MEG Dataset and Analysis Scripts for “The Effects of Task Similarity During Representation Learning in Brains and Neural Networks”

MEG Dataset and Analysis Scripts for “The Effects of Task Similarity During Representation Learning in Brains and Neural Networks.” N., Menghi, W., J., Johnston, S., Viganò, M., A., B., Hinrichs, B., Maess, S., Fusi, C., F., Doeller (2025). This repository contains the MEG RSA data and analysis scripts used in the study “The Effects of Task Similarity During Representation Learning in Brains and Neural Networks.”The dataset was collected to investigate how shared task structures affect learning, generalization, and neural representations in the human brain. Participants were trained on tasks that either shared or differed in their underlying low-dimensional structure, while brain activity was recorded using magnetoencephalography (MEG).The dataset allows users to reproduce the cluster-based permutation analyses reported in the paper and to generate the figures illustrating the MEG results.ContentsMEG data (results of the correlation between empirical and model matrices at different dimensionalities and domains)Behavioral data (behavioural accuracy performance: "Spatual Source Data")Analysis scriptPython package developed to help with retrieving and computing simple operationsData formatData are organized according to a structured folder layout (see README.md in the repository) and include:npy MEG files (numpy).csv behavioral filesPython scripts using MNE-Python for statistical analysis and visualizationUsageThe provided scripts reproduce the statistical tests and figures presented in the manuscript. Users can adapt the code for their own representational analyses or cluster-permutation testing in MEG datasets.