Why Do Placebos 'Work'?

Illustration for article titled Why Do Placebos Work?
Illustration: Benjamin Currie (Gizmodo)
Fake WeekFake WeekThis week, Gizmodo explores fakes, copies, simulations, and what is really “real” anyway.

We all know the placebo effect is a powerful thing—it can ease pain, alleviate depression, turn CBD into a billion-dollar industry, and more. Less widely understood is why it works—how the human brain, in tandem with various other organs, can turn a simple sugar pill into something that, in some cases and for certain ailments, works just as well as the real, patented, super-expensive, active-ingredient-containing thing. For this week’s Giz Asks, we asked the leading experts on the placebo effect to illuminate this process for us.


John Kelly

Distinguished Professor of Psychology at Endicott College and Deputy Director of the Program in Placebo Studies at Harvard Medical School

Actually, placebos do not “work.” Because they are inert, placebos cannot have any direct effects on healthcare outcomes. Instead, the improvements in patient symptoms attributed to the placebo effect are due to the psychological components associated with the treatment ritual and the context within which it takes place. For example, the positive expectations that result from a warm and empathic therapeutic relationship between patient and clinician. This is not to say that placebo effects are “all in the patient’s head” or that they are not “real.” There is abundant evidence from pharmacological and neuroimaging studies that the placebo effect is associated with genuine neurobiological responses in patients. The point is that these effects are attributable to the psychological components of the treatment context, and not to the placebo treatment itself.


Magne Arve Flaten

Professor, Biological Psychology, Norwegian University of Science and Technology, and co-editor of Placebo and Pain: From Bench to Bedside

Placebos work because people expect them to work, which initiates changes in the brain that has consequences for e.g. the experience of pain.

Firstly, the person must believe that the placebo is in effective treatment. Secondly, the person must receive that treatment (a pill, acupuncture or other treatment) against some symptom (pain e.g.). Thirdly, this will induce an expectation in the person that the symptom will be reduced. That expectation is also a process in the brain. Fourthly, that expectation in the brain leads to changes in other parts of the brain, so e.g. the impact of pain in the brain is reduced. That reduction in the symptom is the placebo response.

Luana Colloca

Associate Professor, Pain and Translational Symptom Science, University of Maryland School of Nursing, and co-editor of Placebo and Pain: From Bench to Bedside

A placebo refers to a physiologically inert substance, pill, or intervention that produces a therapeutic effect. The placebo pill in itself does not have any effect. Rather it is the action of taking a pill or undergoing a procedure that produces the beneficial effect. What the placebo leverages is the expectation of relief that is based on the verbal suggestions of benefit, memories of a past beneficial experience, and other cognitive factors. These expectations are what give rise to the placebo effect.

In addition, social learning, the process of observing someone else have a therapeutic effect from an intervention has been shown to elicit the placebo effect. Furthermore, placebos have been found to engage various body systems including opioidergic, dopaminergic, and endocannabinoid systems in producing their modulatory effects. Therefore, placebos work via the creation of expectancies and the activating of these endogenous modulatory mechanisms to produce effects that mimic that of a pharmacological treatment. Namely, expectancies of a therapeutic outcome facilitate the activation of modulatory systems controlling symptoms and recovery processes.

We have demonstrated that the violation of expectancies, such as when a discrepancy between what is expected and what is actually received causes an extinction of placebo effects. Several studies have documented the release of endogenous neuropeptides crucially involved in placebo-induced benefits in pain, Parkinson’s disease, and depression. Placebo effects can be elicited in patients suffering from chronic disorders even when patients appear to be unresponsive to pharmacological interventions. The challenge is to understand why some people respond to placebos with clinically relevant effects, some respond to placebos minimally, and some do not respond to placebos at all. For example, the use of the combination of specific genetic variants within some genes and the careful evaluation of patients’ phenotypes, integrated into clinical practice for assessing the individual’s potential to benefit from placebo effects. Gaining a deeper understanding of the biological and behavioral predictors of placebo effects related healing processes has important implications for precision and personalized medicine and can guide more effective mechanistic-driven therapeutic strategies.


Michael Thase

Professor, Psychiatry, Perelman School of Medicine (University of Pennsylvania)

Within the context of ongoing care or treatment research, the placebo effect represents all aspects of the treatment process other than the specific effect of the intervention (e.g., the passage of time, the expectation of benefit, the sense of being helped, talking to caring people, restoration of morale, etc.). In such settings, the easier to treat the illness and the better the prognosis of the patient/participant, the higher the likelihood of a placebo response.

Part of the response to the actual pill placebo is no doubt classical conditioning (popularized by the image of Pavlov’s dogs, but human have conditioned responses, too). Studies of pain medications, for example, demonstrate the positive placebo responses are mediated by activation of the endogenous opiate (aka endorphin) system.

Expectancy probably plays a larger role in placebo responses than classical conditioning and, with this in mind, the side effects and time course of placebos mimic their matching acting medications (e.g., headache pills takes minutes to hours, placebo antidepressants take weeks, etc.)

