Cortical reorganization and plasticity In the healthy brain

National Institute Of Neurological Disorders And Stroke

Investigators

Linked publications & trials

Abstract

Background: Cortical reorganization occurs in the adult central nervous system, especially during motor skill acquisition. This plasticity contributes to various forms of human behavior including skill learning and memory formation, consolidation, reconsolidation and short- and long-term retention. It is very important to understand the role of these different behavioral processes and of the mechanisms underlying these various forms of human plasticity during skill acquisition to improve skill learning and memory in healthy adults. Findings this year: Our first new finding from FY2025 was reported in the journal, eLife (https://doi.org/10.7554/eLife.102475.3). Activities of daily living rely on our ability to acquire new motor skills composed of precise action sequences. Recent work from our research group has demonstrated that initial learning of a new skill is predominantly expressed as jumps in performance following short rest breaks between practice bouts as opposed to within the practice period, indicating a form of memory consolidation operating over several seconds rather than hours or days as previously observed. Here, we asked if the real-time (i.e. - active at the millisecond level) neural representations of individual actions performed as part of a skilled action sequence incorporate information about the overall skill as learning progresses. An illustrative example would be a person learning to play the piano for the first time. Typically, students begin their training by practicing a short music piece (i.e. – a sequence of individual notes or finger movement actions), where the same note may be played at different locations. Will the brain activity generating the finger movements to play those notes be the same or different? The answer to this question has important implication on how our brains learn novel fine motor skills or encode memories about carrying out important tasks in our daily routines, as well as on the development of non-invasive brain stimulation protocols aimed at enhancing skill learning and performance. We recorded magnetoencephalography (MEG) in healthy adult human subjects while they learned to perform a novel sequential keypress skill task with their non-dominant left hand. The neural representation associated with each keypress action was characterized by computing MEG source-space brain activity features time-locked to the finger movement. Over several minutes of practice, we evaluated changes in the online (within-trial) and offline (between-trial) distance between neural representations of the same finger movement performed at different skill sequence locations. We developed a novel, hybrid spatial scale, brain activity input feature space for finger movement decoding. This approach combined features that emphasized correlated brain activity within local regions that was uncorrelated across the entire brain (i.e. – correlated within but uncorrelated between regions) with those that emphasized uncorrelated activity within selected brain regions into a single set. We then used this hybrid-space feature set to train machine learning decoders that could predict individual sequence-embedded finger movements performed during the skill task with considerable accuracy—approximately 94% on average across all subjects—an improvement of more than 25% over our previous work utilizing source-space brain activity computed over a single spatial scale. Importantly, we found that the neural representation of the same finger movement action performed at different positions within the skill sequence progressively differentiated during rest periods of early learning (i.e. – within the first five minutes of practice). That is, the representation evolved to not only reflect the keypress finger movement, but also where the keypress was performed within the sequence. Furthermore, across participants there was a highly significant linear relationship between the amount of differentiation that occurred and their overall performance gains. Importantly, additional testing on the following day confirmed that this differentiation was retained at least 24 hours later (indicating stability of the formed skill memory) and was specific to the practiced skill sequence (i.e. – the differentiation disappeared when other skill sequences were performed). Our findings that sequence action representations contextually differentiate during early skill learning has important implications for brain-computer interface (BCI) applications used for rehabilitation of motor deficits in patients suffering from brain or spinal cord injury or disease. Incorporating this knowledge into BCI application architecture will be important for improving their performance and utility in clinical and home-based care settings. A second new finding from FY2025 is currently under review at the journal, NeuroImage (https://doi.org/10.1101/2024.11.20.624544). Using a similar experimental paradigm to the study described above, we investigated the specific brain regions driving reorganization of functional brain activity patterns observed during early motor skill learning. Again, we recorded brain activity using MEG while participants learned a novel, moderately difficult, keypress sequence performed with their non-dominant left hand. We used regional source-space estimates to characterize a whole-brain functional network graph describe how individual brain regions interacted with one another. That is, regions displaying temporal activity patterns that were highly correlated are considered to be “functionally connected” regions that actively share information. A small set of regions that show the highest degree of functional connectivity across the whole brain are termed, “hubs”, and are likely to play outsized roles in network reorganization. Importantly, we found that the hub strength of the anterior cingulate cortex (ACC; a region known to be important for detecting errors) and the caudate nucleus (a region of the basal ganglia involved in reinforcement learning) positively correlated with skill performance gains observed across participants. These hub regions were strongly integrated with the hippocampus, parahippocampal cortex, and lingual and fusiform gyri—all important areas for memory formation and consolidation. These findings indicate that the ACC and caudate are targets of interest for non-invasive brain stimulation approaches aimed at facilitating network reorganization and skill learning in neurorehabilitation settings. Finally, a successful collaboration with CUNY culminated with one FY2025 publication published in Imaging Neuroscience (https://doi.org/10.1162/imag_a_00431). Multiple studies have demonstrated that transcranial direct current stimulation (tDCS) of the primary motor cortex (M1) can influence corticospinal excitability and motor skill acquisition. In this preregistered study, 120 participants engaged in a motor skill learning task while receiving tDCS with posterior-to-anterior currents through M1. We employed a double-blind, between-subjects design, with groups of 4 mA, 6 mA, or sham stimulation, while ensuring balanced groups in terms of typing speed. Cortical excitability was assessed via motor-evoked potentials (MEPs) and TMS-evoked potentials (TEPs) before and after motor skill learning with concurrent tDCS. tDCS at these higher intensities was well tolerated, and motor learning correlated with pretraining typing speed. Planned analyses found no dose-response effect of tDCS on motor skill performance or MEP amplitude. This suggests that, under our experimental conditions, tDCS did not significantly modulate motor skill learning or corticospinal excitability. Furthermore, there was no correlation between motor performance and MEP, and thus no evidence for a common neural substrate. Exploratory analyses found an increase in MEP and TEP amplitudes following the sequence learning task. Motor skill gains positively correlated with TEP changes over the stimulated M1, which were more negative with increasing tDCS intensity. The effects of tDCS on motor skill learning and MEPs, if they exist, may require particular experimental conditions that have not been tested here.