What the Research Shows

Pain is one of the best-studied domains in placebo research, especially for experimental and symptom outcomes. That is partly because pain is not a single signal traveling unchanged from an injured body part to the brain. It is an experience assembled from incoming sensory information, attention, memory, emotion, and prediction. Context and care can therefore alter how strongly pain is felt—even when they do not repair the tissue generating it.

A placebo effect is not imaginary pain relief, nor evidence that an injury or disease was never present. It is a change in the nervous system’s processing of pain; the underlying cause may remain and may still require diagnosis and treatment.

The classic finding: expectation changes analgesia

A typical placebo-analgesia experiment gives participants a painful stimulus—heat, pressure, or an experimental procedure—and then offers what they are told is a pain-relieving treatment. In reality, the treatment may be an inert cream, saline injection, or inactive pill. When participants believe relief is likely, many report less pain than they did under a comparison condition.

The effect is not universal or equally large for everyone. Expectations influence pain, but they do not grant unlimited control over it. Across many studies, however, believing that an intervention will help can reduce reported pain and sometimes alter physiological and neurological measures associated with pain processing.

Early placebo research often treated this result too casually. In 1955, Henry Beecher’s influential paper suggested that roughly 35 percent of patients improved because of placebo treatment. That estimate became famous, but later analyses showed that it confused placebo effects with several other possibilities: the natural course of illness, fluctuating symptoms, regression to the mean, reporting changes, and the effects of being observed or treated. A patient who feels better after receiving a placebo has not necessarily improved because of the placebo.

More rigorous studies compare a placebo group with a no-treatment or natural-history group. These designs generally find that placebo effects are most dependable for subjective symptoms—especially pain, nausea, fatigue, anxiety, and some aspects of depression—rather than for directly correcting an infection, healing a fracture, or shrinking a tumor. Pain is therefore an especially useful field of study: it is biologically grounded, but its intensity is also shaped by meaning and expectation.

What the brain is doing

Placebo analgesia can recruit the body’s own pain-regulating systems. Expectation of relief may engage opioid pathways, endocannabinoid signaling, dopamine-related motivation and reward circuits, and descending networks that regulate pain signals in the spinal cord and brain. One important line of evidence comes from studies using naloxone, a drug that blocks opioid receptors. In some experiments, naloxone reduces or abolishes placebo analgesia, suggesting that endogenous opioids contribute to the effect. It does not abolish every placebo response, however. Placebo analgesia is not one mechanism but a family of overlapping mechanisms.

Brain-imaging research also supports this layered account. Expectations can influence activity in regions involved in evaluating threat, attention, emotion, and the sensory dimensions of pain. The brain is not simply deciding that an injury does not exist; it is adjusting the expected significance and intensity of incoming signals.

This helps explain why a placebo can reduce suffering without removing its source. Suppose a person has arthritis in a knee. A reassuring clinical encounter and a credible treatment ritual may lower the pain experienced during movement. The joint remains arthritic; the intervention has not regenerated cartilage. But the person may move with less distress while the medical problem is still assessed and managed.

Expectation and conditioning work together

Expectation is only part of the story. Placebo analgesia is also shaped by learning. If a person repeatedly receives an effective painkiller before a painful procedure, associated sights, words, and sensations—the clinician’s preparation, the medication cup, or the taste of a dissolving tablet—can acquire therapeutic significance. Later, a similar ritual may trigger some of the body’s learned pain-regulating response, even when the active drug is reduced or absent.

Researchers test this through conditioning procedures. Participants may first receive an analgesic alongside a particular treatment cue. On a later occasion, the same cue is paired with an inactive substance. Pain relief can persist, although it usually weakens when the expected benefit is repeatedly contradicted. Familiar routines can matter, but their effects depend on a person’s history and beliefs.

Expectation and conditioning can reinforce each other. A clinician’s confident explanation may create a conscious expectation of benefit, while previous successful treatment may provide a learned one. Conversely, a painful side effect or frightening medical encounter can amplify later symptoms. The same architecture that permits relief also permits nocebo effects.

