What Can Brain Imaging Tell Us About Motor Learning
Post on 21-Jul-2016
What Can Brain Imaging Tell Us About Motor Learning? Chair: Joern Diedrichsen, Motor Control Group, Institute of Cognitive Neuroscience, University College London, London, UK What happens in the human brain when we learn new motor skills? Human neuroimaging should play a key role in answering this question. However, despite hundreds of published studies, we have learned disappointingly little about the neuronal processes underlying learning. In this symposium we will try to point out the pitfalls in the study of motor learning, and to identify the most promising lines of recent research. We believe that there are 3 future challenges that will be the key to novel discoveries in motor learning. First, it will be necessary to tightly integrate anatomical, functional and neuro-chemical imaging methods to understand the underlying neuronal changes. Secondly, we believe it is time to develop linking theories that tell us which neuronal changes to expect with learning and how these changes should become manifest in the measured signals. Finally, to test such theories, we need to progress from a simple description of signal changes to an approach, in which we are able to predict future behavioral performance from current neuronal measures. We will show how white and gray matter structures change under the influence of expertise, how learning can be influenced through transcranial stimulation, and how difference in the responsiveness of certain neurotransmitter systems can predict inter-individual differences in learning. For functional MRI, we will investigate how learning changes neuronal representation and network connectivity. Our discussant, John W. Krakauer will then challenge the speakers to identify the main theoretical questions for future research. Learning Objectives: Having completed this symposium, participants will be able to:
1. Learn about anatomical and functional brain changes induced by motor learning; 2. Recognize the problems and challenges of studying learning using brain imaging; and 3. Identify novel research strategies that address these problems.
Dynamic Brain Correlates of Dexterity and Motor Skill Acquisition Traced with Structural Magnetic Resonance Imaging Hartwig Roman Siebner, Danish Research Center for Magnetic Resonance, Department of MR (DRCMR), Copenhagen University, Hospital Hvidovre, Hvidovre, Denmark The aim of this presentation is to illustrate how magnetic resonance imaging (MRI) can be used to identify structural brain correlates of motor skill acquisition. To highlight its potential, the results of four structural MRI studies will be presented: (i) Diffusion-tensor MRI in children established a link between inter-individual variations in motor performance and regional microstructure (as indexed by regional diffusivity or fractional anisotropy) in the maturing motor system. (ii) In adult individuals, voxel-based morphometry (VBM) of T1-weighted MRIs yielded distinct changes in subcortical grey matter, which correlated with the individual level of musical skill in professional pianists. (iii) Another VBM study on converted left-handers showed that manipulation of dexterity when children learn to write leaves a structural trace in the adult basal ganglia. (iv) Finally, a longitudinal VBM study in patients with focal hand dystonia revealed that immobilisation followed by motor training induced bi-directional changes of grey matter volume. These structural dynamics were paralleled by changes in corticomotor excitability as revealed by transcranial magnetic stimulation. Together, these results show that motor skill and experience dynamically shape regional human brain structure in the motor system. These structural dynamics can be non-invasively traced with structural MRI.
Dynamic Changes in Neurochemistry and Brain Structure with Learning and Brain Stimulation Heidi Johansen-Berg, Oxford Centre for Functional MRI of the Brain (FMRIB), Nuffield Department of Clinical Neurosciences, Oxford, UK This talk will discuss brain correlates of individual differences in motor learning behaviour and will explore ways in which behavioural training and brain stimulation could be used to alter brain structure and to improve learning and recovery. I will present data showing that individual differences in motor behaviour and motor learning can be predicted by variation in white and gray matter structure and also by variation in the responsiveness of the GABA system, as measured by MR spectroscopy. However, these behaviours and brain measures are not fixed; interventions can be used to alter learning and to alter brain structure in both healthy and clinical populations. For example, I will present studies demonstrating that transcranial direct current stimulation of motor cortex accelerates learning and has potential to improve recovery from stroke by altering the balance of cortical excitability. Finally, I will discuss studies showing how behavioural interventions, such as repeated practice, can alter white and gray matter structure and can increase functional recruitment of areas that have been structurally compromised as a result of stroke. Together, these results show how neuroimaging can track dynamic variations in brain structure and neurochemistry relevant to learning, development and recovery. Predicting Learning Based on Large-Scale Network Dynamics in fMRI Scott T. Grafton, UCSB Brain Imaging Center and the Department of Psychological & Brain Sciences, University of California, Santa Barbara, CA, USA Human skill learning requires change in the brain to support novel behaviors such as sequencing, deftness and synergy. Learning dependent changes are commonly observed as regional increases or decreases of brain activity. Change could also occur across large-scale networks irrespective of regional change. Using task dependent fMRI, correlation between cortical or subcortical nodes serves as a proxy for network connectivity. Previous studies show pair-wise changes of correlation weights between isolated regions with skill training. However, the interpretation of isolated changes within larger networks remains ambiguous and tenuous, given the risk for type II error. Here we propose a novel approach for modeling large-scale networks that characterizes an entire network in terms of global parameters. Changes in these global properties are used to make behavioral predictions. To do this, we developed a general statistical framework for the identification of modular architectures in evolving systems that is broadly applicable to disciplines including motor learning, where network adaptability is crucial to the understanding of system performance. Using sequential 3-minute measures of functional connectivity from brain activity, acquired from initial training through mastery of a motor sequencing skill, a strong test of this network-based modeling was considered. We determined if a multi-scale based network properties can be used to predict individual differences in the capacity to learn in future training sessions. Flexibility, measured as the allegiance of nodes to modules summed one experimental session could predict the relative amount of learning in a future session. Similar prediction could not be achieved by pair-wise correlations or univariate measures of regional brain activity. The results demonstrate significant, behaviorally relevant information in global network properties that cannot be observed at the local level. Motor Learning: A Change in Neuronal Representation, Rather than in Activation Joern Diedrichsen, Motor Control Group, Institute of Cognitive Neuroscience, University College London, London, UK Despite 20 years of functional neuroimaging of motor learning we are still lacking a coherent picture
about how motor skills are acquired in a distributed fashion across the brain. I will argue that the main reason for the impasse in field is that research has mostly focussed on changes in the overall neural activity of regions. The deep-rooted problem with this approach is that it is not clear what type of activation changes we should see when a region that undergoes motor learning. One could argue that such a region should increase its activity through learning, as more neurons become recruited in the production of the behaviour. Conversely, one may argue that the region should become less active, as the behaviour is encoded more efficiently. Finally, both processes may occur simultaneously, making motor learning in some regions invisible to traditional fMRI analysis. Here I suggest that a solution to this problem is to study the changes in representation, rather than in activation. Based on network modelling we predict that a network that learns a specific set of motor behaviours should develop neuronal units that are dedicated to the production of these behaviours. In contrast, untrained behaviours will still rely to a large degree on a shared neural substrate. Thus, patterns of activation in the network for trained behaviours should be more dissimilar to each other than the patterns of activation associated with comparable untrained behaviours. In other words, learning should lead to development of a better (more informative) representation of the trained behaviours, even in the absence of overall activity changes. I show here that changes in the distinctiveness of neuronal representation can be detected with multivariate pattern analysis (MVPA). Our data indicates that, for sequential finger movements, the neural measure of representation (but not of activation) correlates with the level of skill shown naturally by participants, and that these measure increase in the supplementary motor area through training. I will furthermore present novel data on the characteristics of activation patterns of highly skilled individuals (musicians).
Session Topic(s): Brain Stimulation Methods Imaging Methods Learning and Memory Motor Behavior