Persistence of hippocampal and striatal multivoxel patterns during awake rest after motor sequence learning

Abstract

Memory consolidation, the process by which newly encoded and fragile memories become more robust, is thought to be supported by the reactivation of brain regions - including the hippocampus - during post-learning rest. While hippocampal reactivations have been demonstrated in humans in the declarative memory domain, it remains unknown whether such a process takes place after motor learning. Using multivariate analyses of task-related and resting state fMRI data, here we show that patterns of brain activity within both the hippocampus and striatum elicited during motor learning persist into post-learning rest, indicative of the reactivation of learning-related neural activity patterns. Moreover, results indicate that hippocampal reactivation reflects the spatial representation of the learned motor sequence. These results thus provide insights into the functional significance of neural reactivation after motor sequence learning.

Keywords: Biological sciences; Cognitive neuroscience; Neuroscience.

Graphical abstract

Figure 1

Experimental design and behavioral results (A) Resting state (RS) scans were acquired immediately prior to (RS1) and after (RS2) the initial training of a motor sequence learning task (Task A) in 55 young healthy individuals as well as following (RS3) a test session (Task B) assessing allocentric (Allo; n = 26) or egocentric (Ego; n = 29) sequence representations. Both tasks consisted of explicitly known 5-element sequences completed with the left non-dominant hand (1 = index finger; 2 = middle finger, 3 = ring finger, 4 = little finger). The egocentric representation Task B was assessed by having the participants complete the same sequence of finger presses as the initial training (i.e., 4-1-3-2-4); however, the spatial locations of the movement responses were now novel as both the keyboard and hand were inverted. For the allocentric representation, the spatial representation of the sequence was preserved by having the participants complete the mirrored finger sequence (i.e., 1-4-2-3-1) after the inversion of the keyboard and hand. Note that all task and RS runs were completed inside the MRI scanner. (B) Participants exhibited significant reductions in the duration to complete a correct sequence (in s) across practice blocks of Task A (magenta), reflective of learning the sequence of movements. Similar results were obtained for Task B (cyan) that followed RS2. These performance improvements were revealed by significant effects of block in one-way repeated measures ANOVAs (14 blocks for Task A and 6 blocks for Task B). As performance speed did not differ between allocentric and egocentric representations of the motor sequence task (see main text for statistical details), the plots above collapsed across the two experimental groups. Large circles represent means, shaded regions represent +/− SEM, small x’s and o’s denote individual data for training and test runs, respectively. (C) Zoomed-in version of behavioral data depicted in panel B to better represent reductions in movement speed as a function of practice. Circles represent means, shaded regions represent +/− SEM Data underlying panels B and C are included in the file Data S1, Figure 1 Data.

Figure 2

Pattern persistence following initial learning (A) MVCS matrices are depicted for an exemplar ROI and participant. Each matrix shows the correlation between each of the n voxels of the ROI with all the other voxels of the ROI during the pre-training resting state (RS1), Task A (i.e., motor sequence learning (MSL) training) and post-training resting state (RS2) runs. The similarity index (SI) between two matrices (RS1 vs. Task A and Task A vs. RS2) is defined as the r-to-z transformed correlation between each pair of MVCS matrices. RS1/Task A and Task A/RS2 SIs were compared to assess whether task-related patterns persisted significantly into post-learning rest (results presented in panel C). (B) Depiction of the hippocampus (HC), putamen (Put) and primary motor cortex (M1) ROIs overlaid on the MNI 152 template as part of the software MRIcroGL available at https://www.mccauslandcenter.sc.edu/mricrogl. (C) Violin plots depicting the Similarity Indices (SI; Fisher Z-transformed correlation coefficients) between RS1 and Task A as well as between Task A and RS2 for the HC, Put, and M1. For the HC and Put, the pattern of neural responses observed during Task A persisted into the subsequent rest, reflective of the reactivation of learning-related neural activity. For all violin plots, individual data points are shown as small colored circles jittered on the horizontal axis within the respective plot to increase visualization (N = 55); white circles and horizontal black lines depict group medians and means, respectively. ∗ indicates p < 0.05 for paired t-test after FDR correction for multiple comparisons. Data underlying the figure are included in the file Data S1, Figure 2 Data.

Figure 3

Functional significance of pattern persistence following Task A Violin plots depicting the Similarity Indices (SI) between MVCS matrices from post-training RS (i.e., RS2) and Task B for the hippocampus (HC; blue) and putamen (Put; red) separately for the egocentric (Ego) and allocentric (Allo) groups. For the HC, SI was greater for the allocentric than the egocentric group, suggesting pattern persistence during RS2 (following initial learning) reflects the reactivation of the allocentric representation of the motor sequence. There were no group differences for the Put. For all violin plots, individual data points are shown as small colored circles jittered on the horizontal axis within the respective plot to increase visualization (N = 28 and 26 in Ego and Allo groups, respectively; one subject in Ego was excluded due to missing data); white circles and horizontal black lines depict group medians and means, respectively. ∗ indicates p < 0.05 for unpaired t-test after FDR correction for multiple comparisons. Data underlying the figure are included in the file Data S1, Figure 3 Data.

Figure 4

Persistence of task patterns following Task B Violin plots depicting the Similarity Indices (SI) between MVCS matrices from Task B and those during RS2 and RS3 for the hippocampus (HC; blue) and putamen (Put; red). For the HC, the pattern of neural responses during Task B was more similar to RS3 than RS2, indicative of persistence. The patterns of neural responses in the Put during Task B were equally similar to the patterns observed during the preceding rest (RS2) and the subsequent rest (RS3) for both allocentric and egocentric conditions. For all violin plots, individual data points are shown as small colored circles jittered on the horizontal axis within the respective plot to increase visualization (N = 54; one subject was excluded due to missing data); white circles and horizontal black lines depict group medians and means, respectively. ∗ indicates p < 0.05 for the paired t-test after FDR correction for multiple comparisons. Data underlying the figure are included in the file Data S1, Figure 4 Data.

Figure 5

Functional significance of pattern persistence following Task B Violin plots depicting the Similarity Indices (SI) between MVCS matrices from the post-test RS (i.e., RS3) and those during Task B for the hippocampus (HC) separately for the egocentric (Ego) and allocentric (Allo) groups. SI was greater for the allocentric than the egocentric group, suggesting pattern persistence during RS3 (following Task B) reflects the allocentric representation of the motor sequence. For all violin plots, individual data points are shown as small colored circles jittered on the horizontal axis within the respective plot to increase visualization (N = 54; one subject was excluded due to missing data); white circles and horizontal black lines depict group medians and means, respectively. ∗ indicates p < 0.05 for independent t-test. Data underlying the figure are included in the file Data S1, Figure 5 Data.