Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Sep 11:6:e24987.
doi: 10.7554/eLife.24987.

Network-wide reorganization of procedural memory during NREM sleep revealed by fMRI

Shahabeddin Vahdat  1 , Stuart Fogel  2 , Habib Benali  3   4 , Julien Doyon  1
Affiliations

Network-wide reorganization of procedural memory during NREM sleep revealed by fMRI

Shahabeddin Vahdat et al. Elife. .

Abstract

Sleep is necessary for the optimal consolidation of newly acquired procedural memories. However, the mechanisms by which motor memory traces develop during sleep remain controversial in humans, as this process has been mainly investigated indirectly by comparing pre- and post-sleep conditions. Here, we used functional magnetic resonance imaging and electroencephalography during sleep following motor sequence learning to investigate how newly-formed memory traces evolve dynamically over time. We provide direct evidence for transient reactivation followed by downscaling of functional connectivity in a cortically-dominant pattern formed during learning, as well as gradual reorganization of this representation toward a subcortically-dominant consolidated trace during non-rapid eye movement (NREM) sleep. Importantly, the putamen functional connectivity within the consolidated network during NREM sleep was related to overnight behavioral gains. Our results demonstrate that NREM sleep is necessary for two complementary processes: the restoration and reorganization of newly-learned information during sleep, which underlie human motor memory consolidation.

Keywords: consolidation; fMRI; human; motor learning; neuroplasticity; neuroscience; sleep; striatum.

PubMed Disclaimer

Conflict of interest statement

The authors declare that no competing interests exist.

