Thought Chart: tracking the thought with manifold learning during emotion regulation

Area of Science:

• Neuroscience
• Computational Neuroscience
• Data Science

Background:

• The brain's complex states can be conceptualized as a high-dimensional manifold.
• Understanding these dynamics is crucial for cognitive and emotion regulation research.
• Existing methods may not adequately capture the intricate, dynamic nature of neural activity.

Purpose of the Study:

• To develop a manifold learning framework (Thought Chart) for reconstructing and visualizing brain state dynamics.
• To explore the underlying neurophysiological dynamics during emotion and cognitive regulation tasks.
• To create a data-driven approach for analyzing complex functional brain data.

Main Methods:

• Utilized electroencephalography (EEG) recordings from 20 healthy participants during resting state and an emotion regulation task.
• Constructed temporal dynamic functional connectomes from EEG data.
• Applied graph dissimilarity space embedding and nonlinear dimensionality reduction (NDR) techniques (k-nearest neighbor and epsilon distance-based) to visualize the manifold in 2D.

Main Results:

• Successfully reconstructed the high-dimensional brain state manifold into a 2D space, termed Thought Chart.
• Demonstrated that different task conditions (resting state vs. emotion regulation) form distinct trajectories within the Thought Chart.
• Identified potential parameters (e.g., distribution, trajectory length) in the 2D space for exploring cognitive load and emotion processing.

Conclusions:

• Thought Chart provides a novel, data-driven method for learning and visualizing the neurophysiological dynamics of brain activity.
• The framework offers insights into the brain's response to cognitive and emotional challenges.
• This approach has potential applications in understanding complex functional brain data and cognitive processes.