Lead Scientists:
Jason Keller (Google Scholar, ORCID)
Josh Dudman (Google Scholar, ORCID)
https://www.biorxiv.org/content/10.1101/2025.02.12.637930v1
CtxCtrlFromSubctxDemo_Rev1.pdf
blueprint thread: https://bsky.app/profile/neurojak.bsky.social/post/3li3fnfvcik2r
In mammals, neocortex plays a critical role in many forms of motor learning. The high dimensionality of the musculoskeletal system allows flexibility but poses significant challenges for efficient learning. A key computational insight is to exploit pre-existing motor patterns to simplify the learning problem, but whether and how neocortex can utilize similar principles for efficient learning remains an open question. To address this, we developed an active avoidance paradigm in which head-fixed mice “jump” by extending against a counterweighted platform. Naïve mice use a pre-existing jump motor pattern to escape a cold platform and can rapidly learn to jump predictively to avoid the cold. Optogenetic inhibition of frontal cortex blocked active avoidance, but left reactive escape jumps intact. Brainwide recordings revealed that persistent, movement-null activity in prefrontal cortex was critical for avoidance and coupled through motor cortex to control predictive jumping. Remarkably, mice covertly learned active avoidance even without trial-and-error exploration of cortically-mediated jumps. We propose that subcortically-mediated movements can ‘demonstrate’ successful motor patterns to enable efficient cortical learning.
SupplVid1_avoidReactExtensions.mp4
SupplVid2_localCtxInhibition.mp4
SupplVid3_acuteDecortication.mp4
FigShare link & DOI coming soon
https://cad.onshape.com (search: “jumpLever” for rig or “miniRIVETS2” for headplates)
video capture: https://github.com/neurojak/pySpinCapture
LabVIEW behavior control: https://github.com/neurojak/jfcRigControl
MATLAB analysis code will be bundled with data and metadata in FigShare above, but for now is available at:
https://github.com/neurojak/CtxCtrlFromSubCtxDemo_AnalysisCode