Figures

Figure 1.
Figure 1.. Experimental design, behavioral performance, and task activation maps.
(A) Experimental procedure. On Day 1, subjects first experienced a screening and adaptation night in the mock scanner, which mimicked conditions experienced in both experimental and control nights. Subjects returned and underwent fMRI scans (Day seven and Day 14) while training (S1) on either the motor skill learning (MSL) or motor control (CTL) task in a counterbalanced order, interleaved by resting-state conditions (RS1 and RS2). This was followed by simultaneous EEG-fMRI sleep recording for up to ~2.5 hr. Subjects were then allowed to sleep for the remainder of the night in the sleep lab. Finally, on the following morning subjects underwent retest fMRI sessions (S2) on the same take as the previous training session (Day eight and Day 15), interleaved by resting-state conditions (RS3 and RS4). Arrows shows the experiment’s timeline. (B) Performance speeds (i.e., inter-key interval) averaged across all subjects show that the learning curves differed between the MSL (red) and CTL (blue) conditions during the learning session (S1). (C) Only the MSL task was consolidated overnight, as indicated by performance gain averaged across subjects (asymptotic performance at the end of S1 compared to the beginning of S2). (D) Color-coded activation maps representing motor sequence-related areas during the learning (S1) and retest (S2) sessions (corrected for multiple comparisons using Gaussian random field theory, cluster level threshold p<0.05). Bar plots illustrate the volume of cortical and subcortical activation in each map. As expected, the connectivity index (CI) within the learning (E) and the consolidated (F) patterns was significantly higher in the MSL compared to the CTL condition. Error bars represent s.e.m.; ** and *** indicate p<0.01 and p<0.001, respectively. DOI: http://dx.doi.org/10.7554/eLife.24987.003
Figure 1—figure supplement 1.
Figure 1—figure supplement 1.. Task-related activation maps during the learning session (S1).
Figure shows the results of block-design analysis related to the MSL (color-coded in red) and CTL tasks (color-coded in blue). Color-coded activation maps indicate Z-score values and are corrected for multiple comparisons using GRF, p<0.05. DOI: http://dx.doi.org/10.7554/eLife.24987.006
Figure 1—figure supplement 2.
Figure 1—figure supplement 2.. Differences in the activation of motor sequence-related areas between the learning (S1) and retest (S2) sessions.
Only subcortical areas including bilateral putamen and cerebellar cortex (lobules V-VI) revealed significantly greater activation during the retest compared to the learning session (contrast: retest [MSL-CTL] – learning [MSL-CTL]; top row), while two cortical clusters, including the superior parietal lobule and anterior intraparietal sulcus bilaterally, showed significantly greater activation during the learning compared to the retest session (contrast: learning [MSL-CTL] – retest [MSL-CTL];bottom row). Color-coded activation maps indicate Z-score values and are corrected for multiple comparisons using GRF, p<0.05. DOI: http://dx.doi.org/10.7554/eLife.24987.007
Figure 2.
Figure 2.. Connectivity index (CI) during resting-state and NREM sleep.
(A, B) show the CI within the learning and the consolidated patterns, respectively, averaged across subjects during the resting-state conditions before (RS1) and after (RS2) the training session (S1), as well as on the following morning before retest (RS3). The learning pattern’s CI was significantly increased in the MSL immediately following training (RS2), while the consolidated pattern’s CI increased significantly only post-sleep (RS3) in the MSL as compared to the CTL condition. (C, D) show, respectively, CI within the learning and the consolidated patterns averaged across subjects during NREM sleep. Only the consolidated pattern’s CI differed significantly between the MSL and CTL nights. Error bars represent s.e.m.; * and ** indicate p<0.05 and p<0.01, respectively. DOI: http://dx.doi.org/10.7554/eLife.24987.008
Figure 2—figure supplement 1.
Figure 2—figure supplement 1.. The group-level spatial maps of four highly-reproducible brain networks extracted during the MSL task.
Each row shows a spatial map extracted from the application of independent component analysis (ICA) on the time-concatenated data of the learning (S1) and retest (S2) sessions during the MSL task across all subjects. The label for each component is provided on the left. Color-coded activation maps indicate Z-score values. DOI: http://dx.doi.org/10.7554/eLife.24987.009
Figure 2—figure supplement 2.
Figure 2—figure supplement 2.. Connectivity index (CI) during resting-state periods within the four control networks reported in Figure 2—figure supplement 1.
(a–d) show the CI within the default mode, visual, left and right fronto-parietal networks, respectively, averaged across subjects during the resting-state conditions before (RS1) and after (RS2) the training session (S1), as well as on the following morning before retest (RS3). No significant difference in CI between MSL and CTL tasks was observed in any of the four control networks (p>0.2, in all cases). Error bars represent s.e.m.; n.s. indicates not significant interaction. DOI: http://dx.doi.org/10.7554/eLife.24987.010
Figure 2—figure supplement 3.
Figure 2—figure supplement 3.. Connectivity index (CI) during non-REM (NREM) sleep within the four control networks reported in Figure 2—figure supplement 1.
(a–d) show the CI within the default mode, visual, left and right fronto-parietal networks, respectively, averaged across subjects during NREM sleep. No significant difference in CI between MSL and CTL tasks was observed in any of the four control networks (p>0.3, in all cases). Error bars represent s.e.m.; n.s. indicates not significant paired t-statistics. DOI: http://dx.doi.org/10.7554/eLife.24987.011
Figure 3.
Figure 3.. Neural correlates of motor sequence memory consolidation during NREM sleep.
The ventrolateral putamen functional connectivity within the consolidated pattern differed significantly between the MSL and CTL nights during NREM sleep. Bar plot illustrates the functional connectivity of the putamen with the rest of structures in the consolidated pattern averaged across subjects during NREM sleep. Importantly, the putamen connectivity within the consolidated pattern during NREM sleep was significantly related to the extent of sleep-dependent performance speed gains on a per subject basis, as depicted in the scatter plot. The color-coded activation map indicates Z-score values and is corrected for multiple comparisons using GRF, p<0.05. Error bars represent s.e.m.; * indicates p<0.05. DOI: http://dx.doi.org/10.7554/eLife.24987.012
Figure 3—figure supplement 1.
Figure 3—figure supplement 1.. Changes in brain functional connectivity related to motor sequence learning during non-REM sleep.
Two ROIs within the consolidated pattern, including the cerebellar cortex (top row) and the putamen (bottom row), showed significant changes in functional connectivity between tasks during non-REM sleep (paired t test (MSL−CTL), corrected for multiple comparisons using GRF, p<0.05). In each row, the left and middle brain slices show the locations of each ROI and the clusters showing significant change in connectivity with that ROI, respectively. The color-coded activation map indicates Z-score values. Each bar plot displays the average values of functional connectivity across subjects during the MSL and CTL nights between the corresponding ROI and the specified clusters. The scatter plots illustrate the linear trend between individual changes in functional connectivity (MSL−CTL) and the associated offline gains in performance speed on a per subject basis. The value of r represents the Pearson correlation coefficient. DOI: http://dx.doi.org/10.7554/eLife.24987.014
Figure 4.
Figure 4.. Neural correlates of motor sequence memory consolidation during post-sleep resting-state periods.
The ventrolateral putamen (A) and the cerebellar cortex (lobules V-VI) (B) functional connectivity within the consolidated pattern differed significantly between the MSL and CTL conditions during post-sleep resting-state periods (RS3) as compared to baseline (RS1). Bar plot illustrates the change in functional connectivity of putamen within the consolidated pattern between RS3 and RS1 scans averaged across subjects in each task condition. The scatter plot in (A) shows that only the putamen functional connectivity was significantly related to the extent of overnight behavioral gains in performance speed. The color-coded activations maps indicate Z-score values and are corrected for multiple comparisons using GRF, p<0.05. Error bars represent s.e.m.; * and ** indicate p<0.05 and p<0.01, respectively. DOI: http://dx.doi.org/10.7554/eLife.24987.015
Figure 4—figure supplement 1.
Figure 4—figure supplement 1.. Functional connectivity within the consolidated pattern during post-sleep resting-state periods (RS3) and baseline (RS1) in each task.
(A and B) illustrate, respectively, the average functional connectivity values of the putamen and cerebellum clusters (see Figure 4) within the consolidated pattern during RS3 and RS1 scans in each task condition. Error bars represent s.e.m.; * indicates significant (p<0.05) interaction between task and resting-state condition. DOI: http://dx.doi.org/10.7554/eLife.24987.016
Figure 4—figure supplement 2.
Figure 4—figure supplement 2.. Neural correlates of motor sequence learning during resting-state periods immediately following training.
The posterior parietal lobule functional connectivity within the learning pattern significantly differed between the MSL and CTL tasks during the resting-state condition immediately following training (RS2) as compared to baseline (RS1). Bar plot illustrates the change in functional connectivity of this area within the learning pattern between RS2 and RS1 scans averaged across all subjects in each task condition. The color-coded activation map indicates Z-score values and is corrected for multiple comparisons using GRF, p<0.05. Error bars represent s.e.m. DOI: http://dx.doi.org/10.7554/eLife.24987.017
Figure 4—figure supplement 3.
Figure 4—figure supplement 3.. Changes in brain functional connectivity related to motor sequence learning during the post-sleep (top row) and the pre-sleep (bottom row) resting-state conditions.
An ROI located in the let putamen showed a significant change in functional connectivity between tasks during the post-sleep resting state condition (RS3) as compared to the baseline (RS1); significant interaction (RS3−RS1)×(MSL−CTL), top row. In contrast, an ROI located in the right posterior parietal cortex (brodmann area, BA7) showed a significant change in functional connectivity between tasks during the pre-sleep resting state condition immediately following learning (RS2) as compared to the baseline (RS1); significant interaction (RS2−RS1)×(MSL−CTL), bottom row. As shown in the scatter plot, only changes in functional connectivity in the post-sleep condition were correlated with the amount of offline gains in performance speed. All maps are corrected for multiple comparisons using GRF, p<0.05. Display conventions are as in Figure 3—figure supplement 1. DOI: http://dx.doi.org/10.7554/eLife.24987.018
Figure 5.
Figure 5.. Temporal dynamics of memory trace during NREM stage two sleep.
(A, B) illustrate the time course of CI change within the learning and the consolidated patterns during NREM stage two sleep, respectively. The consolidated pattern’s CI gradually increased during NREM sleep only in the MSL condition (red curve), while the learning pattern’s CI decreased. CI did not differ significantly over the course of NREM stage two sleep in the CTL night (blue curves). Furthermore, CI within the learning (D) and consolidated (E) patterns did not change significantly over the course of intermittent awakenings distributed throughout the sleep session in the CTL (blue curves) or MSL (red curves) night. (C), (F) illustrate the time course of ventrolateral putamen functional connectivity within the consolidated pattern during NREM stage two sleep and intermittent bouts of awakenings, respectively. Likewise, the putamen functional connectivity with the rest of structures in the consolidated pattern gradually increased only during NREM sleep in the MSL condition. Each data point is calculated using 100 fMRI volumes. Shaded area represents s.e.m. DOI: http://dx.doi.org/10.7554/eLife.24987.019
Figure 5—figure supplement 1.
Figure 5—figure supplement 1.. Robustness of temporal dynamics analysis with respect to the window size.
(A, B) illustrate the CI values, averaged over subjects, within the learning and the consolidated patterns respectively during non-REM stage two sleep (including on average 600 volumes) using different window sizes ranging from n = 50 to 300 volumes. Each curve is calculated using different time windows comprising either the first n fMRI volumes (purple, green) or the last n fMRI volumes (red, blue) in the MSL (purple, red) or the CTL (green, blue) condition. Shaded area represents s.e.m. DOI: http://dx.doi.org/10.7554/eLife.24987.021
See this image and copyright information in PMC

