Analysis of Regularity in the EEG Before/After Working Memory Task

In this paper, we want to test regularity changes in working memory (WM) task with respect to the baseline activity in electroencephalogram (EEG) signals. For this purpose, we used non-linear analysis. The EEG signals were analyzed using three nonlinear measures: approximate entropy (ApEn), permutation entropy (PE) and wavelet entropy. Statistical Wilcoxon exam and Davis-Bouldian criterion is used to select optimal discriminative features. We concluded that the values of the all measures acquired from the frontal and occipital lobes increase during WM task, indicating less regularity and predictability in EEG signals. Among these features, the PE significantly increases in frontal lobes. We suggest this feature in memory-based neurofeedback system to improve memory performance.