policycompression_timememorycosts

Code for the manuscript "Time and memory costs jointly determine a speed-accuracy trade-off and set-size effects" (2024). The authors are Shuze Liu, Lucy Lai, Samuel J. Gershman, and Bilal A. Bari.

https://osf.io/preprints/psyarxiv/fupvy

• Bold words are .m filenames.
• Italicized words are .mat filenames or directory folder names.

All code assumes that the current directory of Matlab is the main folder. Hence, do the following before running any code:

# The variable main_folder contains the path to the main folder on your computer
cd(main_folder)
addpath(genpath(main_folder))

The code runs on Matlab version R2023a.

Online experiments

• Experiment 1: https://gershmanlab.com/experiments/shuze/bandits/experiment/index_exp1.html
• Experiment 2: https://gershmanlab.com/experiments/shuze/bandits/experiment/index_exp2.html
• Experiment 3: https://gershmanlab.com/experiments/shuze/bandits/experiment/index_exp3.html
• Experiment 3 variant with 5 set-size conditions (only appears in Figure S5): https://gershmanlab.com/experiments/shuze/bandits/experiment/index_exp3_old2.html

Main folder

• plot_manuscript_figures.m creates all figures for the manuscript, based on the datafiles and the saved LBA model fits.

data subfolder

• iti_bandit_data_exp1.mat to iti_bandit_data_exp3_old2.mat are the raw datafiles for each experiment. We excluded all participants who, in any test block, had an average RT over 5 seconds. These datafiles are used by plot_manuscript_figures.m above.
• Other .mat files with the suffix train_blocks contain only training block performance. They include both included and excluded participants.
• Other .mat files with the suffix test_blocks contain only test block performance.
• Among the above, .mat files with the suffix _noexclusion do not apply the exclusion criterion. Hence they include both included and excluded participants.

figures subfolder

• It contains all figures generated by plot_manuscript_figures.m above.

utils subfolder

• It contains helper functions for the Blahut-Arimoto algorithm, mutual information and conditional entropy estimation, within-participant errorbar visualization, and color palettes.

lba_models subfolder

• lba_fits_exp1.m and lba_fits_exp3.m perform LBA fits for Experiment 1 and 3. They parse datafiles into LBA-compatible format and fit parameters.

data subsubfolder

• It contains parsed datafiles for Experiment 1 and 3 for subsequent LBA fits.

fits subsubfolder

• It contains LBA fitted parameters for each experiment. They are used by plot_manuscript_figures.m above.

utils subsubfolder

• It contains LBA code adapted from https://github.com/smfleming/LBA. Some of the code was modified for our model fits.