Learning engages transient and sustained cellular mechanisms in the human brain Authors: Guillermina Griffa, Marco Palombo, Abraham Yeffal, Hong-Hsi Lee, Agustin Solano, Susie Y. Huang, Valeria Della-Maggiore, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD, Christian Schnell, PhD Published: 2026 ABSTRACT Structural neuroplasticity supports learning, development, and shapes vulnerability to brain disorders, making it a central priority in neuroscience research. However, progress in humans has remained limited by the inability to probe cellular processes in vivo, leaving mechanistic insight largely dependent on animal models. To address this gap, here we combined the sub-voxel sensitivity of ultra–high-gradient diffusion MRI with the cell-compartment specificity of the Soma and Neurite Density Imaging (SANDI) model to probe structural plasticity directly in the living human brain. By tracking how learning modulates the temporal dynamics of cell bodies and cell processes, we aimed to distinguish plastic from nonplastic biological processes driving changes in microstructure. We found that learning a motor skill triggered two distinct temporal responses: a transient expansion of cell bodies across all brain regions engaged by the task, consistent with a short-lived homeostatic mechanism, and a sustained increase in cell-process density restricted to key motor regions, consistent with structural plasticity. Our approach provides a mechanistic window into human neuroplasticity and marks a significant step toward bridging the gap between animal and human neuroscience. Introduction Experience continually remodels cellular architecture through structural modifications such as synaptogenesis, dendritic spine turnover [1,2], and the reorganization of astroglial processes that stabilize new synapses [3]. Structural plasticity supports learning and development and may influence the pathophysiology of psychiatric and neurodegenerative disorders [4–10]. Much of what is known about its cellular basis comes from invasive studies in animal models. Still, these species are evolutionarily distant and often can only approximate the complexity of human cognition and pathology. This gap underscores the need to study structural plasticity directly in the living human brain. Diffusion MRI (dMRI) offers a promising noninvasive window into cytoarchitecture because it is sensitive to micrometer-scale restrictions imposed by the cell [11,12]. Over the past two decades, diffusion tensor imaging [DTI; 13] has been the primary approach for studying experience-dependent microstructural change in humans. In gray matter, DTI-derived mean diffusivity (MD) is sensitive to learning and correlates with synaptogenesis and astroglial expansion, two hallmarks of structural plasticity in rodents [14–17]. However, because DTI conflates signals from multiple tissue compartments, it cannot distinguish lasting remodeling from transient short-lived homeostatic responses such as swelling that alter cellular morphology without enduring structural impact [18–21]. As a result, DTI remains an indirect marker of plasticity. This lack of biological specificity limits the mechanistic interpretation of diffusion findings across learning, development, and disease. To overcome these limitations, we leveraged the sensitivity of ultra–high-gradient dMRI [22–30] together with the specificity of the compartment-based Soma and Neurite Density Imaging (SANDI) model [31] to separate diffusion signals arising from cell bodies and cell processes. Using the motor sequence learning (MSL) task, a paradigm previously shown by our team to elicit robust functional and DTI changes [32], we tracked how these cellular compartments evolved at baseline, 30 min, and 24 h post-learning. By resolving the temporal dynamics of cell bodies and cell processes, we aimed to distinguish plastic from nonplastic biological mechanisms driving changes in microstructure. SANDI revealed two distinct cellular responses: a transient increase in apparent cell-body density, consistent with a short-lived homeostatic response, and a sustained rise in apparent cell-process density, compatible with structural remodeling. Our approach moves the field beyond descriptive microstructural changes toward mechanistic inference, with broad implications for development and disease. Results The MSL paradigm used here consists of executing a five-element finger sequence, initially learned declaratively, with the nondominant hand [Fig 1A, 33]. In our previous study, this task produced reliable improvements in performance alongside well-characterized functional MRI (fMRI) activation patterns and a distinct temporal pattern of reduction in MD in the hippocampus and precuneus [32]. To investigate the cellular mechanisms underlying these microstructural changes, here we replicated the same paradigm and longitudinal design inside the Connectome scanner [300 mT/m; 25,29,34,35], acquiring ultra–high-gradient multi-shell dMRI at baseline, 30 min, and 24 h later (Fig 1B). Experimental paradigm and design. A) Experimental paradigm. Right-handed participants performed a motor sequence learning (MSL) task involving a 5-item sequence of finger movements on a keyboard, using the four fingers of their non-dominant hand (4-1-3-2-4, with 4 representing the index finger and 1 representing the pinky finger). Initially, subjects memorized the sequence and were then instructed to execute it as quickly and accurately as possible during task blocks (green cross), and relax during rest periods (red cross). A total of 15 blocks were completed. B) Experimental design. fMRI images were obtained during the MSL task following a block design. To assess structural plasticity induced by learning, ultra–high multi-shell dMRI and corresponding T1-weighted images (T1w) were acquired longitudinally at three time points: before learning, 30 min post-learning, and 24 h post-learning. Finally, to assess overnight offline gains, participants completed 8 additional practice blocks 24 h after training; the test was assessed after the MRI session. Before applying SANDI, we first examined if the dataset obtained in the Connectome scanner reproduced the behavioral, fMRI, and DTI results of our previous work conducted on a clinical system. MSL showed the expected performance improvements, dominated by micro-offline gains (MOGs) [36,37] during rest periods between practice blocks rather than during active execution (Fig 2A). fMRI analysis revealed increased activity in the left hippocampus and precuneus during rest periods, consistent with early engagement of the limbic system, while cortico–striatal and cortico–cerebellar circuits supported sequence execution throughout training (Fig 2B). Finally, MD also decreased 30 min after learning in the left hippocampus and right precuneus, with additional reductions in left primary motor cortex (M1) and posterior parietal cortex (PPC), two brain regions central to motor skill acquisition [38,39]. Critically, MD changes in hippocampal and M1 were transient, whereas those in precuneus and PPC persisted overnight (Fig 2C). Behavioral, functional, and DTI changes induced by motor sequence learning (MSL). A) Behavioral results - The left panel shows the MSL curve, depicting the intertap interval -the time elapsed between successive key presses from correctly executed sequences- as a function of blocks (~12 sequences per block). The right panel shows the metrics corresponding to micro-offline gains (MOGs), micro-online gains (MONGs), and total learning, which were obtained based on the cumulative sum of deltas in performance during practice and rest blocks. Specifically, MOGs represent the difference in the mean intertap interval between the last correct sequence of a block and the first correct sequence of the next block. In contrast, MONGs represent the difference between the first and last correct sequence within a block. Total learning = MONGs + MOGs. Note that, in line with our previous work, improvements in performance on the MSL task occurred during rest periods interleaved with practice (t test against zero; p = 0.034), rather than during sequence execution (p = 0.59). Nonsignificant (ns), ***p < 0.001, *p < 0.05 corrected by Bonferroni. B) Functional results - Shown are the whole-brain voxelwise statistical parametric maps (SPMs) for the Task (Task > Rest; left plot) and Rest (Rest > Task; right plot) conditions (p < 0.05, Family-Wise Error [FWE]-corrected). Note that cortico–cerebellar and cortico–striatal networks were more active during task execution (in red), whereas the hippocampus and the precuneus were more active during the rest periods (in green). C) DTI results - Shown are the results from the Sandwich Estimator (SwE) statistical model on mean diffusivity (MD) using a whole-brain threshold-free cluster enhancement (TFCE) approach (p < 0.05 FWE-corrected) across the three dMRI sessions (baseline, 30 min, 24 h); barplots represent the mean % change of MD relative to the baseline ± 95% confidence intervals (CI) for the clusters identified in the whole-brain analysis. Note that MD changes are expressed as % of the baseline to facilitate the comparison to our DTI analysis carried out in a clinical MRI scanner [32]. MSL-induced a rapid (30 min) reduction in MD over the left posterior hippocampus, right precuneus, left primary motor cortex (M1), and left posterior parietal cortex (PCC). Remarkably, while hippocampal and M1 changes were transient, returning to baseline within 24 h, those in the precuneus and PPC persisted overnight. The data underlying this figure can be found at https://zenodo.org/records/20527091. In sum, these results replicate our previous behavioral, fMRI, and DTI findings obtained from a study comparing a learning group to a nonlearning control group performing repeated, non-sequential finger tapping [32]. This replication provides support for the learning specificity of the microstructural changes reported here and strengthens the basis for their cellular-level interpretation. SANDI uncovers distinct cellular contributions to MD dynamics The time course of MD changes across brain regions (Fig 2C) may reflect different phases of the same plasticity cascade or distinct underlying biological mechanisms. Because MD results from the aggregate contribution of multiple tissue compartments, DTI alone cannot distinguish between these hypotheses. To resolve this ambiguity, we applied the SANDI model to the multi-shell dMRI data and examined its compartment-specific estimates within the DTI-defined clusters (Fig 3A and 3B). SANDI is a compartment-based model that decomposes the diffusion signal in gray matter into three fractions: cell soma (FSOMA), cell processes (FNEURITE), and extracellular liquid (FEXTRA), enabling a biologically grounded interpretation of how microstructural changes evolve as a function of time. These compartment-specific measures are further supported by converging evidence from histological atlases, biophysical simulations, and preclinical studies, suggesting sensitivity to cell-body density and underlying microstructural organization [40–47]. Temporal dynamics of microstructural changes induced by MSL: DTI vs. SANDI. A) The MSL task induced a transient decrease in MD (left panel) in the left posterior hippocampus and left primary motor cortex (SwE, p < 0.05 FWE-corrected). These changes were accompanied by a transient increase in apparent cell-body density (FSOMA; right panel; repeated-measures ANOVA, Bonferroni-corrected; hippocampus: p = 0.016; M1: p < 0.001). B) MSL also induced a long lasting reduction in MD (left panel) in the precuneus and posterior parietal cortex (SwE, p < 0.05 FWE-corrected), which was associated with both a transient increase in FSOMA (right panel; precuneus: p = 0.044; PPC: p = 0.043) and a sustained increase in apparent cell-process density (FNEURITE; right panel; repeated-measures ANOVA, Bonferroni-corrected; precuneus: p = 0.024; PPC: p = 0.0071) that persisted at least 24 h post-learning. Bar plots depict the time course of the mean MD/SANDI metrics ± 95% CI for each brain region. *p < 0.05, and **p < 0.05; ***p < 0.001 Bonferroni-corrected. The data underlying this figure can be found at https://zenodo.org/records/20527091. We found that the MD reductions observed 30 min after learning were accompanied by a transient increase in FSOMA across all four regions (left hippocampus, right precuneus, left M1, left PPC; Fig 3A and 3B). Given the short time window, such changes are incompatible with processes like neurogenesis [48,49] and instead point to a rapid, reversible enlargement of cell bodies. The anatomical uniformity of this response across regions with different cytoarchitectonic and embryonic origins is consistent with a non-specific homeostatic mechanism. We propose that learning-induced neural activity causes membrane depolarization, transiently altering the ionic balance of the interstitial space. This perturbation is expected to trigger compensatory responses to buffer the ionic imbalance through cellular uptake, with water following osmotically into the cells and producing transient cell swelling [18–21]. Supporting this interpretation, all four regions were functionally engaged during learning (see Fig A in S1 Text), indicating that FSOMA changes occurred within an active network. The return of FSOMA to baseline at 24 h is consistent with the expected restoration of ionic balance toward physiological levels. In contrast, sustained MD reductions in the precuneus and PPC were associated with an increase in FNEURITE 30 min post-learning that persisted overnight, consistent with an increment in cell-process density (Fig 3B). This pattern aligns with structural plasticity observed in animal models, specifically, with the remodeling of dendritic and astroglial processes leading to new synapses and their stabilization, respectively [2,3,7,10,50,51]. Notably, FEXTRA closely tracked MD changes across all four regions (Fig 3A and 3B), consistent with the interpretation proposed by Assaf and colleagues that MD reductions partly reflect reduced interstitial space due to cellular expansion during plasticity-related states [15,17,52]. To examine whether these compartment-specific changes related to learning, we conducted linear mixed-effects models assessing the relationship between FSOMA and FNEURITE and behavioral measures of learning (see Methods). We found a significant association between the time course of FNEURITE and MOGs (F(1,25) = 6.15, p = 0.020). Specifically, there was a positive relationship at both 30 min and 24 h (β 30 min = 0.007, β 24 h = 0.02), which was significantly stronger at 24 h than at 30 min (time point × MOG interaction: F(1,79) = 3.95, p = 0.05). This suggests that participants showing greater early consolidation underwent more persistent structural remodeling of cell processes. No significant association was found between FNEURITE and overnight memory retention, and no significant interactions between transient FSOMA changes and any behavioral metric were observed, consistent with its interpretation as a nonlearning-specific, activity-dependent response. Together, these findings are consistent with the hypothesis that learning triggers at least two distinct biological mechanisms not distinguishable with DTI alone: a transient homeostatic response affecting cell bodies, and a sustained, regionally specific remodeling of cell processes compatible with structural plasticity. The ability to make this distinction rests on the spatiotemporal dissociation between FSOMA and FNEURITE, whose divergent temporal profiles and contrasting spatial distributions provide the inferential basis for interpreting these compartments as reflecting fundamentally different biological processes. The proposed mechanisms likely represent only a subset of the microstructural changes engaged during early learning, as processes occurring at finer spatial or temporal scales may not be detectable with the SANDI model under the current acquisition parameters. Independent, cross-model validation of SANDI In the whole-brain analyses, SANDI was applied within DTI-defined clusters, maximizing sensitivity but precluding an independent test of cross-model correspondence. To address the potential circularity of this approach, we performed a complementary model-agnostic analysis in the hippocampus. This region was selected because its subfields (head, body, tail) can be reliably segmented longitudinally [53], providing an anatomically defined framework independent of diffusion metrics. DTI and SANDI were fitted independently in each subfield at each time point (baseline, 30 min, 24 h) and their topographical distributions (MD versus FSOMA/FNEURITE) and temporal dynamics were compared. We observed a strong anatomical correspondence between MD and FSOMA across the three hippocampal subfields (Fig 4). Specifically, transient decreases in MD in the tail and body of the left hippocampus, previously reported in MSL and other learning tasks [32,54], were matched by a transient increase in cell-body density. Notably, no significant changes were detected in either the head or the right hippocampus, and MSL did not modulate cell-process density in any subfield. These findings are consistent with the whole-brain spatiotemporal pattern identified by DTI over the left posterior hippocampus. MSL-induced microstructural changes assessed independently using DTI and SANDI. Both MD and SANDI metrics were computed separately across the tail, body, and head subfields of the hippocampi (highlighted in red), which were segmented based on each subject’s T1-weighted images [53]. We observed anatomical overlap for MD and FSOMA metrics across the body and tail of the left hippocampus (repeated-measures ANOVA, Bonferroni-corrected; tail: MD: p = 0.0034; FSOMA: p = 0.022; body: MD: p = 0.034; FSOMA: p = 0.026). Notably, no changes were identified by either model in the hippocampal head or in the right hippocampus. MSL did not affect apparent cell-process density in any of the subfields (p > 0.05). Bar plots depict the time course of mean MD/SANDI metrics ± 95% confidence intervals (CIs). **p < 0.05; ns = nonsignificant. The data underlying this figure can be found at https://zenodo.org/records/20527091. The shared topography between MD and FSOMA confirms cross-model convergence and supports the robustness of SANDI for detecting structural changes in gray matter. Discussion Structural plasticity is instrumental to learning, development, and vulnerability to neurological and psychiatric disorders [55–60]. However, despite decades of research, our mechanistic understanding in humans has remained limited by the inability to probe cellular processes