What the Placebo Effect Actually Is and Is Not

Few phenomena in medicine get misrepresented from as many directions at once as the placebo effect. Advocates of alternative therapies cite it to legitimize treatments that have no specific efficacy; skeptics sometimes wave it away as mere self-deception. Both moves are mistakes.

What the evidence actually shows

The placebo effect is not a single thing. Researchers distinguish several mechanisms that tend to get lumped together under one label. Expectation effects occur when a patient's anticipation of improvement produces genuine physiological changes—measurable shifts in neurotransmitter activity, cortisol levels, and immune markers. Conditioning effects follow a Pavlovian logic: if a patient has previously responded to a treatment, the ritual of receiving something similar can trigger the same biological response even when the active ingredient is absent. Regression to the mean is different again—it is not a placebo effect at all, but a statistical artifact. People tend to seek treatment when symptoms are at their worst, so improvement often follows regardless of what is administered. Conflating these mechanisms produces bad science and worse medicine.

Controlled trials that compare a treated group against a no-treatment group, rather than only against a placebo group, allow researchers to separate genuine placebo responses from natural recovery. When this design is used, the placebo effect shrinks considerably for many conditions. A landmark Cochrane review found that placebo interventions produce modest, sometimes undetectable effects on objective outcomes like survival or laboratory values, while subjective outcomes like pain and nausea show larger and more consistent responses. This pattern is exactly what you would expect if expectation and conditioning work primarily by modulating how the nervous system processes and reports experience.

Open-label placebos and what they complicate

One of the more surprising recent findings is that open-label placebos—placebos given with explicit disclosure that they contain no active ingredient—can still produce measurable relief in conditions like irritable bowel syndrome and chronic lower back pain. This seems paradoxical, but it is consistent with conditioning-based accounts: if the ritual of pill-taking has been reinforced enough times in a patient's history, the conditioned response may run partially independently of conscious belief. It also suggests that the deception traditionally assumed to be necessary for placebo effects is not strictly required.

This matters for medical ethics. If deceptive placebos work partly because of conditioning rather than entirely through false belief, then the case for using them without disclosure weakens further—not just on the grounds that deception violates patient autonomy, but because the deception may not even be doing the causal work assumed. Open-label protocols sidestep the ethical problem while preserving some of the physiological benefit. Whether this approach scales across conditions and patient populations remains an active research question.

Why this matters for evaluating alternative medicine

The placebo effect is frequently invoked to explain why patients report improvement after receiving treatments that trials show to have no specific efficacy—homeopathy, certain forms of energy healing, and various ceremonial interventions. The inference drawn is sometimes: "but it works, so it must have value." This argument needs to be handled carefully rather than simply dismissed.

The relevant question is whether the placebo response produced by an inert treatment is larger than what a similarly delivered, evidence-based treatment would produce, and whether the comparison includes any risks the inert treatment carries—including the risk of displacing effective care. On those tests, inert treatments rarely win. A warm, attentive consultation with an evidence-based practitioner produces comparable expectation and conditioning effects to a warm, attentive consultation with a homeopath, plus the specific efficacy of the real treatment on top. The argument from placebo does not rehabilitate ineffective treatments; it points instead toward improving how effective treatments are delivered.

There is also a harder point. When a patient reports feeling better after an inert treatment, that report can reflect genuine relief from subjective symptoms, or it can reflect social pressure, desire to please the practitioner, or unwillingness to admit the decision to seek the treatment was mistaken. Disentangling these requires more than patient testimony. It requires the controlled designs that alternative practitioners routinely resist subjecting their methods to.

The honest summary

The placebo effect is a real collection of psychobiological phenomena, not a wastebasket for results we do not understand. It operates through expectation, conditioning, and the social context of care. It is most robust for subjective symptoms and smallest for objective disease endpoints. It does not require deception to function, a fact with significant implications for clinical ethics. And it does not, by itself, validate treatments that lack specific efficacy. Understanding it precisely is not a way of dismissing it—it is the only way to use the knowledge productively.