Problems with Placebos

Dr. Jennifer Corns—Postdoctoral Research Fellow—The Value of Suffering Project—University of Glasgow

Consider the fol­low­ing case:

Headache: You have a head­ache. I give you a yel­low sug­ar pill while telling you that it is an aspir­in that will make you feel bet­ter. You take it and your head­ache gets bet­ter.

This case is taken to be a paradig­mat­ic case of a placebo (the sug­ar pill) and a placebo effect (your head­ache gets bet­ter). It is, more par­tic­u­larly, a paradig­mat­ic case of placebo anal­gesia or placebo for pain—the most well-studied and well-accepted type of placebo effect.


We might just think it’s obvi­ous, but what is it about this case that makes the effect a placebo effect?

Consider anoth­er case:

 Disappointment: You are dis­ap­poin­ted. I give you a yel­low sug­ar cook­ie while telling you that it is a well-deserved reward that will make you feel bet­ter. You eat it and your dis­ap­point­ment gets bet­ter.

Your feel­ing bet­ter, in this case, is not a placebo effect, right? Why not? In both of these cases, sug­ar and a kind word have caused you to feel bet­ter.

Maybe it’s because we think that emo­tion­al effects are just not the sort of things that can be placebo effects. But many exper­i­ences, espe­cially emo­tion­al exper­i­ences, are unpleas­ant, and that unpleas­ant­ness appears to be what changes in the typ­ic­al pain placebo cases.

It’s the nasty awful­ness of your pain that placebos make bet­ter and placebos can make the nasty awful­ness of emo­tions bet­ter, too. As Petrovic et al (2005) put it in their influ­en­tial paper (p.963) “… the mod­u­lat­ory pro­cesses in placebo are not spe­cif­ic for placebo anal­gesia, but are rather part of the mech­an­isms involved in the reg­u­la­tion of the emo­tion pro­cesses in gen­er­al.” Along with oth­ers, they have argued that placebos for pain are just an instance of placebo for emo­tion. The unpleas­ant­ness of emo­tions, it is increas­ingly accep­ted, is as sub­ject to placebo treat­ments as the unpleas­ant­ness of pain—and through the same mech­an­isms.

Do you still think Headaches is a case of the placebo effect, but Disappointment is not. Why?

Consider a third case:

Aspirin: You have a head­ache. I give you a yel­low aspir­in while telling you that it is an aspir­in that will make you feel bet­ter. You take it and your head­ache gets bet­ter.

If any­thing is not a placebo effect, we might think, then Aspirin is it. Identifying why, how­ever, turns out to again be dif­fi­cult.

It has turned out that many of the same path­ways, neuro­chem­ic­als, and brain areas involved in paradig­mat­ic cases of anal­gesia, as in Aspirin, are like­wise involved in paradig­mat­ic cases of placebo anal­gesia, as in Headache. And these sim­il­ar­it­ies, as reviewed in Benedetti’s (2009) import­ant book, are not just lim­ited to the ulti­mate effects on symp­toms, but the mech­an­ist­ic means by which symp­tom changes come about. When you take a yel­low aspir­in and a yel­low sug­ar pill, much the same physiolo­gic­al things (rel­ev­ant to the nasty, awful­ness of pain) hap­pen.

Still, we prob­ably think that Headache is a placebo effect, but Disappointment and Aspirin are not. Why?

Placebo research­ers, along with most people, assume that there is some dif­fer­ence between placebo effects and non-placebo effects. A good char­ac­ter­iz­a­tion of the placebo effect will be broad enough to include (at least most of) the placebo effects, and nar­row enough to exclude (at least most of) the non-placebo effects. It will, for instance, include Headache as a placebo effect, but it won’t include Disappointment or Aspirin. Problems giv­ing a char­ac­ter­iz­a­tion of the placebo effect that can do this are called scope prob­lems.


The tra­di­tion­al way to char­ac­ter­ize the placebo effect was to dis­tin­guish between act­ive treat­ment inter­ven­tions and inact­ive or inert ones. A placebo inter­ven­tion was con­sidered to be an inact­ive or inert treat­ment and a placebo effect was the effect of a placebo inter­ven­tion. Using this char­ac­ter­iz­a­tion, we would say that a sug­ar pill and a kind word are inact­ive or inert. So, when a sug­ar pill and a kind word cause you to feel bet­ter, that’s a placebo effect. An aspir­in, how­ever, is an act­ive and potent drug. So, when an aspir­in causes you feel bet­ter, that’s not a placebo effect.

