The Effects of Computer Agents' Nonverbal Behaviors on Users' Cognitions and Perceptions

• Main Authors: Liu, Jia-Ying, 劉佳穎
• Other Authors: Hsu, Yu-Chen
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/29219219419935502074

碩士 === 國立清華大學 === 資訊系統與應用研究所 === 98 === This research explores the external stimuli to human cognition and perception and also discusses the human information reception during the interaction. The experimental system involves embodied computer agents in an interaction scenario. The agents communicate with users via vocal messages and different nonverbal behavior in distinct groups. The independent variable is “nonverbal behavior of computer agents” in four groups: non-semantics, mismatch, hyperactivity, and control. Non-semantics refers to the scenario where the behavior is irrelevant to the communication; mismatch refers to the scenario where the behavior is opposing to the communication. The computer agent in the hyperactivity group exhibits an excessive amount of behaviors. The computer agent in the control group performs no behavior. The dependent variables include “cognitive load, recall and comprehension efficiency, perception, satisfaction and further interaction will”. The experimental results verify that the effect of nonverbal behaviors on human-computer interaction is similar with interpersonal communication. The mismatch group and the hyperactivity group present a high cognitive load than the non-semantics group and the control group. The semantic-mismatched nonverbal behavior of a computer agent distracts uses’ attention and decreases user perception, satisfaction and the will of further interaction. In particular, according to the qualitative data, nonverbal behavior in high frequency not only leads to cognitive load and decrease the information communication efficiency but also creates positive and lively impression of the computer agent to users. This further improves user satisfaction.