Mixed affective responses to music with conflicting cues

Abstract

We examined whether listening to music induces simultaneously happy and sad affective responding. The stimuli were instrumental excerpts from musical recordings that included a broad range of genres. The excerpts varied in tempo (fast or slow) and mode (major or minor), such that they had consistent happy (fast/major), consistent sad (slow/minor), or inconsistent (fast/minor or slow/major) affective cues. Listeners rated how each excerpt made them feel using separate scales for happiness and sadness. When tempo and mode cues conflicted, “mixed” happy and sad feelings were evoked. Mixed ratings on control measures (pleasantness and unpleasantness) did not show the same pattern.

Acknowledgements

Supported by the Social Sciences and Humanities Research Council of Canada.

We thank Jane Campbell for her assistance in recruiting and testing the participants.

Notes

¹Russell and Carroll (1999) proposed that participants may misinterpret unipolar scales as bipolar, such that the low ends of the scales are considered to be the semantic opposites of the high ends. For these participants, the happy and sad scales would be identical but reversed. In the present experiment, this response style would lead to neutral feelings (i.e., neither happy or sad) receiving happy and sad ratings of 3 and 3 (rather than 0 and 0), and for any given excerpt, happy and sad ratings would typically sum to 6 (happy = 6 + sad = 0; happy = 5 + sad = 1; happy = 4 + sad = 2, and so on). Accordingly, participants who felt neutral (rather than happy and sad) in response to music with mixed cues could have relatively high MIN scores (=3). In principle, the observed interaction between tempo and mode could be an artefact of a subset of participants who adopted this unexpected response style.

To test this possibility, we summed each listener's happy and sad ratings separately for each excerpt, and calculated the total number of sums that equalled 6. A median split was used to divide participants into two groups: those with a relatively low number of sums equal to 6 (i.e., participants who treated the scales correctly as unipolar) and those with a relatively high number (i.e., participants who potentially treated the scales as bipolar). Three-way ANOVAs on happy and sad MIN ratings included response style as a between-subjects variable and tempo and mode as within-subjects variables. The three-way (mode×tempo×response style) interaction was not significant in Experiment 1 or in Experiment 2, Fs < 1. In other words, elevated mixed ratings for music with inconsistent cues to happiness and sadness was similar for both groups of participants. Moreover, the two-way interaction between tempo and mode remained robust when participants who had a clearly unipolar response style were analysed separately, Fs(1, 19) = 12.43 and 17.25, in Experiments 1 and 2, respectively, ps<.005.

In additional analyses, we divided participants into two groups based on the number of times their happy and sad ratings summed to 5, 6, or 7. Virtually identical patterns were evident. In short, observed patterns of mixed happy and sad responding cannot be attributed to participants who may have misinterpreted our unipolar scales as bipolar.