Using large-scale EEG datasets and cutting edge ML research, we’re building models that decode images, emotion, and eventually complex cognitive processes.
Our research is broad: spanning from semantic understanding and abstract reasoning to inner speech and intentional planning.
2026
ENIGMA is a new multi-subject decoding model that reconstructs seen images from EEG signals with state-of-the-art accuracy. By learning a shared “visual language” across different brains, the model can adapt to a new user in minutes rather than hours, bringing practical Brain-Computer Interfaces (BCIs) closer to reality.
2025
We introduce the largest public dataset for EEG-image decoding: 6+ hours of recordings per participant across 20 subjects, collected using consumer-grade hardware. We model neural scaling laws and demonstrate log-linear scaling.
2024
MindEye2 introduces a multi-subject fMRI-to-image architecture that trains 40× faster than previous state-of-the-art while matching performance. Accepted to ICML.
2025
The first and only dataset for complex mental imagery reconstruction from fMRI. Accepted to CVPR (Highlight paper).
Using solely evoked EEG responses, our models reconstruct the original stimulus, leading performance across various datasets.
San Francisco
San Francisco
San Francisco
San Francisco
San Francisco
