• info@criticascience.org

Uncertainty

How We Can Learn to Live With It


A preprint of an Israeli study appeared last August in which it was reported that “natural” immunity conferred by having been infected with the virus that causes COVID-19 is stronger than the immunity produced by COVID-19 vaccines. The implication is that if one already had COVID-19, vaccination may not be necessary.

Although the paper had yet to appear in a peer-reviewed journal as of October when this commentary was written, experts seemed to agree that the data appear valid and the conclusions reasonable. Yet the study’s findings seem to contradict previous recommendations: as of September 28, 2021 the CDC was still recommending that people who have previously been infected with the virus that causes COVID-19 should still be vaccinated. “Evidence is emerging that people get better protection by being fully vaccinated compared with having had COVID-19,” the CDC says on its website (accessed October 3, 2021).

Here we have a classic case of new data injecting uncertainty into science. An expert body recommends vaccination even after infection; a new study says that might not be necessary. Predictably, anti-vaccination activists misused this information to insist that vaccination in general is not needed—all you need to do, they asserted, is to get infected with what they think is a mild illness, develop “natural immunity” (a misleading phrase because vaccines induce a “natural” process too) and you’ll be protected far better than if you accept what they think are dangerous vaccines. Thus, uncertainty in science led to a flight into misinformation. For those already hesitant to be vaccinated, this kind of anti-vaccination trope can confirm their prior beliefs and reify the decision not to be vaccinated.

Of course, the new data do not mean that people should deliberately get infected or wait to get COVID-19 instead of getting vaccinated. COVID-19 is a very serious illness that has killed nearly 700,000 people in the U.S. and the vaccines against it are safe and effective at reducing the risk for serious illness, hospitalization, and death from COVID-19. But the new data do mean that there is now a legitimate scientific debate about whether previously infected people need to be vaccinated. It further raises an ethical issue because it would be wrong to give vaccines to people who might not need them, given that many people around the world currently lack access to them.

It is Difficult to Cope with Uncertainty

We do not handle this kind of uncertainty well. Humans in general are not proficient at correctly assessing probability, and inserting uncertainty into the mix confuses us further. In fact, our brains encode uncertainty automatically and rapidly, interfering with our ability to calculate probabilities accurately. A study that appeared in June in the Journal of Experimental Psychology: Learning Memory, and Cognition helps explain some of the dynamics of the way uncertainty unsettles us and even suggests some solutions.

Our brains automatically and rapidly encode uncertainty, making it difficult for us to come to reliable conclusions about relative probabilities of different outcomes (source: Shutterstock). It has been known for some time that when people hold a fixed belief that is challenged by something uncertain, they tend to entertain other explanations and alternative behaviors rather than sticking with what they already know. In the paper under discussion here the authors use the following example to illustrate how uncertainty can modify behavior: imagine Stacey is driving her usual route home, which always takes between 30 and 35 minutes when she is confronted by a traffic jam. It is taking longer than the usual 30 minutes to get home and she is uncertain about what is going on; uncertainty about her usual route introduced by the congestion prompts her to try an alternative route home even though she has no idea if it will be any better. Psychologists call this the “exploration” mode in which people alter their behavior in the face of uncertainty, even though they may have no evidence to suggest that doing so will provide a better outcome.

Two Kinds of Uncertainty

What the recent study showed, however, is that there are at least two different kinds of uncertainty, expected and unexpected uncertainty. Adrian R. Walker and four psychology colleagues demonstrate in the laboratory that when uncertainty is expected, we tend to stay with known explanations and options, what psychologists call the “exploitation” mode because we use what we already know. Only when uncertainty is sudden or unexpected do we enter the exploratory mode. So, returning to Stacey on her drive home, if the drive usually takes anywhere from 30 top 45 minutes depending on traffic and she is used to traffic jams, then she is less likely to explore alternative routes when she confronts one, even though the congestion makes it uncertain how long it will take for her to get home. In this case the uncertainty is expected and so she sticks with what she knows.

Scientists always expect there to be uncertainty in their work. Contrary to what people may think, lots of experiments in all fields of science do not turn out the way the scientists who design and conduct them hypothesize they will. Every experiment is, in a sense, an uncertain venture in which the outcomes are likely to be at least in part not what was predicted or expected. Because the uncertainty is expected, scientists do not immediately trash their hypotheses and go off on completely new lines of investigation when they get a result that is different from what they hypothesized but incorporate the new findings into the design of further experiments.

The public, however, is used to hearing only about experiments that work out. Media report to us on new drugs that perform in clinical trials just as the experimenters hoped they would or about causes of illnesses that are discovered by scientists who found exactly what they were looking for. The element of uncertainty is missing from these reports; we never hear about the countless failed experiments that led to the successful ones, about the myriad of confusing results that scientists have to grapple with before they finally get an experiment to work that reveals something expected and coherent.

Making uncertainty in science something we expect to happen may help protect us from vulnerability to misinformation (source: Shutterstock).

Understanding this difference between expected and unexpected uncertainty suggests some ways to make us less vulnerable to misinformation about health and science. One obvious way might be for scientists and health agencies to be much clearer when they present findings or make recommendations about what they know and what is uncertain. At the very bottom of the CDC’s recommendation to be vaccinated even if previously infected with COVID-19 are these two sentences: “Experts are still learning more about how long vaccines protect against COVID-19. CDC will keep the public informed as new evidence becomes available.” That generic statement is inadequate. What should be written—and written closer to the beginning of the recommendation--is something like “scientists are still uncertain about how strong immunity against COVID-19 is after someone is infected and until we have more evidence that it is adequate to protect from getting seriously ill or dying, we recommend that even previously infected people get vaccinated. This recommendation could change if new evidence shows that the immunity following infection is sufficiently protective.” What we are suggesting here is helping people expect uncertainty so that when it happens it won’t be a shock and won’t motivate us to seek alternative voices that completely deride vaccinations. Doing this should not be hard because as we’ve mentioned scientists are comfortable with the notion that what they do has a very large element of uncertainty. Conveying that to people ought to be straightforward for them and could prove to be a very important step in improving science and health communication.

Where we think this distinction between expected and unexpected uncertainty could really become a powerful one is in science education. When Jack was a student, science was still taught to children as a series of facts to be memorized. No uncertainty was acknowledged. With improvements in our understanding of the best ways to teach science and the introduction of “hands on” science teaching, Sara was more likely to be given “experiments” to carry out. Yet these “experiments” always had a known outcome, so a “right” answer was the expected result. Here again there is no allowance for the uncertainty that is central to all of science.

Let’s give students at all levels of science education exercises to carry out where the result is not entirely certain and in which different students might get different answers. A science curriculum that teaches children that uncertainty is part of the process, to be expected, and to be dealt with rather than the source of panic may help arm adults against anti-science activists and misinformation. Expect uncertainty is the message we need to convey when we communicate about science.

13 views0 comments