TY - GEN
T1 - Chatting with Sheldon in VR
T2 - 2026 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2026
AU - Zhang, Guangtao
AU - Peng, Shen
AU - Dai, Yangbin
AU - Guo, Honglei
AU - Gao, Yang
AU - Li, Lin
AU - Yu, Tao
N1 - Publisher Copyright:
© 2026 IEEE.
PY - 2026
Y1 - 2026
N2 - Recent advances in generative AI and virtual reality (VR) enable highly realistic celebrity embodied conversational agents (ECAs). However, it remains underexplored how visual stylization and voice type shape users' trust and social presence in celebrity ECA interactions. To address this gap, we examine the effects of visual style (Hand-Crafted, AI-Reconstructed, Non-Stylized) and voice type (Cloned, Synthesized) by conducting a 3 × 2 within-subjects study (N=36). Results show that visual style significantly influenced trust and social presence, whereas voice showed no main effect but interacted with style on trust. A post-hoc subgroup analysis based on self-reported familiarity suggests that higher familiarity was associated with higher overall trust and social presence, while lower familiarity corresponded to greater sensitivity to stylistic variation. These findings provide empirical support for extending parasocial interaction frameworks to immersive VR contexts.
AB - Recent advances in generative AI and virtual reality (VR) enable highly realistic celebrity embodied conversational agents (ECAs). However, it remains underexplored how visual stylization and voice type shape users' trust and social presence in celebrity ECA interactions. To address this gap, we examine the effects of visual style (Hand-Crafted, AI-Reconstructed, Non-Stylized) and voice type (Cloned, Synthesized) by conducting a 3 × 2 within-subjects study (N=36). Results show that visual style significantly influenced trust and social presence, whereas voice showed no main effect but interacted with style on trust. A post-hoc subgroup analysis based on self-reported familiarity suggests that higher familiarity was associated with higher overall trust and social presence, while lower familiarity corresponded to greater sensitivity to stylistic variation. These findings provide empirical support for extending parasocial interaction frameworks to immersive VR contexts.
KW - Embodied Conversational Agent
KW - Parasocial Interaction
KW - Virtual Human
KW - Virtual Reality
KW - Visual Stylization
KW - Voice
UR - https://www.scopus.com/pages/publications/105038738923
U2 - 10.1109/VRW70859.2026.00336
DO - 10.1109/VRW70859.2026.00336
M3 - 会议稿件
AN - SCOPUS:105038738923
T3 - Proceedings - 2026 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2026
SP - 1323
EP - 1324
BT - Proceedings - 2026 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops, VRW 2026
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 21 March 2026 through 25 March 2026
ER -