This commentary presents results from a pilot study of emotional response to music. These results lend support to some aspects of the hexpartite theoretical model proposed by Juslin & Västfjäll (J&V). Interpreting these preliminary findings through the lens of the model framework opens new avenues of inquiry demanding fuller exploration in future research.
The pilot study proceeded from a rejection of “the common assumption that musical emotions must be based on a cognitive appraisal,” as J&V write in their short abstract (not printed here). Because young children are less likely to form cognitive appraisals of emotionally inductive stimuli (Harris et al. Reference Harris, Olthof and Terwogt1981; Stein & Levine Reference Stein, Levine, Dalgleish and Power1999), they were selected as participants for the pilot. Twenty excerpts of “classical music” – Western art music composed between the twelfth century and the present – ranging in length from 27 sec to 62 sec were presented in random order to 42 children, ages 3 to 5 years. Variability of section length was allowed to accommodate complete musical phrases or sections. Pieces with English-language texts were excluded, given that children are more likely to respond emotionally to the lyrical, as opposed to musical, content of such works (Morton & Trehub Reference Morton and Trehub2007). In selecting excerpts, emotional valence was not assumed: music was not chosen because it was happy or sad, but rather because it was deemed evocative. Instead of self-report, facial expressions, vocalizations, and body movement were recorded as the dependent variables (Izard Reference Izard1994; Sloboda Reference Sloboda1991; Witvliet & Vrana Reference Witvliet and Vrana2007) for later analysis using Izard's Maximally Discriminative Facial Movement Coding System (Max) (Izard Reference Izard1995).
The same set of excerpts was presented in the same order through a series of iterative pilot phases. In phase 1, the excerpts were played through speakers for an entire preschool classroom. In phase 2, a single child listened through a pair of noise-canceling headphones in an experimental room. Phase 3 repeated this procedure, but with the child in a more familiar, classroom setting. In the fourth and final phase of the pilot, children listened to music through headphones while remaining in their classroom, and were free to put on or take off the headphones whenever they chose. Across all phases of the pilot study, results were broadly similar – a brief period of initial interest gave way to disengagement, marked by an apparent decrease in interest (no facial expressive movement), minimal vocalization, and little body movement. Approximately 4–6 minutes after the music began, children would either remove their headphones (in phases 2–4 of the pilot) or seek another activity.
This “attentional curve” may in part be explained by the first component of J&V's model: brain stem reflex. The initial spike in interest may be less a function of emotional response to the music itself than attenuation to a novel auditory stimulus – a response to “music as sound” (sect. 3.1.1, para. 1). As this novelty wears off, the child's attention falls to a baseline level. Subsequent and fleeting spikes (Fig. 1d) accompany the beginning of new excerpts, and may again be a function of novelty.
This pattern of attention also lends further support to Berlyne's theory that listener preferences are related to arousal by the Wundt curve (an inverse parabolic relationship; see Berlyne Reference Berlyne1971). Berlyne posited that if the arousal potential of a piece of music is misaligned (either too high or too low) relative to the listener's preferences, it will be rejected. The children in this study – well rested following naptime and well fed following a snack – were likely to have a preference for music with a high arousal potential. When the musical excerpts failed to deliver, the children rejected them, less through a demonstration of displeasure than of disinterest or apathy. This rejection may also be a function of genre: Classical music cannot match commercial music for gross aspects of arousal potential such as frenetic energy or volume.
A “genre effect” also lends tentative support to two other mechanisms underlying emotional response. In conversations with children's teachers and parents it was revealed that the primary (and in many cases only) time children listened to classical music was when it was time to go to sleep. The repeated pairing of classical music (conditioned stimulus [CS]) with sleep (unconditioned stimulus [UCS]) would explain a relaxation response through evaluative conditioning. Another aspect of the model, musical expectancy, may also help explain this result. As J&V note, both the pleasure of fulfilled expectation and the displeasure of frustrated expectation are predicated on the listener possessing sufficient knowledge to form an expectation – knowledge that is gained through learning. With limited exposure to classical music, children would not possess the knowledge requisite to forming an expectation. Gaining such knowledge in a relatively rapid fashion might be possible with other genres, but classical music, which does employ self-referential techniques to create coherent structural wholes, makes little use of literal repetition – the sort that would allow expectations to be quickly formed.
In part, young children were chosen for this study to control for the role of emotional contagion and episodic memory, judged (perhaps incorrectly) to be secondary or tangential aspects of emotional response. It was reasoned that young children would be less likely to perceive the emotional character in a piece music and mimic that emotion (Stein & Levine Reference Stein, Levine, Dalgleish and Power1999); their emotional responses would be genuinely their own. They also have had relatively little time to form episodic memories, musically linked or otherwise. The subdued emotional response displayed by children in this study could be taken as preliminary support for either assertion. However, it is interesting to note that when children listened to music in the company of their classmates – as in pilot phases 1 and 4 – they were far more emotionally responsive than when they listened alone. Some degree of emotional contagion may be less one's mimicry of the music than of those nearby.
Considering these preliminary results in the context of J&V's theoretical framework suggests a path for future research. Understanding – even in a hypothetical sense – the mechanisms underlying emotional response to music suggests that studies should be designed to isolate and explore the proportionate role of individual mechanisms in total response. For example, using musically trained and untrained individuals, and varying the level of structural complexity of musical excerpts, could enable a more direct assessment of the role of musical expectancy. In this way, it may eventually be possible to estimate the relative strength of each mechanism in producing emotional response, both in terms of direct and of interaction effects. With a testable model guiding these efforts, it should be possible to produce more consistent and interpretable results.
