This repository contains preprocessed EEG data, stimuli and their ground-truth annotations for the paper **"Investigating the temporal dynamics and modelling of mid-level feature representations in humans"** (Karapetian et al., 2025). Please cite this paper if you use any of the data or the stimulus set. For any questions or to request additional files (e.g., raw EEG data), please email **agnessakarapetian@gmail.com**. More specifically, this repository contains: - ***EEG data*** for the image (N=15) and video (N=20) experiments, preprocessed according to the steps described in the paper - ***Stimuli*** (1440) in .jpg, as images (one frame) and videos (9 frames) - ***Ground-truth annotations*** (5 features, 1440 stimuli) for the image and video stimuli (one annotation per feature and image frame) - ***Metadata*** for the stimulus set, including action identity information for each stimulus (action_indices.csv) and train/test/val splits, along with character identity information (meta_data_anim.mat). Fore more detailed information on the contents of the repository, please consult the readme.txt file. The code for this paper is under https://github.com/Agnessa14/Mid-level-features/.