by MADELINE HUBERTH, GE WANG
Background
It is popularly believed that providing a listener with a description of a piece prior to its performance influences the listener’s experience, as evidenced by the widespread practice of providing program notes or pre-concert lectures to audience members. While the content of the description varies – composer biography, historical context of the work, or structure and content of the work itself may be discussed – it is included with the assumption that it provides the listener with a basic orientation from which to approach the work (Meyer, 1956). Indeed, the inclusion of program notes for music less familiar to audiences, such as electroacoustic works, could aid the listening experience. As shown in the ‘Intention/Reception’ project on electroacoustic music, clues about the composer’s intentions and inspirations improve the listening experience of inexperienced listeners over multiple listens (Weale, 2006).
Such descriptions of composers’ inspirations might include text describing visual imagery, or ‘narrative imagery’. It has been shown in studies of tonal Western classical music that visual imagery imagined by the listener can influence the emotions felt while listening to music. Indeed, Juslin and Västfjäll (2008) posit the conjuring of visual imagery as one of the six main mechanisms by which music may induce emotion in a listener. Cumulatively, studies have shown that listeners seem to respond to mental images in the same way as they would to the corresponding stimuli in the ‘real’ world, for instance, reacting with negative emotions to a mental image of a car crash, or with positive emotions to a mental image of a family enjoying a holiday. As demonstrated specifically in studies of narrative imagery and music, Vuoskoski and Eerola (2015) exposed participants to a sad description (detailing a concentration camp scene) as well as a description of a nature scene before listening to a sad-sounding piece of music. In this study, listeners with the sad description experienced a more intense sadness than those without that description. Furthermore, it has been shown that imagery can be effective in enhancing emotions to music (Band, Quilter, & Miller, 2001) and vice-versa, as described by Bonny (2005). Bonny found that, during ‘guided imagery’ listening experiences where narrative imagery was read to the participants, the addition of music to the narrative imagery increased the visual details, vividness of colors, and intensified emotions experienced by the reading of the imagery.
There is little evidence though that narrative visual imagery can both enhance and temper emotions, especially when the emotions felt in response to the piece alone may be strong. In this study, we use electroacoustic music as stimuli to demonstrate this effect, following the possibility that the genre may induce emotions of high arousal and negative valence, of which the latter still tends to produce positive affect in Western classical music (Gabrielsson, 2002). Furthermore, since electroacoustic music frequently does not use Western harmonic syntax, another important mechanism of emotion induction (Huron, 2006), musical expectancy may be less relevant to emotion induction, placing more import on possible influences such as visual imagery.
Current study
We test here the effects of prior narrative visual imagery on emotions experienced while listening to electroacoustic music. We presented participants with 12 excerpts of electroacoustic music, screened to represent a variety of possible induced emotions, and presented them to participants with one of three descriptive conditions: 1) no description, 2) a description of a nature scene, or 3) a description of an emotional scene involving a person. In the third condition, the text was written to portray a ‘stronger’ experience of the emotions already typically reported by listeners of the piece (evaluated in pre-screening ratings). Emotions were evaluated by using the circumplex model of affect (Russell, 1980) in which the valence (i.e. positivity/ negativity) that a person experiences is evaluated independently of the arousal (i.e. excitement) they feel. In this study, we refer to the four ‘quadrants of arousal-valence space’ as four general types of emotional experiences, in which quadrant I = high-arousal/ high-valence, quadrant II = high-arousal/ low-valence, quadrant III = low-arousal/ low-valence, and quadrant IV = low-arousal/ high-valence. We hypothesize that narrative imagery may both increase the intensity of the felt emotion – likely to be observed through the ‘person-scene’ description – and decrease it, which we expect to observe with the inclusion of the ‘nature-scene’ description.
Method
Participants
105 individuals (50 female) participated in the experiment. The mean age was 37.6 years (SD = 11.3 years). They had a mean of 1.68 years of formal musical training (SD = 2.6 years), and 14.3% claimed to currently play or sing music actively.
Stimuli
The twelve electroacoustic excerpts in the main experiment were selected from a larger initial stimulus set of twenty-nine excerpts extracted primarily from the Computer Music Journal Sound Anthology. Excerpts were initially selected to be cohesive in the affect they might induce. Fifteen participants provided emotion ratings for each of the excerpts, using the same interface as the main experiment. Informed by these ratings, three excerpts from each quadrant of arousal-valence space were chosen for use in the main experiment, with preference given to excerpts that induced stronger emotions and showed minimal variance in emotion ratings. The twelve excerpts were presented to all listeners. A three-sentence description of a nature scene (e.g. clouds, mountains, rivers, forests), and a three-sentence description of a person which was designed to be emotionally evocative, the ‘person-scene’ condition, were written for each excerpt (see examples in Table 1).
Design
The experiment was presented to participants through the Amazon Mechanical Turk website. Individual task presentation was controlled by original JavaScript code running locally in the participant’s browser. The twelve pieces were presented to the participants through the HTML5 audio player, in random order, with each presented on its own page. The descriptive condition was randomly assigned to each excerpt for each participant.
Emotional responses to each excerpt were gauged by the following question: ‘How did you feel while listening to this piece?’ and were measured through self-assessment manikins depicting states of arousal and valence, accompanied by 9-point Likert scales (coded as -4 to 4 for each scale). Participants were also given free-response space to respond to the piece. Orientation of the answer choices and manikins was counterbalanced.
Table 1. Examples of nature scene and person scene descriptions. The first was intended for low-arousal, high-valence, and the second for high-arousal, high-valence.