Hidden treatment versus openly delivered treatment

One of the most revealing research methods compares open and hidden administration of medication. In an open condition, a patient knows when an analgesic is being given and may receive an explanation that it should help. In a hidden condition, the same medication is delivered without the patient knowing when it starts—for example, through an infusion pump. The drug is pharmacologically active in both conditions, but expectation differs.

Across several studies, openly administered analgesics have produced greater reported relief than the identical medication given covertly. In postoperative pain research, for instance, patients may request less additional medication when they know an analgesic has been administered. Hidden medication can still work; it simply lacks the added contribution of positive expectation and therapeutic context. Context does not replace pharmacology, but it can modify pharmacology’s experienced effect.

The reverse is also possible. If patients are warned that a medication may cause severe pain, nausea, or other problems, those expectations can increase symptoms or reduce perceived benefit. What the patient believes about an intervention becomes part of the intervention’s total experience.

The clinician’s words are part of the treatment environment

Communication is not decorative packaging around medical care. It can influence analgesia. Studies comparing positive, neutral, or negative framing have found that statements such as “This treatment should reduce your pain” can improve expected benefit, while emphasizing possible worsening can increase discomfort. Ethical communication is neither exaggerated optimism nor a catalogue of frightening possibilities delivered without proportion.

A credible clinician can say: “This treatment helps many people, though not everyone, and we will monitor how you respond.” Such language offers hope without promising a cure. Warmth, attention, and a clear plan may reduce threat and uncertainty; careful explanation can help a patient distinguish expected discomfort from warning signs that need evaluation.

Pain relief is shaped not only by what is administered, but also by what the person expects, remembers, and understands about what is being administered.

That principle should never become blame. Severe pain is not evidence of insufficient optimism, and persistent pain is not a moral failure. Placebo analgesia offers an additional route to relief alongside appropriate diagnosis and treatment—not an instruction to think pain away, and never a reason to postpone medical advice.

Pain is the clearest example, but it is not the only symptom shaped by expectation and therapeutic context. A placebo effect may alter a genuine experience—nausea, breathlessness, fatigue, bowel discomfort, or tremor—without correcting the disease that produced it. In some conditions that change can be useful; in others, apparent improvement may reflect fluctuating symptoms, selective reporting, or the temporary comfort of being cared for.

Nausea and gastrointestinal symptoms

Nausea is particularly responsive to expectation because it is both a bodily sensation and a prediction. If a person anticipates queasiness after chemotherapy, anesthesia, pregnancy, or a medication, ordinary stomach sensations may be interpreted as signs of impending vomiting. Confidence in a treatment can dampen that alarm. This does not make nausea imaginary; it shows that gut signals are interpreted within a nervous system that forecasts what will happen next.

Research in nausea is complicated by the difficulty of measuring it. Vomiting can be counted, but nausea is largely subjective. A patient’s report is not lesser evidence—it is the symptom itself—though it makes comparison more vulnerable to expectation and communication effects. Trials of anti-nausea treatments often include substantial improvement in placebo groups, especially with careful monitoring and reassurance. Active medicines can still outperform placebo, particularly for preventing vomiting or severe nausea in high-risk settings. Supportive context should therefore accompany antiemetic treatment, not replace it.

Irritable bowel syndrome (IBS) provides another revealing example. IBS involves abdominal discomfort, altered bowel habits, bloating, and a complex interaction between the gut and nervous system. In trials of sham acupuncture, placebo pills, and other interventions, participants often improve even when the specific physical treatment is inert. Improvements have sometimes been larger when care was delivered through a warm, attentive therapeutic relationship rather than minimal contact.

The evidence does not show that IBS is “all in the mind,” nor that a ritual cures its underlying biology. Diet, motility, inflammation, visceral sensitivity, stress, sleep, and the gut microbiome may all contribute. Placebo-supportive care may reduce symptoms and disability, but it should be integrated with appropriate evaluation and evidence-based options such as dietary strategies, psychological therapies, or medication when indicated.

