Classification of the Neural Correlates of Mind Wandering States in EEG Signals

The presence of mind wandering states during attention-demanding activities may have negative consequences for task-related learning and success. Being able to distinguish between various levels of wandering through brain waves would be a fantastic tool in a variety of fields, including job optimization, driver vigilance monitoring, and even gaming. In this paper, we propose an approach that relies on EEG signal to deal with the issue of mind-wandering recognition as a classification challenge. Our model extracts the most important features of EEG data using a signal clipping and autocorrelation techniques then uses a deep neural network (DNN) to classify brain signal into high-level or low-level of wandering. The experiments demonstrated the effectiveness of the DNN model in detecting wandering episodes by achieving 73.5% accuracy which is 8.35% better than the best conventional neural network configuration.