Comparative effects of music and recalled life-events on emotional state

K E Y W O R D S : aesthetic awe, being moved, expression, music and emotion, recall of life-events Music, as a temporal array of auditory stimuli, clearly possesses the means to mimic the patter of feet of children at play, the sound of water bouncing off rocks in a mountain brook, and the slow motion of a profoundly sad person or a funeral procession (cf. Avison, 2003[1752]: 4; Gurney, 1966[1880]: 169; Langer’s (1942: 185) symbolic ideas; Arnheim’s (1954: 434) ‘physiognomics’ in art; Budd’s (1985: 47), Davies’s (2001: 31) and Kivy’s (1989: 46) complementary analyses of the ‘“physiognomy” of musical expression’; and the ‘iconic relationships’ discussed by Sloboda and Juslin (2001: 93)). Through the appropriate choice of register, dynamics, event density, mode, rhythm, melodic contour and harmonic change, among other structural means, music can express, depict, allude to and evoke both the differential auditory patterns commonly associated in the abstract with the fundamental emotions (such as Psychology of Music Psychology of Music Copyright © 2008 Society for Education, Music and Psychology Research vol 36(3): 289–308 [0305-7356 (200807) 36:3; 289–308] 10.1177/0305735607082621 http://pom.sagepub.com distribution. © 2008 Society for Education, Music, and Psychology Research. All rights reserved. Not for commercial use or unauthorized at UNIV CALIFORNIA SAN DIEGO on July 1, 2008 http://pom.sagepub.com Downloaded from 290 Psychology of Music 36(3) joy, sadness and anger), and the specific physical and vocal behavior of a human or animal experiencing and displaying such emotions (cf. Gabrielsson and Lindström, 2001; Hevner, 1936, 1937). In short, music can tell a story about emotions; it can refer to and describe the features of emotions and their display. What is commonly labeled as ‘emotional expression’ or ‘perceived emotion’ in music is really a semiotic issue, one of denotation (Meyer, 1956: 8) or representation.1 Generally, both the performers and listeners of music are intimately familiar with the morphological and acoustic features of emotion-driven behaviour. Therefore, it is not surprising that a skilled guitarist can take a folk tune, such as ‘Greensleeves’, perform it in four versions – ‘happy’, ‘sad’, ‘angry’ and ‘fearful’ – and justifiably expect that these intended emotional expressions will be correctly identified by the listeners (Juslin, 2000). Note, however, that communicating, by musical means, some surface attributes of emotion is very different from ‘communicating emotion’ in the sense of a genuinely happy or sad person conveying his emotions to others – yet this crucial difference is minimized in the very title and abstract of Juslin’s (2000: 1797) article (Konečni, 2003: 337). The above point brought us a step closer to the central issue of our article: the possibility of music being able to induce emotions in the listeners (including in its composers and performers; however, these categories of participants in the music process require separate consideration: see many contributors’ chapters in Juslin and Sloboda (2001), and also Konečni (2003), London (2001–02: 26) and Waterman (1996)). The question of induction of emotion can be broken down into two related parts. One concerns the ability of music to induce emotions that are comparable to those that are induced by real-life events or their intensive recall (cf. Juslin and Zentner, 2001–02: 9; Krumhansl, 1997: 349; Sloboda and Juslin, 2001: 91). The second deals with the possibility that music induces ‘musical emotions’ – states that are different in kind or intensity from those caused by real-life events (that is, events other than music), but are nevertheless, somehow, emotions (cf. Krumhansl, 2002: 45; London, 2001–02: 30). The first question is largely empirical and is directly addressed by the research we report here. The second is largely theoretical and definitional, and one of us has dealt with it by suggesting that the term ‘musical emotions’ (and, more broadly, ‘aesthetic emotions’) be replaced by ‘being moved’ and ‘aesthetic awe’ (Konečni, 2005, 2007; see ‘Discussion’ later). In our experiment, all research participants intensively recalled (the RE factor, Remembering Events, modeled after the Relived Emotion task of Levenson et al., 1991: 30) either ‘the most happy’ (REH level) or ‘the most sad’ event (RES level) in their lives that ‘they could think of at the moment’. They then rated their emotional state ‘now’ (ENOW measure) and ‘at the time the event originally occurred’ (ETHEN measure) on a 13-point bipolar happy–sad scale. All participants also listened to music (the LM factor, Listening to Music) that had been selected on the basis of prior research (Krumhansl, 1997: 340) and our pilot studies: They listened either to Happy (LMH level) or Sad (LMS level) music and then rated, using the same 13-point happy– sad scale, both their emotional state (Induction of Emotion by Music (IEM) measure) and how