Introduction

Mind and body practices are complementary health approaches that focus on facilitating brain-behavior connections, which aim to promote health and well-being by enhancing the ability of the mind to regulate our physical bodies for optimal daily functioning1,2. Among many mind and body practices, mind–body exercise (i.e., Tai Chi, qigong and yoga) is a specific type of practice that incorporates meditation into the execution of movement routines to improve body balance and the flexibility and strength of the musculoskeletal structures3. Unlike conventional physical exercise that specifically emphasizes the awareness of bodily movements4, mind–body exercise emphasizes the coordination between breathing, bodily sensation awareness, and bodily movement execution5,6. According to a National Health Interview Survey conducted in the United States7, mind–body exercise is listed as one of the most common complementary health approaches among adults. Due to its popularity, researchers have continuously studied the beneficial effects of mind–body exercise on human health. Numerous meta-analytic reviews show that mind–body practices are effective in promoting motor, cognitive and affective functioning of both healthy and clinical populations. For instance, Tai Chi has been shown to be effective at improving motor control in patients with stroke8 and Parkinson’s disease9, promoting psychological well-being in older adults10 and enhancing cognitive functions in both healthy11,12 and clinical13,14 elderly populations. Similar positive clinical effects have been observed when qigong15,16,17,18 or yoga19,20,21 is performed.

Given the promising clinical effects of mind–body exercise, the neurobiological mechanisms associated with the observed behavioral changes have been rigorously studied in recent years22,23,24. Specifically, functional magnetic resonance imaging (fMRI) has been widely adopted to investigate how mind–body exercise changes brain activation patterns, connections between different regions25 and brain structures, and some studies have shown that mind–body exercise induces brain changes in the frontoparietal regions. For example, Eyre, Acevedo26 showed that a 12-week yoga training program significantly reduced depressive symptoms and improved visuospatial memory in older adults with mild cognitive impairment compared to conventional cognitive training, with enhanced visuospatial memory performance significantly correlated with a reduction in the functional connectivity between the superior and medial parietal cortices that implies greater neural efficiency26 for visual attention and working memory27. Greater activation in the left ventrolateral prefrontal cortex (PFC), a brain region known for inhibitory control28, was found in long-term yoga practitioners compared to age-matched healthy individuals without regular yoga practice when these individuals were presented negative emotional stimuli29. Tao, Chen30 showed that 12-week Tai Chi and qigong practices in older adults increased regional spontaneous neuronal activity in the dorsolateral PFC and the medial PFC, respectively, which was associated with enhanced memory performance. Moreover, significant increases in both whole-brain white matter31 and hippocampal gray matter32 are evident after long-term mind–body practice. However, statistically negative results have also been reported; some studies reported no statistically significant exercise-induced changes in brain activation or functional connectivity patterns were found between the yoga33/Tai Chi34 practice group and the control group.

Notably, the experimental protocols and fMRI outcome measures vary substantially across the existing empirical studies, which might contribute to inconsistent results. For instance, the duration of mind–body practice in different study protocols is highly heterogeneous. Some studies have investigated the neurobiological effects of long-term mind–body practice33,35, but other researchers have examined how several weeks of mind–body practice potentially modify neural activities26,34,36. Moreover, the nature of the control group also varies. In addition, these empirical studies adopted different methods to analyze fMRI data. Some studies investigated the changes in spontaneous neuronal fluctuations before and after mind–body exercises37,38,39, whereas others investigated the changes in neural connectivity in different regions of interest, including the default mode network (DMN)40 and parietal regions26. Collectively, these differences might complicate interpretations of the overall results, hence limiting our understanding of the neurobiological effects of mind–body practice. Systematic reviews of previously published studies and neuroimaging meta-analyses might help us address the limitations described above. Specifically, using seed-based d mapping with permutation of subject images (SDM-PSI)41, the between-study experimental design heterogeneity discussed above could be controlled by covariate analyses and further explored with meta-regression. In addition, registering consistently coactivated brain regions identified through the meta-analysis on a standardized brain atlas may be possible42, which might help us identify which functional brain networks are consistently modulated by mind–body practice. Finally, the overall study power for the identification of consistently coactivated brain regions by mind–body practice may be enhanced by performing a meta-analysis. Given the aforementioned research gaps, this study aimed to investigate the neurobiological changes (i.e., brain activation, neural connections and structural changes in the brain) associated with mind–body exercise by qualitatively and quantitatively examining currently available fMRI data.