noninvasively. By combining ultra–high-gradient diffusion MRI with the compartment-based SANDI model [31], here we show that the temporal evolution of cell compartments provides relevant information regarding the biological mechanisms supporting learning, linking microstructural dynamics to human brain function. Importantly, converging evidence from histological atlases and preclinical models supports the interpretation of SANDI-derived metrics as reflecting underlying cellular microstructure [40–47]. Prior diffusion MRI work has relied on indirect indices such as MD, which are sensitive to learning but cannot distinguish plastic changes from non-plastic phenomena such as swelling [14–17,32,52,54,61,62]. Multi-compartment models were introduced to improve biological specificity [23,27,63–67], but most rely on white-matter–based geometries and are not optimal for studying gray matter plasticity. By focusing on cellular compartments SANDI provides great potential to explore plasticity in gray matter. Our results show that the mechanisms underlying changes in MD are likely to be heterogeneous and compartment-specific. A key feature of our approach is the longitudinal design, which enabled us to track the temporal dynamics of two cellular compartments, soma and neurites, across three timepoints spanning the early and later phases post-learning. It was precisely the combination of divergent temporal profiles and contrasting spatial distributions—the uniform FSOMA response across cytoarchitectonically distinct regions versus the regionally restricted FNEURITE increase—that provides the inferential basis for distinguishing two biologically distinct processes. We hypothesize that the widespread transient increase in cell-body density reflects homeostatic swelling of neurons and astrocytes that may be triggered by increased activity during early phases of synaptic potentiation [15,18–21]. In contrast, the persistent increase in cell-process density observed in precuneus and parietal cortex is consistent with remodeling of dendritic and glial cell processes leading to new synaptic connections and synaptic stabilization [2,50,51,68,69]. This spatiotemporal dissociation represents a significant conceptual shift in human neuroscience from descriptive diffusion changes toward identifying their biological drivers Persistent increases in cell-process density were restricted to the precuneus and PPC, while the hippocampus and M1 showed only transient FSOMA changes. This temporal dynamics—whereby the hippocampus supports early encoding and reactivation [32,70] while structural remodeling emerges selectively in cortical regions implicated in sequence integration and long-term representation [71,72,73]—is consistent with the systems consolidation framework [74,75], whereby memories initially dependent on the hippocampus are gradually consolidated in distributed cortical networks. Notably, cell-process density increases in the precuneus and PPC were already detectable at 30 min post-learning, indicating that cortical structural remodeling is initiated rapidly after learning. This challenges the classical view of systems consolidation as a slow, sleep-dependent process and is consistent with previous work showing that hippocampal-cortical interactions begin immediately after learning [32,76]. While sleep may further stabilize the structural remodeling initiated during waking, our design and acquisition parameters do not allow us to directly assess its contribution, nor to determine whether our findings are compatible with the synaptic homeostasis hypothesis [SHY, 77], as FNEURITE is unlikely to be sensitive to the fine synaptic changes SHY predicts [31,78]. The ability to dissociate homeostatic from plastic cellular responses is not only relevant for understanding learning, but opens broader possibilities for investigating structural plasticity across development, neurodegeneration, and neuropsychiatric disorders. For example, differentiating transient from persistent cellular changes could help identify biomarkers that distinguish adaptive remodeling from pathological processes such as neuroinflammation or atrophy. Combined with emerging MRI methods that should improve sensitivity to swelling and cell-type discrimination [79–83], this approach may contribute to a biologically interpretable framework for understanding how experience, maturation, and degeneration remodel the brain’s cellular architecture. Despite these advances, several limitations should be considered when interpreting the present findings. SANDI-derived metrics do not provide direct quantification of cellular compartments, but rather model-based inferences about their relative contributions to the diffusion signal. As such, they cannot be directly extrapolated to histological measures of cellular density or volume. In particular, due to the model’s assumption of negligible water exchange between compartments and the relatively long diffusion times used for dMRI acquisition (23 ms), neurite density values are systematically lower than histological estimates—a recognized limitation of diffusion models at these timescales [78]. As a consequence, the neurite compartment is primarily sensitive to larger cellular processes such as axons and dendrites, and is unlikely to capture fine synaptic structures, thereby defining the biological scale at which our inferences apply. Critically, because all analyses were performed relative to each region’s pre-learning baseline, this limitation does not affect the interpretation of the within-subject changes that form the basis of our conclusions. Another consideration is that SANDI yields relative signal fractions rather than absolute proton densities, making the compartment estimates compositionally dependent (i.e., the three fractions are constrained to sum to one within each voxel). More quantitative mapping would require diffusion-relaxometry acquisitions with longer scan times [84,85], calibrated to a proton-density reference such as CSF [86]. Finally, while the present findings provide evidence for distinct mechanisms driving MSL, they are based on a specific experimental paradigm in a healthy young adult sample, and may not generalize to other tasks or cohorts. Manipulating key training parameters -such as duration, intensity, or the implicit/explicit nature of the task- is likely to modulate the magnitude and persistence of compartment-specific changes, as suggested by recent work from our group [54]. In conclusion, tracking the spatiotemporal evolution of cellular compartments provides a biologically grounded framework to interpret microstructural changes in the human brain. By enabling the dissociation of plastic and homeostatic processes, this approach overcomes key limitations of traditional diffusion models and moves the field beyond descriptive changes toward mechanistic inference, thereby helping bridge cellular mechanisms identified in animals with human brain function. Methods Participants Twenty-nine healthy human subjects between 18 and 36 years old (16 female, mean age ± SD = 26.86 ± 5.41 years) were recruited under approval by the Institutional Review Board of Mass General Brigham (IRB protocol number: 2018P002443). All procedures were conducted in accordance with the principles expressed in the Declaration of Helsinki. Written informed consent was obtained from all participants. All participants reported no psychiatric, neurological, or cognitive impairment, nor any history of sleep disturbances. All subjects were right-handed as assessed by the Edinburgh Handedness Inventory [87]. Experimental paradigm Both the learning paradigm and the experimental design used in this study were identical to our previous MSL study [32]. Participants were instructed to learn a 5-item motor sequence using the four fingers of their nondominant (left) hand (4-1-3-2-4, 4 being the index finger and 1 being the pinky finger) as quickly and accurately as possible (Fig 1A). Experimental design and procedure All participants performed the MSL task in a self-paced manner during 15 blocks of 12 sequences each (60 key presses) interleaved with rest periods of 25 s [32]. The task lasted around 15–20 min. Multi-shell Diffusion MRI, T1-weighted, and functional images were acquired throughout the study. fMRI was collected during the MSL task following a block design (task, rest), while dMRI and T1w images were obtained in a longitudinal design at three time points: before practice (baseline), 30 min, and 24 h after learning to track the dynamics of structural plasticity in the short- and long-term (Fig 1B). Finally, to assess overnight offline gains, participants completed 8 additional practice blocks 24 h after training; the test was assessed after the MRI session. MRI acquisition Diffusion MRI, T1-weighted, and fMRI images were acquired on the original 3T Connectome MRI scanner (MAGNETOM CONNECTOM, Siemens Healthcare, Forchheim, Germany), equipped with a maximum gradient strength of 300 mT/m and a slew rate of 200 T/m/s, using a 64-channel phased array head coil [88]. Functional images (BOLD) were obtained following a block design during MSL (15 blocks of self-paced practice alternated with 25 s rest). The acquisition parameters were chosen to match our previous MSL study [32]: voxel size = 3 × 3 × 3 mm3; FOV = 210 mm; 42 slices aligned with the AC-PC line; 10% gap; posterior-anterior (P-A) phase-encoding direction; repetition time (TR) = 1,440 ms; echo time (TE) = 30 ms; multiband acceleration factor = 2; no PAT; bandwidth (BW) = 1,786 Hz/Px; echo spacing = 0.62 ms; EPI factor = 70; flip angle = 69°. A gradient echo (GRE) fieldmap was also acquired for correction of field inhomogeneities using the following parameters: 42 slices; 10% gap; voxel size = 3 × 3 × 3 mm3; field-of-view (FOV) = 210 mm; posterior-anterior (P-A) phase-encoding direction; TR = 716 ms; TE1 = 4.92 ms; TE2 = 7.38 ms; flip angle = 55°; BW = 600 Hz/px. T1w and dMRI images were acquired at baseline, 30 min, and 24 h after training. T1w images were obtained using the multi-echo sequence and the following parameters: TR = 2,530 ms; TE = 1.15 ms; flip angle = 7°; Inversion Time (TI) = 1,100 ms; BW = 649 Hz/Px; FOV = 210 mm; voxel size = 1 × 1 × 1 mm3; parallel acquisition = GRAPPA mode, acceleration factor = 3. The acquisition was performed in sagittal slices. The T1w images were used for the coregistration and subsequent normalization of fMRI images and for the segmentation of each subject’s hippocampal subfields. Multi-shell dMRI images were acquired using a monopolar pulsed-gradient spin-echo, single-shot echo-planar-imaging sequence [89] using the following parameters: diffusion time = 24 ms; gradient pulse duration = 8 ms; TR = 4,200 ms; TE = 55 ms; voxel size = 2 × 2 × 2 mm3; partial Fourier factor = 6/8, GRAPPA acceleration factor = 2; multiband acceleration factor (MB) = 2; anterior-posterior (A-P) phase-encoding direction. dMRI images were acquired in 8 b-values: 32 gradient directions for b-values = 200, 500, 1,200, 2,400 s/mm2, 64 directions for b-values = 4,000, 6,000, 8,000 s/mm2, and 22 interspersed nondiffusion-weighted b = 0 images. The latter were acquired for signal normalization and motion correction, of which five were acquired with reversed phase-encoding direction (posterior-to-anterior) for susceptibility distortion correction. MRI preprocessing Functional MRI. Preprocessing of functional images was conducted primarily using Statistical Parametric Mapping (SPM12, Wellcome Department of Cognitive Neurology). Initial corrections addressed gradient nonlinearity [90]. The functional time series was motion-corrected using the middle volume of the series as a reference. Since the experimental paradigm consisted of a block design, we did not perform slice-timing correction. Susceptibility distortion due to magnetic field inhomogeneities was corrected using the phase and magnitude images of a gradient-echo field map, and the realignment was performed to the middle volume of each functional time series with the aid of FSL’s (FMRIB Software Library) “fslsplit” and “fslmerge” commands. T1w structural images were coregistered to the BOLD images and normalized to the standard MNI152 template. This transformation was subsequently applied to the BOLD data, and the functional images were smoothed using a Gaussian kernel with an 8-mm full-width at half-maximum. T1. Human hippocampal subfields were segmented using FreeSurfer (version 7.0). Each subject’s T1w image was processed through FreeSurfer’s ‘recon-all’ pipeline. The hippocampal head, body, and tail were then extracted along the longitudinal axis using the optimized ‘segmentHA_T1’ function [53]. Diffusion MRI. Preprocessing and normalization of the diffusion MRI images were performed using the FSL (University of Oxford) (version 6.0.3), MRtrix3 (version 3.0.3), and Advanced Normalization Tools (ANTs; Wellcome Department, UCL) (version 2.4.2). Preprocessing steps were applied separately for each scanning session (baseline, 30 min, and 24 h post-training). Images were denoised using MRtrix3’s “dwidenoise”, corrected for Gibbs ringing using MRtrix3’s “mrdegibbs”, and corrected for the nonlinearity of gradients [90]. Geometric susceptibility distortion correction was made using FSL’s “Topup”. Subsequently, head motion, eddy currents, and b-vector rotation were made using FSL’s “eddy” [91]. Finally, bias correction was then applied using MRtrix3’s “dwibiascorrect”. Following preprocessing, FSL’s “dtifit” was used to fit the diffusion tensor model, generating scalar maps for fractional anisotropy (FA) and MD metrics. FA and MD maps were calculated for each subject and each voxel using the 1,200 s/mm² b-shell. Additionally, the SANDI model was fitted to the multi-shell dMRI acquisition using the SANDI MATLAB toolbox (https://github.com/palombom/SANDI-Matlab-Toolbox) [31]. SANDI accounts for cellular contributions to gray matter microstructure by modeling cell bodies as spheres, neurites as sticks, and extracellular space as isotropic Gaussian diffusion, producing three additive fractions. FSOMA tackles the contribution of the cell-body to the diffusion signal; FNEURITE estimates the contribution of cell processes such as neurites and cell processes from glial cells; FEXTRA, which represents the extracellular fluid, is calculated based on the other two fractions (FSOMA + FNEURITE + FEXTRA = 1). To quantify the DTI and SANDI metrics for each hippocampal subfield in native space (Region of Interest –ROI analysis), MD maps and SANDI-derived fractions were registered to each subject’s T1w image using a linear registration algorithm from ANTs. On the other hand, for the whole-brain analysis, DTI and SANDI maps were normalized to stereotaxic space (MNI152) using a custom-made pipeline based on ANTs to minimize across-session test-retest reproducibility error [92]. Finally, the normalized DTI and SANDI maps were smoothed using FSL’s smoothing function with a 4-mm full-width at half-maximum Gaussian kernel. Data analysis and statistics Behavior. MSL was quantified using the intertap interval, defined as the time elapsed between successive key presses during correctly executed sequences [32]. The mean intertap interval was calculated for each correct sequence (~12 sequences per block) and each subject. Then, the intertap intervals from correct sequences were averaged across all subjects for each block. Gains in performance during MSL were assessed as in our previous study [32] into micro-online gains (MONGs), micro-offline gains (MOGs), and total learning. MONGs were calculated as the difference (delta) between the mean intertap interval of the first and last correct sequence within a practice block, while MOGs were computed as the difference between the mean intertap interval of the last correct sequence of a practice block and the first correct sequence of the following block. Total learning represented the sum of MONGs and MOGs. To statistically assess gains in performance for each of these behavioral metrics, we performed t tests against zero, with p-values corrected for multiple comparisons using Bonferroni adjustments. Functional MRI. Statistical analyses of BOLD images were conducted using SPM12. We performed a standard whole-brain voxelwise general linear model (GLM) analysis to identify brain regions more active during motor execution (Task > Rest) and during rest periods (Rest > Task). To determine whether the parameter estimates (beta weights) from this GLM were significantly different from zero, we applied a one-tailed t test for each condition (p < 0.05 corrected for family-wise error–FWE). One subject was excluded from the functional analyses due to a technical inconvenience during acquisition affecting most of the dataset. Additionally, to assess whether the regions showing DTI/SANDI changes were functionally engaged during learning, we extracted the mean BOLD signal during early learning (blocks 1–7) from the four ROIs identified by the DTI analysis and compared it to a reference baseline period. This reference corresponded to the last 30 s of a 1-min fixation interval at the end of the scanning session, during which participants remained still viewing a neutral fixation cross. The first 30 s were excluded to ensure that any residual motor activity following task completion had dissipated. Although brief, this interval provides a reasonable approximation of a resting-state baseline. For each participant, the mean BOLD signal during Task and Rest blocks was normalized by subtracting the fMRI baseline value, and one-tailed one-sample t tests against zero (p < 0.05) were used to determine whether each region showed significant activation above baseline during Task, Rest, or both conditions. Diffusion MRI. To assess overall changes in MD across sessions (baseline, 30 min, and 24 h after learning), we conducted a whole-brain longitudinal analysis using a nonparametric permutation-based F test, using the threshold-free cluster enhancement (TFCE) approach (H = 2, E = 0.15) [93] implemented with the Sandwich Estimator (SwE) toolbox (version 2.0.0). SwE was specifically designed for accurate modeling of longitudinal and repeated-measures neuroimaging data [94]. The SwE analysis was conducted based on 1,000 permutations, and significant clusters were identified using an FWE-corrected p-value < 0.05. To reveal the temporal dynamics for each cluster of the F test, we extracted the median MD for each cluster, subject, and session, expressed it as the percent change relative to the baseline (Fig 2C) and determined the 95% confidence intervals using the summarySEwithin function from the Rmisc package in R [95,96]. Note that % MD changes are shown in Fig 2C to facilitate the comparison between our DTI