Brain Signals: Learning to Understand Without a Teacher express
The brain communicates in short electrical clicks. Previously, to learn this language, thousands of examples with translations from the experimenter were needed. MOJO takes a different path: like learning a language through immersion, it first listens to the 'stream of consciousness' without any hints and finds recurring patterns on its own. Then, given a tiny dictionary, samples of signals linked to movements or words, it turns indistinct noise into clear commands.
Experiments on monkeys and mice showed that even one or two hints dramatically improve the results, outperforming classical algorithms hungry for data.
This cascade reduces uncertainty and allows working with any signals — from single neurons to large groups. It also clarifies the workings of living networks made of carbon. And since self-learning doesn't require expensive labels, vast accumulated archives of brain recordings can be put to use — bringing closer the era of assistive devices for people with paralysis. The idea is not new: as early as John von Neumann pondered the kinship of mind and machine.
Original abstract
as written by the authors arXiv:2607.14086
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.
In a second of recording from a hundred neurons, up to 20,000 electrical discharges flash by — a true orchestra where each impulse plays its own note.
Technology for directly reading motor intentions resembles the neural interfaces from the movie 'Avatar,' where a human mind controls the avatar's body.