# Brain implants for depression and the two-millimetre gap

My Weird Prompts — an AI-generated podcast. Daniel Rosehill sends in a prompt; AI hosts Corn and Herman discuss it. Verify important claims independently.

- Episode: 6649
- Published: 2026-11-17
- Duration: 29:07
- Audio: https://episodes.myweirdprompts.com/audio/deep-brain-stimulation-depression-55419.m4a
- Tags: depression, mathematics, memory, psychiatric-medication, therapy-and-counseling
- Web page: https://myweirdprompts.com/episode/deep-brain-stimulation-depression-55419/
- JSON: https://myweirdprompts.com/api/episodes/deep-brain-stimulation-depression-55419.json

## Summary

Deep brain stimulation is approved for Parkinson's, tremor, and OCD, yet it remains unapproved for treatment-resistant depression. This look at the science traces Helen Mayberg's work on Area 25, the surprising failures of St. Jude Medical's BROADEN and Medtronic's RECLAIM trials, and a new Canadian sham-controlled study where unstimulated patients initially outperformed active ones. It explores how tractography imaging revealed that shifting an electrode by just two millimetres changes which white-matter tracts are engaged, turning the same surgery into two different treatments.

## The prompt

> Deep brain stimulation for treatment-resistant depression. Electrodes are implanted in specific brain circuits for people whom nothing else has helped. Why might it work, what happened in the big trials (including the ones that failed and the later results that revived the field), and where does the science stand now, including the newer personalised, closed-loop approaches?

## Show notes

Deep brain stimulation is FDA-approved for conditions like Parkinson's, OCD, and epilepsy, yet it remains unapproved for treatment-resistant depression—a gap that defines one of psychiatry's most complicated stories. Beginning with neuroscientist Helen Mayberg's circuit-based model targeting Area 25, early open-label results sparked immense hope. However, major sham-controlled trials, including St. Jude Medical's BROADEN and Medtronic's RECLAIM, failed to show benefit over placebo, with unstimulated patients sometimes faring better. A 2026 Canadian crossover study echoed this paradox: no difference during the blinded phase, yet strong open-label response rates at two years. Researchers now point to tractography imaging, which shows that engaging specific white-matter tracts rather than fixed anatomical coordinates—alongside months of individualized tuning—may be the key to making the treatment work.