Fatigue, sleep, and anxiety

Fatigue includes motivation, effort, concentration, mood, sleep quality, and the meaning a person assigns to exertion, not simply energy in the muscles. Expectation can change how effort feels and how much activity seems possible. In trials of cancer-related fatigue, chronic fatigue, and other persistent conditions, placebo groups sometimes report meaningful improvement. But fatigue can also arise from anemia, infection, endocrine disease, medication effects, sleep disorders, depression, or neurological illness. A reassuring ritual must not become an excuse to overlook those possibilities.

Sleep research shows a related phenomenon: people may report better or worse sleep than objective measures suggest. This is sometimes called sleep-state misperception, though the phrase can sound dismissive. The person’s experience of poor sleep is real, even when laboratory measures show more sleep than expected. Placebo interventions can improve perceived sleep quality, confidence about falling asleep, and next-day distress. Objective measures such as sleep duration or brain activity may change less consistently. A ritual may help someone feel more rested without greatly altering sleep architecture.

Anxiety is strongly linked to anticipation of threat. In clinical trials, people assigned to placebo sometimes experience reduced anxiety, especially when they expect relief and receive regular, empathic contact. Yet anxiety disorders can be persistent and impairing, and placebo responses are generally smaller and less durable than responses to well-supported psychological treatments or medication for many patients. Encouragement can support treatment; it should not substitute for treatment.

Depression presents an even more delicate case. Antidepressant trials commonly find substantial improvement in placebo groups. Expectation, frequent clinical contact, natural symptom fluctuation, and entering treatment can all contribute. Some analyses suggest that the difference between antidepressants and placebo is modest on average for less severe depression and larger for more severe symptoms, although interpretation depends on trial design and the individual patient. This does not mean antidepressants “work only because people believe,” nor that depression can be overcome by positive thinking. Depression can involve suicide risk; professional assessment and appropriate care remain essential.

Parkinson’s disease and the body’s performance

Parkinson’s disease demonstrates that expectation can influence observable performance, not merely a questionnaire score. In studies using deceptive placebo administration, some people with Parkinson’s have shown temporary improvement in motor symptoms such as movement speed or rigidity. Brain-imaging research has linked this response to release of dopamine in parts of the striatum, the circuitry affected by Parkinson’s disease. Expecting an effective treatment can therefore engage a biological system central to movement.

The effect has limits. Placebo responses are variable, may be stronger in newly treated or less advanced disease, and often fade with repetition or disappointment. They do not restore lost neurons or halt disease progression. A person may move more easily for a time yet still require medication, rehabilitation, or other neurological care.

Similar patterns appear in other disorders. Asthma illustrates the need for caution: an inert inhaler may make symptoms feel better without necessarily improving lung function to the same extent; this separates felt breathing from airway narrowing and does not justify withholding a prescribed bronchodilator or controller treatment. In some trials of functional neurological symptoms, attention, expectation, and therapeutic framing may influence symptoms, but careful diagnosis is vital because similar complaints can have many causes. In depression, insomnia, IBS, and chronic fatigue, subjective outcomes are not “just reporting,” yet they are especially exposed to context and expectation.

What these findings do—and do not—tell us

Across these examples, three conclusions are safer than a sweeping claim that placebo “heals everything.” First, placebo effects may be more noticeable for subjective symptoms than for correcting underlying disease processes or objective pathology. Second, improvement in a symptom can be genuine without indicating that the underlying disease has been removed. Third, objective outcomes matter. Researchers therefore distinguish self-reported relief from measures such as vomiting frequency, lung function, inflammatory markers, motor performance, hospitalization, or survival.

The distinction is not a contest between “real” and “imagined.” A reduction in nausea or anxiety is real and worth pursuing, but it should not be inflated into recovery from disease. Placebo-supportive care is most responsible when it adds credible hope, attentive communication, and a calming ritual to appropriate diagnosis and treatment.

The practical lesson is modest: improve the person’s experience of illness while continuing to investigate and treat its causes. A clinician who says, “This treatment has helped many people, and we will monitor how you respond,” may strengthen helpful expectation without making a promise. A loved one who offers calm company, a consistent bedtime routine, or confidence in a medical plan may reduce distress without denying uncertainty. Symptom relief and medical evaluation belong together.

Must a person believe a treatment is pharmacologically active for placebo-supportive effects to occur? Research on open-label placebos—placebos given with an explicit explanation that they contain no active drug—offers a surprising, though carefully limited, answer.

