Let me tell you about a doctor who has been treating patients with a wide array of illnesses for many years, always using the same proven therapy, and who, after hours, even helps audiophiles to enjoy listening to music. His name is Dr. Placebo.
He went and did it. Not long ago, our uncle, an impassioned audiophile, bought himself a new cable made of pure gold, a true bargain at just 1000 zlotys (some £200). There would be nothing strange about it – well, nothing very strange, as far as our uncle is concerned anyway – had this not been a USB cable. And USB, as we know, is used to transfer digital data; ones and zeros. To be sure, generally all cables should be of good quality, but a gold cable for 1000 zlotys will transfer digital data in exactly the same way as any cheaper, decent-quality cable.
Of course, we are wasting our time trying to convince our uncle of this. He initially tries to explain that there is less noise because this cable, as he puts it, “loses fewer bits”. Given that we happen to know a bit about error correction and data transmission, we can quickly disregard this line of argument. But then he resorts to his last line of defence, a conclusive one, for which there is no counter-argument: “I have tried both cables, and I can simply hear the difference.”
The problem is that even if our audiophile uncle hears the difference, that does not automatically mean that this difference is real. We know from the work of psychologists like Daniel Kahneman that the very fact we have spent a lot of money on something makes us automatically think it is superior. Our uncle may simply be succumbing to such an illusion. He’s certainly not alone.
Only in our head, or is it real?
The placebo effect has been known about for a long time. Patients who receive a placebo drug may experience a significant improvement in their condition compared to those who have not taken any medication, and often just as strongly as those who have taken the actual treatment being tested. This is why the effects of any tested drug are always compared against a ‘placebo control’ in clinical trials (or even better: against a competitive drug). In itself, the placebo effect is not a bad thing. It is also known that doctors can significantly improve the effects of a real drug through influencing patient expectations, such as by explaining how the drug works or discussing the results of clinical trials. A few years ago, a journal published the results of a study on antidepressants involving over 200 people. Patients received either a placebo pill, St. John’s wort, or sertraline (i.e. Zoloft, a drug introduced in the 1990s). While there were no differences between the groups of patients receiving different treatments (sertraline in this study turned out to be no better than the placebo!), the factor of whether they believed they were receiving effective medication or not turned out to have a decisive influence on whether their well-being actually improved. Patients who thought they were taking placebo pills did not improve as much as others, regardless of which of the three treatments they actually received.
Interestingly, a placebo can work even when patients know they are receiving one. When researchers from Harvard gave placebo pills to patients with irritable bowel syndrome, informing them that they were taking sugar pills without any active substance, they nevertheless observed a marked improvement compared to patients who were not given any treatment. Despite being aware they were taking a placebo, it proved almost as effective as some ‘real’ drugs! This is probably why in some countries (such as Germany) doctors sometimes prescribe homeopathic remedies, even though they don’t actually work – as is known from clinical trials – any better than placebos. They assume that if a placebo can help, why not use this effect in treating patients?
Of course, prescribing expensive medicines that have the same effect as sugar pills surely seems unethical. But the problem is, the more expensive the placebo, the better it works. This has been shown in studies of drugs for Parkinson’s disease, where two placebo pills with different prices were compared against the real drug. Only the more expensive placebo had a similar effect to the real drug.
What’s more, there’s more than just delusions involved here. A famous conversation between Harry Potter and Dumbledore perfectly illustrates this situation: “Of course it is happening inside your head, Harry, but why on earth should that mean that it is not real?”* Although the basis for the placebo effect is inside our minds, the effect itself can have very measurable consequences and cause visible changes in the body. In the most famous experiment of this type, conducted in Japan in the 1960s, participants were blindfolded and informed that a doctor would place the leaf of a poisonous plant on their arm. Although in fact the leaves were harmless, everyone who took part in the experiment got a rash. Although this spectacular result has never been corroborated, many other studies have shown equally amazing results, and the placebo effect has been observed even in animals, such as rats or dogs, which we don’t usually think are susceptible to the power of suggestion.
What Clever Hans taught us
Unfortunately, the placebo effect does not just involve patients. Given that what participants believe can affect their physical condition, might the beliefs of the researcher also influence the results of the research? Indeed, it unfortunately seems so.
In the early 20th century, a German mathematics teacher, Wilhelm von Osten, became quite famous. To be more precise, it was his horse called ‘Clever Hans’ that rose to great fame. Clever Hans knew how to add, multiply, divide and subtract. He answered questions by tapping his hooves. Von Osten considered himself an outstanding teacher and claimed that his methods could even teach a horse. The animal gained such worldwide fame that a scientific commission was set up to investigate the ‘Hans’ case. Initially it was suspected that von Osten was a crook, but it soon turned out that the horse was correctly answering questions asked by others as well, not only his handler. The commission began their investigation by first checking whether the horse could answer a question if the questioner himself did not know the answer, and if he could answer correctly if he did not see the questioner. Ultimately, it turned out that the horse actually knew more about people than about mathematics. He was able to read clues from facial expressions and gestures as to when he should stop tapping his hooves.
