Kral and colleagues at the University of Wisconsin-Madison ran the most rigorous test of meditation and brain structure to date. They combined two three-arm randomized controlled trials with a total of 218 meditation-naive participants: 75 assigned to MBSR, 73 to an active control program, and 70 to a waitlist. They measured gray matter volume, gray matter density, and cortical thickness before and after the programs. The result: no evidence that MBSR produced any brain structural changes at either the whole-brain level or in specific brain regions that earlier studies had highlighted (like the hippocampus or insula). This directly failed to replicate the widely cited Holzel 2011 findings. Published in Science Advances (Vol. 8, eabk3316, 2022). This study had far more participants and better controls than any previous study claiming positive results.

Van Dam and 14 co-authors (including Sara Lazar, who published the original 2005 cortical thickness study) published a detailed critique of meditation research. They identified numerous problems: there is no agreed-upon definition of "mindfulness," making it hard to compare studies. Most studies use inadequate control groups. Sample sizes are very small. Follow-up periods are too short. Confounding variables (like diet, exercise, or personality traits that lead people to meditate) are rarely controlled for. They warned that misinformation about meditation may lead consumers to be harmed, misled, and disappointed. They called for much stricter research standards. Published in Perspectives on Psychological Science (Vol. 13, pp. 36-61).

Fox and colleagues, in the same meta-analysis often cited as evidence for meditation changing the brain, included important caveats. They specifically warned that publication bias is a serious concern, meaning studies that find no brain differences are less likely to be published, skewing the overall picture toward positive results. They also noted that many of the reviewed studies used lenient statistical thresholds, which increases the chance of finding differences that are not real (false positives). The diversity of meditation types, brain scanning methods, and participant groups made it very difficult to draw firm conclusions. Published in Neuroscience and Biobehavioral Reviews (Vol. 43, pp. 48-73, 2014).

Coronado-Montoya and colleagues examined publication patterns in mindfulness research. Of 124 published randomized controlled trials, 87% reported at least one positive outcome in the abstract and 88% concluded mindfulness therapy was effective. This rate of positive findings was 1.6 times higher than what would be expected based on the actual effect sizes. Of 21 trials that had been formally registered (meaning researchers announced their plans in advance), 13 (62%) remained unpublished more than 30 months after completion. None of the 36 systematic reviews and meta-analyses they examined acknowledged that effect sizes might be inflated due to this reporting bias. Published in PLOS ONE (Vol. 11, e0153220).

Goldberg and colleagues reviewed 44 meta-analyses covering 336 randomized controlled trials with a combined 30,483 participants. They found a clear pattern: the strength of the comparison group significantly affected the size of the reported effect. When meditation was compared to doing nothing (waitlist), effects looked large. When compared to an active control (like exercise or education), effects shrank dramatically. They also identified widespread overreporting of significant results, very small sample sizes, lack of active controls, and failure to check whether the meditation was actually taught correctly. The review concluded that many reported meditation benefits are likely overestimated. Published in Perspectives on Psychological Science (Vol. 17, pp. 108-130).

The majority of widely cited studies showing meditation-brain associations (Lazar 2005, Luders 2009, Luders 2012, Luders 2015, Pagnoni 2007) used cross-sectional designs, meaning they scanned meditators and non-meditators once and compared them. This design cannot establish that meditation caused any brain differences. It is equally possible that people born with thicker prefrontal cortex or larger hippocampi are more naturally inclined toward meditation and more likely to sustain a long-term practice. As Fox and colleagues noted in their 2014 meta-analysis: "The critical question of whether meditation is causing the observed differences, or whether pre-existing differences make certain people more likely to engage in intensive meditation" remains unanswered by these study designs.

The most widely cited studies showing meditation changes brain structure had notably small participant groups: Lazar 2005 had 20 meditators, Holzel 2011 had 16 meditators plus 17 controls, Pagnoni 2007 had 13 per group, and Holzel 2010 had 26 total. Small samples in brain imaging studies are known to inflate effect sizes and increase the chance of false positives. When Kral and colleagues (2022) conducted the first large randomized controlled study with at least 70 people per group and proper controls, they found no brain structural changes from meditation. Fox and colleagues (2014) explicitly warned about "the lack of statistical power caused by small samples found in many neuroimaging studies - a problem especially prevalent in the study of meditation practitioners."

A meta-analysis published in Scientific Reports in October 2023 analyzed 11 studies with 581 total participants and claimed to find consistent brain structural changes from mindfulness randomized controlled trials. However, the paper received a retraction notice in 2025 after concerns were raised about how the authors applied their criteria for including and excluding studies. The retraction of a meta-analysis on this exact topic illustrates the fragility of the evidence base and the ongoing methodological challenges in meditation neuroscience research.

Holzel and colleagues'' 2010 study is frequently cited as evidence that meditation shrinks the amygdala. However, the actual findings were more limited than commonly reported. The study found no statistically significant overall main effect of MBSR training on amygdala gray matter density. What they found instead was that individual differences in stress reduction correlated with individual differences in amygdala density changes. The authors themselves stated their results "do not support the conclusion that MBSR training per se leads to decreases in gray matter in this region." This distinction is important: a correlation between two variables within a group is not the same as the training itself causing a change. Published in Social Cognitive and Affective Neuroscience (Vol. 5, pp. 11-17).