Create Your New Reality

What if I told you that a 7-day meditation retreat could rewire your brain, reprogram your cells, activate natural pain-relief mechanisms, and heal nearly every ailment through the power of thought alone? 

And it’s backed by science! 

“Sign me up,” you say.

 

But wait, before you invest $2,500 dollars, let me tell you about the amazing new study published in Communications Biology.

“This intensive non-pharmacological mind-body intervention produces broad short-term neural and plasma-based molecular changes associated with enhanced neuroplasticity, metabolic reprogramming, and modulation of functional cell signaling pathways…” says the abstract. 

“Science is the new language of mysticism,” says the maestro of meditation. 

 

The Paper

Neural and molecular changes during a mind-body reconceptualization, meditation, and open label placebo healing intervention 

 

This paper reports an ambitious, logistically difficult study of 20 participants before and after they attended a week-long meditation retreat. A raft of blood and brain biomarkers were collected by a team of researchers from UC San Diego, a meditation research conglomerate, a clinical trials consulting firm, and the Institute of Advanced Consciousness Studies.  [registration in ClinicalTrials.gov, NCT06615531]

The retreat consisted of three mind-body interventions:

• Reconceptualization (25 hours)
• Meditation (33 hours)
• Guided healing rituals (5 hours) conceptualized as “open label placebo”

There was no way to differentiate any unique biological correlate(s) because the “mind-body” components were always delivered together. I will leave aside (for now) the possibility that all three interventions might be placebos, because THERE WAS NO CONTROL GROUP exposed to a different condition, an alternate treatment, or even a waiting list.

Reconceptualization: 

"Daily lectures emphasized the body’s self-healing abilities, the mind’s capacity to shape lived reality, and the healing power of present-centeredness and mystical-type experiences."

Meditation: 

"All meditations were guided, delivered with atmospheric music, and taught Kundalini techniques, which combine conscious meta-awareness and conscious breathing exercises with slow, ascending, focused interoceptive attention on purported energetic centers along the midline..."

Guided healing: 

These "brought 6–8 “healers” around one “healee” in which the former were instructed to practice loving-kindness compassion meditation while focusing attention on their heart, hands, and on the latter’s body."

 

The study generated a staggering number of data points¹ (~100,000). Collecting and analyzing such a wide array of measures took an extraordinary amount of work, and the authors are to be commended for this. They created a nifty figure of their outcome measures (shown below).

 

Fig. 1A. Outcome measures to capture biological changes associated with brain and body. Created with BioRender. Simpson, S., Jinich, A. (2025). BioRender.com/ryzs1cd.

Before offering my own opinion, here's how the authors presented their work to the public:

GROUNDBREAKING RESULTS: THE DISCOVERY THAT CHANGES EVERYTHING 

-- This is from a paid press release. 

The landmark study demonstrates how intensive meditation can trigger the same profound brain activity previously documented only with psychedelic substances – while simultaneously activating measurable biological transformations throughout the entire body. 

In just seven days, without any pharmaceutical intervention, retreat participants achieved what researchers are calling a "biological reset" – rewired neural networks, boosted cellular nerve cell growth, reprogrammed cellular energy systems, and activated natural pain-relief mechanisms. Research data also shows study subjects' "mystical experience" scores – measured via self-reporting – increased significantly in a group of 20 individuals within the seven-day event. 

 

OK then. Well... My take is different. I suggest some of these claims are not supported by scientific evidence, based on procedural and analytic flaws that occur throughout the manuscript and supplementary materials. 

 

Claim 1. Enhanced Neuroplasticity (greater neurite outgrowth)

“Participants’ anecdotal reports consistently emphasize radical psychological breakthroughs, and previous meditation studies have reported increased BDNF levels consistent with enhanced neuroplasticity. To investigate whether the intervention affected circulating plasma factors conducive to neuroplasticity, we treated cultured glutamatergic PC12 neuroendocrine cells with NGF (nerve growth factor) and 1% pre- and post-intervention plasma and quantified neurite outgrowth length.”

 

Result 1. PC12 cells treated with plasma from all 20 participants

PC12 cells are not primarily glutaminergic. PC12 is a cell line derived from a pheochromocytoma (tumor) of the rat adrenal gland. The cells synthesize, store, and release catecholamines. PC12 cells can be induced to differentiate into neuron-like cells. When treated with NGF, they extend neurite-like processes. Subsequent treatment with a variety of agents can elucidate specific cellular mechanisms.

Here, a problematic choice was that plasma from all 20 participants was combined into one soup, which eliminated individual variation and prevented mechanistic insights. Further, plasma samples from novice and advanced meditators (who showed different molecular profiles in other analyses) were pooled.

 

Claim 2. Metabolic Reprogramming (shift toward glycolytic metabolism)

“Previous studies have characterized meditation as a hypometabolic state and reported enhanced glycolysis in Tibetan Buddhist monks. To test the intervention’s effect on real-time metabolism, we treated BE(2)M17 human neuroblastoma cells with 1% plasma from all participants for 60 min and performed Seahorse XF assays.”

