Leveraging unlabelled data for generalizable neural population decoding

2026-07-15

AI Summary

This work addresses the limited generalization of existing supervised spiking neural decoding models under label scarcity by proposing MOJO, a novel framework that introduces masked autoencoding self-supervision into spike-tokenized models. MOJO jointly leverages self-supervised and supervised training to enable efficient few-shot fine-tuning. It employs spike-level tokenization and a masked autoencoder architecture, making it applicable across species, brain regions, and neural signal modalities. Evaluated on primate motor cortex, multi-region mouse recordings, and human electrocorticographic speech data, MOJO substantially outperforms purely supervised baselines—particularly with limited labels—approaching the performance of specialized neural foundation models while enhancing the interpretability of neural representations and cross-modal generalization capabilities.

Abstract

Robust and accurate neural decoders are integral to neurotechnologies such as brain-computer interfaces and closed-loop experiments. Recent work has shown that tokenizing neural data at the spike level facilitates multi-session pretraining and delivers state-of-the-art decoding performance. However, current spike-based models are restricted to supervised learning (SL), limiting training to datasets with paired behavioural labels. To address this limitation, we introduce MOJO (Masked autOencoder-based JOint training), a training framework for spike-tokenizing models that jointly leverages self-supervised learning (SSL) via masked autoencoding and SL objectives. We evaluate MOJO on three spiking datasets spanning monkey motor cortex during reaching tasks and multi-regional mouse recordings during vision and decision making tasks, demonstrating superior performance over purely SL-trained models. This improvement is especially pronounced when training with limited labelled data, particularly in few-shot finetuning, where only a small amount of labelled data from a new session is available. Incorporating SSL also yields more interpretable neuronal representations, improving performance on brain region classification and spike-statistics prediction without explicit optimization for these tasks. We further show that MOJO generalizes beyond spiking data to human electrocorticography during speech, where it continues to outperform purely SL-trained models and achieves performance comparable to neuro-foundation models (NFMs) designed specifically for continuous signals. Overall, augmenting spike-tokenizing models with SSL improves performance in label-impoverished settings and enables the use of unlabelled data across various tasks and species, while generalizing to other neural modalities. These results suggest a path towards more flexible and scalable data usage when training NFMs.

Problem

• neural decoding
• unlabelled data
• supervised learning
• generalization
• spike-tokenizing

Innovation

• self-supervised learning
• spike-tokenizing models
• masked autoencoding
• neural decoding
• foundation models