In this work, we study neural population dynamics of motor learning with @djoshea, @MattGolub_Neuro, @EricMTrautmann, @SaurabhsNeurons, Stephen Ryu, and @shenoystanford.
-
-
Prikaži ovu nit
-
Neural population dynamics have provided foundational insight into activity patterns and computational principles not readily apparent at single-neuron resolution. Recently, a dynamical system framework has begun to elucidate the neural basis of motor learning.
Prikaži ovu nit -
Collectively, previous studies have observed changes in neural population activity related to the learning process. A remaining challenge is to dissociate neural population dynamics tightly linked to movement parameters (e.g., kinematics and kinetics)…
Prikaži ovu nit -
…from other higher-level aspects of learning. How do changes in neural population dynamics reflect distinct aspects of learning during acquisition of a specific motor skill, as well as retention of multiple skills?
Prikaži ovu nit -
To address this question, we trained rhesus monkeys to learn a curl force field task that elicited new arm-movement kinetics for some (trained reach) but not all reach directions.pic.twitter.com/zsuyx6ityz
Prikaži ovu nit -
We recorded neural activity in dorsal premotor cortex (PMd) and primary motor cortex (M1) using Neuropixels probes, Utah arrays, and V-probes, which provided access to hundreds of neurons simultaneously or pooled over sessions.pic.twitter.com/e47rC7Tbiy
Prikaži ovu nit -
Along certain neural dimensions constructed by linear regression, preparatory states for learning-altered reaches rotate towards preparatory states of their adjacent reaches opposite to the curl field direction. Preparatory states reassociate existing patterns with new movements.pic.twitter.com/ddCiVKJyc0
Prikaži ovu nit -
We found unexpected prominent shifts of preparatory states along a nearly orthogonal neural dimension. These changes were observed uniformly for all reaches including those unaltered by learning. This uniform shift during learning implies formation of new neural activity patternspic.twitter.com/GdBD8MQT7k
Prikaži ovu nit -
During a washout period, movement kinetics gradually reverted, but the learning-induced uniform shift of preparatory activity persisted. This persistent shift may retain a motor memory of the learned field, consistent with faster relearning observed behaviorally and neurally.pic.twitter.com/AwMSaJPbtA
Prikaži ovu nit -
When leaning different curl fields sequentially, distinct uniform shifts occur, each reflecting the identity of the field applied and potentially separating motor memories. E.g., when learning opposite curl fields applied to the same target, preparatory states shift oppositely.pic.twitter.com/H409h5Dn52
Prikaži ovu nit -
Recent behavioral studies by
@HannahSheahan and colleagues have found that when people prepare for different movements associated with different curl fields, they can learn multiple skills without interference that would otherwise hamper learning.pic.twitter.com/w0eQBrGA5I
Prikaži ovu nit -
The uniform shift we found separates preparatory neural states (i.e., initial states) for seeding the local neural dynamics that would evolve in those regions of state space to produce distinct movements. This shift might thus reduce interference between multiple motor skills.pic.twitter.com/8ekZSFMWLV
Prikaži ovu nit -
In conclusion, the neural geometry of these uniform shifts in preparatory activity could serve to organize skill-specific changes in movement production, facilitating the acquisition and retention of a broad motor repertoire.
Prikaži ovu nit -
We thank everyone again for contributing to this work. Thanks to all the members of the Shenoy lab for comments and discussions throughout this project, and to
@MattAntimatt and@marius10p for spike sorting software. More results unpacked in the paper:http://biorxiv.org/content/10.1101/2020.01.30.919894v1 …Prikaži ovu nit
Kraj razgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.