analysis carried out in a clinical MRI scanner [32] and that acquired at the Connectome I scanner. To distinguish between short- and long-term changes in microstructure, we conducted two additional whole-brain analyses on the same MD maps using the same SwE approach described above. We then extracted the SANDI metrics (FEXTRA, FNEURITE, and FSOMA fractions) from the MD clusters, which were used as ROIs to assess longitudinal metrics of these changes (repeated-measures ANOVA; p < 0.05, corrected by Bonferroni). To reveal the temporal dynamics for SANDI metrics, we then computed each fraction’s median for each cluster, subject, and session and the corresponding 95% CI (Fig 3A and 3B). Additionally, we conducted linear mixed-effects models (LMMs) to examine the relationship between the time course of FSOMA and FNEURITE changes (relative to baseline: 30 min – baseline and 24 h – baseline) and behavioral measures of learning (MOGs; micro-online gains, MONGs; and overnight offline gains). Separate models were run for FSOMA and FNEURITE, with time point (30 min, 24 h) as a repeated factor, behavioral metrics as fixed effects, and subjects modeled as a random effect, accounting for between-subject variability. For FSOMA, the key term of interest was the time point × behavioral metric interaction, consistent with a transient response. For FNEURITE, the key term was the main effect of the behavioral metric, consistent with a more persistent structural change. To contrast DTI and SANDI across hippocampal subfields (Fig 4), we carried out independent statistical analyses (repeated-measures ANOVA; p < 0.05, corrected by Bonferroni) for each metric in the tail, body, and head hippocampi used as ROIs [53]. One subject was excluded from the statistical analysis due to missing data at the 24 h post-training MRI session. Supporting information Functional activation during early learning. (DOCX) 19 Dec 2025 Dear Valeria, Thank you for submitting your manuscript entitled "A mechanistic window onto human neuroplasticity through cell-specific diffusion MRI" for consideration as a Short Reports by PLOS Biology. Your manuscript has now been evaluated by the PLOS Biology editorial staff and I am writing to let you know that we would like to send your submission out for external peer review. I should note, however, in the absence of advice from one of our academic editors, that we were not fully convinced whether the conceptual advance is sufficient for us. We will be discussing this with the reviewers and the academic editor after review. However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire. Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. After your manuscript has passed the checks it will be sent out for review. To provide the metadata for your submission, please Login to Editorial Manager (https://www.editorialmanager.com/pbiology) within two weeks, i.e. by Jan 01 2026 11:59PM. Please note that our office will be closed starting on Monday for two weeks, so we will only be able to start inviting reviewers in the second week of January. If your manuscript has been previously peer-reviewed at another journal, PLOS Biology is willing to work with those reviews in order to avoid re-starting the process. Submission of the previous reviews is entirely optional and our ability to use them effectively will depend on the willingness of the previous journal to confirm the content of the reports and share the reviewer identities. Please note that we reserve the right to invite additional reviewers if we consider that additional/independent reviewers are needed, although we aim to avoid this as far as possible. In our experience, working with previous reviews does save time. If you would like us to consider previous reviewer reports, please edit your cover letter to let us know and include the name of the journal where the work was previously considered and the manuscript ID it was given. In addition, please upload a response to the reviews as a 'Prior Peer Review' file type, which should include the reports in full and a point-by-point reply detailing how you have or plan to address the reviewers' concerns. During the process of completing your manuscript submission, you will be invited to opt-in to posting your pre-review manuscript as a bioRxiv preprint. Visit http://journals.plos.org/plosbiology/s/preprints for full details. If you consent to posting your current manuscript as a preprint, please upload a single Preprint PDF. Feel free to email us at plosbiology@plos.org if you have any queries relating to your submission. Kind regards, Christian Christian Schnell, PhD Senior Editor PLOS Biology cschnell@plos.org 13 Feb 2026 Dear Valeria, Thank you for your patience while your manuscript "A mechanistic window onto human neuroplasticity through cell-specific diffusion MRI" was peer-reviewed at PLOS Biology. It has now been evaluated by the PLOS Biology editors, an Academic Editor with relevant expertise, and by several independent reviewers. In light of the reviews, which you will find at the end of this email, we would like to invite you to revise the work to thoroughly address the reviewers' reports. As you will see below, the reviewers find the topic of your study very interesting and are overall positive. However, the reviewers raise a number of important concerns that need to be addressed. While we ask you to address all concerns, we specifically we ask you to focus on (i) sharpening the scope and describe the advance of the study in more clarity and (ii) guiding the reader more clearly about what is being shown, how it is learning related, how we can be certain that these are structural changes, and that they relate to some aspect of learning. We think you will be able to address these concerns with textual revisions and (re)analyzing current data, and do not expect you to collect new experimental data. However, these requests are not simply cosmetic changes. You will see in the reviewer reports that there is some ambiguity in how the reviewers view the advance of you study, so it needs to be very clear what the advance is, and that we can reliably attribute the DTI/SANDI changes to structural change. Given the extent of revision needed, we cannot make a decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is likely to be sent for further evaluation by all or a subset of the reviewers. In addition to these revisions, you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests shortly. We expect to receive your revised manuscript within 3 months. Please email us (plosbiology@plos.org) if you have any questions or concerns, or would like to request an extension. At this stage, your manuscript remains formally under active consideration at our journal; please notify us by email if you do not intend to submit a revision so that we may withdraw it. **IMPORTANT - SUBMITTING YOUR REVISION** Your revisions should address the specific points made by each reviewer. Please submit the following files along with your revised manuscript: 1. A 'Response to Reviewers' file - this should detail your responses to the editorial requests, present a point-by-point response to all of the reviewers' comments, and indicate the changes made to the manuscript. *NOTE: In your point-by-point response to the reviewers, please provide the full context of each review. Do not selectively quote paragraphs or sentences to reply to. The entire set of reviewer comments should be present in full and each specific point should be responded to individually, point by point. You should also cite any additional relevant literature that has been published since the original submission and mention any additional citations in your response. 2. In addition to a clean copy of the manuscript, please also upload a 'track-changes' version of your manuscript that specifies the edits made. This should be uploaded as a "Revised Article with Changes Highlighted" file type. *Re-submission Checklist* When you are ready to resubmit your revised manuscript, please refer to this re-submission checklist: https://plos.io/Biology_Checklist To submit a revised version of your manuscript, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' where you will find your submission record. Please make sure to read the following important policies and guidelines while preparing your revision: *Published Peer Review* Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details: https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/ *PLOS Data Policy* Please note that as a condition of publication PLOS' data policy (http://journals.plos.org/plosbiology/s/data-availability) requires that you make available all data used to draw the conclusions arrived at in your manuscript. If you have not already done so, you must include any data used in your manuscript either in appropriate repositories, within the body of the manuscript, or as supporting information (N.B. this includes any numerical values that were used to generate graphs, histograms etc.). For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5 *Blot and Gel Data Policy* We require the original, uncropped and minimally adjusted images supporting all blot and gel results reported in an article's figures or Supporting Information files. We will require these files before a manuscript can be accepted so please prepare them now, if you have not already uploaded