Finn Brooks: Hey, quick question before we start — have you ever caught yourself doing something repetitive and just thought, wait, is this actually helping me or am I making it worse?
Juniper Vale: I mean — yeah, honestly, more times than I'd like to admit. Why?
Finn Brooks: Because that instinct — that nagging feeling — is basically the whole neurological tragedy of OCD, and Andrew Huberman just laid it out in this Huberman Lab Essentials episode from July 9th, 2026, and I cannot stop thinking about one specific thing he said.
Juniper Vale: Okay, what's the thing?
Finn Brooks: The compulsion is not the symptom — it is the mechanism that keeps the disorder running. Like, wait, actually let me be precise here — the thought-action loop goes: obsessive thought, compulsive response, real genuine anxiety relief, and then the obsession comes back stronger. Not weaker. Stronger. The brain literally rewards you for feeding the loop.
Juniper Vale: That is a brutal design flaw.
Finn Brooks: Right — and it's not a character flaw, it's a circuit disorder, the cortico-striatal-thalamic loop, which is what Huberman frames as the center of all of this. And it touches 2 to 4 percent of people, which is — dude, that is not rare, that is someone in most people's immediate circle. So today we're getting into how that loop works, why relief is literally the trap, and what that means for treatment.
Juniper Vale: And whether understanding the circuit actually translates into helping people — because I think that gap is more complicated than it looks.
Finn Brooks: But that gap — okay, I want to make sure we're actually explaining the engine before we get to what breaks down, because I blew past it.
Juniper Vale: Yeah, exactly — you can't call compulsions the cause without explaining what's generating the pressure in the first place. Think of it like this: your brain has a smoke alarm, and in OCD that alarm gets stuck. It fires even when there's no fire. And the compulsion? That's you yanking the battery out for thirty seconds of quiet. Real quiet. Neurochemically real. But the alarm rewires itself to be louder the next time you do that.
Finn Brooks: The rewiring is the trap.
Juniper Vale: That's the whole thing. And what Huberman is pointing to — the cortico-striatal-thalamic loop — is the actual hardware running that alarm. The cortex, the striatum, the thalamus, all feeding signals back and forth. And the thalamus is acting like a gate, you know, controlling what actually reaches conscious experience. So when that loop gets dysregulated, the thalamus keeps letting the signal through. You can't not hear it.
Finn Brooks: Wait — so the gate is stuck open?
Juniper Vale: Essentially, yeah. And the Yale 2026 Annual Review of Clinical Psychology paper — this is from Christopher Pittenger's group — they frame it as a frontostriatal imbalance. The ventromedial part of the circuit is over-coupled, too loud, and the dorsolateral part is under-coupled, too quiet. The part that would normally pump the brakes has less pull.
Finn Brooks: Okay that is — wait, that's not just 'the alarm is broken,' that's like, the volume knob and the off switch are on opposite ends of a seesaw and one side is always winning.
Juniper Vale: That's a good way to put it. And the part that I think matters most — compulsions aren't just habits. They map onto habit-formation processes in the basal ganglia specifically, which means they become self-reinforcing at the circuit level. It's not stubbornness. The basal ganglia is literally encoding 'do this again.'
Finn Brooks: So when someone says 'just stop' — they're asking the person to override a circuit that is actively practicing the behavior. Every single time.
Juniper Vale: And that's exactly where the treatment story gets uncomfortable — because the main pharmacological answer we've had for decades is SSRIs. That's the anchor. And they work... until they don't, for a lot of people.
Finn Brooks: Okay wait, this is the part that actually got me — like, we have this beautifully mapped circuit and the drug we keep reaching for doesn't have a clean story for how it even fits into that circuit. Serotonin's role in the cortico-striatal-thalamic loop is not — I mean, we can't fully say 'serotonin fixes the gate.' That's not where the science lands.
Juniper Vale: No, that's fair. That's the honest version. And yet SSRIs do outperform placebo in individual patient data meta-analyses — that's real, partial response is real. So dismissing them causes harm.
Finn Brooks: Right — partial win for the circuit people though, because if you can't map serotonin cleanly into the multi-system loop, the incomplete efficacy starts making total sense. Like, of course it's patchy.
