Maze Analysis Repository

MM 7/10/2021

This repo accompanies our article:

Rosenberg, M., Zhang, T., Perona, P., and Meister, M. (2021). Mice in a labyrinth exhibit rapid learning, sudden insight, and efficient exploration. ELife 10, e66175. (https://doi.org/10.7554/eLife.66175)

Also see the submitted version, published as a preprint 1/15/2021:

Rosenberg, M., Zhang, T., Perona, P., and Meister, M. (2021). Mice in a labyrinth: Rapid learning, sudden insight, and efficient exploration. BioRxiv 2021.01.14.426746. (https://doi.org/10.1101/2021.01.14.426746)

Contents of the repo

This repo contains all the data and code needed to reproduce the analysis in the submitted preprint and the revised version.

Rosenberg_Preprint_20210114.pdf: Our submitted manuscript.

Rosenberg_Revision.pdf: Revised version after peer review at ELife.

Maze_Analysis_3A,...,Maze_Analysis_3D. These four jupyter notebooks cover the submitted preprint. They gradually develop the various topics of analysis, starting from raw data, producing figure panels and numerical results for the article along the way. They contain a good number of comments and mathematical sections to guide the user.

Maze_Analysis_3E covers changes and additions made in response to peer review. Some of this material appears in our revised article and the ELife version.

code/: Contains python files with routines accessed from multiple notebooks.

outdata/: A place for data files, both input and output.

outdata - tf files only/: Just the raw data, the starting point for all analysis.

figs/: A place for PDF files that make up the figure panels in the article.

apparatus/: Instructions and files for building the maze used in the article.

How to reproduce all the analysis starting from raw data

0. Read our paper. The version of Jan 2021 is included in the repo. Then read at least the start of Maze_Analysis_3A.
1. Empty the outdata/ directory. Fill it with the contents of outdata - tf files only/. Now you're starting with the raw trajectories of animals in the maze.
2. Empty the figs/ directory.
3. Run the notebooks Maze_Analysis_3A,...,Maze_Analysis_3E in alphabetical sequence.
4. Now the figs/ directory should contain all the figure panels plus a few extras.

How to find code for a specific figure panel

• The names of all the figure panels (as numbered in the preprint of Jan 2021) appear as level-3 headings in the notebooks. Look through these to find your figure of interest. Or...
• In the figs/ directory find the name of the PDF file of interest, and search for that name in the notebooks.

How to view the raw videos

You can find these on Youtube:

• Rewarded animals
• Unrewarded animals