Emotion recognition in music changes across the adult life span

Abstract

In comparison with other modalities, the recognition of emotion in music has received little attention. An unexplored question is whether and how emotion recognition in music changes as a function of ageing. In the present study, healthy adults aged between 17 and 84 years (N=114) judged the magnitude to which a set of musical excerpts (Vieillard et al., 2008) expressed happiness, peacefulness, sadness and fear/threat. The results revealed emotion-specific age-related changes: advancing age was associated with a gradual decrease in responsiveness to sad and scary music from middle age onwards, whereas the recognition of happiness and peacefulness, both positive emotional qualities, remained stable from young adulthood to older age. Additionally, the number of years of music training was associated with more accurate categorisation of the musical emotions examined here. We argue that these findings are consistent with two accounts on how ageing might influence the recognition of emotions: motivational changes towards positivity and, to a lesser extent, selective neuropsychological decline.

Keywords:

• Ageing
• Emotion recognition
• Musical emotions
• Positivity effect

Acknowledgements

This work was supported by grants from the Portuguese Foundation for Science and Technology and the Bial Foundation.

We thank Cristina Queirós, Manuela Cameirão, Maria Luís Almeida, Teresa Limpo and Susana Silva for their help in recruiting participants. We also thank two anonymous reviewers for their valuable comments on earlier versions of the manuscript. The musical stimuli were downloaded from the Isabelle Peretz Research Laboratory website (copyright © Bernard Bouchard, 1998), which we gratefully acknowledge.

Notes

¹The distinction between structural/compositional vs. expressive cues in music is not absolute. Tempo, loudness and articulation, for instance, can be used expressively by the interpreter as well as by the composer. Maybe the most straightforward example is tempo, which can be modulated by the performer in his/her own interpretation, while it can also be predetermined by the composer in musical notation. In the stimuli by Vieillard et al. (2008), tempo was set compositionally, and the other emotional cues were indeed structural, such as mode, dissonance, and pitch range. More importantly, there was only one particular musical excerpt for a given emotion category; no expressive variations of the same musical excerpts were used as stimuli.

²A methodological issue in forced-choice paradigms is that between-group differences in accuracy might not reflect true differences in recognition but rather response bias, i.e., differences in the likelihood of using some labels more than others (see Isaacowitz et al., 2007, for a review). Although our data are not based on a forced-choice paradigm, as a precautionary control we reanalysed the data using the unbiased hit rate Hu (Wagner, 1993). Hu is an estimate of the joint probability that a stimulus category is correctly recognised when it has been presented, and that a response is correct given that it has been used. Hu was calculated for each emotion and participant using the formula Hu=A ²/(B×C), where A corresponds to the number of correctly identified stimuli, B to the number of stimuli presented (14 per category), and C to the total number of responses in that category (i.e., including misclassifications). An analysis of covariance (ANCOVA) computed on the arcsine transformed Hu rates replicated the pattern of results obtained in the analysis of the uncorrected accuracy rates: the interaction between age and emotion, F(6, 333)=10.45, p<.0001, =.16, was significant; there was no age-related decline for happy and peaceful excerpts (ps>.05), only for sad and scary ones (ps<.01).