1. Introduction
Of all the problems that may confront a music psychologist, none is perhaps more important than to explain listeners' reactions to music. Some kind of musical experience is the basis for every musical activity, regardless of whether it involves composing, performing, or listening to music. Several studies have suggested that the most common goal of musical experiences is to influence emotions: People use music to change emotions, to release emotions, to match their current emotion, to enjoy or comfort themselves, and to relieve stress (e.g., Behne Reference Behne, Deliége and Sloboda1997; Juslin & Laukka Reference Juslin and Laukka2004; Sloboda & O'Neill Reference Sloboda, O'Neill, Juslin and Sloboda2001; Zillman & Gan Reference Zillman, Gan, Hargreaves and North1997).
Yet, music's apparent ability to induce strong emotions is a mystery that has fascinated both experts and lay people at least since ancient Greece (Budd Reference Budd1985). “How do sounds, which are, after all, just sounds, have the power to so deeply move those involved with them?” (Reimer Reference Reimer2003, p. 73). To explain how music can induce emotions in listeners is all the more important since music is already used in several applications in society that presume its effectiveness in inducing emotions, such as film music (Cohen Reference Cohen, Juslin and Sloboda2001), marketing (Bruner Reference Bruner1990), and therapy (Bunt & Hoskyns Reference Bunt and Hoskyns2002).
However, despite a recent upswing of research on musical emotions (for an extensive review, see Juslin & Sloboda Reference Juslin and Sloboda2001), the literature presents a confusing picture with conflicting views on almost every topic in the field.Footnote 1 A few examples may suffice to illustrate this point: Becker (Reference Becker, Juslin and Sloboda2001, p. 137) notes that “emotional responses to music do not occur spontaneously, nor ‘naturally’,” yet Peretz (Reference Peretz, Juslin and Sloboda2001, p. 126) claims that “this is what emotions are: spontaneous responses that are difficult to disguise.” Noy (Reference Noy, Feder, Karmel and Pollock1993, p. 137) concludes that “the emotions evoked by music are not identical with the emotions aroused by everyday, interpersonal activity,” but Peretz (Reference Peretz, Juslin and Sloboda2001, p. 122) argues that “there is as yet no theoretical or empirical reason for assuming such specificity.” Koelsch (Reference Koelsch2005, p. 412) observes that emotions to music may be induced “quite consistently across subjects,” yet Sloboda (Reference Sloboda, Riederer and Lahti1996, p. 387) regards individual differences as an “acute problem.” Scherer (Reference Scherer and Bresin2003, p. 25) claims that “music does not induce basic emotions,” but Panksepp and Bernatzky (Reference Panksepp and Bernatzky2002, p. 134) consider it “remarkable that any medium could so readily evoke all the basic emotions.” Researchers do not even agree about whether music induces emotions: Sloboda (Reference Sloboda, Riess-Jones and Holleran1992, p. 33) claims that “there is a general consensus that music is capable of arousing deep and significant emotions,” yet Konečni (Reference Konečni, Juslin and Sloboda2003, p. 332) writes that “instrumental music cannot directly induce genuine emotions in listeners.”
At the heart of all this controversy, we believe, lies the fact that researchers have not devoted enough attention to the question of how music induces emotions. Most writers on the subject acknowledge that this is the most important issue: “Music arouses strong emotional responses in people, and they want to know why” (Dowling & Harwood Reference Dowling and Harwood1986, p. 202). Yet, a search of the literature reveals that surprisingly few articles make any attempt whatsoever to explain the psychological mechanisms that underlie listeners' emotional responses to music. For instance, a search for peer-reviewed articles (in English) in PsycINFO and RILM Abstracts of Music Literature, using the query music* and emotion* and the time limits 1967–2007, revealed 1,033 and 423 articles, respectively, of which a single article in PsycINFO (i.e., Steinbeis et al. Reference Steinbeis, Koelsch and Sloboda2006) and none of the articles in RILM aimed to empirically test a theory about how music induces emotions; 21 articles in each database (2% and 5%, respectively) mentioned a mechanism, or the issue of emotion induction more generally, without reporting any relevant data.Footnote 2 Although these searches may not have uncovered every relevant article, the point is that the great majority of studies of musical emotions have not concerned underlying mechanisms.
We use the term psychological mechanism broadly in this article to refer to any information processing that leads to the induction of emotions through listening to music.Footnote 3 This processing could be simple or complex. It could be available to consciousness or not. However, what the mechanisms discussed here have in common is that they become activated by taking music as their “object.” We adhere to the notion that a defining feature of emotions is that they involve intentional objects: They are “about” something (Frijda Reference Frijda, Kahneman, Diener and Schwarz1999, p. 191). For example, we are sad about the death of a loved one. What are musical emotions about?
One problem with musical emotions is that the conditions for eliciting emotions appear to be different from those in everyday life: In the paradigmatic case, an emotion is aroused when an event is appraised as having the capacity to affect the goals of the perceiver somehow (Carver & Scheier Reference Carver and Scheier1998). Thus, for example, a reviewer's criticism of a manuscript may threaten the author's goal to get it published. Because music does not seem to have any capacity to further or block goals, it seems strange that music can induce emotions. Indeed, it has been denied by some authors that music can induce common “everyday emotions” such as sadness, happiness, and anger (Kivy Reference Kivy1990; Konečni Reference Konečni, Juslin and Sloboda2003; Scherer Reference Scherer and Bresin2003). We suspect that this view rests on the assumption that such emotions need to reflect a cognitive appraisal (see Gabriel & Crickmore [Reference Gabriel and Crickmore1977], Scherer & Zentner [2001], Stratton & Zalanowski [Reference Stratton and Zalanowski1989; Reference Stratton and Zalanowski1991], and Waterman [Reference Waterman1996]) for claims about an important role of cognitive appraisal in emotional responses to music).
The main assumption of appraisal theory is that emotions arise, and are distinguished, on the basis of a person's subjective evaluation of an event on appraisal dimensions such as novelty, urgency, goal congruence, coping potential, and norm compatibility (for an excellent review, see Scherer Reference Scherer, Dalgleish and Power1999). Occasionally, music may lead to the induction of emotions through some of the same appraisal dimensions. Thus, for example, a person may be trying to sleep at night, but is prevented from doing so by the disturbing sounds of a neighbor playing loud music on his or her stereo. In this case, the music becomes an object of the person's irritation because it blocks the person's goal: to fall asleep. Although there is nothing particularly “musical” about this example, it is clear that music can sometimes induce emotions in listeners in this manner (Juslin et al., in press). Such responses can easily be explained by traditional theories of emotion. However, the problem is that the available evidence indicates that this type of emotion is not typical of music listening – most emotional reactions to music do not involve implications for goals in life, which explains why they are regarded as mysterious: “The listener's sad response appears to lack the beliefs that typically go with sadness” (Davies Reference Davies, Juslin and Sloboda2001, p. 37).
Because music does not seem to have goal implications, some researchers have assumed that music cannot induce emotions at all (Konečni Reference Konečni, Juslin and Sloboda2003) – or, at least, that it cannot induce basic emotions related to survival functions (Kivy Reference Kivy1990; Scherer Reference Scherer and Bresin2003).Footnote 4 Some researchers allow for the possibility that music may induce “more subtle, music-specific emotions” (Scherer & Zentner Reference Scherer, Zentner, Juslin and Sloboda2001, p. 381; see also Gurney Reference Gurney1880; Lippman Reference Lippman1953; Swanwick Reference Swanwick1985), the precise nature of which remains to be clarified. This notion is sometimes coupled with the assumption that musical emotions are induced through some unique (but yet unspecified) process that has little or nothing in common with the induction mechanisms of “ordinary” emotions. We reject these views on both theoretical and empirical grounds, and claim that music can induce a wide range of both basic and complex emotions in listeners via several psychological mechanisms that emotions to music share with other emotions.
The primary argument of this target article is that research on music and emotion has failed to become cumulative because music researchers have either neglected underlying psychological mechanisms or assumed that musical emotions reflect a cognitive appraisal. We argue that it is important to look beyond appraisal theory and consider alternative but less obvious ways in which music might induce emotions. While appraisal may be important for many forms of art (Silvia Reference Silvia2005), there are other mechanisms that are far more relevant in the case of music. We claim that if these additional mechanisms are taken into account, there is nothing particularly strange about results that suggest that music induces all kinds of emotions (Gabrielsson Reference Gabrielsson, Juslin and Sloboda2001, Table 19.2).
The problem is that most researchers seem to have mistakenly assumed that musical emotions can be studied and described without regard to how they were induced. Most studies have not controlled for the underlying mechanism, despite their attempts to generalize about the nature of musical emotions. Unfortunately, as discussed further in sections 4.1 and 4.4, failure to distinguish between mechanisms may lead to apparently inconsistent findings and unnecessary controversy among researchers. We believe that the solution to this problem is a more hypothesis-driven approach that takes the characteristics of each mechanism into account. Such an approach is proposed in this article.
In the following, we (a) review evidence from different kinds of sources to show that, despite claims to the contrary, music can induce emotions, (b) present a novel theoretical framework, featuring six psychological mechanisms and 66 hypotheses, that explains how such emotions are induced, (c) consider how this framework might guide future research and help to resolve previous disagreements, and (d) discuss implications for research on emotions in general and musical emotions in particular.
