Real-Time Emotion Recognition for EEG Signals Recollected from Online Poker Game Participants

This study employs a novel emotion elicitation method to recollect EEG signals using an EEG-based BCI commercial device with 14 channels for emotion recognition with machine learning algorithms. We obtained EEG signals of six participants playing a complete Texas hold ’em poker game consisting of several hands in a sit-and-go online tournament with real money at stake. Poker was chosen because it is a game that provokes both positive and negative emotions associated with the sudden changes of the hands. To obtain the EEG readings, we use an EEG-based BCI headset with 14 electrodes, and the participants reported emotions based on self-assessment manikin in the valence – arousal space. This information was processed with kNN, Random Forest, and MLP neural network machine learning algorithms with 68.94, 68.58, and 68.82 values for accuracy, respectively. Therefore, the stimulus of poker game playing evoked emotions that were detectable and classifiable using automatic machine learning systems.