The association between liking, learning and creativity in music.
Level 4 - case-series / case-control
Single-group experimental laboratory study combining behavioral tasks and EEG
PubMed 39152203 · doi:10.1038/s41598-024-70027-z
What was done
Forty non-musicians were trained de novo on an unfamiliar artificial musical grammar. Following training, participants evaluated the pleasantness of final notes in melodies that varied in probability according to a computational model of auditory expectation, while electroencephalography (EEG) was recorded. Participants also composed original musical pieces using the learned grammar, which were subsequently evaluated for creativity by expert raters. The authors analyzed the relationships among note probability, pleasantness ratings, event-related potentials (N100 and P200), and composition creativity.
What was found
An inverted U-shaped relationship occurred between liking and note probability, with intermediate-probability notes rated as most pleasant. These intermediate-probability notes also elicited larger N100 amplitudes at posterior sites and larger P200 amplitudes at frontal sites. Participants rated as producing less creative compositions preferred higher-probability notes, whereas more creative individuals preferred intermediate-probability notes. Evoked brain responses to note probability were reported to be relatively independent of learning and creativity scores. No numerical values, test statistics, or effect sizes were reported in the abstract.
Why it matters
This study links the perceptual appreciation of novelty (auditory prediction error) directly to individual differences in creative production within the same musical domain.
Limits
The study is limited by a small sample size (n = 40) restricted to non-musicians, reducing generalizability to trained musicians or real-world composition. The use of a simplified artificial grammar in a laboratory setting may not reflect naturalistic music perception. The abstract reports directional findings without quantitative effect sizes or confidence intervals.