Similar phenomena can also affect modern clinical trials. The scientific approach taken by the commission in scrutinizing Clever Hans was very innovative at that time. In studying the horse’s case, they used a method that today is one of the cornerstones of clinical trials: double-blinding, which involves concealing the desired outcome from both the research subject and the researcher. For example, the people administering medicines to patients should not know if they are actually giving a placebo or the specific drug being tested. Moreover, when it comes to analysing X-rays or microscope images, and in particular the subjective evaluation of results (such as the patient’s health), the evaluators should not know whether they are looking at the results of a participant from the control group (receiving a placebo, for example) or the tested group. Even a simple blood pressure measurement can give different results depending on what the person taking the measurement feels it should be.
Since not all researchers follow this standard, we can check if double-blinding has any effect on the results simply by comparing the clinical trials in which the principle was applied with those in which this tedious and demanding process was omitted. These types of meta-analyses have been carried out numerous times. In non-double-blind studies there is nearly always an ‘observer bias effect’, whereby the results are usually better and more pronounced. Unfortunately, while double-blinding is standard in clinical trials carried out on humans, the preceding studies carried out with animals rarely apply researcher-blinding, despite the fact that costly human trials depend on the positive outcome of animal tests. A few years ago, the third phase of clinical trials on a promising drug for treating stroke (codenamed NXY-059), carried out on over 3000 patients, ended with a huge disappointment. The drug turned out to be ineffective, although it had clearly worked when it was previously tested on animals, protecting them from the effects of stroke. It turned out that the devil was in the details. In most of the animal trials, the experimenters had not been blinded, but in double-blinded trials the effects of the drug were minimal or non-existent. If all the animal trials had been researcher-blinded, no clinical trials would have taken place at all. Of course, this is not just about huge costs. First of all, the study was carried out on patients suffering from a stroke (so it is better for them if a drug actually works). Second, the pool of money available for research is limited, so it may have been better spent on testing some other, more effective drug.
But let’s go back to music and hearing. Here, as well, blinding causes many seemingly obvious effects to disappear. Although some experienced musicians say that even when blindfolded they will recognize a Stradivarius, in reality it is difficult to confirm this. Claudia Fritz, a researcher from the French National Center for Scientific Research (CNRS), has been studying the perception of musical instruments for years. Working with Joseph Curtin (a well-known American violin maker), over the past few years she has published a series of studies in which top-class violinists tried to distinguish between a Stradivarius violin and contemporary high-end instruments. Various measures were taken so that musicians could fully appreciate the sound of each instrument, but at the same time not be able to recognize the instruments by their appearance. All studies thus far have ended with the same conclusion: while musicians can distinguish certain instruments from one other, and even agree with each other that some of them are clearly better, they are not actually able to distinguish a Stradivarius from a brand new (though very high-class) instrument. Although the research was conducted on a small number of musicians and only a handful of instruments, any significant effects would have been visible even in this situation.
Professional sound equipment is tested using the ABX method, based on double-blinding. This method could be used with my audiophile uncle. First, we play him some music using his new gold cable, telling him that he is listening to music with this cable (version A). Then we replace the gold cable with a standard one (version B), and again we tell him what’s what. If there is a difference, our uncle should be able to pick up on it. Then it’s time for X (hence the name of the method – ABX). We randomly pick one of the two cables and connect it to the device, then ask him which one it is. (Note: for this purpose, it will be necessary to have someone else do the selection; someone who cannot distinguish one cable from another, because we have to be blinded ourselves. If Hans the horse was able to read the facial expressions of humans, then our own uncle will surely be able to sense from us which cable we have connected).
We repeat the test many times. If the number of correct hits differs significantly from 50-50, the expense was objectively justified, and we will become famous for showing that the zeros transmitted by a gold cable differ from the zeros transmitted by a regular supermarket cable. Unfortunately for audiophiles, one extremely important drawback of the ABX method is that it almost always turns out that listeners are unable to distinguish between speakers or amplifiers, not to mention cables. In other words, most often there is no difference, and the listeners make random guesses. There have been many attempts at explaining this phenomenon, but so far none that is backed by facts. Most likely the results will be the same as in the majority of such tests carried out by both amateurs and professionals. Our uncle will be wrong on average 50% of the time, which means that there is probably no difference between the cables.
Some fans of top-end audio equipment, however, point out an important issue: that in the ABX test, the listener cannot enjoy the pleasure of being aware that they are listening to a beautiful, expensive device (I’m not kidding, I have seen such arguments). In other words, if the effectiveness of a drug can be affected by the patient’s expectations (especially when the drug is expensive), then how we perceive music will surely be influenced by our expectations. Expectations that have little to do with bits or frequencies. So let’s not tell our uncle about the results of our experiment. Why spoil the fun for him?
And as for medicine, the choice of examples discussed in this article is obviously biased. The vast majority of authorized medicines have won their head-to-head battles against a placebo. Even sertraline has passed this test many times, with only a single slip-up, which I referenced in this article.
*The quote comes from the book by J.K. Rowling, Harry Potter and the Deathly Hallows.
Translated from the Polish by Daniel J. Sax