Since the early 2000s, neuroscience has produced a large literature on meditation, a smaller literature criticising it, and a very large quantity of popular reporting that overstates both.

This article tries to separate what is reasonably established from what is plausible, what is contested, and what is simply marketing.

How the field started #

Two strands converged.

The Mind and Life Institute, founded in 1987 by the fourteenth Dalai Lama, the neuroscientist Francisco Varela, and the entrepreneur Adam Engle, established regular dialogues between Buddhist practitioners and scientists and actively encouraged research. The Dalai Lama’s stated position — that if science conclusively disproves a Buddhist claim, the claim must go — gave the collaboration an unusual footing.

Clinical mindfulness research (MBSR and MBCT) created demand for a mechanism. If eight weeks of practice reduces relapse in depression, something is presumably happening in the brain.

Early studies that drew attention include Richard Davidson and Kabat-Zinn’s 2003 work reporting left-sided prefrontal activation and improved antibody response after MBSR, and Antoine Lutz and Davidson’s 2004 study finding unusually high-amplitude gamma synchrony in long-term Tibetan practitioners during compassion meditation.

What is reasonably well supported #

Meditation produces measurable acute state changes. During practice, EEG and fMRI show consistent differences from rest: changes in alpha and theta activity, altered connectivity, and — in experienced practitioners — distinctive gamma patterns. This is not in serious dispute.

Attention networks are engaged and, with practice, change. Studies of focused-attention practice show recruitment of the dorsolateral prefrontal cortex and anterior cingulate, regions associated with sustained attention and conflict monitoring. Longitudinal studies, including intensive retreat studies such as the Shamatha Project, find improvements in sustained attention and reductions in attentional lapses that persist for months.

The default mode network behaves differently. The DMN — medial prefrontal cortex, posterior cingulate, and associated regions — is active during rest, mind-wandering, self-referential thought, and autobiographical processing. Judson Brewer’s 2011 work found reduced DMN activity and altered connectivity in experienced meditators, across several practice types.

This is the finding with the most obvious doctrinal resonance. A network associated with self-referential narration being quieted by practices aimed at non-self is a striking correspondence — and the correspondence should be held loosely, for reasons discussed below.

Long-term practitioners differ from controls. Studies of monastics with tens of thousands of hours find differences in cortical thickness, gyrification, white matter integrity, and gamma activity. Sara Lazar’s work reported greater cortical thickness in prefrontal and insular regions in practitioners.

Pain processing changes. Fadel Zeidan’s work has found that meditation-related analgesia involves different mechanisms from placebo and is not blocked by naloxone, suggesting it does not operate through the endogenous opioid system.

What is weaker than reported #

Cross-sectional studies cannot show causation. A great many findings compare long-term meditators with non-meditators. People who complete ten thousand hours of meditation differ from people who do not in many ways besides the meditation, including in whatever disposed them to start.

Effect sizes are mostly modest and shrink as trial quality improves. Structural brain change findings in particular have been hard to replicate. A 2022 preregistered study by Kral and colleagues with a substantial sample found no significant structural changes after MBSR, contradicting earlier smaller reports.

Publication bias and small samples. The early literature was dominated by small studies from committed researchers. A 2018 review in Perspectives on Psychological Science (Van Dam et al.) documented systematic overstatement, inconsistent definitions, and inadequate control conditions across the field.

Popular claims that are not supported: that meditation “rewires the brain” in any sense more specific than that all learning changes the brain; that it reliably increases grey matter; that it slows brain ageing; that specific numbers of minutes produce specific outcomes. These circulate widely and do not follow from the evidence.

The interpretation problem #

Even where findings are solid, what they mean is not straightforward.

Neural correlates are not explanations. Showing that DMN activity decreases during practice describes what the brain is doing; it does not explain the experience, and it does not validate the tradition’s account of it. Every mental state has neural correlates, including ones the tradition regards as delusion.

