Practicing meditation: Does it really work?
By Ute Kreplin
Among the promised psychological and physical benefits of meditation are the elimination or reduction of stress, anxiety and depression, as well as bipolar disorder, eating disorders, diabetes, substance abuse, chronic pain, blood pressure, cancer, autism and schizophrenia. It is a panacea for the individual. There are also apparent interpersonal and collective effects.
Mindfulness and other Buddhist-derived meditation techniques, such as compassion and loving-kindness meditation, can perhaps increase pro-social emotions and behaviours, yielding greater social connection and altruism, tampering aggression and prejudice. “If every eight-year-old in the world is taught meditation,” the Dalai Lama purportedly said, “the world will be without violence within one generation.” The quote is widely shared online.
Such a useful activity naturally finds a variety of applications. Meditation techniques have been deployed in the military with the aim of increasing the wellbeing and work effectiveness of soldiers. Snipers are known to meditate in order to disengage emotionally from the act of killing, to steady the hand that takes a life (the element of peacefulness associated with meditation having been rather set aside). Corporations counteract stress and burnout with meditation which, on the surface, is an amiable aim, but it can also help create compliant workers. And in schools, meditation interventions aim to calm children’s minds, offering students the ability to better deal with the pressure of attaining high grades. Here, too, the goal is to reduce misbehavior and aggression in a bid to increase pro-sociality and compliance.
Psychological research often upholds this optimism about the efficacy of meditation. Indeed, studies on the pro-social effects of meditation almost always support the power of meditation – the power not only of transforming the individual but of changing society. So it appears well-grounded that meditation might improve socially advantageous behaviour. This brings with it the prospect of applications in a variety of contexts, where it might find its use in social conflicts, such as mitigation of war and terrorism.
The problem, however, is with the research that bolsters such claims.
Last year, the experimental psychologists Miguel Farias, Inti A Brazil and I conducted a systematic review and meta-analysis that examined the scientific literature behind the claim that meditation increases pro-sociality. We looked at randomized controlled studies, where meditators were compared with non-meditating individuals, and reviewed more than 20 studies that evaluated the effect of various types of meditation on pro-social feelings and behaviours such as how compassionate, empathetic or connected individuals felt.
The studies we reviewed used a variety of methodologies and interventions. For example, one used an eight-week meditation intervention called “mindfulness-based stress-reduction”. Individuals learned how to conduct mindful breathing and to practice “being in the moment”, letting go of their thoughts and feelings. Meanwhile the control group, with which the meditators were compared, engaged in a weekly group discussion about the benefits of compassion. Another study compared guided relaxation (participants listening to an audio recording about deep breathing and unwinding) with a control group that simply did nothing in a waiting room. Most studies required participants to fill in questionnaires about their experience of the meditation intervention, and their levels of compassion towards themselves and others. Some studies also included behavioral measures of compassion, in one case assessed by how willing a person was to give up a chair in a (staged) full waiting room.
Initially, the results were promising. Our meta-analysis indicated that meditation did indeed have a positive, though moderate, impact on pro-sociality. But digging deeper, the picture became more complicated. While meditation made people feel somewhat more compassionate or empathetic, it did not reduce aggression or prejudice, nor did it improve how socially connected one felt. So, the pro-social benefits are not straightforward, but they are apparently measurable. The issue is the way in which those benefits were measured.
To fully dissect the studies, we conducted a secondary comparison to see how methodological considerations would change our initial findings. This analysis looked at the use of control groups and whether the teacher of the intervention was also an author of the study, which might be an indication of bias. The results were astounding.
Let’s start with the control groups. The purpose of the control group is to isolate the effects of the intervention (in our case, meditation) and to eliminate unintentional bias. The importance of adequate control conditions was first brought to light by the discovery of the placebo effect in drug trials, which is when a treatment is effective even though no active agent (or drug) is used. To avoid this effect, each group in a drug trial receives identical treatments, except one group receives a placebo (or sugar pill) and the other gets the real drug. Neither the experimenter nor the participants know who is in which trial (this is called a double-blind design), which helps to eliminate unintentional bias. This way they can tell if it’s the active agent that is effective and not something else.
But the use of adequate controls is tricky in studies that look at behavioral change, because it is harder to create a control group (or placebo) when the treatment is not just a pill but an action. The control has to be similar to the intervention but lack some important components that differentiate it from the experimental counterpart. This is known as an active control. A passive control group simply does nothing, compared with the group that has the intervention.
Meditation did indeed improve compassion when the intervention was compared with a passive control group, that is, a group that completed only the questionnaires and surveys but did not engage in any real activity. So, participants who undertook eight weeks of loving-kindness meditation were found to have improved compassion following the intervention – compared with a passive waiting-room control group.
