Here we provide the code to reproduce the results of our data resource paper:
"A large and rich EEG dataset for modeling human visual object recognition".
Alessandro T. Gifford, Kshitij Dwivedi, Gemma Roig, Radoslaw M. Cichy

If you experience problems with the code, please create a pull request or report the bug directly to Ale via email ([email protected]).

Please visit the dataset page for the data, paper, dataset tutorial and more.

Here you will find some useful videos on our EEG dataset.

To run the code first install Anaconda, then create and activate a dedicated Conda environment by typing the following into your terminal:

curl -O https://raw.githubusercontent.com/gifale95/eeg_encoding_model/main/environment.yml
conda env create -f environment.yml
conda activate eeg_encoding

Alternatively, after installing Anaconda you can download the environment.yml file, open the terminal in the download directory and type:

conda env create -f environment.yml
conda activate eeg_encoding

The source, raw and preprocessed EEG dataset, the training and test images and the DNN feature maps are available on OSF. The ILSVRC-2012 validation and test images can be found on ImageNet. To run the code, the data must be downloaded and placed into the following directories:

• Source EEG data: ../project_directory/eeg_dataset/source_data/.
• Raw EEG data: ../project_directory/eeg_dataset/raw_data/.
• Preprocessed EEG data: ../project_directory/eeg_dataset/preprocessed_data/.
• Training and test images; ILSVRC-2012 validation and test images: ../project_directory/image_set/.
• DNN feature maps: ../project_directory/dnn_feature_maps/pca_feature_maps.

• 00_data_collection: Matlab (Psychtoolbox) code used for data collection.
• 01_data_preparation: convert the source EEG data into raw EEG data, reformat the resting state data, and extract behavioral results.
• 02_eeg_preprocessing: preprocess the raw EEG data.
• 03_dnn_feature_maps_extraction: extract the feature maps of all images using four DNN architectures (AlexNet, ResNet-50, CORnet-S, MoCo), and downsample them using principal component analysis (PCA).
• 04_synthesizing_eeg_data: synthesize the EEG responses to images through linearizing and end-to-end encoding models.
• 05_synthetic_data_analyses: perform the correlation, pairwise decoding and zero-shot identification analyses on the synthetic EEG data.
• 06_plotting: plot the analyses results.

Here you will find a Colab interactive tutorial on how to load and visualize the preprocessed EEG data and the corresponding stimuli images.

If you use any of our data or code, partly or as it is, please cite our paper:

Gifford AT, Dwivedi K, Roig G, Cichy RM. 2022. A large and rich EEG dataset for modeling human visual object recognition. NeuroImage, 264:119754. DOI: https://doi.org/10.1016/j.neuroimage.2022.119754