici, A.D., 2006b. From air oscillations to music and speech: functional magnetic resonance imaging evidence for fine-tuned neural networks in audition. J Neurosci. 26, Tillmann, B., Janata, P., Bharucha, J.J., Activation of the inferior frontal cortex in musical priming. Ann N Y Acad Sci. 999, Tillmann, B., Lebrun-Guillaud, G., Influence of tonal and temporal expectations on chord processing and on completion judgments of chord sequences. Psychol Res. 70, Tse, C.-Y., Penney, T.B., On the functional role of temporal and frontal cortex activation in passive detection of auditory deviance. Neuroimage. 41,

110 Winkler, I., Karmos, G., Naatanen, R., Adaptive modeling of the unattended acoustic environment reflected in the mismatch negativity event-related potential. Brain Res. 742,

111 Figure Legends Figure Legends Fig. 1 - Stimuli. T=Tonic, S=Subdominant, Sn=Neapolitan Subdominant, D=Dominant, T3= Inverted tonic, Tm=Mistuned tonic, Sm=Mistuned Subdominant. Top box: Musical notation of the chords use for the stimulus, only in the key of C for illustration purposes. The arrows point to the mistuned note in Tm and Sm respectively. Bottom box: Illustration of the musical cadence. The left column shows the type of condition. The right columns show the 7 positions of the cadence and the key deviant chords (3, 5, 7). Fig. 2 - Top: Top view of the ERP analysis in all 128 electrodes. Bottom: Difference waves shown in channel 21 (Fz) between ERP responses to harmonically incongruous chords, illustrating the ERAN in the three different cadence positions (3, 5, 7). Fig. 3 - Top: Top view of the ERP analysis in all 128 electrodes. Bottom: Difference waves shown in channel 21 (Fz) between responses to mistuned chords, illustrating the MMN in the three different cadence positions (3, 5,7). Fig. 4 - Mean location of the regional sources (RS) of all participants in the Cartesian plane. Here we show the RS separately for Condition (ERAN, MMN) and Position (3, 5, 7). Top image illustrates the coordinates in the y (anterior-posterior) and x (right-left) axes. Bottom image illustrates the coordinates in the z (top-bottom) and x axes. Left coordinates represent the left hemisphere, while right coordinates represent the right hemisphere. The sources were symmetrically constrained, therefore hemisphere representation is irrelevant. Fig. 5 - The left column corresponds to the ERAN and the right column to the MMN. Inside each column there are two sub-columns. Left sub-columns: discrete source analysis of the ERAN and MMN responses for all experimental conditions. Right sub-columns: distributed (SPM) analysis of the ERAN and MMN responses for all experimental conditions. The first row shows the sagittal view, the second row shows the

112 coronal view, and the third row shows the transversal view (A=Anterior, P=Posterior, L=Left, R=Right, IRS=Individual Regional Sources, MRS=Mean Regional Sources). The color bar represents the T- statistic. Fig. 6 - Mean scores of a) Pleasantness and b) Fittingness behavioral ratings in the y axis. Each experimental condition and position is showed in the x axis. (Std= Standard condition).

113 Figure 1 T T 3 S D Sn Tm Sm Position Standard: Neapolitan: Mistuned: T D T T3 S 34 D T Sn Sn Sn Tm Sm Tm

114 Figure 2

115 Figure 3

116 Figure 4 y x z x Position 3 Position 5 Position 7 ERAN MMN

117 Figure 5 Click here to download high resolution image 38

118 Figure 6 a 5 4 Mean Scores b Mean Scores Std E3 E5 E7 M3 M5 M7

119 Table I Table I. Regional Sources MMN x y z Location STG STG ERAN x y z Location IFG IFG Table 1. Mean average coordinates of the MMN and the ERAN regional sources in the discrete source analysis. The coordinates are in Talairach space. (STG= Superior Temporal Gyrus; IFG= Inferior Frontal Gyrus).

120 Table II Table II. Distributed analysis results MMN: Superior temporal gyrus x y z Z score p value < < < < < < <.001 ERAN: Inferior frontal gyrus x y z Z score p value < < < < < < < < < < < < < < < < < < < < < <.001

121 Table 2. Talairach coordinates of the significant peaks localized by the small volume correction (threshold p <.001 uncorrected). IFG includes BA 44, 45 and 47, whereas STG includes BA 41 and 42.

122

123 Elsevier Editorial System(tm) for The Journal of Pain Manuscript Draft Manuscript Number: JPAIN-D Title: Superior analgesic effects of mental arithmetic versus unfamiliar music and sounds: The role of emotional impact and cognitive styles Article Type: Clinical Report Keywords: pain; music; emotion; personality; analgesia Corresponding Author: Dr. Eduardo Adrian Garza-Villarreal, M.D., Ph.D. Corresponding Author's Institution: University of Aarhus First Author: Eduardo Adrian Garza-Villarreal, M.D., Ph.D. Order of Authors: Eduardo Adrian Garza-Villarreal, M.D., Ph.D.; Elvira Brattico, MSc, PhD; Lene Vase, Ph.D.; Leif Østergaard, MD, MSc, PhD, DMSc; Peter Vuust, MSc, BA, PhD Abstract: Previous studies have shown a superior analgesic effect of favorite music over other passive or active distractive tasks. However, it is unclear what mediates this effect. In this study we investigated to which extent distraction, emotional valence and cognitive styles may explain part of the relationship. Forty-eight healthy volunteers received heat stimuli during an active mental arithmetic task (PASAT), and passive listening to music (Mozart), environmental sounds (rain and water), and control (noise). The participants scored the conditions according to affective scales and filled out questionnaires concerning cognitive styles (Baron - Cohen and self-report). Active distraction with PASAT led to significantly less pain intensity and unpleasantness as compared to music and sound. In turn, both music and sound relieved pain significantly more than noise. When music and sound had the same level of valence they relieved pain to a similar degree. The emotional ratings of the conditions were correlated with the amount of pain relief and cognitive styles seemed to influence the analgesia effect. These findings suggest that the pain relieving effect previously seen in relation to music may be at least partly mediated by distraction, emotional factors and cognitive styles rather than by the music itself.

124 *Manuscript Superior analgesic effects of mental arithmetic versus unfamiliar music and sounds: The role of emotional impact and cognitive styles Eduardo A. Garza Villarreal 1,2,*, Elvira Brattico 4, Lene Vase 1,3, Leif Østergaard 1,5, Peter Vuust 1,2 1 Center of Functionally Integrative Neuroscience, University of Aarhus, Denmark 2 Royal Academy of Music, Aarhus, Denmark 3 Department of Psychology, University of Aarhus, Denmark 4 Cognitive Brain Research Unit, Cognitive Science, Institute of Behavioral Science, University of Helsinki & Centre of Excellence for Interdisciplinary Music Research, University of Jyväskylä, Finland 5 Department of Neuroradiology, Aarhus University Hospital, Denmark *Corresponding Author: Eduardo Adrian Garza Villarreal Center for Functionally Integrative Neuroscience, Danish Neuroscience Center, Aarhus University Hospital, Nørrebrogade 44 Building 10G , Aarhus C, Denmark Mobile: Work: Fax: [email protected], [email protected] URL:

125 Abstract Previous studies have shown a superior analgesic effect of favorite music over other passive or active distractive tasks. However, it is unclear what mediates this effect. In this study we investigated to which extent distraction, emotional valence and cognitive styles may explain part of the relationship. Forty-eight healthy volunteers received heat stimuli during an active mental arithmetic task (PASAT), and passive listening to music (Mozart), environmental sounds (rain and water), and control (noise). The participants scored the conditions according to affective scales and filled out questionnaires concerning cognitive styles (Baron Cohen and self-report). Active distraction with PASAT led to significantly less pain intensity and unpleasantness as compared to music and sound. In turn, both music and sound relieved pain significantly more than noise. When music and sound had the same level of valence they relieved pain to a similar degree. The emotional ratings of the conditions were correlated with the amount of pain relief and cognitive styles seemed to influence the analgesia effect. These findings suggest that the pain relieving effect previously seen in relation to music may be at least partly mediated by distraction, emotional factors and cognitive styles rather than by the music itself. Perspective The current study provides evidence that the analgesic effects of music are mediated by the distractibility of the stimulus, the emotional response it elicits in the person and the individual cognitive style. These findings could help clinicians prescribe music as an adjuvant for pain management, with the highest analgesic effect possible. Furthermore, this study provides suggestions to composers interested in enhancing the therapeutic impact of music. Keywords: pain, music, emotion, personality, analgesia

