标题： 心理负荷估计中的差异：数据可视化评估中自我报告与基于EEG的测量方法 显示英文标题

标题： Discrepancies in Mental Workload Estimation: Self-Reported versus EEG-Based Measures in Data Visualization Evaluation

摘要： 准确评估心理工作量（MW）对于理解可视化任务中的认知过程至关重要。 虽然基于脑电图（EEG）的测量方法正在成为传统评估技术（如自我报告测量）的有前景的替代方法，但研究不同方法之间一致性的文献有限。 在一项初步研究中，我们观察到基于EEG的和自我报告的MW测量之间可能存在差异的迹象。 受到这些初步观察的激励，我们的研究通过涉及可视化任务的实验进一步探讨了基于EEG的和自我报告的MW评估方法之间的差异。 在实验中，我们采用两项基准任务：可视化素养评估测试（VLAT）和空间可视化（SV）任务。 在可视化任务期间，使用32通道系统以128 Hz的采样率从参与者那里记录EEG信号。 对于每个参与者，使用基于图注意力网络（GAT）架构的EEG模型来估计MW，并将这些估计值与传统MW测量进行比较，以检查潜在的差异。 我们的研究结果揭示了任务难度与基于EEG的MW估计之间的显著差异，以及在不同任务难度水平下基于EEG的MW测量与自我报告的MW测量之间的差异。 此外，观察到的模式表明存在可能无法仅通过自我报告捕捉的无意识认知努力。 显示英文摘要

摘要： Accurate assessment of mental workload (MW) is crucial for understanding cognitive processes during visualization tasks. While EEG-based measures are emerging as promising alternatives to conventional assessment techniques, such as selfreport measures, studies examining consistency across these different methodologies are limited. In a preliminary study, we observed indications of potential discrepancies between EEGbased and self-reported MW measures. Motivated by these preliminary observations, our study further explores the discrepancies between EEG-based and self-reported MW assessment methods through an experiment involving visualization tasks. In the experiment, we employ two benchmark tasks: the Visualization Literacy Assessment Test (VLAT) and a Spatial Visualization (SV) task. EEG signals are recorded from participants using a 32-channel system at a sampling rate of 128 Hz during the visualization tasks. For each participant, MW is estimated using an EEG-based model built on a Graph Attention Network (GAT) architecture, and these estimates are compared with conventional MW measures to examine potential discrepancies. Our findings reveal notable discrepancies between task difficulty and EEG-based MW estimates, as well as between EEG-based and self-reported MW measures across varying task difficulty levels. Additionally, the observed patterns suggest the presence of unconscious cognitive effort that may not be captured by selfreport alone.