The prob­lem, as people even­tu­ally real­ized, is that if the sug­ar pill and kind word have made you feel bet­ter, then they are not inact­ive or inert. After all, they made you feel bet­ter!

This prob­lem with the tra­di­tion­al char­ac­ter­iz­a­tion of the placebo effect is called the placebo para­dox. As Koshi and Short (2007) put it (p.10): “If placebos are inert sub­stances, they can­not cause an effect. If an effect occurs, the placebos are not inert.” In the face of this prob­lem, most people have giv­en up the tra­di­tion­al char­ac­ter­iz­a­tion of the placebo effect.

But the same prob­lem arises for many of the oth­er attempts to char­ac­ter­ize the placebo effect. Some try think­ing of placebo effects as the effects of fake, illus­ory, or sham inter­ven­tions. But if the inter­ven­tions work, why aren’t they real? And if they aren’t real, how can they cause improve­ment?

Placebo research­ers are still look­ing for a suc­cess­ful char­ac­ter­iz­a­tion of the placebo effect. Though I lack the space to dis­cuss them all here, all char­ac­ter­iz­a­tions require that we are able to neatly sep­ar­ate out some inter­ven­tions and out­comes from oth­ers. They require, that is, that we can sep­ar­ate out the legit­im­ate treat­ment inter­ven­tions and outcomes—like tak­ing an aspir­ing and feel­ing better—from the sup­posedly ille­git­im­ate ones—like tak­ing a sug­ar pill and feel­ing bet­ter.

It seems to me that we should, instead, accept as equally legit­im­ate those inter­ven­tions that work when we test them. If an inter­ven­tion works, then we should use it and drop whatever biases we might have had against its legit­im­acy before we tested it.

We might think, how­ever, that we need to identi­fy placebos and placebo effects pre­cisely we need to test which inter­ven­tions are legit­im­ate.

Randomized con­trolled tri­als (RCTs) are the gold stand­ard for test­ing treat­ment inter­ven­tions. In an RCT, people are ran­dom­ized to at least two inter­ven­tions called treat­ment arms, and the out­comes in these arms are com­pared. RCTs can vary (among oth­er ways) in the num­ber and types of inter­ven­tions com­pared, the types of factors used to select the people, and the type of ana­lys­is used to eval­u­ate the out­comes.

Important for us is that RCTs almost always involve a “placebo arm”: some people are ran­domly assigned to receive what is sup­posed to be a placebo inter­ven­tion. The out­come for those receiv­ing the placebo inter­ven­tion is taken to be the placebo effect and it is taken to be a sign of legit­im­acy for a treat­ment that the outcome(s) in its arm(s) is great­er than that meas­ured in the placebo arm(s). Any treat­ment inter­ven­tion that is not more effect­ive than a placebo inter­ven­tion is con­sidered to be inef­fect­ive.


The issues here are com­plic­ated, but it seems to me that we do not need the notion of placebos or placebo effects for RCTs—we will simply need to stop think­ing of the “placebo arms” and the out­comes in those arm as placebos. Instead, we can and should re-conceptualize RCTs as involving the com­par­is­on of out­comes of inter­ven­tions that are sim­il­ar and dif­fer­ent in spe­cified ways. This we can do without dis­tin­guish­ing a dis­tinct placebo effect or class of such effects.

It is import­ant to com­pare the out­comes of inter­ven­tions both to under­stand how they work and to inform decisions about which inter­ven­tions are safe, effect­ive, and worth devel­op­ing. All this we can, and should, do without identi­fy­ing any­thing as a placebo or placebo effect. Nunn (2009) nicely puts the point this way (p.338):

In a post-placebo era, exper­i­ments will simply com­pare some­thing with some­thing else. That is, they will com­pare exper­i­ment­al con­di­tions: one group gets these con­di­tions and anoth­er group gets those con­di­tions. The report of every meth­od­o­lo­gic­ally accept­able exper­i­ment will describe the con­di­tions that have been com­pared… Eventually we will have stand­ard descrip­tions for com­monly com­pared things. Legislation will reflect those stand­ards. We gain trans­par­ency, hon­esty, and clar­ity.

If an inter­ven­tion is effect­ive, then we should use it. If it works, then it should be accep­ted as being as real, potent, and legit­im­ate as any oth­er inter­ven­tion. Calling some effect­ive treat­ment inter­ven­tion a placebo and a real effect a placebo effect under­mines the legit­im­acy of those treat­ments and effects.

