Neural codes of probability

This repository accompanies the paper entitled A non-monotonic code for event probability in the human brain. by Cedric Foucault (1,2,#,§), Tiffany Bounmy (1,#), Sébastien Demortain (1), Bertrand Thirion (3), Evelyn Eger (1), Florent Meyniel (1,4,§)

1. Cognitive Neuroimaging Unit, NeuroSpin (INSERM-CEA), University of Paris-Saclay, 91191 Gif-sur-Yvette, France
2. Sorbonne University, Doctoral College, F-75005 Paris, France
3. Inria, CEA, University of Paris-Saclay, Palaiseau, France
4. GHU Paris, psychiatrie et neurosciences, Hôpital Sainte-Anne, Institut de neuromodulation, 75014 Paris, France

#: co-first authors, §: corresponding authors cedric.foucault@gmail.com and florent.meyniel@cea.fr.

Authors of the code

• Tiffany Bounmy
• Sebastien Demortain (preliminary aspects)
• Cédric Foucault
• Florent Meyniel

Structure of the repository

The repository contains several folders:

• behavior_analysis: scripts to analyze the behavioral data collected in the MRI
• fmri_analysis: scripts for fMRI data analysis
• ProbLearnTask: scripts of the task used for data collection in the MRI scanner
• simulation: scripts for simulation analyses
• TransitionProbModel: Ideal observer model to infer probabilities and confidence from a sequence
• utilities: functions and parameters used in many scripts

Runing the code

• scripts should be executed from the root of the codebase. For instance: python behavior_analysis/correlate_subject_model_estimate.py
• before executing any scripts, the root directory of the codebase should be added to your PYTHONPATH. From the root directory, run: export PYTHONPATH="$PYTHONPATH:$(pwd)". This ensures that Python can locate and import the modules used in the scripts.
• from scratch: all scripts are provided, from data preprocessing to the final results
• reprocude the figures of the paper: the results in the paper can be generated, without starting from scratch, by executing the scripts located in fmri_analysis/second_level_analyses, this will use pre-computed results and output the tables and figures saved in output/results
• the notebook notebook_encoding_model_using_a_basis_set.ipynb explains the methods behind the versatile encoding model used in the paper.

System requirement

Python 3.7