Methods

Literature search

The Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines43 were used to guide the implementation of this review (PROSPERO registration number: CRD42021248984; see Table S1 for the PRISMA checklist and access the review protocol via https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=248984). All authors first decided on the keywords and electronic databases to search, followed by the first title/abstract/keyword search of the electronic databases Scopus, Embase, ScienceDirect and PubMed conducted on April 2022 using the keywords (“mindful*” OR “mind–body” OR “yoga” OR “tai chi” OR “qigong”) AND (“magnetic resonance imaging” OR “functional magnetic resonance imaging” OR “MRI” OR “fMRI”) to identify relevant papers. Full search strategies were reported in Table S2. An updated search using the same keywords and search engines was performed in June 2023 to ensure the most up-to-date articles were included in the analysis. References from a previously published review44 were also screened. No limitations on the dates of publication were set.

Study inclusion for the systematic review and meta-analysis

The systematic review included published randomized controlled trials (RCTs) and cross-sectional studies investigating the effects of mind–body exercise compared to regular physical exercise (active control)/waitlist control groups. After removing duplicated records, titles and abstracts of the records were screened. Records that were (1) not empirical studies (e.g., review articles, book chapters, conference proceedings and editorials), (2) nonhuman studies and (3) not involving mind–body practice as an intervention were excluded. During the full-text screening, studies that (1) investigated the brain activations during meditative state only, (2) did not report standard stereotactic coordinate space results (either Talairach or MNI), and (3) did not compare the effects of mind–body exercise with a physical exercise/waitlist control group were excluded. Only studies that fulfilled all inclusion criteria were included in the systematic review. For meta-analysis, only fMRI studies comparing whole-brain seed-based resting-state functional connectivity (rsFC)45 between mind–body exercise and active/waitlist control groups were included. The whole screening process was conducted independently by the co-first, second and third authors and recorded separately on each Excel spreadsheet. The first author decided to include or exclude papers when discrepancies appeared. Corresponding authors of the included papers were contacted if there was missing or unclear information.

Data extraction and recoding

Data were independently extracted from the included papers and recorded by the second and third authors. The extracted data were recorded in an Excel spreadsheet independently by the co-first author, second and third authors and were cross-checked by the first author. Demographic characteristics (i.e., participants’ diagnoses, mean age, ratio of sex), experimental details (i.e., types of mind–body exercise, average duration of practice in years, and control group), and outcome measures (i.e., the MRI paradigm and analysis methods, main outcomes in the active group) were extracted. In particular for studies included in the meta-analysis, as we aimed to pool the peaks with significant group differences between mind–body exercise group and control groups reported in previous studies46,47, seed center coordinates and peak coordinates of between-group differences were extracted from each study.

Data recoding was performed to facilitate covariate analyses in the coordinate-base meta-analysis. First, to recode the seed center coordinates of individual studies in a standardized manner such that covariate analyses were possible, the seed center coordinates were registered within the seven resting-state brain networks reported by studies that used large-scale data from human cerebral48,49, cerebellar50, and striatal51 parcellation studies. The seven functional networks included the default mode network (DMN; coordinates other task-positive networks), somatomotor network (SMN; for motor control and execution), frontoparietal network (FPN; coordinates goal-directed behavior), dorsal attention network (DAN; for top-down attention control), ventral attention network (VAN; detects salient stimuli), limbic network (LIM; for emotional processing), and visual network (VIS; processes incoming visual information). Second, regarding the control group conditions, given some studies utilized control groups with participants performing regular physical exercise (e.g. treadmill walking/usual physical exercise regime) while other studies adopted waitlist control groups (e.g. not doing mind–body exercise while also not engaging in any kinds of physical exercise regime), studies with physical exercise as a control condition were recoded as having an “active control group”, while studies with waitlist control groups were recoded as having a "waitlist control group”.

Quality of reporting appraisal and risk of bias within the studies

The revised Cochrane risk-of-bias tool for randomized trials (RoB 2)52 and Joanna Briggs Institute’s (JBI) critical appraisal checklist for analytical cross-sectional studies53 were used to evaluate the methodological quality of the included RCTs and cross-sectional studies, respectively. Bias arising from the randomization process, bias due to deviations from intended interventions, bias due to missing outcome data, bias in measurement of the outcome, and bias in selection of the reported result were assessed using RoB 2. Three risk of bias categories (low, high or some concerns) were possible for each domain. The overall rating of the five domains yielded the “overall risk-of-bias judgment” for each of the included studies. Using the JBI critical appraisal checklist, the included cross-sectional studies were qualified, as suggested by Ma, Wang54. By counting the number of “yes” responses from the critical appraisal items among the total items of the study, the quality of each study was then assessed. Criteria with “not applicable” responses were excluded, while “unclear” responses were coded as “no” and did not fulfill the quality criteria. “Yes” scores for 0–33% of the JBI items were assigned a “low” overall quality rating, while 34–66% “yes” scores from the JBI questions were assigned a “medium” rating, and 67% or more “yes” scores for the JBI items were assigned a “high” rating.