But have we isolated the effects of meditation or are we simply demonstrating that doing something is better than doing nothing? It might be that compassion improved simply because individuals spent eight weeks thinking about being more compassionate and felt good about having engaged in a new activity. An active control group (e.g., participants taking part in a discussion about compassion) is a more effective tool to isolate the effects of the meditation intervention because both groups have now engaged in a new activity that involves cultivating compassion. And here the results of our analysis suggest that meditation per se does not, alas, make the world a more compassionate place.
A well-designed control condition allows studies with a double-blind design. Developing an effective placebo for a meditation intervention is often said to be impossible, but it has in fact been done – and with considerable success. In the heydays of transcendental meditation research in the 1970s, Jonathan C Smith developed a 71-page manual describing the rationale and benefits of a meditation technique. He gave the manual to a research assistant, who was unaware that the technique was completely made-up – therefore, a placebo – and who then proceeded to give a lecture to participants in the control group about the merits of the technique. (When it came to the actual placebo technique, participants were instructed to sit quietly for 20 minutes twice per day in a dark room, and to think of anything they wanted.) The point is, the placebo can work in studying meditation, it’s just not often used.
Double-blind designs can help to eliminate the accidental bias of the participants through the researcher. These biases have a longstanding history in psychology and are called experimenter biases (when the experimenter inadvertently influences the participant’s behavior) and demand characteristics (when participants behave in a way that they think will please the experimenter). The importance of avoiding experimenter bias and demand characteristics was discussed as early as the 1960s. Recent work indicates that experimenter biases remain – particularly in the study of meditation.
In light of the discussion around experimenter bias and demand characteristics, it is surprising to find that, in 48 per cent of the studies we looked at, the meditation intervention was taught by one of the study’s authors, often its lead author. More importantly, little attempt was made to control for any potential bias that an enthusiastic teacher and researcher might have had on the participants. Such a bias is often not intentional but stems from subconsciously giving preferential treatment or being particularly enthusiastic to participants in the experimental group. The prevalence of authors as teachers was so great that we decided to look at it statistically in our meta-analysis. We compared studies that had used an author with studies that had used an external teacher or other form of instruction (e.g., an audio recording). We found that compassion increased only in those studies where the author was also the teacher of the intervention.
Experimenter bias often goes hand-in-hand with demand characteristics, where participants behave or respond in a way that they think is in line with the expectations of the researcher. For example, participants might respond – regardless of their true feelings – more enthusiastically on a questionnaire about compassion because the researcher herself was enthusiastic about compassion. The media buzz around meditation – which portrays it as a cure for a range of mental-health problems, the key to improved wellbeing and to changing one’s brain for the better – is also very likely to feed back to participants, who will expect to see benefits from a meditation intervention. Yet, almost none of the studies we examined controlled for expectation effects, and this methodological concern is generally absent in the meditation literature.
The prevalence of experimenter bias is only one side of the coin. Another troubling but rarely discussed bias concerns data-analysis and reporting. Interpreting statistical results and choosing what to highlight is challenging. Data do not speak for themselves: they are interpreted by academics whose minds are not blank states. Academics often tread a thin line between the duty of impartial data-analysis and their own beliefs, desires and expectations. In 2003, Ted Kaptchuk of Harvard Medical School summarized a number of interpretative biases that have become widespread in science reporting: confirmation bias, rescue bias (finding selective fault with an experiment to justify an expectation), and “time will tell” bias (holding on to an expectation discounted by data because additional data might in fact support it), among others. All were overwhelmingly present in the meditation literature we reviewed.
The most common bias we encountered was a “confirmation bias”, in which evidence that supports one’s preconceptions is favored over evidence that challenges these convictions. Confirmation bias was particularly prevalent in the form of an over-reporting of marginally significant results. When using statistical testing, a p-value of 0.05 and below typically indicates that the results are statistically significant in psychological research. But it has become common practice to report results as “trends” or as “marginally significant” if they are close to, but don’t quite reach the desired 0.05 cut-off. The problem is that there is little consensus in psychology as to what might constitute “marginal significance”, which in our review ranged from p-values of 0.06 to 0.14 – hardly even marginal. (It is debatable whether p-values are not the most accurate way to conduct science anyway, but we should stick to the rules if we are using this type of testing.)
Being liberal with statistical methods that were designed to have clear cut-offs increases the chance of finding an effect when there is none. A further problem with the use of “marginal significance’ is reporting it free from bias. For instance, in one study the authors reported a marginally significant difference (p = 0.069) in favour of the meditation intervention relative to the control group. However, on the following page, when the authors reported a different set of results that did not favour the meditation group, they claimed the exact same p-level as non-significant. When the results confirmed their hypothesis, it was “significant” – but only in that case. In fact, the majority of studies in our review discussed the marginally significant as equal to statistically significant.