the music sounded (Expression or Depiction of Emotion by Music (EDEM) measure). Note that the designations of happy and sad music should have the adjectives in quote-marks; even though the designations follow folk parlance and music distribution. © 2008 Society for Education, Music, and Psychology Research. All rights reserved. Not for commercial use or unauthorized at UNIV CALIFORNIA SAN DIEGO on July 1, 2008 http://pom.sagepub.com Downloaded from Konečni et al.: Music and recalled life-events 291 researchers’ perhaps unfortunate habit of going along with it (cf. Kivy, 1989: 153), it is important for theoretical reasons to remember that music is not a sentient being (Budd, 1985: 37; Davies, 2001: 25–6). The mentioned aspects of the design and its other features – for example, both the RE and LM factors also had neutral controls – allowed us, above all, to compare the emotions arising in the non-musical ‘real world’ to those potentially induced by music. To the best of our knowledge, this is the first attempt in the literature to make this comparison directly and experimentally.2 The question of induction of emotion by music contains several additional complications. One is the classical question of the role of associations that people have when listening to music (e.g. Avison, 2003[1752]: 4; Goldstein, 1980: 127; Kivy, 1989: 29; Scherer et al., 2001–02: 150; Sloboda and Juslin, 2001: 94; Waterman, 1996: 59). The bubbling vivacity of Vivaldi’s ‘Spring’ may make someone who had never heard it before think of a budding love affair and of life’s bountiful offerings and seemingly endless possibilities, in addition to producing someone else’s explicit ‘our tune’ or ‘my time in Venice’ memories. A causal model of the form M E that many music psychologists implicitly espouse is theoretically very different from the one in which the causal effect of music on emotion is fully dependent on the interpolated cognitive content (M Assoc E). In the latter model, music has an effect on emotions only because it gives rise to mental associations; even though non-musical events may also trigger such associations, music may be, for a variety of reasons, a particularly powerful trigger.3 In our experiment, one-half of the participants did the RE task first, followed by the LM (RE1LM2 level of the RELM Order factor), whereas the other half encountered the reverse order of tasks (RE2LM1). The RELM factor did not simply control for order effects. Our thinking was that the participants who listen to music in the RE1LM2 order would be more likely to have vivid associations to the music (especially in the congruent REH/LMH and RES/LMS cells) than those in the RE2LM1 order of the two tasks – which is, of course, not to say that listening to music without prior pondering of real-life emotional events precludes associations with emotional content. Another matter that complicates the research on the induction of emotion by music lies in the prevailing romantic-sentimental folk theories about the inescapable link between music and emotion that most participants bring to experiments – and the response ‘errors’ that they consequently tend to commit. Specifically, a considerable amount of unpublished interview data from our laboratory shows that the participants’ tendency is to take into account the music’s message about emotion when asked to rate, on some intensity scale, their own emotional response to the music – with the extent of this misattribution depending on the details of the instructions. The common outcome of such a tendency is to inflate the ratings of own emotion. This is a problematic conceptual and measurement issue of considerable significance for the field of music and emotion (Gabrielsson, 2001–02: 124; Juslin and Zentner, 2001–02: 11; Kivy, 1989: 162; Meyer, 1956: 8; Scherer and Zentner, 2001: 379–80). In the present experiment, the issue was addressed by having one-half of the participants, after listening to music, answer first the question about how the music sounded, before telling us how they themselves felt (EDEM1IEM2 level of the EDEM/IEM Order factor), whereas the other half first reported about their own state and then rated the music (EDEM2IEM1 level). In addition to this being another control distribution. © 2008 Society for Education, Music, and Psychology Research. All rights reserved. Not for commercial use or unauthorized at UNIV CALIFORNIA SAN DIEGO on July 1, 2008 http://pom.sagepub.com Downloaded from 292 Psychology of Music 36(3) order factor, we thought that, for the IEM measure, the participants would be less likely to be rating the sound, even though they were asked to rate themselves, in the condition in which the rating of the music was salient by being made first (EDEM1IEM2). A complementary possibility, for the EDEM measure, was that having rated their own state first, in the EDEM2IEM1 order, the participants would subsequently additionally increase (that is, make more extreme) their EDEM ratings. We say ‘additionally,’ because the above analysis