Game theory study suggests how memory and cooperation evolved

A key feature of human societies is our ability to cooperate. Yet in many scenarios, it would seem more rational to act selfishly and freeload off the generosity of others. So what factors support a tendency toward cooperative behavior?

A new study by Joshua Plotkin, a professor in the Department of Biology in the School of Arts & Sciences, and Alexander Stewart, a former postdoctoral researcher at Penn who is now at University College London, identifies a driving force: memory.

In their work, which uses the principles of game theory, they find that the capacity for longer memories promotes the emergence of cooperation. They also find, perhaps intuitively, that cooperative strategies are more likely to evolve in smaller groups rather than larger ones.

Though the findings could be used to help understand the evolution of even the simplest one-celled bacteria, the researchers are particularly intrigued by what they have to say about the intricacies of human social interactions.

“Our analysis sheds light on the circumstances that govern behavior in social situations,” says Plotkin. “We see that longer memories allow for a wider range of behaviors that are cooperative, to the mutual benefit of all the individuals in a given group.”

The study used a common economic experiment called the public-goods game, in which players can decide how much of a personal resource to share with a common pool, which is then divided equally among all the players in the game. In the most cooperative scenario, all players contribute selflessly and reap the greatest rewards. If one player is selfish, though, generous players can lose out. The game is played over and over; players with successful strategies are allowed to “reproduce” and pass their strategy on to the next round.

Plotkin and Stewart varied the number of players in a group as well as the memory capacity of those players. Players with longer memories could base their strategies upon the actions of their opponents in previous rounds.

The researchers showed that cooperative strategies were more likely to be sustained if group sizes were small and memories were long, likely because larger groups reduced the likelihood that a player would be punished for bad behavior, and enhanced memory capacity led to more nuanced strategies, including those that used punishment effectively.

“A stronger memory allowed players to weed out the rare defector,” Plotkin says.

Interestingly, they found that longer memories co-evolved with enhanced cooperation in a kind of feedback loop; if a group was sufficiently small, longer memories evolved and drove an increase in cooperative strategies; and that increase leads to greater benefits for the group as a whole.

“That kind of situation, where you go from a simpler system to one that is more complex, is a great example of what evolution does,” Stewart says. “It leads to more and more complexity.”

Next for Stewart and Plotkin is to test whether the relationships between memory, group size, and cooperation seen in their analyses hold up in an experimental game with human players.