Jhana · ACAM-J · 7T fMRI · n=20

Jhana: Yang 2025, twenty meditators, eight absorptions on 7T

JM Thomas · 17 August 2026

I've run access concentration on a breath until applied thought thinned. Search this week: jhana, jhana meditation, jhana science, ACAM-J, absorption meditation neuroscience. Kripalu lists a jhana week 28 August to 4 September 2026. Samatha Centre Wales lists 28 August to 6 September. This page stays with the 7T file.

Eight named absorptions in one sit

Sparby and Sacchet, Mindfulness 15 (2024) 1375-1394, DOI 10.1007/s12671-024-02367-w. Eight contemporary manuals. A shared skeleton, and real fights over how deep each rung goes. This page uses their ACAM-J label: advanced concentrative absorption meditation, jhana type. Doctrine stays in the monastery. The scanner gets eight reportable states.

Joy is early. Formlessness is late. That order is the whole trick. You do not skip to infinite space because a retreat brochure used the word bliss. You hold an object until the first rung reports, then you let the next one arrive. Hold the object.

Front of the head goes quiet

Prefrontal cortex, orbitofrontal cortex, and the brainstem arousal nodes Yang named (locus coeruleus and inferior medullary reticular formation) dropped in ReHo as the ladder deepened, while visual cortex, precuneus-posterior cingulate, parietal, and somatomotor rose. Early states carried more of the anterior drop. J7 and J8 carried the posterior rise. Frontal quiet. Field loud.

The principal connectivity gradient, G1, flattened. Control tasks kept the usual unimodal-to-transmodal ladder. Absorption compressed it. Geometric eigenmodes, power and energy, drew a U: drop through the mid jhanas, recover later. Yang printed that U. Quote it as a U, not as a percent you can paste on a thumbnail.

Partial least squares tied ReHo to phenomenology at r=0.80. Posterior sensory-attentional loadings sat with stability, width of attention, and correspondence of the named state. Anterior control loadings sat against narrative thought and extra sensation. Neurosynth decoding pointed the same way: attention and executive maps up, affect and autobiographical maps down. Stronger toward J8.

That is the mechanism the search string wanted. Absorption reallocates local synchrony away from frontal narrative work and toward a denser posterior field, and you still report being awake enough to name the rung you are in. Stay awake. Name it.

Twenty people in a 7 Tesla

n=20, ages 26 to 67 (mean 43.95, SD 15.10), with a mean 17.40 years of practice (SD 11.59) and an estimated 20,500 hours (SD 17,289): eighteen intermediate, two deep, Mass General Brigham IRB plus Witten/Herdecke, Siemens MAGNETOM Terra, 32-channel coil, 1.1 mm isotropic BOLD, TR 2.9 s. Trained cohort. Hard hours.

Each person ran their usual sequence, eyes closed. Access concentration through J8, then afterglow. The lab asked for at least 3 minutes on a rung so ReHo would compute. A sit under 2 minutes got another run. Button presses marked transitions. One person skipped the J6-J7 and J7-J8 presses so the formless would hold. 22 ACAM-J runs. Mean complete run 33.60 minutes, SD 15.40.

Controls were counting down from 10,000 by fives, and recalling the last week. Rest got skipped on purpose. Adepts meditate when you say rest. Five people who also sat case studies ran the controls twice. 25 control sessions in all. Some still leaked a little jhana into the count. Yang says the leak did not erase the ACAM-J versus control difference.

After each full ladder they rated 1 to 10: stability, width, quality of the named state, early versus late sights/sounds/body/narrative, then the factor for that rung. Bliss and joy on J1-J2. Contentment on J3. Equanimity on J4. Formlessness on J5-J8. Missing ratings got MICE imputation. Say that. The phenomenology is trained self-report, locked to the BOLD, on a preprint.

One teacher, fifty-six runs

Same meditator as Yang 2024a and Ganesan 2024. Five days of EEG. Five days of 7T. Access concentration, J1 through J8, afterglow. Buttons through J6. No buttons on J7 and J8. A press would have dumped the formless. Average dwell, all runs: J1 77 s, J2 92 s, J3 108 s, J4 75 s, J5 69 s, J6-J8 together 180 s.

Within-network modularity fell in DMN, frontoparietal, visual, and salience, strongest in the formless. Global functional connectivity rose, especially DMN and visual in J5-J8. Default-mode desegregated. EEG broadband power dropped, delta through alpha the loudest. Lempel-Ziv complexity rose. Bliss on J2 tracked higher DMN global connectivity, lower DMN modularity, and lower beta/gamma. Width of attention tracked higher LZ.

Some fMRI contrasts moved with which control you pick. Counting split more cleanly than memory. EEG stayed consistent across both. Chowdhury printed that split. Keep it. An n=1 with 56 runs is a high-density case on one nervous system, which is exactly as strong and as narrow as that sentence sounds. One teacher.

Hagerty 2013, Neural Plasticity 2013:653572, DOI 10.1155/2013/653572, PMID 24303217, scanned one person on about two fMRI runs and pointed at nucleus accumbens. Reward. Old n=1. Yang is the group lock. Chowdhury is the dense multimodal lock. Hagerty is the early photograph.

Access concentration is the door

Yang's people already had that door. Mean 20,500 hours. Chowdhury's teacher had twenty-five years. A first-week sit can still train the door, and it still cannot honestly wear the 7T ReHo map from a cohort that spent decades getting the object to hold. Write a 1-to-10 on how long the object held. Write whether joy showed up without you pushing it. Stop there if the second number is theater. Log it.

Lab sits used a button. Home sits use a log. If you need a press to know you moved, you probably did not move. Chowdhury skipped the last two presses for a reason. Formless rungs hate being announced.

Kripalu, Jhourney, Brasington, Samatha Centre: those are retreat clocks. They can teach the ladder. They did not print Yang's n=20. Do the retreat if you want a teacher. Quote the preprint if you want the scan.

Sankalpa is a different sit

You can run both in a week. Mixing them in one hour is a personal experiment. Calling the mix Yang n=20 is how a lab number gets invented. Yoga nidra wants the watcher offline so a sentence can land. Jhana wants the watcher absorbed so eight qualities can be named. Keep the verbs.

Educational mechanism page by JM Thomas. A clinician owns diagnosis and treatment. Yang 2025 is a 7T preprint in 20 trained meditators. Chowdhury 2025 is an n=1 multimodal case, 27 fMRI plus 29 EEG. Hagerty 2013 is an n=1 reward-system snapshot. Talk to a clinician if unusual distress, dissociation, or a psychiatric history is on the table.

Papers this page cites