5.5citescore ORIGINAL RESEARCH article Front. Hum. Neurosci., 16 March 2015 Sec. Brain Health and Clinical Neuroscience Greater widespread functional connectivity of the caudate in older adults who practice kripalu yoga and vipassana meditation than in controls 1. Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Charlestown, 2. Bender Institute of Neuroimaging, Justus Liebig Universitt Giessen, Giessen Germany Article metr View details Mechanisms Underlying Movement- Contemplativ 17 articles Maxime Taquet Rohan Dixit Britta K. Hlzel Bradford C. Dickerson Sara W. Lazar ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy Cookies Settings Reject non-essential cookies Accept cookies There has been a growing interest in understanding how contemplative practices affect brain functional organization. However, most studies have restricted their exploration to predefined networks. Furthermore, scientific comparisons of different contemplative traditions are largely lacking. Here we explored differences in whole brain resting state functional connectivity between experienced yoga practitioners, experienced meditators, and matched controls. Analyses were repeated in an independent sample of experienced meditators and matched controls. Analyses utilizing Network- Based Statistics () revealed difference components for yoga practitioners > controls and meditators > controls in which the right caudate was a central node. Follow up analyses revealed that yoga practitioners and meditators had significantly greater degree centrality in the caudate than controls. This greater degree centrality was not driven by single connections but by greater connectivity between the caudate and numerous brain regions. Findings of greater caudate connectivity in meditators than in controls was replicated in an independent dataset. These findings suggest that yoga and meditation practitioners have stronger functional connectivity within basal ganglia cortico-thalamic feedback loops than non-practitioners. Although we could not provide evidence for its mechanistic role, this greater connectivity might be related to the often reported effects of meditation and yoga on behavioral flexibility, mental health, and well-being. Introduction and Tables Zalesky et al., 2010 ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy There is a growing interest in the neural correlates of meditation practice. While initial studies focused on the meditative state or the effects of meditation on brain activation during a specific task, more recent studies also have investigated the effect of ongoing regular meditation experience on the resting state of the brain ( ; ). These studies have provided first insights in how meditation affects functional brain connectivity at rest. An important limitation of these studies is that they only investigate differences in connectivity between nodes of the default mode network ( ) without accounting for the complex network structure that these connections underpin. Recent models of the brain as a complex network has furthered the understanding of its resting state and provided robust methods to compare its properties amongst subjects based on graph theory. These methods refrain from comparing the fMRI signal at every voxel, thereby increasing the statistical power of group comparisons. Therefore this approach is particularly useful for studying the brain resting state between groups of healthy subjects, for which differences may be Further, the above-mentioned studies focused only on practitioners of meditation. There is much theoretical debate about how various contemplative practices may be similar or different, both in terms of mechanisms and effects. There is a growing interest in understanding how different contemplative practices compare (). A study directly comparing different practices may therefore provide invaluable insights into the neural processes involved and provide concrete evidence as to how these practices differ, or not. Jang et al., 2010Brewer et al., 2011Kilpatrick et al., 2011Taylor et al., 2012 Buckner et al., Brewer et al., 2011 ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy In a recent, hypothesis driven study, we addressed these issues, and investigated global resting state brain functional network properties of yoga- and meditation practitioners ( ). Here we use explorative methods on the same dataset to investigate local differences in the brain resting state functional networks of individuals with extensive meditation or yoga practice compared to demographically matched controls. Unlike previous studies, we use a data- driven approach to reliably identify the differences in networks between the groups across the entire brain, without limiting ourselves to any a priori sub- network or region and without the need of a specific hypothesis. To strengthen confidence in the main finding, we repeated analyses with a second, independent dataset of experienced meditators and controls. Results will be discussed in the light of recent research on the role of the basal ganglia. Materials