Mathematical model sheds light on the hidden psychology behind authoritarian decision-making

A recent study suggests that authoritarian leaders are most likely to initiate major democratic reforms when they possess a specific blend of impulsive optimism and rational calculation. By using mathematical models to map how political leaders process risk and reward, the research provides evidence that the path to reform often looks risky in the short term but beneficial in the long term. The findings were published in the American Journal of Political Science.

Eugene Yu Ji, a postdoctoral researcher in cognitive science and computer science at the University of Waterloo and the Mila Quebec Artificial Intelligence Institute, conducted the study to understand the psychological mechanisms behind major political changes. Ji, who was a postdoctoral teaching fellow at the University of Chicago during the study, wanted to explain why powerful leaders sometimes voluntarily give up control. Previous research tends to focus on the broad social and economic pressures facing a government rather than the people in charge.

Ji designed this research to bridge the gap between those large-scale political dynamics and the individual cognitive processes that happen in a leader’s mind. The goal was to build a model that predicts how leaders handle rare but highly impactful decisions. These situations include yielding power, initiating massive economic shifts, or transitioning a country to a democracy.

“One motivation is the gap between psychology and other social science disciplines in addressing many real-world problems,” Ji told PsyPost. “In comparative politics, how authoritarian leaders make decisions is a central question, and fundamentally a psychological one, yet psychologists have offered few answers, largely because it is difficult to study by most traditional empirical means. This led me to develop the authoritarian decision-making modeling framework.”

To investigate these decision-making processes, Ji created a mathematical framework based on a dynamical systems approach. A dynamical system is a mathematical method used to describe how a complex state evolves over time, similar to how astronomers predict the movement of planets. In this context, the researcher used two mathematical equations to map how a leader’s thoughts and emotions change as they consider the future consequences of a potential reform.

The model focuses on two distinct types of cognitive processing that happen simultaneously. The first is metacognition, which is the slow, deliberate, and conscious process of thinking about one’s own thinking. In behavioral science, this mental effort is often called System 2 thinking.

System 2 thinking allows a person to handle complex analytical tasks and make logical judgments based on intentional effort. In the mathematical model, this represents how a leader rationally calculates the potential benefits of a reform and the uncertainty surrounding those future rewards.

The second type of processing is subcognition, which involves fast, automatic, and emotional responses to immediate situations. Scientists frequently refer to this rapid, intuitive reaction as System 1 thinking.

System 1 operates with little to no conscious effort, relying on gut feelings and past experiences to make quick assessments. In the model, this represents a leader’s impulsive and risk-averse reactions to potential threats or losses. The researcher suggests that major political decisions require both systems to interact, balancing deep logical analysis with strong emotional instincts.

Ji applied this mathematical model to six historical and contemporary cases in East and Southeast Asia to test its accuracy. The sample included China in the early 1990s and the 2010s, Taiwan in the late 1980s, and Singapore from the 2000s to the present. It also included Myanmar in the early 2010s and Vietnam in the 2010s.

The researcher assigned numerical scores to each region based on historical data. These scores reflected domestic economic performance, the strength of the ruling party, and international geopolitical pressures. These numbers were then plugged into the equations to see if the math could predict whether a major reform, a minor reform, or no reform would occur.

When evaluating Taiwan in the late 1980s, the researcher noted that the ruling party enjoyed strong economic growth but faced indirect pressure from the United States. This specific combination of strengths and pressures created a high degree of risk-driven optimistic thinking in the mathematical model. The equations predicted this would lead to a major reform, which aligns with Taiwan’s actual transition to democracy.

In contrast, the model analyzed China in the early 1990s following a period of intense political unrest. The mathematical variables indicated a scenario where leaders expected the risks of reform to simply outweigh the benefits. As a result, the equations accurately predicted that only minor economic reforms would happen, with no major democratic shifts.

Overall, the calculations successfully predicted the actual historical outcomes for five of the six cases examined. The model accurately mapped the major democratic reforms in Taiwan and Myanmar, as well as the lack of political reform in Singapore and modern China. The only exception was Vietnam, where the model predicted a major reform, but only minor to moderate political shifts actually took place.

“Autocrats can be sophisticated decision-makers, but their behavior is not beyond scientific understanding, and psychology can provide indispensable perspectives and tools to understand and analyze it,” Ji said.

The model suggests that major reforms only happen under very specific psychological conditions. A leader must anticipate what the researcher calls a “bittersweet path.” This means the leader expects uncertainties and risks to initially spike after a reform is announced, representing the “bitter” phase.

However, the leader must also believe that these uncertainties will eventually drop to levels lower than before the reform began, which represents the “sweet” phase. The researcher found that rational calculation alone is usually not enough to prompt a leader to embrace this chaotic middle period. Instead, the model provides evidence that a leader needs a certain degree of wishful thinking driven by their impulsive, emotional subcognition.

This optimistic impulse allows them to override their natural aversion to risk and push through the initial instability of a major political transition. Small shifts in the political environment can alter these psychological variables and completely change the outcome. For instance, slight changes in economic performance or party strength can make a leader feel the risks are too high, halting the reform process entirely.

“What surprised me most is how powerful and productive the cognitive modeling approach can be for addressing a major political science question,” Ji told PsyPost. “The practical significance lies in showing that important psychological processes shaping contemporary life, even those difficult to study through controlled experiments, can still be meaningfully approached. Cognitive modeling provides a way for psychology to engage with these types of questions that matter across the social sciences.”

While the mathematical model offers a new way to look at political history, the researcher notes some caveats. The study does not confirm the biological reality of the two thought systems used in the equations, but rather uses them as a practical framework to understand human behavior.

“This is a modeling study and does not affirm or deny the validity of the dual-process framework of system-1 and system-2, which are interpreted quite differently within psychology and across the social sciences,” Ji explained. “Rather, it shows that this framework can serve as useful heuristics for certain questions and can function as an interdisciplinary bridge in certain settings.”

The mathematical framework also leaves out cross-generational leadership dynamics. For example, the equations do not account for how an aging leader might behave differently than a younger successor. The model currently treats the decision-making process as an isolated event, rather than a long chain of interconnected choices involving multiple politicians. Future research is expected to incorporate more complex scenarios.

“One goal is to develop a more integrative and dynamic account of system-1 and system-2 processes, bridging behavioral economics, where the framework is often seen as transformative, and psychology, where it is sometimes viewed as overly simplistic,” Ji explained. “More broadly, this work opens a pathway to study rare but important social scientific phenomena through cognitive modeling, and I have already begun several interesting and important projects building on this framework, especially on rare but consequential decision-making under radical uncertainty across social sciences.”

“This work suggests that psychology can contribute even more powerfully to core questions in other social sciences than is often assumed. It also highlights the value of theoretical and modeling approaches in extending psychology’s reach beyond traditional empirical settings (especially given the field’s strong empiricist orientation).”

The study, “Unraveling authoritarian reform decision-making: A metacognitive–subcognitive model,” was authored by Eugene Yu Ji.