## Sources

- [Alagapan S, Choi KS, Heisig S et al., Cingulate dynamics track depression recovery with deep brain stimulation, *Nature*, 2023, PMID 37730990](https://pubmed.ncbi.nlm.nih.gov/37730990/) — (retrieved 2026-11-17)
- [Ghicobbe P, Rizvi SJ, Ceniti AK et al., Deep Brain Stimulation to the Subgenual Cingulate Gyrus for Treatment-Resistant Depression: A Randomized Controlled Trial and 2-Year Long-Term Follow-Up, *Canadian Journal of Psychiatry*, 2026, PMID 41171999](https://pubmed.ncbi.nlm.nih.gov/41171999/) — (retrieved 2026-11-17)
- [Gholamali Nezhad F, Tassone VK, Khoo Y et al., Lack of neuropsychological effects following short-term subcalosal cingulate gyrus deep brain stimulation in treatment-resistant depression: a randomised crossover study, *BMJ Mental Health*, 2025, PMID 40021211](https://pubmed.ncbi.nlm.nih.gov/40021211/) — (retrieved 2026-11-17)
- [Attali D, Tiennot T, Manuel TJ et al., Deep transcranial ultrasound stimulation using personalised acoustic metamaterials improves treatment-resistant depression in humans, *Brain Stimulation*, 2025, PMID 40311843](https://pubmed.ncbi.nlm.nih.gov/40311843/) — (retrieved 2026-11-17)
- [Riis TS, Feldman DA, Kwon SS et al., Noninvasive Neuromodulation of the Subcalosal Cingulate and Depression With Focused Ultrasonic Waves, *Biological Psychiatry*, 2025, PMID 39396736](https://pubmed.ncbi.nlm.nih.gov/39396736/) — (retrieved 2026-11-17)
- [Ramasubbu R, Brown EC, Mouches P et al., Multimodal imaging measures in the prediction of clinical response to deep brain stimulation for refractory depression: A machine learning approach, *World Journal of Biological Psychiatry*, 2024, PMID 38185882](https://pubmed.ncbi.nlm.nih.gov/38185882/) — (retrieved 2026-11-17)
- [Sendi MSE, Waters AC, Tiruvadi V et al., Intraoperative neural signals predict rapid antidepressant effects of deep brain stimulation, *Translational Psychiatry*, 2021, PMID 34728599](https://pubmed.ncbi.nlm.nih.gov/34728599/) — (retrieved 2026-11-17)
- [Stakweather CK, Sugrue LP, Cajigas I et al., Stereoelectroencefalography Electrode Implantation for Inpatient Workup of Treatment-Resistant Depression, *Neurosurgery*, 2024, cited by 10](https://pubmed.ncbi.nlm.nih.gov/39283114/) — (retrieved 2026-11-17)
- [Subcalosal cingulate deep brain stimulation for treatment-resistant depression: a multisite, randomised, sham-controlled trial, *The Lancet Psychiatry*, published 2017-10-04](https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366) — (17)30371-1/abstract (retrieved 2026-11-17)
- [Closed-loop neuromodulation in an individual with treatment-resistant depression, *Nature Medicine*, published 2021-10-04](https://www.nature.com/articles/s41591-021-01480-w) — (retrieved 2026-11-17)
- [State-dependent responses to intracranial brain stimulation in a patient with depression, PMC8284979](https://pmc.ncbi.nlm.nih.gov/articles/PMC8284979/) — (retrieved 2026-11-17)
- [Lampros M, Symeou S, Alexiou G et al., Applications of machine learning in deep brain stimulation for major depressive disorder: a systematic review and meta-analysis, *Neurosurgical Review*, 2025, PMID 41051633](https://pubmed.ncbi.nlm.nih.gov/41051633/) — (retrieved 2026-11-17)
- [Figee M, Riva-Posse P, Choi KS et al., Deep Brain Stimulation for Depression, *Neurosurgery*, 2022, cited by 123](https://pubmed.ncbi.nlm.nih.gov/35817944/) — (retrieved 2026-11-17)
- [Johnson KA, Okun MS, Scangos KW et al., Deep brain stimulation for refractory major depressive disorder: a comprehensive review, *Molecular Psychiatry*, 2024, cited by 69](https://pubmed.ncbi.nlm.nih.gov/38287101/) — (retrieved 2026-11-17)
- [Unadkat P, Quevedo J, Soares J et al., Opportunities and challenges for the use of deep brain stimulation in the treatment of refractory major depression, *Discover Mental Health*, 2024](https://pubmed.ncbi.nlm.nih.gov/38483709/) — (retrieved 2026-11-17)
- [Gonzále V, Rumià J, Brell M et al., From lobotomy to connectomics: a critical review of contemporary psychosurgery, *Frontiers in Psychiatry*, 2026](https://pubmed.ncbi.nlm.nih.gov/42109284/) — (retrieved 2026-11-17)
- [Martine-Nunez AE, Rozell CJ, Litle S et al., Proceedings of the 12th annual deep brain stimulation think tank: cutting edge technology meets novel applications, *Frontiers in Human Neuroscience*, 2025](https://pubmed.ncbi.nlm.nih.gov/40070487/) — (retrieved 2026-11-17)
- [Dong J, Dai M, Guo Z et al., The Targets of Deep Brain Stimulation in the Treatment of Treatment-Resistant Depression: A Review, *Brain and Behavior*, 2025, cited by 4](https://pubmed.ncbi.nlm.nih.gov/40321033/) — (retrieved 2026-11-17)
- [Idlett-Ali SL, Salazar CA, Bell MS et al., Neuromodulation for treatment-resistant depression: Functional network targets contributing to antidepressive outcomes, *Frontiers in Human Neuroscience*, 2023, cited by 20](https://pubmed.ncbi.nlm.nih.gov/36936612/) — (retrieved 2026-11-17)
- [Dandekar MP, Diaz AP, Rahman Z et al., A narrative review on invasive brain stimulation for treatment-resistant depression, *Revista Brasileira de Psiquiatria*, 2022](https://pubmed.ncbi.nlm.nih.gov/34468549/) — (retrieved 2026-11-17)
- [Ramasubbu R, Lang S, Kiss ZHT, Dosing of Electrical Parameters in Deep Brain Stimulation (DBS) for Intractable Depression: A Review of Clinical Studies, *Frontiers in Psychiatry*, 2018, cited by 93](https://pubmed.ncbi.nlm.nih.gov/30050474/) — (retrieved 2026-11-17)
- [The Neorocritic, BROADEN Trial of DBS for Treatment-Resistant Depression No Better than Sham, 2017-10](https://neorocritic.blogspot.com/2017/10/broaden-trial-of-dbs-for-treatment.html) — (retrieved 2026-11-17)
- [David Dobbs, Why a 'Lifesaving' Depression Treatment Didn't Pass Clinical Trials ... but could still be a groundbreaking therapy, *The Atlantic*, 2018-04-17](https://www.theatlantic.com/science/archive/2018/04/zapping-peoples-brains-didnt-cure-their-depression-until-it-did/558032/) — (retrieved 2026-11-17)
- [Closing the Loop on Deep Brain Stimulation for Treatment-Resistant Depression, PMC5871707](https://ncbi.nlm.nih.gov/pmc/articles/PMC5871707/) — (retrieved 2026-11-17)
- [*Psychiatric News*, Does DBS Have a Future in Depression Treatment?, 2020-03](https://psychiatryonline.org/doi/full/10.1176/appi.pn.2020.3a22) — (retrieved 2026-11-17)

