Humans form social networks in almost all aspects of life. From families to fraternities to Facebook groups, people are drawn to network with others. New research from Associate Professor of Mathematics Timothy Killingback and Computer Science Lecturer Swami Iyer reveals another benefit to networks: They encourage us to behave more cooperatively.
In Killingback and Iyer’s new paper, published in the journal PLOS Computational Biology, the researchers investigate how individuals respond to social dilemmas when situated in different kinds of networks.
“A social dilemma is where what’s best for an individual to do is different from what would be the best kind of behavior for a whole group or society to do,” Killingback said.
Killingback and Iyer used a classic social dilemma called “The Donation Game” to test out whether individuals are more likely to take selfless actions when they are networked with other people. In this scenario, individuals could choose either to pay a small cost to give another person a large benefit, or do nothing. That seems like an easy choice—why bother to sacrifice anything of yourself? But, if both individuals pay the small cost, they both get the large benefit. The group would be better off acting cooperatively than “defectively,” or selfishly.
“These altruistic or cooperative acts occur in all sorts of systems of biology and human social systems,” said Killingback. “Without cooperative behavior, all kinds of complicated organizational behavior can’t get established.”
Using computer models, Killingback and Iyer created thousands of individuals. They were programmed to “learn” from the past actions of themselves and their neighbors whether selfish or selfless choices would earn them greater benefits. When the individuals were partnered with other individuals at random, they were more likely to make selfish decisions over time. But once the researchers restricted the individuals to interacting with a smaller group of interconnected “neighbors,” they began to behave more cooperatively.
“The reason why it works is that it stops this possibility of cooperative strategy being able to be exploited by the bad actors,” said Killingback. “Because you are in this connected network, you get this higher benefit of cooperative interaction. If an individual was using a defective strategy, they’d have to be on the outside of this cluster.”
The level of cooperative behavior depended on the type of network. Networks that have too many interconnections don’t encourage cooperative behavior, because the interactions are too random.
Additionally, networks that are made up of individuals with a similar number of connections (for instance, individuals with lots of connections matched with others with lots of connections), were more likely to result in cooperation, This is actually how most real-world networks look. Data from social networking sites indicates that people with large or small numbers of connections link up with other people who have similar numbers of connections.
“We’d like to understand why this is such a common phenomenon, this altruistic behavior in all these different social dilemmas,” said Killingback. “Individuals do normally interact in some form of network, so that provides a sensible idea for why this is such a common phenomenon.”