dataset-affective-music-BCI

Dataset: A closed-loop, music-based brain-computer interface for emotion mediation

This repository contains the dataset accompanying the publication:

Ehrlich, S. K., Agres, K. R., Guan, C., & Cheng, G. (2019).
A closed-loop, music-based brain-computer interface for emotion mediation.
PLOS ONE, 14(3), e0213516. https://doi.org/10.1371/journal.pone.0213516

The dataset is organized around two experimental studies:

• Study I (listening): music listening experiment (affective responses / ratings)
• Study II (BCI): closed-loop affective BCI experiment (EEG + adaptive music feedback)

Documentation for each study (protocol, data format, variables, etc.) is included in the respective study folders.

---

Repository structure

dataset-affective-music-BCI/
├── study I (listening)/
│   ├── data_all_participants.mat
├── study II (BCI)/
│   ├── P01      # EEG adn meta data of participant 1
│   ├── P02      # EEG adn meta data of participant 2
│   └── ...
├── study_I_documentation.pdf
├── study_II_documentation.pdf
└── README.md

---

Study descriptions (brief)

Study I – Listening experiment

Purpose: establish affective responses to music stimuli and collect ground truth labels for emotion-related targets.

Design: Participants listened to music excerpts and provided self-report ratings (e.g., arousal/valence-related measures).
This study provides behavioral affect annotations and stimulus-related data that can be used for modeling affective responses to music.

Folder: study I (listening)/

Study II – Closed-loop affective BCI experiment

Purpose: evaluate a closed-loop affective BCI that adapts algorithmic music generation based on the user’s brain state, aiming at emotion mediation.

Design: Participants interacted with a system that collected EEG signals, extracted affect-relevant information, and used this information to adapt the music in real time (neurofeedback loop).
This study provides EEG recordings, event markers, system states, and additional behavioral measures depending on condition.

Folder: study II (BCI)/

---

About the dataset

Intended use

This dataset supports research in:

• affective computing and passive BCI
• music-induced emotion and music psychology
• neuroergonomics / human–AI interaction
• machine learning models for affect prediction and mediation

Modalities and measures (overview)

Depending on the study, the dataset includes:

• EEG recordings (Study II)
• music stimuli / musical parameters
• affect labels from self-report ratings (Study I and/or Study II)
• timestamps / event markers for synchronized analysis

For exact data formats, EEG configuration, label definitions, and event descriptions, consult the documentation inside each folder.

---

Citation

If you use this dataset in academic work, please cite:

@article{ehrlich2019closedloop,
  title   = {A closed-loop, music-based brain-computer interface for emotion mediation},
  author  = {Ehrlich, Stefan K. and Agres, Kat R. and Guan, Cuntai and Cheng, Gernot},
  journal = {PLOS ONE},
  volume  = {14},
  number  = {3},
  pages   = {e0213516},
  year    = {2019},
  doi     = {10.1371/journal.pone.0213516}
}