## Transcript

Corn: A surgeon drills two small holes in your skull, threads electrodes down into the brain, and wires them to a battery in your chest. The goal: to lift a depression that twenty years of medication never touched. It sounds like either the future of psychiatry or its cautionary tale, and here's the thing — it's been both, twice.

Corn: That's roughly where Daniel's prompt begins. He wants the whole arc of deep brain stimulation for treatment-resistant depression — why anyone thought it would work, what happened in the big sham-controlled trials, the ones that failed and the later results that revived the field, and where things stand now with personalised and closed-loop devices.

Corn: One footnote before we start. Daniel usually picks the topics himself, but this one came out of a rare reverse — he asked the friendly AI down the road what the show might be missing, the corners of mental health nobody talks about, and this is what came back. So if you've been waiting for us to go here, you have company.

Herman: And it's a good time to go here, because the field is in motion. We've got a new sham-controlled trial this year, we've got ultrasound knocking on the door from the outside. But let's set the stage, because DBS itself is not experimental — it's boring, in the best sense.

Herman: The hardware is a pulse generator implanted under the skin of the chest, leads running up to electrodes sitting in specific spots in the brain, delivering controlled electrical impulses. Over two hundred and forty-four thousand people have been implanted worldwide in three decades, mostly for Parkinson's, essential tremor, dystonia. The FDA has approved it for tremor since ninety-seven, Parkinson's since two thousand two, dystonia, OCD, epilepsy.

Corn: Everything except the one we're talking about.

Herman: Right. Depression is the studied-but-never-approved indication, and that gap is the whole story. Now, treatment-resistant depression itself — the standard definition is major depression that hasn't responded adequately to at least two different antidepressants, at a decent dose, for a decent duration. Inadequate response usually means less than a twenty-five percent reduction in symptoms.

Corn: A quarter. Two full drug trials and you're a quarter better.