them. Please carefully read our guidelines for how to prepare and upload this data: https://journals.plos.org/plosbiology/s/figures#loc-blot-and-gel-reporting-requirements *Protocols deposition* To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols Thank you again for your submission to our journal. We hope that our editorial process has been constructive thus far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments. Sincerely, Christian Christian Schnell, PhD Senior Editor PLOS Biology cschnell@plos.org ------------------------------------ REVIEWS: Reviewer #1 (Yaniv Assaf): This is a strong and conceptually important manuscript that makes a clear advance in human neuroplasticity research. By combining ultra-high-gradient diffusion MRI with the SANDI model, the authors provide one of the first in vivo demonstrations that learning-related microstructural changes in the human brain can be temporally and compartmentally dissociated into transient, non-plastic processes (cell-body expansion) and sustained, region-specific structural plasticity (increased cell-process density). This directly addresses a long-standing limitation of diffusion MRI studies, which have been unable to distinguish swelling or homeostatic responses from genuine remodeling. Methodologically, the work is rigorous and well controlled. The longitudinal design, replication of prior behavioral/fMRI/DTI effects, use of ultra-high-gradient data, and convergent validation via hippocampal subfield analyses substantially strengthen confidence in the findings. The interpretation is biologically grounded, well integrated with animal literature, and avoids overstatement. Importantly, the results move the field beyond descriptive diffusion changes toward mechanistic inference, which is particularly valuable for studies of learning, development, and disease. The main limitations (model assumptions, relative—not absolute—compartment fractions, and reliance on a single learning paradigm) are appropriately acknowledged and do not undermine the central conclusions. Any remaining issues are largely matters of framing and clarification, particularly with respect to the biological specificity of the model and the extent to which these findings can be generalized beyond the present task and acquisition regime. These do not represent conceptual or technical weaknesses of the study. Overall, this is a high-quality and high-impact contribution that is very well aligned with the scope and standards of PLOS Biology. I therefore strongly recommend acceptance, with at most minor revisions focused on sharpening the discussion and boundaries of interpretation, rather than on additional analyses or changes to the core conclusions. Reviewer #2: This is an interesting and fun manuscript to read. It makes a valuable contribution by showing that different aspects of an MRI (diffusion) signal can provide insight into different cellular processes, which are differentially affected by learning. It is a nice mix of technical development (validation) along with more cognitive/biological issues. Many interesting questions arise from these insights that I don't articulate here because they are beyond the scope of the current work; nonetheless, they speak to the importance (and potential impact) of the work. There are several aspects that I struggled to understand, and so the work could benefit from clarification. These center around two major issues: (a) the relationship between the recorded signal and cellular changes; (b) new insights provided into motor memory processing. The following questions sometimes deal with one or other of these questions or at times are a bit overlapping. Perhaps the greatest support for the two signals (FSOMA and FNEURITE) tapping into different biological processes is that they dissociate across time (and space). The FNEURITE signal as would be expected last longer (present at 24 hours) and for specific brain areas. It should be possible to highlight this important strength more in the work. Whilst the logic that these signals are tapping into different biological processes is sound there is not much evidence in the study that they are --- other than this very nice temporal dissociation. What is the relationship between the functional activation (see Fig 2 - having these was a nice strength to the work) and the FNEURITE and FSOMA signals? Are they overlapping, and so come about due to energetic changes, and therefore consistent with the (energetic) cellular changes that are thought to be the basis for the different (MRI) signal changes? It was hard to understand by what was meant by "homeostatic changes". This phrase is repeatedly used; however, it was not clear to me how the changes related to homeostasis. Learning is the perturbing event, and there is then a process to return the biological system to a set-point (homeostasis), which in the case of say blood loss (perturbing event) is vasoconstriction (homeostatic process) to achieve a consistent set point (blood pressure). Beyond learning being the perturbing event, I am confused and unclear about all the other aspects of homeostasis (the process, the set point). The relationship between changes interpreted as cell body changes (FSOMA) and those that are interpreted as cell process (FNEURITE) changes could be revealing and strengthen the perspective provided in the work. Specifically, the cell body increasing in size could be due to protein synthesis required for distal cell process changes (FNEURITE). Could those brain areas (precuneus and parietal) showing lasting changes have cell bodies in those brain areas showing short term changes (FSOMA)? It was also not clear to me how this work fits in or alters our understanding of offline memory processing? Is the long-term storage of a sequence not within the hippocampus but in the cortex (such as the parietal cortex)? Is this what happens due to consolidation, and perhaps sleep? Given that sleep is meant to decrease synaptic density (synaptic downscaling) would there not have been an expected decrease in the FNEURITE signal (broadly) across many neural circuits? Is it possible to account for this apparent discrepancy between these results. I suspect it comes down to baseline and when it was taken relative to sleep. However, I think it might be nice to make this issue explicit so that other readers do not need to think this through for themselves. How many of these changes are perhaps due to performance of a sequence as opposed to learning a sequence? Would the human brain be unaffected by making repeated movements without any serial structure? Minor/Specific comments: On page 3, the sentence starting "In sum, these….." I did not understand how this related to the earlier paragraph that seemed to discuss learning as opposed to performance. On page 7, I didn't understand the value provided by this analysis. I am sure it is valuable. I think it could be just better articulated how and what it adds. Reviewer #3: The manuscript by Griffa et al. investigates neuroplasticity in response to motor learning using novel diffusion MRI methods. The main aim is to overcome previous limitations of conventional diffusion MRI methods in capturing biologically specific brain changes in humans. Longitudinal studies have started to investigated learning-induced neuroplasticity in humans approx. 20 years ago. Diffusion MRI is sensitive to neuroplasticity in grey matter regions and several studies over the past 10 years have consistently shown motor learning-related changes in hippocampal networks using diffusion MRI. The current study leverages the specificity of new MRI hardware (connectome MRI scanner) and software (SANDI) to cell-specific grey matter changes (e.g., astrocytes, neurons). SANDI is one of many new methods proposing increased biological specificity of parameters derived from in vivo diffusion MRI data. Griffa et al. obtained three diffusion MRI scans per participant: before, 30 min. as well as 24h after the first practice session of a new finger sequence task. Results show short-term and transient increases in a cell-specific SANDI marker in hippocampus and motor cortex as well as short-term but persistent (remained at least until 24h) neurite-specific changes in posterior parietal cortex and precuneus. The authors conclude that the results differentiate between plastic (persistent) and non-plastic (transient) changes and provide a mechanistic window into human neuroplasticity relevant to bridge the gap between human and animal neuroscience research. The evidence is based on (1) a well-established learning paradigm (finger sequence learning) suitable for application within the MR-environment, (2) a beneficial combination of diffusion-based MRI and task-based fMRI for localizing relevant brain areas and (3) a remarkable consistency of previous findings (partly from the same group) on hippocampal involvement in early motor sequence learning using conventional DTI methods. Overall, the results appear in line with previous DTI findings and the new methods (SANDI, connectome MR scanner) enabled the discovery of dissociable temporal dynamics in presumably cell- and neurite-specific brain changes underlying the DTI signal - which is indeed a leap forward for the human imaging field and a stimulus for new hypotheses testable with cross-species animal and human MRI studies. However, there are major concerns related to the validity of conclusions which I describe below: 1. The authors did not include a control group that would allow them to relate SANDI changes to learning (rather than unspecific changes over time). This is unusual, in part, because the sequence learning paradigm has been used for decades using well-controlled study designs. The authors mentioned the consistency of DTI- and fMRI findings with their previous work (that included a non-learning control group), but due to the lack of additional correlations between SANDI changes and learning parameters, the nature of cell- and neurite-specific changes remain elusive. The lack of behavioral correlations is unexpected given the wealth of learning parameters captured by the study design (micro-offline gains, micro-online gains, retention performance…). A proper control group or correlations with learning parameters would be essential to associate brain changes to behavior. 2. With respect to 1., I think that the evidence for concluding "distinct plastic and non-plastic changes" underlying the SANDI signals is low due to the following reasons: first, there are lots of relevant microstructural changes induced during initial skill learning and MRI may only capture the tip of the iceberg. Theory suggests non-linear brain changes during motor skill acquisition (see Hille et al., 2024). Labelling a transient change (fig. 3A) as "non-plastic" ignores the large number of ongoing changes causally relevant for learning (e.g., the clustered addition and pruning of new synaptic spines) but potentially not detectable with diffusion MRI. Second, dissociable time courses should be inferred from direct statistical comparisons of SANDI changes (e.g., using analyses of variance) rather than separate region-wise analyses (as done throughout the manuscript). 3. The authors interpret the increase in soma fraction as potential correlate of activity-induced cell swelling. In the abstract, they note that "We found that learning a motor skill triggered two distinct temporal responses: a transient expansion of cell bodies across all brain regions engaged by the task…". Right-handed participants practiced the finger sequence using their non-dominant left hand and fMRI results indicate task-induced activity in the contralateral right M1 (fig. 2B). However, FSOMA changes appeared in left but not right M1 (figs. 2, 3) which argues against brain activity as driver of FSOMA changes. In addition to that, the hippocampus was more active in the resting periods between practice blocks compared to the actual task execution (fig. 2). It is not clear, however, if this is the results of a deactivation during practice blocks or an activation during rest. Together, the relation between brain activity and cell-specific transient change remains unclear. 4. SANDI offers readouts of cell diameter in addition to the fractions of soma, neurite and free water. Cell diameter was not analysed in the current study but this parameter would be more specific to the proposed cell swelling mechanism (rather than soma fraction). 5. The empirical basis of cell-specific (rather than neurite-specific) plasticity through short-term learning remains to be shown in the intro. Cited studies largely reflect neurite-specific mechanisms but then, why was SANDI and not another diffusion model used in the context of this study? 6. Please also describe the evidence from histological validation studies indicating cell-specific mechanisms as a correlate of SANDI changes. 28 Apr 2026 Submitted filename: Authors_response.pdf 1 May 2026 Dear Valeria,, Thank you for your patience while we considered your revised manuscript "A mechanistic window onto human neuroplasticity through cell-specific diffusion MRI" for publication as a Short Reports at PLOS Biology. Your revised study has been evaluated by the PLOS Biology editors and the Academic Editor, but not yet by the original reviewers. Before we will send the manuscript back to the reviewers, however, we would like you to address the editorial comments from the previous decision letter more clearly, because we feel that these haven't been directly addressed. This includes issues about advance. For example, is it actually possible for synaptic changes to be observed after only 30 minutes? Is there any evidence in the literature for this, for example from animal studies? I have pasted the excerpt from the previous decision letter below. I encourage you to add your response to these editorial comments to the rebuttal letter after your response to Reviewer 3's comments, so the reviewers have access to this discussion too. Given the extent of revision needed, we cannot make a decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is likely to be sent for further evaluation by all or a subset of the reviewers. In addition to these revisions, you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests shortly. We expect to receive your revised manuscript within 3 months. Please email us (plosbiology@plos.org) if you have any questions or concerns, or would like to request an extension. At this stage, your manuscript remains formally under active consideration at our journal; please notify us by email if you do not intend to submit a revision so that we may withdraw it. **IMPORTANT - SUBMITTING YOUR REVISION** Your revisions should address the specific points made by each reviewer. Please submit the following files along with your revised manuscript: 1. A 'Response to Reviewers' file - this should detail your responses to the editorial requests, present a point-by-point response to all of the reviewers' comments, and indicate the changes made to the manuscript. *NOTE: In your point-by-point response to the reviewers, please provide the full context of each review. Do not selectively quote paragraphs or sentences to reply to. The entire set of reviewer comments should be present in full and each specific point should be responded to individually, point by point. You should also cite any additional relevant literature that has been published since the original submission and mention any additional citations in your response. 2. In addition to a clean copy of the manuscript, please also upload a 'track-changes' version of your manuscript that specifies the edits made. This should be uploaded as a "Revised Article with Changes Highlighted" file type. *Re-submission Checklist* When you are ready to resubmit your revised manuscript, please refer to this re-submission checklist: https://plos.io/Biology_Checklist To submit a revised version of your manuscript, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' where you will find your submission record. Please make sure to read the following important policies and guidelines while preparing your revision: *Published Peer Review* Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details: https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/ *PLOS Data Policy* Please note that as a condition of publication PLOS' data policy (http://journals.plos.org/plosbiology/s/data-availability) requires that you make available all data used to draw the conclusions arrived at in your manuscript. If you have not already done so, you must include any data used in your manuscript either in appropriate repositories, within the body of the manuscript, or as supporting information (N.B. this includes any numerical values that were used to generate graphs, histograms etc.). For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5 *Blot and Gel Data Policy* We require the original, uncropped and minimally adjusted images supporting all blot and gel results reported in an article's figures or Supporting Information files. We will require these files before a manuscript can be accepted so please prepare them now, if you have not already uploaded them. Please carefully read our guidelines for how to prepare and upload this data: https://journals.plos.org/plosbiology/s/figures#loc-blot-and-gel-reporting-requirements *Protocols deposition* To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols Thank you again for your submission to our journal. We hope that our editorial process has been constructive thus far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments. Sincerely, Christian Christian Schnell, PhD Senior Editor PLOS Biology cschnell@plos.org === Excerpt from the previous decision letter: While we ask you to address all concerns, we specifically we ask you to focus on (i) sharpening the scope and describe the advance of the study in more clarity and (ii) guiding the reader more clearly about what is being shown, how it is learning related, how we can be certain that these are structural changes, and that they relate to some aspect of learning. We think you will be able to address these concerns with textual revisions and (re)analyzing current data, and do not expect you to collect new experimental data. However, these requests are not simply cosmetic changes. You will see in the reviewer reports that there is some ambiguity in how the reviewers view the advance of you study, so it needs to be very clear what the advance is, and that we can reliably attribute the DTI/SANDI changes to structural change. 