Sunday, 16 September 2012

The Placebo Effect

The placebo effect is a well-recognised concept in popular culture. On the surface it seems like a relatively easy construct to understand but there are still questions in science over exactly how it works and even questions over how it is to be best defined. One extreme perspective from Rupert Sheldrake (in "The Science Delusion"1) suggests that the placebo effect is a demonstration of the mind's ability to manipulate reality and that the very concept of a placebo contradicts the materialist assumptions of science, thus demonstrating the inadequacy of science to explain our universe.

I, on the other hand, will either demonstrate why Sheldrake's understanding of the placebo effect is flawed or I will deconstruct and solve some of the deepest issues surrounding the mind-body problem and question the fundamental nature of our metaphysical reality. I assume the former is more likely.


In common parlance, the term "placebo" is often used as a way of describing or explaining the claimed benefits of some pseudoscientific treatment; for example, a skeptic, when presented with the anecdote of their mother being cured of the cold after taking some homeopathic concoction, may assert that it was "just the placebo effect". In a more formal sense, medical research uses placebo conditions to judge the comparative effectiveness of their drug, with the implication that the drug has no effect if it is no better than placebo. In both of these cases (and among others) there appears to be a few common elements, namely that:
  1. there is a benefit (perceived or actual) that needs explaining
  2. there is an illusionary or confounding influence distinct from the substance used as a placebo
  3. there are conceptual (and ethical) issues in viewing the placebo as a "real" treatment.
These elements have been combined and restructured in numerous ways over the years, to suit different agendas, applications, and interpretations, and this has left us with a broad range of formal definitions. Kirsch, for example, defined it as: “substances, given in the guise of active medication, but which in fact have no pharmacological effect on the condition being treated”2, which emphasises the inert nature of the placebo. This definition, however, is complicated by the existence of 'active placebos', where active compounds are used to mimic the side effects of the drug being tested without having any beneficial effect on the condition being treated (like using lorazepam to produce the known side effects of sleepiness and drowsiness associated with the use of the pain killers morphine and gabapentin, without having any pain killing effects of its own3). To complicate things further, other researchers (like Shapiro and Morris) have suggested that it is important to include the possibility of placebo therapies, like sham surgeries4.

Stewart-Williams and Podd looked at some of these definitions and argued that the placebo effect should be defined as: "a genuine psychological or physiological effect, in a human or another animal, which is attributable to receiving a substance or undergoing a procedure, but is not due to the inherent powers of that substance or procedure"5. The advantage of this approach is that it accounts for: a) perceived and actual improvements, b) human and animals results, c) active substances which are inert in relation to the condition being treated, and d) different forms that placebos can take (e.g. pills and surgeries).

Even with this definition in mind, we need to be aware of the confusion over conflating improvements in the placebo arm with the placebo effect itself. This distinction was raised by Ernst and Resch in their article "Concept of True and Perceived Placebo Effects"6 where they urge us to distinguish between confounding effects such as the natural course of the disease, regression towards the mean, and unidentified parallel interventions, which all can contribute to an improvement in the control condition that is entirely unrelated to the placebo intervention that has been introduced. The "true" placebo effect, they argue, is what is left over when these extraneous effects are removed, as demonstrated in this diagram:


When we eliminate the effects that Ernst and Resch warn us of, we are left with the "true" placebo effect. There are essentially two main theories proposed to explain how the placebo effect works: the expectancy theory, and classical conditioning.

Expectancy theory is the explanation which is commonly described in discussions on the placebo effect. In a nutshell, this approach suggests that a placebo creates an expectation of an effect, and that this expectation results in the actualisation of the effect; and so, in a sense, it is a formal description of what is sometimes referred to as "positive thinking". This seems fairly straight-forward but we now have to ask how expectations can produce these effects. Turner et al.7 argued, in part, that it could be a result of 'anxiety reduction', in that the expectation that you will feel better after being treated will reduce the anxiety you are feeling which is suggested to increase the functioning of your immune system. Other possibilities for the anxiety reduction explanation suggest that the expectation of a relief of symptoms could result in the patient resuming normal daily routines, which in turn helps alleviate the symptoms they were initially experiencing.

The problem with this understanding of expectancy theory, as commented on by Kirsch8, is that it fails to account for the observation of the placebo effect occurring in healthy individuals, and it fails to account for undesirable side effects which can result from taking a placebo treatment (sometimes termed the 'nocebo' effect). To accommodate these objections, Kirsch proposed the 'response expectancy theory', which is the anticipation of particular responses (automatic, subjective, and behavioral) to situational cues. What this means is that if the patient expects a treatment to produce a pain relieving effect, then they will experience a relief in pain; and similarly, if the patient expects a treatment to produce a decrease in depression, then they will experience a decrease in depression. The advantage here is that it does not rely on reducing anxiety to produce its effects, as anxiety reduction could not explain, for example, studies that show placebos increasing alertness in subjects. Despite this, however, the expectancy theory approach still struggles to explain situations where an actual physiological response occurs as a result of a placebo.

