The 2026 PNPL Competition focuses on a single, ambitious task: Word Classification. Given a segment of MEG brain data recorded while a subject listens to spoken English, predict the specific word being heard.
Word Classification
Build a model that maps MEG recordings to the specific word being heard by the participant. This is a significant step up from last year's phoneme-level classification, requiring models to capture richer temporal and semantic information from brain signals.
Evaluation metric and vocabulary details will be announced with the dataset release.
The Dataset: LibriBrain100
This year's competition uses LibriBrain100, a major expansion of the original LibriBrain dataset to over 100 hours total:
Deep single-subject data
~80 hours of recordings from the original LibriBrain subject, expanded from last year's ~50 hours.
32 new subjects × ~40 minutes each
MEG data from 32 additional subjects, each contributing ~40 minutes of Sherlock Holmes audiobook listening. Enables research on cross-subject generalisation.
Full task specification, evaluation metrics, and baseline models will be released on July 1, 2026.