Pretraining across many mice sometimes helps but gains vary by task
A large collection of brain data from many mice shows pretraining can help transfer to new animals but results depend on the task and data alignment.
- Publication
- arXiv
- Stage
- Preprint
- What we read
- Summary of the paper
- Authors
- Alexandre Andre, Shivashriganesh P. Mahato, Vinam Arora, Keshav Balaji, Divyansha Lachi, Nanda H. Krishna, Jingyun Xiao, Yizi Zhang, Ximeng Mao, Wenrui Ma, Han Yu, International Brain Laboratory, Daniel Birman, Niccolò Bonacchi, Gaelle A. Chapuis, Joana A. Catarino, Felicia Davatolhagh, Mayo Faulkner, Laura Freitas-Silva, Fei Hu, Julia M. Huntenburg, Anup Khanal, Inês Laranjeira, Petrina Lau, Guido T. Meijer, Nathaniel J. Miska, Jean-Paul Noel, Alejandro Pan-Vazquez, Georg Raiser, Cyrille Rossant, Karolina Z. Socha, Anne E. Urai, Miles J. Wells, Steven J. West, Olivier Winter, Blake Richards, Guillaume Lajoie, Cole Hurwitz, Mehdi Azabou, Matthew R. Whiteway, Liam Paninski, Eva L. Dyer
- Universities and research institutions
- University of Pennsylvania, Mila - Quebec Artificial Intelligence Institute, Université de Montréal, Stanford University, Columbia University, Allen Institute, ISPA - Instituto Universitário, William James Center for Research
Institution metadata: OpenAlex record ↗
What they did and found
Researchers collected brain recordings from 139 mice across 276 brain regions. They used 126 mice to train a general model and tested 13 mice left out on three task groups. Pretraining often helped compared with single-session use, but no method was best for all tasks.
Why it matters
Using large shared brain data can speed up learning for new animals, but results are uneven and you should tailor the approach to your task.
The work shows training on data from many animals can help new animals, but gains vary by task and by method used. What task are you aiming to improve?
Also, matching training data to the goal matters: some gains come from extra filtering and setup, and some approaches do not beat simple single-session baselines.
Original sources · 1
- BrainWideBench: Benchmarking large-scale pretraining and across-animal transfer in multi-region neural recordings ↗arXiv · 2026-09-18
Check the original paper for its authors, methods, version and access terms.
What remains uncertain
The study looked only at cross-animal data from one species and one measurement type; results depend on which animals were left out and how well training data match the task; no method is best for all tasks.
What is your take?
Ask a question, add useful context or share a different perspective. Keep the conversation respectful and grounded.