2. Does music really induce emotions?
Studies of music and emotion have been conducted off and on since psychology's birth at the end of the nineteenth century (Gabrielsson & Juslin Reference Gabrielsson, Juslin, Davidson, Scherer and Goldsmith2003). The majority of studies have focused on how listeners perceive emotions expressed in the music. Similarly, most theories of music and emotion have focused on the representational features of music that enable listeners to perceive emotions (e.g., Clynes Reference Clynes1977; Cooke Reference Cooke1959; Langer Reference Langer1957). However, perception of emotions is primarily a sensory or cognitive process that does not necessarily say anything about what the listener himself or herself is feeling, since perception of emotions may well proceed without any emotional involvement (Gabrielsson Reference Gabrielsson2002; Harré Reference Harré, Hjort and Laver1997). Hence, induction of emotions must be studied in its own right.
With an increasing number of studies devoted to exploring emotional responses to music, we are in a good position to answer more definitively the long-standing question of whether music really can induce emotions. However, the answer to this question depends on how emotion is defined. Table 1 offers working definitions of affective terms used in this article, based on the emerging consensus in research on affect (e.g., Davidson et al. Reference Davidson, Scherer and Goldsmith2003, p. xiii; Juslin & Scherer Reference Juslin, Scherer, Harrigan, Rosenthal and Scherer2005, Table 3.1; Oatley et al. Reference Oatley, Keltner and Jenkins2006, pp. 29–31).
Table 1. Working definitions of affective terms used in this target article
Although researchers may not agree on a precise definition of emotions, they largely agree on the characteristics and components of an emotional response (e.g., Izard Reference Izard2007). As shown in Table 1, emotions are typically described as relatively brief, though intense, affective reactions to potentially important events or changes in the external or internal environment that involve several subcomponents: (a) cognitive appraisal (e.g., you appraise the situation as “dangerous”), (b) subjective feeling (e.g., you feel afraid), (c) physiological arousal (e.g., your heart starts to beat faster), (d) expression (e.g., you scream), (e) action tendency (e.g., you run away), and (f) regulation (e.g., you try to calm yourself) (e.g., Ekman Reference Ekman1992a; Johnson-Laird & Oatley Reference Johnson-Laird and Oatley1992; Scherer Reference Scherer and Borod2000b). Each of these six components can be used to measure emotions, though researchers debate the extent to which different components are synchronized during an emotional response (cf. Frijda Reference Frijda, Kahneman, Diener and Schwarz1999; Scherer Reference Scherer and Borod2000b).
To demonstrate that music can evoke “real” emotions, one should provide evidence that music produces reactions in all of the aforementioned emotion components. Such evidence comes from many different strands of research and is summarized in Table 2. Although each source of evidence is associated with its own set of problems, the combined evidence is quite compelling. If these findings do not reflect emotions, as some have argued, what exactly do they reflect? Most of the evidence was collected in Western societies, though there is evidence from anthropology and ethology that emotional reactions to music occur in all human societies of the world and are not simply inventions of the Western world (Becker Reference Becker, Juslin and Sloboda2001; Reference Becker2004; Eibl-Eibesfeldt Reference Eibl-Eibesfeldt1989). Music appears to induce a wide range of both basic and complex emotions (e.g., Gabrielsson Reference Gabrielsson, Juslin and Sloboda2001, Table 19.2; Juslin & Laukka Reference Juslin and Laukka2004, Table 4; Sloboda Reference Sloboda, Riess-Jones and Holleran1992, Table 1; Wells & Hakanen Reference Wells and Hakanen1991, Table 1), something that a theory of musical emotion must be able to account for. There is also preliminary evidence of synchronization of emotion components in response to music (Lundqvist et al., in press).
Table 2. Summary of evidence of emotional reactions to music in terms of various subcomponents
Most studies of musical emotions have relied merely on self-report, which could be subject to demand characteristics (i.e., the total sum of cues that convey the researcher's hypothesis to the participant and thus may influence the participant's behavior; Orne Reference Orne1962). It is therefore promising that several studies have reported effects of musically induced emotions on indirect measures that should be less sensitive to demand characteristics (see Table 3). These findings, which suggest that music can be just as effective as other emotion-elicitation techniques, offer further evidence that music induces emotions in listeners. Though these studies are sometimes referred to as studies of mood induction, we claim that music usually induces emotions rather than moods,Footnote 5 because listeners' reactions focus on an “object” (the music, or more specifically certain information in the music processed relative to individual and situational factors), they last only for a limited duration (ca. 5–40 mins; Västfjäll Reference Västfjäll2002a, p. 192; see also Panksepp & Bernatzky Reference Panksepp and Bernatzky2002), and they involve autonomic responses (Krumhansl Reference Krumhansl1997). These aspects are associated with emotions rather than moods (Table 1; Beedie et al. Reference Beedie, Terry and Lane2005). However, there is one emotion component for which evidence is lacking – the cognitive appraisal. This raises the primary question of how, exactly, musical emotions are induced.
Table 3. Examples of findings from studies that used indirect measures of musically induced emotions
* Note: Description refers to effects of positive (happy) as compared to negative (sad) emotions.
3. How does music induce emotions?
Consider the following example of a listener's emotional responses during a concert (possible induction mechanisms are indicated within the parentheses and are further explained in section 3.1):
Klaus arrived just in time for the concert on Friday evening … He sat down and the music began. A sudden, dissonant chord induced a strong feeling of arousal (i.e., brain stem reflex), causing his heart to beat faster. Then, when the main theme was introduced, he suddenly felt rather happy – for no apparent reason (i.e., evaluative conditioning). In the following section, the music turned more quiet … The sad tone of a voice-like cello that played a slow, legato, falling melody with a trembling vibrato moved him to experience the same sad emotion as the music expressed (i.e., emotional contagion). He suddenly recognized the melody; it brought back a nostalgic memory from an event in the past where the same melody had occurred (i.e., episodic memory). When the melody was augmented by a predictable harmonic sequence, he started to fantasize about the music, conjuring up visual images – like a beautiful landscape – that were shaped by the music's flowing character (i.e., visual imagery). Next, the musical structure began to build up towards what he expected to be a resolution of the tension of the previous notes when suddenly the harmonics changed unexpectedly to another key, causing his breathing to come to a brief halt (i.e., musical expectancy). He thought, “This piece of music is really a cleverly constructed piece! It actually made me reach my goal to forget my trouble at work.” Reaching this goal made him happy (i.e., cognitive appraisal).
This fictitious, although empirically inspired, example gives an idea of the phenomena that need to be explained by a satisfactory model of musical emotions. One thing should be apparent from this brief example: there is no single mechanism that can account for all instances of musically induced emotion. Yet, although several authors have acknowledged that there may be more than one mechanism (Berlyne Reference Berlyne1971; Dowling & Harwood Reference Dowling and Harwood1986; Meyer Reference Meyer1956; Robinson Reference Robinson2005; Scherer & Zentner Reference Scherer, Zentner, Juslin and Sloboda2001; Sloboda & Juslin Reference Sloboda, Juslin, Juslin and Sloboda2001), there has been no attempt to develop a complete theoretical framework with a set of hypotheses. In fact, few of the theories proposed have even been properly tested.
In the following sections of this article, we outline a new theoretical framework featuring six psychological mechanisms that we hypothesize are involved in the musical induction of emotions: (1) brain stem reflexes, (2) evaluative conditioning, (3) emotional contagion, (4) visual imagery, (5) episodic memory, and (6) musical expectancy. We suggest that these mechanisms (along with cognitive appraisal) can explain most emotions induced by music in everyday life.Footnote 6 It must be noted at the outset that, though we consider it necessary to distinguish among the mechanisms for research purposes (sect. 3.2), the mechanisms are not mutually exclusive. Instead, they should be regarded as complementary ways through which music might induce emotions. Our framework builds partly on the work of pioneers in the field (Berlyne Reference Berlyne1971; Meyer Reference Meyer1956), as well as on more recent ideas (Juslin & Sloboda Reference Juslin and Sloboda2001). However, by synthesizing theories and findings from several domains, we are able to provide the first set of hypotheses that may help researchers to distinguish between the mechanisms. We first describe each mechanism separately (in sect. 3.1) and then present the hypotheses (in sect. 3.2). Because few studies so far have investigated these mechanisms in regard to music, the description of each mechanism is broad and preliminary.
3.1. Psychological mechanisms
3.1.1. Brain stem reflex
This refers to a process whereby an emotion is induced by music because one or more fundamental acoustical characteristics of the music are taken by the brain stem to signal a potentially important and urgent event. All other things being equal, sounds that are sudden, loud, dissonant, or feature fast temporal patterns induce arousal or feelings of unpleasantness in listeners (e.g., Berlyne Reference Berlyne1971; Burt et al. Reference Burt, Bartolome, Burdette and Comstock1995; Foss et al. Reference Foss, Ison, Torre and Wansack1989; Halpern et al. Reference Halpern, Blake and Hillenbrand1986). Such responses reflect the impact of auditory sensations – music as sound in the most basic sense.
The perceptual system is constantly scanning the immediate environment in order to discover potentially important changes or events. Certain sound qualities are indicative of change, such as sudden or extreme sounds, sounds that change very quickly, or sounds that are the result of strong force or large size. Sounds that meet certain criteria (e.g., fast, loud, noisy, very low- or high-frequenced) will therefore produce an increased activation of the central nervous system. The precise physiological processes underlying such brain stem responses are not completely understood, although evidence suggests that they occur in close connection with the reticular formation of the brain stem and the intralaminar nuclei of the thalamus, which receive inputs from the auditory system. The brain stem is an ancient structure of the brain that subserves a number of sensory and motor functions including, but not limited to, auditory perception and the mediation and control of attention, emotional arousal, heart rate, breathing, and movement (Joseph Reference Joseph2000). The reticular system is in a position to quickly induce arousal so that attention may be selectively directed at sensory stimuli of potential importance. The system exerts its widespread influences on sensory and motor functions and arousal through neurotransmitters such as norepinephrine and serotonin. While the system may be activated and inhibited by the amygdala, hypothalamus, and orbitofrontal cortex, it may also be activated independently of these structures in a more reflex-like manner (Lipscomb & Hodges Reference Lipscomb, Hodges and Hodges1996; Tranel Reference Tranel, Lane and Nadel2000).
