Edited by:
Reviewed by:
*Correspondence:
†These authors have contributed equally to this work.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Yoga combines postures, breathing, and meditation. Despite reported health benefits, yoga’s effects on the brain have received little study. We used magnetic resonance imaging to compare age-related gray matter (GM) decline in yogis and controls. We also examined the effect of increasing yoga experience and weekly practice on GM volume and assessed which aspects of weekly practice contributed most to brain size. Controls displayed the well documented age-related global brain GM decline while yogis did not, suggesting that yoga contributes to protect the brain against age-related decline. Years of yoga experience correlated mostly with GM volume differences in the left hemisphere (insula, frontal operculum, and orbitofrontal cortex) suggesting that yoga tunes the brain toward a parasympatically driven mode and positive states. The number of hours of weekly practice correlated with GM volume in the primary somatosensory cortex/superior parietal lobule (S1/SPL), precuneus/posterior cingulate cortex (PCC), hippocampus, and primary visual cortex (V1). Commonality analyses indicated that the combination of postures and meditation contributed the most to the size of the hippocampus, precuneus/PCC, and S1/SPL while the combination of meditation and breathing exercises contributed the most to V1 volume. Yoga’s potential neuroprotective effects may provide a neural basis for some of its beneficial effects.
Yoga originates in India and is increasingly practiced by Westerners (
Yoga offers several documented health benefits including, but not limited to, improvement of depressive, anxious and stressful states and the relief of various painful conditions (
In the current report, we revisit our data set to address whether the number of years of yoga experience, the amount of weekly yoga practice, and the different aspects of yoga practice impact specific brain regions. Brain differences related to experience and amount of practice within a group of yoga practitioners would suggest that yoga contributes to changing brain anatomy. Indeed, both short-term and long-term increased training and/or performance have been associated with GM increases in human adults in a wide range of cognitive tasks (
Additionally, if different aspects of yoga practice such as postures, breath control techniques, and meditation contributed differently to brain changes it would further suggest that yoga practice contributes to changing brain anatomy. For example, meditation and physical activity are associated with structural differences in brain regions that do not completely overlap. Meditators were repeatedly shown to have larger hippocampal (
Finally, previous studies have shown that global brain GM declines with age (
This study was approved by McGill University Institutional Review Board. Data were drawn from the same study population described in
Group matching criteria and yoga practice characteristics.
Yogis ( |
Controls ( |
||
---|---|---|---|
Sex | Five males |
Five males |
|
Handedness | Nine right-handed |
Nine right-handed |
|
Age (years) | 37.0 ± 6.6 | 36.7 ± 7.3 | |
Body mass index | 21.6 ± 2.1 | 22.6 ± 2.8 | |
Education (years) | 15.9 ± 1.6 | 15.5 ± 2.1 | |
Exercise (h/week) | 5.2 ± 3.1 | 4.7 ± 3.5 | |
Yoga experience (years) | 9.6 ± 2.8 |
||
Weekly yoga practice (h/week) | 8.6 ± 4.1 |
||
Physical postures (%) |
66 ± 21 |
||
Yoga teachers (N) |
11 |
Subjects participated in one MRI scanning session including a 10-min anatomical scan and a 15-min diffusion tensor imaging (DTI) scan (DTI results are reported elsewhere;
Anatomical images were preprocessed with the VBM8 toolbox
Total GM volume [including the cerebellum and expressed as % of total intracranial volume (TIV)] was correlated with age separately for each group using Pearson correlations in SPSS PASW Statistics 18.0.
For the yoga practitioners, we performed two separate whole-brain VBM regression analyses while controling for age and using the number of years of yoga experience and the weekly amount (hours) of yoga practice as predictors [voxel-wise threshold
Once the significant clusters related to the number of hours of weekly yoga practice were identified in the whole-brain regression analyses, we extracted their volumes (in arbitrary units) for each yoga practitioner using MarsBar toolbox for SPM and used the number of weekly hours devoted to the practice of postures, breath control, and yoga-related meditation to determine which aspect or combination of aspects of the total personal yoga practice (excluding teaching) best predicted the size of the identified brain regions using standard multiple regression analyses (SPSS PASW Statistics 20.0) and commonality analyses (SAS 9.3). Regression commonality analysis enables the partitioning of the
The number of years of yoga experience was correlated with the number of hours spent teaching yoga each week using Pearson correlations in SPSS PASW Statistics 18.0.
