ROAMM

# ROAMM: Reading Observed At Mindless Moments **ROAMM** is a large-scale multimodal dataset featuring simultaneous **EEG and eye-tracking** data collected during naturalistic reading with **span-level mind-wandering annotations**. ROAMM provides a benchmark dataset for MW detection and EEG-to-text decoding tasks, and enables the study of attention-related degradation in language decoding from brain activity in naturalistic reading. ## Dataset Status - **Synchronized ML Dataset:** For researchers looking for the pre-processed, synchronized EEG and eye-tracking data (Pickle format), please navigate to: `derivatives/synced/` - **Linguistic Content:** Reading materials (words with coordinate information) are stored in `derivatives/stimuli/wiki_stories`. Each word is assigned a unique key to enable mapping fixated words back to their original corpus. - **Raw EEG (BIDS):** **Work in Progress.** We are currently converting the full raw EEG dataset for all participants into BIDS-compliant format. ## Project Details - **Task:** Naturalistic reading of standardized articles with retrospective self-report paradigm (ReMind task). - **Participants:** 44 subjects (50+ hours of data). - **Modalities:** - EEG (BioSemi ActiveTwo 64 channels). - Simultaneous Eye-Tracking (SR Research EyeLink 1000 Plus). - Span-level mind-wandering annotations. - Reading comprehension scores (page-level, multiple-choice questions). ## Structure This repository follows the Brain Imaging Data Structure (BIDS). - `participants.tsv`: Demographic information (age, sex, handedness, ADHD/Reading Disability status). - `derivatives/synced/`: Synchronized multi-modal data frames ready for Machine Learning pipelines. ## Code Repository Baseline models and machine learning pipelines are available at: **[https://github.com/GlassBrainLab/ROAMM](https://github.com/GlassBrainLab/ROAMM)** ## Publication & Citation The dataset paper describing the collection, synchronization, and baseline modeling of this data is available on OpenReview: [https://openreview.net/forum?id=zqLPdt09fE](https://openreview.net/forum?id=zqLPdt09fE) If you use ROAMM in your research, please cite: ```bibtex @inproceedings{sun2026roamm, title = {{ROAMM}: A Benchmark Dataset for Multimodal Human Attention Decoding and {EEG}-to-Text Modeling During Naturalistic Reading}, author = {Haorui Sun and Ardyn Vivienne Olszko and Niharika Singh and David C. Jangraw}, booktitle = {Forty-third International Conference on Machine Learning}, year = {2026}, url = {https://openreview.net/forum?id=zqLPdt09fE} } ```