Brain stem reflexes to music rely on the early stages of auditory processing. When an auditory signal reaches the primary auditory cortex, the signal has already undergone a number of analyses by such brain structures as the superior olivary complex, the inferior colliculus, and the thalamus (Koelsch & Siebel Reference Koelsch and Siebel2005). Accordingly, alarm signals to auditory events that suggest “danger” may be emitted as early as at the level of the inferior colliculus. Brain stem reflexes are “hard-wired.” Thus, for instance, the perceived pleasantness and unpleasantness of sensory consonance and dissonance reflect how the hearing system divides frequencies into critical bandwidths: If the frequency separation of two tones is either very small or larger than the critical bandwidth, the tones will be judged as consonant. If the separation is about one-fourth of a critical band, the tones will be judged as maximally dissonant (Lipscomb & Hodges Reference Lipscomb, Hodges and Hodges1996). Sensory dissonance is suggestive of “danger” in natural environments, because it occurs in the “threat” and “warning” calls of many species of animals (Ploog Reference Ploog, Papousek, Jürgens and Papousek1992). Dissonance may thus have been selected by evolution as an unlearned negative reinforcer of behavior (Rolls Reference Rolls, Coan and Allen2007).
Brain stem reflexes are quick and automatic, as shown by evidence of rapid and pre-attentive categorization of subtle timbral differences associated with different emotions (Goydke et al. Reference Goydke, Altenmüller, Möller and Münte2004), and affective priming effects of consonant and dissonant chords (Sollberger et al. Reference Sollberger, Reber and Eckstein2003). Brain stem reflexes to music may function even prior to birth, as indicated by findings that playing loud music to fetuses produces heart rate accelerations and increased motor responses, whereas soft music produces moderate heart rate decelerations and reduced movement (for a review, see Lecanuet Reference Lecanuet, Deliège and Sloboda1996).
The arousal-inducing properties of music were investigated and theorized by Berlyne (Reference Berlyne1971).Footnote 7 According to Berlyne's theory, listeners will prefer musical stimuli that induce an “optimum” level of physiological arousal. If the “arousal potential” of the music is too high, listeners will reject the music. Similarly, if the arousal potential is too low, listeners will reject the music. Hence, Berlyne hypothesized that listeners' preferences are related to arousal (or some aspect of it, such as speed or loudness) in the form of an inverted U-shaped curve (the Wundt curve). Berlyne's theory has received some empirical support from experimental studies (for a review, see North & Hargreaves Reference North, Hargreaves, Hargreaves and North1997). In addition, several studies have shown that listeners use music to regulate their arousal in order to obtain optimal arousal (DeNora Reference DeNora, Juslin and Sloboda2001; Thayer Reference Thayer1996). However, what is judged as “optimal” by a listener varies depending on the situation (North & Hargreaves Reference North, Hargreaves, Hargreaves and North1997) and on personality characteristics of the listener (McNamara & Ballard Reference McNamara and Ballard1999). Thus, it may be difficult to predict arousal responses without taking individual and contextual factors into consideration. Brain stem reflexes can explain the stimulating and relaxing effects of music, and how mere sounds may induce pleasantness and unpleasantness. However, it is unclear how the mechanism could explain the induction of specific emotions.
3.1.2. Evaluative conditioning
This refers to a process whereby an emotion is induced by a piece of music simply because this stimulus has been paired repeatedly with other positive or negative stimuli. Thus, for instance, a particular piece of music may have occurred repeatedly together in time with a specific event that always made you happy (e.g., meeting your best friend). Over time, through repeated pairings, the music will eventually come to evoke happiness even in the absence of the friendly interaction.
Evaluative conditioning (EC) is also referred to as affective learning, fear conditioning, emotional conditioning, and preference conditioning, but regardless of the term used, it seems to refer to the same phenomenon – a special kind of classic conditioning that involves the pairing of an initially neutral conditioned stimulus (CS) with an affectively valenced, unconditioned stimulus (US). After the pairing, the CS acquires the ability to evoke the same affective state as the US in the perceiver. Regardless of the term used, and of whether positive (e.g., liking) or negative (e.g., fear) states are conditioned, the process appears to have the same characteristics.
Firstly, an EC may occur even if the participant is unaware of the contingency of the two stimuli (Field & Moore Reference Field and Moore2005; Hammerl & Fulcher Reference Hammerl and Fulcher2005), which may not be true for other forms of classic conditioning (e.g., Lovibond & Shanks Reference Lovibond and Shanks2002). Indeed, it has been reported that an EC response can be both established and induce emotions without awareness (Martin et al. Reference Martin, Stambrook, Tataryn and Beihl1984; Öhman & Mineka Reference Öhman and Mineka2001). Attention may even hamper effects of EC (De Houwer et al. Reference De Houwer, Baeyens and Field2005). This characteristic of EC has some interesting implications for musical experiences: It has been found that, sometimes, pieces of music induce emotions for no apparent reason (e.g., Juslin et al., in press). EC offers a possible explanation of this phenomenon. Furthermore, it generates the prediction that we might react with positive emotions to music that we think is of poor quality simply because the music has occurred repeatedly in previous pleasant situations. Such effects could presumably be demonstrated in listening experiments that use established paradigms for conditioning (Lavond & Steinmetz Reference Lavond and Steinmetz2003), along with indirect measures of emotion (Table 3). Secondly, EC seems to be more resistant to extinction than are other forms of classic conditioning (LeDoux Reference LeDoux, Cacioppo, Berntson, Adolphs, Carter, Davidson, McClintock, McEwen, Meaney, Schacter, Sternberg, Suomi and Taylor2002). (Extinction refers to the process whereby postacquisition presentations of the conditioned stimulus, e.g., a specific piece of music, without the unconditioned stimulus, e.g., a happy event, leads to a gradual elimination of the previously acquired response; De Houwer et al. Reference De Houwer, Thomas and Baeyens2001, p. 858). Hence, once a piece of music has been strongly associated with a specific emotional outcome, this association could be quite persistent. Thirdly, EC seems to depend on unconscious, unintentional, and effortless processes (De Houwer et al. Reference De Houwer, Baeyens and Field2005; LeDoux Reference LeDoux, Cacioppo, Berntson, Adolphs, Carter, Davidson, McClintock, McEwen, Meaney, Schacter, Sternberg, Suomi and Taylor2002), which involve subcortical brain regions such as the amygdala and the cerebellum (Balleine & Killcross Reference Balleine and Killcross2006; Johnsrude et al. Reference Johnsrude, Owen, White, Zhao and Bohbot2000; Sacchetti et al. Reference Sacchetti, Scelfo and Strata2005).
Although this mechanism seems to be generally acknowledged as a powerful source of emotions in music (see Berlyne Reference Berlyne1971, p. 33; Dowling & Harwood Reference Dowling and Harwood1986, pp. 204–5; Hanslick Reference Hanslick and Payzant1854/1986; Sloboda & Juslin Reference Sloboda, Juslin, Juslin and Sloboda2001, pp. 94–95), very few studies so far have investigated EC responses to music. There are two possible reasons for this. Firstly, the responses are often highly personal and idiosyncratic (i.e., different listeners have different learning histories, with a few notable exceptions), which may seem to render them more difficult to study systematically. Secondly, because EC responses are not strongly related to the music as such – the music merely acts as a conditioned stimulus – they have been regarded as “irrelevant” responses to music and, thus, unworthy of study (Hanslick Reference Hanslick and Payzant1854/1986). However, if EC is a strong and frequent source of music-induced emotions in everyday life, the mechanism should be part of a credible framework for musical emotions. Which element of the musical stimulus that best serves as the conditioned stimulus as well as its degree of generalization and discrimination are issues that remain to be investigated. The melody (or theme) of the music could be especially effective, though studies of fear conditioning have shown that even a simple tone can be effective in establishing a fear association (LeDoux Reference LeDoux, Cacioppo, Berntson, Adolphs, Carter, Davidson, McClintock, McEwen, Meaney, Schacter, Sternberg, Suomi and Taylor2002).
Blair and Shimp (Reference Blair and Shimp1992) reported that when participants were originally exposed to a piece of music in an unpleasant situation, they later held a less favorable affective attitude towards a product presented together with the music than did participants who had not been pre-exposed to the same conditioning. Similarly, Razran (Reference Razran1954) found, in a series of experiments, that affective attitudes (as indexed by ratings and characterizations) towards pieces of music, paintings, and photographs could be modified by free lunches – at least when participants were unaware of the aim to condition them. It should be noted that music commonly occurs in situations where music listening is not the only or the primary activity (Juslin & Laukka Reference Juslin and Laukka2004; Sloboda & O'Neill Reference Sloboda, O'Neill, Juslin and Sloboda2001) and where subtle conditioning processes outside of awareness could easily occur. Thus, it seems plausible that EC could account for many of our emotional responses to music in everyday life.Footnote 8
3.1.3. Emotional contagion
This refers to a process whereby an emotion is induced by a piece of music because the listener perceives the emotional expression of the music, and then “mimics” this expression internally, which by means of either peripheral feedback from muscles, or a more direct activation of the relevant emotional representations in the brain, leads to an induction of the same emotion. For instance, the music might have a sad expression (e.g., slow tempo, low pitch, low sound level) that induces sadness in the listener (Juslin Reference Juslin, Juslin and Sloboda2001). Evidence that music with a specific emotional expression can give rise to the same emotion in the listener has been reported in several studies (e.g., Kallinen & Ravaja Reference Kallinen and Ravaja2006; Lundqvist et al., in press).