Comment in

  • Shifting memories.
    Ngo HV, Staresina BP. Ngo HV, et al. Elife. 2017 Sep 11;6:e30774. doi: 10.7554/eLife.30774. Elife. 2017. PMID: 28892465 Free PMC article.

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Albouy G, Sterpenich V, Vandewalle G, Darsaud A, Gais S, Rauchs G, Desseilles M, Boly M, Dang-Vu T, Balteau E, Degueldre C, Phillips C, Luxen A, Maquet P. Neural correlates of performance variability during motor sequence acquisition. NeuroImage. 2012;60:324–331. doi: 10.1016/j.neuroimage.2011.12.049. - DOI - PubMed
    1. Albouy G, Fogel S, King BR, Laventure S, Benali H, Karni A, Carrier J, Robertson EM, Doyon J. Maintaining vs. enhancing motor sequence memories: respective roles of striatal and hippocampal systems. NeuroImage. 2015;108:423–434. doi: 10.1016/j.neuroimage.2014.12.049. - DOI - PubMed
    1. Ambrose RE, Pfeiffer BE, Foster DJ. Reverse replay of hippocampal place cells is uniquely modulated by changing reward. Neuron. 2016;91:1124–1136. doi: 10.1016/j.neuron.2016.07.047. - DOI - PMC - PubMed
    1. Beck AT, Rial WY, Rickels K. Short form of depression inventory: cross-validation. Psychological Reports. 1974;34:1184–1186. - PubMed
    1. Beck AT, Epstein N, Brown G, Steer RA. An inventory for measuring clinical anxiety: psychometric properties. Journal of Consulting and Clinical Psychology. 1988;56:893–897. doi: 10.1037/0022-006X.56.6.893. - DOI - PubMed

Publication types

Grants and funding