When the placebo is labeled honestly

In an open-label study, participants are told directly: this pill contains no active medication, yet research suggests that taking it in a supportive treatment ritual may still reduce certain symptoms. The aim is to test whether benefit can occur without deception.

One influential trial, led by Ted Kaptchuk and colleagues in 2010, studied people with irritable bowel syndrome (IBS). Participants were told that they were receiving placebo pills, described as inert substances with no active ingredients. They were also told that placebos might work through mind–body processes and that taking the pills consistently mattered. Compared with a no-treatment control group, the open-label placebo group reported greater improvement in overall IBS symptoms, symptom severity, and relief.

The result was intriguing precisely because participants were not tricked. It was not evidence that a sugar pill repaired intestinal inflammation, eliminated an infection, or corrected the underlying causes of IBS. It suggested that a transparent therapeutic ritual—taking a pill, receiving an explanation, and participating in regular clinical follow-up—could influence how symptoms were experienced and regulated.

Similar findings have appeared in other conditions, though not always with the same strength. A small 2016 randomized trial of people with chronic low-back pain found that participants who were openly given placebo pills reported greater reductions in pain and disability than those receiving usual care alone.^[3] In migraine research, a 2014 study by Kam-Hansen and colleagues compared several ways of labeling identical migraine tablets, including an openly labeled placebo. The placebo label produced less benefit than the active-drug label, but it did better than receiving no treatment at all. The study also showed how strongly the meaning attached to a treatment can alter its effect: an active medicine labeled as placebo helped less than the same medicine labeled accurately.

Open-label placebo studies have also examined fatigue, chronic pain, cancer-related symptoms, and other persistent complaints. Some report improvements; others find little or no advantage over control conditions. The pattern is therefore not “placebos work for everything.” It is closer to this: in some symptom-focused conditions, especially where expectation, attention, stress, and nervous-system regulation influence the experience, an honest therapeutic ritual may provide a modest additional benefit.

An open-label placebo is not a magic pill without medicine. It is a test of whether an honest, structured therapeutic context can help the body and brain regulate symptoms.

Why might an honest ritual help?

The apparent contradiction—knowing a pill is inert while still benefiting from it—becomes less mysterious when belief in a drug’s chemical action is separated from the broader effects of a treatment ritual.

First, the ritual can create an expectation of care and possible improvement. A person may not expect the pill to alter intestinal bacteria or repair tissue, but may reasonably expect that taking part in a carefully explained process could make symptoms more manageable. That expectation can shift attention away from constant threat-monitoring and influence systems involved in pain, nausea, fatigue, and gut sensation.

Second, rituals draw on learning. If taking medication in the past was repeatedly followed by relief, the setting, timing, packaging, and act of swallowing a pill may acquire reassuring associations. An openly labeled placebo does not erase those learned cues. The pill may function as a reminder of previous treatment, rest, professional attention, or the expectation that symptoms can settle.

Third, regular use can organize behavior. Taking a tablet twice daily may prompt someone to pause, breathe, notice symptoms without panic, follow a routine, and reconnect with a treatment plan. Stress, attention, sleep, and autonomic arousal can affect symptom intensity, but these changes are not proof that the placebo has treated a disease process.

The surrounding relationship matters too. In the IBS trial, participants received an explanation, a treatment rationale, and contact with researchers who took their symptoms seriously. Some benefit may therefore have come from that credible, respectful encounter rather than from the inert pills alone. An open-label placebo should not be imagined as a standalone object; its effects, if present, belong to a package of meaning, routine, attention, and expectation.

What these studies do—and do not—establish

The evidence is promising but still preliminary. Many open-label placebo trials are small, short, or difficult to blind. Participants may know they are receiving an unusual intervention and report improvement partly because they want to cooperate with researchers. A no-treatment control can show that outcomes differ from doing nothing, but it cannot identify which ingredient produced the difference. The benefit might arise from the pills, the explanation, clinical attention, the daily routine, or natural fluctuation in symptoms.