and Methods Participants The first study consisted of 47 participants: 16 yoga practitioners, 16 meditation practitioners, and 15 controls. The three groups were matched for age, sex, education, and handedness. Yoga practitioners were primarily trained in the Kripalu Yoga () tradition and had an average of 13,534 (SD = 9,950) hours of yoga experience. Meditators were primarily trained in Vipassana (a.k.a. insight or mindfulness) meditation had an average of 7,458 h (SD = 5,734) of meditation experience. Controls had less than 4 yoga or meditation classes in the past year and less than 10 classes in their lifetime. See Table for Faulds, 2005 Goldstein and Kornfield, 2001 ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy the demographic characteristics of each group. Participants provided written informed consent and were compensated $100 for their time. The study was approved by the Partners Human Research Committee, Massachusetts General Hospital (protocol 2005P001392). Other data from these subjects has been published elsewhere (). Handedness Comparison of demographic variables between controls, yoga practitioners, and meditators for the original dataset. For the replication study we used data of a subset of individuals who participated in a previously published study (). Resting state BOLD data was available for 13 Vipassana meditation practitioners and 16 controls with little or no meditation experience (less than 4 classes in the past year and less than 10 classes in their lifetime). Meditators had an average experience of 4,831 (SD = Gard et al., 2014b Lazar et al., 2005 ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy 3,738) hours. See Table for the demographic characteristics of each group. Participants provided written informed consent and were compensated $100 for their time. The study was approved by the Partners Human Research Committee, Massachusetts General Hospital Handedness Comparison of demographic variables between controls, and meditators for the replication dataset. Image Acquisition For the original study data was collected on a Siemens 1.5 Tesla Avanto MRI scanner (Erlagen, Germany) at the Martinos Center for Biomedical Imaging. Structural images were acquired using a T1-weighted magnetization prepared rapid acquisition gradient echo (MPRAGE) sequence (128 sagittal slices, slice thickness = 1.33 mm, TR = field of view = 256 mm 256 mm, matrix = 192 mm 192 mm). A 5 min ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy functional resting state scan was acquired using a gradient echo T2- weighted sequence (TR = 2.5 s, TE = 40 ms, FA = 90, field of view = 320 mm Twenty five sagittal slices with 1 mm gap (voxel size: 3.13 mm 3.13 mm 5 mm) were acquired inter-leaved. For the replication study data was collected on a Siemens 1.5 Tesla Sonata MRI scanner (Erlagen, Germany) at the Martinos Center for Biomedical Imaging. Structural images were acquired using a T1-weighted magnetization prepared rapid acquisition gradient echo (MPRAGE) sequence (128 sagittal slices, slice thickness = 1.33 mm, TR = 2.73 s, TE = 3.39 ms, flip angle = 7, field of view = 192 mm). A 6.7 min functional resting state scan was acquired using a gradient echo T2-weighted sequence mm). Twenty-five sagittal slices with 1 mm gap (voxel size: 3.13 mm 3.13 mm 5 mm) were acquired inter-leaved. Participants of both the original and the replication study were instructed not to meditate during the resting state scan. Demographics To test if groups were successfully demographically matched for age and education, ANOVAs and independent sample t-tests (two-tailed) were conducted for the original and the replication study, respectively. To evaluate comparability on gender and handedness, -tests were conducted for both studies. Data Preprocessing For both studies resting state data were slice time corrected, realigned, coregistered to individual T1-weighted ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy images, normalized, and spatially smoothed with at 5 mm kernel using SPM8 (Wellcome Department of Cognitive Neurology, London, UK). Next, in the original study the first eight volumes of the functional time series were discarded to allow for stabilization of the MR signal. The remaining 112 volumes were further preprocessed using the Connectivity toolbox (). In the replication study the first five volumes were discarded and the remaining 95 were further processed in the same way as the data from the original study. Mean white matter signal, mean CSF signal, six motion parameters, and the first order motion derivative were regressed out of the data. Finally, the residual time series were band-pass filtered with a window of 0.0080.09 Hz. Anatomical Parcellation and Time Series Extraction Resting state scans were parcellated into 116 regions of interest (ROIs; 90 cortical and subcortical, and 26 cerebellar) using the Automated Anatomical Labeling (AAL; ) template in the Wake Forest University (WFU) Pickatlas version 2.5 (). For each ROI, the average (of all voxels in the ROI) preprocessed time-series was extracted, resulting