Herman: At best, yes. And nearly thirty percent of people with major depression fall into that bucket. We're not talking about a rare fringe — this is a huge population, and by some estimates around a hundred million people worldwide.

Corn: So the stakes are enormous, the patient pool is enormous, and the surgery is approved for five other things. Why isn't this settled?

Herman: Because the brain doesn't care how the petition is worded. The story starts with a neuroscientist named Helen Mayberg and a piece of tissue called Area 25.

Herman: The subgenual cingulate — Area 25 — is a pair of pale-pink curves sitting underneath the corpus callosum, each about the size and shape of a newborn's crooked finger. Mayberg's work going back to her imaging studies suggested that in depressed people this region is hyperactive — a junction box stuck broadcasting despair. And her model wasn't the old chemical-imbalance story. It was a circuit story: depression as specific networks locked in a pathological pattern, not a global shortage of serotonin.

Corn: Which reframes what you're trying to do. You're not topping up a tank.

Herman: No. You're interrupting a signal. And here's what I find elegant — and I want to be careful with that word, because elegant mechanisms have burned this field before —

Corn: Noted for the record.

Herman: The mechanism is thought to be disruption. You're not stimulating an underactive area back to life, the pacemaker metaphor everyone reaches for. You're mimicking what a lesion would do — disrupting pathologically elevated, oversynchronised traffic in a misfiring network — but reversibly. Turn it off, the effect goes away. Unlike ablation, nobody's burning a bridge.

Corn: That reversibility is what makes it science, isn't it. You can run a blinded trial because you can switch the thing off without the patient knowing.

Herman: And that's rare in surgery! It's one of the few neurosurgical procedures that even permits a proper double-blind design. Keep that in the back of your mind, because it becomes the instrument of the field's humiliation.

Herman: So, two thousand three to two thousand six. Mayberg tries the surgery on twenty people who had been torturously depressed for years despite every treatment that exists. In two thousand five she publishes the first six — who did amazingly well. The other fourteen later showed similar progress. Most cut their depression scores by around half. Over a third left depression behind entirely.

Corn: For people who'd tried everything, a third in remission is... that's not a marginal result.

Herman: It was seismic. Tom Insel, then director of the National Institute of Mental Health, called it a new way of understanding depression. And the momentum was real — St. Jude Medical designed a trial called BROADEN, which stood for Brodmann Area twenty-five deep brain neuromodulation, very proud of that. The plan was two hundred patients across twenty neurosurgery centres, devices in everyone, turned on in two-thirds, off in one-third, and nobody with patient contact would know which was which.

Corn: Textbook.

Herman: Textbook. What they got was ninety patients, and the trial published in Lancet Psychiatry in twenty seventeen. Sixty active, thirty sham, six months. The response rate, averaged over months four through six — twenty percent active, seventeen percent sham. Not statistically significant. Three points.

Corn: Three points is a rounding error wearing a lab coat.

Herman: The futility analysis is where it gets painful. The statisticians had predicted a forty percent response in the stimulation group against eighteen and a half in controls. Actual: twenty against seventeen. The model concluded the trial had a seventeen percent chance of success if it continued. That didn't technically meet the pre-set futility threshold — but the sponsor ended enrolment anyway.

Corn: Ninety patients instead of two hundred. If they'd needed those numbers, "seventeen percent chance of success" is a kind sentence.

Herman: It is. And RECLAIM, the Medtronic-sponsored trial, went the same way, maybe worse. That one targeted the ventral capsule and ventral striatum — the internal capsule, the reward circuitry region. All thirty patients got implants; everyone went through an optimisation phase first, hunting for stimulation settings that produced acute mood effects. Then half had their devices switched off for four months. The blind was verified — it held. And the review board stopped the trial because the data showed no sign the stimulation was helping.

Corn: Define "no sign."

Herman: The patients who weren't being stimulated improved more than the patients who were.

Corn: The machine made it worse.

Herman: Numerically, in that window, yes. That's the sentence that should keep anyone in this field awake.

