Cognitive Reframing of Negative Thoughts through Human-Language Model Interaction

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 9977–10000 July 9-14, 2023 ©2023 Association for Computational Linguistics Cognitive Reframing of Negative Thoughts through Human-Language Model Interaction Ashish Sharma♠ Kevin Rushton♢ Inna Wanyin Lin♠ David Wadden♣ Khendra G. Lucas♢ Adam S. Miner♥♡ Theresa Nguyen♢ Tim Althoff♠ ♠Paul G. Allen School of Computer Science & Engineering, University of Washington ♢Mental Health America ♣Allen Institute for Artificial Intelligence ♥Department of Psychiatry and Behavioral Sciences, Stanford University ♡Center for Biomedical Informatics Research, Stanford University {ashshar,althoff}@cs.washington.edu Abstract A proven therapeutic technique to overcome negative thoughts is to replace them with a more hopeful “reframed thought.” Although therapy can help people practice and learn this Cognitive Reframing of Negative Thoughts, clinician shortages and mental health stigma commonly limit people’s access to therapy. In this paper, we conduct a human-centered study of how language models may assist people in reframing negative thoughts. Based on psychol- ogy literature, we define a framework of seven linguistic attributes that can be used to reframe a thought. We develop automated metrics to measure these attributes and validate them with expert judgements from mental health practi- tioners. We collect a dataset of 600 situations, thoughts and reframes from practitioners and use it to train a retrieval-enhanced in-context learning model that effectively generates re- framed thoughts and controls their linguistic attributes. To investigate what constitutes a “high-quality” reframe, we conduct an IRB- approved randomized field study on a large mental health website with over 2,000 partici- pants. Amongst other findings, we show that people prefer highly empathic or specific re- frames, as opposed to reframes that are overly positive. Our findings provide key implications for the use of LMs to assist people in overcom- ing negative thoughts. 1 Introduction Negative thoughts are a natural part of human cog- nition. However, for people experiencing men- tal health challenges, such thoughts are often entrenched, automatic and emotionally trigger- ing, making it difficult to overcome them in-the- moment (Beck, 1976). An evidence-based, well- established therapeutic intervention to overcome negative thoughts is Cognitive Reframing, in which a negative thought is replaced with a more hope- ful “reframed thought,” which offers an alternative perspective on one’s situation (Beck, 1976). For example, imagine a person with a situation “I’m submitting a research paper to ACL 2023” has a thought “This paper is going to get rejected.” A possible way to reframe this thought is to say “This paper has some chance of getting accepted due to its novel methodology and potential impact.” Psychotherapy research suggests that for a re- framed thought to be effective, it must be (a) relat- able to the individual, (b) helpful in overcoming the negative thought and (c) memorable to be acces- sible the next time a similar thought arises (Beck, 1976; Burns, 1980). However, understanding what characterizes a relatable, helpful and memorable reframe is challenging and unknown. Professional therapists can support people in coming up with such highly effective reframed thoughts. However, barriers like clinician …

Content truncated to 500 words