“Learning to cooperate: The evolution of social rewards in repeated interactions”

Posted on

Slimane Dridi and Erol Akçay

READ THE PAPER

Biologists Slimane Dridi and Erol Akçay at the University of Pennsylvania demonstrate in a new mathematical model that evolution may shape the learning system of biological organisms to have preferences for increasing the chances of survival of others. The authors base their theoretical work on previous experimental discoveries made by other scientists, indicating that humans as well as other species find cooperation rewarding. An action is rewarding if the brain region specialized in processing rewards (the same region that is activated when we expect to eat sweet or salted foods) is activated when performing this action.

Starting from this observation, the authors explored the possibility that natural selection could favor individuals who would be rewarded by the mere action of cooperating with a social partner. Combining the theories of reinforcement learning and natural selection, they investigate the situation where individuals in a population engage in social interactions that offer the possibility to cooperate or defect. These capture many of the interactions that occur in society and nature, such as the current problem of investing efforts to protect the climate, or the work performed by soldiers in ant colonies.

The authors show that two main types of individuals are likely to take over an evolving population: individuals who the authors call “conditionally other-regarding”, meaning that they find mutual cooperation rewarding but are averse to exploitation (i.e., the situation where they cooperate but their partner defects), and selfish individuals who are only rewarded by defecting, which is the optimal action from a materialistic standpoint. Purely altruistic individuals who completely sacrifice their material gain to help others are generally not favored by natural selection. However, they can coexist with conditionally other-regarding individuals. This research helps us better understand the psychological motives behind the learning dynamics of helping behaviors in society and nature.

Invasion analysis and utilities in simulations for various benefit-to-cost ratios (b/c).
(© 2018 The University of Chicago Press)