126 Introduction Music is valued for its aesthetical and recreational qualities, but plays an important role in medicine for treatment and management of pain as well. Recent studies report analgesic effects of music comparable to NSAIDs drugs (e.g. ibuprofen) 1. Furthermore, listening to music reduces the dosage of sedatives and analgesic medication, and has benefits to the overall wellbeing 2-8. However, most studies investigating the analgesic effects of music do not systematically control for factors known to influence pain such as distractibility, emotional factors, familiarity, and the personality of the participants. Distraction from a painful stimulus reduces its intensity and unpleasantness 9-11, whereas focusing attention on the pain increases it In a recent study, Mitchell, et al. 14 reported that self-chosen music increased pain tolerance more than an active distraction task consisting of mental arithmetic. However, in this study the music was chosen by the participants, as this research group had shown that self-chosen music increases pain tolerance more than music selected by the experimenter 15. Thus, preference, intentionality and familiarity, could possibly enhance the individual drive to listen attentively to the music and the consequent distraction from the pain. The emotional factors experienced by listening to music that supposedly influence pain perception are valence and arousal 1, Music with positive valence (pleasant music), significantly reduces pain perception as shown in a recent study by Roy et al. 1, in which pleasant vs. unpleasant musical pieces were presented while the participants received painful stimuli. Their conclusion was that positive emotional valence contributes to the analgesic effect of music. Emotional valence has been shown to contribute to the analgesic effect with non-musical stimuli Although in this study the participants did not choose the music, the songs were fairly

127 popular and possibly familiar to more than one participant. The other important emotional factor modulating pain perception is arousal Therefore it has been suggested that music with a slow tempo may reduce heart rate, anxiety levels 20 and pain 5, 21. Nevertheless, although the above described results are clear-cut and relevant to the field, in neither of these studies on music and pain were the cognitive styles of the participants taken into account. Personality traits influences susceptibility to pain 22, possibly by determining the individual emotional reactions to events and by predisposing to certain moods Furthermore, cognitive styles, such as being emotional or having a tendency to focus on systematic and analytic structures, have a close relation with music listening styles 28. Therefore, cognitive styles could possibly influence pain when listening to music. Gender differences in pain perception with and without music have also been reported in several studies 17, 29, with contradicting results. However, none of these studies exploring the role of individual differences on pain experience with music have fully controlled for the analgesic effect of distraction, emotional connotations, or familiarity. In this study we wanted to investigate the analgesic effects of an active mental arithmetic task compared to passive listening of music. As a second aim, we wanted to investigate the effects of the different passive listening tasks, characterized by similar levels of valence, arousal and liking, on perception of pain. As a third aim, we wished to explore the individual variation in the analgesic potential of the active and passive tasks related to gender, and cognitive styles, by administering the Baron-Cohen Empathizer Systemizer and self-report questionnaires. We hypothesized that music listening would reduce pain more than noise and environmental sound listening and more than the mental arithmetic task. We also expected that the emotional

128 connotations of the tasks, and the individual gender and cognitive styles would influence the pain. Materials and Methods Participants Forty eight participants (24 male, 24 female), aged between 19 and 39 years (mean = 24, SD = 4) took part in the experiment. All of them were native Danish speakers. They were recruited placing advertisements around the city and on a research recruitment website. The participants were healthy, right handed, they reported normal hearing and had minimum to no musical training. They had not consumed any analgesic medication in the 24 hours prior to the experiment. Written informed consent was obtained from all participants. Participants received compensation for taking part in the experiment. Ethical permission was obtained from The Research Ethical Committee for Mid-Jutland Region, Denmark. Thermal stimuli and pain measures The thermal stimuli was produced by a 3 x 3 cm contact thermode (Pathway model ATS from Medoc Ltd. Advanced Medical System, Israel) on the forearms. The pain limits and threshold were investigated for each participant. Calibration trials were performed as as control for individual differences in pain perception. The method was adapted from Price et al We presented two trials with four different temperatures: 42, 43, 45, and 47 o C presented in a random order. Each stimulus was separated by approximately 15 20s, and the participants rated pain intensity and unpleasantness on the Visual Analog Scale (VAS) for each temperature. The goal temperature had to be below 47 o C and reflect pain ratings between mm (moderate

129 to high) in the VAS. This goal temperature was kept constant during the entire experiment to avoid high variability of the VAS scores. To avoid habituation, the thermode was changed to a slightly different skin location on the forearm after every two experimental conditions. Each painful stimulus consisted of a plateau of 16 s with a rise/fall time of 2 s. The baseline temperature was 35 C. The thermal stimulus was measured using the VAS for pain intensity and unpleasantness. The scale ranged from no pain (left end of the scale) to very intense or very unpleasant (right end) 31. Auditory stimuli (for the active and passive tasks) We selected the auditory stimuli of our study from a pilot experiment, where 18 participants listened to a pool of 16 environmental sounds and 19 music excerpts. The environmental sounds were recordings from nature (Sound effect library, Sound Ideas The sounds were four examples of each: fire, water, rain and wind. The music pieces were 19 different Mozart string compositions, virtually unknown to the layman. The participants were required to rate the stimuli according to valence (0 = unpleasant, 10 = pleasant), arousal (0 = environmental, 10 = stimulating), and liking (0 = not liked, 10 = liked). The results showed that rain and water for the environmental sounds and the String Quartet No. 1 in G major, K. 80/73f (1770) Adagio and Divertimento in E flat, K. 563 Adagio for the music pieces were rated as the most pleasant, liked and least arousing (Fig 1). Therefore, we used these auditory stimuli for the passive tasks of this study. Each auditory stimulus lasted 300 s (5 min) and was normalized. Additionally, we included pink noise (less distressing than white noise) as a control. To control for the familiarity of the music, after the experiment the participants were asked if they had previously listened to the music piece.

130 The PASAT (Paced Auditory Serial Addition Test) was chosen as the only active task 32. The PASAT consisted of a woman s voice in Danish saying numbers every three seconds. The task for the participant consisted in adding together two of the numbers at a time, while remembering the last number for the whole duration of the condition. This condition also lasted 300 s (5 min) (Fig 1). Emotional measures After the experiment, the participants rated the auditory stimuli in 10-point Likert scales for valence (0 = unpleasant, 10 = pleasant), arousal (0 = environmental, 10 = stimulating) and liking (0 = does not like, 10 = likes) (VLA), to corroborate the emotional ratings reported in the pilot study and measure the emotions associated with the stimuli. Cognitive styles The participants answered the Baron-Cohen Empathizer Systemizer quotient and a self-report questionnaire. The Baron-Cohen Empathizer Systemizer Quotient divides the population into three groups: Empathizers (more empathic and social), Systemizers (attracted to patterns in objects and events) and Balanced (in between). Empathizers are thought to be attracted to the emotional content of the music, whereas the Systemizers might be attracted to musicianship and performance level 28. For the self-report, they were asked a simple binomial question with only one possible answer: Do you think of yourself as more: Emotional or Rational?. Other psychometric scales, addressing questions irrelevant to the present paper, were also used and the results will be reported elsewhere (Garza-Villarreal et al., in preparation).