 

Result 2A. Pooled Plasma (loss of individual metabolic signatures)

Once again, the claim of “significant” changes in glycolysis is based on comparing two pooled samples (basically n=1 vs. n=1). We can't determine which plasma components (or which individual participants) drove the metabolic changes.

Result 2B. Preselected Proteins (Fig. 4E, glycolysis proteomics heat map)

The authors didn't explain their criteria for choosing 19 proteins involved in glycolysis and oxidative phosphorylation (out of hundreds). Was the selection biased in some way? Were important proteins were missing?

Claim 3. Modulation of Functional Cell Signaling Pathways (proteomics)

“To investigate the intervention’s effects on the plasma proteome, 7596 proteins were quantified with the SomaScan Assay v4.1.   ...  Volcano plot analysis (Fig. 5A) revealed 21 significantly altered proteins. Cofilin-2 (COF2) and Enoyl-CoA hydratase were significantly upregulated, which suggests enhanced cellular processes related to cytoskeletal regulation and fatty acid metabolism.”

 

Result 3A. Only 21 out of 7596 proteins were significantly altered (far less than chance)

How did that happen?? The proteomic and metabolomic analyses were exploratory and hypothesis-free, but... with 7,596 proteins tested at p < .05, the expected number of false positives is ~380 proteins (5% of 7,596). Statistical correction for multiple comparisons was inconsistently applied across measures. Even within the Proteomics category, Table 5 says p<.05 but the Methods (p. 14) indicate that FDR correction was applied (false discovery rate).

Result 3B. Tiny Effect Size (Fig. 5A, linear fold change of 0.25 in either direction) 

This means there was only a 25% increase or decrease in these proteins after the retreat, which may be within technical or biological noise levels. Most studies require at least 1.5 fold changes for biological relevance. Elsewhere, for the protein-protein interaction networks in Fig. 5B, they used either p<.05 OR fold change >.5, which is still problematic. Use more stringent criteria instead (FDR < 0.05 AND fold change > 1.5).

Across the entire study, the chance of false positive results is high.

 

Claim 4. Upregulation of Anti-Inflammatory and Inflammatory Markers (dynamic process of immune modulation)

“To assess whether the intervention elicited inflammatory or anti-inflammatory cascades, we examined a panel of 23 inflammatory and 21 anti-inflammatory proteins (Fig. 5E). We found significant upregulation of inflammatory markers ... Interestingly, we also observed a significantly upregulated anti-inflammatory markers index... Concurrent activation of both pathways suggests a dynamic process of immune modulation, possibly reflecting enhanced cellular turnover or repair mechanisms.”

Result 4A. Upregulation of Inflammatory Markers is Bad (negative effects of the retreat?)

This result is certainly the opposite of what would be expected. Some studies suggest that meditation can reduce inflammatory markers (lower IL-6, TNF-α, CRP, etc.). For an intervention that ostensibly promotes well-being and neuroplasticity, increased inflammation is counterintuitive and potentially harmful. The authors didn't consider alternative explanations, such as: 

• Stress – the intensive retreat format (10-12 hours daily) could be physically/psychologically stressful
• Sleep Deprivation – long meditation hours might reduce sleep and trigger inflammation 
• Dietary Changes – retreat food and eating patterns might differ from normal

Result 4B. Incorrect Statistical Reporting (namely, incorrect effect sizes)

This occurs elsewhere in the paper, but here are three examples. 

t = 2.25, p = 0.03, Cohen's d = 0.09  0.50
t = -2.09, p = 0.04, Cohen's d = -0.65  -0.47
t = 3.81, p = 0.0001, Cohen's d = 0.15  0.85

Claim 5. Alterations in Functional Brain Activity (related to meditation)

“To characterize the neural signature of the meditative state, participants underwent structural and blood-oxygenation-level-dependent (BOLD) functional MRI scans during rest (5 min) and meditation (15 min). 
...
fMRI data showed that this meditation style functionally disrupts the default mode and salience networks (responsible for self-referential thought and allostatic regulation) and cerebellum-prefrontal predictive processing circuits involved in integrating internal models with external sensory data.”

 

Result 5A. Explosion of fMRI Features (an excess of networks, regions, and parcellations)

The authors chose redundant and superfluous ways to analyze the data, increasing the risk of false positive results.

Seven Canonical Resting State Networks
- within-network connectivity: ~50-100 features per network
- between-network connectivity: 21 pairs
- subtotal: ~400-700 features

Eight ROIs (regions of interest)
- all pairwise connections: 28 unique pairs

Harvard-Oxford Atlas (48 regions)
- all pairwise connections: 48×47/2 = 1,128 features

Power Atlas (264 regions)
- all pairwise connections: 264×263/2 = 34,716 features

Total fMRI Features: ~36,500+

 

Result 5B. Significant Results Caused by Head Motion? (difference between rest and meditation)

During the 5 minute resting state scan, participants were told to “not move, keep eyes open, stay awake, and think about whatever you want, but do not meditate.”  During the 15 minute meditation scan, participants were instructed to “not move, listen to the guided meditation soundtrack, and meditate as suggested by the audio while keeping your eyes open.”

