Imagine scrolling through your feed and seeing a headline about a group you dislike: “Members of this group are lying about their taxes.” It feels true, doesn’t it? It fits the story already in your head, it flatters your sense of who is good and who is bad, and you might share it without a second thought. Now imagine the exact same headline, except it’s about a group you love. Suddenly you want evidence. You question the source. You wonder if the story is biased. You might even click away. That frustrating, familiar reaction is what psychologists call “partisan bias,” and it’s one of the most important forces behind the spread of misinformation. In simple terms, people overaccept information that supports their identity and underaccept information that threatens it. We don’t just evaluate claims based on how true they are; we evaluate them based on how they make us feel about ourselves and our groups. A claim that says, “your side is right” gets treated as a gift. A claim that says, “your side is wrong” gets treated as an attack. This is true across politics, religion, sports fandom, and even trivial everyday loyalties. But for a long time, researchers couldn’t agree on why this happens. Is it because our thoughts are twisted by loyalty, or is it because we simply know more facts that support our side? That question might sound small, but it has enormous consequences for how we fight fake news, how we talk across divides, and whether we can ever be honest with ourselves about our own blind spots.
For years, the most common way of studying partisan bias was to take real-world groups, like Democrats and Republicans, and show them political statements. The results were always the same: conservative participants were more likely to accept conservative-friendly misinformation, and liberal participants were more likely to accept liberal-friendly misinformation. On the surface, that sounds like clear evidence of bias. But there was a problem. Democrats and Republicans don’t just differ in their motivations; they differ in their knowledge. People who identify as liberal have spent years absorbing liberal news, being exposed to liberal arguments, and noticing when liberal claims are debunked. Conservative people have done the same in their own information environment. So when a Democrat accepts a pro-Democrat claim, it could be because he or she is being biased, but it could also be because that claim actually feels more familiar, and familiarity can masquerade as truth. Likewise, when a Republican rejects a pro-Democrat claim, it could be because of identity protection, but it could also be because he or she has genuinely seen more counter-evidence over the years. As the authors of this new paper put it, previous work couldn’t tell the difference between two competing explanations. The first explanation says that people are motivated reasoners: we weigh evidence with an invisible thumb on the scale, demanding little from flattering information and demanding a mountain of proof from threatening information. The second explanation says that people are simply more knowledgeable about identity-congenial information: we know more true things that support our side, so our judgments are actually based on knowledge, not bias. Decades of research struggled to separate these two forces because they always used real-world identities with real-world histories. Then Tyler Hubeny and colleagues at the University of Texas at Austin came up with a clever way out of this trap.
Instead of studying identities that people had carried with them for years, they created identities out of thin air. In two experiments involving a total of 1,411 adult participants recruited online through Prolific, people were randomly assigned to one of two artificial groups. These groups were brand new. They had no history, no political platform, no real-world rivals, and no reason for anyone to feel strongly about them. Participants had no pre-existing knowledge about these groups because the groups didn’t exist before the study. This was the key insight: if partisan bias appears even for random, meaningless identities, then the bias cannot be explained by differential knowledge. It would have to be caused by something deeper, something about the way the human mind clings to and protects whatever identity it has, however arbitrary that identity might be. The experiments used a method from signal-detection theory to measure two separate things. The first was the “acceptance threshold,” which is the amount of evidence a person demands before deciding that something is true. Some people have low thresholds: they accept claims eagerly, even with little evidence. Other people have high thresholds: they need strong proof. The second was “truth sensitivity,” which is a more objective measure of how well a person can actually distinguish true statements from false ones. By separating these two abilities, the researchers could determine whether differences between groups were driven by a biased standard of judgment or by an actual difference in knowledge.
The results were striking and consistent across both experiments. Participants who saw statements that were favorable to the group they had just been assigned to were much more willing to accept those statements as true. They had lower acceptance thresholds, meaning it didn’t take much evidence to convince them. But when the exact same kind of statement was unfavorable to their group, their thresholds shot up. Suddenly they were skeptics. They demanded more proof, pointed out weaknesses, and hesitated before saying something was true. This happened even though the groups were completely arbitrary and no one had any real reason to care. It happened even though participants had no prior knowledge differences, because the groups were invented by the researchers. There was no fact-checking history to fall back on, no community of interpreters, no deeply held political philosophy. And yet the bias was still there. The brains of participants treated “what is good for the group” as “what is true” in a very real, measurable way. The signal-detection analysis made this especially clear. The differences between congenial and uncongenial statements showed up in the acceptance thresholds, not in an overall ability to tell truth from falsehood. In other words, participants were not more perceptive, smarter, or better informed when judging flattering statements. They were simply willing to accept flattering statements with much less evidence. This is exactly what identity-protective motivated reasoning looks like: not the inability to recognize truth, but the unwillingness to require proof when the truth makes us feel protected.
These findings have big implications for how we think about misinformation and how we try to correct it. Many current interventions assume that people fall for fake news because they lack accurate information. The solution, according to this logic, is to provide more facts, debunk false stories, and improve media literacy. But this study suggests that approach is incomplete. People don’t just eat misinformation because they’re ignorant. They eat it because it tastes like loyalty. It makes them feel like a good member of their group. If that’s the case, then simply telling someone they’re wrong can backfire. Correcting a false claim can feel like an attack on their identity, which pushes them even deeper into resistance. The researchers argue that interventions need to address identity protection itself. For example, self-affirmation techniques can help people feel secure enough to tolerate information that challenges their group. If people are reminded that they are good, moral, and valued independent of the group, they don’t need to protect the group as fiercely. Other potential strategies include framing truthful information in a way that is compatible with someone’s values rather than hostile to them, using messengers who share the audience’s identity, and reminding people before they encounter information that accuracy is a shared value, not another partisan weapon. In a world where misinformation is addictive, it might be more effective to reduce the emotional need for false beliefs than to simply try to replace them with facts. This study reminds us that sharing misinformation is often not about being fooled. It’s about being a faithful member of a tribe, even when the tribe is nothing more than a random label assigned a few minutes ago.
Of course, no single study can answer every question. The identities used in these experiments were deliberately weak, so in some ways the fact that bias appeared is remarkable. If a randomly assigned group with no history can trigger partisan bias, imagine how much stronger the effect must be for real political identities. But real-world identities also come with other complexities. When participants in these experiments judged congenial and uncongenial statements, they had no punishing consequences for being wrong. They didn’t have to defend their answer to a friend or a political ally. They didn’t have a lifetime of social pressure weighing on them. In real life, the pressure is even stronger, but so are the opportunities for nuance and reflection. The finding that bias appears even in minimal groups tells us that identity protection is a deep and fundamental feature of human cognition, not just a side effect of polarized media or partisan culture. It also suggests that we are all vulnerable. This is not a problem of one political party or one type of person. It’s a problem of being human. The study does not say that knowledge is irrelevant. Prior knowledge still shapes what we believe, and readers who know more will usually be better at sorting true from false. But this study proves that knowledge is not the only force. Even when knowledge differences are absent, the motivational pull of identity can change how much evidence we demand. That is a sobering thought, but it is also a hopeful one. If bias is wired into the way we socialize and identify, then we can design our environments to reduce that bias. We can build cultures that reward curiosity over loyalty, that make it safe to change our minds, and that separate “this information makes my group look good” from “this information is true.” But we need to do it with a clear-eyed understanding of what we are fighting. We are not just fighting bad facts. We are fighting the deep human desire to belong. The first step is admitting that ”true” and “good” can feel almost identical in our minds, and that we all have a lower bar for the information we want to believe.