MINT: A toolbox for the analysis of multivariate neural information coding and transmission

Area of Science:

• Neuroscience
• Computational Neuroscience
• Information Theory

Background:

• Information theory is a key framework for understanding brain information processing and neural network behavior.
• Traditional methods focused on small neural populations; recent advances allow analysis of larger networks and inter-area communication.
• Multivariate information theory and dimensionality reduction are crucial for analyzing complex neural data from large populations or multiple brain regions.

Purpose of the Study:

• Introduce the Multivariate Information in Neuroscience Toolbox (MINT).
• Provide a tool combining advanced information theoretic methods with statistical approaches for robust neural data analysis.
• Demonstrate MINT's utility with simulated and real neural data from electrophysiology and calcium imaging.

Main Methods:

• Development of the Multivariate Information in Neuroscience Toolbox (MINT).
• Integration of multivariate information theory and dimensionality reduction techniques.
• Application of statistical tools for robust estimation from limited empirical datasets.

Main Results:

• MINT successfully analyzes information encoding and communication across neural populations and brain areas.
• The toolbox is applicable to various brain activity measurement modalities, including electrophysiology and calcium imaging.
• MINT facilitates reverse engineering of information processing and emergent functions from neural interactions.

Conclusions:

• MINT offers a powerful, integrated approach for neuroscience research.
• The toolbox enhances the study of information flow and processing within complex neural systems.
• MINT is freely available for Linux, Windows, and macOS, supporting diverse neuroscience research needs.