Results
Since we were most interested in the intensification of emotion, ANOVAs were conducted to test the effect of condition on arousal ratings for pieces in higher or lower arousal quadrants, as well as the effect of condition on valence ratings in higher or lower valence quadrants. There was a significant effect of narrative imagery condition on arousal ratings for higher-arousal pieces, (F(2,623) = 7.90, p < .001), in which the person-scene description significantly increased arousal ratings compared to the no description condition (Mpers = 0.74, Mnone = 0.29, p < .001). Nature descriptions produced marginally-significantly lower arousal ratings than when no description was present (p = .086). However, arousal ratings did not significantly differ by descriptive condition for lower-arousal pieces. For lower valence pieces, there was a significant effect of description on valence ratings (F(2,626) = 11.14, p < .001), with post-hoc tests revealing that the nature description resulted in significantly higher valence ratings (Mnat = -0.63, Mnone = -1.15, p < .01). The person-scene description resulted in marginally-significantly lower valence scores (p = .065). In higher valence pieces, there was also a significant effect of description on valence ratings (F(2,624) = 6.60, p < .005), in which the nature condition did not serve to neutralize valence ratings, but rather resulted in significantly higher ratings (Mnat = .46, Mnone = .08, p < .05). The person-scene description resulted in higher valence ratings than presentation without description (Mpers = .68, Mnone = .08, p < .001).
Figure 1. Effects of condition on arousal and valence ratings. Mean and standard error are shown, and significant post-hoc tests are indicated.
Conclusions
Overall, our data show that narrative imagery can have both an intensifying and neutralizing effect on listeners’ emotions while listening to electroacoustic music. Compared to arousal and valence ratings reported when the piece was presented without a description, the person-scene descriptions resulted in significantly higher arousal ratings for pieces pre-rated to induce above-moderate arousal. However, they did not result in lower arousal ratings for pieces pre-rated to induce below-moderate arousal. The addition of the person-scene description with higher-valence pieces resulted in higher valence ratings, though it did not result in significantly lower valence ratings from lower-valence pieces. Nature descriptions resulted in significantly higher valence ratings in both lower and higher-valence pieces. Whilst a pattern of how descriptions may influence emotions felt while listening to electroacoustic music is not readily available from this data, the data do illustrate that descriptions can both intensify and temper emotions.
It was surprising that nature descriptions resulted in higher valence ratings for pieces in both the low valence and the high valence quadrants, as previous studies did not find a significant effect of change of emotion when using a description about nature (Vuoskoski & Eerola, 2015). Perhaps either our nature descriptions were more evocative than used in past studies, or the emotions felt while listening to our chosen stimulus set of electroacoustic music were somehow more influenced by description. We previously mentioned that visual imagery may be a particularly powerful emotion-inducing mechanism in electroacoustic music, though further controlled studies comparing genres will be needed to validate and further explore this.
We found that inclusion of the description influenced the reported induced emotions felt during the piece – the effect of descriptions from the perspective of the listener was qualitatively observed through the last question for each piece: a blank text field, with the instruction ‘please briefly comment on your listening experience’. Nearly all participants filled out this field for all excerpts. We noted comments that either alluded to the imagery description (e.g. ‘I could imagine standing in the middle of a forest’), or commented on some emotional reaction (e.g. ‘this piece made me feel happy and calm’). From these data, it seemed that participants, when presented with the nature scene, commented most frequently on imagery when compared to the other groups (18%), followed by person-scene (16%), and no description (10%). Of the three conditions, the person-scene condition elicited the most comments reporting some emotional reaction (35%). Perhaps more surprisingly, the no-description condition elicited the second most comments related to emotion (26%), while the nature scene only elicited comments related to emotion in 20% of the responses. These reactions seem to underscore our results, suggesting a connection between prior descriptions and imagery and emotional reaction in the listener. We hope to draw from these and other responses in further research.
To the authors’ knowledge, this is the first study of its nature to show that narrative imagery can have an effect on a range of affects. However, in practice, a program note would often include biographical or descriptive information about the composer’s life and the structure of the work – any narrative imagery may not be so bare to the listener’s attention. It remains to be seen how an emotional effect of narrative imagery compares to when the listener is concurrently taking in other information about the piece.
Notes
Address for correspondence: Madeline Huberth, Stanford University.
Email: mhuberth@ccrma.stanford.edu.
Musical selections, narrative descriptions, and code used to serve the experiment in Amazon Mechanical Turk can be found at: http://ccrma.stanford.edu/~mhuberth/ICMEM2015.html
References
Band, J., Quilter, S., & Miller, G. (2001). The influence of selected music and inductions on mental imagery: Implications for practitioners of Guided Imagery and Music. Journal of the Association for Music and Imagery, 8, 13-33.
Bonny, H. & Savary, L. (2005). Music & our mind: Listening with a new consciousness. Gilsum, New Hampshire: Barcelona.
Gabrielsson, A. (2002). Emotion perceived and emotion felt: same or different?. Musicae Scientiae, 5(1), 123-147.
Huron, D. (2006). Sweet anticipation music and the psychology of expectation. Cambridge, Massachusetts: MIT Press.
Juslin, P. & Västfjäll, D. (2008). Emotional response to music: the need to consider underlying mechanisms. Behavioral and Brain Sciences, 31, 559-621.
Meyer, L. (1956). Emotion and meaning in music. Chicago: University of Chicago Press.
Russell, J. (1980). A Circumplex Model Of Affect. Journal of Personality and Social Psychology, 39, 1161-1178.
Vuoskoski, J. & Eerola, T. (2015). Extramusical information contributes to emotions induced by music. Psychology of Music, 43(2), 262-274.
Weale, R. (2006). Discovering How Accessible Electroacoustic Music Can Be: The Intention/Reception project. Organised Sound, 11, 189-200.