131 We then grouped the participants according to the Baron-Cohen Empathizer-Systemizer (Baron-Cohen), the Emotional-Rational self-report (E/R) and Gender for the statistical analysis. The Baron-Cohen classification led to three groups of 16 participants: Empathizers, Systemizers and Balanced (each with 8 male, 8 female) whereas the Emotional-Rational classification consisted of two groups of 24 participants each (12 male, 12 female). Procedure The participants were contacted prior to the experiment to explain the details about the thermal stimulation and to arrange the appointment for the experiment. First, they were asked to answer the psychometric questionnaires. Then, when entering the laboratory, they were told their task was to rate the effects of different types of sounds while receiving pain. The experiment took place in a sound proof white room without windows. Instructions for all participants were exactly identical and given by the same male experimenter, who was the only person present during the experiment. The participants were trained to use the VAS and were familiarized with the thermal stimuli by investigating pain limits and threshold. They were also trained to perform the PASAT. They were seated comfortably in a chair in front of a monitor and were given a mouse to rate pain using a computerized VAS. To minimize confounds, a panel wall stood between the experimenter and the participant to avoid visual contact, and the participants were told that the experimenter could not see their pain scores once the experiment started. The auditory stimuli were presented using headphones (Philips Hi Fi Stereo headphones SH P8900) at an individual comfortable sound intensity level which remained constant throughout the experiment. The auditory and thermal stimuli were presented and controlled by a computer using

132 Presentation software (Version 14.0, The individually measured painful temperature was kept constant. The paradigm included six conditions: a control (Noise), an active distraction task (PASAT), two environmental sounds (Rain, Water), and two musical excerpts (Music1, Music2).We chose two sounds and two music pieces because: First, having more than one auditory stimuli of each condition type increases statistical power allowing the study of the effects of emotional nuances. Second, the two auditory stimuli were used to induce a placebo effect discussed in another paper (Garza-Villarreal, et al. in preparation). Each condition lasted 300 seconds (5 minutes) for a total time of 30 minutes per run, two runs per subject, separated by one minute of rest (Fig 2). During each condition, the auditory stimuli were presented for the first 140 s where the participants listened passively (or actively for the PASAT). During the following 160 s, they received four consecutive painful thermal stimuli. After each painful stimulus, the participants had 20 seconds to rate it for intensity and unpleasantness. The conditions were randomized, making sure the two environmental sounds (Rain/Water) and the music pieces (Music1/Music2) did not follow each other. After the experiment, the participants rated the auditory stimuli using the Valence, Liking and Arousal (VLA) and reported that the music pieces were unfamiliar to them. Statistical analysis The statistical analysis was performed using SPSS version 17.0 (SPSS Inc., Chicago IL). We examined the Pain and Emotion ratings. The dependent variables were pain intensity (PI) and pain unpleasantness (PU). For the first analysis, we averaged the scores of Rain/Water into a new condition named Sounds, and the Music1/Music2 into Music to test for differences

133 between the four conditions: control, music, sound and PASAT. We then performed repeated measures ANOVA, with Condition (4 levels: Noise, PASAT, Sounds, Music) as within-subject factors, and Baron-Cohen (3 levels: Balanced, Empathizers, Systemizers), Emotional-Rational (E/R) (2 levels: Emotional, Rational), Gender (2 levels: Males, Females) as between-subject factors. In a second repeated-measures ANOVA, we investigated the subtler differences between conditions by replacing the 4-level within-subject Condition factor with a 6-level Condition factor including Noise, PASAT, Rain, Water, Music1 and Music2. We performed single pre-hoc contrasts using Noise as the contrasting condition. Pairwise contrasts were done to investigate differences within the other conditions and between groups. Type I errors were controlled for by using Mauchly s test and the Greenhouse Geisser epsilon when appropriate. The alpha level for all statistical analysis was.05, unless stated otherwise. Results Analgesic effect of the passive stimuli Table 1 and Fig 3 show the descriptive statistics of each condition. In the first repeated-measures ANOVA we obtained a significant main effect of Condition (see Table 2) indicating that the pain ratings were significantly different for each condition and in particular that the unpleasantness and intensity pain ratings of PASAT, Sounds and Music were significantly lower than Noise. In the second repeated-measures ANOVA, there was also a difference between Conditions (see Table 3), with lower intensity and unpleasantness pain ratings in all conditions as compared with Noise, except for Rain which did not differ in PU ratings from Noise. Studying

134 the pairwise comparisons only between passive conditions, we found that whereas PI ratings of Rain, Water, Music1 and Music2 did not differ from each other, PU ratings differed between Rain and Music1 (p=.005), and between Music1 and Music2 (p <.001). Hence, among the sound conditions, Rain and Music2, characterized by more negative valence (Table 1 and Fig 3), had the worst analgesic effects, and Music1, characterized by the most positive valence of the sound conditions, had the best effect. Analgesic effect of the active stimulus In the first repeated-measures ANOVA the pairwise comparisons showed that the PASAT pain ratings were significantly lower than Sounds and Music (Table 2), meaning the PASAT reduced pain perception more than the other conditions. In the second repeated-measures ANOVA the pairwise comparisons showed significant differences on PI between PASAT and all the conditions (Table 3). For PU, pairwise comparison showed significant differences between PASAT and Noise, Rain, Water, Music, Music2. In sum, the PASAT task produced strongest analgesic effect compared with the other conditions. Emotion Table 1 and Fig 3 show the descriptive statistics of the emotions. In the first repeated-measures ANOVA the pairwise comparison showed no significant difference between Sounds and Music (PI: p =.473, PU: p =.810) (Table 2). This means that both conditions had similar pain ratings.

135 In the second repeated-measures ANOVA the pairwise comparisons showed no significant difference between the Rain, Water, Music1 and Music2 for PI (Table 3). However, it showed significant differences for PU of Rain vs Music1, p=.005, and Music1 vs Music2, p <.001. Rain was no different than Noise. Water, Music1 and Music2 had lower pain ratings than Noise. In sum, Sounds and Music conditions reduced pain in a similar way, except for the Rain and Music2 which had the worst effect and Music1 which had the strongest effect. Cognitive style In the first repeated-measures ANOVA, we did not find any main effects of the cognitive style and gender on the pain ratings in the Baron-Cohen, Emotional-Rational, and Gender between-subjects factors. However, we observed significant interactions for Condition x Baron- Cohen only for PI (Table 2) and for both PI and PU for Condition x Gender x Baron-Cohen. The first interaction suggests that the groups characterized by different cognitive styles based on the Baron-Cohen questionnaire rated pain differently depending on the condition: More precisely, the Systemizers (irrespectively of their gender) perceived less pain intensity during PASAT than the Balanced and Empathizers. The second interaction shows that the pain ratings differed for each condition depending on the gender and cognitive style of the participants: male Systemizers had lower PI ratings in the PASAT than the female Systemizers and the Empathizers and Balanced participants; however, they rated with higher scores the PU for all conditions except PASAT. Female Empathizers reported higher PI and PU ratings of all conditions except PASAT. The second analysis replicated the results of the first one (Fig 4), with significant interactions for Condition x Baron-Cohen only for PI and for Condition x Gender x Baron- Cohen.

136 Discussion The passive tasks of listening to music and environmental sounds as well as the active task of performing mental arithmetic (PASAT) reduced pain as compared with a noise control condition. Moreover, the mental arithmetic task produced the best analgesic effects compared with music and environmental sounds. In turn, the analgesic effects of music and environmental sounds were similar when they shared comparable levels of valence, liking and arousal. This suggests emotion plays a key role in music and sound induced analgesia. The cognitive style and gender of the participants influenced the analgesic effects of the primary tasks, including music. Systemizers perceived less pain during the PASAT, whereas the Female Empathizers reported more pain during all conditions except the PASAT. In general, these results suggest that distraction, emotion and individual differences in cognitive style and gender, and not the stimulus itself, should be taken into account as central mechanisms in sound-induced analgesia. Distraction PASAT, the active distraction task, was the condition obtaining the lowest pain ratings compared with the passive listening conditions. This finding is discrepant with the observations by Mitchel et al. 14, who showed that music had better analgesic effects than PASAT. However, there are several differences in the experimental design of our study that may explain this. First, they elicited pain using cold pressor, a technique that is thought to emulate chronic pain, whereas we used localized heat eliciting acute pain 35. Moreover, although in their study they measured both pain tolerance and intensity, they only found a difference in pain tolerance and not in pain intensity. The analgesic effect of distraction could vary depending on the type and duration of the

137 pain. Second, in Mitchell et al. the music was self-chosen and highly liked. Familiarity with the music piece could possibly produce more distraction and perceived control, hence more pain relief 15, than the unfamiliar music selected by the experimenter in our study. The analgesic effect of the PASAT can be considered to reflect how distraction reduces pain 10, 12. In music, distraction might be induced by high valence, high liking and low arousal. These emotional elements could provide strong entrainment and deviate attention away from the noxious stimulus. An alternative explanation for the analgesic mechanism involved in the PASAT condition could be stress-induced analgesia (SIA), where exposure to a stressful stimulus suppresses pain 36. Performing mental arithmetic while receiving and rating pain may provide enough stress to elicit this survival mechanism, which the Mozart music in the present study would not. Therefore, we suggest that the analgesic mechanisms activated by PASAT as a task condition, are distraction and stress, which in the present study outperformed the effects of passive listening to unfamiliar music and environmental sounds. Our findings highlight the need in future studies for controlling the actual amount of distraction effects, without assuming a priori higher distraction in active tasks than in passive listening conditions. Emotion Our results indicate that environmental sounds and Mozart music reduced pain perception more than the control noise condition, and therefore had an analgesic effect. However, we did not find any significant difference between analgesic effect of environmental sounds and music overall. It may be that the perceived emotions of the stimuli played a bigger role than whether their informational content, namely whether the stimuli consisted of music or environmental sounds. A detailed analysis of each condition showed slight differences between the pain ratings

