COS 68-2 - Decision-making in complex environments: How to ensure sustainability when individuals operate under cognitive and information constraints?

Abstract: Numerous natural common pool resources (NCPRs) play a critical role in maintaining the well-being of societies, including fish populations, forests, and the climate. The long-term sustainability or collapse of these resources depends on how people make decisions. Human decision-making, however, represents a primary source of uncertainty in the study of the impacts of management policies. Reducing this uncertainty allows for better planning by resource managers.


In social-ecological systems with economic decision-making, expectations are a core component of human decision-making and, therefore, a key determinant of system dynamics. Studies on NCPR sustainability typically assume perfectly rational expectations, despite extensive opposing empirical evidence. This paper aims to address the gap in the literature by exploring the role of individuals’ expectations in the long-term sustainability of NCPRs. Specifically, it investigates the effect of the length of agents’ memories, the extent to which they consider outcomes in the future, and the degree to which they observe other agents’ actions. System dynamics under bounded rationality are then compared to a perfectly rational benchmark.


The research employs a game-theoretical agent-based model where agents’ payoffs depend on the state of the NCPR. This dependence generates feedback between the resource and agents’ strategic action. Agents observe other players’ actions through an information-sharing network between agents, which constrains information. They use these observations to form expectations on the future state of the NCPR, using the Heuristics Switching Model. The Heuristics Switching Model relies on extrapolative rules that imply cognitive limitations. Based on these expectations, agents choose how much to extract from the resource.





Preliminary results suggest that increasing the size of agents’ memories has no significant influence on system dynamics. Increasing the extent to which agents consider outcomes in the future leads to a lower level of extraction from the NCPR, but only if agents perceive that their actions significantly impact the state of the NCPR. Additionally, results indicate that agents with more central positions in the network – and hence with more information – make more accurate forecasts. A higher density of connections improves systemwide forecast accuracy and leads to less extraction. Finally, the study finds that boundedly rational agents, on average, extract more than perfectly rational agents.





Overall, the research highlights the crucial role of expectations in decision-making and their impact on NCPR depletion. The study provides insights into which management policies can prevent this depletion and promote the long-term sustainability of NCPRs in complex environments.