Text or talk? The impact of chatbot identity disclosure, emotional support, and interaction modality on users’ emotional reactions and neural activity

Measuring users' emotional reactions and subsequent neural activity towards chatbots' design cues could significantly enhance our understanding of chatbots and potentially optimise the design of human-chatbot interaction. This study conducted a mixed experimental design with two within-subject factors (chatbot identity disclosure and emotional support) and one between-subject factor (interaction modality) to explore users' subjective evaluation and neural activity. Results revealed that not disclosing the chatbot's identity or providing emotional support during the interaction could enhance user' positive emotional reactions. Notably, results showed a three-way interaction effect on users' HbO₂ concentration changes in the right DLPFC. Specifically, in the text-based interaction condition where chatbots' identity was disclosed, providing emotional support from chatbots elicited a higher concentration increase of HbO₂. Similarly, when the chatbot provided emotional support, disclosing the chatbot's identity induced a higher concentration increase of HbO₂. These findings offer theoretical contributions and practical implications for the design of human-chatbot interaction.