Confirmation bias is difficult to overcome. Journals rely on reviewers to spot them, but because some of these biases have become standard practice (through the reporting of marginally significant effects, say) they often slip through. Reviewers and authors also face academic pressures that make these biases more likely since journals favour the reporting of positive results. But in the study of meditation there is another complication: many of the researchers, and therefore the reviewers of journal articles, are personally invested in meditation not only as practitioners and enthusiasts but also as providers of meditation programs from which their institutions or themselves financially profit. The overly positive view of meditation and the fierce fight to protect its untarnished reputation make it harder to publish negative results.
My aim is not to discredit science, but scientists do have a duty to produce an evidence base that aims to be bias-free and aware of its limitations. This is important because the inflated results for the power of meditation fuel magical beliefs about its benefits. Mindfulness websites market it as a “happy pill, with no side effects”; it is said it can bring world peace in a generation, if only children would breathe deep and live in the moment. But can we be sure that there are no unexpected outcomes that neither benefit the individual nor society? Is it possible that meditation can fuel dysfunctional environments and indeed itself create a path to mental illness?
The utilization of meditation techniques by large corporations such as Google or Nike has created growing tensions within the wider community of individuals who practice and endorse its benefits. Those of a more traditional bent argue that meditation without the ethical teachings can lead into the wrong kind of meditation (such as the sniper who steadies the killing shot, or the compliant worker who submits to an unhealthy work environment). But what if meditation doesn’t work for you? Or worse, what if it makes you feel depressed, anxious or psychotic? The evidence for such symptoms is predictably scarce in recent literature, but reports from the 1960s and ’70s warn of the dark side of transcendental meditation. There is a danger that those few cases that receive psychiatric attention are discounted by psychologists as having had a predisposition to mental illness.
In The Buddha Pill (2015), Miguel Farias and Catherine Wikholm take a critical look at the symptoms of depression, anxiety, restlessness, mania and psychosis that are triggered directly by meditation. They argue that the prevalence of adverse effects has not been assessed by the scientific community, and it is easy to think that the few anecdotal cases that might surface are due to an individual’s predisposition to mental-health problems. But a simple search on Google shows that reports of depression, anxiety and mania are not uncommon in meditation forums and blogs. For example, one Buddhist blog features a number of reports on adverse mental-health effects that are framed as “dark nights”. One blogger writes:
“I’ve had one pretty intense dark night, it lasted for nine months, included misery, despair, panic attacks, inability to concentrate (to the point that it was difficult to do simple tasks), inability to socialize (because of bad feelings, but also because I had a hard time following and understanding what others were saying, due to lack of concentration), loneliness, auditory hallucinations, mild paranoia, treating my friends and family badly, long episodes of nostalgia and regret, obsessive thoughts (usually about death), etc, etc, etc.”
In Buddhist circles, these so-called “dark nights” are part of meditation. In an ideal situation, “dark nights” are worked through with an experienced teacher under the framework of Buddhist teachings, but what about those who don’t have such a teacher or who meditate in a secular context?
The absence of reported adverse effects in the current literature might be accidental, but it is more likely that those suffering from them believe that such effects are a part of meditation, or they don’t connect them to the practice in the first place. Considering its positive image and the absence of negative reports on meditation, it is easy to think that the problem lies within. In the best-case scenario, one might simply stop meditating, but many webpages and articles often frame these negative or ambivalent feelings as a part of meditation that will go away with practice. Yet continuing to practice can result in a full-blown psychotic episode (at worst), or have more subtle adverse effects. For example, in 1976 the clinical psychologist Arnold A Lazarus reported that a ‘young man found that the benefits he had been promised from transcendental meditation simply did not emerge, and instead of questioning the veracity of the exaggerated claims, he developed a strong sense of failure, futility, and ineptitude’.
In a best-case scenario, individuals will have a psychiatrist or experienced meditation teacher to guide them, but those who practice alone can be left isolated in the claws of mental ill-health. Lazarus warned that meditation is not for everyone, and we need to consider individual differences and be aware of adverse effects in its application in a secular context. “One man’s meat is another man’s poison,” he once said about transcendental meditation. Researchers and therapists need to know both the benefits and the risks of meditation for different kinds of people – it is not unvarnished good news.
In The Buddha Pill, Farias and Wikholm write:
“[We] haven’t stopped believing in meditation’s ability to fuel change but [we are] concerned that the science of meditation is promoting a skewed view: meditation wasn’t developed so we could lead less stressful lives or improve our wellbeing. Its primary purpose was more radical – to rupture your idea of who you are; to shake to the core your sense of self so that you realize there is ‘nothing there’. But that’s not how we see meditation courses promoted in the West. Here, meditation has been revamped as a natural pill that will quiet your mind and make you happier.”
There must be a more balanced view of meditation, one that understands the limitations of meditation and its adverse effects. One day there will be a more complete picture of this potent and poorly understood practice. For now, our understanding is mostly warped.
Ute Kreplin is lecturer in psychology at Massey University in New Zealand. Her research has been published in Nature and Neuropsychologia, among others.