Prediction errors from distinct perspectives induce separable EEG features for brain-computer interface.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Cognitive Science

Background:

• Error detection is vital for adaptive behavior and advanced brain-computer interfaces (BCI).
• Error-related potentials (ErrP) signal prediction errors in BCI but struggle to differentiate self-caused from other-caused errors.
• This limitation hinders BCI application in social contexts.

Purpose of the Study:

• To investigate the electroencephalogram (EEG) differences between first-person perspective (1PP) and third-person perspective (3PP) prediction errors.
• To assess the feasibility of distinguishing these error types for improved BCI functionality.
• To enhance BCI capabilities for collaborative tasks and social interactions.

Main Methods:

• Utilized virtual reality (VR) to induce 1PP and 3PP prediction errors in 22 healthy subjects.
• Recorded electroencephalogram (EEG) data during the VR tasks.
• Analyzed event-related potentials (ERP), event-related spectral perturbation (ERSP), inter-trial coherence (ITC), and employed a shrinkage discriminant canonical pattern matching (SKDCPM) algorithm.

Main Results:

• ErrP from 1PP errors appeared significantly earlier than 3PP errors.
• Greater ERSP and ITC in the prefrontal theta and alpha bands were observed for 1PP errors.
• Accurate decoding of 1PP vs. 3PP errors achieved 76.4%± 9.13% accuracy.

Conclusions:

• This research successfully differentiates prediction errors based on perspective (1PP vs. 3PP) using EEG.
• The findings provide a more granular classification of error types, crucial for sophisticated BCI.
• This work lays the foundation for improved two-person collaborative brain control and human-machine hybrid intelligence.