in a 116 (ROIs) 112 (volumes) time-series matrix for each subject. Time-series extraction was done with the Connectivity toolbox Network Analysis Networks are defined as a set of nodes connected by links. In the context of functional network analysis, the nodes are each of the 116 regions of interest and the links represent the strength of the connections between them. Anatomical parcellation and time series extraction result in a time-series matrix Whitfield-Gabrieli et al., 2010 Mazoyer et al., 2002 Maldjian et al., 2003 Whitfield-Gabrieli et al., 2010 ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy for each subject. The correlations between each pair of time series of each such correlation matrix were computed, resulting in a 116 116 correlation matrix. The elements of this matrix are therefore real numbers between -1 and 1. All negative entries were set to zero so that all elements belong to [0,1], which is a necessary step to obtain a network with positive weights ( These matrices essentially define networks wherein the (i,j) entry of the matrix is the strength of the connection between the i-th and j-th ROI. These networks are weighted (because the connections can have any value between zero and one) and undirected [because the (i,j) entry of the matrix equal the (j,i) entry]. Our choice of using weighted networks instead of unweighted ones (obtained by further binarizing positive weights) is motivated by () showing that analysis of weighted networks is more reliable and () showing that binarization results in a loss of valuable information. The networks were then analyzed using NetworkX (). Network-Based Statistics One-to-one comparisons between groups for each connections in the network would result in many comparisons to be made. These comparisons may lack statistical power due to the need to correct for multiple comparisons. To explore differences in resting state brain functional connectivity between yoga practitioners, meditators, and controls, while considering the entire brain network, we therefore employed Network-Based Statistics (NBS) which detects clusters of connections (instead of individual connections) that significantly differ between group. NBS is a solution to the multiple comparison problem. NBS and McGonigle, 2011 Wang et al., 2011 Barrat et al., 2004 Hagberg et al., 2008 ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy assumes that edges contributing to population differences tend to appear in connected components ( ). Introducing this assumption decreases the number of comparisons and unveils clusters of edges that significantly differ between the groups. More specifically, we used the NBS method for the comparisons yoga practitioners > controls, meditators > controls, and yoga practitioners versus meditators (two-sided test). For the comparisons involving controls we used one-sided tests, based on previous studies that found greater resting stage connectivity in meditators compared to controls (). Much like cluster-based statistics, NBS requires a threshold on the t-statistics (or equivalently on the p-value) of individual edge differences. Connected components are subsequently defined in the binary network of supra-threshold edges. To explore spatially small, hence interpretable, subnetworks, we used a relatively severe initial threshold of p < Degree Centrality Network-based statistics limits the number of comparisons by automatically and reliably identifying subnetworks of interests. To further investigate the central role that the caudate (central node of the detected subnetworks) plays in the functional networks for yoga practitioners > controls and meditators > controls, degree centrality of the caudate was computed for each subject (Eq. 1). Degree centrality was chosen as it is conceptually the simplest measure of nodal importance in a network. Degree centrality is defined as: Zalesky et al., Brewer et al., 2011 ArticlesResearch TopicsEditorial board Frontiers ... ArticlesResearch TopicsEditorial board All journalsAll articlesSubmit your research We use cookies Our website uses cookies that are essential for its operation and additional cookies to track performance, or to improve and personalize our services. To manage your cookie preferences, please click Cookie Settings. For more information on how we use cookies, please see ourCookie Policy where deg(v) is the weighted degree of the node v (i.e., the sum of the strengths of its connections) and n is the total number of nodes in the network. A larger degree centrality therefore implies that the node is more connected to the rest of the network. We compared the degree centrality of controls, meditators, and yoga practitioners for the left and right caudate nuclei. Since the assumption of homogeneity of variances was not met, we used a Welchs test to assess the equality of means in the population. The test was followed up by independent two-tailed t-tests comparing yoga practitioners, meditators, and controls pairwise. To validate the findings from these analyses, we tested the hypothesis that medi