机制(生物学)
功能可见性
化学
效力
神经科学
计算机科学
作用机理
动作(物理)
透视图(图形)
集合(抽象数据类型)
作者
Yimin Wang,Jiahui Wang,Yu Zou,Yiran Peng,安静,Yonglin Dai,Wenrui Liang,Haowei Guo,Yurun Chen,Ziyue Zhang,Zhangyan Yan,Jiaqi Li,Xin Lyu
标识
DOI:10.1080/10447318.2026.2660852
摘要
Photorealistic Al-driven metahumans are increasingly used not only as productivity tools but also as relational interfaces that invite self disclosure. In emotionally sensitive dialogue, however, the same cues that make disclosure feel safe may also shape reliance, legitimacy, and users' ability to pause, redirect, or exit. To explain when these affordances support responsible self-disclosure—and when they instead produce withdrawal, overload,or miscalibrated reliance—we propose the Perception-Actualization Mechanism of Affordance (PAMA), linking perceptual enablers (PEs) to their actualized effects in emotional variation (EV) and cognitive behavioral engagement (CBE). Using are search through design-informed process, we developed a disclosure oriented metahuman (Xuefei) and tested it in a mixed-methods study combining pre-/post-measures (N = 178), explainable ML with SHAP, and follow-up interviews (N =133). Results showed reliable post-interaction gains, identified higher-order disclosure pathways, and informed responsibility-oriented design criteria and auditable guard rails for deep dialogue metahuman systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI