Masked and unmasked error-related potentials during continuous control and feedback.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Error-related potentials (ErrPs) are crucial for brain-computer interfaces (BCIs) with discrete feedback.
• Decoding ErrPs in continuous feedback scenarios remains a significant challenge in BCI research.

Purpose of the Study:

• To investigate electroencephalographic (EEG) signatures of cursor control loss in continuous feedback tasks.
• To compare ErrPs under normal versus jittered (masked) continuous feedback conditions.
• To assess the feasibility of asynchronous ErrP detection in a simulated BCI setting.

Main Methods:

• Developed a continuous cursor control task using a joystick with normal and jittered visual feedback.
• Analyzed time-locked EEG data to identify error-related potentials (ErrPs).
• Performed time-locked and asynchronous classification analyses to distinguish correct vs. error trials and masked vs. unmasked errors.

Main Results:

• Time-locked analysis showed typical ErrP components, with masked errors delayed and having lower amplitudes than unmasked errors.
• Classification of correct vs. error trials was accurate (TPR 81.8%), but masked vs. unmasked errors were indistinguishable (chance level).
• Asynchronous ErrP detection achieved a true positive rate (TPR) of 64.9% and true negative rate (TNR) of 84.0%.

Conclusions:

• Time-locked classification failed to differentiate between masked and unmasked errors.
• Asynchronous ErrP detection proved robust, indicating that feedback modality does not impede real-time error detection in BCIs.