Inferring users’ perceptions of Virtual Environments (VEs) is essential for Virtual Reality (VR) research. Traditionally, this is achieved through assessing users’ affective states before and after being exposed to a VE, based on standardized, self-assessment questionnaires. The main disadvantage of questionnaires is their sequential administration, i.e., a user’s affective state is measured asynchronously to its generation within the VE. A synchronous measurement of users’ affective states would be highly favorable, e.g., in the context of adaptive systems. Drawing from nonverbal behavior research, we argue that behavioral measures could be a powerful approach to assess users’ affective states in VR. In this paper, we contribute by providing methods and measures evaluated in a user study involving 42 participants to assess a users’ affective states by measuring head movements during VR exposure. We show that head yaw significantly correlates with presence, mental and physical demand, perceived performance, and system usability. We also exploit the identified relationships for two practical tasks that are based on head yaw: (1) predicting a user’s affective state, and (2) detecting manipulated questionnaire answers, i.e., answers that are possibly non-truthful. We found that affective states can be predicted significantly better than a naive estimate for mental demand, physical demand, perceived performance, and usability. Further, manipulated or non-truthful answers can also be estimated significantly better than by a naive approach. These findings mark an initial step in the development of novel methods to assess user perception of VEs.

推断用户对虚拟环境（VE）的看法对于虚拟现实（VR）研究至关重要。传统上，这是通过基于标准化的自我评估调查表评估使用者接触VE之前和之后的情感状态来实现的。问卷的主要缺点是它们的顺序管理，即，用户的情感状态与其在VE中生成的情感状态是异步测量的。用户的情感状态的同步测量将是非常有利的，例如，在自适应系统的情况下。从非语言行为研究中得出的结论，我们认为行为量度可能是评估VR中用户情感状态的有力方法。在本文中，我们通过提供一项由42位参与者组成的用户研究进行评估的方法和措施来做出贡献，以通过测量VR暴露期间的头部运动来评估用户的情感状态。我们表明，偏航与存在，心理和身体需求，感知的性能以及系统可用性显着相关。我们还将发现的关系用于基于偏航的两个实际任务：（1）预测用户的情感状态，以及（2）检测可操纵的问卷答案，即可能不真实的答案。我们发现，对于心理需求，身体需求，感知的表现和可用性，可以比对天真的估计更好地预测情感状态。此外，与天真的方法相比，操纵或非真实答案的估计也要好得多。