A Study of Detecting Human Decision by Interpreting Neural Signal via Electroencephalography (EEG) and Machine Learning in a (Yes/No) Paradigm

In this study, we analyzed the neural phenomenon of decision-making from the perspective of cognitive modes. The main purpose of this study is to investigate the detectability of human decisions from brain signals by a computerized program. We delved into the neural signals to understand the brain functions of the cognitive modes in a binary decision-making context. The decision-making process is based on two different cognitive modes, which are intuitive and analytic. As they are opposite and conflicting by definition and characteristics, we aim to find out whether the mode significantly affects the detectability of the decision. We introduced time limitation vs. adequate time, unclear knowledge vs. clear knowledge, and absence of information during decision-making vs. the availability of information to create an intuitive cognitive mode and analytic mode, respectively. We used a modified Artificial Grammar Learning (AGL) design combined with a Yes/No binary context to create a binary decision-making event. We collected the brain signal during the decision-making using the noninvasive Electroencephalography (EEG) technique. For the investigation of the detectability of the decision, we used the supervised Machine Learning technique for classification analysis. We achieved 95% average accuracy in Analytic mode and 93% accuracy in intuitive mode. With the aim of efficient design, we also identified the top 5 channels for each mode, which can provide up to 80% average accuracy for detecting decisions. This study also suggests that the best Machine Learning algorithms for this purpose are the KNN (K-Nearest Neighbor) and RF (Random Forest) classifiers.