Classical conditioning, on the other hand, does a better job of explaining the processes behind the placebo effect that could account for these measurable differences. The basic idea stems directly from the work of Ivan Pavlov who was a physiologist studying the relativity uninteresting topic of salivary responses in dogs. As we would expect, to make them salivate he present them with food and he would then measure the resulting effect - however, what was not expected at the time was that otherwise irrelevant stimuli were also capable of eliciting salivation in dogs. So what was happening was that the dogs would begin to salivate before the food was presented, and Pavlov deduced that there must be some other cue that the dogs were using to predict that food was about to come (like the sound of the researcher's footsteps as they came down the hallway with the food). In order to test this, he looked at what would happen when an entirely neutral stimulus (the sound of a bell) was associated with food by ringing the bell at the same time as the food was presented. The conclusion of this experiment is what he deemed to be a "conditioned stimulus", which in simple terms just means that the sound of the bell (through repeated pairings with the food) had taken on the value of the food, and so in order to make the dogs salivate all he had to do was to ring the bell - the food was no longer necessary to produce this response.

That is all very interesting but what does it have to do with the placebo effect? The answer to that lies in another experiment that was also studying something unrelated to placebos and stumbled upon an interesting (and somewhat disturbing) effect. Ader and Cohen9 had been studying the effects of an immunosuppressant by adding the substance to their rats' sweetened water, but when an abnormally high number of their subjects started dying from otherwise minor ailments, they suspected that something else was going on. Utilising the same basic methodology as that used by Pavlov, they discovered that by pairing cyclophosphamide (an immunosuppressant) with saccharine (sweetened water), they were able to effectively reduce the immune system of the rats. So, like Pavlov's bell, the saccharine was given the ability to be able to elicit a basic biological response.


The Ader and Cohen finding was demonstrated to be consistent with the classical conditioning research in that a stronger response was generated when the saccharine was paired with a higher dose of cyclophosphamide, the effect of the immunosuppressant was a function of the schedule of reinforcement, and the effect was extinguished after repeated exposure to saccharine without it being paired with the cyclophosphamide10. What all of this demonstrates, and the most important part of this demonstration, is that the basic autonomic functions underlying our ability to deal with certain diseases can be substantially and unconsciously altered by our everyday experiences with the world. So when we take our regular pain killers for a headache, our body is not only reacting to the substance we are ingesting; we are also necessarily responding to the context and cues which are intrinsically linked to the act of swallowing a pill and the consequences that follow.

The Stewart-Williams and Podd paper goes into greater detail in trying to distinguish between the expectancy theory and classical conditioning (spoiler alert: they conclude that the correct answer is a combination of the two), but the findings that are briefly discussed above help us nail down a definition for a potentially fuzzy concept. We can be confident in our claims that the placebo effect can cover actual, as well as perceived, benefits as our main theories can cover both eventualities, and we have a framework for understanding unusual results; like the finding that the placebo stills works even when the subject knows that it is a placebo11. With all of this in mind, it is still useful to consider the complications of this situation as summarised by Colloca and Benedetti12:
It is important to stress, and there is confusion on this point, that the real placebo response is a psychobiological phenomenon that can be due to different mechanisms, which include the expectation of clinical benefit and Pavlovian conditioning. In other words, there is not one single placebo effect, there are many, so we need to look for different mechanisms in different conditions.
Arguably, most importantly of all, this understanding of the placebo effect does not force us to reconsider the fundamental nature of reality, or suppose the existence of some immaterial mind which inexplicably bends our material world to our whims and fancies. For a scientist, a theory that does not rely on magic is a good thing.


1. Sheldrake, R. (2012). The Science Delusion: Freeing the spirit of enquiry. Coronet, London.

2. Kirsch, I. (1985). The logical consequences of the common-factor definition of the term placebo. American Psychologist, 40, 237–238.

3. Gilron I., Bailey J.M., Tu D., Holden R.R., Weaver D.F., Houlden R.L. (2005). "Morphine, gabapentin, or their combination for neuropathic pain". New England Journal of Medicine, 352 (13): 1324–34.

4. Shapiro, A. K., & Morris, L. A. (1978). The placebo effect in medical and psychological therapies. In S. L. Garfield & A. E. Bergin (Eds.), Handbook of psychotherapy & behavior change (2nd ed., pp. 369–410). New York: Wiley.

5. Stewart-Williams, S. & Podd, J. (2004). The placebo effect: dissolving the expectancy versus conditioning debate. Psychological Bulletin. 130, 324–340.

6. Ernst E., Resch K.L. (1995). Concept of true and perceived placebo effects. BMJ. 311, 551-3.

7. Turner, J. A., Deyo, R. A., Loeser, J. D., Von Korff, M., & Fordyce, W. E. (1994). The importance of placebo effects in pain treatment and research. Journal of the American Medical Association, 271, 1609–1614.

8. Kirsch, I. (1997). Specifying nonspecifics: Psychological mechanisms of placebo effects. In A. Harrington (Ed.), The placebo effect: An interdisciplinary exploration (pp. 166–186). Cambridge, MA: Harvard University Press.

9. Ader, R., & Cohen, N. (1975). Behaviorally conditioned immunosuppression. Psychosomatic Medicine, 37, 333–340.

10. Ader, R. (1985). Conditioned immunopharmacological effects in animals: Implications for a conditioning model of pharmacotherapy. In L. White, B. Tursky, & G. E. Schwartz (Eds.), Placebo: Theory, research, and mechanisms (pp. 306–323). New York: Guilford Press.

11. Kaptchuk T. J., et al. (2010). Placebos without deception: a randomized controlled trial in irritable bowel syndrome. PLoS ONE 5, e15591.

12. Colloca, L., and Benedetti, F. (2005). Placebos and painkillers: is mind as real as matter? Nature Reviews Neuroscience, 6, 545–552.

No comments:

Post a Comment