Defensive Motivational Nodes (DMN): Inferring Psychological States from Language via Defense Mechanisms and Affective motivation

Resumen

Language encodes latent psychological signals. We introduce Defensive Motivational Nodes (DMN), a two-axis representation that pairs one defense mechanism with one affective motivation per utterance. Grounded in Vaillant’s hierarchy and a synthesis of Russell’s circumplex with Plutchik’s primary emotions, DMN captures how and why people express emotions. We created a 10 × 8 taxonomy, an annotation manual, and a 300-utterance quadri-lingual synthetic corpus , achieving substantial agreement (κ ≈ 0.78; α ≈ 0.80). Incorporating DMN tags into GPT-4 prompts yielded more nuanced, empathetic replies than baseline outputs. We outline five ethical pillars—autonomy, non-maleficence, non-stigmatization, cultural humility, beneficence—and highlight a Franco-Georgian-Korean collaboration that mitigates WEIRD bias. DMN opens new avenues for psychologically aware NLP in clinical, educational, and cross-cultural contexts. The associated open dataset is available at Zenodo (DOI: 10.5281/zenodo.16269463 )

Áreas y palabras clave

Cómo citar (APA 7)

Kim, R. S. (2025). Defensive Motivational Nodes (DMN): Inferring Psychological States from Language via Defense Mechanisms and Affective motivation. https://doi.org/10.31234/osf.io/v3wxh_v1