Recruitment creates another uncertainty. People willing to enroll in a study of open-label placebos may already be curious about mind–body medicine or receptive to the possibility of benefit. Their response may not predict what will happen in a broader population. Studies also vary in wording, dosing frequency, condition, and outcome measures. A reduction in self-reported discomfort is meaningful, but it is different from an improvement in an objective marker of disease.

Duration remains an open question. Some trials last only a few weeks. We know less about whether benefits persist for months or years, whether the ritual loses power through familiarity, or whether it can be refreshed by changing the routine. Better evidence is also needed about who benefits, the average effect size, and whether open-label placebos add anything when people already receive high-quality education, reassurance, behavioral treatment, or medication.

There is a further ethical boundary. Honest labeling does not make every placebo use appropriate. A placebo should not replace diagnosis, antibiotics for a bacterial infection, insulin, cancer treatment, or evaluation of new and alarming symptoms. Nor should a person be pressured to take one or told that failure to improve reflects inadequate optimism. The evidence concerns selected symptoms and carefully designed contexts—not a general license to treat illness with suggestion.

For now, open-label research supports a balanced conclusion. Deception is not required for every placebo-supportive effect, and a transparent ritual may help some people with persistent symptoms. The most plausible benefit comes from the interaction of expectation, conditioning, attention, and therapeutic context. That possibility is worth exploring, provided it remains an addition to responsible medical care rather than a substitute for it.

The question worth carrying forward is whether an honest ritual adds meaningful relief beyond the time and attention of an ordinary, supportive consultation.

A promising study can still support a misleading headline. In placebo research, several questions are easily blended together: Did participants improve? Did they improve more than a comparison group? Was the change caused by the placebo-related context? Was it large enough to matter in daily life? And would the result appear again in another population, with another research team?

Learning to separate those questions protects the useful insight that expectation and care can influence symptoms without turning suggestion into a cure.

Start with the comparison group

The first question is not simply whether people receiving a placebo improved. People often improve while waiting for treatment because symptoms naturally fluctuate, an illness is resolving, they have changed other aspects of their behavior, or they remember their worst moments more vividly than ordinary ones. Improvement in a placebo group, by itself, does not demonstrate a placebo effect.

The more informative comparison is often a no-treatment or usual-care group. If both groups improve equally, the study may be showing the natural course of symptoms, regression to the mean, or the effects of being observed. If the placebo group improves more than the no-treatment group, that difference is stronger evidence for a placebo effect—though expectations, contact with researchers, and other study features may still contribute.

A different comparison is an active-treatment group. If a placebo performs worse than a genuine treatment, it may still have produced a modest effect while the active treatment produced a larger one. Conversely, if placebo and active treatment appear similar, we need to know whether the trial was designed to establish equivalence or was simply too small to detect a difference. “No statistically significant difference” does not automatically mean “the treatments work equally well.”

The details matter. Was allocation randomized? Were participants and outcome assessors blinded where possible? Did the groups begin with similar symptom severity? Did one group receive more attention or more reassuring communication? A placebo is not only a pill; the surrounding encounter can differ too.

Look at the outcome, not just the label

Placebo research is most persuasive when it measures outcomes plausibly influenced by expectation and context, such as pain intensity, nausea, fatigue, itching, breathlessness, or perceived well-being. These symptoms are real, but they are also shaped by attention, interpretation, emotion, and nervous-system regulation.

That is different from showing that a placebo eliminated a tumor, cleared a bacterial infection, repaired damaged tissue, or lowered a dangerous blood pressure to a safe range. A person may feel substantially better without the underlying disease changing in the same way. Both facts can matter, but they must not be confused.

Ask whether the study measured:

• a subjective outcome, such as a rating of pain or fatigue;
• an objective outcome, such as lung function, blood glucose, wound healing, or a laboratory marker; or
• a meaningful real-world outcome, such as walking farther, sleeping better, returning to work, or reducing rescue medication.

Subjective outcomes are not “merely imaginary.” Pain ratings, for example, describe a genuine experience and are important clinical outcomes. But a large change in a symptom score does not prove that the disease process has been reversed. The strongest studies often include both symptom measures and objective or functional measures, allowing us to see what changed—and what did not.

