The tarball files in sequence 'data.tar.gzxx' and 'scripts_E1.tar.gz' etc contain scripts and data underlying Experiment 1 in Acunzo, Grignolio, & Hickey (2025, PLOS Biology). 

The tarball files in sequence 'data_E2.tar.gzxx' contain scripts and data underlying Experiment 2 in the paper. 

There is a separate readme file for E2 (prepared by David Acunzo). 

Sha512sum output is provided for E1 and E2 files separately in 'sha512sum_output_E1.txt' and 'sha512sum_output_E2.txt'. These can be used to check for data corruption during download of the files. 

For E1, the EEG data is provided in EEGLAB 'set' format for MATLAB. The data provided has been preprocessed. Trials with eye movements have been removed, and ICA has been used to remove artifactual components associated with blinks and residual eye movement activity. Details on preprocessing are provided in the paper. 

Analysis scripts are provided. These are unlikely to run immediately on your own hardward due to 2 issues. 

The first is that we relied heavily on cluster computing for data analysis, and therefore the scripts are optimized for our cluster. Note that the analyses are computationally heavy; you are likely to require HPC to run this kind of analysis. 

The second issue is that we make use of an adapted version of the Cosmomvpa toolbox for MATLAB (www.cosmomvpa.org). The adaptations allow us to keep output from each iteration during cross validation (so that we can relate EEG characteristics in each sample to decoding accuracy). We have not provided these modified scripts, as our scripts have also been adapted in other ways that are sure to cause confusion, but the key change is addition of an output variable to cosmo_crossvalidate.m that keeps the models generated in each fold. These can be captured within the loop that begins on line 186. You need to additionally adapt any upstream functions calling cosmo_crossvalidate.m so that this new variable is organized and ultimately passed to your top-level call.

This all means that the scripts are provided as a set of detailed notes on the analysis, and to provide a general path for you to follow if you plan on using the technique, rather than as a package you can run. 

The analysis scripts contain some details on EEG event code organization. Code and stimuli for the task is also provided in 'task_E1.tar.gz'. In spite of the label, this file contains the tasks for E1 and E2. These differ in minor ways associated to the recording of unimodal EEG or EEG/MRI. The task code relies on psychtoolbox (http://psychtoolbox.org/). For both experiments, event triggers were conveyed via a labjack I/O unit; the code will not run as is without this hardware installed, and code referencing this hardward will need to be removed / commented out to get it to run otherwise.  

Don't hesitate to reach out if you have conceptual questions. 

Clayton Hickey
Feb 3rd, 2025
c.m.hickey@bham.ac.uk