Navigating emotions: contingent affective governance in an AI-powered mental health APP

Abstract

This study examines how affective governance operates on AI-mediated mental health platforms through a case study of Xin Dao Diary, a Chinese app combining emotional self-tracking, peer support, and AI “animal therapists.” Affective governance involves the strategic modulation of users’ emotions through design, algorithms, and rules. While existing literature often treats this as a unidirectional and structurally determined process, we argue it is highly dynamic. We propose the concept of contingent affective governance to capture how these strategies emerge, adapt, and occasionally backfire through continuous feedback loops between platform mechanisms and user emotions. Methodologically, we combine a walkthrough analysis to trace intended governance logics with a diary study and an examination of user-generated content to reveal everyday negotiations, resistance, and public contestation. Our analysis identifies three core platform mechanisms: normalized expressive design, algorithmic empathy matching, and gamified limits and incentives. On the user side, three key dynamics emerge: co-constructive care, autonomy in resistance, and public contestation. These findings demonstrate how the app extracts emotional data while facilitating perceived therapeutic benefits, producing differentiated, and provisional forms of biopolitical control. The article reframes affective governance as an evolving process of mutual shaping between platforms and users.