Matching Memories: The Role of Emotion in Music Recommendations for Autobiographical Memories

Dementia remains a major challenge in our aging society, with Alzheimer’s disease being the most common cause. People with dementia often recall fewer autobiographical memories, which can diminish their sense of self and reduce quality of life (QOL). In the absence of a cure for dementia, interventions that support QOL are therefore essential. This can be addressed by using nonpharmacological treatments, such as music therapy, which can play a beneficial role in elderly care. This study focused on a music recommender system that suggests songs to individuals based on their descriptions of positive autobiographical memories. The objective was to identify whether musical emotion congruency was associated with generating music recommendations fitting with individual memories, and whether this was affected by age. A total of 71 Dutch and French participants in three age groups (18-25, 40-60, 65+) described 87 positive memories and received song recommendations based on the person and...Show moreDementia remains a major challenge in our aging society, with Alzheimer’s disease being the most common cause. People with dementia often recall fewer autobiographical memories, which can diminish their sense of self and reduce quality of life (QOL). In the absence of a cure for dementia, interventions that support QOL are therefore essential. This can be addressed by using nonpharmacological treatments, such as music therapy, which can play a beneficial role in elderly care. This study focused on a music recommender system that suggests songs to individuals based on their descriptions of positive autobiographical memories. The objective was to identify whether musical emotion congruency was associated with generating music recommendations fitting with individual memories, and whether this was affected by age. A total of 71 Dutch and French participants in three age groups (18-25, 40-60, 65+) described 87 positive memories and received song recommendations based on the person and location associated with each memory, either with or without emotion matching. The recommender system used emotional ratings from the Spotify API to match the emotion of a recommended song to that of a described memory. Results comparing participants’ own song ratings showed that the Spotify audio features were only partially reliable (valence p = .076; arousal p = .024). However, emotional alignment between a song and a memory emerged as an important factor in recommendation quality (valence p < .001; arousal p = .018). Moreover, differences in age group did not influence the maximum experienced rating of fit (p =.370). These findings reveal both strengths and weaknesses of the proposed music recommender system and highlight concrete aspects that should be refined in future research. Ultimately, integrating such systems into clinical practice may increase positive emotions linked to autobiographical memories and thereby contribute to an improved QOL for people with Alzheimer’s disease.Show less

Leiden University Student Repository

Documents

In Collections

Matching Memories: The Role of Emotion in Music Recommendations for Autobiographical Memories