Page  00000001 Psychophysiology of Musical Emotions Carol L. Krumhansl Department of Psychology, Cornell University clk4@cornell.edu Abstract This paper focuses on emotional reactions to music, taking the approach of correlating psychophysiological measures with dynamic changes in emotional reactions to music. A philosophical debate about musical emotions concerns whether music produces emotional changes in listeners (the "emotivist" position) or simply expresses emotions that listeners recognize in the music (the "cognitivist" position). Psychophysiological measures were recorded while listeners heard two excerpts chosen to represent each of three emotions: sad, fear, and happy. The measures covered a fairly wide spectrum of cardiac, vascular, electrodermal, and respiratory functions. The sad excerpts produced the largest changes in heart rate, blood pressure, skin conductance and temperature. The fear excerpts produced the largest changes in blood transit time and amplitude. The happy excerpts produced the largest changes in the measures of respiration. These emotion-specific physiological changes only partially replicated those found for non musical emotions. However, the physiological effects of music observed generally support the emotivist view of musical emotions. 1 Introduction A philosophical debate about musical emotions concerns whether music produces emotional changes in listeners (the "emotivist" position) or simply expresses emotions that listeners recognize in the music (the "cognitivist" position). Peter Kivy (1990), one of the strongest proponents of the latter position, distinguishes between musical "cognitivists" and musical "emotivists" as follows: "Those I am calling musical emotivists believe that when, under normal circumstances, musical critics, theorists, or just plain listeners call a piece of music (say) "sad,", it is because it makes us sad when we listen to it; and what they mean by "sad" music, I will assume, is music that normally arouses sadness in the normal listeners. The musical cognitivists, like the emotivists, believe that it is proper sometimes to describe music in emotive terms. But unlike the emotivists, they do not think that sad music is sad in virtue of arousing the emotion in the listeners. Rather, they think the sadness is an expressive property of the music which the listener recognizes in it..." Kivy rejects the emotivist position absolutely on the grounds "that there are no behavioral symptoms of listeners actually experiencing [emotions] when attending to music, is... decisive." (p. 151) The cognitivist position is represented also by Leonard Meyer (1956, p. 8), who states: "... it may well be that when a listener reports that he felt this or that emotion, he is describing the emotion which he believes the passage is supposed to indicate, not anything which he himself has experienced." He dismisses the importance of physiological changes in response to music, such as heart rate and skin conductance, as being inconclusive on two grounds. First, "...no relation can be found between the character or pattern of the musical selection evoking the response and the particular physiological changes which take place." Second, "... it seems more reasonable to assume that the physiological changes observed are a response to the listener's mental state rather than to assume that tone as such, in some mysterious and unexplained way, bring these changes about directly." (p. 11) Meyer then goes on to offer a theory that ascribes musical emotions to variations in tension and release from tension operating over short spans of time, rather than to more general characteristics that may establish moods over longer durations. The cognitivist position stands in stark contrast to the assumption either implicit or explicit in studies of musical emotions, that musical emotions are of the same kind felt in non musical situations. Most available studies use verbal reports of emotional responses to music, and these find considerable consistency across individuals. The results of these studies, and

Page  00000002 advances in psychological theory and methodology for studying emotions, motivate further study of musical emotions. The experiment reported here takes the approach of measuring psychophysiologal changes during music listening. According to a recent survey of the field of psychophysiology (by Kenneth Hugdahl, 1995, p. 8), "Psychophysiology is the study of brain-behavior relationships in the framework of peripheral and central physiological responses." Recording of psychophysiological responses is regarded as a "window" into the brain and mind. These responses include measures of the central nervous system (through electroencephalograms, event-related potentials, and more recently brain imaging techniques) and peripheral nervous system (electrodermal activity, heart and blood circulation, respiration, and muscular activity). The present study exclusively considers measures of peripheral nervous system function. 