Social norms changed dramatically during the pandemic. Six-foot distancing and mask wearing became de facto rules in some communities, while elsewhere, wearing a mask was seen as an invitation for harassment. A new study in Proceedings of the Royal Society B asks how people learn new social norms and how flexible those norms can be. Using a series of simple video games, author Uri Hertz showed that some norms are easier to change than others. The least flexible behavioral patterns turned out to be those that actively harm others.
To study how people learn group expectations, Hertz, a cognitive neuroscientist at the University of Haifa, Israel, recruited 276 participants to play a simple video game. Players moved around a digital grid collecting stars and could zap each other by pressing a button. In each game, a single participant played with several other “players” that were actually preprogrammed bots. There were four possible social environments. The first environment was a culture of active harm, in which the bots frequently zapped other players to eject them from the game. The second was a culture of active helping, in which zapping was frequent but beneficial, conferring a bonus on the zapped player. The third and fourth environments were passive social cultures, where zapping was harmful or beneficial, respectively, but bots avoided zapping other players. Every participant played two trials of the game with different social environments.
No matter which environment players started out in, they readily adapted to match the social norms in their first trial. Then in the second trial, most participants easily switched to a new social norm, for instance transitioning from beneficial zapping to zap avoidance. “But one behavior was resilient to this change,” Hertz says. People who experienced an actively harmful zapping environment in the first trial kept zapping through trial two, even when the social norm switched to zap avoidance. An analogy in the pandemic might be heckling people who wear masks, a kind of active harm, says Leor Hackel, a psychologist at the University of Southern California (USC) in Los Angeles, who was not involved in this new work. In communities where confronting mask wearers is now the norm, it might be harder to learn tolerance than in the reverse scenario, he says, in which nobody is heckled.
“The question is what underlies this effect,” Hertz says. After the video game trials, Hertz applied several cognitive models to the data. The model that best predicted how the participants behaved was one in which players learned to mimic actions (zapping) more easily than nonactions (zap avoidance), and in which positive behaviors were attributed to individuals while negative behaviors were attributed to the group. For instance, if a player saw a blue bot giving a beneficial zap to another bot, the player was likely to assume something nice about the blue bot as an individual. But if that same player saw the blue bot harmfully zapping another bot, then the player would become hostile to the whole group. “Negative things tend to make you more suspicious towards everyone,” Hertz says. Combined, behaviors that are both active and negative are doubly biased and thus harder to overcome, which is perhaps why harmful zapping was the hardest behavior to change. Players who had seen harmful zapping in the first trial might assume that even avoidant bots could become hostile at any moment in the second trial.
Understanding when and why people struggle to change is a pressing question in psychology, on which this study sheds some light, USC’s Hackel says. Neuroscientist Oriel FeldmanHall at Brown University in Rhode Island notes that teaching new and realistic social norms is often challenging in the lab because subjects arrive with existing beliefs. The methods are “the cleverest part of this experiment,” she says, because they offer a workaround: a video game to test social behavior removed from any familiar social context.
Next up, Hertz hopes to run similar experiments using cliques of bots to test how participants learn to recognize in-groups. “Think about high school. There are the cool guys and the nerds,” he says. Nobody hands out a popularity manual, yet everyone knows the social order. “We don’t know a lot about how people infer these things that are group attributed,” Hertz says. So far, what does seem clear is that “big active things” drive behavioral change.
Other recent papers recommended by Journal Club panelists: