基于EEG的多因素认知任务脑力负荷研究

EEG-based Study on Mental Workload in Multi-factor Cognitive Tasks

目的研究不同类型、不同难度的认知任务组合情况下,脑力负荷变化情况的精细表征。方法设计一种基于逻辑运算、工作记忆和运动执行的脑力负荷诱发范式,利用该范式开展24名男性受试者参与的实验,采集受试者主观量表评分、任务绩效和脑电图(EEG)信号,并计算EEG信号多个频带的功率特征。结果主观量表和任务绩效分析表明,计算难度、N-back等级均能诱发出不同等级的脑力负荷;EEG信号分析表明,脑力负荷的增加伴随着前额叶theta波增强和alpha波的减弱;利用支持向量机(SVM)构建脑力负荷分类模型,能实现平均75%单因素三分类正确率和81.7%的脑力负荷三分类正确率;利用逐步回归模型可实现对脑力负荷的预测。结论EEG信号的频域特征能够反映多因素认知任务的脑力负荷变化情况,可以对认知因素水平和脑力负荷进行分类和连续预测。

Objective To study the accurate characterization of mental workload changes in different types of cognitive tasks with different difficulties.Methods A mental workload induction paradigm was designed based on logic operation,working memory and exercise execution.Twenty-four male subjects were recruited to participate in the experiment adopting this paradigm.The subjective rating,task performance and electroencephalogram(EEG)signals were collected and the band power characteristics of multiple bands of EEG signals were calculated.Results The subjective rating and task performance analysis showed that both the computational difficulty and the N-back level could induce different levels of mental load.The EEG signal analysis showed that the increase of mental load was accompanied by the enhancement of the theta wave and the attenuation of the alpha wave.The support vector machine(SVM)was used to construct the mental load classification model,which achieved an average accuracy of more than 75%of the single factor in three-way classifications and 81.7%of three-way mental workload classifications.The prediction of mental workload could also be achieved using a stepwise regression model.Conclusion The frequency domain features of EEG signals can reflect the changes of mental load of multi-factor cognitive tasks when factors change.Thus the classification and continuous prediction of cognitive factors and mental workload can be realized.