Development of an EEG-Based Method for Detecting Flow State Using a Wearable Headband in a Game Environment.

Area of Science:

• Neuroscience
• Cognitive Science
• Human-Computer Interaction

Background:

• Flow state is a peak mental condition linked to heightened productivity and well-being.
• Understanding the neural underpinnings of flow is an active area of research.
• Consumer electroencephalography (EEG) devices offer accessible tools for studying brain activity.

Purpose of the Study:

• To develop and validate a low-channel EEG-based method for detecting the flow state.
• To explore the efficacy of entropy-based features from EEG signals for flow detection.
• To assess the performance of a Random Forest classifier in identifying flow using consumer-grade EEG data.

Main Methods:

• Acquired 29 EEG recordings from participants playing a Tetris game using a consumer EEG device.
• Preprocessed EEG signals, including denoising and Discrete Wavelet Transform (DWT) for signal decomposition.
• Extracted Slope Entropy, Distribution Entropy, and Spectral Entropy features from EEG sub-bands.
• Employed a Random Forest classifier with Random Sampling (RS) and Leave-One-Subject-Out (LOSO) cross-validation.

Main Results:

• The Random Forest classifier achieved a mean accuracy of 93% with Random Sampling validation.
• Leave-One-Subject-Out (LOSO) cross-validation yielded an average accuracy of 82%.
• The proposed method demonstrated robust performance in detecting the flow state from low-channel EEG data.

Conclusions:

• The developed method shows significant promise for real-time flow state detection in practical applications.
• Consumer-grade EEG devices, combined with advanced signal processing and machine learning, can effectively identify flow.
• This research contributes to a deeper understanding of the neural correlates of flow and its detection.