Listeners discern affective variation in computer-generated musical sounds.

We carried out two experiments to test the relationship between real-time perception of structural change in stylistically unusual musical sounds, and perception of its affect (arousal and valence). Computer music was used because of its unfamiliarity and our capacity to control it in ecologically appropriate ways. In experiment 1, thirteen participants unselected for musical training participated in tasks to detect segmentation and changes in affect. Changes in affect occurred upon detection of segmentation; but not all algorithmically distinct segments conveyed distinct affect. Short segments followed by long segments led to greater changes in arousal and valence at the point of segmentation than vice versa. In experiment 2, intra-segment sound transitions were introduced. Sixteen musicians performed the same affect task as in experiment 1, and a novel change in sound task. Participants were slow to respond to a continuous transition, but quick to respond to instantaneous transitions. Contrary to literature on the perception of affect in more familiar music, the musician participants in experiment 2 differed more in their ratings of arousal than of valence, in spite of a strong correlation of arousal with the composition of the stimuli. These findings are discussed in relation to the positive valence attributed to the more familiar sounds in both experiments.