This mechanism is related to the vast literature on emotional expression in music. It has been suggested that expression may be an “iconic” source of emotion (Dowling & Harwood Reference Dowling and Harwood1986). The term iconic refers to the fact that the structures of music show formal similarities to the structures of expressed (Kivy Reference Kivy1980) or felt (Langer Reference Langer1957) emotions. Numerous studies have shown that listeners are able to perceive specific emotions in pieces of music (Gabrielsson & Juslin Reference Gabrielsson, Juslin, Davidson, Scherer and Goldsmith2003), and that even children as young as 3 or 4 years may be able to recognize basic emotions in music (Cunningham & Sterling Reference Cunningham and Sterling1988). But how exactly does perception of an emotion in the music lead to induction of the same emotion in the listener?
Lipps (Reference Lipps1903) was probably the first to postulate a mechanistic account of empathy, where the perception of an emotional gesture in another person directly induces the same emotion in the perceiver without any appraisal process (e.g., Preston & de Waal Reference Preston and de Waal2002). Modern research has confirmed that people may “catch” the emotions of others when seeing their facial expressions (Hatfield et al. Reference Hatfield, Cacioppo and Rapson1994) or hearing their vocal expressions (Neumann & Strack Reference Neumann and Strack2000). Previous research on emotional contagion has focused mostly on facial expression. For example, people exposed to pictures of facial expressions of emotions spontaneously activate the same face muscles (as shown by electromyography) even when the pictures are processed outside of awareness. Moreover, they report feeling the same emotions (Dimberg et al. Reference Dimberg, Thunberg and Elmehed2000). It has been argued that emotional contagion facilitates the mother-infant bond (Darwin Reference Darwin1872), as well as social interaction in general (Preston & de Waal Reference Preston and de Waal2002). In support, such contagion seems to create affiliation and liking (e.g., Lakin et al. Reference Lakin, Jefferis, Cheng and Chartrand2003), which is arguably beneficial for social interaction.
Recent research has suggested that the process of emotional contagion may occur through the mediation of so-called mirror neurons discovered in studies of the monkey premotor cortex in the 1990s (e.g., di Pellegrino et al. Reference di Pellegrino, Fadiga, Fogassi, Gallese and Rizzolatti1992). It was found that the mirror neurons discharged both when the monkey carried out an action and when it observed another individual (monkey or human) performing a similar action (Rizzolatti & Craighero Reference Rizzolatti and Craighero2004). These mirror neurons appeared to be located in the ventral premotor regions of the brain, regardless of the type of stimulus. Direct evidence for the existence of mirror neurons in humans is lacking so far, but a large amount of indirect evidence suggests that a mirror-neuron system exists also in humans. For example, several studies have shown that when individuals observe an action carried out by another individual, the motor cortex may become active in the absence of overt motor activity (Rizzolatti & Craighero Reference Rizzolatti and Craighero2004). De Gelder et al. (Reference de Gelder, Snyder, Greve, Gerard and Hadjikhani2004) reported that observing fear expressions in body language increased activity in motor areas of the brain, in addition to those associated with emotion, which is consistent with the notion of a mirror mechanism.
How may emotional contagion be applied to music? Because music often features expressive acoustical patterns similar to those that occur in emotional speech (for a review, see Juslin & Laukka Reference Juslin and Laukka2003), it has been argued that we become aroused by the voice-like aspects of music via a process in which a neural mechanism responds quickly and automatically to certain stimulus features, which leads us to mimic the perceived emotion internally. According to the super-expressive voice theory (e.g., Juslin Reference Juslin, Juslin and Sloboda2001), what makes a particular performance of music on, say, the violin, so expressive is the fact that it sounds a lot like the human voice, whereas at the same time it goes far beyond what the human voice can do in terms of speed, intensity, and timbre. For example, if human speech is perceived as “angry” when it has fast rate, loud intensity, and a harsh timbre, a musical instrument might sound extremely “angry” by virtue of its even higher speed, louder intensity, and harsher timbre. This aspect should render music a particularly potent source of emotional contagion.
While the notion of emotional contagion admittedly remains speculative in relation to music, a recent functional magnetic resonance imaging (fMRI) study by Koelsch et al. (Reference Koelsch, Fritz, von Cramon, Müller and Friederici2006) indicated that music listening activated brain areas related to a circuitry serving the formation of premotor representations for vocal sound production (no singing was observed among the participants). Koelsch et al. concluded that this could reflect a mirror-function mechanism, and the findings render tentative support to the notion that listeners may mimic the emotional expression of the music internally. Precursors of emotional contagion via facial and vocal expression have been observed as early as the first year of development (Soussignan & Schaal Reference Soussignan, Schaal, Nadel and Muir2005), but remain to be explored in relation to music. We assume that emotional contagion mainly involves “basic” emotions with distinct nonverbal expressions (Juslin & Laukka Reference Juslin and Laukka2003; Laird & Strout Reference Laird, Strout, Coan and Allen2007).
Some authors have pointed out that music does not sound very much like vocal expressions, except in special cases (Davies Reference Davies, Juslin and Sloboda2001). Why, then, should we respond to music as though it were a vocal expression? One possible explanation is that the expressions are processed by a domain-specific and autonomous “module” of the brain (Fodor Reference Fodor1983), which reacts to certain features in the stimulus. This module does not “know” the difference between a vocal expression and other acoustic expressions, and will react in the same way (e.g., registering anger) as long as certain cues (e.g., high speed, loud dynamics, rough timbre) are present in the stimulus. This modular theory remains to be tested, but some support, in terms of Fodor's (1983) suggested characteristics of a module, was summarized by Juslin and Laukka (Reference Juslin and Laukka2003, p. 803). Thus, it is plausible that listeners' emotions to music sometimes reflect social, modular responses to the voice-like and emotion-specific acoustic patterns of the music.Footnote 9
3.1.4. Visual imagery
This refers to a process whereby an emotion is induced in a listener because he or she conjures up visual images (e.g., of a beautiful landscape) while listening to the music. The emotions experienced are the result of a close interaction between the music and the images.Footnote 10
Visual imagery is usually defined as an experience that resembles perceptual experience, but that occurs in the absence of relevant sensory stimuli. The study of visual imagery has an old, but confused, status in psychology, marked by much controversy (Kolers Reference Kolers1983). Much of the controversy has concerned its ontological status: Does visual imagery involve a distinctively “pictorial” representation of events in mind, or does it reflect a “propositional” representation? Kosslyn (Reference Kosslyn1980) argued that the images themselves are quasi-pictorial representations, whereas the generative, long-term structure of imagery is propositional (e.g., similar to a TV set whose output is a picture, but whose mechanisms for generating this picture are better expressed in discrete symbols of electronics). The pictorial view is supported by findings that many of the brain regions that are activated during visual perception are similarly activated when a person is involved in visual imagery (Farah Reference Farah and Gazzaniga2000; Ganis et al. Reference Ganis, Thompson, Mast, Kosslyn and Gazzaniga2004). In accordance with theories of symbolic development (Piaget Reference Piaget1951), one could assume that visual imagery develops during the preschool period, when children create increasingly complex symbolic representations of the external world (Gärdenfors Reference Gärdenfors2003; for empirical evidence, see Kosslyn et al. Reference Kosslyn, Margolis, Barrett, Goldknopf and Daly1990).
Mental images have been regarded as “internal triggers” of emotions (Plutchik Reference Plutchik1984), and studies have revealed that visual imagery associated with different emotions involves different imagery contents (Lyman & Waters Reference Lyman and Waters1989), as well as different patterns of physiological response (Schwartz et al. Reference Schwartz, Weinberger and Singer1981). It has been suggested that musical stimuli are especially effective in stimulating visual imagery (Osborne Reference Osborne1980; Quittner & Glueckauf Reference Quittner and Glueckauf1983), and a few studies have indicated that imagery can be effective in enhancing emotions to music (Band et al. Reference Band, Quilter and Miller2001–2002; see also Västfjäll Reference Västfjäll2002a, p. 183).
The precise nature of this visual imagery process remains to be determined, but listeners seem to conceptualize the musical structure through a metaphorical nonverbal mapping between the music and so-called image-schemata grounded in bodily experience (Bonde Reference Bonde2006; Lakoff & Johnson Reference Lakoff and Johnson1980); for example, hearing melodic movement as “upward.” We argue that listeners respond to mental images much in the same way as they would to the corresponding stimuli in the “real” world – for example, reacting with positive emotions to a beautiful nature scene (see Figure 2.4. in Bradley & Lang [Reference Bradley, Lang, Coan and Allen2007], for examples of affective responses to various pictures).
Osborne (Reference Osborne1989) reported certain recurrent “themes” in visual imagery to music, such as nature scenes (e.g., sun, sky, ocean) and out-of-body experiences (e.g., floating above the earth), but the results were probably affected by the particular musical style used (“spacey, synthesized electronic music with simple structure, some free form, and much repetition,” p. 134). Indeed, it has been suggested that certain musical characteristics, such as repetition, predictability in melodic, harmonic, and rhythmic elements, and slow tempo, are especially effective in stimulating vivid imagery (McKinney & Tims Reference McKinney and Tims1995).