A few things:

• The guided meditation has a bombastic audio soundtrack, while the resting state is silent 
• It would have been beneficial to compare auditory cortex activity for meditation vs. rest
• Breathing exercises during meditation can increase head movement

The authors admit this show-stopping confound of greater head motion during the meditation scan. 

“Participants moved more during meditation than rest, a potential confound revealed by the significant effect of task (meditation, rest) on mean framewise displacement on a two-way (task × time) repeated measures ANOVA (n = 19, F(1,18) = 25.1, p = 0.00009, η²p = 0.58).”

Thus, the functional connectivity differences between meditation and rest could explained by head motion, not by changes in brain activity. Even tiny movements (0.5 mm) create spurious correlations between regions. The effect size above indicates that 58% of variance in movement was explained by meditation vs. rest. The whopping p-value above was reported in the abstract, but in a different context:

“Meditation decreases functional integration in the default mode (p = 0.00009) and salience networks (p = 0.000003).”

Result 5C. Errors in the Framewise Displacement Spreadsheet (supplementary material)

Framewise displacement (FD) is a quality control metric that considers changes in six head motion parameters from one frame to the next. The authors said their results “were robust to excluding BOLD runs with mean framewise displacement > 0.3 mm, indicating they were not due to higher meditation-associated head motion.” I wasn't sure what they meant by "run" (an entire 5 min block? how many blocks tossed? why wasn't a more stringent cutoff used?). 

At any rate, the Supplementary Material includes a temporal string of FD values for all subjects and conditions. Some numbers in the spreadsheet looked like this, '0.5053053200000001 (an extraneous apostrophe) while the pre and post values for the Rest condition were identical.

Claim 6. Machine Learning Models Discriminated Pre- and Post-Meditation States with High Accuracy

“We applied machine learning to identify the most biologically relevant features across time point (pre/post) and experience level (novice/advanced) datasets. Each dataset was preprocessed (log-transformed and auto-centered) and missing data was imputed to ensure feature scaling, normalization consistency, and data integrity and comparability across modalities. Post-preprocessing, ELISA, metabolomics, transcriptomics, and proteomics datasets were concatenated into a single feature matrix used as input for an eXtreme Gradient Boosting (XGBoost) classifier chosen for its ability to handle high-dimensional data. ... both models achieved strong discrimination between pre- and post-meditation states (XGBoost AUC = 0.86; Random Forest AUC = 0.90).”

Result 6. Extreme Overfitting (with no way to validate in an independent sample)

Unless I'm mistaken, they dumped ALL the data, including ~36,500 fMRI features, into one giant classifier. This guarantees overfitting. With 100,000 features and 20 samples, the model can memorize the training data. In contrast, 1 feature per 10 samples is the minimum recommended for stable ML. There was no validation set, because you can't split 20 samples into training and testing sets. There was no independent cohort, so you can't see if the findings generalize to another population. The impossibly high AUCs of 0.86-0.90 are meaningless: you could use random numbers and get these values. 

Then there were correlations between self-report scores on the Mystical Experiences Questionnaire and the top 14 features per model, which I'll skip for now.

Major Weaknesses

1. No control group (can't separate meditation from the passage of time)
2. Massive overfitting (100,000 features, 20 people)
3. Major motion confound
4. Uncorrected multiple comparisons
5. Inconsistent statistics
6. Possible p-hacking and post hoc explanations

 

Undisclosed Conflicts of Interest

The meditation teacher (Dr. Joe Dispenza) is an author on the paper. He declared his employment with Encephalon, Inc., which offers the meditation retreats. However, the $10 million award (gift) to the senior author (Dr. Hemal H. Patel) from InnerScience Research Fund isn't fully explained. Dr. Joe owns InnerScience. So the person who ran the week-long retreat for 1,444 attendees also funded the research project and co-authored the article. Dr. Patel recently received another $2.45 million from InnerScience.

 

Next time we'll take a closer look at Dr. Joe Dispenza.

 

Footnote

¹ To preview, my take is that ~100,000 data points per participant (n=20 or less) were fed into machine learning (ML) models. Oh no.

 

Reference 

Jinich-Diamant A, Simpson S, Zuniga-Hertz JP, Chitteti R, Schilling JM, Bonds JA, Case L, Chernov AV, Dispenza J, Maree J, Amkie Stahl NE. et al. Neural and molecular changes during a mind-body reconceptualization, meditation, and open label placebo healing intervention. Communications Biology. 2025 Nov 6;8(1):1525.

 

 

The highly speculative model below interprets the findings within a Bayesian brain framework. The figure shows superficial pyramidal cells (layers 2/3), deep pyramidal cells (layers 5/6), a neuromodulatory cell, and information flow. It's a cellular-level circuit diagram based on:

• Group-level fMRI (resolution: ~3mm³)
• Blood samples
• 20 people
• Motion artifacts

Fig. 9. Potential cortical implementation. Hierarchical predictive coding scheme showing how the three mind-body techniques may synergistically facilitate a more flexible and adaptive predictive system.