Predicting Human Response in Feature Binding Experiment Using EEG Data

The ability to predict future human responses via analysis of electroencephalogram (EEG) signals is an active area of research from the viewpoint of understanding human brain and behavior, brain computer interface (BCI) applications, and neuro-rehabilitation of patients suffering with neurological symptoms and disorders. In this work, we predict human responses via analysis of EEG signals of healthy young adults collected during an experiment of visual feature binding. The subjects were asked to detect changes in color-shape binding of four objects shown in two successive screens with a gap of 1500 ms. The behavioral experiment comprised 96 trials as EEG data was collected simultaneously from a 21-electrode machine. The EEG data was pre-processed and artifacts were removed using independent component analysis (ICA). Feature reduction was carried out using principal component analysis (PCA) and linear discriminant analysis (LDA). A number of machine learning classifiers were trained on the EEG data of 15 subjects to predict the response of the subject in the color-shape binding experiment. Results are promising and show that EEG signal analysis can help in building relevant tools for the neuro-rehabilitation of subjects suffering with impairments in visual feature binding or tools for BCI applications.