<h1>Overview</h1>
This repository contains models fitted with GrOWL and LASSO to predict behaviorally-estimated semantic similarity structure among viewed images of familiar animate and inanimate items based on neural recordings obtained via ECoG from 10 participants. The study is described in the following preprint https://doi.org/10.1101/2022.10.27.514039 .
<h2>Workflow</h2>
The model fitting and subsequent analyses were implemented in Matlab. The data are stored in `.mat` files, the native Matlab storage format. The code for the model fitting is part of the WISC MVPA toolbox: https://doi.org/10.5281/zenodo.6975270 .
The fitted models are then read back into Matlab and processed by the `correlation_analysis.m` script, which relies on the WISC MVPA toolbox linked above and an additional repository called `RSL_correlation_analysis` https://doi.org/10.5281/zenodo.7566607 . This will produce `.csv` formatted text files organized in a directory tree.
Finally, once the correlation analyses has completed, figures can be generated using **R** scripts in the `ecog-rsl-figures` repository https://doi.org/10.5281/zenodo.7566609 .