When Higher Intelligence Makes You Worse at Reasoning
Here's something that should worry every person who believes they're rational: the more mathematically capable you are, the less likely you are to think clearly about politically charged issues.
That counterintuitive finding comes from research by Dan Kahan, a scholar studying how people process scientific information. His work suggests that something other than raw intelligence determines how well we reason with data—and that thing is political identity.
The Skin Cream Test
Kahan's study presented participants with what looks like a simple problem. Imagine a fake clinical trial testing a new skin cream for rashes. Participants who used the cream had improved rashes 81% of the time, while those in the control group improved 65% of the time.
Most people look at these numbers and conclude the cream works—it's obvious which group had better outcomes. The larger number sits with the treatment group. This is what researchers call "proportional reasoning": comparing two ratios rather than raw totals.
But here's where it gets interesting. When Kahan tested 1,111 Americans on this problem, something unexpected happened. Participants with higher numeracy scores—meaning those more comfortable working with numbers and quantitative data—were significantly more likely to get the right answer. The math-savvy participants correctly identified that no treatment actually worked better than the control; in fact, the cream appeared to make rashes worse.
This is exactly what you'd expect: more mathematical skill leads to better reasoning when the problem is purely statistical. No politics involved. Just numbers.
When Politics Enters the Picture
Kahan then flipped the script. Instead of skin cream, he presented participants with a made-up study on gun control—comparing cities that enacted stricter firearm regulations against those that didn't—and tracked whether crime rose or fell.
The results were radically different.
When shown data suggesting gun control reduced crime, higher numeracy helped participants find the correct answer. But when shown data suggesting gun control increased crime—a conclusion aligned with conservative politics—numeracy suddenly stopped mattering. Highly numerate people scored no better than those who struggled with math.
The same pattern appeared in reverse for liberal participants. When the data showed gun control reducing crime, their reasoning improved with increasing numeracy. But when the data showed it increasing crime—a politically threatening result—all that mathematical ability vanished. They performed just as poorly as people with low numeracy scores.
This is the polarization effect. The research also found identical patterns for questions about climate change and fracking.
The Paradox of Science Literacy
Perhaps most troubling: Kahan's data showed that people who score higher on general science literacy are actually MORE polarized on politically contentious topics than those with lower scores. More knowledge doesn't lead to better reasoning—it leads to more rigid defense of existing beliefs.
The opposite holds for people who score high on something the researchers called "science curiosity." Those participants remained consistent regardless of political alignment.
What Can Be Done
Kahan identified two promising approaches. First, avoid partisan framing entirely. Instead of talking about "gun control" or "climate change," focus on specific local policies that address concrete problems—like bipartisan Florida lawmakers working together on sea-level rise without debating whether climate change is human-made.
Second, cultivate curiosity rather than knowledge. People who are curious about how the world works don't show the same polarization effect as those who simply acquire information.
The Hard Truth
The beliefs most people hold aren't formed by carefully weighing evidence. They're shaped by tribal loyalty—to political party, to social group, to cultural identity. Questioning those beliefs risks ostracism from the community that matters most.
This isn't new. It's just more visible now because everyone can shout into an infinite digital space. But awareness might be the first step toward change.
"Most people will continue to hold partisan beliefs regardless of evidence—all while believing they arrived at these positions through careful analysis."
Bottom Line
The core insight is sound: when problems are apolitical, numeracy helps reasoning. When they're political, higher intelligence makes you worse—which explains why policy debates feel impossible no matter how much expertise exists.
The biggest vulnerability is that solutions remain theoretical. Avoiding partisan language and fostering curiosity help at the margins, but they don't address the deeper problem: most people aren't interested in changing their minds, even when presented with contradictory evidence. The research confirms this. Changing behavior requires something harder than awareness—it requires genuine discomfort.