The question of how to get people to work together has bedeviled society for millennia. Now a large-scale field experiment testing how to get more than 2,400 participants to prevent blackouts in the real world is supporting theoretical work on how to get people to cooperate that until now was largely tested only with small experiments in the lab, findings detailed in this week’s Proceedings of the National Academy of Sciences.
Mathematical biologist Martin Nowak at Harvard University and his colleagues investigated what they consider a defining aspect of human cooperation, the concept of indirect reciprocity, where one’s behavior toward a person is based on that person’s reputation for what they have done to others. (When it comes to direct reciprocity, on the other hand, your behavior toward a person is based on what that person has done to you.)
A key factor of indirect reciprocity is the ability to observe the actions of others. The chance to see what others are doing promotes cooperation in theoretical models and experiments with small groups playing games in the laboratory, but until now, there was little evidence that such observability could foster large-scale cooperation in real world settings.
To investigate whether past findings regarding observability and cooperation hold true on large scales, Nowak and his colleagues worked together with the Pacific Gas and Electric (PG&E), a power utility company that operates the majority of Northern California’s retail residential electricity market. They analyzed how best to get people to volunteer with the SmartAC program, which is designed to prevent blackouts.
The program in question is aimed to reduce excessive use of air conditioning during times when demand for electricity is high, improving the stability of the electrical grid in all of California. Volunteering with the program means the utility installs a device for free that remotely curbs air conditioners on days with unusually high demand or in cases of unexpected plant or transmission failures. Few people typically participate.
The scientists targeted residents of 15 homeowners associations in Santa Clara County in California, asking people to volunteer in the program via flyers. Sign-up sheets were posted in communal areas near homes, such as shared mailbox kiosks. The researchers varied whether these public posted sheets required residents either to print their names or only a randomly generated code. Participants were not aware they were taking part in an experiment.
Residents who could see what their neighbors did were nearly three times more likely to participate in the program than residents who could not. This effect was nearly seven times greater than offering a $25 incentive for signing up, the company’s previous policy.
“We all kind of knew that reputation would matter for these kinds of things, but what was surprising was how big that effect was compared to financial incentives,” said study coauthor Moshe Hoffman, an economist at the University of California, San Diego. “It was just so big an effect for so little cost.”
“Many firms, non-profit organizations, and well-meaning government entities would like to increase prosocial behavior, to fight global warming, decrease waste, increase voter turnout, support disaster relief, and much more,” Hoffman added. “In many of these cases, it is quite feasible to make contributions observable, thereby taking full advantage of people’s strong inborn concerns for reputation.”
The scientists also varied whether the materials participants were given said that signing up would benefit others, or whether the program was just a new feature the company offered. The effect of observability was less when the program was not described as benefiting the public good. This supports theoretical work that predicts that observability generally benefits cooperation only if participation benefits others.
The researchers also discovered the effect of observability was greater in apartment buildings than with row houses and individual homes. They suggest this is because residents in apartment buildings are more likely to interact with their neighbors in public spaces, boosting the role of ongoing relationships and observability. In addition, observability dramatically increased participation among homeowners but not with renters, which Nowak and his colleagues suggest is due to how renters, being more transient, are likely to be less invested in relationships with their neighbors.
“A natural next step is looking at how other theories related to cooperation work in field settings—for example, dynamic social networks, in which people have control over who they interact with,” said study coauthor David Rand, a psychologist at Yale University.
“Everyone on our team is obsessed with cooperation,” Hoffman said. “We hope to continue doing research figuring out why people cooperate and how to get people to cooperate more. While we have ample opportunities to study cooperation in the laboratory or in mathematical models on our chalkboards and computers, our biggest obstacle, like so many academics, is finding ways out of the ivory tower and into the real world. We would love to find firms or policy makers or other do-gooders who want to collaborate with us to figure out how to get consumers or citizens to be more cooperative in the real world setting closest to their heart.”