It is a sadly familiar scenario: a lone attacker commits a horrendous act of violence. In the immediate aftermath, the motive is unclear. Yet many people are quick to conclude he must be mentally ill; others just as quickly assume he is a terrorist.

Kellogg’s Nour Kteily wanted to understand the psychological underpinnings of this stark divide.

In a series of three experiments—two of which involved surveying participants immediately following a violent attack—Kteily and coauthors conclude that the driving force for this divide goes beyond stereotypes about terrorists. Instead, people are driven by a desire to protect their own views of the world and of the groups with which they associate themselves.

If a person who shares, say, our partisan beliefs or religion commits a heinous act, our instinct is to explain that away, Kteily says. Attributing violence to mental illness is one way to do that when there is no obvious motive.

But, the research shows, even after a motive becomes clear, people are still inclined to find ways to distance themselves from an in-group perpetrator.

“There is a desire, conscious or unconscious, to distance your group from nasty behavior, a desire to protect your cherished identity,” says Kteily, an assistant professor of management and organizations.

The consequences of this phenomenon go beyond simply reinforcing one’s own beliefs. When participants assumed a perpetrator was mentally ill, they were more likely to endorse a lenient punishment. The opposite was true when participants deemed the perpetrator a terrorist.

Motivated Reasoning

The backdrop for these findings is our increasingly partisan and polarized world.

“We’re getting more ideologically divided,” Kteily says. “So understanding the processes that lead to or that influence how we process information is incredibly important.”

Kteily and coauthors, Masi Noor, of Keele University in the UK, and Birte Siem and Agostino Mazziotta, both of FernUniversität in Germany, began with an established psychological concept called motivated reasoning.

“People will fill in the blanks with things that confirm their existing beliefs.”

People engage in motivated reasoning to lessen discomfort with an idea. For example, parents generally believe their children are good people. So when faced with the uncomfortable fact that their child started a schoolyard fight, parents may be motivated to explain away that reality by insisting that the other kid started it.

This motivation extends beyond our family to larger groups—both demographic groups like race or religion, and partisan groups that we voluntarily join.

Mental Illness versus Terrorism

The researchers’ first two studies tested whether people’s partisan beliefs predicted whether they would label a violent attacker as mentally ill or a terrorist.

It is jarring when someone from your group commits a violent act, the researchers reasoned. An easy way to protect your positive beliefs about your group is to blame the violent behavior on mental illness. After all, if people are not in control of their faculties, they—and by extension the group you both belong to—bear less culpability for their actions. On the other hand, if someone from a different group commits the same action, you will be more willing to attribute the violence to terrorism.

The first study was conducted on the day of the Brexit vote in Britain in 2016. A week earlier, a British man named Thomas Mair murdered Jo Cox, a Member of Parliament who was a strong supporter of the “Remain” campaign that advocated staying in the European Union.

The researchers quickly put together their study. They surveyed roughly 200 British participants online about their partisan feelings on the Brexit vote and about Mair’s motivation in stabbing Cox.

Importantly, on the day of the Brexit vote, Mair’s motives remained unclear.

Participants who supported the “Leave” campaign, and thus could see themselves being grouped with Mair if his motivations ended up being political, were more likely to say that Mair was mentally ill than a terrorist. Conversely, those who identified as “Remain” supporters—and thus would be less likely to find themselves in the same partisan group as Mair—tended to label him a terrorist.

“Any time there’s ambiguity about someone’s motives, that allows room for people to bring their own perspectives to bear on it,” Kteily says. “People will fill in the blanks with things that confirm their existing beliefs.”

A second study showed a similar pattern. German participants were surveyed immediately after a Syrian refugee injured 15 people at a German festival in 2016. Participants who identified themselves as pro-immigration were more likely to say the perpetrator was mentally ill. Those who self-identified as anti-immigration were more likely to call the refugee a terrorist.

Distancing the Attacker

Next, the researchers tested what happens when a perpetrator’s motive is clear. If people cannot attribute violent behavior to mental illness, will they find other ways to protect their own beliefs and cherished social identities?

These findings indicate that stereotypes are not the only force at work when someone is labeled either mentally ill or a terrorist.

The researchers surveyed about 500 Americans online. First, they assessed how strongly participants identified themselves with America through questions such as, “Being American is an important part of how I see myself.”

Next participants read about “Mr. A,” a college student arrested after brandishing a rifle and yelling at passersby. Some participants read that Mr. A had a history of heavy involvement in political gatherings; others read that he was bipolar; a third group got no information about his background.

Participants were then asked questions about Mr. A, including whether his parents were likely born in the U.S. and whether he was likely Christian, Jewish, Muslim, or atheist. The researchers wondered whether participants would distance themselves from Mr. A by saying he had foreign-born parents or assuming he belonged to a minority religion, which would allow them to protect their own sense of American-ness.

“Yeah, maybe this person is American,” Kteily says of their motivated reasoning. “But he isn’t really American. He’s not really one of us.”

Indeed, those who read about Mr. A’s political activities were more likely to distance themselves from him compared to those who thought he was mentally ill.

But there was further nuance to this result: It occurred most clearly when participants highly identified as American and therefore presumably cared more about protecting the American identity. When participants had a low level of American identification, they did not distance themselves more from the politically motivated Mr. A than from the mentally ill one.

These findings also indicate that stereotypes are not the only force at work when someone is labeled either mentally ill or a terrorist, Kteily explains.

One could imagine that participants labeled Mr. A a Muslim when they learned he was politically motivated not because they wanted to distance him from their group, but because they stereotypically associated Islam with political violence.

However, the findings suggest this was not the case. Participants who strongly identified as American would presumably not have been any more or less exposed to stereotypes of Muslims being terrorists than participants who weakly identified as Americans. Yet the results differed between these two groups in how likely they were to assume Mr. A was Muslim.

The results also highlight why ascribing political versus mental-health motives to violent attackers matters: participants who believed Mr. A was politically motivated advocated for harsher punishment than those who thought he was mentally ill.

Growing Partisanship

This desire to protect one’s own group may help explain how one person’s terrorist is another’s mentally troubled perpetrator. But it does not make the discussions around these events any easier, Kteily says—especially as we dig in deeper and deeper into our respective partisan camps.

Kteily, who discussed his research in the week after the Parkland, Florida school shooting, referenced the heated debates about gun control that followed.

“People are bringing to bear information that fits preexisting beliefs. That’s part of why the debate becomes so fraught,” he says. “Any given set of data can be chopped and served up differently, and people do that in systematic ways that align with their partisan identity. We’re not unbiased consumers of information. Rather, we go looking within information to find what we already believe.”