Decoding Intrinsic Fluctuations of Engagement From EEG Signals During Fingertip Motor Tasks.

Area of Science:

• Neuroscience
• Rehabilitation Engineering
• Human-Computer Interaction

Background:

• High mental engagement is crucial for effective motor rehabilitation.
• Flow experience, a state of deep engagement, is an ideal but challenging goal in rehabilitation tasks.
• Existing flow research often lacks objective, continuous measures, especially during motor tasks.

Purpose of the Study:

• To decode intrinsic fluctuations of flow experience from electroencephalogram (EEG) signals during a virtual reality fine fingertip motor task.
• To address the gap in flow research concerning real-time fluctuations and neural correlates.
• To develop a machine learning model for classifying high-flow versus low-flow states.

Main Methods:

• A virtual reality-based fine fingertip motor task with adaptive difficulty was employed.
• Motor behavioral measures were used to represent and label flow states, overcoming sparse self-reporting.
• EEG signals were analyzed using spectral power and coherence features to train a machine learning-based neural decoder.

Main Results:

• The neural decoder achieved over 80% classification accuracy in distinguishing high-flow from low-flow states.
• High-frequency bands in EEG activities were identified as significant contributors to flow decoding.
• Increased power in parietal-occipital electrodes and global coherence in alpha and beta bands were observed during high-flow periods.

Conclusions:

• Decoding intrinsic flow fluctuations during fine motor tasks using EEG is feasible with high accuracy.
• The findings support the use of EEG-based neural decoding for objective flow assessment in rehabilitation.
• This methodology provides a foundation for future interventions aimed at manipulating flow and enhancing engagement in motor rehabilitation.