Corn: And now, a brief call from Corn's mother. Dorothy, are you there?

Dorothy: Hello dear, it's your mother, with Dorothy's History of Toast.

Dorothy: The pop-up toaster was invented by Charles Strite, a mechanic in Minnesota, who was fed up with the burnt toast in his factory canteen. He filed the patent in nineteen nineteen.

Dorothy: And I suppose that's the way of it, dear, a man fixes one thing and the rest of the world carries on burning its breakfast.

Dorothy: Anyway, I've got a load of washing on and I should see to it. Give my love to Herman. Bye bye now.

Herman: Give her my love back, Corn.

Corn: She's got a load of washing on, Herman, she's busy.

Herman: Fair enough. Where were we — sham beating active. So here's the temptation, and I want to name it because a lot of coverage falls into it: the temptation is to say the trials were rigged, or that the blind leaked, and therefore the open-label results were right all along. But the sham problem cuts both ways, and it got stranger this year.

Herman: The Canadian trial, published this year, thirty-five patients, single centre — the largest single-centre sham-controlled study in the literature. Crossover design: two three-month phases, everyone eventually gets stimulation, randomised order. No significant difference between groups at three months or six months. All groups improved. And at the end of the blinded phase, the group that had been off the whole time — the OFF-OFF group — had numerically lower depression scores than the group that had been on the whole time.

Corn: The group that got the least treatment did the best. Again.

Herman: Within the blinded window. Then you flip to the open-label follow-up, everyone's device on, and the response rates are sixty-five point seven percent at twelve months, sixty-nine at eighteen, seventy-three at twenty-four. Cognitive testing showed no deterioration — one measure, category fluency, actually improved after three months of active stimulation, though that didn't survive multiple-comparison correction, so I'd file it under interesting, not established.

Corn: Herman, sit with that pair of numbers for a second. Blinded: nothing. Open-label: three-quarters responding at two years. What do you actually do with that as a clinician? You spent decades in medicine — how do you hold both?

Herman: Honestly? Uncomfortably. There are three live explanations and none of them is comfortable. One: the blind is doing something strange — there may be perceptible effects of stimulation that let patients figure out their assignment, and once someone believes they're in the active arm, expectation does the rest. Two: regression to the mean and natural course are enormous in a population this severe, people enrol at their worst. Three: the active stimulation in those trials wasn't hitting the right circuits for those particular brains, and the benefits only emerge after months of individualised tuning — tuning that happened during the open-label phase, after the blinded window had closed.

Corn: And that third one is the explanation the field itself chose to run with.

Herman: Because there's evidence for it. And this is where the revival begins. Between 2008 and 2012 — the BROADEN era — targeting meant fixed anatomical coordinates. Same spot, every brain, on the theory that Area 25 is Area 25. What tractography imaging has since shown is that what predicts benefit isn't the grey matter point — it's which white-matter tracts your electrodes happen to sit near and engage. The bundles of cables, not the junction box. Two patients can have electrodes a couple of millimetres apart, in the same named region, and be getting entirely different treatments.

Corn: So the same surgery, two millimetres apart, is two different drugs.

Herman: That's the cleanest way to put it I know. And it explains the trial failures without declaring the mechanism dead: a fixed coordinate in a heterogeneous population means some fraction of implanted patients got a well-placed intervention and a larger fraction got, effectively, a near miss. Averaged together, that washes out against sham.

Corn: I'd like to push on something, though. The wash-out explanation assumes the near-miss is neutral. But in RECLAIM the unstimulated patients did better than the stimulated ones. A near miss isn't zero — it's slightly negative.

Herman: That's a fair hit, and I don't have a clean answer. Stimulation at an unhelpful site isn't necessarily harmless — there can be mood effects, anxiety, agitation when you're in the wrong tract. My guess is it's a mix of small negative effects in some, nothing in most, and the whole thing swamped by regression to the mean in a group of thirty. But you're right that "the targeting was just off" shouldn't be waved around as if it explains everything cleanly.

