OmniMouse: Scaling properties of multi-modal, multi-task Brain Models on 150B Neural Tokens

arXiv cs.AI / 4/22/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The OmniMouse study trains multi-modal, multi-task “brain models” on a large neural dataset (3.1M neurons from 73 mice, 323 sessions) totaling over 150B neural tokens recorded during natural and controlled stimuli plus behavior.
The models can flexibly perform three tasks at test time—neural prediction, behavioral decoding, and neural forecasting—individually or in combination.
OmniMouse achieves state-of-the-art results, outperforming specialized baselines across nearly all evaluation regimes for these neural modeling tasks.
The paper finds scaling is reliable with more data, but performance gains from larger model size saturate, suggesting brain modeling may be data-limited even with very large recordings.
The authors propose that the observed scaling behavior could indicate phase transitions in neural modeling, where richer and larger datasets might produce qualitatively new capabilities akin to emergent effects in large language models.

Abstract

Scaling data and artificial neural networks has transformed AI, driving breakthroughs in language and vision. Whether similar principles apply to modeling brain activity remains unclear. Here we leveraged a dataset of 3.1 million neurons from the visual cortex of 73 mice across 323 sessions, totaling more than 150 billion neural tokens recorded during natural movies, images and parametric stimuli, and behavior. We train multi-modal, multi-task models that support three regimes flexibly at test time: neural prediction, behavioral decoding, neural forecasting, or any combination of the three. OmniMouse achieves state-of-the-art performance, outperforming specialized baselines across nearly all evaluation regimes. We find that performance scales reliably with more data, but gains from increasing model size saturate. This inverts the standard AI scaling story: in language and computer vision, massive datasets make parameter scaling the primary driver of progress, whereas in brain modeling -- even in the mouse visual cortex, a relatively simple system -- models remain data-limited despite vast recordings. The observation of systematic scaling raises the possibility of phase transitions in neural modeling, where larger and richer datasets might unlock qualitatively new capabilities, paralleling the emergent properties seen in large language models. Code available at https://github.com/enigma-brain/omnimouse.