In controls, whole brain GM negatively correlated with age [GM volume (
The whole brain regression analyses conducted in yogis revealed that the number of years of yoga experience was positively correlated with GM volume in clusters located in the left mid-insula, left frontal operculum (Brodmann area [BA] 44), right middle temporal gyrus (BA 21) and left OFC (BA 47;
The number of hours devoted weekly to yoga postures, yoga-related meditation, and stand-alone breath control exercises for each yogi are presented in
Correlation matrix associated with the multiple regression analysis.
Postures | Yoga-related meditation | Stand-alone breath control exercises | |
---|---|---|---|
Postures | |||
Yoga-related meditation | |||
S1/SPL | |||
V1 | |||
Hippocampus | |||
Precuneus/PCC |
Multiple regression results.
Adjusted |
Beta | PM | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Postures | 0.852 | 0.726 | 0.644 | 0.007 | 0.549 | 0.024 | 0.591 | 0.349 | 0.441 | 0.194 | 0.694 | 0.481 | 0.324 |
Meditation | 0.852 | 0.726 | 0.644 | 0.005 | 0.372 | 0.144 | 0.729 | 0.531 | 0.262 | 0.069 | 0.856 | 0.732 | 0.271 |
Breath | 0.852 | 0.726 | 0.644 | 0.005 | 0.333 | 0.175 | 0.393 | 0.154 | 0.241 | 0.058 | 0.461 | 0.213 | 0.131 |
Postures | 0.736 | 0.542 | 0.405 | 0.015 | 0.510 | 0.084 | 0.523 | 0.274 | 0.410 | 0.168 | 0.711 | 0.505 | 0.267 |
Meditation | 0.736 | 0.542 | 0.405 | 0.008 | 0.275 | 0.386 | 0.611 | 0.373 | 0.194 | 0.038 | 0.830 | 0.689 | 0.168 |
Breath | 0.736 | 0.542 | 0.405 | 0.010 | 0.318 | 0.307 | 0.337 | 0.114 | 0.230 | 0.053 | 0.458 | 0.210 | 0.107 |
Postures | 0.873 | 0.761 | 0.690 | 0.017 | 0.512 | 0.024 | 0.506 | 0.256 | 0.411 | 0.169 | 0.580 | 0.336 | 0.259 |
Meditation | 0.873 | 0.761 | 0.690 | 0.012 | 0.335 | 0.157 | 0.754 | 0.569 | 0.236 | 0.056 | 0.864 | 0.746 | 0.253 |
Breath | 0.873 | 0.761 | 0.690 | 0.018 | 0.475 | 0.050 | 0.525 | 0.276 | 0.344 | 0.118 | 0.601 | 0.362 | 0.249 |
Postures | 0.902 | 0.813 | 0.757 | 0.006 | 0.337 | 0.076 | 0.554 | 0.307 | 0.271 | 0.073 | 0.614 | 0.377 | 0.187 |
Meditation | 0.902 | 0.813 | 0.757 | 0.013 | 0.704 | 0.005 | 0.854 | 0.729 | 0.497 | 0.247 | 0.947 | 0.896 | 0.601 |
Breath | 0.902 | 0.813 | 0.757 | 0.001 | 0.071 | 0.713 | 0.358 | 0.128 | 0.052 | 0.003 | 0.397 | 0.158 | 0.025 |
The model including the three predictors accounted for approximately 73% of the hippocampal volume variance. The
Commonality matrix.