A special feature of the imagery mechanism is that the listener is very much able to influence the emotions induced by the music. Although images might come into the mind unbidden, in general a listener may conjure up, manipulate, and dismiss images at will. Larson (Reference Larson1995) has speculated that music offers a medium for adolescents, in particular, through which they may conjure up strong emotional images around which a temporary sense of self can cohere. The music is like a “fantasy ground” for exploring possible selves during the important process of resolving a personal identity in late adolescence (see also Becker Reference Becker, Juslin and Sloboda2001; DeNora Reference DeNora, Juslin and Sloboda2001).
Visual imagery in relationship to music has been discussed most extensively in the context of music therapy (Toomey Reference Toomey1996). Helen Bonny developed a method, Guided Imagery and Music (GIM), where a “traveler” is invited to “share” his or her images as they are experienced in real time during a pre-programmed sequence of music (see Bonny & Savary Reference Bonny and Savary1973). Music-induced imagery may produce a state of deep relaxation, with health benefits such as reduced cortisol levels (McKinney et al. Reference McKinney, Antoni, Kumar, Tims and McCabe1997). However, there seem to be large individual differences with regard to the ability to generate visual images (Marks Reference Marks1973).
Visual imagery may occur in connection with episodic memories (discussed in sect. 3.1.5), although it seems necessary to distinguish the two mechanisms, because a musical experience may evoke emotions when a listener conjures up images of things and events that have never occurred, in the absence of any episodic memory from a previous event in time. Moreover, visual imagery is more strongly influenced or shaped by the unfolding structure of the music than is episodic memory, for which the music mainly serves a retrieval cue. In the words of Meyer (Reference Meyer1956), “it seems probable that … image processes play a role of great importance in the musical affective experiences of many listeners” (p. 258).
3.1.5. Episodic memory
This refers to a process whereby an emotion is induced in a listener because the music evokes a memory of a particular event in the listener's life. This is sometimes referred to as the “Darling, they are playing our tune” phenomenon (Davies Reference Davies1978).
Research has suggested that music often evokes memories (e.g., Gabrielsson Reference Gabrielsson, Juslin and Sloboda2001; Juslin et al., submitted; Sloboda Reference Sloboda, Riess-Jones and Holleran1992). When the memory is evoked, so also is the emotion associated with the memory (e.g., Baumgartner Reference Baumgartner1992). Such emotions can be rather intense, perhaps because the physiological reaction patterns to the original events are stored in memory along with the experiential content, as proposed by Lang (Reference Lang1979). Baumgartner (Reference Baumgartner1992) reported evidence that episodic memories evoked by music tend to involve social relationships (e.g., past or current romantic partners, time spent with friends).Footnote 11 However, the memories can involve all kinds of events, such as vacations, movies, music concerts, a victory in a boxing match, the death of a grandfather, or childhood memories (Baumgartner Reference Baumgartner1992; see further examples in Gabrielsson Reference Gabrielsson, Juslin and Sloboda2001, p. 439). Indeed, music accompanies most important human activities from the cradle to the grave (Gregory Reference Gregory, Hargreaves and North1997), although due to childhood amnesia listeners are unlikely to recall much from the first years of their life (Reisberg & Heuer Reference Reisberg, Heuer, Reisberg and Hertel2004). Children's ability to recall and converse about episodic memories develops slowly across the preschool years (e.g., Fivush & Sales Reference Fivush, Sales, Reisberg and Hertel2004; Perner & Ruffman Reference Perner and Ruffman1995), and episodic memory is the type of memory that begins to decline first as a result of aging (e.g., Tulving Reference Tulving2002). Both kinds of developmental trends should be observable in listeners' emotional reactions to music based on episodic memory.
Episodic memory is one of the induction mechanisms that have commonly been regarded as less “musically relevant” by music theorists, but recent evidence suggests that it could be one of the most frequent and subjectively important sources of emotion in music (see Juslin et al., in press; Sloboda & O'Neill Reference Sloboda, O'Neill, Juslin and Sloboda2001). Many listeners actively use music to remind them of valued past events, which indicates that music can serve an important nostalgic function in everyday life. The music may help to consolidate a listener's self-identity. Furthermore, a retrospective memory study by Sloboda (Reference Sloboda, Wilson and Roehmann1989) has indicated that strong and positively valenced childhood memories of musical events may be important in determining which individuals will pursue a high level of involvement in music later in life.
In previous research, most researchers have regarded both conditioning and episodic memory as cases of memory-based or associative mechanisms (Dowling & Harwood Reference Dowling and Harwood1986; Scherer & Zentner Reference Scherer, Zentner, Juslin and Sloboda2001; Sloboda & Juslin Reference Sloboda, Juslin, Juslin and Sloboda2001). However, there are good reasons to view these as partly separate and independent mechanisms. Although evaluative conditioning is a form of memory, episodic memory is different in that it always involves a conscious recollection of a previous event in time that preserves much contextual information. Also, unlike conditioning, episodic memory appears to be organized in terms of a hierarchical structure with three levels: life-time periods, general events, and event-specific knowledge (Conway & Rubin Reference Conway, Rubin, Collins, Gathercole, Conway and Morris1993). Furthermore, the two kinds of memory have partly different process characteristics and brain substrates (sect. 3.2). Hence, they should be distinguished in research on musical emotions in order to not yield inconsistent findings.
One important characteristic of episodic memory, more generally, is the common finding that people tend to recall more memories from their youth and early adulthood (15–25 years of age) than from those periods that precede or follow it. This is referred to as the reminiscence bump, and may be explained by the fact that many self-defining experiences tend to occur at this stage of life development (Conway & Holmes Reference Conway, Holmes, Braisby and Gellatly2005, p. 513). In this context, it should be noted that music seems to play a very prominent role in adolescents' lives and, particularly, in relation to the development of a self-identity (Laiho Reference Laiho2004). Hence, we would expect episodic memories associated with music to be particularly emotionally vivid and frequent with regard to music from young adulthood, as indeed seems to be the case. Schulkind et al. (Reference Schulkind, Hennis and Rubin1999) found that older adults preferred, knew more about, as well as had stronger emotional responses to music popular during their youth than to music popular later in life. Further, both younger and older adults were more likely to retrieve a spontaneous autobiographical memory when they were cued by a song that moved them emotionally. Holbrook and Schindler (Reference Holbrook and Schindler1989) also found that participants showed the greatest liking for music that was popular during their youth. Hence, one reasonable prediction could be that emotional reactions to music involving episodic memory more commonly involve events from one's youth and early adulthood than from other periods in one's life. Empirical evidence suggests that nostalgia may be one of the more common responses to music (Juslin et al., in press).
3.1.6. Musical expectancy
This refers to a process whereby an emotion is induced in a listener because a specific feature of the music violates, delays, or confirms the listener's expectations about the continuation of the music. For instance, the sequential progression of E-F# sets up the musical expectation that the music will continue with G# (Sloboda Reference Sloboda, Riess-Jones and Holleran1992). If this does not happen, the listener may become, for instance, surprised.
This psychological mechanism has been most extensively theorized by Meyer (Reference Meyer1956), in what could well be the most cited volume on music and emotion ever. Meyer's theory was inspired by Aiken's (Reference Aiken1950; cited in Meyer Reference Meyer1956, p. 25) ideas regarding musical expectations, as well as by contemporary psychological theories of perception (e.g., the Gestalt school) and emotions (e.g., Dewey's conflict theory of emotions). However, Meyer was the first theorist to develop the notion of musical expectancy in a convincing and thorough manner. It should be noted that musical expectancy does not refer to any unexpected event that might occur in relationship to music. A simple form of unexpectedness (e.g., the sudden onset of a loud tone) would instead be an example of the mechanism called brain stem reflex (see sect. 3.1.1). Similarly, more general surprising features of an event that involves music (e.g., that a concert was better than the listener had expected) would instead be an example of the cognitive appraisal mechanism. Musical expectancy refers to those expectancies that involve syntactical relationships between different parts of the musical structure (Narmour Reference Narmour1991; Patel Reference Patel2003).
Like language, music consists of perceptually discrete elements, organized into hierarchically structured sequences according to “well-formedness” rules. Thus, it is a common view among music theorists that most musical styles are, in principle, describable by a grammar (Lerdahl & Jackendoff Reference Lerdahl and Jackendoff1983). It is only through the perception of this syntax that the relevant musical expectations arise. These expectations are based on the listener's previous experiences of the same musical style (Carlsen Reference Carlsen1981; Krumhansl et al. Reference Krumhansl, Louhivuori, Toiviainen, Järvinen and Eerola1999). Emotional reactions to music are induced when the listener's musical expectations are somehow disrupted, for instance, by new or unprepared harmony (for examples, see Steinbeis et al. Reference Steinbeis, Koelsch and Sloboda2006).
The musical expectancy mechanism is notable for its strong dependence on learning (Meyer Reference Meyer1956). Evidence that musical expectancies depend much on cultural learning comes from the fact that such responses are not shared by young children. For instance, Sloboda (Reference Sloboda, Wilson and Roehmann1989) noted that 5-year-old children were unable to reject gross chordal dissonances as “wrong.” By the age of 9, however, they were overtly laughing at the “wrong” chords and scoring at an adult level. Another test in the same study focused on the ordering of the chords that could be either conventional (ending with a cadence) or “scrambled” (ending without resolution). On this test, children did not achieve adult levels of performance until the age of 11. Evidence of age differences have also been reported with regard to sensitivity to tonal hierarchies (Krumhansl & Keil Reference Krumhansl and Keil1982) and implied harmony (Trainor & Trehub Reference Trainor and Trehub1994). Although the ability to detect syntactical violations can be observed early (Jentschke et al. Reference Jentschke, Koelsch and Friederici2005), responses arising from musical expectancies also depend on sufficient exposure to the musical style in question.
