A thorough understanding of placebo effects is critical to the science of medicine, and to understanding how objectively worthless interventions can become so popular. A recent article in Nature: Scientific Reports presents a series of experiments that take a look at psychological placebos, where the intervention is not medical, but purely psychological. The results are not surprising, but they are instructive for both practitioners and researchers.
I have been writing about placebo effects since this blog began, because it is perhaps the most misunderstood and abused concept in medicine. Further, it lies at the heart of a deliberate campaign to weaken scientific standards in medicine and confuse the public in order to promote fanciful and ineffective treatments. Essentially, the public is being sold a misconception – that placebo effects are real biological effects that can be exploited to produce real healing. However, decades of actual research tell a very different story.
Placebo effects are a combination of non-specific effects of the therapeutic interaction, and an illusion of biased perception. Measured placebo effects are generally subjective and short-lived. They result from things like having an improved mood at the expectation of being treated, a positive therapeutic interaction, getting medical attention, and being compliant with treatment. These are non-specific effects, meaning that they derive from the medical interaction, not from any real physiological effect of the treatment itself. The illusory effects include subjective validation, regression to the mean, and confirmation bias. We tend to seek treatment when our symptoms are at their worst, so they are bound to improve. Many illnesses are self-limiting – they will get better on their own, no matter what we do. And we will tend to subjectively interpret our symptoms in line with our expectations and desires, perhaps out of a motivation to justify the decision to seek a specific treatment.
This is why placebo-controlled double-blind assessments are critical in clinical trials. Without them, we can’t really be sure of the results.
This is the point at which many people ask, “What’s the difference?” What does is matter if someone feels better because of a physiological effect or placebo effect, as long as they feel better? It matters a great deal. First, the person may not actually feel better, but has convinced themselves that they do in order to justify the treatment. This could be dangerous, if the illness being treated is something like asthma. This could delay presentation for medical care, increasing the risk of death.
Further, placebo effects are powerful at convincing people that a treatment works when in fact it doesn’t. This leads to numerous harms – seeking ineffective treatments at considerable cost and lost opportunity, delaying or replacing effective treatment, and leading people to unscientific and even bizarre beliefs about health and disease. This also makes them vulnerable to charlatans, whether knowing or themselves deceived.
Scientifically the harm is also great. Without properly accounting for placebo effects, science cannot progress. If researchers cannot distinguish between a real physiological treatment and non-specific treatment effects, they will not know which medical theories are more correct, and which are blind alleys. Medical research will then chase its tail, and not make any real progress. We don’t have to assume this will happen – we historically saw this happen, for literally thousands of years. Bizarre notions of health and disease flourished in every culture prior to scientific medicine, largely driven by placebo effects. Real progress was stymied. Understanding and controlling for placebo effects is largely responsible for the progress medicine has made in the last century.
Psychological placebo
All of the above also applies to psychological interventions, which can be even more challenging to tease apart genuine effects from placebo effects. There are a few reasons for this. First, many outcomes in psychological treatment are inherently subjective – how do patients feel? We can find objective markers in some cases, like lost days of work, use of symptomatic medication, or observed behavior, but there is still a huge subjective component.
But also – the non-specific therapeutic interventions are actually a large part of the point of treatment. Clients are supposed to form a positive therapeutic relationship with their therapist, and this alone can have a positive impact. It is still, however, critically important to tease apart specific from non-specific therapeutic effects, because otherwise the science of therapy will not advance, and patients and practitioners will waste a lot of time engaging in theatrical placebos without any inherent benefit.
This brings us to the current study, which is actually a good example of the kind of treatments that tend to thrive in the world of mental health because of the challenge of detecting placebo effects. The researchers did a series of three experiments on a total of 421 subjects. Subjects were exposed to green images on a computer screen, either a dot, a flux (a moving combination of green, white, and yellow colors), or a morph (pulsating green and yellow circles). In each of the studies there was a control condition in which subjects were exposed to the green video and were told only that they were a control, with an emotionally neutral technician. In the placebo condition the subjects were given a rationale for why the green video should improve their mood. In the placebo-plus condition, the technician was also instructed to be friendly, empathic, and positive. In the third study only, the green morph, there was a fourth group with a friendly technician, but without the therapeutic rationale.
So all the interventions were just placebos, and the researchers just made up the alleged rationale for why looking at a video of green color for six minutes should improve their mood. The results in all cases showed no effect on mood, alertness, or calmness for the control condition. The placebo condition showed a positive effect only for calmness. The placebo plus condition showed a positive effect for mood and calmness (but not alertness). In the third experiment, the positive interaction without the rationale showed no benefit. All results were greatest in the short term, but still present (although diminished) at 7 days. They were not measured beyond 7 days.
These results mostly confirm conventional wisdom and prior research – that when you combine an explanation for why an intervention should help, with a positive caring therapist, people feel less stressed and have a positive mood, at least in the short term. I was a little surprised that the positive interaction alone did not have any significant effect, but that was looked at in only on group and so there is limited data.
Taken at face value, this research suggests that a therapist can essentially make up whatever bull they want, and as long as they are friendly and positive, their clients will feel better in the short term. The implication of this is that any intervention will seem to work in these conditions.
Again, the problem with this is that then practitioners assume the underlying theory must be correct, when we know that this is not true. Let’s take EMDR, for example, the idea that having a patient move their eyes back and forth has a specific therapeutic benefit. Unless we control properly for the clear nonspecific effects of doing any intervention, we cannot know if the underlying theory of EMDR has any merit. It probably doesn’t, in my opinion, and the neurological explanations are contrived and unconvincing. The clinical research so far is not able to exclude non-specific effects as the sole cause.
There are also non-benign interventions that are based on pure magical thinking, or result in interventions that are inherently abusive or dangerous, like attachment therapy. Also, people seeking psychological treatments are an especially vulnerable population, and therapists are in an especially powerful position. It is unfortunately easy, for example, in this context to convince clients that they were abused when they were younger, when it never happened. This lead to an entire industry of what we now know is false memory syndrome.
The subjective nature of psychological symptoms and therefore interventions means that researchers need to be more careful in designing studies, and to carefully, even obsessively, control for nonspecific placebo effects. Otherwise what they are doing is the very definition of pseudoscience – it has all the trappings of science and may seem superficially convincing, but is not actually advancing our understanding of human psychology and treatment.
