New neural activity patterns emerge with long-term learning

Oby, Emily R.; Golub, Matthew D.; Hennig, Jay A.; Degenhart, Alan D.; Tyler-Kabara, Elizabeth C.; Yu, Byron M.; Chase, Steven M.; Batista, Aaron P.

doi:10.1073/pnas.1820296116

Cited by 168 publications

(209 citation statements)

References 37 publications

(53 reference statements)

Supporting

Mentioning

198

Contrasting

Order By: Relevance

“…during VMR learning, and suggests that reusing existing activity patterns may be a common strategy at least partially benefiting different learning contexts [6][7][8]38 .…”

Section: Introductionmentioning

confidence: 99%

Skill-specific changes in cortical preparatory activity during motor learning

Sun

Golub

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Animals have a remarkable capacity to learn new motor skills, but it remains an open question as to how learning changes neural population dynamics underlying movement 1 . Specifically, we asked whether changes in neural population dynamics relate purely to newly learned movements or if additional patterns are generated that facilitate learning without matching motor output. We trained rhesus monkeys to learn a curl force field 2 task that elicited new arm-movement kinetics for some but not all reach directions 3,4 . We found that along certain neural dimensions, preparatory activity in motor cortex reassociated existing activity patterns with new movements. These systematic changes were observed only for learning-altered reaches. Surprisingly, we also found prominent shifts of preparatory activity along a nearly orthogonal neural dimension. These changes in preparatory activity were observed uniformly for all reaches including those unaltered by learning. This uniform shift during learning implies formation of new neural activity patterns, which was not observed in other short-term learning contexts 5-8 . During a washout period when the curl field was removed, movement kinetics gradually reverted, but the learning-induced uniform shift of preparatory activity persisted and a second, orthogonal uniform shift occurred. This persistent shift may retain a motor memory of the learned field 9-11 , consistent with faster relearning of the same curl field observed behaviorally and neurally. When multiple different curl fields were learned sequentially, we found distinct uniform shifts, each reflecting the identity of the field applied and potentially separating the associated motor memories 12,13 . The neural geometry of these shifts in preparatory activity could serve to organize skill-specific changes in movement production, facilitating the acquisition and retention of a broad motor repertoire. Motor learning encompasses a wide range of phenomena, from relatively low-level mechanisms for calibrating movement parameters, to making high-level cognitive decisions about how to act in a novel environment 1 . Motor adaptation has been a long-standing and widely used paradigm for studying motor learning. Decades of behavioral studies have demonstrated many intriguing phenomena during motor adaptation, such as the error-driven calibration of movements, generalization of learned skills to a new context, savings (faster relearning) or memory retention, and interference between learning multiple skills 3,4,12,[14][15][16][17][18] . Yet their neural mechanisms, in particular the underlying neural population dynamics, remain largely unknown.In the field of motor control, neural population dynamics have provided foundational insight into activity patterns and computational principles not readily apparent at single-neuron resolution 19,20 . Recently, a dynamical system framework has started to help elucidate the neural basis of motor learning 5,8,21-23 . Collectively, these experiments have observed changes in neural population st...

show abstract

“…during VMR learning, and suggests that reusing existing activity patterns may be a common strategy at least partially benefiting different learning contexts [6][7][8]38 .…”

Section: Introductionmentioning

confidence: 99%

Skill-specific changes in cortical preparatory activity during motor learning

Sun

Golub

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Besides the original experimental study showing that monkeys can not adapt to outsidemanifold perturbations on the timescale of a single experiment [Sadtler et al, 2014], there is new evidence showing that outside-manifold learning is indeed possible if the training lasts for several days and follows a specific incremental protocol [Oby et al, 2019]. We have shown so far that outside-manifold learning should not be possible, as the correct feedback weights needed for successful retraining can not be learned.…”

Section: Feedback Signal For Outside-manifold Perturbation Can Potentmentioning

confidence: 80%

“…The second scenario is rewiring in the local circuit. In this study we focused on the latter and tested to what extent it can explain experimental observations [Sadtler et al, 2014, Oby et al, 2019.…”

Section: Discussionmentioning

confidence: 99%

“…Interestingly, the timescale of within-manifold learning is in the order of minutes to hours [Sadtler et al, 2014]. In contrast, outside-manifold learning is not possible within a single session, but instead requires progressive training over multiple days [Oby et al, 2019]. Due to the timescale difference, it has been hypothesized that within-manifold learning is related to fast motor adaption and outside-manifold learning is related to skill learning.…”

Section: Introductionmentioning

confidence: 99%

“…Furthermore, learning the feedback signal from scratch was only possible for within-manifold perturbations. To enable feedback learning for outside-manifold perturbations we had to introduce an incremental strategy, which resembles the progressive training used in experiments [Oby et al, 2019].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Neural manifold under plasticity in a goal driven learning behaviour

Feulner

Clopath

2020

Preprint

View full text Add to dashboard Cite

Neural activity is often low dimensional and dominated by only a few prominent neural covariation patterns. It has been hypothesised that these covariation patterns could form the building blocks used for fast and flexible motor control. Supporting this idea, recent experiments have shown that monkeys can learn to adapt their neural activity in motor cortex on a timescale of minutes, given that the change lies within the original low-dimensional subspace, also called neural manifold. However, the neural mechanism underlying this within-manifold adaptation remains unknown. Here, we show in a computational model that modification of recurrent weights, driven by a learned feedback signal, can account for the observed behavioural difference between within-and outside-manifold learning. Our findings give a new perspective, showing that recurrent weight changes do not necessarily lead to change in the neural manifold. On the contrary, successful learning is naturally constrained to a common subspace.

show abstract

Deciphering the Neuronal Population Code

Khanna

Smith

2020

Neural Engineering

View full text Add to dashboard Cite

New neural activity patterns emerge with long-term learning

Cited by 168 publications

References 37 publications

Skill-specific changes in cortical preparatory activity during motor learning

Skill-specific changes in cortical preparatory activity during motor learning

Neural manifold under plasticity in a goal driven learning behaviour

Deciphering the Neuronal Population Code

Contact Info

Product

Resources

About