Meyer discussed emotions in an approach characteristic for his time (i.e., as undifferentiated arousal; see Duffy Reference Duffy1941), but he observed that mere arousal through interruption of musical expectancies has little value. To have any aesthetic meaning, the arousal or tension must be followed by a satisfying resolution of the tension. In fact, Meyer (Reference Meyer1956) appeared open to the possibility that this musical play with expectations may lead to the induction of specific emotions, such as apprehension/anxiety (p. 27), hope (p. 29), or disappointment (p. 182), but these ideas have still not been tested. In fact, while highly influential and respected, Meyer's theory has not stimulated much research on musical emotions (but see Sloboda Reference Sloboda1991), perhaps because the theory is difficult to test. For example, a piece of music could produce several different expectations at different hierarchical levels of the music, and these expectations could also vary for different listeners. Therefore, it is difficult to understand or predict exactly what the listener is responding to in a particular situation.
In recent years, however, researchers have developed novel models of expectancy (Hellmuth Margulis Reference Hellmuth Margulis2005; see also Huron Reference Huron2006), which should make it more feasible to test predictions experimentally. Neurophysiological methods might be useful in this regard. It has been found that violations of musical expectancy activate the same brain areas that have been previously implicated in violations of syntax in language (Koelsch et al. Reference Koelsch, Gunter, van Cramon, Zyset, Lohmann and Friederici2002a; Maess et al. Reference Maess, Koelsch, Gunter and Friederici2001). Patel (Reference Patel2003; Reference Patel2008, Ch. 5) has therefore suggested that syntactical processing in both language and music shares a common set of processes for syntactical integration (localized in Broca's area) that operate on distinct structural representations for music and language. Evidence that expectancy violations can induce emotions was recently reported by Steinbeis et al. (Reference Steinbeis, Koelsch and Sloboda2006). Thus, it seems likely that some of our emotions to music reflect the disruption of style-specific expectations.
3.2. How can the mechanisms be distinguished?
How may we describe the relationships among the different mechanisms? We propose that it could be useful to think of the mechanisms as consisting of a number of (more or less) distinct brain functions that have developed gradually and in a specific order during the evolutionary process, from sensations (brain stem reflexes) to syntactical processing (musical expectancy) (Gärdenfors Reference Gärdenfors2003). We regard the mechanisms as information-processing devices at various levels of the brain that use various means to track significant aspects of the environment, and that may produce conflicting outputs (Griffiths Reference Griffiths, Evans and Cruse2004; Teasdale Reference Teasdale, Dalgleish and Power1999).Footnote 12 They all take music as their “object”, treating the music – rightly or wrongly – as featuring some kind of information that warrants an emotional response. However, note that the emotion induced is not the result of an appraisal of the music on several dimensions relative to the listener's motives, needs, or goals. Because the mechanisms depend on distinct brain functions with different evolutionary origins, each mechanism should possess unique characteristics. Hence, Table 4 presents a set of preliminary hypotheses regarding the characteristics of each mechanism. The mechanisms are listed in the approximate order in which they can be hypothesized to have appeared during evolution (Gärdenfors Reference Gärdenfors2003; see also Joseph Reference Joseph2000; Reber Reference Reber1993; Tulving Reference Tulving1983).Footnote 13
Table 4. Hypotheses regarding the characteristics of six psychological mechanisms through which music might induce emotions
The hypotheses can be divided into two subgroups: The first subgroup concerns characteristics of the psychological mechanism as such. Thus, Survival value of brain function describes the most important benefit that each brain function brought to those organisms that possessed this brain function.Footnote 14 Visual imagery, for example, enabled an organism to “simulate” important events internally, through self-conjured images in the absence of direct sensory input, which meant that overt and potentially dangerous action plans could be tested and evaluated before they were implemented in the external world. Information focus specifies broadly the type of information that each mechanism is processing. For instance, evaluative conditioning (EC) focuses on covariation between events. Ontogenetic development concerns the approximate time in human development when respective mechanisms might begin to have a noticeable effect on emotional responses to music. Brain stem reflexes to music could be functional even prior to birth, whereas responses involving musical expectancy do not develop fully until somewhere between the ages of 5 and 11. Key brain regions describes those regions of the brain that have been most consistently associated with each mechanism in imaging studies. Note that musical emotions can be expected to involve three kinds of brain regions: (1) regions usually involved when music is perceived, such as the primary auditory cortex; (2) regions usually involved in the conscious experience of emotions regardless of the precise cause of the emotions (e.g., the rostral anterior cingulate and the medial prefrontal cortex; e.g., Lane Reference Lane, Lane and Nadel2000, pp. 356–358); and (3) regions involved in emotional information-processing that partly differ depending on the mechanism inducing the emotion. Hence, although musical emotions are likely to involve several brain regions (Peretz Reference Peretz, Juslin and Sloboda2001), the hypotheses in Table 4 focus on the last type of regions, especially those that can help researchers to discriminate among mechanisms. For instance, the experience of conscious recollection of an episodic memory is associated with activation of the hippocampus brain region. Cultural impact/learning refers to the relative extent to which each mechanism is influenced differently by music that varies from one culture to another. For example, brain stem reflexes reflect primarily “hardwired” responses to simple features that are not affected much by learning, whereas musical expectancy reflects learned schemata about specific styles of music that differ from one culture to another and that make listeners from different cultures react differently to the same piece of music.
A second subgroup of characteristics (see Table 4) concerns the precise nature of the emotion induction process associated with each mechanism. Induced affect specifies which affective states might be expected to be induced, depending on the mechanism. For example, whereas emotional contagion might be expected to induce only “basic” emotions, which have more or less distinct nonverbal expressions of emotion, visual imagery might be expected to induce all possible human emotions. Induction speed refers to how much time each mechanism requires, in relation to other mechanisms, for an emotion to occur in a particular situation. For example, brain stem reflexes can induce emotions very quickly (in less than a second), whereas musical expectancy can be expected to require more time (at least a number of seconds) because some of the musical structure has to unfold in order for any musical expectation to occur that can be confirmed or violated. Degree of volitional influence refers to the extent to which the listener himself or herself could actively influence the induction process (e.g., through focus of attention, active recall, self-activation). For instance, reactions that involve EC may be involuntary and automatic, whereas reactions that involve visual imagery may be strongly influenced by the way the listener actively chooses to entertain some inner images and themes rather than others. Availability to consciousness is the extent to which at least some aspects of the induction process are available to the listener's consciousness, so that the listener may be able to explain his or her response. For example, if a piece of music evokes a strong episodic memory, the listener will have a conscious recollection of a previous event and some inkling of the reasons (e.g., the appraisal) that made this event evoke the emotion that is now re-experienced. Conversely, EC responses to music can be both learned and aroused outside conscious awareness. Therefore, a listener who experiences a musical emotion via this mechanism could be completely unable to explain any aspect of the induction process. Modularity refers to the extent to which the induction process of each mechanism functions as an independent and information-encapsulated module that may be activated in parallel with other psychological processes.Footnote 15 For instance, emotional contagion can be described as highly modular, because it may be activated independently of other processes, and is not influenced by the information of other modules (e.g., we respond to the expressive characteristics of the music as if they came from a person expressing emotions in the voice even if we know, at some cognitive level, that the music is not a voice). Dependence on musical structure refers to the extent to which the induction depends on the precise structure or style of the music that the listener is hearing. At one extreme, the structure of the music is not very important as such – it mainly functions as a “retrieval cue.” This is the case for evaluative conditioning and episodic memory. At the other extreme, the precise pattern of the musical structure strongly determines the nature of the induced response. This is the case for musical expectancy.
Empirical findings of relevance to the hypotheses shown in Table 4 could come from a broad range of research domains such as memory, development, emotional expression, evolutionary psychology, neuropsychology, learning, clinical psychology, and psychophysiology, as well as music psychology and music therapy. A selected number of representative sources that render theoretical or empirical support to each hypothesis have been included in Table 4. As much as possible, we have tried to include sources that involve music, although most sources focus on the mechanism more generally, as explored in fields other than music. Hence, further research is needed to test most of the hypotheses in regard to music. We acknowledge that some of the hypotheses are imprecise and mainly descriptive. This reflects the current lack of research on these issues. However, we argue that even simple predictions in terms of “high” and “low” can be tested in experiments that contrast one mechanism against another. Such tests could help to render the hypotheses more specific.
We propose that the testing of the new framework could involve an approach consisting of an interplay between field studies (diary studies, questionnaires) and experimental studies. Field studies that enable researchers to study listeners' emotional reactions to music in their natural environment could generate hypotheses about possible causal factors. These factors could then be formalized in a preliminary model, which is evaluated in experiments. These experiments may suggest the need for further knowledge about specific factors, wherefore further field studies may be needed. By combining the approaches, we may eventually arrive at general principles that can form the basis of a more detailed model of the induction process, featuring a description of the time-course and the inter-relationships of the different mechanisms. Field studies are required, because if there are several mechanisms that can induce musical emotions, and their importance varies depending on the situation, only by sampling a wide variety of situations can we hope to capture all the mechanisms. On the other hand, certain mechanisms, such as conditioning, may be difficult to demonstrate other than in a controlled laboratory setting. Field studies will have to focus on self-reports – although with the possible addition of ambulatory physiological measures (see Fahrenberg & Myrtek Reference Fahrenberg and Myrtek1996). Laboratory studies may involve any combination of the measures listed in Table 2, as well as indirect measures (Table 3), to maximize the validity of conclusions about induced emotions.
