Decoding consumer purchase decisions: exploring the predictive power of EEG features in online shopping environments using machine learning

Abstract

This study explores the use of machine learning algorithms and EEG data to predict consumer purchasing decisions in an online shopping context. In a field experiment with 66 participants, we observed 328 decision-making instances. Two key EEG features—power spectral density (PSD) and prefrontal asymmetry index (PAI)—were analyzed. We employed traditional classifiers (KNN, RF, SVM) and a shallow neural network to assess predictive performance. Various feature selection methods (F-Score, T-value, Correlation, PCA, recursive feature elimination) were utilized. The results showed the SVM with a Gaussian kernel achieving the highest accuracy of 87.1%. Results underscore the significant roles of frontal and occipital regions in purchasing decisions. The PAI indicated the prefrontal cortex’s critical role in processing cognitive and emotional dimensions, while occipital PSD features highlighted visual processing and attention management in an online shopping environment. These findings affirm the potential of EEG features in consumer behavior analysis and emphasize the importance of advanced machine learning for interpreting neural decision-making correlates. This study provides a foundation for enhancing neuromarketing strategies and creating more engaging customer experiences, suggesting a comprehensive framework for improving marketing effectiveness and understanding consumer decision-making.