Quantifying the Nonlinear Dynamic Complexity of Cognitive EEG by Conditional Entropy Based on Different Entropy Measures

The brain is a typical nonlinear complex system influenced by various factors. This study aims to explore the changes in conditional entropy during memory tasks to quantify cognitive Electroencephalogram(EEG) activity under different memory loads and to evaluate the effectiveness of conditional entropy in revealing the working memory process. A total of 18 graduate students from Nanjing University participated in the memory task experiment, during which EEG signals were collected synchronously and analyzed using conditional entropy. Two methods were employed for calculating conditional entropy: K-nearest neighbor estimation and kernel estimation. The kernel estimation method utilized Sigmoid, Gaussian, and Heaviside kernel functions. Particularly in the kernel estimation method employing Gaussian and Heaviside functions, better results and stability were obtained, demonstrating optimal performance. The study findings indicate that as working memory load increases, the complexity of cognitive processing in the brain also rises, especially in brain regions associated with working memory, attention, and task execution, such as the frontal pole (FP1), middle frontal gyrus (F8), and central region (Cz),