Corn: So the trials failed. But the field didn't abandon the idea — it went back to the drawing board on targeting, and that's where the revival comes from.

Herman: And the revival has two prongs. The first is connectomic targeting with recording devices, and the landmark is the twenty twenty-three Nature paper from Alagapan and colleagues. Ten patients, subgenual cingulate DBS, but with a new generation of device that can record the local electrical activity — the local field potential — while stimulating. At twenty-four weeks, ninety percent showed a robust clinical response. Seventy percent were in remission.

Corn: Ninety and seventy. In the same disease where the big trials couldn't beat sham by three points.

Herman: And open-label, yes — hold that caveat, it's load-bearing. But the recording is the real news. Using signals from six of the participants, they ran an explainable AI approach to find changes in the local field potential that tracked the patient's actual clinical state — a biomarker. Distinct from the transient artefact of stimulation itself, sensitive when the clinicians adjusted the therapy, accurate at capturing each person's recovery. For the first time, the device isn't just delivering current and hoping — it's reading the circuit back.

Corn: Which answers the other problem Daniel's prompt gestures at — how do you measure depression objectively? The clinic's instrument is a questionnaire and a sad conversation.

Herman: And subjective symptom reporting is exactly what the review literature flags as a bottleneck — the twenty twenty-four Molecular Psychiatry review by Johnson, Okun, Scangos and colleagues says the effectiveness has been demonstrated in meta-analyses and open-label work but the randomised trials are mixed, and one of the key challenges is tracking symptoms objectively to evaluate the therapy. If the biomarker work holds up, you can dose against a signal instead of against a mood scale.

Corn: There was a lovely detail in that study too — recovery predicted before anyone switched anything on.

Herman: Yes — the variable recovery trajectories were predicted by the degree of preoperative damage to the structural integrity and connectivity of the targeted white-matter network. Read that again: how well you do was partly visible in the wiring before the surgery. Which immediately raises the obvious clinical question — should you implant at all in the patients whose networks look badly damaged? That's a selection question, and the field is starting to answer it. A twenty twenty-four study of nineteen patients found that regional brain glucose metabolism predicted response with eighty-four percent accuracy — better than blood flow or grey-matter volume, both around seventy-four. And a -analysis of machine-learning models in this space pooled sensitivity of about zero point seven four. Not good enough to deny someone surgery, but good enough to start a conversation about who's a good candidate.

Corn: And that matters because this is not a risk-free intervention. You're drilling a skull.

Herman: The second prong is the closed-loop approach, and I want to be very precise about its status, because it's the part that gets over-sold. There is exactly one published individual. Nature Medicine, twenty twenty-one. A woman with treatment-resistant depression, and the workup was extraordinary — days of intracranial recording and focal stimulation across multiple sites, mapping which contacts produced which emotional responses, to find a personalised biomarker and a personalised treatment location.

Corn: What was her biomarker?

Herman: Gamma activity — fast oscillations — in the amygdala, appearing when her depressive symptoms came on. The treatment site was the ventral capsule and ventral striatum, the reward-circuitry region that RECLAIM had targeted with such disappointment. They implanted a chronic sensing-and-stimulating device, and when the amygdala signature appeared, it delivered about one milliamp at that site. On demand, only when needed.

Corn: A demand pacemaker for mood.

Herman: And she had rapid and sustained improvement. But the authors themselves are explicit that generalisation requires further work. One person. One biomarker, one site, one wiring diagram that is hers. The follow-up protocol now being used — the staged workup — involves implanting ten recording electrodes across five brain regions bilaterally and monitoring for days before you even choose the permanent target. This is bespoke medicine in the most literal sense: the intervention is designed after the implant, not before.

Corn: Which is a strange sentence. Usually you design the treatment and then give it to the patient.

