Placebo Effect: How It Works, Nocebo & Limits

What Is the Placebo Effect?

The placebo effect is a real improvement in symptoms that comes from receiving a treatment with no active ingredient, such as a sugar pill or saline injection. It is driven by the person’s expectations, past learning, and the context of care, and it is strongest for subjective symptoms like pain, nausea, and fatigue.

At a glance:

Word origin: Latin for “I shall please.”
Main mechanisms: expectation (what you believe will happen) and conditioning (what your body has learned to associate with treatment).
Strongest for: pain, nausea, fatigue, and other symptoms people report and rate themselves.
Weakest for: objective disease processes such as tumor growth, infection, or lung function.
Its evil twin: the nocebo effect, where negative expectations produce real side effects.
Why it matters: it is the reason modern drug trials compare new treatments against a placebo rather than against nothing.

A Short History of the Placebo

For most of medical history, many treatments had no specific effect beyond the ritual of being treated. Physicians knew that patients sometimes improved anyway, and “placebo” came into medical use by the late eighteenth century to describe a remedy given more to please the patient than to act on the disease.

The modern interest dates to the mid-twentieth century. In 1955 the anesthesiologist Henry Beecher published “The Powerful Placebo,” which pooled a set of studies and claimed that roughly a third of patients improved on placebo alone. The paper was enormously influential and helped make the placebo-controlled trial the standard for testing new drugs.

It was also flawed. Later reanalyses pointed out that Beecher had not separated the placebo effect from other reasons people get better, such as the natural course of an illness. That distinction, between what the placebo itself causes and what simply happens over time, is the most important idea for understanding placebos correctly.

What Counts as a Placebo Effect (and What Does Not)

When people in the placebo group of a trial improve, it is tempting to credit the placebo. But several things are happening at once:

Natural history. Many conditions, including colds, back pain, and episodes of depression, improve on their own over time.
Regression to the mean. People tend to seek treatment, or enroll in trials, when their symptoms are at their worst. Symptoms that are unusually severe are statistically likely to be less severe at the next measurement, regardless of treatment. Our guide to research methods explains why this trips up so many studies.
Reporting bias. Participants may report feeling better to please researchers or because they expect to, without any real change in how they feel.
Other care. Being in a study often means more attention, more check-ups, and better adherence to other treatment.

The true placebo effect is only what remains after these are accounted for: the change caused by the experience of receiving treatment. The cleanest way to measure it is to compare a placebo group with a no-treatment group, which most drug trials do not include.

When researchers Asbjørn Hróbjartsson and Peter Gøtzsche reviewed trials that did include a no-treatment group, in a 2001 analysis in the New England Journal of Medicine, they found little evidence that placebos had powerful clinical effects in general. The exception was a possible modest benefit for subjective outcomes, particularly pain. That finding has shaped the field since: the placebo effect is real, but narrower than Beecher’s figure implied.

How It Works: Expectation and Conditioning

Two psychological mechanisms explain most placebo effects. They often work together.

1. Expectation

If you believe a treatment will reduce your pain, that belief changes how your brain processes pain signals. Psychologist Irving Kirsch described this as response expectancy: expecting a particular bodily or emotional response helps bring it about. Expectations come from many sources: what a doctor says, what the packaging looks like, what you have read, and what has worked for you before.

The context of treatment shapes expectation strongly. Studies of placebo pills have found that factors such as the number of pills, their color, and whether a treatment is presented as a branded product can influence how well people report it works. More invasive-seeming treatments, such as injections or procedures, tend to produce larger placebo responses than pills. The warmth and confidence of the clinician also appear to matter.

2. Classical conditioning

The body can learn to respond to the cues that accompany a real drug. This is the same classical conditioning that Ivan Pavlov demonstrated with dogs and bells. If a particular pill has repeatedly been followed by pain relief, the act of taking a similar-looking pill can come to trigger some of that relief by itself.

A striking demonstration came from psychologist Robert Ader and immunologist Nicholas Cohen in 1975. They paired saccharin-flavored water with a drug that suppresses the immune system in rats. Later, the flavored water alone produced measurable immune suppression. The study helped found the field of psychoneuroimmunology and showed that conditioned placebo-like responses are not limited to how people feel; in some cases they involve physiological systems directly.

