DataHigh: graphical user interface for visualizing and interacting with high-dimensional neural activity.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Data Visualization

Background:

• Neural population activity, involving tens to hundreds of neurons, can be simplified using dimensionality reduction techniques.
• Latent variables from dimensionality reduction define a lower-dimensional space for studying neural population dynamics.
• Visualizing this reduced-dimensional space is crucial for understanding neural activity variations across time, trials, and conditions.

Purpose of the Study:

• To develop an interactive visualization tool for exploring high-dimensional neural population activity.
• To address the limitations of 2D/3D plotting for data with optimal dimensionality greater than 3.
• To facilitate intuitive and hypothesis-driven exploratory data analysis in neuroscience.

Main Methods:

• Development of a MATLAB graphical user interface (GUI).
• Implementation of interactive navigation through a continuum of 2D projections of the reduced-dimensional space.
• Application of the GUI to visualize neural population activity from premotor and motor cortices during reaching tasks.

Main Results:

• The GUI enables dynamic visualization of neural population activity, including single-trial and trial-averaged data.
• Interactive tools like movie playback and summary statistics (e.g., covariance ellipses) enhance data exploration.
• Demonstrated utility in analyzing neural population dynamics during motor tasks.

Conclusions:

• The developed GUI provides a versatile platform for visualizing and analyzing complex neural population data.
• Interactive exploration of multiple 2D projections is critical for avoiding misleading interpretations.
• This visualization approach, combined with dimensionality reduction, can advance the understanding of neural population coding.