Effects of Empathetic Responses, Agent Identity, and Gender Match on Users’ Self-Disclosure of Depression and Perceptions in Human–Chatbot Interactions

This experimental study examined how a chatbot’s response style (empathetic versus informative), agent identity (generic versus anthropomorphized), and gender match (same versus different user/chatbot gender) influence users’ perceptions of interactions with chatbot and self-disclosure of depressive experiences. We utilized a 2 × 2 × 2 between-subjects experimental design with 417 American adult participants interacting with a text-based chatbot. Results showed that participants exposed to empathetic responses reported greater willingness to disclose depressive experiences. Neither the anthropomorphic agent’s identity nor gender match had a significant effect on the outcomes. Additionally, although trust positively predicts self-disclosure and satisfaction, it did not mediate the effects of empathetic responses on users’ satisfaction. These findings underscore the significance of emotional responsiveness compared to an informational approach in human–machine health communication. Implications for affective computing and digital mental health assistance tools are discussed.