How People Learn and Think

BY SIMA DIMITRIJEV AND MARYANN KARINCH

WE KNOW THAT THE FUTURE WILL BE DIFFERENT FROM the past. To deal with this uncertainty, we often turn to experts whose knowledge can help us make the best decisions. As statistician and risk analyst Nassim Nicholas Taleb described in Fooled by Randomness, 1 even venture capital investors like to follow proven formulas to manage their investment risks. People trust math, especially when experts use it with links to the exact sciences. However, it’s unhelpful when this trust goes beyond the use of math as a calculation tool.

From Proven Science to Trial and Error?

 A striking example of people’s trust in math models is how the world reacted to the unknown coronavirus that was spreading from the Chinese city of Wuhan at the beginning of 2020. The new virus created a fear of uncertainty. To deal with this fear, people turned to scientists who offered advice and predictions. An epidemiological model from the Imperial College London, a respected public research university, played into peoples’ desire for certainty. As one of the most influential models, it had a decisive impact on the world’s initial response to the coronavirus. This model projected that—under the unmitigated scenario—more than two million people would die in the U.S. and more than half a million in the UK within two to three months. In the summary of the paper, the authors stated that “optimal mitigation policies” might reduce deaths by half.3 As shocking as these projections were, they implied an unusual behavior of the unknown virus. It should be common sense that modeling of unknown behavior cannot be “certain science,” but that didn’t alleviate people’s fear caused by the projected death estimates.

On the other hand, it was quite certain that prolonged lockdowns would lead to severe economic damage. However, we didn’t have models predicting the number of deaths due to this devastation of businesses around the world. Also, these deaths would happen in a distant future and most people didn’t fear them—they feared the near term threat. The epidemiological models provided the only “certainty” about the imminent future. So it seemed the only reasonable option was to accept and implement an unprecedented lockdown of people, that was certain to wipe out tens of millions of jobs.

In addition, mathematical models in economics are not as trusted as epidemiological models. Thinking that way, let’s take another step to have a look at the most trusted math, which formulates the laws in physics. In this context, Newton’s second law of motion is a classic example to illustrate a problem with this trust. Newton’s first law addresses forces in balance; the second law describes how objects behave when all forces are not balanced. A general prediction based on this second law, the law of acceleration, is that a relatively small force is needed to, say, accelerate any light car—regardless of specific differences in color, engine type, body shape, tire pressure, and so on. Surprisingly, this prediction includes a light car with deflated tires!

The trick here is that the force in Newton’s law is the total force, defined as the difference between the engine force and the opposing friction. This is indeed a trick because the friction force—due to either a tire burst or any other factor—is unpredictable. In practical terms, the generality of Newton’s law is limited to ideal frictionless conditions, which ignore the fact that friction is responsible for more than 50 percent of the vehicle’s fuel consumption.

It’s unwise to disparage the simplicity of this insight. We want you to see that the inability of such a general law of physics to predict a tire burst is surprisingly similar to the inability of economic models to predict events such as the extent to which businesses collapsed due to a pandemic. We want you to know that the widespread illusion about the absolute generality of the laws of physics works for two reasons: (1) their mathematical formulation resonates with people’s desire for certainty and (2) their inherent idealizations make them elegant.

The elegance of idealized laws moves science away from reality. This problem is not limited to exact sciences. Riskless business models are just as alluring as frictionless models in physics. The dream is to kick a business off and to watch it move forward forever, just as a kicked ball should roll smoothly in perpetuity according to Newton’s first law (that is, an object in motion stays in motion unless something stops it). The scientist in us likes these idealized models even if they are not useful, which the Nobel laureate physicist Richard Feynman explained in the following way: “Science is like sex: sometimes something useful comes out, but that is not the reason we are doing it.”4

Of course, it’s good when science creates a useful surprise. However, idealized laws, absolute logic, and perfect symmetries are platonic concepts that are not only different from real sex but also sometimes lead to unhelpful surprises. Although empirical knowledge about specific friction forces is not scientifically elegant, engineers must use it to design reliable cars, because no surprise on the road is elegant.