The abstract is read at the publisher; the summary is JClub's.
A multimodal digital twin trained on freely moving mouse V1 predicts spiking and recovers canonical tuning, including orientation, spatial frequency, and contrast differences between putative excitatory and inhibitory cells.
- Why it matters: Classical head-fixed studies miss how active movement reshapes visual input and couples it to behavior, leaving a gap between controlled physiology and natural active vision.
- What they did: The model learned to forecast V1 spiking from retinocentric video plus locomotion, pupil, and eye/head motion, then was frozen and probed in silico under controlled stimulus conditions.
- The result: The learned representation generalized across stimulus classes, animals, and recording regimes, transferred to new mice from a few minutes of data, matched head-fixed natural-movie geometry, and showed locomotion and pupil dilation share a gain increase while only locomotion shifts sensitivity to high spatial frequencies.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.