138 of Music1 and Music2. Music1 had the lowest pain ratings of the four passive conditions, whereas Music2 had the highest pain ratings. On the other hand, unlike Water, Rain did not significant effect on pain. These findings may be explained by the differences between stimuli in their emotional ratings of valence and arousal, since it is known that these emotional dimensions influence pain. Rain was rated lowest in valence and liking as compared with the other three auditory stimuli (Music1, Music2, and Water). Overall, this suggests that emotional perception of the stimuli greatly influences pain perception, rather than stimuli themselves. In other words, it is not the sound or music per se, but the emotional value we attribute to it. These findings are consistent with the Roy et al. 1 study in which it was found that positive emotional valence contributes to the analgesic effect of music. The current emotional state can indeed influence the autonomic nervous system and modulate pain perception Several studies report that pleasant stimuli reduce pain, whereas unpleasant stimuli increase pain 19, 27, 40, 45. Music elicits strong emotions 46-47, which can be pleasant or unpleasant depending on different elements such as the listener s personal preference, sound intensity, and external context Besides the connection between valence and liking, music with low arousal is generally perceived as more pleasant 51. Several studies show relaxing music reduces pain perception 5, 21. However, the similarity between the effects of environmental sounds and Mozart music in our study could also be explained by unfamiliarity of the stimuli. Some studies suggest that self-chosen music has a more significant impact on the amount of pain relief, possibly due to familiarity, nostalgia and perceived subjective control Familiarity by itself could lead to more distraction, which in our study seems to be of more importance for pain relief, as reflected by the superior analgesic effect of the PASAT.

139 Cognitive style In our study, Systemizers perceived less pain during mental arithmetic. Systemizing is the drive to analyze the variables, derive the underlying rules that govern the behavior of a system and to control and construct them 33-34, 52. Thus, Systemizers might find the mental arithmetic more entraining and distracting than the other personalities. This could explain the reduction in their physical pain perception in the PASAT condition. On average, males spontaneously systemize to a greater degree than do females 34. Our results show that Female Empathizers reported higher pain intensity and pain unpleasantness in all conditions, except the PASAT. Empathizing is the drive to identify another person s emotions and thoughts, and to respond to these with an appropriate emotion, to predict behavior and to care about the feelings of others. On average, females spontaneously empathize to a greater degree than do males 33-34, 52. In contrast, however, the Baron-Cohen Empathizer-Systemizer and the Emotional- Rational questionnaires did not have any significant effect on pain under the passive listening conditions. This contradicts previous findings that the cognitive styles determine individual strategies to music listening 28. Thus, the Baron-Cohen questionnaire, that was created to assess individuals with autistic traits, may not clearly reflect important characteristics related to the analgesic effect of sounds and music. Gender alone did not have any effect on pain perception either. Gender differences in pain perception have been reported in several studies 17, 29. Nevertheless, the effect of gender varies in each study and no final conclusions can be made as of yet 53. In summary, unfamiliar Mozart music is not superior in relieving pain than environmental sounds when controlled for distractibility of the stimulus, emotional connotations, familiarity, and the cognitive style of the participants. The results also show that cognitive style

140 influences the analgesic effects of the task conditions, including music. However, females and males did not differ in their pain perception during the music and sound conditions. Hence, in designing music for analgesic purposes these factors should be taken into consideration. It is noteworthy that an active task with mental arithmetic had the best analgesic effect than passive listening to music and environmental sounds, especially for male Systemizers. Nevertheless, in the clinical context, music used as an analgesic adjuvant or treatment would be preferable over mental arithmetic, as the PASAT would be highly arousing and stressful for the patient. Furthermore, the PASAT task is highly dependent on individual cognitive abilities and mental state and could be unfeasible by a patient whereas music listening is affordable by and pleasant to everybody. Future studies should aim at using neuroimaging methods, such as fmri, to further understand the neural mechanisms behind the analgesic effects of secondary tasks like music and environmental sound listening. Acknowledgements We would also like to thank the people to help made this possible: Arne Møller for his help and support; Joshua Skewes and Daniel Campbell-Meiklejohn for their invaluable help programming the paradigm; Else-Marie Jegindø, Nanna Brix Finnerup, Troels Staehelin Jensen, and rest of the Danish Pain Research Center for their insight, advices and for allowing the use of the thermode. Finally, thanks to Pierre Rainville for his input. Figure Legends

141 Figure 1: Auditory stimuli presented in the experiment. The left column shows the stimulus type, the central column shows the stimulus name, and the right column shows an image representing each stimulus. Figure 2: The experimental paradigm. a: Example of a condition. Here we show the Control condition, with noise as the auditory stimulus and the four pain/thermal stimuli. b: A run, consisted of six conditions of five minutes each, for a total of 30 minutes. The full experiment consisted of two runs. Figure 3: Mean pain and emotion ratings. The X axis shows the conditions and the Y axis shows the mean scores for all participants. a: Pain ratings of the first statistical analysis, where Rain and Water scores are averaged into Sounds, and Music1 and Music2 are averaged into Music. b: Emotion ratings of the first statistical analysis. c: Pain ratings of the second statistical analysis. d: Emotion ratings of the second statistical analysis. Figure 4: Mean pain ratings divided into groups according to the Baron-Cohen questionnaire. The X axis shows all conditions, and the Y axis shows the mean pain ratings. The top graph represents pain intensity ratings, whereas the bottom graph represents pain unpleasantness ratings. Table Legends Table 1: Descriptive statistics of the pain and emotion ratings. SD = Standard deviation, PI = Pain intensity, PU = Pain unpleasantness. Table 2: The table shows the results of the first repeated-measures ANOVA. Within-subject factors = Condition, between-subject factors = Baron-Cohen, Emotional-Rational and Gender. n.s. = Not significant.

142 Table 3: The table shows the results of the second repeated-measures ANOVA. Within-subject factors = Condition, between-subject factors = Baron-Cohen, Emotional-Rational and Gender. n.s. = Not significant. Disclosures This study was supported by Ulla og Mogens Folmer Andersen s Fond and The Danish National Research Foundation. The authors declare that there are no conflicts of interest. References 1. Roy M, Peretz I, Rainville P: Emotional valence contributes to music-induced analgesia. Pain 134: Allred KD, Byers JF, Sole ML: The effect of music on postoperative pain and anxiety. Pain Manag Nurs 11: Podder L: Effects of music therapy on anxiety levels and pain perception. Nurs J India 98: Klassen JA, Liang Y, Tjosvold L, Klassen TP, Hartling L: Music for pain and anxiety in children undergoing medical procedures: a systematic review of randomized controlled trials. Ambul Pediatr 8: Nilsson U: The anxiety- and pain-reducing effects of music interventions: a systematic review. AORN J 87: Laopaiboon M, Lumbiganon P, Martis R, Vatanasapt P, Somjaivong B: Music during caesarean section under regional anaesthesia for improving maternal and infant outcomes. Cochrane Database Syst Rev:CD Mitchell LA, MacDonald RAR, Knussen C, Serpell MG: A survey investigation of the effects of music listening on chronic pain. Psychology of Music 35: Huang ST, Good M, Zauszniewski JA: The effectiveness of music in relieving pain in cancer patients: a randomized controlled trial. Int J Nurs Stud 47: Miron D, Duncan GH, Bushnell MC: Effects of attention on the intensity and unpleasantness of thermal pain. Pain 39: Tracey I, Ploghaus A, Gati JS, Clare S, Smith S, Menon RS, et al.: Imaging attentional modulation of pain in the periaqueductal gray in humans. J Neurosci 22: Brooks JC, Nurmikko TJ, Bimson WE, Singh KD, Roberts N: fmri of thermal pain: effects of stimulus laterality and attention. Neuroimage 15: Villemure C, Bushnell MC: Cognitive modulation of pain: how do attention and emotion influence pain processing? Pain 95:

