Ever since Donald Trump handed down his executive order temporarily halting all immigration from seven majority-Muslim nations for three months and barring refugees from Syria indefinitely, the social-media outpouring from liberals has focused, understandably, on how unfair the policy is to Muslims.

Those who didn’t decry the injustice of it all instead highlighted how important it is to protect refugees from harm (myself included.) They point out that Anne Frank’s family tried in vain to secure asylum in the U.S. in the 1940s. “Anne Frank today is a Syrian girl,” The New York Times Nicholas Kristof wrote.

Those are poignant, strong arguments against the policy. But according to one fascinating line of psychological research, they’re not likely to work on conservatives.

Rather than emphasizing concern over the harm that might come to refugees, says Matt Feinberg, an assistant professor at the University of Toronto, to truly persuade Trump supporters on the matter it would have been be better to go with something like this:

“These refugees and immigrants are just like our family members who came to America in years past to seek a better life. All our ancestors wanted was to live the American dream, and that’s why today’s immigrants and refugees have chosen to come to America, so they too can live that same American dream that brought our families here. That dream is what our nation was founded on, it is what brought our grandparents and great-grandparents to this great land, and it is the great success story that these immigrants want to be a part of.”

It’s a message high on patriotism and loyalty—two “moral frames” that research shows are more important to conservatives than are traditionally more liberal values, like reciprocity and caring.

Feinberg and his co-author, Stanford University sociologist Robb Willer, have extensively studied how it is that liberals and conservatives—two groups that now seem further apart than ever on their policy preferences—can convert people from the other side to their way of seeing things. One reason this is so hard to do, they explain, is that people tend to present their arguments in a way that appeals to the ethical code of their own side, rather than that of their opponents.

For example, when Feinberg and Willer asked liberals to write an op-ed aiming to convince conservatives of the value of same-sex marriage, most wrote something to the tune of, “Why would we punish these people for being born a certain way? They deserve the same equal rights as other Americans.” The problem is, research on thousands of people around the world, summed up in something called Moral Foundations Theory, has shown that liberals are more likely than conservatives to endorse fairness-based arguments like these. Meanwhile, just 8 percent of the liberals in Willer and Feinberg’s study were able to craft an argument that would appeal to conservatives’ value of loyalty toward your own kind. (So something like, “Our fellow citizens of the United States of America deserve to stand alongside us ... We should lift our fellow citizens up, not bring them down.”) What’s worse, some of them picked an argument that directly contradicted what many conservatives value, with arguments like, “your religion should play no part in the laws of the United States.”

As part of the same study, which they published last year in Personality and Social Psychology Bulletin, Feinberg and Willer tried to see if this type of “moral reframing” would be more effective. Previously, they had found that conservatives were more likely to endorse environmental protections when researchers activated their concerns about purity, rather than the more liberal concern about “harm”: A picture of a forest covered in rotting garbage, in other words, performed better with Republicans than a forest of tree stumps. This time, the researchers tested four different hot-button political issues, each time trying to reframe it in terms of the values that the Moral Foundations Theory tells us are more important for the opposite political side. Again, for liberals that’s “harm and fairness (e.g. benevolence, nurturance, equality, social justice),” and for conservatives, “group loyalty, authority, and purity (e.g., patriotism, traditionalism, strictness, religious sanctity).”                                             

First, they had participants read an article that advocated for the support of universal health care, using either a “fairness” argument (“health coverage is a basic human right”), or in terms of “purity” (a high uninsured rate means more “unclean, infected, and diseased Americans”). Conservatives who read the purity argument were much more supportive of universal health care, and, surprisingly, even Obamacare.

Next, they tried the same principle with liberals and military spending. Those who read an argument saying we should maintain high levels of military spending because it’s a poverty-fighting tool— “through the military, the disadvantaged can achieve equal standing”—were more likely to support a robust defense budget than those who read a paper on how the military “ensures the United States is the greatest nation in the world.”

Finally, they crafted an argument that making English the official language of the U.S.—a traditionally conservative stance—would actually lead to “fairer” outcomes for immigrants by helping them avoid discrimination. Liberals were more likely to warm to that view, the researchers found, but so were a separate group of participants who just happened to really value fairness. That suggested the reason the reframed messages work is because they resonate with deeply held moral values.

In a later study that’s currently under review, Feinberg and Tilburg University’s Jan Völkel found this even worked to get conservatives to dislike Donald Trump, and liberals to disavow Hillary Clinton. Conservatives were less likely to support Trump if arguments against him were presented in terms of his patriotism— “has repeatedly behaved disloyally towards our country to serve his own interests”—rather than a tendency to overlook the marginalized (“his unfair statements are a breeding ground for prejudice.”) Liberal participants, meanwhile, were more likely to be swayed by Clinton’s ties to Wall Street than by the incident in Benghazi.                                                                       

So if it’s so easy, why don’t more people—either in studies or in real life—try this strategy?

“We tend to view our moral values as universal,” Feinberg told me. That “there are no other values but ours, and people who don't share our values are simply immoral. Yet, in order to use moral reframing you need to recognize that the other side has different values, know what those values are, understand them well enough to be able to understand the moral perspective of the other side, and be willing to use those values as part of a political argument.”

Some people just can’t bring themselves to take that last step, he said, even if they know it’s more effective. And perhaps the reason it’s so difficult is because politics is so deeply intertwined with our personal values. When something is important to us, it’s usually for a reason, and it’s hard to break free of those reasons, even for political expediency’s sake. To do so would take an abundance of empathy, and that’s in short supply all around these days.

What’s more, not every researcher buys that it is quite so easy to persuade ideological opponents in the current climate, where people are changing their avatars to “#Resist” and “#MAGA.” “This [research] assumes that both sides are rational and at least partially open to hearing a different point of view,” said Blair Kidwell, a Florida International University professor whose consumer psychology research was cited by Feinberg and Willer. He says Trump is spearheading a “war on facts and even information itself,” which is causing many conservatives to distrust anyone but a fellow Trump supporter. “This is something, in my opinion, that cannot be fixed simply by appealing to conservative’s authority, purity ,and duty,” he added.

Still, there’s one thing Feinberg said definitely won’t work. In the wake of the executive order, Feinberg said he saw lots of liberals lobbing ad-hominem attacks, such as “you're being un-American” or “you’re making the Statue of Liberty cry.”

“People typically do not do well when attacked,” he said, “this could simply push them to be more staunch in their position.”

If you can’t persuade your political foes, that is, you can at least try not to make the conflict worse.