4. Implications
4.1. Resolving previous disagreements
One implication of the new framework is that it can resolve many disagreements in the field. Specifically, apparent contradictions of different approaches may be reconciled by observing that they focus on different psychological mechanisms. For example, one recurring theme in studies of music and emotion concerns the role of the person experiencing the emotion in the causal process. At one extreme is the case where the emotion is induced automatically and involuntarily (see Peretz Reference Peretz, Juslin and Sloboda2001); at the other extreme is the case where the person uses the music as a resource in a more active process of emotion construction (see DeNora Reference DeNora, Juslin and Sloboda2001; see also Meyer Reference Meyer1956, p. 11). These different views can be reconciled by observing that different mechanisms may be involved in each case: For instance, emotion induction through evaluative conditioning may really be direct and involuntary, whereas emotion induction through visual imagery may require active engagement of the listener. Only consideration of the mechanism involved can resolve this kind of argument.
The framework can also help to explain some previous disagreements about which emotions music can induce in a listener. Some researchers argue that music can induce basic emotions (Krumhansl Reference Krumhansl1997), while others deny that this is possible (Scherer Reference Scherer and Bresin2003). Some researchers argue that music can induce only “broad” positive and negative emotions (Clark Reference Clark1983), whereas others argue that music can induce a range of both basic and complex emotions (Gabrielsson Reference Gabrielsson, Juslin and Sloboda2001). However, as shown in Table 4, which emotions music can induce could depend on the precise mechanism involved. For example, emotional contagion may be limited to more basic emotions, whereas visual imagery may induce all possible emotions. Hence, although certain emotions (e.g., happiness, sadness, calm, nostalgia) may be especially common with regard to music (Juslin et al., submitted), we should be careful not to rule out the induction of other emotions. Which emotions music can induce depends on the functions of the music in a particular situation (e.g., using music to relax or to evoke nostalgic memories), and may thus vary considerably from one context to another. This implies that researchers should avoid settling prematurely on a particular conceptualization of emotions (e.g., discrete, dimensional, component, or music-specific) before more data regarding the frequency of different emotions to music in everyday life have been collected.Footnote 16
4.2. Musical emotions versus other emotions
A recurrent issue in research on musical emotions is whether musical emotions are somehow qualitatively different from other emotions in everyday life. Swanwick (Reference Swanwick1985), for example, suggests that “emotions in ‘life’ … and emotions we might experience as a result of engaging with music are not the same” (p. 29) (although he admits that “we are left trying to understand how ‘feelings’ in music relate to feelings in general,” p. 35). Similarly, Lippman (Reference Lippman1953) warns researchers not to
fall into the easy trap … of assuming that because musical and extramusical events both evoke emotions, they must evoke the same emotions … It is no more possible for a musical composition actually to arouse an instance of … sadness than it is for the stimulus of such an emotion to arouse the very emotion produced by a musical composition. (Lippman Reference Lippman1953, p. 563)
In contrast, the present framework implies that music recruits largely the same mechanisms as do other stimuli that induce emotions, and that the emotions evoked by music are largely similar. Some emotions may be more common than others in response to music, but the same is true of most other types of stimuli for emotions. For instance, some emotions might be more common than others in response to animals. Some emotions might be more common than others in response to sport events. Still, we would not propose a set of qualitatively unique emotions for each of these types of events. The burden of proof lies, in our view, on those who claim that there are music-specific emotions. Which are those emotions? What is their nature? So far we have not seen any evidence for the existence of music-specific emotions. A more parsimonious view is that there is one set of emotions that can be evoked in different ways and to different degrees by different stimuli. This view is consistent with findings from several studies suggesting that music evokes mostly the same emotions as other stimuli (Gabrielsson Reference Gabrielsson, Juslin and Sloboda2001; Juslin & Laukka Reference Juslin and Laukka2004; Juslin et al., in press; Sloboda Reference Sloboda, Riess-Jones and Holleran1992; Wells & Hakanen Reference Wells and Hakanen1991). What is unique about musical emotions is not the underlying mechanisms or the emotions they evoke, but rather the fact that music – unlike most other stimuli for our emotions in everyday life – is often intentionally designed to induce emotions, using whatever means available.
4.3. Relationships among mechanisms
Another implication of the framework is that music could induce so-called mixed emotions, because different mechanisms might be activated simultaneously at different levels. Thus, for example, a piece of music could make a listener happy because of the happy expression of the piece (emotional contagion), but at the same time make the listener sad because the piece reminds him or her of a sad event in the past (episodic memory). Thus, the end result may be a bitter-sweet feeling of both happiness and sadness. Instances of mixed emotions have been commonly reported in the literature (e.g., Gabrielsson Reference Gabrielsson, Juslin and Sloboda2001, p. 440), but no explanation has been offered previously. The current explanation requires that more than one mechanism can be activated at the same time – which remains to be demonstrated. However, this issue is not unique for musical emotions: It remains unclear to what extent emotions can generally reflect the output from many mechanisms simultaneously (Izard Reference Izard1993). In any case, the existence of mixed emotions speaks against using the “circumplex model” (Russell Reference Russell1980) to study musical emotions, since it precludes feeling both sad and happy at the same time (Larsen et al. Reference Larsen, McGraw and Cacioppo2001).
The possible co-activation of different psychological mechanisms – at least those that do not interfere with each other's information processing – suggests that an important task for future research is to examine possible interactions between different mechanisms. The mechanisms proposed here may seem simple: How can the extremely diverse music experiences reported by listeners in previous studies be reconciled with the simple theories proposed to account for these experiences? Part of the answer may be that the richness of our experiences comes from the complex interactions among these mechanisms, even within a single musical event. What mechanisms may be activated depends on several factors in the music (e.g., what information is available in the music?), the listener (e.g., is the listener's attention focused on the music?), and the situation (e.g., what are the circumstances of the listening context?). Thus, individual mechanisms may be expected to correlate with specific musical styles, listener states, listener activities, and listening situations. We see no a priori reason to assume that the mechanisms cannot be activated in isolation from each other, since they focus on different types of information and engage partly different brain regions (see Table 4). However, this is an empirical question to be resolved by further research.
One further implication is that emotions to music should change qualitatively across the life span, as the relative impact of the different psychological mechanisms changes. Preliminary evidence that there is a developmental trajectory for emotional responses to music has been reported (Schmidt et al. Reference Schmidt, Trainor and Santesso2003; Sloboda Reference Sloboda, Wilson and Roehmann1989), but more systematic study of such life-span changes seems warranted (see Table 4, Ontogenetic development). We would expect that emotional reactions to music proceed in a more or less orderly progression during the development, where listeners' reactions first focus on acoustic sensations (i.e., brain stem reflexes), then on the emotional expression in the music (i.e., emotional contagion), and then on more stylistic or formal characteristics of the music (i.e., musical expectancy). It should be noted that Swanwick and Tillman's model of musical skill development proposes a somewhat similar developmental trajectory (Swanwick Reference Swanwick2001). In both cases, the trajectory might reflect a gradual maturation of the child's cognitive functioning, as well as cultural learning. Thus, we would expect musical emotions to become increasingly multifaceted during the development, with increasing occurrence of mixed emotions (see also Larsen et al. Reference Larsen, To and Fireman2007; Peters et al. Reference Peters, Hess, Västfjäll and Auman2007).
4.4. The cost of neglecting mechanisms
The most important implication of the proposed framework for future research in the field is that it will not be sufficient to induce and study musical emotions in general. For data to contribute in a cumulative fashion to our knowledge, researchers must try to specify as far as possible the mechanism involved in each study. Otherwise, studies will produce results that are inconsistent, or that cannot be interpreted clearly. Lack of control with respect to mechanisms may also increase individual differences in listeners' responses, because without a systematic manipulation of stimuli, different listeners may activate different mechanisms to the “same” musical stimulus, with resulting differences in response (Table 4). While a neglect of mechanisms has been the rule rather than the exception, there are areas where this problem becomes particularly salient. A case in point is provided by the recent series of brain-imaging studies of musical emotions. Numerous brain regions have been implicated in these studies – including, but not limited to, thalamus, cerebellum, hippocampus, amygdala, cingulate cortex, orbitofrontal cortex, midbrain/periaqueductal gray, insula, Broca's area, nucleus accumbens, visual cortex, and supplementary motor areas (Bauer Alfredson et al. Reference Bauer Alfredson, Risberg, Hagberg and Gustafson2004; Blood & Zatorre Reference Blood and Zatorre2001; Blood et al. Reference Blood, Zatorre, Bermudez and Evans1999; Brown et al. Reference Brown, Martinez and Parsons2004; Gosselin et al. Reference Gosselin, Samson, Adolphs, Noulhiane, Roy, Hasboun, Baulac and Peretz2006; Koelsch et al. Reference Koelsch, Fritz, von Cramon, Müller and Friederici2006; Menon & Levitin Reference Menon and Levitin2005). However, different brain regions have been activated in different studies, without any clear explanation of why these differences occur.
