Onpode
Cover art for Why belief changes biology — the neural mechanisms behind expectation effects

Why belief changes biology — the neural mechanisms behind expectation effects

November 20, 2026 · 14 min

Juniper Vale & Finn Brooks

Hey — before we start, I need you to explain something to me, because you sent me a PET scan image at eleven p.m. with no context. Okay, fair, I owe you context. The image was from a study on Parkinson's Disease patients. Researchers gave them a placebo — genuinely inert, no active compound —…

Placebo and nocebo effects are psychobiological responses driven by expectation, learning, and social context — not deception or imagination. A placebo effect is a measurable improvement following an inert treatment, attributable to the brain-mind response to the treatment context rather than any pharmacological action.

0:0013:32
Get the next episode on Health →

Follow it free — new episodes land in your feed.

Or make your own — any topic, in minutes

More Onpode episodes on Health →

About this episode

Placebo and nocebo effects are psychobiological responses driven by expectation, learning, and social context — not deception or imagination. A placebo effect is a measurable improvement following an inert treatment, attributable to the brain-mind response to the treatment context rather than any pharmacological action.

Grounded in 10 sources
A Comprehensive Scientometric Analysis of Placebo and Nocebo Effects in Pain Research: From Placebo Mechanisms to Clinical Trials ↗ · doi.org
Placebo and nocebo effects on pain through the lens of the predictive brain: Neurobiological mechanisms and translational implications ↗ · doi.org
The Roles of Endogenous Opioids in Placebo and Nocebo Effects: From Pain to Performance to Prozac. ↗ · doi.org
Justice for Placebo: Placebo Effect in Clinical Trials and Everyday Practice ↗ · pmc.ncbi.nlm.nih.gov
Effects of open-label placebos in clinical trials: a systematic review and meta-analysis | Scientific Reports ↗ · nature.com
Placebo and nocebo effects: from observation to harnessing and clinical application | Translational Psychiatry ↗ · nature.com
Clinical neuroscience and neurobiology of placebo and nocebo effects ↗ · pmc.ncbi.nlm.nih.gov
The neuroscience of placebo effects: connecting context, learning and health ↗ · pmc.ncbi.nlm.nih.gov
Communication, expectancy, and placebo response in clinical trial ... ↗ · thelancet.com
Placebo Effects: Neurological Mechanisms Inducing Physiological, Organic, and Belief Responses—A Prospective Analysis ↗ · mdpi.com
Read transcript

Juniper Vale: Hey — before we start, I need you to explain something to me, because you sent me a PET scan image at eleven p.m. with no context.

Finn Brooks: Okay, fair, I owe you context. The image was from a study on Parkinson's Disease patients. Researchers gave them a placebo — genuinely inert, no active compound — and then ran PET imaging on the striatum afterward. And the scan showed measurable endogenous dopamine release. Like, the brain produced dopamine. Because it expected to.

Juniper Vale: And Parkinson's is specifically a disease of dopamine loss in the striatum.

Finn Brooks: Exactly — so this isn't some peripheral system, this is the broken part. The part that makes hands shake. And the placebo partially fixed it. Not metaphorically. Neurochemically.

Juniper Vale: You know, people hear 'placebo' and they hear 'fake medicine.' But if someone's hand is tremoring at nine in the morning, they swallow an inert pill, and within the hour the tremor eases — and we can point to real dopamine on a scan as the reason — what exactly is fake about that?

Finn Brooks: Nothing! That's — nothing is fake about it. The word is doing all this work to distance us from something that is just... biology the brain initiated on its own.

Juniper Vale: Which makes the design problem for Randomized Placebo-Controlled Trials really interesting, right? Because the whole architecture of an RCT is built around subtracting the placebo response to find the drug's 'real' effect. But if the placebo response involves real dopamine release —

Finn Brooks: What are we subtracting?

Juniper Vale: What are we subtracting. Yeah.

Finn Brooks: And then the deeper thing — I keep wondering where this stops. Like the brain can apparently negotiate dopamine in the striatum just on the basis of expectation. Which systems can it actually reach, and which are just completely off the table? That feels like the real question.

Juniper Vale: And that's exactly where naloxone comes in, because it kind of answers your question — which systems, and how do we know.

Finn Brooks: Wait, the overdose reversal drug?

Juniper Vale: Same drug, yeah. So what researchers did — and this is the landmark naloxone-blockade experiment — they gave people a placebo for pain, let the relief kick in, and then administered naloxone, which is an opioid antagonist, it blocks opioid receptors. And the pain relief just... vanished. Completely reversed.

