Should AI Have Humanlike Self-disclosure in the Mental Health Intervention? The Moderated Mediation Effect of Negative Emotion

Abstract

With the increasing concerns about psychological issues and limited mental healthcare resources, chatbots have emerged as a promising therapeutic solution. This study investigates the psychological mechanisms underlying human-chatbot interactions in mental healthcare, focusing on the roles of self-disclosure strategies, emotion recognition, and users' negative emotions. Through a 2 × 3 between-group factorial design involving 197 subjects, this research explores how chatbot self-disclosure strategies and emotion recognition capabilities influence users' perceived empathy and authenticity, with users' negative emotions as a moderating factor. By examining how different chatbot therapeutic interventions influence user perceptions across negative emotion states, this research provides practical insights for designing personalized therapeutic chatbot interventions in mental healthcare.