Estimating neuronal firing density: A quantitative analysis of firing rate map algorithms

Area of Science:

• Neuroscience
• Computational Neuroscience
• Spatial Cognition

Background:

• Analysis of spatially tuned neurons (grid, place, boundary cells) relies on firing rate maps.
• Current mapping approaches lack quantitative parameter investigation, leading to study variability.
• Parameter choices significantly impact spatial map accuracy.

Purpose of the Study:

• To quantitatively assess the impact of tuning parameters on spatial firing rate map accuracy.
• To identify optimal parameters for minimizing error and balancing performance metrics.
• To compare the effectiveness of different mapping approaches.

Main Methods:

• Utilized simulated neural datasets to test various firing rate map generation approaches.
• Examined the influence of tuning parameters, recording duration, and firing field size.
• Evaluated methods based on error, computation time, place field detection, and data extrapolation.

Main Results:

• Identified parameter subsets yielding low-error spatial maps for each approach.
• Smoothed bivariate and averaged shifted histograms demonstrated fast computation with accurate maps.
• Adaptive smoothing/binning and kernel smoothed density estimation effectively handled low positional sampling.

Conclusions:

• Bivariate histograms with spatial smoothing are recommended for most spatial mapping tasks.
• The choice of mapping approach and parameters significantly affects neural data analysis accuracy.
• Quantitative investigation provides a foundation for standardized and reliable spatial neural mapping.