We would argue that the main problem is that that neuropsychological studies have tended to simply present “emotional music” to listeners without manipulating, or at least controlling for, the underlying induction mechanism.Footnote 17 This makes it exceedingly difficult to understand what the obtained neural correlates actually reflect in each study (“It is not possible to disentangle the different subcomponents of the activation due to limitations of this experimental design”, Bauer Alfredson et al. Reference Bauer Alfredson, Risberg, Hagberg and Gustafson2004, p. 165). Given the aim of studying emotional reactions to music, one would expect the manipulation of musical stimuli to be essential to the task. Yet, stimuli have been selected non-systematically (e.g., instrumental songs of the rembetika style, joyful dance tunes, listener-selected music). The fact that different studies have reported activations of different brain regions does suggest that different mechanisms were involved. But, after the fact, there is no way of knowing. This shows that musical emotions cannot be studied without regard to how they were induced. On the other hand, if researchers could manipulate separate induction mechanisms in future listening experiments, they would be better able to explain the obtained brain activation patterns. Indeed, to the extent that we can obtain systematic relations among mechanisms and brain regions, we might eventually be able to discriminate among the mechanisms based on brain measures alone. However, no study published so far has quite the specificity needed to contribute to that goal.
4.5. Implications for emotion research
The present framework might have some broader implications, as well. Thus, for instance, the study of musical induction of emotions along the lines suggested here could benefit the field of emotion as a whole. A serious problem in studying emotions has been the methodological and ethical difficulties involved in inducing strong emotions in the laboratory. Many studies in the field of emotion either lack experimental control (when using naturalistic settings) or achieve only a limited variation in target emotions and limited ecological validity (when using laboratory settings) (see Parrott & Hertel Reference Parrott, Hertel, Dalgleish and Power1999). Music could evade some of these problems by offering new paradigms for emotion induction, especially with regard to positive emotions, which have tended to be neglected in previous research. Musical structure is easy to manipulate in psychological experiments and is a frequent source of emotion in everyday life. Thus, studies of music could provide an additional source of evidence concerning emotions.
The unique characteristics of the various induction mechanisms (see Table 4) will be crucial when researchers design experiments that aim to induce a specific emotion. Specifically, it is important that the study involves an induction procedure that allows for the induction of that emotion. Some procedures may limit the kind of emotions that can be induced depending on the mechanism involved (e.g., Table 4, Induced affect). Some mechanisms require particular acoustic characteristics in the stimulus (e.g., emotional contagion), others require a prolonged encoding phase (e.g., evaluative conditioning), and still others require sufficient listening time in order for a sufficient amount of structure to unfold (e.g., musical expectancy). Thus, to facilitate studies of musical emotions, we should try to create standard paradigms and tasks that reliably induce specific emotions in listeners through each of the mechanisms outlined here earlier. This would be analogous to the different tasks used to measure distinct memory systems (Tulving Reference Tulving1983). A more systematic and theoretically informed approach to the manipulation of musical stimuli would be a significant advance compared to the mostly intuitive selection of stimuli in current studies using music as an emotion-elicitation technique (Eich et al. Reference Eich, Ng, Macaulay, Percy, Grebneva, Coan and Allen2007; Västfjäll Reference Västfjäll2002a).
Possible stimuli and procedures for inducing different kinds of musical emotions can already be found in the literature, although they need further evaluation and refinement. For instance, paradigms aimed at activating brain stem reflexes could rely on psycho-acoustic models that specify quantitative relationships between sound stimuli and auditory perception (Zwicker & Fastl Reference Zwicker and Fastl1999). Paradigms aimed at activating the evaluative conditioning mechanism could use established procedures from studies of conditioning (Lavond & Steinmetz Reference Lavond and Steinmetz2003). Paradigms aimed at activating the emotional contagion mechanism could create stimuli based on similar emotion-specific patterns of acoustic cues in speech and music (Juslin & Laukka Reference Juslin and Laukka2003, Table 7), perhaps also using timbres that are “voice-like,” such as those of the cello and the violin. Paradigms aimed at activating the visual imagery mechanism could rely on extensive programs of music developed especially for the purpose of stimulating imagery to music in therapy (Bruscia & Grocke Reference Bruscia and Grocke2002, e.g., Appendices B–L). Paradigms aimed at activating the musical expectancy mechanism could rely on both stimuli and procedures that have already been used to explore syntactical processing in music perception (Koelsch et al. Reference Koelsch, Gunter, Friederici and Schröger2000). Perhaps the most difficult mechanism for musical emotion induction to activate in a controlled way in the laboratory is episodic memory, because the laboratory situation is not conducive to establishing the strong personal significance needed to encode an emotional episodic memory.
To explore the mechanisms and test the hypotheses in Table 4 fully, we need not only be able to activate each mechanism. To separate the effects of different mechanisms, we must also be able to suppress or eliminate particular mechanisms in individual cases. Although space does not permit a detailed exposition of experimental set-ups in this target article, we propose that this could be done in two principal ways. Firstly, one could manipulate stimuli in such a way as to withhold or eliminate information required for a specific mechanism to be activated (the principle of information impoverishment). Musical structures are easy to manipulate, and there are sophisticated techniques in acoustics that enable researchers to standardize a stimulus with regard to certain acoustic features, while leaving others intact. Secondly, one could design the procedure in such a manner that it will prevent the type of information processing required for a particular mechanism to be activated (the principle of interference). This could be done in a number of ways. One approach could be to force listeners to allocate the cognitive resources needed for a specific mechanism to a task instead; for instance, one could use an experimental task that recruits attentional resources to such an extent that visual imagery, also dependent on these resources, will be made impossible. Another possibility could be to use a neurochemical interference strategy; for example, it has been shown that blocking of a specific class of amino acid receptors (N-methyl-D-aspartate or NMDA) in the lateral amygdala can interfere with the acquisition of evaluative conditioning (Miserendino et al. Reference Miserendino, Sananes, Melia and Davis1990). Yet another form of interference involves the use of transcranial magnetic stimulation (Pascual-Leone et al. Reference Pascual-Leone, Davey, Rothwell, Wassermann and Puri2002). By disrupting brain activity at crucial times and locations, one may prevent specific mechanisms from becoming activated by a musical stimulus.
Another implication concerns the role of cognitive appraisal relative to other mechanisms. A common characteristic of human behavior is that it is multiply determined (Brunswik Reference Brunswik1956). This is true also for emotions, although the possibility of multiple induction mechanisms that interact has been somewhat neglected in previous research (but see Izard Reference Izard1993). It is usually assumed that appraisals account for the lion's share of emotions in everyday life, but there is little formal evidence so far to support this notion – primarily because it is difficult to test the notion using the type of “post hoc” self-reports of emotions that have dominated in studies of cognitive appraisal to date (Frijda & Zeelenberg Reference Frijda, Zeelenberg, Scherer, Schorr and Johnstone2001). A crucial question is to what degree the additional mechanisms described here play a role in non-musical emotional episodes. The present framework implies that there is no simple “one-to-one” relationship between cognitive appraisals and emotions. Instead, there are several mechanisms that – singularly or together – determine emotional outcomes, according to the precise conditions of the situation. Ellsworth (Reference Ellsworth, Ekman and Davidson1994) acknowledges that musical emotions pose “a real threat to the generality of appraisals as elicitors of emotion” (p. 195). To the extent that a great deal of our emotional responses in everyday life involve mechanisms such as conditioning, contagion, and episodic memory, an approach similar to that advocated in this target article could be fruitful also in understanding non-musical emotions. Does this mean that what we claim about music – that emotions cannot be studied without regard to how they were evoked – is true of non-musical emotions as well? To the extent that the received view is correct – namely, that non-musical emotions are mostly induced through cognitive appraisal (Ellsworth Reference Ellsworth, Ekman and Davidson1994; Scherer Reference Scherer, Dalgleish and Power1999) – the issue of controlling for the underlying mechanism may not be as important outside the musical domain. However, this is an empirical question that awaits further research.
5. Concluding remarks
It could appear that our claim that musical emotions must be investigated with regard to their underlying mechanisms is uncontroversial, and that all music researchers would agree. Yet, this is not how research has been conducted, which is ultimately what counts. Studies thus far have produced data that are collectively confusing and internally inconsistent, mainly because researchers have been considering only the induced emotions themselves, instead of trying to manipulate the underlying mechanisms in a systematic manner. We argue that much progress may be achieved, provided that more rigorous theoretical and methodological approaches are adopted. Considering the crucial implications that such an endeavor could have for both basic and applied research in music psychology and psychology in general, this opportunity should not be missed. For instance, it has been increasingly recognized that music may have positive effects on physical health and subjective well-being (e.g., Khalfa et al. Reference Khalfa, Dalla Bella, Roy, Peretz and Lupien2003; Pelletier Reference Pelletier2004). We suggest that many of these effects are mediated by the emotions that the music induces. A better understanding of the mechanisms underlying these emotions could therefore be of great importance for applications, such as music therapy.
Meyer (Reference Meyer1956), one of the pioneers in this field, argued that “given no theory as to the relation of musical stimuli to affective responses, observed behavior can provide little information as to either the nature of the stimulus, the significance of the response, or the relation between them” (p. 10). In other words, amassing data on listeners' emotional reactions to music is not fruitful, unless one is able to interpret these data in the light of an explanatory theory. In this target article, we have proposed a theoretical framework and a set of hypotheses that may aid researchers in exploring the manifold and different mechanisms that relate music to emotions – all musical emotions are not created equal.
ACKNOWLEDGMENTS
This research was supported by the Swedish Research Council through a grant to Patrik N. Juslin. Part of this article was first presented at the EuroScience Open Forum, Stockholm, Sweden, August 2004. We are grateful to John Sloboda, Barbara Finlay, Aaron Williamon, Petri Laukka, Simon Liljeström, and a number of anonymous reviewers for useful comments on previous versions of the article. We dedicate this article to the memory of Leonard Meyer, who passed away on December 30th, 2007.
