Large Language Models are Capable of Offering Cognitive Reappraisal, if Guided

• DOI: 10.48550/arxiv.2404.01288
• Publication Date: 2024-04-01
• Authors: Hongli Zhan; Allen Zheng; Yoon Kyung Lee; Jina Suh; Junyi Jessy Li; Desmond C. Ong
• Keywords: FOS: Computer and information sciences; Computer Science - Computation and Language; Computation and Language (cs.CL)

Large language models (LLMs) have offered new opportunities for emotional support, and recent work has shown that they can produce empathic responses to people in distress. However, long-term mental well-being requires self-regulation, where a one-time response falls short. This takes first step by engaging with cognitive reappraisals, strategy from psychology practitioners uses targetedly change negative appraisals an individual makes of the situation; such is known sit at root human experience. We hypothesize psychologically grounded principles could enable advanced capabilities LLMs, design RESORT which consists series reappraisal constitutions across multiple dimensions be used as LLM instructions. conduct first-of-its-kind expert evaluation (by clinical psychologists M.S. or Ph.D. degrees) LLM's zero-shot ability generate medium-length social media messages asking support. fine-grained showed even LLMs 7B scale guided are capable generating help users reappraise their situations.