Qualities of music-evoked autobiographical memories are associated with auditory features of the memory-evoking music.

Studies of music-evoked autobiographical memories (MEAMs) show that music is a potent cue for retrieving vivid and self-relevant memories. However, whether and how musical features are able to predict the qualities of MEAMs - including their emotional qualities, phenomenological characteristics and retrieval efficiency - remains unclear. In our study, a sample of 233 adult participants identified a piece of music that evoked an autobiographical memory (AM) before providing a written description of the memory, and then evaluating its emotional and phenomenological content. Participants were then presented with excerpts of ten songs that were popular during their childhood and early adulthood and reported the same details for any AMs evoked. Features of all songs were extracted using the Spotify Web API and subjected to principal components analysis for dimension reduction. This revealed a primary auditory feature component - characterised by low energeticness and high acousticness - that was found to predict several qualities of the memory. Specifically, results showed that low energetic - high acoustic songs were associated with AMs characterised emotionally by aesthetic appreciation, adoration, calmness, romance and sadness, while high energetic - low acoustic songs were associated with AMs high in memory energeticness, amusement and excitement. Phenomenologically, AMs associated with low energetic - high acoustic songs were described as less social, and more vivid, unique and important, and, in terms of retrieval efficacy, tended to be retrieved more slowly. Our findings show for the first time the extent to which the qualities of MEAMs can be predicted by music's stimulus features. Further, by taking into account how the AMs were evoked, and subjective factors related to the memory-evoking music such as liking and familiarity, our study provides insights into possible mechanisms underlying music-assisted memory encoding and retrieval. We discuss the implications of our findings for understanding the links between perception, emotion and memory processes, and make suggestions for future work that can advance this research area.