Neural Dynamics of Social Cognition: A Single-trial Computational Analysis of Learning under Uncertainty

This repository contains Matlab code for reproducing the results in:

Charlton, C. E., & Hauke, D. J., et al. (2025). Neural Dynamics of Social Cognition: A Single-trial Computational Analysis of Learning under Uncertainty.

Members of the project

Supervision: Daniel J. Hauke, Andreea O. Diaconescu

Contributors: Vladimir Litvak, Michelle Wobmann, Christina Andreou, Renate de Bock, Stefan Borgwardt, Volker Roth

Project description

We acquired EEG data from 43 control participants during an advice-taking task. This data was modeled to explore the computational mechanisms that govern social cognition and learning under uncertainty.

Submodules

This repository utilizes submodules. If you're cloning the repository, make sure to get the associated submodules as well.

The current submodules include:

• SPM12

Getting Started

1. Clone this repository along with its submodules using the following command:

git clone --recurse-submodules https://github.com/colleenc11/compi_ioio.git

If you've already cloned the repository without its submodules, initialize and fetch them with:

git submodule update --init --recursive

2. Open Matlab and navigate to the COMPI_IOIO/code directory.
3. Initialize your environment by running compi_setup_paths script.

Running Pipeline in Matlab

The models were implemented in Matlab (version: 2023a; https://mathworks.com) using the HGF toolbox (version: 6.0). This toolbox is part of the open-source TAPAS (Frässle et al., 2021) software collection, available here.

Steps to run the pipeline:

1. Open Matlab and navigate to the COMPI_IOIO/code directory.
2. From within this directory, run the 'compi_master_eeg' script.
3. To configure analysis options, navigate to the code/configs/options directory and modify the compi_ioio_options file.

Third Party Code

The repository also contains the following external, freely available Matlab tools:

• the function notBoxPlot by Rob Campbell, available here.
• the TNUEEG toolbox for preprocessing, aligning with the methods presented in the study by Weber et al., 2022.