Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2015 Sep 18;13(9):e1002263.
doi: 10.1371/journal.pbio.1002263. eCollection 2015.

Sleep-Dependent Reactivation of Ensembles in Motor Cortex Promotes Skill Consolidation

Affiliations
Free PMC article

Sleep-Dependent Reactivation of Ensembles in Motor Cortex Promotes Skill Consolidation

Dhakshin S Ramanathan et al. PLoS Biol. .
Free PMC article

Abstract

Despite many prior studies demonstrating offline behavioral gains in motor skills after sleep, the underlying neural mechanisms remain poorly understood. To investigate the neurophysiological basis for offline gains, we performed single-unit recordings in motor cortex as rats learned a skilled upper-limb task. We found that sleep improved movement speed with preservation of accuracy. These offline improvements were linked to both replay of task-related ensembles during non-rapid eye movement (NREM) sleep and temporal shifts that more tightly bound motor cortical ensembles to movements; such offline gains and temporal shifts were not evident with sleep restriction. Interestingly, replay was linked to the coincidence of slow-wave events and bursts of spindle activity. Neurons that experienced the most consistent replay also underwent the most significant temporal shift and binding to the motor task. Significantly, replay and the associated performance gains after sleep only occurred when animals first learned the skill; continued practice during later stages of learning (i.e., after motor kinematics had stabilized) did not show evidence of replay. Our results highlight how replay of synchronous neural activity during sleep mediates large-scale neural plasticity and stabilizes kinematics during early motor learning.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1
Fig 1. Behavioral paradigm.
(A) Illustration of behavioral task [33], in which animals are required to reach through a narrow slit to grasp and retrieve a pellet into their cage. (B) Picture of animals grasping a pellet. Green line shows an example reach trajectory from which kinematic analyses are performed. (C) Overall behavioral/recording paradigm involved four blocks: a 2-h block of spontaneous recording (Sleep1), a block of skilled motor training (Reach1), another 2-h block of spontaneous recording (Sleep2), followed by a final reach block (Reach2).
Fig 2
Fig 2. Comparison of online and offline changes in skilled motor behavior.
Motor learning was characterized across two major parameters: speed and accuracy. (A) Speed was quantified for every trial by measuring the time from reach onset to retraction of forelimb. (B) Unlike changes in speed, animals showed a general improvement in accuracy across the learning block (Reach1), but accuracy was not significantly different after sleep (Reach2). For both plots we used a moving average window of ten trials. (C,D) Effects were quantified across all animals (n = 5). Error bars represent mean ± SEM; *** p < 0.001.
Fig 3
Fig 3. Comparison of online and offline changes in neural activation.
(A) Change in movement-related activation of a single neuron during Reach1early, Reach1late, and Reach2early. Dotted line is the time of reach onset. Traces below represent a Bayesian adaptive regression spline fit of the respective PETH. (B) Curves show respective cumulative distributions of the single neuron PETH time to peak (red dot represents median timing). While there was not a significant change in the distribution for online learning (Kolmogorov-Smirnoff test, p = 0.9 comparing Reach1early versus Reach1late, there was a significant shift after sleep (p < 0.001 comparing Reach2early to both Reach1early and Reach1late, n = 4 animals, 99 units). (C) Change in task-related neural modulation. Task related modulation index is ratio of baseline firing to the peak instantaneous firing rate n = 4 animals, 99 units). Error bars show standard error of the mean (S.E.M.) * p < 0.05, *** p < 0.001.
Fig 4
Fig 4. Sleep restriction prevents offline gains and temporal shifts.
(A) Representative example of lack of improvement in speed after sleep-restriction (moving window average of 10 trials). (B) Comparison of effects of sleep restriction versus sleep (n = 5 each) on changes in motor speed and “neural speed” (i.e., shift in PETH peak; n = 99 neurons from sleep animals, n = 80 neurons from sleep-restricted animals). Error bars show S.E.M. *** p < 0.001 (C) Cumulative distribution of single neuron peak of PETH (non-significant based on Kolmogorov Smirnoff test).
Fig 5
Fig 5. Reactivation of task-related neural ensembles during slow-wave sleep.
(A) Example of reactivation events prior to and after learning (i.e., Sleep1 versus Sleep2 events in respective blue and red boxes). Also shown are the activation strengths of reactivation events during the initial NREM epochs from Sleep1 and Sleep2. (B) Across all animals, there was a significant reactivation of task-related ensembles (p < 0.001, sign-rank test). Quantification was based on the entire recorded NREM sleep. (C) Linear correlation between single neuron reactivations and neural modulation during Reach2. To estimate “single neuron reactivation,” we first calculated the overall ensemble reactivation [(mean activation during NREM from Sleep2)–(mean activation during NREM from Sleep1)] and then multiplied this value with the principle component weight for each neuron. Plot shows the regression analysis between each neuron’s reactivation and subsequent temporal shift (r = -0.41, p < 0.001). Error bars show S.E.M. * p < 0.05, ** p < 0.01,*** p < 0.001.
Fig 6
Fig 6. Relationship of ensemble reactivation to NREM oscillations.
(A) Event-triggered z-scored LFP for fast-spindle oscillatory frequencies filtered at 13–16 Hz. Post-hoc differences (p < 0.05) indicated by points above graph. (B) Event-triggered LFP for slow-spindle frequencies (9–12 Hz). There were no significant differences across any time point. (C) Event-triggered LFP for slow oscillation frequencies (0.5–4 Hz). (D) Comparison of the coefficient of variation (CV) for each oscillatory band. For fast spindles and slow oscillations, for each point that was significantly different between the two groups, we compared the CV. After learning, the CV was significantly reduced, suggesting more consistent temporal locking after learning. For slow-spindle oscillations, because no points were different between the two groups, we instead used the points that were significantly different in the fast-spindle frequency. (E) Comparison of the z-scored temporal coupling between reactivation-triggered slow wave and fast-spindle oscillation before and after learning. (F) Changes in mean event-triggered spindle amplitude. We calculated the instantaneous analytic amplitude of the fast spindles at the slow-wave trough for each reactivation. There was a significant increase in spindle amplitude. (G) Increase in the probability that a reactivation event was associated with a fast-spindle oscillation. Error bars show S.E.M. * p < 0.05, ** p < 0.01,*** p < 0.001.
Fig 7
Fig 7. Lack of increased reactivation with continued motor learning.
(A) We performed the same experiment as described for day 1 (Sleep1/Reach1/Sleep2/Reach2) on subsequent days to assess whether there was evidence of continued reactivation on these days. (B) Movement speed trends during Reach1 and Reach2 on day 2 for one animal. (C) Comparison of respective reactivations for subsequent training days (sign-rank p = 0.4). (D) Comparison of reactivations on day 1 versus subsequent days grouped into deciles. Mean differences were compared for each decile (*** p < 0.001). We found significant differences between groups at every decile, though clearly the greatest difference occurs at the “top” decile of reactivation events. (E) Correlation between top ten percentile reactivation events and offline changes in movement speed. We found a significant correlation (Spearman’s Rho = -0.76, p < 0.05) in which reactivation predicted offline motor improvements. Error bars show S.E.M. * p < 0.05, ** p < 0.01,*** p < 0.001.

