Register to compete in the BIOMAG 2026 Le Petit Prince (LPP) Decoding Challenge! Work with naturalistic MEG and fMRI data to decode brain activity, with a focus on cross-modality transfer and robust generalization across subjects.
Register Now Learn More
Explore cutting-edge brain data analysis and advance neuroscience
The BIOMAG 2026 Data Challenge is an official satellite event of the BIOMAG 2026 conference in Beijing, China. The challenge aims to promote reproducible and generalizable methods at the intersection of natural language processing, cognitive neuroscience, and brain decoding.
Recent advances in large language models (LLMs) have opened new frontiers for understanding how the brain processes language. By aligning neural representations with computational language models, researchers can now decode semantic content from brain activity with unprecedented accuracy—making this an exciting time to explore cross-modal and cross-subject brain decoding.
Task: Naturalistic listening of Chinese version of The Little Prince
Dataset 1 — fMRI: Part of the publicly available Le Petit Prince multilingual naturalistic fMRI corpus (Li et al., 2022).
Li, J., Bhattasali, S., Zhang, S., Franzluebbers, B., Luh, W.-M., Spreng, R. N., Brennan, J. R., Yang, Y., Pallier, C., & Hale, J. (2022). Le Petit Prince multilingual naturalistic fMRI corpus. Scientific Data, 9(1), 530. [doi]
Dataset 2 — MEG: First-time release for data challenge! Our newly collected MEG data (Li's Lab) uses exactly the same Chinese stimulus as fMRI, enabling direct cross-modality alignment. Chinese as a tonal language also offers unique challenges and opportunities for neural decoding.
Lab Website: LAnguages, Machines & Brains Lab
| Scanner | 3T GE Discovery MR750 |
| Head Coil | 32-channel |
| Participants | 35 (15 female), Mean age: 19.9 |
| Stimuli | The Little Prince CN audiobook |
| Duration | 5954s (~99 min) |
| TR / TEs | 2000 ms / 2.8, 27.5, 43 ms |
| Resolution | 3.75 × 3.75 × 3.8 mm |
| System | TRIUX™ neo 306-channel |
| Channels | 102 mag + 204 grad |
| Participants | 32 (16 female), Mean age: 26.6 |
| Stimuli | The Little Prince CN audiobook |
| Duration | 5954s (~99 min) |
| Sampling Rate | 100 Hz |
| EOG / HPI | 2 channels / 5 coils |
Advance decoding approaches that extract meaningful information from MEG and fMRI signals under realistic experimental conditions. Emphasis is placed on methods that generalize across individuals and are robust to variability in neural responses.
By providing MEG and fMRI data collected with identical stimuli and timing, the challenge promotes methods that align, transfer, or jointly model neural representations across modalities. Leverage the complementary temporal and spatial properties of MEG and fMRI.
Using continuous narrative stimuli as a shared paradigm, the competition provides a standardized benchmark for evaluating decoding methods on complex, time-varying brain data. The focus remains on methodological generality rather than task-specific tuning.
Save the Key dates for the BIOMAG 2026 Data Challenge
Follow these steps to get started with the challenge
Create your account and complete the registration form
Access your personal challenge portal
Get MEG and fMRI data bundles from your portal
Submit your draft and final reports through the portal
How submissions will be scored (Total: 100 points)
Demonstrate a clear rationale for the chosen decoding method.
Use appropriate statistical tests to evaluate the significance of the findings.
Clearly interpret the outcomes in the context of cross-subject decoding. Discuss any unexpected findings.
Compare decoding results with fMRI and MEG.
Present a well-organized and clear report. Include relevant figures and replication-friendly code with clear comments.
Describe measures taken to address reliability concerns. Replicate findings on the second half of the dataset.
Recognition for the top performing teams
Submission requirements and guidelines
Teams are required to submit a written report supplemented by figures and codes (max 8 pages). The report should comprehensively address all the evaluation criteria. The provided code should be replication-friendly and with clear comments.
Make explicit how concerns on overfitting and reliability were addressed (e.g., consider developing measures on one part of the data set and test on the second part). Address potential concerns complicating the interpretation (e.g., field spread and event-related fields). At the end of the challenge, all submitted materials will be made public.
Meet the team behind BIOMAG 2026 LPP Decoding Challenge
City University of Hong Kong
City University of Hong Kong
City University of Hong Kong
City University of Hong Kong
City University of Hong Kong
Beijing Language and Culture University, local committee of BIOMAG2026
The Hong Kong Polytechnic University
Fudan University
Fudan University
Fudan University
Join researchers worldwide in this exciting data challenge
Register Now