Data analysis

Narrative synthesis was performed for studies included in the systematic review. Effect sizes (Cohen’s d) and 95% confidence intervals (CI) were calculated using the formula reported in Lipsey and Wilson55 for all included experiments based on the significance threshold adopted by individual papers. the SDM-PSI software version 6.2141was used to pool the brain imaging data for the meta-analysis. The recommended data preprocessing pipeline was used41. Briefly, preprocessing of extracted fMRI coordinate data was performed with anisotropy = 1, isotropic full width at half maximum (FWHM) set to 20 mm, and a voxel size of 2 mm on a gray matter mask. The between-group analysis was performed with rsFC seed network as a covariate to investigate the effects of mind–body exercise on rsFC. Meta-regression analyses with mean age of participants (years) mean years of mind–body practice as independent variable and peak coordinates of between-group differences (mind–body exercise vs. control group) as dependent variable was performed to explore how duration of mind–body practice modulates the changes in brain activation. Regarding the significance threshold, we first identified significant clusters with the threshold p = 0.05 (uncorrected), Z > 1 and a cluster size (k) larger than 10 voxels, followed by performing voxel-wise error corrections using 1000 permutations to obtain the null distribution of cluster sizes that pass the threshold of p < 0.005, and that distribution was used to set a minimum cluster size56. Each of the local peaks within the significant clusters was classified based on their location within the seven functional networks to understand how mind–body exercise modulate the functional brain networks (i.e., visual, somatomotor, dorsal attention, ventral attention, frontoparietal, limbic and default mode networks). The heterogeneity between studies was indicated by I2 test with I2 = 25%, 50% and 75% corresponds to low, medium and high heterogeneity respectively57, which was calculated for statistically significant peak coordinates yielded from the between-group or meta-regression analyses. Harbord Egger bias tests58 were used to assess “small study effects”. Certainty assessment was conducted using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach59.

Results

Study selection

A total of 2600 records were retrieved from multiple electronic databases and references. After the removal of duplicate records, 1437 abstracts were screened. After the exclusion of nonempirical non-fMRI studies and studies that did not involve mind–body interventions, the full text of 487 studies was screened for inclusion in the systematic review. Based on the inclusion and exclusion criteria, 34 fMRI studies (with 42 comparisons) were included, with 26 studies included in the narrative synthesis and 8 studies (with 13 comparisons) included in the meta-analysis. The flow of the article selection process is outlined in Fig. 1.

Figure 1
figure 1

Flowchart of the article screening process.

Full size image

Risk of bias within studies

The summary of the authors’ judgment of the risk of bias of the included RCTs and cross-sectional studies is shown in Fig. 2. For the included RCTs (n = 16; Fig. 2a), nine studies had a low overall risk of bias, while the remaining seven studies warranted some concerns. All studies showed a low risk of bias in deviations from the intended interventions and missing outcome data. One study showed unclear bias in selective reporting of results. Unclear bias in outcome measurement was noted in two studies, and an unclear randomization process was noted in five studies. For the included cross-sectional studies (n = 18; Fig. 2b), four studies were rated as having an unclear overall risk of bias, and four studies were rated as having a high overall risk of bias. All studies showed a low risk of bias in defining participants’ inclusion criteria (D1), the use of valid, reliable and objective measurement tools (D3 and D4), outcome measurement (D7) and the choice of statistical analysis (D8). Four of 18 studies showed an unclear risk of bias in the participants’ demographic characteristics (D2). Five of 18 studies showed unclear risk of bias in the identification of confounding factors (D5). Four studies did not report strategies to control for the stated confounding factors (D6).

Figure 2
figure 2

Risk of bias summaries: review authors’ judgements about each risk of bias item for each included (a) RCT and (b) cross-sectional study.

Full size image

Study characteristics

Table 1 lists the demographic characteristics, experimental design, neuroimaging outcome measures and results of the 35 studies included in this review. Detailed narrative and quantitative syntheses are presented in the subsequent paragraphs.

