De Novo Brain-Computer Interfacing Deforms Manifold of Populational Neural Activity Patterns in Human Cerebral Cortex

Area of Science:

• Neuroscience
• Cognitive Science
• Biomedical Engineering

Background:

• Human brains exhibit remarkable adaptability, modulating innate activities for novel tasks and environments.
• The sensorimotor system acquires diverse activity patterns to enhance behavioral performance.
• Understanding neural adaptation during brain-computer interface (BCI) learning is crucial for BCI development.

Purpose of the Study:

• To map neural repertoire acquisition during BCI tasks to performance improvements.
• To analyze net neural populational activity during voluntary BCI modulation.
• To investigate the impact of different BCI classifier types on neural adaptation.

Main Methods:

• Recorded whole-head high-density electroencephalograms (EEGs) from human participants.
• Applied dimensionality reduction to capture cortical activity pattern changes.
• Analyzed neural manifold dynamics in relation to BCI classifier interactions.

Main Results:

• Systematic interactions were found between neural activity patterns and BCI classifiers.
• Neural manifolds stretched with motor-related EEG features for fixed classifiers, but not adaptive ones.
• Manifolds deformed orthogonally to the boundary for novel classifiers with fixed, unnatural features.

Conclusions:

• Neural plasticity is induced by BCI operation requiring fixed activities, irrespective of biological naturalness.
• Macroscopic neural adaptation principles may explain learning diverse behaviors and environmental adaptation.
• BCI design can leverage these principles to enhance user learning and adaptation.