Skip to main content

Different strokes for different folks

by Trish Barker

Simulations using NCSA’s Abe help psychologists better understand how human beings carve out cultural niches.

“There is no such thing as a person without culture,” says University of Illinois psychology professor Dov Cohen. While we all have our individual characteristics and quirks, we’re also greatly influenced by the distinct ways of living, acting, and even thinking that surround us from the moment we’re born.

For example, various experiments indicate that Western culture is generally more individualistic and analytical, while East Asian culture is more collectivist and holistic. In one often-cited experiment, members of both cultures were asked to describe what they saw in an aquarium. Westerners tended to single out the big fish in the foreground, while the Easterners were more likely to describe the scene as a whole.

Note the qualifications in that paragraph, however. “Generally.” “Tended to.” “More likely.” Culture does not stamp out cookie-cutter people. There are always individuals who don’t conform to the general trend.

“Usually, you consider that unexplained behavior ‘error,'” says third-year graduate student Ivan Hernandez. “Which is like saying, we don’t really know why it’s going on, but let’s kind of ignore it for now.”

Cohen thought a computer model could help him better understand this intra-culture variability. Cohen, Hernandez, and Karl Dach-Gruschow, who earned his PhD in psychology from Illinois in spring 2010, used Northwestern’s NetLogo modeling environment to develop an agent-based model of cultural evolution. Using the Abe supercomputer at NCSA, the team repeatedly ran their model, generating multiple cultural outcomes and comparing them to what’s known from observation and experiments. Their preliminary results were presented in 2010 at The Conference on Honor in Barcelona and have been accepted for a presentation at the 2012 conference of the Society for Personality and Social Psychology.

Evolution in action

The model begins with a population of agents. Each one is defined by two characteristics—positive reciprocity (how likely is the agent to return a favor?) and negative reciprocity (how likely is the agent to pay back a wrong?).

High positive reciprocity and high negative reciprocity are characteristic of many “honor cultures,” often found among stateless people or in other environments where government can’t be relied upon to provide support and right wrongs.

“They’ll go through hell and high water to pay back a favor, and they pay back their debts, but they also make good on their threats,” Cohen says.

Other agents were a mix: high negative reciprocity and low positive reciprocity, low negative reciprocity and low positive reciprocity, etc.

The agents interact following simple rules. Agents that are high on negative reciprocity can command a certain amount of deference; but if two agents high on negative reciprocity meet and one is thought to have cheated the other, a feud can develop, resulting in a huge loss of resources for both parties. When any agent is depleted to zero resources, that agent disappears from the system. Agents that build up a certain level of resources can reproduce, generating offspring who are likely to share their parent’s characteristics.

“The computer simulations are incredibly useful, because they let you model evolution in action, the transmission of cultural ideas or ways of doing things from one generation to the next, from one adult to his or her successors,” Cohen says.

“It lets you examine how systems and the people within them develop,” Dach-Gruschow adds.

Each run of the model continues through hundreds of generations, and the model was run more than 1,200 times. That required more computer power than a simple laptop or even small cluster could provide.

“It quickly became apparent that if you were going to do this seriously…you were going to need a lot of computing resources,” Cohen says. Celso Mendes, an Illinois computer scientist who frequently works with NCSA, recommended that Cohen apply for a start-up allocation on the center’s supercomputers. “What they gave us really made this project possible, and every time we had a problem, it was a phone call away.

“NCSA has been so good to us. It’s a great resource that’s just tremendously helpful.”

Room for difference

The culture generated by the interaction of agents in the model looked remarkably like the real world—a mix of different types of people.

“All of the types continue to exist, some at lower levels,” Cohen says. “In honor cultures, one of the most prevalent types are honor people, but their exact opposite is also quite prevalent: People who in our model we call the ‘adventitious’—they have no inclination to cooperate, they will minimally appease you if you absolutely make them, they will cheat if they can. But they also aren’t punishers either. And we’ve found that in our lab experiments, too.”

The results indicated that intra-culture variability is not error or noise, but instead that it’s logical for individuals within a culture to follow a variety of strategies, some in step with the cultural norms and others diverging from it. Cohen, Hernandez, and Dach-Gruschow compare the results to the way animals carve out niches in their ecosystem. Among midshipman fish, for example, it’s generally the largest males who win mates. But very tiny male midshipman fish are able to stealthily mate with female fish without being observed by their rivals. Both types of male can be successful.

“There is room, even when the predominant culture is one thing, to have people following a different cultural rule set,” Hernandez says.

Future research potential

As these preliminary results from their modeling efforts are presented at conferences, Cohen and Hernandez are gathering valuable feedback from their peers. They plan to continue to work with the model, trying different conditions and parameters. For example, would making the world of the model more settled lead to a less honor-based culture, which would support the current theory of how honor cultures evolve and devolve? What impact does a market economy have on culture? Law enforcement? Can the model illuminate the process through which one type of culture shifts to a different type of culture?

“It’s hard to understand the origins of culture. We study the differences, but where do the differences come from originally?” Hernandez says. It’s usually not possible to watch cultures evolve over many generations—unless the “people” are agents in a computer models. “This type of approach allows that to be possible.”

Team members
Dov Cohen
Karl Dach-Gruschow
Ivan Hernandez

Disclaimer: Due to changes in website systems, we've adjusted archived content to fit the present-day site and the articles will not appear in their original published format. Formatting, header information, photographs and other illustrations are not available in archived articles.

Back to top