Predicting human cooperation: sensitizing drift-diffusion model to interaction and external stimuli.

Abstract

Human cooperation arises naturally and is essential for the development of successful societies. This study aims to identify which aspects of the interaction influence societal cooperation and defection. Specifically, we investigate human cooperation within the framework of the Multiplayer Iterated Prisoner's Dilemma game, modelling the decision-making process by using the drift-diffusion model (DDM). We propose a novel Bayesian model for the evolution of the DDM parameters, based on the nature of interactions experienced with other players. This approach enables us to predict the evolution of the expected rate of cooperation within the population. We successfully validate our model using an unseen test dataset-separated from the training one-and apply it to explore three strategic scenarios known from previous research to affect cooperation: (i) manipulation of co-players, (ii) the use of rewards and punishments, and (iii) time pressure. Our model successfully explains the test dataset and behaves consistently with established findings in the literature on human behaviour in these simulated scenarios. These results support the potential of our model as a foundational tool for developing and testing strategies that foster cooperation, improving our ability to study, understand and intervene in scenarios where individual and collective interests conflict.