Coefficient | % TOTAL | |
---|---|---|
Unique to postures (x1) | 0.194 | 26.722 |
Unique to meditation (x2) | 0.069 | 9.504 |
Unique to breath control (x3) | 0.058 | 7.989 |
Common to x1, x2 | 0.308 | 42.424 |
Common to x1, x3 | -0.058 | -7.989 |
Common to x2, x3 | 0.250 | 34.435 |
Common to x1, x2, x3 | -0.095 | -13.085 |
Total | 0.726 | 100 |
Unique to postures (x1) | 0.168 | 30.996 |
Unique to meditation (x2) | 0.038 | 7.011 |
Unique to breath control (x3) | 0.053 | 9.779 |
Common to x1, x2 | 0.223 | 41.144 |
Common to x1, x3 | -0.053 | -9.779 |
Common to x2, x3 | 0.177 | 32.657 |
Common to x1, x2, x3 | -0.064 | -11.808 |
Total | 0.542 | 100 |
Unique to postures (x1) | 0.169 | 22.208 |
Unique to meditation (x2) | 0.056 | 7.359 |
Unique to breath control (x3) | 0.118 | 15.506 |
Common to x1, x2 | 0.260 | 34.166 |
Common to x1, x3 | -0.095 | -12.484 |
Common to x2, x3 | 0.331 | 43.495 |
Common to x1, x2, x3 | -0.078 | -10.250 |
Total | 0.761 | 100 |
Unique to postures (x1) | 0.073 | 8.979 |
Unique to meditation (x2) | 0.246 | 30.258 |
Unique to breath control (x3) | 0.002 | 0.246 |
Common to x1, x2 | 0.366 | 45.018 |
Common to x1, x3 | 0.009 | 1.107 |
Common to x2, x3 | 0.259 | 31.857 |
Common to x1, x2, x3 | -0.142 | -17.466 |
Total | 0.813 | 100 |
The model including the three predictors accounted for approximately 54% of the volume variance of S1/SPL with the number of hours devoted to postures having the greatest influence (highest
The model including the three predictors accounted for about 76% of V1 volume variance. The
The model including the three predictors accounted for approximately 81% of the volume variance of precuneus/PCC. Yoga-related meditation was by far the best predictor of GM precuneus/PCC volume (high beta coefficient, large structure coefficient, and product measure value). This was confirmed by the commonality analysis revealing that nearly a third of the explained variance of the precuneus/PCC GM volume was explained by meditation alone, and 45% by the combination of meditation and postures.
In summary, the combination of postures and meditation contributed the most to the size of the hippocampus, precuneus/PCC, and S1/SPL while the combination of meditation and breath control exercises contributed the most to V1 size (
If we exclude an outlier who had the least amount of yoga experience but was teaching many hours/week and one teacher for whom we had no data about the number of hours of weekly teaching, we find a positive correlation between the number of years of yoga experience and the number of hours spent teaching yoga each week (
These data suggest that yoga practice has a neuroprotective effect against the well-documented age-related whole-brain GM degradation, which was evident in our control group. The data also revealed that increasing experience (years of yoga practice) had a differential effect on the brain than did increasing weekly hours of yoga practice. Whole-brain regression analyses showed that more years of yoga experience was associated with increasing GM volumes in clusters located in the left insula, left frontal operculum, right middle temporal gyrus, and left OFC, while more hours devoted to yoga weekly was associated with increasing GM volumes in the right S1/SPL, left hippocampus, midline precuneus/PCC, and right V1 cortex. Finally, postures, breathing exercises, and meditation contributed differently to the structural changes of the four brain areas associated with the amount of weekly yoga practice, consistent with the different nature of the processing taking place in those structures.