Juniper Vale: Okay, but here's the part I want to make concrete — picture someone sitting in a clinic waiting room, six weeks into sertraline, nothing has shifted. And they're thinking, am I just the wrong kind of OCD patient? The NIMHANS meta-analysis actually names who that person might be: high baseline severity, poor insight going in, comorbid tics. Those are the three predictors of SSRI non-response they pulled out.
Finn Brooks: Wait, comorbid tics specifically?
Juniper Vale: Specifically tics, yeah. Which — I mean, that's actionable. If you know those are the flags, you can identify those patients earlier instead of running out a full trial and calling it failure.
Finn Brooks: And there's the longitudinal fMRI work — they used paroxetine in drug-naive OCD patients, looking at striatal changes before and after treatment, and that's where you can actually see the circuit responding or not. And if nothing moves early? Meta-analytic evidence says early non-improvement predicts eventual treatment failure. That six-week waiting room person is not failing — that is data.
Juniper Vale: That reframe matters so much. And when SSRIs aren't enough, antipsychotic augmentation is on the table — adjunctive, not first-line, but it exists. The toolkit isn't just one thing.
Finn Brooks: Which actually sets up the harder question — and we'll get into this — Huberman's episode can hand someone the circuit map, but there's no version of that episode that tells you what your specific circuit is going to do with a drug.
Juniper Vale: And that's the thing that doesn't get said cleanly enough — because the ENIGMA-OCD Consortium just ran a transformer model on resting-state fMRI from 1,706 participants, and the overall AUROC was 0.653. Which is... I mean, that's modest. That's not broken science, but it's not a diagnostic tool.
Finn Brooks: Wait — some sites hit 0.427?
Juniper Vale: Worse than chance at some sites, yeah. The leave-one-site-out performance ranged from 0.427 to 0.819. Same model, same circuit knowledge, completely different reliability depending on where you run it.
Finn Brooks: Okay that — no, that genuinely breaks the story. Because Huberman Lab Essentials is designed to reach people who will never read the preprint, which is actually good, I want that, but if someone hears 'we've mapped the circuit' and then assumes the science can already tell them what their brain will do — they might just... skip the ERP waitlist? Like, 'I understand my diagnosis now, I'm good.'
Juniper Vale: That's the real risk. Exposure and Response Prevention — actually sitting with the feared stimulus and not doing the compulsion, repeatedly, in a clinical process — that takes sustained work with a therapist. No episode conveys that duration.
Finn Brooks: So the calibrated version is what, exactly?
Juniper Vale: The circuit framing is real and consistent — decades of neuroimaging, the Yale paper, all pointing the same direction. Understanding OCD as a loop disorder rather than a willpower failure is a genuine shift and it matters. But — and this is the part that actually holds up — that map cannot tell you how your circuit responds to sertraline, or whether ERP will click in eight weeks or twenty. The science hasn't solved individual prediction yet. 0.653 is the honest number.
Finn Brooks: So Huberman's episode democratizes the knowledge — that part's real — but it can't close the gap between the population-level circuit story and what any one person's brain is actually going to do.
Juniper Vale: That's it. That's the defensible claim. Understanding the circuit is a real shift — bypassing the clinical process because you understand it is where it goes wrong.
Finn Brooks: So my hot take at the top was right — compulsions are the trap — I just, I mean, I kind of glossed over the part where the trap is built by a circuit we can describe in extraordinary detail and still cannot predict. Like, we can draw the cortico-striatal-thalamic loop on a whiteboard and we still can't tell you whether sertraline is going to do anything for your specific version of it. That's — yeah. That's the actual state of play.
Juniper Vale: And that's not going to change with a better scan. The next question isn't sharper neuroimaging — it's why the same broken loop responds so differently from person to person. Even the three-year pediatric follow-up comparing sertraline against continued CBT for non-responders, that data exists, and long-term guidance is still limited. We've had decades of frontostriatal research and individual variation is still the open problem.
Finn Brooks: Which is almost exactly where we started — that nagging feeling of, wait, is what I'm doing helping or making it worse? Turns out that question is still genuinely hard, even for the researchers.
Juniper Vale: Yeah. Unresolved is sometimes the honest answer. Good thinking it through with you.
