虚拟现实中多通道信息呈现方式的认知负荷定量研究

目的 针对虚拟现实体验系统中的不同信息呈现方式，对使用者交互效率影响难以确定及量化的问题，进行定量研究。方法 通过搭建虚拟现实场景，以信息呈现通道的类别与数量作为变量，展开跟踪–检测响应双任务实验;通过记录任务行为数据中的跟踪误差和响应时间，以及生理数据中的瞳孔直径大小，分析并讨论在不同通道刺激影响因素下的实验中，任务绩效及眼动生理反应变化规律;同时结合主观负荷评价数据，建立了基于BP神经网络的多通道认知负荷模型，以认知负荷作为交互效率的综合评价指标，量化任务执行效率。结果 信息呈现的通道类别及其数量对任务效率均有显著的影响。结论 信息呈现通道数量与任务绩效及生理反应呈一定程度的正相关;多个任务使用相同通道呈现信息会损害所有任务的绩效，增加认知负荷。模型输出的负荷值与主观认知负荷评估值吻合较好，相对误差为8.2%，验证了其有效性。

Quantitative Study on Cognitive Load of Multi-channel Information Presentation in Virtual Reality

Aiming at the problem that different information presentation methods in the virtual reality experience system are difficult to determine and quantify the impact on user interaction efficiency, the work aims to conduct quantitative research. By building a virtual reality scene, the category and number of information presentation channels were used as variables to carry out the track-detection response dual-task experiment. By recording the tracking error and response time in the task behavior data, as well as the pupil diameter in the physiological data, the changes in task performance and eye movement physiological response in experiments under different channel stimulation factors were analyzed and discussed. At the same time, combined with subjective load evaluation data, a model of multi-channel cognitive load based on BP neural network was established. The cognitive load was used as a comprehensive evaluation index of interaction efficiency to quantify the task execution efficiency. The results showed that the type and number of channels for information presentation had a significant impact on task efficiency. The number of channels for information presentation is positively correlated with task performance and physiological response to a certain extent. Using the same channel for multiple tasks to present information will harm all the performance of all tasks and increase the cognitive load. At the same time, the cognitive load value of the model is in good agreement with the evaluation value of subjective cognitive load, with an absolute error of 8.2%, which verifies its effectiveness.