There are also negative placebo effects, sometimes called nocebos, in which expectancy or classical conditioning elicits worsening.


Fabrizio Benedetti

Professor of Physiology and Neuroscience at the University of Turn Medical School, and the author of Placebo Effects: Understanding the mechanisms in health and disease, among other books

A placebo is a treatment, be it pharmacological or not, that has no specific therapeutic properties for the condition being treated, e.g. fresh water, flour, an empty syringe. In other words, it is a fake therapy which the patient believes to be true. The placebo effect is the effect that follows the administration of a fake therapy. However, it is important to stress that what matters is not so much the fake therapy itself, namely, water or flour, but the psychosocial context around the patient and the treatment or, in other words, the whole ritual of the therapeutic act. Therefore, the placebo effect is a psychological effect deriving for the patient’s expectations, trust, hope, beliefs. The patient undergoes a therapeutic ritual, say an injection, s/he believes it is true, expects a benefit, and sometimes this is enough to produce an improvement.

Placebos, i.e. fake therapies, work because expectations, hope, trust, beliefs are capable of triggering in the patient’s brain the same mechanisms that are activated by drugs. Therefore, placebos and drugs share a common mechanism of action. It is important to point out that this holds true for some medical conditions only, such as pain, motor performance, anxiety, depression, that it, all those conditions whereby psychological factors are important in the course of illness. Conversely, placebos cannot kill the bacteria of pneumonia, nor can they stop cancer growth or prevent pregnancy.

Although it may sound weird, sometime placebos work even though patients know what they are receiving is a fake treatment. After all, this is not surprising, as we are conditioned to many rituals in our life. It is not different from watching a horror movie. You know everything is fake: the victim is an actor, the knife is made of plastic, blood is actually tomato juice. None the less, you are scared and do have physiological reactions: increase in heart beating, sweat, shivers, and the like.

In conclusion, placebos work because psychological factors are crucial in many circumstances. The doctor’s words are sometimes as important as drugs, and today neuroscience tells us that words hit the same targets of drugs. Actually, it would be better to say that it is drugs that hit the same targets of words, as words and social interaction emerged during evolution much earlier than drugs.


Charlotte Blease

PhD, Program in Placebo Studies, Harvard Medical School

There are many senses in which a placebo can “work.” So, strictly speaking, we can’t answer unless we know what sense of placebo we are talking about.

There are two nuanced ways in which the term placebo is currently used. First, for centuries doctors adopted a kind of medical vernacular when they speak of giving patients ‘placebos.’ In clinical settings docs have (and surveys show, frequently still do) prescribe pills or suggest treatments that our beguiling MDs don’t believe will actually work. Rather, they want to instill some hope in us patients (or get rid of us). For example, doctors might prescribe antibiotics for a recurrent viral infection—even though they know that the ‘ingredients’ will not target the malady. Of course, this raises many ethical questions. Notwithstanding, in this sense, a placebo works, if it gets rid of us pestering patients, calms us down, or makes us feel a bit better (which I’ll get to later).

Second, placebos in research mean something rather different. Here placebos are ‘controls’—or a kind of measuring tool—used to test for the efficacy of new healthcare interventions such as drugs. In randomized clinical trials, there are typically three groups of participants: one group of patients is randomly allocated to the new treatment; another group is given a ‘placebo’; and a third group is on a waitlist. Placebo treatments should ideally be identical to the real treatment except for those components that the medical researchers believe to be crucial to it working. For example, if you are testing a new antibiotic, ideally the placebo should look and taste just like the real thing (but it shouldn’t be an antibiotic). Participants and the clinicians dispensing the treatments should also be blind to the allocation. This is to control for all surprisingly noisy ways in which human psychology can interfere with reporting in clinical trials. For example, we respond differently when we’re being observed, prodded, and analyzed (called Hawthorne effects). We also people-please (called “responder biases”)—for example, we might report to our clinician we feel better when, in fact, we don’t (like telling a waiter the meal was great, when it was mediocre). And sometimes, when we expect we are going to get better, we actually do feel better. These are called placebo effects: the genuine, salubrious psychobiological effects that are studied by health scientists. So—taken together—for a placebo to work in a clinical trial it has to control for all of this noise. That way, researchers can zero in on the actual effects of the treatment.

Now, if we are really asking “Why do drugs that have no active ingredients help us to get better?” the honest answer is that it depends on: (a) our symptoms; and (b) the stuff that happens around the prescribing of the placebo pills. Placebo effects don’t shrink tumors but they are effective for pain, depression, fatigue, and other symptoms. To further complicate matters, placebos won’t do their magic—i.e. give rise to placebo effects—unless we believe that they’ll work, and this, in turn, may be modified by how empathetic our practitioner is, or how competent we believe the doctor dispensing the placebos to be. To confuse matters further, we may not need a ‘placebo’ or a doctor to elicit these beneficial effects. We may just need someone caring who exudes just the right amount of care and attention.