143 Kenntner-Mabiala R, Andreatta M, Wieser MJ, Muhlberger A, Pauli P: Distinct effects of attention and affect on pain perception and somatosensory evoked potentials. Biol Psychol 78: Mitchell LA, MacDonald RA, Brodie EE: A comparison of the effects of preferred music, arithmetic and humour on cold pressor pain. Eur J Pain 10: Mitchell LA, MacDonald RA: An experimental investigation of the effects of preferred and relaxing music listening on pain perception. J Music Ther 43: Rhudy JL, Williams AE, McCabe KM, Russell JL, Maynard LJ: Emotional control of nociceptive reactions (ECON): do affective valence and arousal play a role? Pain 136: Kenntner-Mabiala R, Gorges S, Alpers GW, Lehmann AC, Pauli P: Musically induced arousal affects pain perception in females but not in males: a psychophysiological examination. Biol Psychol 75: Williams AE, Rhudy JL: Emotional modulation of autonomic responses to painful trigeminal stimulation. Int J Psychophysiol 71: de Tommaso M, Sardaro M, Livrea P: Aesthetic value of paintings affects pain thresholds. Conscious Cogn 17: White JM: Effects of relaxing music on cardiac autonomic balance and anxiety after acute myocardial infarction. Am J Crit Care 8: Good M, Anderson GC, Ahn S, Cong X, Stanton-Hicks M: Relaxation and music reduce pain following intestinal surgery. Research in Nursing & Health 28: Ploner M, Lee MC, Wiech K, Bingel U, Tracey I: Prestimulus functional connectivity determines pain perception in humans. Proc Natl Acad Sci U S A 107: Barrett FS, Grimm KJ, Robins RW, Wildschut T, Sedikides C, Janata P: Music-evoked nostalgia: affect, memory, and personality. Emotion 10: Brattico E, Jacobsen T: Subjective appraisal of music: neuroimaging evidence. Ann N Y Acad Sci 1169: Taenzer P, Melzack R, Jeans ME: Influence of psychological factors on postoperative pain, mood and analgesic requirements. Pain 24: Zelman DC, Howland EW, Nichols SN, Cleeland CS: The effects of induced mood on laboratory pain. Pain 46: Berna C, Leknes S, Holmes EA, Edwards RR, Goodwin GM, Tracey I: Induction of depressed mood disrupts emotion regulation neurocircuitry and enhances pain unpleasantness. Biol Psychiatry 67: Kreutz G, Schubert E, Mitchell LA: Cognitive Styles of Music Listening. Music Perception 26: Rhudy JL, Williams AE: Gender differences in pain: do emotions play a role? Gend Med 2: Price DD, Milling LS, Kirsch I, Duff A, Montgomery GH, Nicholls SS: An analysis of factors that contribute to the magnitude of placebo analgesia in an experimental paradigm. Pain 83: Scott J, Huskisson EC: Graphic representation of pain. Pain 2: Gronwall DM: Paced auditory serial-addition task: a measure of recovery from concussion. Percept Mot Skills 44: Wakabayashi A, Baron-Cohen S, Uchiyama T, Yoshida Y, Kuroda M, Wheelwright S: Empathizing and systemizing in adults with and without autism spectrum conditions: cross-cultural stability. J Autism Dev Disord 37: Baron-Cohen S: Autism: the empathizing-systemizing (E-S) theory. Ann N Y Acad Sci 1156: Mitchell LA, MacDonald RAR, Brodie EE: Temperature and the cold pressor test. The Journal of Pain 5: Butler RK, Finn DP: Stress-induced analgesia. Prog Neurobiol 88: Barrett LF, Mesquita B, Ochsner KN, Gross JJ: The Experience of Emotion. Annual Review of Psychology 58:

144 Rhudy JL, McCabe KM, Williams AE: Affective modulation of autonomic reactions to noxious stimulation. Int J Psychophysiol 63: Rainville P, Bao QV, Chretien P: Pain-related emotions modulate experimental pain perception and autonomic responses. Pain 118: Rhudy JL, Williams AE, McCabe KM, Nguyen MA, Rambo P: Affective modulation of nociception at spinal and supraspinal levels. Psychophysiology 42: Williams AE, Rhudy JL: Supraspinal modulation of trigeminal nociception and pain. Headache 49: Wiech K, Farias M, Kahane G, Shackel N, Tiede W, Tracey I: An fmri study measuring analgesia enhanced by religion as a belief system. Pain 139: Wiech K, Tracey I: The influence of negative emotions on pain: behavioral effects and neural mechanisms. Neuroimage 47: Rhudy JL, Meagher MW: The role of emotion in pain modulation. Curr Opin Psychiatr 14: Rhudy JL, Williams AE, McCabe KM, Rambo PL, Russell JL: Emotional modulation of spinal nociception and pain: the impact of predictable noxious stimulation. Pain 126: Blood AJ, Zatorre RJ: Intensely pleasurable responses to music correlate with activity in brain regions implicated in reward and emotion. Proc Natl Acad Sci U S A 98: Goldstein A: Thrills in response to music and other stimuli. Physiological Psychology 8: Roy M, Mailhot JP, Gosselin N, Paquette S, Peretz I: Modulation of the startle reflex by pleasant and unpleasant music. Int J Psychophysiol 71: Juslin PN, Vastfjall D: Emotional responses to music: the need to consider underlying mechanisms. Behav Brain Sci 31:559-75; discussion Brattico E, Jacobsen T, De Baene W, Glerean E, Tervaniemi M: Cognitive vs. affective listening modes and judgments of music - An ERP study. Biol Psychol. 51. Green AC, Baerentsen KB, Stodkilde-Jorgensen H, Wallentin M, Roepstorff A, Vuust P: Music in minor activates limbic structures: a relationship with dissonance? Neuroreport 19: Baron-Cohen S, Richler J, Bisarya D, Gurunathan N, Wheelwright S: The systemizing quotient: an investigation of adults with Asperger syndrome or high-functioning autism, and normal sex differences. Philos Trans R Soc Lond B Biol Sci 358: Fillingim RB, King CD, Ribeiro-Dasilva MC, Rahim-Williams B, Riley JL, 3rd: Sex, gender, and pain: a review of recent clinical and experimental findings. J Pain 10:

145 Figure 1 CONTROL - Noise DISTRACTION - PASAT SOUNDS { Rain Water MUSIC { Music1 Music2

146 Figure min 10 min 15 min 20 min 25 min 30 min a pain rest pain rest pain rest pain 20 s 20 s 20 s 20 s 140 s s b AUDITORY STIMULUS AUDITORY + THERMAL STIMULUS x2

147 Figure 3 a Intensity Unpleasantness c Intensity Unpleasantness Pain score (VAS) Pain score (VAS) Noise PASAT Sounds Music 0 Noise PASAT Rain Water Music1 Music2 b 10 d Valence Liking Arousal 9 8 Valence Liking Arousal 7 7 Emotion score Emotion score Noise PASAT Sounds Music 0 Noise PASAT Rain Water Music1 Music2

148 100 Figure 4 80 Pain intensity score Balanced Empathizers Systemizers Noise PASAT Rain Water Music1 Music2 100 Pain unpleasantness score Balanced Empathizers Systemizers Noise PASAT Rain Water Music1 Music2

149 Table 1 Table 1 Mean (+SD) ratings of pain and emotion for all conditions. PAIN EMOTION PI PU Valence Liking Arousal Analysis 1 Noise (+16.48) (+20.09) 3.9 (+1.90) 3.2 (+2.43) 5.0 (+2.43) PASAT (+17.97) (+21.05) 4.5 (+2.03) 5.0 (+2.63) 7.4 (+2.63) Sounds (+16.98) (+20.66) 7.3 (+2.04) 7.4 (+1.94) 3.6 (+2.35) Music (+16.60) (+20.53) 7.8 (+1.85) 7.5 (+1.97) 4 (+2.55) Analysis 2 Rain (+16.93) (+21.27) 6.7 (+1.98) 6.9 (+2.01) 3.8 (+2.02) Water (+17.19) (+20.16) 7.9 (+1.96) 7.9 (+2.64) 3.3 (+1.72) Music (+16.25) (+19.62) 8.0 (+1.96) 7.8 (+2.86) 4.0 (+2.00) Music (+17.06) (+21.41) 7.5 (+1.70) 7.1 (+2.24) 4.0 (+1.91)