Many inter­ven­tions appear to be lumped in as placebos, even though they are effect­ive, because we think they are all ille­git­im­ate. If it’s not a pill, a needle, or a knife, then it is med­ic­ally sus­pect. This lump­ing mat­ters since some inter­ven­tions are inef­fect­ive; some sup­posed placebos are more effect­ive, for some things, than oth­ers. These lumped inter­ven­tions should be sep­ar­ated out and more thor­oughly and trans­par­ently tested so that we can bet­ter under­stand if, when, and how they work, and so that we can make decisions about when they are safe, effect­ive, and worth devel­op­ing. Not all “placebos” are cre­ated equal. But if we keep dis­tin­guish­ing placebo treat­ments from all oth­er treat­ments, then our biases against their legit­im­acy, des­pite their effic­acy, are likely to per­sist.

Our beliefs about legit­im­acy are, I think, the biggest con­trib­ut­or to explain­ing why we think some inter­ven­tions and effects are placebos while oth­ers are not. Think again of Headache, Disappointment, and Aspirin. A cook­ie and a kind word is, we might think, a legit­im­ate inter­ven­tion for disappointment—so, not a placebo effect. An aspir­in, we might think, is a legit­im­ate inter­ven­tion for a headache—so, not a placebo effect either. A sug­ar pill and a sug­ges­tion for a head­ache is, how­ever, illegitimate—so, any effect caused by this inter­ven­tion is a placebo effect.


Sometimes, how­ever, maybe what you need is a cook­ie and a kind word, and not anoth­er aspir­in. Not just for dis­ap­point­ment, but for a head­ache, too. If we stop think­ing of cook­ies and kind­ness as placebos, then maybe we’ll start to bet­ter under­stand how they help. Sometimes they really, legit­im­ately, do.



Benedetti, F.. 2009. Placebo Effects: Understanding the Mechanisms in Health and Disease. New York: Oxford University Press.

Koshi, E.B. and Short, C.A. 2007. Placebo the­ory and its implic­a­tions for research and clin­ic­al prac­tice: a review of the recent lit­er­at­ure. Pain Practice 7(1), pp. 4–20.

Nunn, R. 2009. It’s time to put the placebo out of our misery. British Medical Journal 338, b1568.

Petrovic, P, Dietrick T, Fransson P, Andersson J, Carlsson K, and Ingvar M. 2005. Placebo in Emotional Processing—Induced Expectations of Anxiety Relief Activate a General Modulatory Network. Neuron, 46 (6), pp. 957–969.


3 thoughts on “Problems with Placebos”

  1. I’m a little bemused by this art­icle: it’s per­fectly inter­est­ing when it’s talk­ing about defin­i­tions and semantics re: dis­ap­point­ment and head­aches, but when it starts talk­ing about these things as they apply to drugs, it’s kinda non­sense:

    “A placebo inter­ven­tion was con­sidered to be an inact­ive or inert treat­ment and a placebo effect was the effect of a placebo inter­ven­tion. Using this char­ac­ter­iz­a­tion, we would say that a sug­ar pill and a kind word are inact­ive or inert. So, when a sug­ar pill and a kind word cause you to feel bet­ter, that’s a placebo effect. An aspir­in, how­ever, is an act­ive and potent drug. So, when an aspir­in causes you feel bet­ter, that’s not a placebo effect.”

    This isn’t true. When aspir­in makes you feel bet­ter, it’s well-known (by doc­tors, research­ers, my mum…) that it’s partly due to the placebo effect. By exclu­sion, the dif­fer­ence in effic­acy between aspir­in and placebo is assumed to be a func­tion of aspirin’s chem­ic­al struc­ture, rather than effects that we have estab­lished are not due to its chem­ic­al struc­ture. That’s what we refer to as an inert sub­stance, even though noth­ing is per­fectly inert. That’s why you com­pare an inter­ven­tion to placebo rather than noth­ing (although doing noth­ing is not inert either, since there’s no default set­ting for human life). In most of the best tri­als, you com­pare your inter­ven­tion to the stand­ard of care AND placebo, because a sug­ar pill or a salt injec­tion, alone, are actu­ally really mean things to give to someone who’s really ill.

    I’ve nev­er seen it stated out­right, because it would take an expert not to under­stand this intu­it­ively, but the placebo effect is a legit­im­ate effect. The point of inter­ven­tion­al medi­cine is that we’re all try­ing to do bet­ter than just giv­ing pos­it­ive vibes to patients
    (though that should be part of any treat­ment). The aim of giv­ing a sug­ar pill is to quanti­fy and exclude the part of a treat­ment’s effect­ive­ness caused by by a) the placebo effect and b) the effect of the pill’s bulk and mass, and oth­er mech­an­ic­al factors. It is not an exer­cise in den­ig­rat­ing the placebo effect, only account­ing for it.