Finn Brooks: Okay hold on. That means the placebo was working *through* actual endogenous opioids the brain was releasing on its own.

Juniper Vale: Exactly. Because if it were just — I mean, if it were a reporting shift, like 'I feel better because I expect to feel better,' naloxone wouldn't touch it. You can't block a mood with an opioid antagonist. But it did block it. Fully.

Finn Brooks: That's like — the brain heard 'relief is coming' and started synthesizing morphine. Its own morphine.

Juniper Vale: That's not an analogy, that's literally what happened.

Finn Brooks: And so where does that signal go? Like the brain makes the opioids — but then what, they just float around and hope they find pain somewhere?

Juniper Vale: No — and this is the part that actually got me. There's a 2009 fMRI study, Eippert and colleagues, and they showed that placebo analgesia reduces pain-related activity in the dorsal horn of the spinal cord. The ipsilateral dorsal horn. That's not brain activity. That is the spinal cord, where the pain signal first arrives.

Finn Brooks: It travels *down*?

Juniper Vale: It travels down. Through the prefrontal-cingulate circuits, through the Periaqueductal Grey in the brainstem — which is this key relay station in descending pain control — and all the way to the spinal cord, quieting the signal almost at the source. That's a top-down neurochemical command.

Finn Brooks: So it's not that the brain reinterprets a pain signal coming up — it's actively suppressing the signal before it even fully arrives. And you've got opioids doing some of that work, dopamine from the Parkinson's data doing some of it — are there other systems, or is it mostly those two?

Juniper Vale: Endocannabinoids are also implicated. So you've got three neurochemical systems — endogenous opioids, dopamine, endocannabinoids — all physically altered by what the brain *expects* is about to happen. And that's where I want to slow down, because the word 'expectation' still sounds soft. But we've just walked through pharmacological proof. This is not a mental mood. It's a cascade.

Finn Brooks: Three neurochemical systems — okay, but that's exactly what makes me want to break something, because if the brain can reach the striatum, the spinal cord, the opioid receptors — why can't it reach a bacterial infection? Like, why is there a wall?

Juniper Vale: That's the question, right. And the honest answer is — the placebo effect is robust and well-replicated for pain, for motor performance, for the subjective stuff. But when you look at instrument-measured biological endpoints — infection markers, tumor size, cellular damage — it's mostly absent.

Finn Brooks: Mostly.

Juniper Vale: Mostly — and that word is doing real work. I mean, I don't want to draw a clean line and say 'placebo stops here' as if that's proven. It's genuinely an unresolved research question. We know where it reliably works. We don't have a complete map of the ceiling.

Finn Brooks: Okay but — wait, is there actually a mechanistic reason though? Like, is there something about bacteria that the brain's predictive system just... has no cable connected to?

Juniper Vale: Think of it like this. Pain and motor function are perception-dependent systems — they're literally built to be modulated by context, attention, expectation. The whole point of descending pain control is that the brain negotiates the signal. But a bacterial load, tumor cells dividing — those aren't running on prediction-sensitive firmware. The brain doesn't have a direct dispatch line.

Finn Brooks: So it's a — it's a hardware versus software thing. Some systems are just not wired to receive that top-down command.

Juniper Vale: That framing works, yeah — but overclaiming in either direction is unjustified right now. Saying placebo can't touch immune signaling is different from saying it probably can't. Those are different scientific claims.

Finn Brooks: Right — but this is what actually breaks the clean version of this whole story. Open-label placebos. You strip out the deception entirely — the doctor literally says 'this pill is inert, we know it's inert' — and you still get a beneficial effect compared to no treatment. How does the brain's predictive machinery even fire if you've told it there's nothing to predict?

Juniper Vale: That's the part that reshuffles everything. Because OLPs still work — smaller effect than deceptive placebos, meta-analyses show honesty reduces but doesn't eliminate the response — but they work. Which means the mechanism isn't just 'fooling the brain into expecting relief.'

Finn Brooks: So what is it then? Like, the ritual? The relationship? The act of being cared for?

Juniper Vale: Probably all of that, and we can't fully disentangle them yet. The nervous system seems to respond to the structure of care itself — independent of deception. And actually, that opens something we need to get to, because the dark mirror of all this is that the same mechanism runs in reverse — and what it does to informed consent is... genuinely disturbing.

Finn Brooks: Oh no. Yeah. Yeah, let's go there.