150 Table 2 Table 2 1 st Analysis results (repeated-measures ANOVA). Pain Intensity Pain Unpleasantness Analgesic Effect Condition F (2.04, 73.54) = 34.29, p <.001 F (1.81, 65.16) = 17.64, p <.001 Contrasts Noise vs. PASAT F (1, 36) = 51.35, p <.001 F (1, 36) = 32.38, p <.001 Noise vs. Sounds F (1, 36) = 8.11, p <.01 F (1, 36) = 32372, p <.001 Noise vs. Music F (1, 36) = 8.60, p <.01 F (1, 36) = 12.34, p <.001 Distraction Pairwise comparisons PASAT vs. Sounds p <.001 p <.001 PASAT vs. Music p <.001 p <.01 Emotion Pairwise comparisons Sounds vs. Music n. s. n. s. Cognitive style Baron-Cohen E/S n. s. n. s. Emotional-Rational n. s. n. s. Gender n. s. n. s. Interactions Condition x Baron-Cohen E/S Condition x Gender x Baron-Cohen E/S F (4, 73.54) = 2.82, p <.05 n. s. F (4, 73.54) = 4.86, p <.001 F (3.6, 65.16) = 4.38, p <.01

151 Table 3 Table 3 2 nd Analysis results (repeated-measures ANOVA). Pain Intensity Pain Unpleasantness Analgesic Effect Condition F (3.37, ) = 22.90, p <.001 F (3.41, ) = 11.45, p <.001 Contrasts Noise Vs PASAT F (1, 36) = 51.35, p <.001 F (1, 36) = 32.38, p <.001 Noise vs. Rain n. s. n. s. Noise vs. Water F (1, 36) = 14.03, p <.001 F (1, 36) = 17.45, p <.001 Noise vs. Music1 F (1, 36) = 10.00, p <.01 F (1, 36) = 17.48, p <.001 Noise vs. Music2 F (1, 36) = 5.47, p <.05 F (1, 36) = 6.37, p =.01 Distraction Pairwise comparisons PASAT vs. Rain p <.001 p <.001 PASAT vs. Water p <.001 p <.01 PASAT vs. Music1 p <.001 p <.05 PASAT vs. Music2 p <.001 p <.001 Emotion Pairwise comparisons Rain vs. Water n. s. n. s. Rain vs. Music1 n. s. p <.01 Water vs. Music1 n. s. n. s. Water vs. Music2 n. s. n. s. Music1 vs. Music2 n. s. p <.001 Cognitive style Baron-Cohen E/S n. s. n. s. Emotional-Rational n. s. n. s. Gender n. s. n. s. Interactions Condition x Baron-Cohen E/S Condition x Gender x Baron-Cohen E/S F (6.75, ) = 2.43, p <.05 n. s. F (6.74, ) = 4.05, p <.001 F (6.81, ) = 3.126, p <.01

152

153 1 The placebo effect in music-induced analgesia Eduardo A. Garza Villarreal 1,2,*, Elvira Brattico 4, Lene Vase 1,3, Leif Østergaard 1,5, Peter Vuust 1,2 1 Center for Functionally Integrative Neuroscience, University of Aarhus, Denmark 2 Royal Academy of Music, Aarhus, Denmark 3 Department of Psychology, University of Aarhus, Denmark 4 Cognitive Brain Research Unit, Department of Psychology, University of Helsinki 5 Department of Neuroradiology, Aarhus University Hospital, Denmark Keywords: Music, placebo analgesia, pain, emotion Abbreviated title: The placebo effect in music-induced analgesia. *Corresponding Author: Eduardo Adrian Garza Villarreal Center for Functionally Integrative Neuroscience, Danish Neuroscience Center, Aarhus University Hospital, Nørrebrogade 44 Building 10G , Aarhus C, Denmark Mobile: Work: Fax: [email protected], [email protected] URL:

154 2 Abstract Music is known to have analgesic effects. However, the underlying mechanisms are still largely unknown. As placebo effects are known to exist in relation to analgesia, it would be of interest to know to which extent placebo effects contribute to music induced analgesia and to know the cognitive and emotional factors that may mediate this effect. Forty eight participants listened to Mozart music (active condition), sounds (placebo condition) and noise (control condition) while they received and rated pain intensity and pain unpleasantness on VAS scales. Verbal suggestions of pain relief were given for the active and placebo conditions. Participants rated valence, arousal and liking as well as perceived emotional intensity of the auditory stimuli. Music and sound both reduced pain intensity and unpleasantness significantly more than noise but there was no significant difference in the pain relieving effect of music and sound. The verbal suggestions for pain relief did not influence pain ratings and this was independent of whether participants believed the verbal suggestions or not. The magnitude of the placebo analgesia appeared to be larger in relation to auditory stimuli that were perceived as happy as opposed to stimuli that were perceived as sad. These findings suggest that the analgesic effect of music may be partly a placebo effect influenced by emotional feelings but not verbal suggestions.

155 3 Introduction Recent studies show that pleasant and relaxing music reduces pain significantly [8,13-14,24,26]. The proposed mechanisms behind the analgesic effect of music are distraction and emotion [14,20,24] (Garza-Villarreal et al, submitted). However, there is an analgesic mechanism which has yet to be investigated: the placebo effect. Research of pharmacologically induced analgesia, the drug under study is usually tested up against a placebo agent in order to see whether there is a specific pharmacological effect of the drug. In relation to music induced analgesia, it is more challenging to choose a placebo control. Nevertheless, sounds share many of the same qualities as music without having the specific musical components [12,22]. Therefore, sound may act as a placebo. To separate placebo analgesia from natural fluctuations it is important to test them up against a no treatment control condition [21]. Therefore, sound should be tested up against noise as it is auditory input without positive or negative emotions. In this manner, the specific analgesic effect of music can be calculated as the difference in pain between the music and sound conditions, and the placebo analgesic can be calculated as the difference between pain in the sound and the noise conditions. Different mechanisms in placebo analgesia have been proposed. Placebo analgesia has been shown to be influenced by verbal suggestions [15-16] and several studies have shown that the stronger the verbal suggestion the larger the effect [4,21,27]. However, there have also been studies showing that not all placebo effects are influence by verbal suggestion [2]. Benedetti and colleagues have, for example, shown that verbal suggestion does not influence hormonal

156 4 secretion, such as increase and decrease of growth hormone and cortisol. So far, it is not known whether verbal suggestions influence placebo effects in relation to auditory stimuli. Recently, studies have also started to demonstrate that emotional feelings may be central to the placebo effect [18,28]. Petrovic et al showed activation of the rostral anterior cingulate cortex (racc) and the orbitofrontal cortex (OFC) correlated with placebo analgesia. Vase et al showed that ratings for expected pain levels, desire for pain relief and anxiety decreased, correlating with the placebo effect. However, in this study the effect was not mediated by opioids, suggesting that a reduction in negative emotions contributed to the placebo effect. Neuroimaging studies have also shown that dopaminergic responses and activation of regions related to emotion, such as the orbitofrontal cortex (OFC), insular cortex (IC) and rostral anterior cingulate cortex (racc), are involved in the placebo effect [18-19,25,29-30]. However, at this point it is not known to which extent emotional feelings contribute to placebo effects in relation to music. Given the general influence of emotion on music perception [6,9,23], it would be relevant to test this. Furthermore, as a reduction in negative emotions have been primarily related to placebo effects, it would be important to test if auditory stimuli that are associated with either positive emotional feelings (e.g. feeling happy) or negative emotional feelings (e.g. feeling sad), give rise to different magnitudes of placebo analgesia. In this study the placebo analgesia was investigated in relation to music by having healthy volunteers listen to Mozart music (active), environmental sounds (placebo) and noise (control) conditions while they received and rated pain intensity and pain unpleasantness on VAS scales. Verbal suggestions for pain relief were given in the music and sound conditions to test if they would increase the magnitude of active and placebo analgesia and perceived emotional intension of the auditory stimuli was rated. We hypothesize that placebo analgesia was

157 5 present in the music, and that it was mediated by verbal suggestions of pain relief and positive emotional feelings. Materials and Methods Participants Forty eight participants (24 male, 24 female), aged between 19 and 39 years (mean = 24, SD = 4) took part in the experiment. All of them were native Danish speakers. They were recruited placing ads around the university and in a research recruitment website. The ad specifically mentioned the use of Mozart music in a pain experiment to induce expectation. The participants were healthy, right handed, they reported normal hearing and had minimum to none musical training. They had not consumed any analgesic medication in the 24 h prior to the experiment. Written informed consent was received from all participants. Participants received compensation for taking part in the experiment. Ethical permission was obtained from The Research Ethical Committee for Mid-Jutland Region, Denmark. Thermal Stimuli and pain measures The thermal stimuli was produced by a 3 x 3 cm contact thermode (Pathway model ATS from Medoc Ltd. Advanced Medical System, Israel) on the forearms. The pain limits and threshold were investigated for each participant. We performed calibration trials to control for individual differences in pain perception. The method was similar to that of Price et al [21], with modifications. We presented two trials with four different temperatures: 42, 43, 45, and 47 o C in random order. Each stimulus was separated by approximately 15 20s, and the participants rated pain intensity and unpleasantness in the Visual Analog Scale (VAS) for each

