"It's just placebo" functions as a dismissal — the implication being that nothing happened.
The research says otherwise. Placebo effects are measurable, have identifiable neurobiological mechanisms, and produce real changes in symptoms. Understanding them properly is more interesting than either dismissing or overclaiming them.
What placebo effects are
An improvement in symptoms following an intervention with no specific therapeutic mechanism for those symptoms.
Critically, this is distinguished from other things that get confused with it.
Natural history. Many conditions improve on their own. Improvement after any intervention partly reflects this.
Regression to the mean. People seek treatment when symptoms are at their worst. Subsequent measurement tends towards their average regardless of treatment.
Reporting effects. People may report improvement to please a clinician or to justify participation.
Studies comparing placebo groups against no-treatment groups — rather than assuming placebo group improvement is the placebo effect — find smaller effects than the raw placebo group change, which tells you how much of the apparent effect is these other factors.
What's genuinely established
Even after accounting for the above, real effects remain, particularly in some domains.
Pain. The best-studied. Placebo analgesia has been shown to involve endogenous opioid release — demonstrated by the finding that blocking opioid receptors reduces placebo analgesic effects. That's about as direct a mechanistic demonstration as this field offers.
Brain imaging has shown changes in pain-processing regions during placebo analgesia.
Parkinson's disease. Studies have shown dopamine release in response to placebo administration in Parkinson's patients, with corresponding motor improvement.
Nausea, anxiety, depression, fatigue. Substantial placebo responses documented, which is one reason trials in these areas require large samples.
Notably, these are all conditions with a substantial subjective or centrally mediated component. Placebo effects on objective disease processes — tumour size, infection clearance, fracture healing — are not established, and this is the crucial limit.
The mechanisms
Two principal ones, and they operate differently.
Expectation. Conscious belief that improvement will occur. Manipulable through information, and dose-dependent — stronger expectations produce larger effects.
Conditioning. Learned associations from prior experience. Repeatedly pairing a treatment with an effect can produce the effect from the inert stimulus alone, demonstrated in both animals and humans.
Conditioning explains why placebo effects are stronger in people with prior positive experience of a treatment, and why they can occur without conscious expectation.
Open-label placebo
The most surprising finding in this area, and one that overturns an assumption.
Trials have administered placebo openly — telling participants explicitly that they're receiving an inert substance with no active ingredient — and still found symptom improvement in several conditions including irritable bowel syndrome, chronic low back pain and cancer-related fatigue.
These are relatively small studies and the finding has replicated across several. It suggests deception isn't necessary, which changes the ethical picture considerably.
The mechanism is unclear. Proposals include conditioning operating below conscious belief, and the ritual of treatment itself carrying effect.
The context effect
Related and clinically relevant. Elements surrounding a treatment influence outcomes.
Research has found that factors including the clinician's manner, the time spent, the perceived expense of treatment, the invasiveness of the procedure and the setting all modulate response.
Injections produce larger placebo effects than tablets. More frequent dosing produces larger effects than less. A warm, confident clinician produces better outcomes than a brusque one.
That last finding is significant beyond placebo research: it suggests the relational aspects of care have measurable effects on outcomes, which is an argument for consultation time that's usually made on other grounds.
Nocebo
The inverse and considerably less discussed. Negative expectations produce negative effects.
Participants given inert substances report side effects at substantial rates, matching those listed for the active drug they believe they might be taking.
This has real clinical consequences. Information about potential side effects, delivered in certain ways, may increase their occurrence. There's an ethical tension between informed consent and not inducing symptoms through information.
What this means practically
Several things.
Dismissing an effect as placebo doesn't mean the person didn't improve. It means the improvement wasn't caused by the specific mechanism claimed.
That distinction matters for treatment decisions. A treatment working through placebo effects is real relief and doesn't have the specific effect claimed — which matters enormously if you're relying on it to treat a serious disease process rather than a symptom.
The ethical position most commonly taken is that maximising placebo effects around evidence-based treatment is legitimate and good practice — clear communication, confidence, attention, adequate time. Prescribing inert treatment while implying it has specific effects is not.
And for anyone evaluating a therapy: the question isn't whether people feel better, because they frequently do. It's whether they feel better than they would have with an equally elaborate ritual containing nothing. That's what controlled trials measure, and it's why they exist.
General information only. Treatment decisions should be discussed with a qualified healthcare professional.
Why trials need control groups
Everything above amounts to an argument for a methodological point worth stating plainly. If improvement occurs after an intervention, you cannot conclude the intervention caused it, because natural history, regression to the mean and placebo effects all produce improvement on their own.
The only way to isolate a specific effect is to compare against a group receiving everything except the active component. That is not bureaucratic obstruction; it is the minimum required to distinguish a treatment from a ritual.
Which is also why testimonials, however sincere, cannot establish that something works. The people giving them genuinely improved. What nobody can tell from a testimonial is whether they would have improved anyway.