Comment in

  • To Replay, Perchance to Consolidate.
    Genzel L, Robertson EM. Genzel L, et al. PLoS Biol. 2015 Oct 23;13(10):e1002285. doi: 10.1371/journal.pbio.1002285. eCollection 2015 Oct. PLoS Biol. 2015. PMID: 26496145 Free PMC article.

Similar articles

See all similar articles

Cited by 34 articles

See all "Cited by" articles

References

    1. Fitts PM. Cognitive aspects of information processing: III. Set for speed versus accuracy. J Exp Psychol. 1966;71(6):849–57. - PubMed
    1. Fitts PM, Peterson JR. Information capacity of discrete motor responses. J Exp Psychol. 1964. February;67:103–12. - PubMed
    1. Shmuelof L, Krakauer JW. Are We Ready for a Natural History of Motor Learning? Neuron. 2011. March 11;72(3):469–76. 10.1016/j.neuron.2011.10.017 - DOI - PMC - PubMed
    1. Adams JA. Historical review and appraisal of research on the learning, retention, and transfer of human motor skills. Psychol Bull. 1987;101(1):41–74.
    1. Dayan E, Cohen LG. Neuroplasticity Subserving Motor Skill Learning. Neuron. 2011. November 3;72(3):443–54. 10.1016/j.neuron.2011.10.008 - DOI - PMC - PubMed

Publication types

Grant support

This work was supported by fellowship awards from the Department of Veterans Affairs, Veterans Health Administration (http://www.va.gov/) to DSR and from the American Heart Association (http://www.heart.org/HEARTORG/) to TG. This work was also supported by awards to KG from the Department of Veterans Affairs, Veterans Health Administration (http://www.va.gov/, CDA B6674W); from the Northern California Institute for Research and Education (NCIRE, www.ncire.org) and start-up funds from the UCSF Department of Neurology (http://neurology.ucsf.edu). KG also holds a Career Award for Medical Scientists from the Burroughs Wellcome Fund (http://www.bwfund.org/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Feedback