embodied-vr-feedback-3d-motor-imagery-bci - SKILL.md Agent Skill

name: embodied-vr-feedback-3d-motor-imagery-bci description: "Embodied Virtual Reality feedback reshapes neural representations to support continuous 3D motor imagery decoding in brain-computer interfaces. First systematic investigation of embodied VR feedback during real-time 3D virtual limb control. Use when: (1) Designing VR-based BCI systems, (2) Studying motor imagery neural representations, (3) Comparing VR vs screen feedback modalities, (4) Investigating longitudinal BCI training effects. Activation: embodied VR feedback, motor imagery BCI, 3D virtual limb, VR vs screen, continuous BCI, neural representations reshaping, sensorimotor-parietal" license: Complete terms in LICENSE.txt metadata: arxiv_id: "2605.29677" published: "2026-05-28" authors: "Niall McShane, Attila Korik, Karl McCreadie, Naomi Du Bois, Darryl Charles, Damien Coyle" journal: "Nature Biomedical Engineering (submitted)" zenodo_doi: "10.5281/zenodo.16047021" tags: [embodied-vr, bci, motor-imagery, neural-representations, continuous-decoding, vr-feedback, cnn-lstm, longitudinal-training]

Embodied VR Feedback Reshapes Neural Representations

arXiv:2605.29677 | Submitted: 2026-05-28 | Journal: Nature Biomedical Engineering (submitted)

Overview

首次系统性调查实时 3D 虚拟肢体控制中的具身 VR 反馈如何通过运动想象驱动，以及反馈模态和纵向训练如何塑造神经表征和解码性能。10 个受试者，10 个纵向 sessions。

Core Innovation

Embodied VR Feedback System

First Systematic Investigation: 首次系统性调查具身 VR 反馈在实时 3D 虚拟肢体控制中的作用
Longitudinal Training: 10 个受试者，10 个纵向训练 sessions
Real-time 3D Control: 实时 3D 虚拟肢体控制，由运动想象驱动

Three Evaluation Strategies

Fixed Decoder Generalisation (FDG): 实际在线性能，固定解码器泛化
Sequential Adaptive Training (SAT): 定期重新训练，顺序适应训练
Within-Session Reconstruction (WSR): 会话内上限估计，会话内重建

CNN-LSTM Decoder

VR 下 imagined movement correlations: r = 0.762
Screen feedback baseline: r = 0.672
VR 优势显著: 8.9-13.0% (p <= 0.002, d = 1.42-2.05)

Key Results

Performance Comparison

Feedback	Correlation	Improvement
VR	r = 0.762	8.9-13.0%
Screen	r = 0.672	baseline

VR 在所有策略和运动维度上显著优于 screen
VR 优势在固定解码器（无重新训练）下持续存在
具身 VR 反馈产生本质上更可解码和可泛化的神经表征

Statistical Analysis

Linear Mixed-Effects Model: 反馈模态和运动轴主效应稳健
无交互效应
所有运动维度 VR 优势显著

Neurophysiological Findings

Enhanced Desynchronisation

Sensorimotor-Parietal: VR 产生更强的 sensorimotor-parietal 去同步化
Motor-Frontal Connectivity: 增强 motor-frontal 功能连接

Anterior Insula Engagement

Pervasive Engagement: 所有频率波段的前部脑岛参与
Real Movement Patterns: 与真实运动执行相关的模式

Superior Parietal Lobule Coupling

Increased Coupling: 增加的上顶叶耦合
Spatial Processing: 空间处理相关

Neural Representation Reshaping

VR vs Screen Feedback

VR: 具身空间反馈
Screen: 传统屏幕反馈
Reshaping Effect: VR 重塑神经表征

Brain Network Changes

Sensorimotor Network: 感觉运动网络激活增强
Parietal Cortex: 顶叶皮层去同步化增强
Frontal-Motor Connectivity: 额叶-运动连接增强

Applications

Continuous BCIs

Next-generation BCI design principles
Intuitive motor control
Neurorehabilitation

VR-based Training

Embodied feedback design
Longitudinal training protocols
Performance enhancement

Neural Rehabilitation

Stroke rehabilitation
Motor recovery
Spatial feedback therapy

Key Design Principle

Embodied Spatial Feedback: 具身空间反馈作为下一代连续 BCI 的关键设计原则

Why VR Outperforms Screen?

Embodied Experience: 具身体验增强运动想象
Spatial Representation: 空间表征更接近真实运动
Neural Engagement: 神络参与更强
Functional Connectivity: 功能连接增强

Implementation

VR System Requirements

Embodied Virtual Reality: 具身虚拟现实系统
3D Virtual Limb: 3D 虚拟肢体渲染
Real-time Feedback: 实时反馈系统

Decoder Architecture

CNN-LSTM Decoder:
- CNN: Spatial feature extraction
- LSTM: Temporal sequence processing  
- Output: 3D movement trajectory

Training Protocol

10 Sessions: 10 个训练 sessions
10 Participants: 10 个受试者
3 Strategies: FDG, SAT, WSR

Clinical Relevance

Neurorehabilitation Applications

Stroke rehabilitation
Spinal cord injury recovery
Motor function recovery

BCI Design Principles

Embodied feedback as key design principle
Spatial VR feedback enhances decoding
Longitudinal training protocols

Data Availability

Zenodo DOI: https://doi.org/10.5281/zenodo.16047021

Citation

@article{mcshane2026embodiedvr,
  title={Embodied Virtual Reality Feedback Reshapes Neural Representations to Support Continuous Three-Dimensional Motor Imagery Decoding},
  author={McShane, Niall and Korik, Attila and McCreadie, Karl and Du Bois, Naomi and Charles, Darryl and Coyle, Damien},
  journal={arXiv preprint arXiv:2605.29677},
  year={2026},
  note={Submitted to Nature Biomedical Engineering}
}