Advancing Strategic Decision Science Since 2014
Dr. Aris Thorne heads the Behavioral Strategy Group at the Nevada Institute of Game Theory. With a background in experimental economics and cognitive psychology, her work focuses on the systematic ways human decision-making departs from the 'fully rational, self-interested agent' model that underpins much of classical game theory. In this interview, she sheds light on the importance of these deviations and how the Institute is incorporating them into more accurate, powerful models of strategic interaction.
A: 'Irrational' is a loaded term—we prefer 'boundedly rational' or 'behaviorally influenced.' Some of the most robust phenomena include inequity aversion, where people will sacrifice their own payoff to punish someone they perceive as unfair, even in one-shot anonymous games. This upends predictions of the ultimatum game, for instance. There's also strong reciprocity—a tendency to reward cooperative behavior and punish defectors, which is crucial for the emergence of cooperation in repeated games. Overconfidence is another major factor, leading players to overestimate their probability of success in competitive auctions or negotiations. We also see a powerful effect of framing; the way a strategic choice is presented can dramatically shift decisions, even if the underlying payoffs are identical.
A: We run a battery of controlled laboratory experiments with human subjects, as well as online experiments that give us a larger, more diverse sample. A typical study might involve a series of interactive games—prisoner's dilemmas, trust games, bargaining games—where we carefully control the information, incentives, and context. We use eye-tracking and biometrics in the lab to measure attention and emotional arousal. We also conduct field experiments in collaboration with local businesses or using online platforms, where the stakes are real. The key is to isolate specific behavioral hypotheses and see if they hold under scrutiny.
A> One fascinating project looked at 'level-k' reasoning in poker tournaments we hosted locally. Classical theory assumes common knowledge of rationality—I'm rational, you're rational, I know you're rational, etc. In reality, people operate at different depths of strategic thought. We found that most amateur players operate at level-1 ('I think about what I would do') or level-2 ('I think about what you think I will do'). Very few consistently reach higher levels. This has profound implications for designing AI opponents, for regulatory policy in markets, and even for military strategy. It means you can often model complex interactions with surprisingly shallow hierarchies of belief.
A> We develop modified utility functions and solution concepts. For example, we might take a standard utility function (which is just a function of one's own monetary payoff) and augment it with terms that capture aversion to inequality or a preference for reciprocity. We then solve for the equilibrium of this 'behavioral game.' Another approach is to use agent-based modeling, where we simulate populations of agents with different behavioral rules—some purely rational, some altruistic, some vengeful—and see what macro-level outcomes emerge from their interactions. This hybrid approach, combining clean theory with messy human data, is where the most exciting advances are happening.
A> Immense. If we design an auction or a market mechanism assuming perfect rationality, it might fail spectacularly when real people use it. Understanding behavioral tendencies allows us to 'debug' these mechanisms in advance. For instance, in designing a carbon credit trading scheme, if we know participants are prone to myopic loss aversion, we can structure the trading intervals and information feedback to mitigate panic selling. In corporate strategy, understanding that rivals might be overconfident or reactive allows for more effective competitive positioning. Our work helps build bridges between how we *should* act if we were perfectly rational machines and how we *do* act as emotional, social humans, leading to better outcomes in business, law, and policy.
Dr. Thorne's work exemplifies the Institute's commitment to grounding theoretical elegance in empirical reality. The Behavioral Strategy Group is currently expanding its research into cross-cultural differences in strategic play and the developmental origins of strategic thinking in children. By uncovering the roots and patterns of human strategic behavior, the Institute aims to create models that are not only more predictive but also more humane, acknowledging the complex social and psychological fabric within which all our games are played. This research is fundamental to the next generation of economic theory and its application to real-world problems.