Psychological Needs as Credible Song Signals

This thesis proposes a new framework, Psychological Needs as Credible Song Signals, to explore how contemporary songs may convey fundamental psychological needs, echoing song-like vocalization patterns in primates and other social animals. Grounded in the Temporal Need-Threat model of ostracism, an evolutionarily stable strategy for social influence, the framework suggests that music preferences as ostracism coping may align with lyrical expressions of four core psychological needs: self-esteem, self-control, belonging, and recognition.

English song lyrics are curated from manually selected tracks, entries in a published database, and Spotify playlists based on ostracism-related keywords and then annotated by ChatGPT-4o with human validation. The Chi-square goodness-of-fitness test indicates a significant quantitative difference between lyrics selected using ostracism-related keywords and random selections, suggesting a prevalence of psychological need signals consistent with findings from ostracism research. Preliminary and extended experiments using decoder-only and encoder-only transformers demonstrate that song lyrics can be classified as credible signals of psychological needs, though achieving high classification accuracy remains a challenge.

These findings highlight the framework’s potential for song content analysis to support young people coping with social exclusion. Rather than solely recognizing listeners’ emotions, identifying psychological needs can offer a privacy-friendly alternative to emotion-tracking technologies. Furthermore, it can steer music recommendations to focus on listeners’ deeper, enduring needs rather than transient emotions, offering a more accurate measure of listeners’ intentions.