Real-Time EMG-EEG-Based BCI: Random Dot Motion Decision Task

Overview

This repository contains Python code for a real-time closed-loop Brain-Computer Interface (BCI) experiment investigating the "point of no return" in sensorimotor decision-making. The experiment implements a Random Dot Motion (RDM) task with evidence termination paradigm to test the double integration framework in motor control.

Research Objectives

The experiment aims to determine:

1. When a motor action becomes irreversible after initial motor preparation begins
2. The minimum time required for an EMG burst to trigger a completed action
3. Whether changing perceptual evidence after EMG onset can affect action completion

Key Features

• Real-time EMG monitoring with <10ms latency for onset detection
• Continuous Random Dot Motion stimulus presentation to avoid visual evoked potentials
• Closed-loop paradigm where visual evidence terminates based on EMG activity
• Precise timing control with hardware-accelerated PsychoPy implementation
• Lab Streaming Layer (LSL) integration for EMG-PsychoPy communication
• Adaptive delay mechanisms (10, 20, 30ms increments) between EMG onset and evidence termination

Technical Specifications

Hardware Requirements

• High-refresh monitor (60Hz minimum, 120Hz recommended)
• EMG recording system with real-time processing (1000Hz+ sampling)
• EEG system (optional, for offline analysis)
• Response buttons/keypads
• Lab Streaming Layer (LSL) compatible setup

Stimulus Parameters

• Dot Motion: 16.7 dots/degree², 2-3 pixel dots, 5°/s speed, 50ms lifetime
• Coherence Levels: 50% (high) and 0% (termination)
• Aperture: 7.1° diameter circular field
• Fixation: Continuous 5×5 pixel white square (0.5° visual angle)
• Frame Rate: 60fps minimum

Trial Structure

• ITI: Randomly varied (3, 6, 9 seconds)
• High Coherence Phase: Up to 1.5 seconds
• Evidence Termination: Triggered by EMG onset + variable delay (Δt)
• Confidence Rating: 1-7 scale (post-response)

Installation

# Clone the repository
git clone https://github.com/yourusername/emg-eeg-bci-experiment.git
cd emg-eeg-bci-experiment

# Install Python dependencies
pip install -r requirements.txt

# Required packages include:
# - psychopy>=2023.1.0
# - numpy>=1.21.0
# - pylsl>=1.16.0
# - scipy>=1.8.0
# - matplotlib>=3.5.0

Configuration Options

# Test EMG connection
python src/emg_interface.py --test

# Calibrate timing precision
python tests/timing_validation.py

# Run practice session
python src/main_experiment.py --participant P001 --practice

Data Output

The experiment generates several data files:

• Behavioral data: Response times, accuracy, confidence ratings
• EMG data: Onset times, burst characteristics, response completion
• Stimulus timing: Coherence change timestamps, frame timing validation
• Trial metadata: Condition parameters, trial outcomes

Key Implementation Features

Real-Time EMG Processing

• High-pass filtering (10Hz cutoff) with full-wave rectification
• Threshold detection (2-3 SD above baseline)
• Maximum 10ms latency from onset to PsychoPy trigger

Timing Precision

• Hardware-accelerated stimulus presentation
• Frame drop monitoring and trial exclusion
• Photodiode validation for critical timing events

Closed-Loop Control

• Seamless coherence transitions (50% → 0%) without VEPs
• Variable EMG-to-termination delays (Δt: 10, 20, 30ms)
• Continuous fixation maintenance throughout trials

Experimental Design

• 10-12 blocks of 60-80 trials each
• Counterbalanced motion directions (left/right)
• Systematic variation of EMG-coherence delays across blocks
• Adaptive procedures to focus on individual "point of no return"

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/emg-optimization)
3. Test timing accuracy with provided validation tools
4. Commit changes with detailed descriptions
5. Submit pull request with performance metrics

Contact

[Hamed Ghane] - [[email protected]] Project Link: https://github.com/HamedGhanes/Emg-EEG-BCI Project Lab: https://mphiliastides.org/

Acknowledgments

• Lab Streaming Layer (LSL) community for real-time data streaming
• PsychoPy development team for precise stimulus control