Cognition, body, and mind: A three‐in‐one coordinate‐based fMRI meta‐analysis on cognitive, physical, and meditative trainings

“

TL;DR

This meta-analysis of fMRI studies found that cognitive, physical, and meditative training each increase brain activity in distinct but neighboring regions of the medial prefrontal cortex, specifically the dorsal midcingulate cortex, pre-supplementary motor area, and dorsal anterior cingulate cortex, respectively. These regions are linked to different networks supporting performance monitoring, action control, and attentional control. The findings suggest that combining these training types could enhance cognitive function more effectively than any single approach, with implications for designing interventions to boost cognitive reserve and prevent decline.

Full summary

4,991 CHARS

**Background:** The increasing prevalence of age-related cognitive decline has spurred interest in non-pharmacological 'brain training' interventions. Cognitive, physical, and meditative trainings are popular approaches, but their comparative efficacy and underlying neural mechanisms remain unclear. This study aimed to meta-analyze fMRI studies to identify the distinct and overlapping neural correlates of these three training types in healthy, non-expert individuals, and to assess their effects on cognitive performance.

**Methods:** The authors conducted a coordinate-based meta-analysis using Activation Likelihood Estimation (ALE) on fMRI studies published up to December 2022. Separate literature searches identified studies on cognitive training (n=22 studies, 22 experiments), physical training (n=22 studies, 22 experiments), and meditative training (n=20 studies, 20 experiments). Inclusion criteria required studies on healthy adults (18-99 years), reporting whole-brain task-related fMRI coordinates for pre- vs. post-training contrasts. Studies on experts, clinical populations, or using ROI/SVC analyses were excluded. ALE analyses identified consistent clusters of activation change for each training type. Conjunction and contrast analyses compared activations across training types. Functional connectivity of resulting clusters was assessed using Neurosynth. Publication bias was evaluated via correlation between sample size and foci, and fail-safe N (FSN) analysis. For behavioral effects, a separate meta-analysis of a subset of studies (cognitive: 20, physical: 14, meditative: 5) reporting pre/post cognitive measures was performed using random-effects models to calculate pooled Cohen's d effect sizes.

**Key Results:** - **Neuroimaging:** Each training type was associated with increased brain activity in distinct but neighboring medial prefrontal regions. Cognitive training engaged the anterior and posterior dorsal midcingulate cortex (adMCC/pdMCC; peak MNI: -8, 16, 50; z=5.32, p<.001). Physical training engaged the pdMCC and pre-supplementary motor area (pre-SMA; peak MNI: -7, 4, 51; z=4.90, p<.001). Meditative training engaged the dorsal anterior cingulate cortex (dACC; peak MNI: -9, 31, 28; z=4.18, p<.001). No significant overlapping activations were found between any pair of training types in conjunction analyses. Direct contrasts confirmed these specificities (e.g., Cognitive > Physical: -10, 22, 46, z=2.52, p=.006; Meditative > Cognitive: -10, 31, 28, z=2.03, p=.021). Functional connectivity maps showed that the cognitive training cluster coactivated with the salience network (dorsal anterior insula, lateral prefrontal cortex), the physical training cluster with sensorimotor and fronto-parietal networks, and the meditative training cluster with the frontopolar cortex and posterior cingulate. FSN analysis confirmed robustness against publication bias (cognitive: 7<n<29; physical: 6<n<30; meditative: 5<n<24). - **Behavioral:** Cognitive training showed a significant large positive effect on cognitive performance (d=0.83, 95% CI [0.5357, 1.1292], p<.0001), with high heterogeneity (I²=88.4%). Physical training showed a significant moderate positive effect (d=0.54, 95% CI [0.3258, 0.7467], p=.0001), with moderate heterogeneity (I²=32.9%). Meditative training showed a non-significant effect (d=-0.18, 95% CI [-2.3291, 1.9584], p=.8221), with high heterogeneity (I²=86.3%). Moderator analyses for cognitive training revealed significant subgroup differences for Gender (Q=13.81, p=.001) and Experimental Design (Q=11.77, p=.003). For physical training, significant subgroup differences were found for Age (Q=7.66, p=.005; larger effect in >40 subgroup: d=0.69) and Training Duration (Q=6.25, p=.001; larger effect for long duration).

**Clinical Implications:** The study provides a neural framework for understanding how different training types enhance cognitive control. The distinct engagement of fronto-medial regions (adMCC for monitoring, pre-SMA for action control, dACC for attentional/emotional control) suggests that combined interventions could synergistically boost cognitive function by targeting multiple control processes. The behavioral data support the efficacy of cognitive and physical training for improving cognition, with physical training showing particular benefit for older adults. The lack of significant behavioral effect for meditative training may be due to the small number of studies with cognitive outcome measures, highlighting a critical gap in the literature. The findings advocate for standardized neurocognitive and neuroimaging protocols to enable more robust comparisons and inform the development of tailored, combined interventions (e.g., cognitive + physical, or cognitive + meditative) for healthy aging and clinical populations. The identified clusters also represent potential targets for neuromodulation techniques like transcranial electrical stimulation to enhance training effects.