The quest for interpretable models of neural population activity

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Brain function relies on coordinated activity of large neuronal populations.
• Advanced neural recording technologies offer unprecedented access to population activity.
• Extracting insights requires analytical tools linking population activity to system function.

Purpose of the Study:

• To develop analytical tools for understanding brain function from neural population activity.
• To motivate the use of latent variable models for low-dimensional descriptions of neural data.
• To explore new approaches for interpretable descriptions of neural components.

Main Methods:

• Utilizing latent variable models to describe neural population activity.
• Relating low-dimensional activity descriptions to experimental variables.
• Investigating uncontrolled variables like attention, arousal, and behavior.
• Developing novel approaches beyond traditional low-dimensional visualizations.

Main Results:

• Latent variable models provide a framework to analyze complex neural population dynamics.
• These models can link neural activity to experimental conditions and internal states.
• Emerging methods offer more interpretable insights into neural components.

Conclusions:

• Latent variable models are crucial for interpreting large-scale neural recordings.
• Advancements in these models are essential for understanding system-level brain function.
• Future research focuses on enhancing the interpretability of neural component descriptions.