L-FAME: Longitudinal Focused Attention Meditation EEG Dataset and Benchmark
Angqi Li, Ab Basit Rafi Syed, Hamzeh Alzweri, Taosheng Liu, Barry H. Cohen, Saiprasad Ravishankar
arXiv (Cornell University) May 21, 2026 DOI: 10.48550/arxiv.2605.22893 (opens in new tab) via OpenAlex
Summary
AI-generated from the abstractA new dataset, L-FAME, provides EEG recordings and psychological assessments from 74 healthy college participants who were randomly assigned to one of three meditation practices (two mantra-based techniques and one breath focus) and measured before and after a six-week training period. The dataset supports three benchmark tasks: distinguishing resting from meditation states, classifying which specific technique was used, and testing whether machine learning models can generalize across the two time points. Baseline results using classical and deep learning methods are reported, and the dataset, preprocessing code, and evaluation framework are publicly released to advance computational meditation research and EEG-based machine learning.
Study at a glance
| Characteristics | Randomized controlled trial Longitudinal Peer reviewed |
|---|---|
| Sample size | 74 |
| Population | Healthy college participants |
| Interventions | SA-TA-NA-MA Hare Krishna Breath Focus |
| Duration | Six-week training period |
| Topics | Meditation |
| Keywords | Benchmark surveying Electroencephalography Preprocessor Suite |
| Key finding | The L-FAME dataset enables three classification tasks—cognitive state decoding, fine-grained meditation technique classification, and cross-session adaptation—with baseline results from classical and deep learning models. |
Abstract
We introduce a novel Longitudinal Focused Attention Meditation Electroencephalography (L-FAME) dataset and an accompanying benchmark, designed to foster research into the neural effects of various meditation practices and the evolution of these effects over a six-week training period. The dataset contains EEG recordings and psychological assessments from 74 healthy college participants, collected at two distinct time points: pre-intervention and post-intervention. Participants were randomly assigned to one of three distinct meditation groups: two mantra-based techniques (SA-TA-NA-MA and Hare Krishna) and one Breath Focus practice. Leveraging this unique longitudinal and comparative dataset, we propose a benchmark suite comprising three distinct classification tasks: (1) cognitive state decoding to distinguish between resting and meditation states, (2) fine-grained classification of the specific meditation techniques, and (3) cross-session adaptation to evaluate model generalization across the longitudinal time gap. We provide comprehensive baseline results for these tasks utilizing a range of classical machine learning algorithms and deep learning architectures. The complete dataset, preprocessing pipelines, and benchmark evaluation code will be publicly released, offering a valuable resource and a standardized framework for the development and comparison of new analytical methods in computational meditation research and EEG-based machine learning. The dataset is available at https://huggingface.co/datasets/L-FAME-Dataset-Benchmark/L-FAME