EEG-Based Evaluation of Classifying Attention States Between Single and Dual Tasks

Everyone in their daily lives, occasionally, needs to juggle more than one task simultaneously. Although performing these tasks sequentially, one at a time, would result in optimal task performance, it is inevitable to multitask. Under such multitasking scenarios, a mixture of different attention paradigms is required to achieve the best performance outcome. Earlier studies have mainly focused on investigating their subjects’ cognitive performance under dual tasks or single tasks separately. Not much research had been conducted comparing single tasks and dual tasks based on attention detection using electroencephalography (EEG). We designed an EEG experiment consisting three common cognitive tasks with single-tasking and dual-tasking paradigms to classify the attention levels of subjects. We collected data from twenty-five adolescents after seeking ethical approval and receiving parental consent. We used six bandpower features with machine learning and statistical analysis to evaluate attention detection performance among different task pairs. From our analysis, though there were less statistically significant differences between the mean p-value (p = 0.21) of accuracy between single tasks and dual tasks, it was also found that there was only 2% accuracy improvement obtained in dual tasks compared with respective single tasks in the subject-independent cross-validation of attention classification.