2 Experiment The experiment is of an exploratory nature. It sought to obtain a fairly wide spectrum of physiological measures of cardiac, vascular, electrodermal, and respiratory function on the same subjects while listening to music. These were taken in the laboratory of Robert Levenson. The measures were taken continuously (at onesecond intervals) during the music. Six excerpts were chosen to represent (two selections each) the emotions of sad, fear, and happy. One question of interest is whether different configurations of physiological responses correspond to these different emotions. Each of the excerpts was approximately three minutes in duration. Each excerpt was preceded by a 90-second period of silence during which baserate measures were collected. These values were subtracted from the pre-music measure to control for individual differences and overall changes that might occur during the experimental session. In addition, listeners gave self-report ratings of their emotional responses during the excerpts. However, the primary emphasis is on dynamic changes that occurred during the musical excerpts. Therefore, independent groups of listeners made dynamic ratings of the degree of sadness, fear, and happiness they experienced while listening. The six selections did, as indicated by self-report measures, produce the intended emotions. Both groups of listeners, those who gave dynamic emotion quality ratings and those whose physiological measures were taken, reported very similar responses to the selections. Intersubject consistency was also very strong. The subjects judged their response to the excerpts by Albinoni (Adagio in G minor for Strings and Orchestra) and Barber (Adagio for Strings) to be strongest for the emotion of sad. They judged their response to the excerpt by Holst (Mars -- the Bringer of War from The Planets) and Mussorgsky (Night on Bare Mountain) to be strongest for the emotions of anxious, afraid, and surprised. They judged their responses to the excerpt by Vivaldi (La Primavera -- Spring -- from The Four Seasons) and Hugo Alfven (Midsommarvaka) to be strongest for the emotion of happy, followed by amused and contented. The dynamic ratings of emotion over time also showed the intended emotions of the chosen excerpts. All six excerpts were judged, by different groups of listeners, on a second-bysecond basis for the degree of sadness, fear, and happiness. As expected, the ratings of sad were highest for the Albinoni-Barber pair, the ratings of fear were highest for the Holst-Mussorgsky pair, and ratings of happy were highest for the Vivaldi-Alfven pair. We can now turn to the psychophysiological measures themselves. The data are complex, with twelve different physiological measures and second-by-second measures for six three-minute long excerpts. The first result was that the presence of music significantly affected all twelve physiological measures, and the direction of the changes was the same for all three excerpt types. The changes were: longer cardiac interbeat interval (increased IBI, slower heart rate), longer pulse transmission times to ear and finger (increased FPTT and EPTT) and reduced blood amplitude at the finger (decreased FPA), decreases in respiration intercycle interval, depth, and respiration-sinus asynchrony (decreased ICI, which means a faster breathing rate, and decreased RD and RSA), higher systolic, diastolic, and mean arterial blood pressure (increased SBP, DBP, and MAP), and lower skin conductance and finger temperature (decreased SCL and TEM). Thus, the presence of music had effects on all of these physiological measures. The data argues against the cognitivist position taken, for example, by Kivy (1990) and his claim that there are no behavioral indicators that listeners experience emotions when listening to music.

Page  00000003 Not only do listeners verbally report emotional responses to music with considerable consistency, music appears to produce physiological changes as well. The central question that remains, then, is whether there are emotion-specific patterns in the psychophysiology. This question has a long and vexed history going back to William James (1890) who proposed that different emotions are associated with unique patterns of physiological changes. It was hypothesized that these patterns differentiate the emotions, in other words, identification of the emotion felt is determined by a reading of the physiological responses. An influential study by Schacter and Singer (1962) undercut this position by showing that undifferentiated arousal resulted in different reports of emotions depending on the subject's cognitive response to external events. More recently, the first position -- the physiological differentiation of emotions -- has been strengthened by a series of studies by Paul Ekman, Robert Levenson and colleagues (e. g., Ekman, Levenson, & Friesen, 1983; Levenson, Ekman, & Friesen, 1990; Levenson, 1992, 1994; see however, Zajonc and McIntosh, 1992; Cacioppo, Klein, Berntson, and Hatfield, 1993). These studies suggest consistent differences in some physiological correlates of anger, fear, sadness, and happiness. A number of analyses were conducted to assess emotion-specific physiology in the present study. The primary analysis correlated the dynamic physiology ratings with the dynamic ratings of emotion quality (sadness, fear, and happiness). Strongest correlations with ratings of sadness were found for: cardiac interbeat interval (positive correlation with IBI), the three measures of blood pressure (positive correlations with SBP, DBP, MAP), skin conductance level (negative correlation with SCL), and finger temperature (negative correlation with TEM). Strongest correlations with ratings of fear were found for: finger and ear pulse transmission time (positive correlations with FPTT and EPTT), and finger pulse amplitude (negative correlation with FPA). Strongest correlations with the ratings of happiness were found for the respiration measures of intercycle interval, respiration depth, and respiration-sinus asynchrony (negative correlations with ICI, RD, and RSA). Thus, the dynamic emotion ratings of sad, fear, and happy, and were aligned with changes in different physiological systems. Sad was associated with changes in measures of cardiac and electrodermal systems, fear with changes in cardiovascular measures, and happy with respiration measures. These results suggest that musical emotions may indeed be reflected in psychophysiological measures, supporting the emotivist position that musical emotions are felt emotions. But, are these changes like those found with non musical emotions? First, I will consider the summary by Zajonc and McIntosh's (1992) of the directed facial action task and the relived emotions task in the work of Paul Ekman and Robert Levenson. In the directed facial action task, where most of the significant effects have been found, subjects posed different emotional expressions. Little agreement was found with the present results. In fact, quite a few of the differences found in those studies were opposite those found in the present experiment. Possibly, as Frans Boiten (1996) has suggested, the physiological changes in the directed facial actions task may depend more on the difficulty of producing the facial expression than the underlying emotion. Better agreement was found with the summary of data by Cacioppo et al. (1993). Of all seven cases with significant differences in this and the reviewed studies, the direction of the difference was the same. Emotion-specific differences were found for heart rate (Averill, 1969, using film manipulation of emotion; Tourangeau and Ellsworth, 1979, also using film manipulation), for finger temperature, skin conductance level, and finger pulse amplitude (Stemmler, 1989, real life manipulation, where fear was induced by a scary radio play and music and unexpected darkness, and happy was induced by a nice experimenter, extra monetary bonus, and a shorter experiment), and both systolic blood pressure and diastolic blood pressure changes (Averill, 1969, using a film manipulation). It may be significant that all the manipulations in these studies were extended over time as were the musical excerpts in this experiment. Thus, the question of emotion-specific physiology remains somewhat elusive. However, the results encourage future empirical research and theorizing about the connections between cognitive and emotional responses to musical sounds.

Page  00000004 References Averill, J. R. (1969). Autonomic response patterns during sadness and mirth. Psychophysiology, 5, 399-414. Boiten, F. (1996). Autonomic response patterns during voluntary facial action. Psychophysiology, 33, 123-131. Cacioppo, J. T., Klein, D. J., Berntson, G. G., & Hatfield, E. (1993). The psychophysiology of emotion. In M. Lewis & J. M. Haviland (Eds.), Handbook of emotions. New York: Guilford. Ekman, P., Levenson, R. W., & Friesen, W. V. (1983). Autonomic nervous system activity distinguishes between emotions. Science, 221, 1208-1210. Hugdahl, K. (1995). Psychophysiology: The mind-body perspective. Cambridge, MA: Harvard University Press. James, W. (1890/1950). Principles of psychology. New York: Dover. Kivy, P. (1990). Music alone: Philosophical reflections on the purely musical experience. Ithaca: Cornell University Press. Levenson, R. W. (1992). Autonomic nervous system differences among emotions. Psychological Science, 3, 23-27. Levenson, R. W. (1994). The search for autonomic physiology. In P. Ekman & R. J. Davidson (Eds.), The nature of emotion. New York: Oxford University Press. Levenson, R. W., Ekman, P., & Friesen, W. V. (1990). Voluntary facial action generates emotion-specific autonomic nervous system activity. Psychophysiology, 27, 363-384. Levenson, R. W., Ekman, P., Heider, K., & Friesen, W. V. (1992). Emotion and autonomic nervous system activity in the Minangkabau of West Sumatra. Journal of Personality and Social Psychology, 62, 972-988. Meyer, L. B. (1956). Emotion and meaning in music. Chicago: University of Chicago Press. Schacter, S., & Singer, J. E. (1962). Cognitive, social, and physiological determinants of emotional state. Psychological Review, 69, 379-399. Stemmler, D. G. (1989). The autonomic differentiation of emotions revisited: Convergent and discriminant validation. Psychophysiology, 26, 617-632. Tourangeau, R., & Ellsworth, P. C. (1979). The role of facial response in the experience of emotion. Journal of Personality and Social Psychology, 37, 1519-1531. Zajonc, R. B., & McIntosh, D. N. (1992). Emotions research: Some promising questions and some questionable promises. Psychological Science, 3, 70-74.