Psychoacoustical and psychophysiological correlates of the emotional impact and the perception of music
Frederik Nagel · 2007
Music plays an important role in everyday human life. One important reason for this is the capacity of music to influence the emotions of the listeners. Sometimes music is even accompanied by physical reactions, the so-called “strong experiences of music” (SEMs). Very pleasurable emotions are referred to as “chills”, which denote shivers down the spine or goose pimples. The present study consists of two experiments investigating both the role of psychoacoustical features and social backgrounds of participants, and physiological correlates of chills. An adequate study of emotions in music should be based on the real-time measurement of self-report data using a continuous response method. Therefore, a recording system is proposed in this thesis which considers two important aspects: firstly, experimental and technical standards for continuous measurement should be taken into account for a better comparison of results, and secondly, the recording system should be open to the inclusion of multimodal stimuli. Based on these two primary considerations, four basic principles of the continuous measurement of emotions are addressed: (a) the dimensionality of the emotion space; (b) the synchronization of media and self-reported data; (c) the interface construction for emotional responses; and (d) an integrated software solution. Researcher-developed software (EMuJoy) is presented as a freeware solution for the continuous measurement of responses to different media. Participants (n = 38) in the first experiment listened to both pre-selected musical pieces and their own favorite emotion-inducing pieces and reported their chill experiences. They were asked to report their emotions in real-time using a two-dimensional emotion space in a computer software. Additionally, psychophysiological measurements, such as heart rate (HR) and skin conductance (SC), were taken. Films were created from the time-series of all self-reports as a synopsis for seven musical pieces heard by all participants. This technique of data visualization allows a very aesthetic method of data analysis and provides the opportunity to investigate commonalities of emotional self-reports as well as differences between individuals. In addition to the presentation of the films, possible applications are discussed in areas such as social sciences, musicology and the music industry. Psychoacoustical parameters of musical excerpts, each 20 s in length, with chill-reports in the middle of the time window were analyzed and compared with musical excerpts of the same length without chill responses. A significant increase of loudness in the frequency range between 8 and 18 Barks (920-4400 Hz) was found in excerpts with which chills were experienced. The relationship between self-reported chills and the psychophysiological parameters HR, HR variability (HRV) and SC were also investigated in this exploratory study. Standard deviation of instantaneous heart rate was found to be greater in musical pieces where chills occurred than in those without chills. SC increased by 79.8% and HR by 70.6% when chills were reported, regardless of the specific musical piece. However, the correlation between chills and physiology is weak. The maximum skin conductance response lagged behind the onset of self-reported chills by 3.5 s; and the maximum heart rate lagged behind by 1.5 s. Crosscorrelating skin conductance and heart rate allows the quantification of the lag both between as well as within participants. The physiological time-series data showed a typical pattern for the mean values when chills were experienced. It could, however, be demonstrated that though physiology is not appropriate as a sole indicator of strong emotions, it is necessary in order to validate psychological self-reports when chills are defined as emotional reactions with the cooccurrence of both physiological changes and the experience of the chill. In a second experiment, choral singers (n = 54) listened to 174 musical pieces selected from the first experiment. Participants had to rate the familiarity, arousal, liking, and chillpotential, and to indicate whether they experienced chills themselves. Participants were able to rate the chill-potential of musical pieces with high coherence (Cronbach’s α = .85). Furthermore, there is high accordance in the rating of arousal and liking (α = .95; α = .94), while the real chill experience is participant-specific (α = .70). Regression analysis showed that subjectively perceived arousal can mainly be explained by loudness (R² = .51). However, variance in liking and chill-potential cannot be explained so easily. Participants were also asked about their criteria for rating the chill-potential of excerpts. Harmony, melody, climax and course of loudness were the four most important parameters. Also a gender-related effect of music in the selected 174 excerpts was observed: females experienced more chills than males (females: 32.1 chills per subject, males: 15.2 chills per participant), while arousal, liking and chill-potential were rated systematically higher by women than by men.