Table 1 Effects of mind–body exercise on different neuroimaging outcomes (34 studies, 42 experiments).
Full size table

Can mind–body exercise modulate resting-state brain activities?

Twenty-four studies showed that both long-term and short-term mind–body practice induced changes in resting-state brain activities (Table 1). Most of the studies reported that mind–body exercise is capable of modulating the synchronization between different brain regions. Among studies that reported changes in brain synchronization, 13 experiments (reported in 8 studies) were included in the meta-analysis based on the inclusion criteria. In addition, some studies have shown that mind–body exercise modulates spontaneous neuronal fluctuations.

Nine studies reporting changes in rsFC, as indexed by the statistical correlations of BOLD signals between selected brain regions, were included in the narrative synthesis only. These studies showed that mind–body exercise is capable of reducing rsFC of the cortical brain regions within the DMN60, visual network61, as well as between bilateral periaqueductal gray (PAG) and the right medial prefrontal cortex63, while increasing rsFC between the DMN and hippocampus38,62, between PAG and parietotemporal regions63, and increasing rsFC within the DAN35. The remaining three studies showed significant changes in local brain region synchronization, as indexed by changes in regional homogeneity32,64 and nodal efficiency65.

Thirteen experiments reported in eight studies were included in the SDM-PSI coordinate-based meta-analysis33,34,36,40,66,67,68,69. With seed location and control condition as covariates, increased activation was observed in the DMN with a peak at the left anterior cingulate gyri and decreased activation was observed in the VAN with a peak at the left supramarginal gyrus in the mind–body exercise group when compared to the control group (uncorrected p < 0.05; Table 2). These peaks did not survive voxel-corrected significance tests. Meta-regression between brain activation and years of mind–body practice showed that, the activation difference between treatment and control group of right inferior parietal gyri (BA40) increases with increasing years of mind–body practice (voxel-corrected p = 0.003; Fig. 3). Meta-regression of mean age of participants did not reveal significant correlation with brain activation patterns at the whole brain level (p > 0.05). Regarding heterogeneity and small study effects, the left anterior cingulate gyri (I2 < 0.01%; Harbord Egger bias tests p-value = 0.93), left supramarginal gyrus (I2 = 1.79%; Harbord Egger bias tests p-value = 0.94) and right inferior parietal gyri (I2 < 0.01%; Harbord Egger bias tests p-value = 0.99) peaks showed low heterogeneity and nonsignificant small study effects.

Table 2 Effects of mind–body exercise on the functional connectivity of resting-state networks.
Full size table
Figure 3
figure 3

(a) The activation of right inferior parietal gyri (BA40, peak voxel indicated by ‘+’ symbol), which lies within the default mode network (red), increases with years of mind–body practice (voxel-corrected p < 0.005). (b) A scattered plot showing the positive relationship between the activation of right inferior parietal gyri and years of mind–body practice.

Full size image

Regarding spontaneous neuronal fluctuations that are indicated by the fractional amplitude of low-frequency fluctuations (fALFF), five studies reported significant changes in fALFF after mind–body exercise. Three studies reported changes in fALFF in multiple regions in the frontal, temporal and hippocampal regions after several weeks of mind–body exercise30,37,38,99, and one study reported network-level changes in fALFF after more than 14 years of Tai Chi practice39.

Can mind–body exercise modulate brain activities associated with task performance?

Seven studies reported mind–body exercise-associated changes in brain activation (indicated by changes in BOLD signals) and functional connectivity in practitioners when they performed cognitive and affective tasks (Table 1). Regarding changes in BOLD signals, both short-term and long-term yoga practice induce changes in brain changes in the frontoparietal areas when participants perform various affective tasks. For instance, long-term yoga practice was found to induce a decrease in the activity of the right dlPFC during emotional picture viewing and an increase in activity of the left vlPFC during the emotional Stroop task29, while short-term practitioners were shown to have increased activation in the postcentral gyrus, left superior parietal lobule and right anterior supramarginal gyrus during emotional video viewing tasks70. Changes in cortical and subcortical activation patterns were also evident when long-term yoga practitioners engaged in the color Stroop task and tasks with low (i.e., 1-back) and high (i.e., 2-back) working memory loads71. Some evidence has revealed that short-term Tai Chi practice induces changes in frontal activations during task switching8. Regarding changes in task-based FC, long-term yoga practice strengthens FC within the frontoparietal network during the color Stroop task. During working memory tasks, FC is strengthened between the PFC and various brain regions, including the parietal brain areas, cerebellum and insula, while it is reduced within parietal brain areas, cerebellum and insula71. Some evidence revealed that long-term Tai Chi practice induced stronger frontostriatal FC during a decision-making task embedded with emotional components72.