“Meditation” is not one thing. Focused attention, open monitoring, loving-kindness, koan practice, deity visualisation, and shikantaza are different operations with different aims, and lumping them produces meaningless averages. The better recent work distinguishes them, and finds that they do differ neurally.

The traditions’ claims are not about brains. Buddhism claims that craving can end, that rebirth ceases, and that dukkha is extinguished. No imaging study addresses any of this. A demonstration that meditation improves attention is not evidence for or against the tradition’s actual claims.

The “Buddhism and science” problem #

There is a broader issue, and Donald Lopez’s Buddhism and Science: A Guide for the Perplexed sets it out well.

The claim that Buddhism is uniquely compatible with science is about 150 years old and had a specific purpose: it was a response to Christian missionary attacks, deployed by Asian reformers who needed an argument that their tradition was rational where their opponents’ was superstitious. See Buddhist Modernism.

The claim has been sustained since by a selective procedure: findings that fit are highlighted, doctrines that do not fit — the realms, Mount Meru, the supernormal powers, rebirth — are reinterpreted or quietly dropped. Lopez’s point is that a tradition which only ever agrees with current science is not being tested by it.

This is not an argument against the research, which is legitimate. It is an argument against the framing that treats neuroscience as validating Buddhism.

Where the collaboration has been genuinely productive #

Two areas stand out.

First-person method. Varela’s proposal of “neurophenomenology” — using trained contemplatives’ detailed introspective reports to generate hypotheses about neural data — took seriously the possibility that long-term practitioners have observational skills relevant to the science. Whether this has delivered is arguable; that it is a serious proposal is not.

Adverse effects. Ironically, the most clearly useful contribution has been documenting that meditation can go wrong. Willoughby Britton’s Varieties of Contemplative Experience study catalogued a wide range of difficult experiences in practitioners, many of which the traditions describe but which modern secular programmes had not warned about. See When Meditation Is Difficult.

What a practitioner should take from this #

The research is not the reason to practise. If the evidence were weaker than it is, the tradition’s claims would be unaffected, because they are not claims about brain structure.

Modest effects are still effects. Improved attention and reduced rumination are worth having, and the evidence for them is reasonable.

Be sceptical of specific numbers. Claims about how many minutes produce what changes are not supported, and the practice traditions themselves have never worked that way.

The measurable and the important are not the same set. Attention and stress markers are measurable. Whether craving has weakened is not, and is what the tradition cares about.

Common questions #

Does meditation change the brain? All sustained activity changes the brain. Whether meditation produces the specific structural changes early studies reported is now genuinely uncertain, with several failed replications. Functional changes during and after practice are better established.

How much practice is needed? The clinical programmes use about 45 minutes daily for eight weeks and show effects. The long-term practitioner findings involve tens of thousands of hours. Between those, the evidence is thin, and the traditions have always emphasised consistency over duration.

Is meditation better than exercise or therapy? Generally, head-to-head trials find mindfulness interventions comparable to other active treatments rather than superior. For relapse prevention in recurrent depression, MBCT performs comparably to maintenance antidepressants.

Do Buddhists care about this research? Variously. The Dalai Lama has actively promoted it. Many traditional teachers regard it as beside the point. Some scholars regard the enthusiasm for it as a symptom of modernism rather than a vindication of the tradition.

Sources & further reading #

Donald S. Lopez Jr., Buddhism and Science: A Guide for the Perplexed (University of Chicago Press, 2008) — the essential corrective to the compatibility narrative. Daniel Goleman & Richard J. Davidson, Altered Traits (Avery, 2017) — a survey by insiders that is unusually candid about weak studies. Nicholas Van Dam et al., “Mind the Hype,” Perspectives on Psychological Science 13 (2018). Antoine Lutz, Heleen Slagter, John Dunne & Richard Davidson, “Attention Regulation and Monitoring in Meditation,” Trends in Cognitive Sciences 12 (2008) — on distinguishing practice types. Evan Thompson, Why I Am Not a Buddhist (Yale, 2020) — a philosopher’s critique of Buddhist modernism and the neuroscience framing.