Herman: It's the inversion. And that inversion is expensive, it's slow, it's a hospitalisation — and it might be the only honest way to do it, if the disease really is heterogeneous. That's the definition problem sitting underneath everything. Treatment-resistant depression is defined by failing two antidepressants. Not by any biology — by a treatment history. It's entirely possible that the label covers several different circuit disorders that all present as depression, in which case a fixed target was never going to work for everyone, and the connectomic, closed-loop approaches aren't a refinement — they're the acknowledgement that the category itself was too coarse.

Corn: We touched this exact nerve in the episode about whether therapy and pills produce brain changes that last — the question there was whether the intervention rewrites the circuit or rides on it. Here it's the same question with the skull open. Corn... sorry, Herman, that's yours.

Herman: It is, and the honest version is nobody knows yet whether DBS for depression rewires anything permanently or whether it's a very expensive ongoingsubscription. The two-year open-label data says the benefit persists while the device works — that's what we can say.

Corn: Now the part I've been waiting for. The outsiders.

Herman: The ultrasound work! Yes. Two strands, and both matter. The first, published last year in Biological Psychiatry: twenty-two people with treatment-resistant depression, randomised, double-blind, sham-controlled — low-intensity focused ultrasound aimed at the subgenual cingulate from outside the head. Functional MRI during stimulation showed a target-specific decrease in SCC activity — that's real engagement of the same region all the electrodes are aimed at — and in the per-protocol sample, real stimulation beat sham on depression scores at twenty-four hours.

Corn: Sham-controlled. Positive. Those four words haven't appeared together in this episode yet.

Herman: In the non-invasive arm, they have. And the second strand is wilder — the first human use of metalens-based transcranial ultrasound. The problem with getting ultrasound through a skull is that bone distorts the beam, so this group used patient-specific metamaterials — a lens 3D-printed to correct for each person's skull aberrations — targeting the individual's white-matter tracts of the subgenual cingulate. Note that: tract-based targeting, the same lesson the electrode people learned, applied from outside the head.

Corn: How many people?

Herman: Five. Open-label. Five days, twenty-five sessions of five minutes each. By day five, depression severity down an average of sixty-one percent, range thirty to eighty-four. No serious adverse events.

Corn: Five days. The electrode people need two years of tuning.

Herman: I would put a hand on that enthusiasm — five people, open-label, and we've spent this whole episode learning exactly what five people and an open-label design can do to your judgement. Mayberg's first six patients also did amazingly well. But the direction of travel is what matters: if you can reach the deep circuits without drilling, the entire risk-benefit calculation changes. No haemorrhage risk, no battery replacements, no lead revisions — and a hundred million people worldwide with this condition. The authors make exactly that point: implantation poses significant risks and limits widespread use. You cannot scale brain surgery to a hundred million people. You can scale a clinic with an ultrasound wand and a printed lens.

Corn: And if the dose is five days rather than two years of programming, you can run the trials faster and cheaper too.

Herman: The trials are the choke point for everything. Where does the science actually stand, November twenty twenty-six? No published, adequately powered, positive randomised sham-controlled trial of subgenual cingulate DBS for treatment-resistant depression. None. No regulatory approval anywhere for DBS in depression. The closed-loop personalised approach rests on one published individual. The best sham-controlled trial we have is this year's Canadian study, and it found nothing in the blinded phase while its open-label extension produced two-thirds to three-quarters response. That's the state of play, honestly held.

Corn: It's a strange kind of progress. The field didn't answer "does it work." It dissolved the question into "for whom, at which tracts, measured how."

Hilbert: — and tuned, that's the part you're both circling. It's not the operation, it's the two years of appointments afterward.

Hilbert: I rang a man this afternoon while you were going on about the Canadian numbers. Met him on a train from Liège in oh eight, sat opposite me the whole way, right hand shaking so he couldn't fill in his crossword. Essential tremor, not depression. Had the implants done in Bristol. We exchanged addresses because he wanted the name of a glue I'd mentioned, and he rang me back about the glue, which is more than most people manage.