    The argu­ment that fol­lows in the lat­ter part of the art­icle is that we should dis­card a use­ful short­hand that is invari­ably well-understood in the clin­ic­al world — and can be under­stood by any­one else who reads the very good Wikipedia art­icle on the sub­ject — to elev­ate the legit­im­acy of an already legit­im­ate and estab­lished concept.

    I also have an issue with this part, which con­flates the sub­ject of the art­icle with anoth­er, dif­fer­ent prob­lem:

    “If it’s not a pill, a needle, or a knife, then it is med­ic­ally sus­pect.”

    While there’s a lack of fund­ing for unprof­it­able inter­ven­tions, that is a dif­fer­ent prob­lem entirely and won’t be solved by award­ing cred­ib­il­ity to the sug­ar pill industry (for it is an industry, about 10% of the size of pharma, in the form of homeo­paths, tra­di­tion­al medi­cine vendors, nutri­tion­ists and so on). The prob­lem is with the
    profit motive, not the par­tic­u­lar people pur­su­ing it. You can com­pare non-pharmaceutical inter­ven­tions to each oth­er very eas­ily. In that case, the con­trol would be a sim­il­ar inter­ven­tion (not a sug­ar pill) e.g. unstruc­tured con­ver­sa­tion vs. CBT. You could also com­pare dif­fer­ent strategies of giv­ing placebos, and in fact it has
    been done: green placebos work best for affect­ive dis­orders, for example; placebos work bet­ter if the doc­tor thinks he’s really giv­ing fentanyl for pain than if he thinks it’s a placebo; inert injec­tions work bet­ter than inert pills (and have more side effects) etc ad inf. That is not to say these things are chem­ic­ally act­ive. It does mean that they can be used to design tri­als, e.g. a blue sug­ar pill should be used in a tri­al for Viagra and a white one for Lipitor.

    In the end, sug­ar pills are not legit­im­ate drugs. When a medi­cine can be proven more effect­ive than placebo, then it is not chem­ic­ally inert and no longer an accept­able placebo. The goal then should be to identi­fy how it works and see if some­thing else can work even bet­ter using sim­il­ar prin­ciples. This does­n’t mean that we should be try­ing to improve on giv­ing someone chocol­ate for dis­ap­point­ment: not everything should be par­celled off to the pharma industry. I agree as far as the homespun wis­dom at the end goes, you should­n’t always take aspir­in for a head­ache. But all the pos­it­ive think­ing in the world isn’t going to cure chron­ic hep­at­it­is.

    1. Formatting suffered because I had to email this to myself, apo­lo­gies.

  2. Thanks for your response, but I think you have mis­un­der­stood me.

    I think you are right that most people now recog­nize that things called placebos—like sug­ar pills and salt injections—aren’t inert. The his­tory of the placebo effect was to define it that way, but most people now do recog­nize that it’s a prob­lem­at­ic defin­i­tion.

    Unfortunately, people do still call sup­posed placebo inter­ven­tions “sham,” “phony”, and so on, how­ever. And there are some good reas­ons to think that most non-experts think that placebo effects aren’t as legit­im­ate or real. These ways of try­ing to res­cue or explain the notion are just as bad, but I didn’t have space in the post to talk about these many ways people have tried to save the defin­i­tion.

    That wasn’t the point of the post.

    The point here was that, while some people—though not all—now recog­nize the troubles with our old defin­i­tion and we still don’t have a good new one, we con­tin­ue to act as if all effects of inter­ven­tions can be treated as either placebo or non-placebo.
    That over­sim­pli­fies.

    The prob­lem isn’t that there is no dif­fer­ence between a sug­ar pill and an aspir­in. Of course there is! Denying that would be non­sense. There’s also a dif­fer­ence between a sug­ar pill and a salt injec­tion, for instance, and a dif­fer­ence between aspir­in and morphine.

    I guess I should have been clear­er that the point of the post wasn’t to say that we should ignore these dif­fer­ences; we shouldn’t! The chem­ic­al struc­tures of all these things mat­ter! Of course they do! I want us to bet­ter under­stand how our inter­ven­tions work. To do that, I think we shouldn’t try to lump them all into one of two cat­egor­ies: act­ive, legit­im­ate drugs and inact­ive, ille­git­im­ate placebos. This hides the import­ant dif­fer­ences that we want to learn more about.

Comments are closed.