This repository contains the code for performing the analysis, computational modelling, and plotting described in the manuscript Bayesian confidence in optimal decisions.
Author (excluding submodules): Joshua Calder-Travis
Some parts of the code need the VBL toolbox to be installed in order to run. Each session before using the code all the folders and subfolders in the respository should be added to the matlab path.
To run the plotting functions it is necessary to download the data and model fit results from the associated OSF project. All data and fits are contained in subfolders of the folder "DataAndFits". The path of the "DataAndFits" folder on your local system needs to be specified, so that the code can load the appropriate files. To do this, a matfile named '''confDataAndFitDir.mat''' should be saved in the main folder of the code respository. This matfile should contain a single variable dataDir, which contains the full file path of the "DataAndFits" folder on your system.
To make a wide range of plots of data and fits, including the plots in the paper, call makePlotsForEmpiricalPaper().
To perform the fitting instead of using the fitting results supplied in the associated OSF project the functions runCrossValFits and fitModels can be used with appropriate inputs to either run cross-validated model fitting, or fitting to all data without train-test splitting, respectively. See comments on those functions for more detail. The function produceDefaultModelLists provides a way of conveniently producing a list of the name-codes for the key models. Users wishing to run the fitting on a computer cluster will need to adapt the script mT_runOneJob.sh and mT_submitAllJobs.sh to their specific environment and cluster scheduling system. Results from a cluster can be collected and analysed using the mT_analyseClusterResults function.
Reviewed all code Jan 2024. The code has been carefully checked. Nevertheless, no guarantee or warranty is provided for any part of the code. See license for further details.