Hilbert: He told me the surgery was a Wednesday and he was home Friday. Then the work started. Eleven programming sessions over nineteen months. Eleven. Each one, a technician with a handset the size of a pocket calculator changes numbers, asks how he feels, sends him home for a fortnight to find out. He said the worst part was the half-second after they adjust it, a sensation like a coin dropping into water somewhere behind his eyes, and then nothing you can name.

Hilbert: Corn, you did forty minutes on whether it works and one line on the battery, which is the bit that actually recurs. His got replaced in a second surgery under local, he was awake and complaining throughout. Herman, keep your levels up, your left channel's dropped twice since the ultrasound fellow.

Hilbert: Anyway. He said the thing nobody prints: for the first four months he couldn't sign his name and it came back wrong. Better, but wrong. Looked like someone else's handwriting. He kept a sample in his wallet, both versions, to show people what the machine did. I asked him whether he'd do it again and he said he'd do it Tuesday. It's a Tuesday, so I expect he's somewhere doing it.

Corn: Nineteen months of tuning for a tremor, where the target's been settled for decades. Now imagine that tuning against a questionnaire.

Herman: That's actually the most sobering thing anyone's said today. The Parkinson's and tremor programming is hard enough — and the target's a motor circuit, you can watch the hand and adjust. In depression, the adjustment signal is the patient's self-report, weeks later. The biomarker work is precisely the attempt to swap that out for something you can read in real time.

Hilbert: The handset's the problem as well. He said the technician had three screens and a laminated card. Laminated. Nineteen months and the card was laminated.

Corn: The card being laminated suggests it was written more than once, which is more than can be said for some trial protocols.

Herman: I'll grant the programming burden, though the recording devices are the first real attempt at it — the Nature study's whole point was that the signal tracks recovery, so the handset gets a screen worth trusting. Eventually.

Corn: Eventually is carrying a lot of weight in this field.

Herman: It always has. That's the pattern of the whole twenty-year arc, isn't it — Mayberg's six patients, the crash, the connectomic rebuild, and every stage taking a decade longer than the people inside it expected. The twenty twenty-six critical review in Frontiers in Psychiatry frames it as psychosurgery becoming circuit-based medicine — less an invasive surgical discipline, more a way of steering networks. I buy the framing. I'd just note that medicine became circuit-based the day it started measuring circuits instead of asking people how they felt on a scale of one to ten, and that day hasn't quite arrived.

Corn: Whereas Hilbert's man was measured in crosswords, signatures, and whether he'd do it on a Tuesday. Cheap instruments, honest readings.

Herman: The signature thing isn't trivial either — handwriting change after DBS is documented in the movement-disorder literature, and it's exactly the kind of objective, functional readout the depression field keeps saying it needs and doesn't have.

Corn: So here's where we land. The open question I'd watch is the simplest one: can anyone, anywhere, run a sham-controlled trial of subgenual cingulate DBS for depression that actually comes out positive? Because until that happens, the field is asking regulators and patients to accept two decades of open-label conviction and one remarkable closed-loop case of one — against a trial record that keeps saying no. And behind that, the race: ultrasound reaching the same tracts without a drill. If five people and a printed lens can do what two years of programming does, the surgeons are out of the loop.

Herman: The optimistic read is that this is one of the rare stories where both the sceptics and the believers have been right the whole time — the trials really did fail, and the mechanism really is probably sound. The error was thinking one coordinate could be everyone's cure.

Corn: Thanks as always to Hilbert Flumingtop at the desk, who fixed our levels mid-story and didn't ask for thanks.

Corn: For more along these lines, there's episode twenty-five twenty-seven, Do Brain Changes from Therapy or Pills Actually Last; episode eighteen, Beyond the GPU; and episode twenty-five twenty-nine, Depression Subtypes. This has been My Weird Prompts. If you found this worth your time, subscribe, tell a friend, and leave a review — it helps more than you know.

Herman: We'll be back soon. See you tomorrow, brother.