Can mind–body exercise induce structural changes in the brain?

Eight studies showed that mind–body exercise induced structural changes in the brain, with 7 studies indexed these structural changes in terms of gray matter volume. Multiple studies showed that short-term (i.e., 3 months) Qigong (Baduanjin) practice promotes an increase in gray matter volume the right hemisphere, namely the right precentral and middle occipital gyri74, as well as in the right hippocampus38,98, while for long-term Tai Chi and Yoga practitioners32,73, the gray matter volume is increased in multiple cortical and subcortical areas, including frontal, temporal and occipital lobes, as well as limbic, parahippocampal areas and cerebellum. In contrast, short-term yoga practice is found to induce grey matter reduction in frontoparietal75, as well as temporal regions76. One study showed that long-term Tai Chi practice promotes the overall integrity and efficiency of the white matter network, as evidenced by an increase in the normalized clustering coefficient and the characteristic path length31.

Certainty assessment

A summary of findings and certainty assessment was reported in Table 3. Overall speaking, the certainty of the evidence reported above was low, mainly due to small number of studies, indirectness and inconsistencies.

Table 3 Summary of findings and certainty assessment.
Full size table

Discussion

By systematically synthesizing the current fMRI literature, this review aimed to examine the effects of mind–body exercise on brain activation, functional neural connections and structural changes in the brain. After performing a systematic literature search, 34 empirical studies were included for either narrative or meta-analytic review. The following results were obtained from the qualitative and quantitative analyses of fMRI studies: (1) mind–body exercise modulates the rsFC of the DMN, VAN and DAN; (2) mind–body exercise-induced changes in activation patterns in the frontal regions, as well as the changes in frontoparietal functional connections; (3) changes in spontaneous neuronal fluctuations at rest in frontal, temporal and hippocampal regions are evident in short-term mind–body practice practitioners; and 4) short-term Qigong and long-term Tai Chi/Yoga practice induces changes in the gray matter volume in various cortical and subcortical brain regions. The clinical implications of these results are discussed in the subsequent paragraphs.

Mind–body exercise modulates functional connections in the brain at rest, as evidenced by changes in correlations of BOLD signals between brain regions, regional homogeneity and nodal efficiency. Specifically, the functional connections in brain regions within the DMN were found to be enhanced when compared to control conditions, and importantly, the increase in activation of a brain region within DMN (i.e. right inferior parietal gyri) is shown to be significantly associated with years of mind–body practice. The DMN has long been suggested to be responsible for self-awareness77 and the coordination of task-positive networks78. Previous studies have shown that the functional connections between the DMN and task-positive functional networks increase only when an individual is preparing to perform cognitive tasks79. The capability of mind–body exercise to enhance rsFC within the DMN may imply that mind–body exercise might promote the readiness to perform goal-directed tasks of an individual. It is interesting to note that the SDM-Z value of the meta-regression is negative at the Y-axis, and one might be tempted to associate this with the possible attenuation effect on the well-documented age-related decline in the right inferior parietal gyri80,81,82. We recommend cautious interpretation of results given the number of studies are limited for the meta-regression. Further studies are needed to explore how mind–body exercises maintain the function of the parietal network and to examine the associations between behavioral improvements and mind–body exercise-induced changes in DMN.

The systematic review and meta-analysis revealed that mind–body exercise modulates the functional connections of dorsal (DAN) and ventral (VAN) attention networks at rest. The DAN and VAN are the two brain functional networks that are established to be responsible for attention and cognitive control83,84. While the DAN, a network that exerts top-town control, is responsible for regulating the processing of external stimuli in facilitation of the successful engagement of goal-directed behavior, the VAN is a “bottom-up” network that is responsible for the detection of behaviorally relevant stimuli85. These two networks interact with each other during attention shifting; specifically, the VAN interrupts the activation of DAN to allow an individual to reorient his attention to a new external stimulus84. As revealed by our results, the upregulation of DAN and downregulation of VAN at rest may imply that mind–body exercise reduces the reactivity towards external stimuli while also enhancing the ability to focus on goal-directed behaviors. Indeed, this postulation is consistent with previous findings showing that mindfulness-based interventions reduce reactivity to temptations (e.g. food cues)86 and enhance attentional control87 in practitioners.