Juniper Vale: The dark mirror is nocebo. And it's not a metaphor — if the brain can manufacture endogenous opioids from positive expectation, it can manufacture pain from negative expectation. Same machinery, opposite direction.

Finn Brooks: Wait, real pain? Not just 'I feel worse' — like, measurable?

Juniper Vale: Real pain. And there's a neurochemical substrate for it — cholecystokinin, CCK. It's a neuropeptide, and it basically does the opposite of what endogenous opioids do. It inhibits them. So when negative expectation fires up CCK, it's not just failing to produce pain relief — it's actively blocking the pain relief you might have gotten anyway.

Finn Brooks: So CCK is like... the nocebo's own chemical. It's not just the absence of the good stuff.

Juniper Vale: It has its own substrate, yeah. Which is what makes nocebo a real pharmacological phenomenon and not just pessimism with consequences.

Finn Brooks: Okay but — hang on, because I want to push to the place that actually keeps me up at night, which is — where does this land in clinical trials? Because if nocebo is a real neurochemical event triggered by negative expectation, and you're running a randomized placebo-controlled trial, and you read patients their informed consent form listing every possible side effect—

Juniper Vale: You just dosed the control arm.

Finn Brooks: You just — yeah. You handed them the nocebo. Through a legal document.

Juniper Vale: And the data reflects it. In RCTs, patients in the placebo arm report side effects that match the side-effect profile they were warned about — not random symptoms, the specific ones from the consent form. Which means the apparent harm rate in the control group is partly an artifact of disclosure. It inflates how dangerous the placebo looks, and it muddies what we can say about the drug.

Finn Brooks: No, but that's — wait, that's genuinely disturbing. Because informed consent exists to protect patients. It's not optional. You cannot legally or ethically withhold that information. But the act of delivering it is causing real physiological harm in a subset of people. That's not a tension, that's a — I mean, there's no clean move here, right?

Juniper Vale: There isn't. And I don't want to paper over that. The researchers who study this aren't saying 'so we should hide things from patients.' They're saying — and this is the part that's genuinely unresolved — word choice inside a fully truthful disclosure is not neutral. You can say 'some patients experience discomfort' or you can say 'this may cause severe pain.' Both can be accurate. They do not produce the same CCK response.

Finn Brooks: So the clinician's framing — how they say it, the specific words — that is itself a pharmacological act. It's not bedside manner. It's dosing.

Juniper Vale: That's the hardest version of this, and I think it's the right one. We spend enormous effort regulating what goes in a pill. The sentence a doctor says before handing it over is operating on the same biological system — and right now it's almost entirely unregulated.

Finn Brooks: And that's — I mean, that's where I keep getting stuck, because we've spent this whole time building the case that placebo is real, the neurobiology is real, the naloxone reversal is real, the dopamine on the PET scan is real — and then it turns out the trial design we use to measure drugs is soaked in the very thing it was supposed to subtract. Like, a 'placebo-controlled' trial was always supposed to mean expectation-controlled. But it doesn't. Both arms are running on expectation-driven neurochemistry. The placebo arm has real opioid release, real top-down spinal suppression, real therapeutic ritual baked in. And so if a drug only barely outperforms that — wait, is the drug weak, or is the placebo response just so large it's buried the signal?

Juniper Vale: Both could be true simultaneously. That's what makes it hard to untangle.

Finn Brooks: And we might be throwing out drugs that actually work, or — worse — keeping ones that don't, because we can't see through the expectation layer clearly enough.

Juniper Vale: The thing I genuinely can't settle — and I've been turning this over — is the clinical encounter itself. If a doctor's word choice is pharmacologically active, operating on the same opioid and CCK systems as an actual compound, then where does the treatment end? I mean, we regulate the pill. We don't regulate the sentence delivered with it. And I don't know what the right framework even looks like for that.

Finn Brooks: No, that one I can't close either.

Juniper Vale: It's not a rhetorical question. I actually don't know.

Finn Brooks: I think that's — honestly, that might be the most uncomfortable place this topic lives. It's not 'placebos are powerful, cool, let's use them.' It's 'we may have been practicing medicine in ways that accidentally dose people with nocebo for decades, through legally required paperwork, and we don't have a clean path out of that.' That's not a satisfying endpoint.

Juniper Vale: No. It isn't. I think the brain being a pharmacy is a genuinely wild and kind of beautiful fact — and then you follow it all the way and you end up somewhere really uncomfortable. Both things are true.

Finn Brooks: Yeah. Good conversation. I'm going to be weird about informed consent forms for a while now, so — thanks for that.