1 Department of Psychology, Concordia University, Montreal, Québec, Canada.
2 School of Health, Concordia University, Montréal, Québec, Canada.
3 Brain Language Lab, Freie Universität Berlin, Berlin, Germany.
4 Charité Universitätsmedizin, Berlin, Germany.
5 Department of Neurology, Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany.
6 Department of Nuclear Medicine and Radiobiology, Université de Sherbrooke, Sherbrooke, Québec, Canada.
7 Department of Biomedical Engineering, McGill University, Montreal, Québec, Canada.
8 McConnell Brain Imaging Centre, Montreal Neurological Institute, Montreal, Québec, Canada.
9 Department of Neurology and Neurosurgery, McGill University, Montreal, Québec, Canada.
10 Clinic for Cognitive Neurology, Leipzig, Germany.
11 Department of Physics, Concordia University, Montreal, Québec, Canada.
12 Montreal Heart Institute, Montreal, Québec, Canada.
13 Full Brain Picture Analytics, Leidein, the Netherlands.

Ultra-high field magnetic resonance imaging at sub-millimeter resolution opens the possibility to assess subtle longitudinal changes in brain structure and function. Identifying the regions and putative mechanisms involved in learning specific motor sequences is a key step towards understanding neuroplasticity. To disambiguate sequence-specific learning from simple motor execution, the present study trained an experimental group over five consecutive days on a complex motor sequence and contrasted them with a control group who performed a simple sequence. Both groups were scanned at 4 time points with magnetic resonance imaging at 7 Tesla with MP2RAGE. Training-related grey matter (GM) structural plasticity was assessed with voxel-based morphometry (VBM: time by group interaction) on T1w uniform intensity images and followed up with a focused investigation of quantitative T1 values (qT1) to probe the physiological changes driving GM plasticity. Our interaction analyses identified significant differences in the precuneus, superior parietal cortex (SPC), and angular gyrus (AG) where GM increases were greater in the experimental group. Of these regions, only the left SPC (during slow learning) exhibited sequence-specific plasticity-where the magnitude of change was greater in the experimental vs. the control group. Interestingly, while the experimental group showed increased GM volume, the control group was characterized by decreases. Post hoc comparisons revealed that the control group exhibited increasing T1 across days 1, 2 and 5, while the experimental group initially increased (day 1 to 2) and then decreased (day 2 to 5), suggesting that the two groups underwent differential physiological change as a function of training. These findings align with invasive animal studies showing that learning simple tasks is characterized by angiogenesis while learning complex sequences initially leads to synaptogenesis and is then followed by intracortical myelination as the sequence is well learned. Our findings provide initial evidence in humans that learning a complex sequence induces GM increases that are likely driven by synaptogenesis for initial encoding and myelogenesis for later learning and consolidation, while habituating to a simple repeated task leads to GM decreases due to attenuated blood flow.