Conditioning helps explain why placebo responses can occur even when expectations are low, and why prior experience with an effective drug tends to strengthen later placebo responses.

What Happens in the Brain

Placebo effects are not imaginary. For some conditions, researchers have traced them to specific brain chemistry.

Pain and endogenous opioids. In a 1978 study of patients after dental surgery, Jon Levine, Newton Gordon, and Howard Fields found that placebo pain relief could be blocked by naloxone, a drug that blocks opioid receptors. This suggested that placebo analgesia works partly by triggering the body’s own painkillers. Later brain-imaging work found that placebo treatment reduced activity in pain-processing regions.
Parkinson’s disease and dopamine. Imaging studies have shown that people with Parkinson’s who expect to receive medication can release dopamine in response to a placebo, and some show short-term motor improvement.
The open-hidden paradigm. Researchers including Fabrizio Benedetti have shown that the same active painkiller works noticeably less well when given secretly, by a hidden infusion, than when the patient sees it being given. Part of every real treatment’s effect, in other words, is a placebo component.

This research is a good example of how psychology and neuroscience meet: a belief, a thoroughly psychological event, has measurable effects on neurotransmitter systems. For more on the psychology of pain specifically, see our guide to chronic pain psychology.

The Nocebo Effect

The nocebo effect (Latin, “I shall harm”) is the reverse: negative expectations produce real negative symptoms. It runs on the same mechanisms of expectation and conditioning.

Nocebo effects are visible in almost every drug trial. People taking the placebo regularly report side effects, often the very side effects listed in the consent form for the active drug, such as headache, nausea, dizziness, or fatigue. Some stop taking the placebo because of them.

A clear example comes from statins, the cholesterol-lowering drugs. Many people stop taking statins because of muscle aches and other symptoms. In the SAMSON trial, published in 2020, participants who had previously stopped statins rotated through months of statin, placebo, and no pills without knowing which was which. Most of the symptom burden they reported while taking statins was also reported while taking placebo, suggesting that much of it was a nocebo response rather than a direct drug effect.

Nocebo effects raise a genuine ethical tension. Patients have a right to know about possible side effects, yet the way risks are described can make those side effects more likely. Researchers are studying ways of framing information, for example emphasizing how many people do not experience a side effect, that keep patients informed without priming symptoms. Nocebo processes may also contribute to the way worry about symptoms can worsen them, a pattern discussed on our health anxiety page.

Open-Label Placebos

For decades it was assumed that a placebo only works if the patient is deceived. Open-label placebo research challenges that assumption.

In a 2010 study led by Ted Kaptchuk at Harvard, people with irritable bowel syndrome were given pills and told plainly that they were placebos containing no medication, but that placebo pills had been shown to produce improvement through mind-body self-healing processes. Compared with a no-treatment group, those taking the open-label placebo reported greater symptom improvement.

Similar trials have since been run for chronic low back pain, cancer-related fatigue, and other conditions, with generally encouraging results for self-reported symptoms. The findings are interesting but come with caveats:

Most studies are small and cannot be blinded, since participants know whether they received a pill. That makes it hard to rule out reporting bias.
Benefits are almost entirely on subjective measures.
The explanation given to participants, which is itself designed to raise expectations, may be part of the active ingredient.

Even so, the research suggests that the ritual of treatment, and conditioning built up from past treatment, can help some people without deception, which matters for the ethics of using placebo effects in practice.

The Placebo in Drug Trials

Because people improve for so many reasons other than a drug’s chemistry, the only way to know whether a drug works is to compare it with something that looks identical but lacks the active ingredient. That is the logic of the randomized, double-blind, placebo-controlled trial:

Randomized: participants are assigned to the drug or placebo by chance, so the groups are comparable.
Double-blind: neither participants nor the researchers assessing them know who received what, which prevents expectations on either side from biasing the results.
Placebo-controlled: the drug’s real effect is the difference between the two groups, not the total improvement in the drug group.

The antidepressant debate

Placebo response is a particular issue in psychiatry. In a 2008 meta-analysis of data submitted to the US Food and Drug Administration, Irving Kirsch and colleagues argued that the average difference between antidepressants and placebo was small, and larger in more severe depression. The paper provoked a long debate. Other researchers have argued that average differences understate the benefit for some patients and that antidepressants have a clear effect, even if a large share of improvement in trials also occurs on placebo. What is not in dispute is that placebo response in depression trials is substantial. That is one reason the decision to take medication is best made with a prescriber; see our pages on depression and psychopharmacology. Do not stop a prescribed medication without medical advice.

