Game Theory Promised To Explain Human Decisions. Did It?

As things stood, the US and the Soviet Union were on a collision course for a full-on nuclear war. Now imagine you’re the president of the United States in this boiling geopolitical moment; one wrong move could throw your adversary a deadly advantage or, worse, trigger a mutual exchange of nuclear weapons, resulting in countless casualties across the world. What do you do?

Presented like this, the Cuban Missile Crisis can be seen as a game of Chicken, whereby two rivals are set on a destructive course of action unless one of them gives way first. The winner in such games is the one who can hold their nerve the longest and not deviate from their strategy, while the loser is the one who “chickens out” and gives way first.

This has been a recurring perspective of this historical event for decades. In fact, some people see the crisis as a quintessential example of game theory – the mathematical study of strategic interactions – in action. Since the Cold War, game theory has often been held in almost reverential acclaim for its supposed power to predict people’s actions and decisions (especially within the world of economics). But is it deserving of this high status?

I expect there are few people who, even if they’ve not heard of game theory, have not been influenced by it in some indirect way. Sure, few people can say their lives have necessarily been impacted by the role this theory played in the Cold War’s strategic logic or foreign policy decision making, but what about its quieter role in influencing political ideologies during the 1980s and 1990s (neoliberalism is tightly bound up with game theory) or its subtle effect on state education models or healthcare programs, especially in the UK?

How about in the world of science? Ever heard of The Selfish Gene? This influential model was proposed by Richard Dawkins in 1976 and attempted to clarify natural selection by focusing on gene-level selection and, as you may have guessed by now, it was heavily informed by game theory.

Although game theory shouldn’t be understood as a coherent single idea, but more like a set of tools applied differently at different times and by different people, it has been a barely visible force behind many major aspects of modern history.

Game theory can be understood as a mathematical framework for analyzing strategic interactions between people who are understood to be rational decision-makers. Its promises are wide, including offering insights into how individuals, businesses, and even nations make decisions when the results depend on others’ actions.

In its purest form, game theory may be a mathematical framework that strives for objectivity, but it does so on broad assumptions about human behavior that characterize it as rational and seeking to maximize their own payoff. So, in every interaction, game theory would suggest people behave in ways that will always gain them the advantage, regardless of how it affects others. But here is the rub – people are wonderfully irrational.

A game of chicken in Cuba

As mentioned above, the Cuban Missile Crisis has consistently been a go-to case study for people developing game theory models. One of the first to do so was Thomas Schelling, an American economist and professor of foreign policy who was one of the pioneers of the theory. In his 1966 book, Arms and Influence, Schelling saw the US and the Soviet Union as being caught in the game of “Chicken” mentioned above. To win, either side had to set up its strategy and push forward with it in order to force the other to deviate from theirs, ultimately resulting in cooperation. To put it in Schelling’s terms, one player “defects” from cooperation and tries to push the other into “swerving” into cooperation.

In the context of the Cuban Missile Crisis, Schelling believed the US achieved their “defection” when President Kennedy televized the threat of a nuclear retaliation to the Soviet Union on October 22, 1962. After this threat, so the story goes, Kennedy continued to force forward by establishing a naval blockade around Cuba. This forced the Soviets to swerve rather than letting the situation escalate into a war – Premier Nikita Khrushchev publicly agreed to dismantle the missile sites on Cuba. The US were victorious. They had achieved victory by sticking to a credible threat and holding their nerve longer than their adversary.

It’s a nice story, one that demonstrates how game theory can be used to understand even the most complex and high-tensioned political problems. However, is it accurate? Over the decades, more information has appeared that undermines this interpretation of events.

Rather than a game of brinkmanship, both the US and the Soviet Union opened back-channel communications to de-escalate the situation, and neither attempted to induce the other to back down with more intimidating action. There was no taste for conflict on either side. Moreover, did anyone even win this game of Chicken in reality? Sure, the Soviets dismantled their missiles in Cuba, but that was only in exchange for the US dismantling theirs in Turkey and agreeing to avoid invading Cuba.

Far from being an illustrative example of game theory in practice, the peaceful resolution of the Cuban Missile Crisis was brought about due to individual restraint, empathy, and caution – all behaviors game theory would deem irrational in the context of a game.

Since Schelling, as the political scientist Frank C. Zagare has noted, multiple new models have been created as new historical evidence has emerged and the field of game theory has evolved. Newer explanations (such as those by Nigel Howard or Steven Brams) have been created, which use the Cuban Missile Crisis as an anchor point. However, these too have been riddled with problems and fall short of a full explanation.

Irrational humans

Since game theory became the intellectual darling of economists and some political thinkers, some researchers have attempted to bring it down from its pedestal. There have now been two Nobel Prizes in economics awarded to those who have challenged the traditional core assumptions of game theory, undermining the idea that humans are rational beings.

According to some work, those who are most likely to actually behave like the model “rational human” originally envisaged by game theorists are young kids and those who score higher for psychopathy (a chilling thought when you think how deeply traditional game theory informed world events). Of course, there are complexities to this, but the point still stands that the original model did not expect humans to act as they really do.

Since these shortcomings came to light, game theory has evolved. In particular, behavioral game theory has emerged as a way to incorporate psychological elements and learning into the package. This model accounts for those “messier” human behaviors, especially altruism, which defied the original perspectives. However, even this approach has its limits.

Behavioral game theory operates under the assumption that if we can know the types of systemic deviations from the predictions of game theory – which include individual character or aspects like cognitive overload in a given situation – then we can predict outcomes correctly. It’s a sound rescue operation in principle, but these “deviations” are still defying game theorists’ expectations. The problem here is that these deviations are inconsistent, and, more importantly, people’s beliefs do not also match their choices.

This has often meant that behavioral game theory is more descriptive than it is predictive, or to put it another way, it is better at describing how people act in certain limited strategic situations rather than offering predictions or guides for handling practical, real-world scenarios. People are too difficult to model, and there is no single rule that can be applied across the complex social and cultural worlds we inhabit.

In addition to this, much of the research into behavioral game theory has been based in lab experiments that tend to include low stakes, artificial setups (including the classic prisoner’s dilemma), and are tracked across one-shot or short-term games. But you can only simulate so much in a lab before the results become inconsistent with real-world situations.

In many cases, the decisions that matter – be they related to marriage, business, politics, or even war – involve significant stakes, higher emotions, more complex power dynamics, and more influential history. There is also the over-reliance on individual choice, which may be applicable in some situations, but often decisions are the result of collective influences (families, friends, society, governments).

Whether or not game theory can ever overcome these issues is certainly unclear right now. Some scholars, such as Gale Lucas and colleagues, have suggested it is a hollow model that shouldn’t be resurrected, at least not without significant rethinking. But while academics may have more modest views of this once influential mathematical tool, it still holds a powerful place in popular culture. So maybe the game isn’t over yet.