A new statistical method to analyze Morris Water Maze data using Dirichlet distribution

Abstract

The Morris Water Maze (MWM) is a behavioral test widely used in the field of neuroscience to evaluate spatial learning memory of rodents. However, the interpretation of results is often impaired by the common use of statistical tests based on independence and normal distributions that do not reflect basic properties of the test data, such as the constant-sum constraint. In this work, we propose to analyze MWM data with the Dirichlet distribution, which describes constant-sum data with minimal hypotheses, and we introduce a statistical test based on uniformity (equal amount of time spent in each quadrant of the maze) that evaluates memory impairments. We demonstrate that this test better represents MWM data and show its efficiency on simulated as well as in vivo data. Based on Dirichlet distribution, we also propose a new way to plot MWM data, showing mean values and inter-individual variability at the same time, on an easily interpretable chart. Finally, we conclude with a perspective on using Bayesian analysis for MWM data.