Predictability of players’ actions as a mechanism to boost cooperation

Abstract

Cooperation plays an important role in evolutionary game theory, but it is still not adequately understood which mechanisms can make cooperation faster. However, the speeds of processes are crucial in social science because which scenario will prevail in societal processes with respect to globalization, emigration, or politics in general, mainly depends on the speed of the underlying societal processes. While cooperation mechanisms conceived to boost pro-social behavior mainly rely on random selections of players yet there is inconsistency between individuals’ actual behavior and random choices due to cognitive bias. Here applying sequence of decisions, we introduce an entropy-based method in order to test the limits of predictability in selecting players for interaction in a pairwise prisoner’s dilemma. We demonstrate that sequential information embedded in real entropy is helpful to predict the action of the next player, significantly increasing the fraction of cooperators, explicitly represented by one characteristic parameter λ. The numerical results reveal that cooperation can be significantly boosted when λ is large or exponential distributed. This emphasizes that high predictability of selecting players at individual level and heterogeneous predictability of selecting players at group level are factors of boosting cooperation. Our entropy-based method provides a credible way to model actions of players and suitably combines information theory with evolutionary game theory.