Design of Persona-Based Interactive Interfaces and Their Impact on Human Self-Perception

Grounded in the frameworks of affective computing and large-language-model technology, this paper systematically sorts out the conceptual evolution and core techniques of persona-based interactive interfaces (PBIIs) and, by means of a randomized controlled experiment, verifies their comprehensive effect on users’ self-perception. The study first constructs a four-level closed-loop architecture of “Perception – Understanding – Generation – Feedback.” Sixty university students are then recruited for a 14-day intervention; pre- and post-tests using a Self-Efficacy Scale and an Emotion-Awareness Test are compared. Results show a significant rise in self-efficacy for the experimental group (ΔM = +0.70, p < 0.01) and a 12 % increase in emotion-recognition accuracy, validating the synergistic mechanism of affective mirroring and verbal persuasion. Qualitative interviews further reveal potential risks such as emotional dependence and cognitive dissonance. In view of these findings, this paper proposes PBII design principles centred on uncertainty management and user-agency cues, offering theoretical and practical reference for responsible application in education and mental-health fields.