Introduction: The main purpose of this research is to explore and analyze changes in various frequency bands of brain signals among two distinct groups: memorizers and non-memorizers of the Quran. This investigation focuses on the execution of visual memory tests using Cantab software, with an emphasis on selecting optimal feature channels and employing different classifiers. Materials and Methods: First, brain signals were recorded from 15 Quran memorizers and 15 non-Quran memorizers during the performance of delayed matching to sample (DMS), Paired Associates Learning (PAL), and Spatial Recognition Memory image memory tests using Cantab software. Following appropriate pre-processing, non-linear features such as Lyapunov profile, correlation dimension, entropy, and detrended fluctuation analysis parameters were extracted. The selection of relevant channels was performed using T-TEST, Sequential Forward Selection, and Genetic Algorithm (GA) methods. Classification involved the use of multi-layer perceptron (MLP), Support Vector Machine, and naïve Bayes algorithms. Results: The selected optimal channels were primarily associated with frontal, parietal, and occipital brain regions involved in the attention network and visual memory of Quran memorizers. In most instances, the average power of low-frequency components in brain signals was found to be higher in memorizers than in non-memorizers. The MLP neural network, utilizing optimal channels selected by the GA method, demonstrated the highest accuracy between memorizers and non-memorizers at 94.79%. Conclusion: Analysis of EEG data revealed that the power ratio of low-frequency components, the power ratio of low-to-high-frequency components, and the power ratio of theta to beta bands indicated an increase in relaxation and patience among the memorizer group during the retrieval phase of visual memory. This enhanced concentration and attention, leading to a higher percentage of correct answers and increased reaction time in the memorizer group during the implementation of visual memory tests using Cantab software. The MLP neural network, employing features selected by the GA method, particularly sample and approximate entropy in D, A5 sub-bands, and in the occipital, parietal, and central brain regions, achieved a superior accuracy percentage in the implementation of the DMS test.