158 6 temperature. The goal temperature had to be below 47 o C and reflect pain ratings between mm (moderate to high) in the VAS. This goal temperature was kept constant during the entire experiment to avoid high variability of the VAS scores. To avoid habituation, the thermode was changed to a slightly different skin location on the forearm after every two experimental conditions. Each painful stimulation consisted of a plateau of 16 s with a rise/fall time of 2 s. The baseline temperature was 35 C. The thermal stimulus was measured using the VAS for pain intensity and unpleasantness. The scale ranged from No Pain (left end of the scale) to Very Intense or Very Unpleasant (right end). Auditory Stimuli We used three different 5-min auditory stimuli: Music, Sounds and Noise. The music pieces were different Mozart string compositions ( String Quartet No. 1 in G major, K. 80/73f (1770) Adagio and Divertimento in E flat, K. 563 Adagio ). The sounds (Rain and Water) were recordings from nature and the environment (Sound effects library, Sound Ideas These auditory stimuli were chosen for their matching high valence and liking, and low arousal ratings. Each auditory stimulus lasted 300 s (5 min) and was normalized. Verbal Suggestions To determine if there was an effect of verbal suggestion, we made two different false suggestions to each participant. The first suggestion was about the Sounds: According to our previous studies, we have found that the sound of (Rain or Water) is better to reduce pain than (the opposite sound). The second suggestion was about the Music: One of the musical pieces is

159 7 from Mozart and the other is from a composer called Salieri, when in reality both musical pieces were from Mozart. Furthermore, we added Mozart music has been shown to reduce pain, and this is also part of the Mozart effect, to get advantage of the popularity of this term. To balance the suggestions and because the participants listened to all sounds and music, we divided them into four groups depending on the combination of suggestions given: Rain-Music1 (Rain is better than Water, Music1 is Mozart and so forth), Water-Music1, Rain-Music2, Water-Music2. After the experiment, each participant was asked if they believed the suggestions they were given. Emotional Measures To determine the effects of emotion, after the experiment the participants rated the auditory stimuli in a different scale for valence (0 = unpleasant, 10 = pleasant), arousal (0 = relaxing, 10 = stimulating) and liking (0 = does not like, 10 = likes) (VLA), to corroborate the emotional ratings reported in the pilot study. They also rated the stimuli for Perceived Emotional Intention (PEI), a scale which rates 6 different emotions for each auditory stimulus: happiness, sadness, fear, anger, surprise and peacefulness. Each emotion was rated from 0 (not perceived) to 10 (extremely perceived) [10,24]. Procedure As an attempt to induce more suggestion in the participants, the recruitment announcement specifically informed them: In this study we wish to investigate if Mozart music reduces pain in people. Prior to the experiment the participants were informed about the details

160 8 about the thermal stimulation and to arrange the appointment for the experiment. When entering the laboratory, the participants were told their task was to rate the effects of different types of sounds and music while receiving pain. They were also told that the sound and the music could have an analgesic effect. The experiment took place in a sound proof white room without windows. Instructions for all participants were exactly identical and given by the same male experimenter, who was the only person present during the experiment. The participants were trained to use the VAS and were familiarized with the thermal stimuli by investigating pain limits and threshold. They seated comfortably in a chair in front of a monitor and were given a mouse to rate pain using a computerized VAS. To avoid confounds, a panel wall stood between the experimenter and the participant to avoid visual contact, and the participants were told the experimenter could not see their pain scores once the experiment started. The auditory stimuli were presented using headphones (Philips Hi Fi Stereo headphones SH P8900) at an individual comfortable sound intensity level which remained constant throughout the experiment. The auditory and thermal stimuli were presented and controlled by a computer using Presentation software (Version 14.0, The individually measured painful temperature was kept constant. The paradigm consisted of six conditions: A Control (Noise), an active distraction task (PASAT), two relaxing sounds (Rain, Water), and two musical excerpts (Music1, Music2). Each condition lasted 300 seconds for a total time of 30 minutes per run, two runs per subject, and one minute of rest in between. Each condition consisted on: an auditory stimulus playing for the whole duration of the condition, and the thermal stimuli. During the first 140 s, the participants listened to the auditory stimulus. During the following 160 s, they received four consecutive painful stimuli while they still listened to the auditory stimulus. After each painful stimulus, the

161 9 participants had 20 s to rate it for intensity and unpleasantness. The conditions were randomized, making sure the two relaxing sounds (Rain/Water) and the music (Music1/Music2) never appeared after the other. After the experiment, the participants rated the auditory stimuli using the Valence, Liking and Arousal (VLA) and the PEI questionnaires, and they indicated whether they believed the verbal suggestion or not. Analysis First, the three conditions: Music (active), sound (placebo) and noise (control) were compared to calculate to which extent there was a specific effect and a placebo effect in music induced analgesia. Second, we tested whether verbal suggestions increased the efficacy of active and placebo induced analgesia and whether participants believed the verbal suggestions given. Third, the contribution of emotional feelings to the efficacy of active and placebo was tested. The dependent variables were pain intensity (PI) and pain unpleasantness (PU) in each condition. The independent variables were the verbal suggestions and the emotional factors (VLA, PEI) The hypotheses were investigated using Repeated-measures ANOVAs. For the first analysis, the scores of Rain/Water were averaged into a new condition named Sounds, and the Music1/Music2 into Music to test only for differences between control, placebo and active conditions. Then Repeated Measures ANOVA were performed with within-subject factors: Condition (3) (Noise, Sounds, Music). Single pre-hoc contrasts were performed using Noise as the contrasting condition. Pairwise comparisons were done to investigate differences between Sounds and Music using Bonferroni correction. To investigate the influence of verbal suggestion and emotional feelings, we performed Repeated Measures ANOVAs with within-subject factors: Condition (4) (Rain, Water, Music1,

162 10 Music2), verbal Suggestion (2) (Yes, No), between-subject factor: Belief (Yes, No),Emotion (2) (Happy, Sad) For all statistical analyses, type I errors were controlled for by using Mauchly s test and the Greenhouse Geisser epsilon when appropriate. The alpha level for all statistical analysis was.05, unless stated otherwise.the statistical analysis was performed using SPSS version Results The first analysis, showed a significant effect of Condition: PI, F (1.66, 78.05) = 8.461, p <.001, and PU, F (1.78, 83.51) = , p <.001. This means the pain ratings were significantly different for each condition. For the PI, the contrasts showed significant effects of: Noise vs Sounds, F (1, 47) = 10.22, p <.01 and Noise vs Music, F (1, 47) = 10.64, p <.01. Pairwise comparison showed no significant differences between Sounds and Music (p = 1.00). For the PU, the contrasts showed significant effects for: Noise vs Sounds F (1, 47) = 30.83, p <.001, Noise vs Music, F (1, 47) = 16.19, p <.001. Pairwise comparison showed no significant differences between Sounds and Music (p <.001). These results show that music did not relieve pain significantly better than sound, but sound did relieve pain significantly better than noise. In order words, the pain relieving effect of music seem to be due to a placebo analgesia effect Verbal suggestion did not significantly increase the pain reliving effect of active music or placebo sounds (not significant). Similar results were found for participant who believed and did not believe the verbal suggestion (present results) so the lack of an effect for verbal suggestion did not seem to arise from participants not believing the suggestion. There was found significant interaction between suggestion and emotion for the happy condition (Water) F (1, 11) = 14.67,