Global brain GM declines with age (
More GM related to long-term experience or skill proficiency have been reported in a number of populations like meditators (
In the current study, most brain regions’ volume correlating with the number of years of yoga experience was located in the left hemisphere. This was also the case in another study of hatha yoga practitioners with strikingly similar age, education, and yoga experience (
The mid-insula is implicated in autonomic integration (
Larger OFC GM volume could be related to better emotional regulation with increasing yoga experience (
It is unclear why larger left frontal operculum volumes (BA 44) were found in the most experienced yogis since traditionally the function of BA 44, a part of Broca’s area, is ascribed to language (see
Finally, more yoga experience was associated with larger GM volume in the right middle temporal gyrus (BA 21). A somewhat comparable region, although in the left hemisphere (peak
Short-term activity-dependent plasticity in the adult human brain is a known phenomenon and can occur very rapidly. These activity-dependent GM changes are accompanied by perceptual and/or performance changes, and usually regress shortly after the activity is terminated. For example, it was previously demonstrated that short-term but repeated presentation of painful stimuli over 8 days increases both pain thresholds and GM density in S1 cortex and that these changes recede after regular nociceptive input is discontinued (
Yoga involves interoceptive awareness and focused attention. During the practice of yoga postures, attention is consciously directed to the breath, body alignment and position, and emotional state. This might be related to alterations in S1 cortex, which contains a representational map of the entire body receiving increasing sensory input with increasing hours of weekly practice, as well as alterations in SPL, involved in the voluntary orienting of attention (
Larger GM volume in V1 with increasing weekly practice may be related to some yogic practices involving visualization of objects or scenery, such as some meditation/relaxation techniques including
Spending more time doing yoga each week was associated with larger left hippocampal GM volume. Previous studies found that both hippocampi were activated in yoga teachers during
Together these findings suggest that more is better when it comes to the frequency of yoga practice notably as far as somatosensation, attention, self-relevant processing, and stress regulation are concerned. This dose-response should be taken into account in eventual longitudinal studies evaluating the effects of yoga practice on regional brain volume changes.
Our relatively small sample size may have prevented us from identifying subtler GM differences. Matching groups on the amount of exercise performed outside yoga may be viewed as a limitation given that postures may be considered a form of physical exercise and, as mentioned before, physical activity and cardiovascular function can impact brain structure. The global GM differences observed between groups could hypothetically be entirely attributed to this extra amount of physical activity. However, postures are an integral part of yoga practice and are more than a simple physical activity, as practitioners are trained to do them with mindful awareness. Further, it can be argued that a relatively small proportion of
The use of a higher-resolution head coil (such as the recently developed 32-channel coil) might have permitted us to detect more subtle GM differences, though recent research (
Despite the cross-sectional nature of this study, the current findings (the correlations between GM volume of several brain areas and yoga experience and practice frequency, the relationship between different aspects of yoga practice and changes in different brain areas, and the possible evidence against age-related whole-brain GM decline in yogis) suggest that yoga practice contributes to the observed brain differences. However, we cannot totally exclude that our sample of yogis had fundamentally different brains to begin with, predisposing them to adopt yoga practice and/or persevering on that path. In fact, some of our results suggest that yoga practitioners might have started with smaller brain volumes in certain brain areas before reaching higher levels of experience. This was particularly evident for the brain areas correlating with the number of years of experience. For example,
In conclusion, regular practice of yoga may have neuroprotective effects against whole brain age-related GM decline. Additionally, our results suggest that more weekly regular yoga practice is associated with larger brain volume in areas involved in bodily representation, attention, self-relevant processing, visualization, and stress regulation. Distinct components of yoga practice (postures, breathing exercises, and meditation) or combination of these predicted GM volumes of these brain areas differently, in keeping with the nature of the processing taking place in those structures. Furthermore, certain brain changes continue to occur after several years of practice, as reflected by the link between increasing yoga experience and increasing brain volume in areas subserving autonomic integration, emotional processing and regulation, hierarchical sequential organization, and in a brain area implicated in either the monitoring of the transition between innocuous to painful sensation or in experiences characterized by insights into the unity of all reality and feelings of peace and joy. Most of these experience-related changes were located in the left hemisphere suggesting that increasing years of yoga practice progressively tunes the brain toward a parasympathetically driven mode and positive affective states. Together these findings provide a neural basis for some of the beneficial effects of yoga. Finally, the current study involved yoga practitioners who were otherwise typical North Americans. As such, if the observed structural brain variances are indeed related to yoga training, they should be within the reach of the average person and not reserved to a select few.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This research was supported in part by the Intramural Research Program of the NIH, National Center for Complementary and Alternative Medicine. We thank the team at the McConnell Brain Imaging Centre of the Montreal Neurological Institute for expert MRI data acquisition.