Practical problems

Blinding is harder than it sounds. If a drug has noticeable side effects, such as a dry mouth, participants may guess they are on the real thing, which can strengthen their expectations and inflate the apparent effect. Some trials use “active placebos” that mimic side effects without the intended therapeutic action. Similar problems arise in trials of surgery, psychotherapy, and psychedelic-assisted therapy, where a convincing placebo is difficult or impossible to design.

Placebo surgery

Sham surgery sounds extreme, but it has been informative. In a 2002 trial led by J. Bruce Moseley, patients with knee osteoarthritis who received a sham arthroscopic procedure, with skin incisions but no actual work inside the joint, reported outcomes similar to those who received the real procedure. Results like this have led to some common operations being reconsidered.

What Placebos Cannot Do

Popular accounts sometimes present the placebo effect as proof that the mind can heal anything. The evidence does not support that.

A 2011 asthma study led by Michael Wechsler makes the point neatly. Patients received, on different occasions, an albuterol inhaler (an active drug), a placebo inhaler, sham acupuncture, or no treatment. On patients’ own ratings, the placebo inhaler and sham acupuncture helped about as much as albuterol. But on an objective test of lung function, only albuterol produced a substantial improvement. Patients felt better on placebo while their airways did not.

That pattern recurs across the research:

Placebos do not shrink tumors, clear infections, mend broken bones, or reverse the progression of degenerative disease.
Their effects are mostly on symptoms people perceive, such as pain, nausea, fatigue, and mood, rather than on the underlying disease.
Effects tend to be modest and variable. They differ widely between individuals and conditions, and no reliable “placebo personality” has been identified.
Feeling better can be risky when it masks a condition that needs real treatment, as the asthma study illustrates.

This is why placebo research is no argument for unproven remedies. A treatment that “works” only as well as a placebo has, by definition, no specific effect, and it may delay care that does. Thinking critically about health claims, including our own tendency to notice evidence that confirms what we already believe, is covered in our guides to confirmation bias and critical thinking.

Related concepts

Hawthorne effect: changes in behavior that come from being observed or studied.
Demand characteristics: cues in a study that lead participants to behave as they think the researcher wants.
Experimenter expectancy: researchers’ expectations subtly influencing results, the reason for double-blinding.

Frequently Asked Questions

Is the placebo effect real?

Yes, but it is narrower than often claimed. Placebos can produce genuine changes, particularly in pain, where they can trigger the body's own opioid system. However, much of the improvement seen in placebo groups is due to other factors, such as illnesses getting better on their own and regression to the mean. The true placebo effect is clearest for subjective symptoms.

What is the difference between the placebo and nocebo effect?

The placebo effect is an improvement caused by positive expectations or learned associations with treatment. The nocebo effect is the reverse: negative expectations produce real symptoms, such as side effects reported by people taking a dummy pill in a drug trial. Both rely on the same mechanisms of expectation and conditioning.

Can a placebo work if you know it is a placebo?

Some studies suggest it can. In open-label placebo trials, such as a 2010 study of irritable bowel syndrome, patients told they were taking inert pills still reported more improvement than a no-treatment group. The studies are mostly small and measure self-reported symptoms, so the findings are promising but not definitive.

Why do drug trials use placebos?

Because people improve for many reasons besides a drug's chemistry, including expectations, extra attention, and natural recovery. Comparing a drug against an identical-looking placebo, with neither participants nor researchers knowing who received which, isolates the drug's specific effect.

What can't placebos do?

Placebos do not shrink tumors, cure infections, heal fractures, or reverse disease progression. Their effects are mainly on how symptoms feel rather than on the underlying condition. In a 2011 asthma study, a placebo inhaler made patients feel better but did not improve their lung function, while the real drug did.

How does the placebo effect work in the brain?

Expectation and conditioning change how the brain processes signals from the body. In pain, placebo relief can be blocked by naloxone, showing that it involves the body's own opioids. In Parkinson's disease, expecting medication can trigger dopamine release. Imaging studies show reduced activity in pain-processing regions after placebo treatment.

Placebo Effect: How It Works, the Nocebo Effect, and What Placebos Cannot Do