should be viewed against the backdrop of our expectation, based on years of pilot data, that the ratings of the ‘expressivity’ of the music (EDEM) would overall be more extreme than the ratings of own emotional reaction to music (IEM). After all, the Happy and Sad pieces were chosen for their maximum objective respective valence in the domain of instrumental classical music by both Krumhansl (1997) and us, whereas many of the participants may well have been indifferent (or even hostile) to the classical idiom. The last, but not least, issue of concern when contemplating the induction of emotion by music concerns the often rather different viewpoints of the psychology-ofemotion and music-psychology researchers. In general, emotion theorists do not regard fluctuations in peripheral arousal or, for example, changes in cerebral blood flow in brain areas that are also involved with emotion, motivation and consumption, as sufficient evidence to indicate emotion (e.g. Konečni, 2003: 333, 2005: 38; Konečni et al., 2007; Scherer and Zentner, 2001: 363), whereas music and neuro-music psychologists often do (e.g. Blood and Zatorre, 2001: 11823; Peretz, 2001: 116; Rickard, 2004: 373). Our view (see Figure 1) is that one can be highly aroused, for example, by the rhythmic aspects of the music, one can be foot-tapping and dancing, one can be startled by the dynamics and the dissonances and yet experience no identifiable emotion whatsoever (in accord with Konečni’s, 2007, prototypical emotion-episode model, PEEM). The physiological record alone does not equal the presence of emotional experience: ratings of one’s subjective state or verbal reports of one’s experience are indispensable to show the induction of emotion (cf. Gabrielsson, 2001–02: 128; Konečni, 2007). In our view, music can add to the physiological substrate of genuine emotions, but not cause the emotions themselves. This view would augment the above-mentioned M Assoc E model. In short, if for no reason other than the fact that the matter of the induction of emotion by music is very complicated and multi-faceted, the ‘cognitivist’–’emotivist’ dichotomy (Kivy, 1990: 146), often cited by music psychologists, seems to be a considerable oversimplification (although we, unlike many music psychologists, agree in general with Kivy’s rejection of the ‘emotivist’ position). The position on the music – emotion relationship that we have presented so far is summarized in Figure 1. It serves as the background for the experimental design and the predictions. Before presenting our predictions, we shall examine several experiments that are usually quoted as supportive of the idea that music is an inductor of emotion. The foremost of these is Krumhansl’s (1997) study, in which one group of 10 Cornell participants was ‘instructed to continuously adjust ... the slider on the display to indicate the amount of sadness they experienced while listening’ (p. 340) to six three-minute musical selections (including two meant to represent ‘sadness’, two ‘happiness’ and two ‘fear’). Three other groups of 10 participants each were asked to adjust the slider distribution. © 2008 Society for Education, Music, and Psychology Research. All rights reserved. Not for commercial use or unauthorized at UNIV CALIFORNIA SAN DIEGO on July 1, 2008 http://pom.sagepub.com Downloaded from Konečni et al.: Music and recalled life-events 293 to ‘judge fear, happiness, and tension, respectively’ (p. 340) in response to the same six music selections. To ask participants to report their ‘sadness’ continuously (in emotion theory and in real life a powerful fundamental emotion), in response to six three-minute musical selections of different valence appears a somewhat odd task, one that seems particularly prone to incline the participant toward judging the music instead of her own state. In the Berkeley part of Krumhansl’s study (in R.W. Levenson’s laboratory), involving 38 participants, although there were physiological differences in response to the various musical selections (as would also be expected from our viewpoint expressed above), ‘few of the correlations between self-reports and average physiological measures were either significant or marginally significant ... [n]or did they correspond with correlations using dynamic measures of physiology and emotion’ (Krumhansl, 1997: 347) – again, as one might expect from the standpoint that the physiological response does not necessarily equal or lead to emotional experience. Krumhansl (1997: 349) took an important additional analytic step and compared her physiological data to those obtained in three major published studies of emotion physiology, finding ‘little correspondence with [her] results’. Another well-known study that is often imprecisely cited by music psychologists as showing the induction of emotion by music was performed by Nykliček et al. (1997). In fact, all 25 (Dutch) research participants, who listened, in the ‘music selection stage’, Music Representation, Denotation (onomatopoeia, ‘physiognomics’)