Consider the size of the study

A study with twenty participants may reveal an intriguing pattern, but it can produce unstable results. A few unusually large improvements may shift the average. Small studies are also more likely to miss a real effect, and positive small studies are more likely to attract attention than negative ones.

Larger studies generally provide more precise estimates, but size is not everything. Look for who participated, how many completed the study, and whether many participants dropped out. If researchers analyze only those who completed treatment, the result may look better than it would in ordinary practice.

The participants matter as much as the number. A placebo intervention tested in adults with long-standing irritable bowel symptoms may not tell us what to expect in children with acute abdominal pain or someone with a newly diagnosed inflammatory disease. The closer the study population is to the situation of interest, the more cautiously—but more usefully—it can be applied.

Statistical significance is not the same as importance

Researchers often report a p value or a confidence interval. Statistical significance asks whether the observed difference would be relatively surprising if there were no true difference under the study’s assumptions. It does not tell you how large the benefit is, how long it lasts, or whether an individual person is likely to notice it.

A very large study can find a statistically significant improvement of one point on a hundred-point symptom scale—an effect that may be real but barely noticeable. A small study may fail to reach statistical significance despite a potentially meaningful effect simply because it lacks precision.

For practical decisions, ask three further questions:

1. How large was the average difference?
2. How uncertain is that estimate? Look at the confidence interval rather than only the headline number.
3. What proportion of people experienced a worthwhile improvement?

The average can conceal different individual responses. If ten people improve substantially and ninety do not, the intervention may still be valuable for some, but it is not accurate to say that it “works” uniformly.

Duration changes the meaning of a result

A benefit measured after ten minutes, one week, or six months answers a different question each time. Contextual effects may be strongest at the beginning of treatment, persist as learning develops, or fade when expectations change. Symptoms themselves may fluctuate.

Check whether the study included follow-up after the initial result. Was the improvement maintained? Did participants need repeated rituals or continued contact? Did researchers measure unwanted effects, additional medication use, or daily functioning? A temporary reduction in discomfort can be worthwhile, but it should not be described as durable recovery unless the evidence supports that claim.

Replication is a test of humility

One study is a piece of evidence, not a verdict. Confidence grows when independent researchers reproduce a finding and results remain broadly consistent across settings. Replication may reveal that an effect is smaller than first reported, applies only to certain symptoms, or depends on unusually skilled communication or carefully selected participants.

Reviews and meta-analyses can help, but their conclusions depend on the quality of the studies included. If studies differ greatly in interventions, populations, and outcomes, a single pooled number may create an illusion of precision. Publication bias also matters: studies showing an exciting effect are more likely to be published than studies finding little or no effect.

Read the claim at the same scale as the evidence

A careful conclusion might say: In some people with persistent pain, an engaging and credible treatment context was associated with a modest improvement in reported symptoms over several weeks. An overstatement might say: The placebo effect heals the body or placebos work just as well as medicine. The first respects the study’s population, outcome, size, and duration. The second turns a limited finding into a universal promise.

Watch especially for claims that placebo effects cure disease, replace effective treatment, work equally well for everyone, or depend on positive thinking alone. Placebo-related mechanisms are neither magic nor moral tests. People vary in expectations, prior experiences, symptom biology, trust, culture, and sensitivity to suggestion. Some respond strongly; some minimally; some not at all. No one should be blamed for failing to improve.

Finally, ask who funded the research and who stands to benefit from the conclusion. A financial conflict does not prove that a study is false, but it is a reason to examine methods and independent replication carefully. Also distinguish a peer-reviewed research paper from a commercial testimonial, a press release, or a dramatic social-media summary.

The most responsible reading leaves room for both possibility and limits: placebo-supportive effects can meaningfully influence symptoms and treatment experience, yet they are not a universal cure and cannot be assumed to alter every underlying disease. That disciplined middle ground is where useful hope becomes safer care.

A practical way to apply this discipline is to read every study through four questions: What is the care action? What context is shaping the experience? What result actually changed? What safety edge still applies? This keeps the interpretation anchored to the evidence. A response is not automatically an effect, and an effect on a symptom is not automatically an effect on disease. Context is part of care, not a substitute for care.