Kim Junhui, Yoshida Takako
Department of Mechanical Engineering, School of Engineering, Tokyo Institute of Technology, Tokyo, Japan.
Front Psychol. 2024 May 17;15:1364076. doi: 10.3389/fpsyg.2024.1364076. eCollection 2024.
Inconsistent jittery temporal delays between action and subsequent feedback, prevalent in network-based human-computer interaction (HCI), have been insufficiently explored, particularly regarding their impact on the sense of agency (SoA). This study investigates the SoA in the context of eye-gaze HCI under jittery delay conditions.
Participants performed a visual search for Chinese characters using a biresolutional gaze-contingent display, which displayed a high-resolution image in the central vision and a low-resolution in the periphery. We manipulated the delay between eye movements and display updates using a truncated normal distribution (μ to μ + 2 σ) with μ ranging from 0 to 400 ms and σ fixed at 50 ms. Playback of recorded gaze data provided a non-controllable condition.
The study revealed that both reported authorship and controllability scores, as well as the fixation count per second, decreased as μ increased, aligning with trends observed under constant delay conditions. The subjective authorship weakened significantly at a μ of 94 ms. Notably, the comparison between jittery and constant delays indicated the minimum value (μ) of the distribution as a critical parameter influencing both authorship perception and visual search time efficiency.
This finding underscores the importance of the shortest delay in modulating SoA. Further examining the relative distribution for fixation duration and saccade amplitude suggests an adaptation in action planning and attention distribution in response to delay. By providing a systematic examination of the statistical attributes of jittery delays that most significantly affect SoA, this research offers valuable implications for the design of efficient, delay-tolerant eye-gaze HCI, expanding our understanding of SoA in technologically mediated interactions. Moreover, our findings highlight the significance of considering both constant and variable delay impacts in HCI usability design, marking a novel contribution to the field.
在基于网络的人机交互（HCI）中，动作与后续反馈之间存在不一致的抖动时间延迟，这一现象尚未得到充分研究，尤其是其对代理感（SoA）的影响。本研究在抖动延迟条件下，对眼动HCI背景下的代理感进行了调查。
参与者使用双分辨率注视相关显示器对汉字进行视觉搜索，该显示器在中央视觉区域显示高分辨率图像，在周边区域显示低分辨率图像。我们使用截断正态分布（μ至μ + 2σ）来操纵眼动与显示更新之间的延迟，其中μ范围为0至400毫秒，σ固定为50毫秒。回放记录的注视数据提供了不可控条件。
研究表明，随着μ的增加，报告的作者身份和可控性得分以及每秒注视次数均下降，这与在恒定延迟条件下观察到的趋势一致。在μ为94毫秒时，主观作者身份显著减弱。值得注意的是，抖动延迟与恒定延迟之间的比较表明，分布的最小值（μ）是影响作者身份感知和视觉搜索时间效率的关键参数。
这一发现强调了最短延迟在调节代理感方面的重要性。进一步研究注视持续时间和扫视幅度的相对分布表明，在应对延迟时，动作规划和注意力分配会发生适应性变化。通过系统地研究对代理感影响最显著的抖动延迟的统计属性，本研究为高效、耐延迟的眼动HCI设计提供了有价值的启示，扩展了我们对技术介导交互中代理感的理解。此外，我们的研究结果突出了在HCI可用性设计中考虑恒定和可变延迟影响的重要性，这是该领域的一项新贡献。