Ted Kaptchuk

Professor of Medicine and Professor of Global Health and Social Medicine at Harvard Medical School, and Director of the Harvard’s Program in Placebo Studies and the Therapeutic Encounter (PiPS)

Perceptions and placebo responses are “Bayesian inferences.” People/patients have chronic pain in all kinds of situations: they have pathophysiology but no symptoms; symptoms and not pathophysiology and everything in between. This means that perceptions/sensations (including symptoms like pain and fatigue) are not only information the body senses and registers there like a computer print-out. The bottom-up (afferent pathways) are getting signals all the time. The nerves only send up the “difference that makes a difference.” Most of the signals don’t go upwards. If they did the brain would be in overload and fry.

The decision of what this difference is ultimately... figured out using an architecture of Bayesian inference. Is this symptom real or am I getting [a] false signal? When patients enter the ritual/drama of healing/take placebos… this negotiating sometimes (not always) favors the healing encounter… The nerves allow healing to happen… That is, start to feel less pain or fatigue because the healing that is happening is allowed to register.


Beth Darnall

Associate Professor of Anesthesiology, Perioperative and Pain Medicine (Adult MSD) and, by courtesy, of Psychiatry and Behavioral Sciences (General Psychiatry & Psychology (Adult)) at the Stanford University Medical Center

People often think placebo means fake treatment. In reality, placebo effects are comprised of non-specific contextual factors, an individual’s beliefs about a treatment, and historical factors. There are multiple mechanisms that explain why placebos work.

Consider the example of pain. If we believe a treatment will reduce our pain, we are more likely to engage with the treatment. Similarly, if we like our doctor and believe they have our best interests at heart, we are more likely to engage with a treatment. We are more likely to be attuned to pain reductions if we are expecting them (confirmation bias). Having expectations for pain relief may reduce attention to pain and anxiety about pain, both of which are associated with reduced pain processing in the brain. Researchers who study placebo analgesia show that it is associated with hormonal changes, autonomic responsivity, and brain activity, thereby underscoring that the placebo exerts multiple psychobiological effects and is multifactorial.

On the flip side, if we think something will increase our pain, we can experience nocebo effects—actual increased pain. One interesting study conducted by Bingle, Tracey, and colleagues examined the effect of expectations on response to heat pain testing and intravenous opioid administration in healthy individuals. Everyone who participated in the experiment experienced heat pain (thermode placed on the hand) and an intravenous solution under three different conditions. In the first condition, while participants received heat pain, they were told they were receiving a powerful painkiller through the IV. In the second condition, while they received heat pain they were told they were receiving an inert solution through the IV. In the third condition, while they received heat pain they were told they were receiving a solution that would increase their pain. There was deception used in this study. In actuality, in all three conditions participants were receiving remifentanil, an opioid medication. Every participant underwent each of the described three experimental conditions, and the only thing that was different was the participants expectations as to what they would experience—more or less pain. The researchers found that when participants were told they were receiving a powerful painkiller, the analgesic benefit of remifentanil was doubled relative to when participants believed they were receiving the inert solution. When participants believed they were receiving a medication that would increase their pain (nocebo), the analgesic benefit of remifentanil was completely abolished. The researchers correlated the participants pain reports with neuroimaging findings showing increased or reduced pain processing based on their expectations and experience of actual pain. This study does not mean that pain is not real, or that it is all in our minds. Rather, it illustrates the power of our mind. These results are especially important to consider in the face of today’s medical climate where patients who have taken prescription opioids long term are suddenly being denied access to the medication, or are being force tapered. These circumstances will understandably cause people to experience nocebo responses, and this has been shown to contribute to reduced opioid analgesia while opioids are decreasing. This is a recipe for patient suffering. It is my wish that we use placebo/nocebo science to inform more compassionate patient care.


Do you have a question for Giz Asks? Email us at tipbox@gizmodo.com.


Share This Story

Get our newsletter



I think you made a bit of a mistake in only talking to researchers studying the psychological angle of placebo, psych and neurology doctors. While that portion of placebo is a thing its very much over emphasized in discussion of placebo and isn’t the only thing going on. It mostly effects self reported outcomes or things like pain perception or feelings of wellbeing which can not be tracked through objective measures and effects.

But the vast bulk of placebo effects, the stuff that crops up as results in studies or trackable impacts. Are the result of statistical artifacts and observation effects. Things like regression to the mean. And the way that people in a study are, by nature of being studied, paying much more attention to their health and taking more active health measures. Basically boiling down to capturing things that happen normally in all conditions, or the effects if intervening in any way shape or form. When you control for these things the placebo effect basically disappears. This end of it represents the vast bulk of what we hear about with placebo, and the narrow focus on the psychological aspects is what creates the borderline magical perception of the subject that so many people have.

Science Based Medicine has covered placebo pretty regularly, with a focus on the full subject and its impacts on research and alt med marketing.