Apart from resting-state neural connections, mind–body exercise also modulates brain activation patterns and functional connectivity when individuals perform cognitive/affective tasks. Notably, mind–body exercise specifically modulates task-related brain activation patterns in the frontal cortex and the functional connections between frontal and other brain regions. The effect of abnormal frontal lobe activation patterns on behaviors has been extensively studied in individuals with multiple neurodevelopmental, neuropsychiatric and neurodegenerative diagnoses, such as autism88, depression89, and dementia90. As revealed in the currently available literature, mind–body exercise appears to alter brain activation patterns specifically in the prefrontal brain regions, which signifies its applicability as a treatment for patients with frontal abnormalities. Indeed, previous clinical studies have provided positive empirical evidence supporting its applications2,91. Notably, the studies that investigated the effects of mind–body exercise on brain activity during active task performance are heterogeneous in terms of study designs. More studies involving similar cognitive/affective tasks are needed to support meta-analytic synthesis, which will help us further understand how mind–body exercise modulates task-based brain activities and functional connections.

In addition, spontaneous neuronal fluctuation at rest, as indexed by fALFF changes, is modulated by short-term mind–body exercise. Because fALFF has been shown to be associated with cognitive control abilities92,93, changes in fALFF in multiple brain regions may imply a network-based modulation of neuronal activities, although further investigations are needed to dissect the relationship between regional-specific fALFF changes and behavioral enhancement. Moreover, some evidence shows that short-term Qigong and long-term mind–body exercise may increase gray matter volume in various brain regions, including the frontal, temporal and occipital lobes, as well as limbic and parahippocampal areas and cerebellum. These brain regions are involved in various cognitive processes, which have been shown to shrink with increasing age94,95, and the reduction in volume is associated with age-related functional decline96. Long-term mind–body practice reverses age-related neural degeneration, implying that these exercises may play a role in delaying aging in the general population. The finding that short-term Qigong promotes grey matter volume changes in both frontal and hippocampal regions is encouraging; future studies should investigate the effects of long-term Qigong practice on grey matter increment. Future cross-sectional studies might consider studying the associations between years of training and changes in gray matter volume to determine the length of time participants must engage in mind–body exercise practice to yield observable changes in brain structure, and longitudinal studies may help address whether the observed changes in gray matter volume are long lasting.

Limitations

To the best of our knowledge, this was the first study that comprehensively investigated the effects of mind–body exercise on brain activation, neural connectivity and brain structures by means of narrative synthesis and coordinate-based meta-analysis, yet several limitations were noted. Although we attempted to retrieve all possible published fMRI studies that investigated the effects of mind–body exercise on brain activities and narratively synthesized studies that did not report whole-brain rsFC between-group difference, only 13 experiments were included for coordinate-based meta-analysis, which limited the power of our analysis, resulting in small effect sizes observed in the meta-analysis. Regarding the study characteristics, although we attempted to control for the between-study heterogeneity by conducting covariate analyses in the meta-analysis, we were aware of the unaddressed heterogeneity across studies in both narrative synthesis and meta-analysis, for example the heterogeneity induced by the inclusion of both RCTs and cross-sectional studies, participants of different age and biological sexes. We encourage future neuroimaging studies examining the effects of mind–body exercise at the whole-brain level, such that larger-scale meta-analytic reviews that controls for the between-study heterogeneity could be performed, which would deepen our understanding about how mind–body exercise can promote functional changes in different regions of the brain. In addition, researchers may also consider investigating the specific effects of mind–body exercise on DMN, VAN, DAN and their behavioral correlates. Last but not least, it is critical to study the differential effects of meditation-only, conventional-exercise-only, and mind–body exercise on the brain in future studies.

Conclusion

This review examined the neurobiological effects of mind–body exercise on brain activation, functional neural connections and structural changes in the brain. A systematic literature search yielded 34 relevant empirical studies that were included in the review, and data from 13 fMRI experiments were included in the meta-analysis. The results show that mind–body exercise modulates the rsFC of task-negative and attentional control networks, while also changes frontal activation patterns and frontoparietal functional connections during various cognitive tasks. Additionally, preliminary data show that short-term mind–body practice alters spontaneous neuronal fluctuations at rest in frontal, temporal and hippocampal regions, and short-term Qigong practice is further shown to induce both cortical and subcortical grey matter increment. We recommend that future studies include both neuropsychological and neurophysiological/neuroimaging techniques to further understand the neural mechanisms underpinning mind–body exercise.