Utilizing sensory prediction errors for movement intention decoding: A new methodology.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• The brain utilizes forward models to predict sensory consequences of actions.
• Decoding human movement intentions from brain activity is crucial for advanced assistive technologies.
• Current methods often require extensive user training and may impose cognitive load.

Purpose of the Study:

• To develop and validate a new methodology for decoding human movement intentions.
• To leverage forward models and prediction errors for enhanced intention decoding.
• To improve the accuracy and efficiency of brain-computer interfaces for movement intention detection.

Main Methods:

• Proposing a method involving subliminal sensory stimulation of the modality relevant to the intended movement.
• Decoding movement intentions by identifying prediction errors between expected and actual sensory feedback.
• Utilizing electroencephalography (EEG) to capture brain responses to subliminal galvanic vestibular stimulation (GVS) during a wheelchair turning task.

Main Results:

• Achieved a median single-trial decoding accuracy of 87.2% across participants.
• Demonstrated high performance with zero user training and within 96 ms of stimulation.
• Confirmed no additional cognitive load due to the subliminal nature of the stimulation.

Conclusions:

• Subliminal sensory stimulation combined with prediction error decoding significantly improves movement intention detection from EEG.
• This methodology offers a promising, efficient, and low-cognitive-load approach for brain-computer interfaces.
• The findings have implications for developing intuitive control systems for assistive devices.