3 May 2026 Submitted filename: Authors_response_auresp_3.pdf 20 May 2026 Dear Valeria, Thank you for your patience while we considered your revised manuscript "A mechanistic window onto human neuroplasticity through cell-specific diffusion MRI" for publication as a Short Reports at PLOS Biology. This revised version of your manuscript has been evaluated by the PLOS Biology editors, the Academic Editor and two of the original reviewers. Based on the reviews and on our Academic Editor's assessment of your revision, we are likely to accept this manuscript for publication, provided you satisfactorily address the remaining points raised by Reviewer 3. Please also make sure to address the following data and other policy-related requests: * We would like to suggest a different title to improve its accessibility for our broad audience: Learning induces transient cell-body expansion and sustained increases in neurite density in the human brain * Please add the links to the funding agencies in the Financial Disclosure statement in the manuscript details. * Please include the approval/license number of the ethical approval for the experiments. * Please include information in the Methods section whether the study has been conducted according to the principles expressed in the Declaration of Helsinki. * Please provide information whether the participants provided written or verbal consent. * Please move the methodological details from the supplementary information file to the main manuscript methods sections. We do not have a word count limit. * DATA POLICY: You may be aware of the PLOS Data Policy, which requires that all data be made available without restriction: http://journals.plos.org/plosbiology/s/data-availability. For more information, please also see this editorial: http://dx.doi.org/10.1371/journal.pbio.1001797 Note that we do not require all raw data. Rather, we ask that all individual quantitative observations that underlie the data summarized in the figures and results of your paper be made available in one of the following forms: 1) Supplementary files (e.g., excel). Please ensure that all data files are uploaded as 'Supporting Information' and are invariably referred to (in the manuscript, figure legends, and the Description field when uploading your files) using the following format verbatim: S1 Data, S2 Data, etc. Multiple panels of a single or even several figures can be included as multiple sheets in one excel file that is saved using exactly the following convention: S1_Data.xlsx (using an underscore). 2) Deposition in a publicly available repository. Please also provide the accession code or a reviewer link so that we may view your data before publication. Regardless of the method selected, please ensure that you provide the individual numerical values that underlie the summary data displayed in the following figure panels as they are essential for readers to assess your analysis and to reproduce it: 2A (right panel), 2C (left part of each panel), 3AB, 4 and S1. NOTE: the numerical data provided should include all replicates AND the way in which the plotted mean and errors were derived (it should not present only the mean/average values). Please also ensure that figure legends in your manuscript include information on where the underlying data can be found, and ensure your supplemental data file/s has a legend. Please ensure that your Data Statement in the submission system accurately describes where your data can be found. * CODE POLICY Per journal policy, if you have generated any custom code during the course of this investigation, please make it available without restrictions. Please ensure that the code is sufficiently well documented and reusable, and that your Data Statement in the Editorial Manager submission system accurately describes where your code can be found. More information on our Code Policy, what and how to share can be found here: https://journals.plos.org/plosbiology/s/code-availability Please note that we cannot accept sole deposition of code in GitHub, as this could be changed after publication. However, you can archive this version of your publicly available GitHub code to Zenodo. Once you do this, it will generate a DOI number, which you will need to provide in the Data Accessibility Statement (you are welcome to also provide the GitHub access information). See the process for doing this here: https://docs.github.com/en/repositories/archiving-a-github-repository/referencing-and-citing-content As you address these items, please take this last chance to review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the cover letter that accompanies your revised manuscript. In addition to these revisions, you may need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests shortly. If you do not receive a separate email within a few days, please assume that checks have been completed, and no additional changes are required. We expect to receive your revised manuscript within two weeks. To submit your revision, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' to find your submission record. Your revised submission must include the following: - a cover letter that should detail your responses to any editorial requests, if applicable, and whether changes have been made to the reference list - a Response to Reviewers file that provides a detailed response to the reviewers' comments (if applicable, if not applicable please do not delete your existing 'Response to Reviewers' file.) - a track-changes file indicating any changes that you have made to the manuscript. NOTE: If Supporting Information files are included with your article, note that these are not copyedited and will be published as they are submitted. Please ensure that these files are legible and of high quality (at least 300 dpi) in an easily accessible file format. For this reason, please be aware that any references listed in an SI file will not be indexed. For more information, see our Supporting Information guidelines: https://journals.plos.org/plosbiology/s/supporting-information *Published Peer Review History* Please note that you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details: https://plos.org/published-peer-review-history/ *Press* Should you, your institution's press office or the journal office choose to press release your paper, please ensure you have opted out of Early Article Posting on the submission form. We ask that you notify us as soon as possible if you or your institution is planning to press release the article. *Protocols deposition* To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols Please do not hesitate to contact me should you have any questions. Sincerely, Christian Christian Schnell, PhD Senior Editor cschnell@plos.org PLOS Biology ------------------------------------------------------------------------ Reviewer remarks: Reviewer #2: The authors have done a great job of revising this work. The third paragraph of the discussion is an example for us all - it very nicely balances the observations with their interpretation. It makes it clear what is known against what is thought. Reviewer #3: Thank you very much for your responses to my concerns. The manuscript improved significantly. The important behavioral correlations suggest a link between diffusion changes and persistent improvements in performance. In your response to point 3, the lateralization issue was not sufficiently addressed in my opinion. I got the point that left-hand practice engages contralateral AND ipsilateral M1 (consistent with ealier studies), but this does not explain why the highly active right M1 was not affected by diffusion changes. This seems inconsistent with the interpretation that practice-induced increases in brain activity drive neural plasticity. Please provide an explanation for this result. 25 May 2026 Submitted filename: Authors Response.pdf 4 Jun 2026 Dear Dr Della-Maggiore, Thank you for the submission of your revised Short Reports "Learning engages transient and sustained cellular mechanisms in the human brain" for publication in PLOS Biology. On behalf of my colleagues and the Academic Editor, Edwin Robertson, I am pleased to say that we can in principle accept your manuscript for publication, provided you address any remaining formatting and reporting issues. These will be detailed in an email you should receive within 2-3 business days from our colleagues in the journal operations team; no action is required from you until then. Please note that we will not be able to formally accept your manuscript and schedule it for publication until you have completed any requested changes. While you attend to those requests to come, please also make sure to reference the source data in the figure legends, for example: "The data underlying this Figure can be found at https://zenodo.org/records/20527091." Please take a minute to log into Editorial Manager at http://www.editorialmanager.com/pbiology/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process. PRESS We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with biologypress@plos.org. If you have previously opted in to the early version process, we ask that you notify us immediately of any press plans so that we may opt out on your behalf. We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/. Thank you again for choosing PLOS Biology for publication and supporting Open Access publishing. We look forward to publishing your study. Sincerely, Christian Christian Schnell, PhD Senior Editor PLOS Biology cschnell@plos.org