163 11 p <.01. This means placebo analgesia was larger when the stimulus was perceived as happy versus when it was perceived as sad. Discussion In this study we investigated placebo effects in relation to music induced analgesia and the possible mechanisms behind it. Music and sounds induced a significant reduction of pain intensity and pain unpleasantness as compared to noise. However, music and sound did not differ in their analgesic effect. These results suggest that the analgesic effect of music may be due to placebo analgesia. Verbal suggestions of pain relief did not increase the magnitude of the placebo analgesia, not even for the participants who did believe the verbal suggestions. However, positive emotions perceived in the placebo conditions influenced the analgesic effect. This is the first study to show that placebo effects exist in relation to music induced analgesia. This finding is important for at least two reasons. First, it suggests that it may not be the music per se that gives rise to the pain relief seen in music induced analgesia. Second, it shows that placebo analgesia effects may not only exist in relation to pharmacological treatment [18], transmagnetic treatment [11] and neurological degeneration [1], but also in relation to music induced analgesia. To our surprise the placebo analgesia in relation to auditory stimuli were not influenced by verbal suggestions for pain relief. This is in contrast with previous studies showing that verbal suggestions increase placebo analgesia in relation to cream [5,21]. One explanation could be that verbal suggestions given in the current study were not efficient. However, as there were no differences in pain relief between the participants who believed the verbal suggestions and participants who did not believe the verbal suggestions, this does not seem to be an obvious explanation. Another explanation could be that verbal suggestions influence placebo effects

164 12 differently depending on the vehicle by which they are induced. Benedetti and colleagues have previously shown that not all placebo effects are influenced by verbal suggestions [2]. In their study, verbal suggestions influenced placebo effects in relation to pain and motor performance, but not in relation to hormonal secretion such as an increase or decrease in growth hormone and cortisol [2]. Similarly it could be relevant in future studies to investigate whether the efficacy of verbal suggestions depend on the vehicle by which the placebo effect is induced. Interestingly, the current study showed that placebo analgesia in relation to music was mediated by positive emotional feelings. Previous studies have primarily shown that placebo analgesia can be related to a decrease in negative emotions such as anxiety [17,28]. To our knowledge this is the first study to directly investigate how negative and positive emotions influence the placebo effect. The finding that positive emotional feelings are related to a larger magnitude of placebo analgesia than negative emotional feelings is in agreement with the general literature on emotion and pain. Because the analgesic mechanisms of music have been describe to be partly emotional [7,24], the placebo effect of music could be influence by emotion instead [3,18,25]. This is supported by our result where the happy placebo condition was affected by the suggestion made. Future imaging studies would help determine the neural correlates of the placebo mechanisms. Overall, placebo analgesia seems to be the main mechanism in analgesia induced by music. This effect is not influenced by suggestion however, it does seem to be mediated by emotion mechanisms. Acknowledgements

165 13 This study was supported by grants from the University of Aarhus and Aarhus University Hospital. The authors declare that there are no conflicts of interest. We would also like to thank the people to help made this possible: Arne Møller for his help and support. Joshua Skewes and Daniel Campbell-Meiklejohn for their invaluable help programming the paradigm. Else-Marie Jegindø, Nanna Brix Finnerup, Troels Staehelin Jensen, and rest of the Danish Pain Research Center for their insight, advices and allowing the use of the thermode. Finally, thanks to Pierre Rainville for his input. References [1]BenedettiF.Placeboanalgesia.NeurolSci2006;27Suppl2:S [2]BenedettiF,PolloA,LopianoL,LanotteM,VighettiS,RaineroI.Consciousexpectationand unconsciousconditioninginanalgesic,motor,andhormonalplacebo/noceboresponses.j Neurosci2003;23(10): [3]ChungSK,DonaldDP,VerneGN,MichaelER.Revelationofapersonalplaceboresponse:Itseffects onmood,attitudesandfutureplaceboresponding.pain2007;132(3): [4]DePascalisR,IwahashiM,TamuraM,PadlanEA,GonzalesNR,SantosAD,GiulianoM,SchuckP, SchlomJ,KashmiriSV.Graftingof"abbreviated"complementaritydeterminingregions containingspecificitydeterminingresiduesessentialforligandcontacttoengineeraless immunogenichumanizedmonoclonalantibody.jimmunol2002;169(6): [5]DePascalisV,MaguranoMR,BellusciA.Painperception,somatosensoryeventrelatedpotentials andskinconductanceresponsestopainfulstimuliinhigh,mid,andlowhypnotizablesubjects: effectsofdifferentialpainreductionstrategies.pain1999;83(3): [6]FritzT,JentschkeS,GosselinN,SammlerD,PeretzI,TurnerR,FriedericiAD,KoelschS.Universal recognitionofthreebasicemotionsinmusic.currbiol2009;19(7): [7]GarzaVillarrealE,BratticoE,VaseL,OstergaardL,VuustP.submitted. [8]HuangST,GoodM,ZauszniewskiJA.Theeffectivenessofmusicinrelievingpainincancerpatients:a randomizedcontrolledtrial.intjnursstud2010;47(11): [9]JuslinPN,VastfjallD.Emotionalresponsestomusic:theneedtoconsiderunderlyingmechanisms. BehavBrainSci2008;31(5):559575;discussion

166 14 [10]KenntnerMabialaR,GorgesS,AlpersGW,LehmannAC,PauliP.Musicallyinducedarousalaffects painperceptioninfemalesbutnotinmales:apsychophysiologicalexamination.biolpsychol 2007;75(1):1923. [11]KrummenacherP,CandiaV,FolkersG,SchedlowskiM,SchonbachlerG.Prefrontalcortexmodulates placeboanalgesia.pain2010;148(3): [12]LangnerG.Periodicitycodingintheauditorysystem.HearRes1992;60(2): [13]MitchellLA,MacDonaldRA.Anexperimentalinvestigationoftheeffectsofpreferredandrelaxing musiclisteningonpainperception.jmusicther2006;43(4): [14]MitchellLA,MacDonaldRA,BrodieEE.Acomparisonoftheeffectsofpreferredmusic,arithmetic andhumouroncoldpressorpain.eurjpain2006;10(4): [15]MontgomeryG,KirschI.TheeffectsofsubjectarmpositionandinitialexperienceonChevreul pendulumresponses.amjclinhypn1996;38(3): [16]MontgomeryGH,KirschI.Classicalconditioningandtheplaceboeffect.Pain1997;72(12): [17]MortonDL,WatsonA,ElDeredyW,JonesAK.Reproducibilityofplaceboanalgesia:Effectof dispositionaloptimism.pain2009;146(12): [18]PetrovicP,DietrichT,FranssonP,AnderssonJ,CarlssonK,IngvarM.Placeboinemotional processinginducedexpectationsofanxietyreliefactivateageneralizedmodulatorynetwork. Neuron2005;46(6): [19]PetrovicP,KalsoE,PeterssonKM,IngvarM.Placeboandopioidanalgesiaimagingashared neuronalnetwork.science2002;295(5560): [20]PlonerM,LeeMC,WiechK,BingelU,TraceyI.Prestimulusfunctionalconnectivitydeterminespain perceptioninhumans.procnatlacadsciusa2010;107(1): [21]PriceDD,MillingLS,KirschI,DuffA,MontgomeryGH,NichollsSS.Ananalysisoffactorsthat contributetothemagnitudeofplaceboanalgesiainanexperimentalparadigm.pain 1999;83(2): [22]RauscheckerJP.Corticalprocessingofcomplexsounds.CurrOpinNeurobiol1998;8(4): [23]RoyM,MailhotJP,GosselinN,PaquetteS,PeretzI.Modulationofthestartlereflexbypleasantand unpleasantmusic.intjpsychophysiol2009;71(1):3742. [24]RoyM,PeretzI,RainvilleP.Emotionalvalencecontributestomusicinducedanalgesia.Pain 2008;134(12): [25]ScottDJ,StohlerCS,EgnatukCM,WangH,KoeppeRA,ZubietaJK.Individualdifferencesinreward respondingexplainplaceboinducedexpectationsandeffects.neuron2007;55(2): [26]ShabanloeiR,GolchinM,EsfahaniA,DolatkhahR,RasoulianM.Effectsofmusictherapyonpain andanxietyinpatientsundergoingbonemarrowbiopsyandaspiration.aornj2010;91(6): [27]VaseL,PetersenGL,RileyJL,3rd,PriceDD.Factorscontributingtolargeanalgesiceffectsinplacebo mechanismstudiesconductedbetween2002and2007.pain2009;145(12):3644. [28]VaseL,RobinsonME,VerneGN,PriceDD.Increasedplaceboanalgesiaovertimeinirritablebowel syndrome(ibs)patientsisassociatedwithdesireandexpectationbutnotendogenousopioid mechanisms.pain2005;115(3): [29]WagerTD.Expectationsandanxietyasmediatorsofplaceboeffectsinpain.Pain2005;115(3): [30]WagerTD,ScottDJ,ZubietaJK.Placeboeffectsonhumanmuopioidactivityduringpain.ProcNatl AcadSciUSA2007;104(26):

167

168