# Confirmation Bias
Sixteen twenty. Francis Bacon sits in his study, candlelight making shadows dance across parchment. And he writes a sentence that should have changed everything:
"The human understanding, when it has once adopted an opinion, draws all things else to support and agree with it."
Four hundred years later? We're still doing exactly what he warned us about. We're still building elaborate castles out of evidence that confirms what we already believe... and calling it thinking.
This is confirmation bias.
And here's what makes it so tricky — it doesn't *feel* like bias when you're doing it. It feels like clarity. It feels like finally seeing what was there all along.
Let me show you what I mean.
Nineteen sixty-six. Peter Wason — a psychologist at University College London with a reputation for elegant torture — has a puzzle. He shows people four cards on a table.
E. K. Four. Seven.
The rule is simple: if a card has a vowel on one side, it must have an even number on the other. Your job? Flip the minimum number of cards to test if the rule is true.
Most people flip the E card. Good start. And then... the four card. Which feels logical, right? You're checking if the rule holds.
But that's not testing the rule. That's *confirming* it.
The card you actually need to flip? The seven. Because if that card has a vowel on the back, the rule is broken. The even number tells you nothing new.
But almost nobody flips the odd number.
Out of Wason's first hundred subjects, only five got it right. Five.
And here's where it gets strange. Even when Wason told people they were doing it wrong... even when he walked them through the logic step by step... they'd do it again the same way. One subject — a mathematics student — got it wrong three times in a row. Apologized profusely each time. And then, you can almost hear Wason's exhausted sigh in the research notes, did it wrong a *fourth* time.
The bias isn't about intelligence. It's about how thinking works at the architectural level.
Nineteen seventy-nine. Three researchers — Lord, Ross, and Lepper — gather people with strong opinions about capital punishment. Some believers, some opponents. They give everyone the same evidence: studies that support the death penalty, studies that oppose it. Mixed bag. Balanced.
Then they ask: has your opinion changed?
Not only did opinions not change... they got *stronger*.
Pro-death-penalty people found the pro studies more convincing and poked holes in the anti studies. Anti-death-penalty people did the exact reverse. Same evidence. Opposite conclusions. Everyone left more certain they were right.
The gap between the two groups didn't shrink. It doubled.
The technical term for what happened is "biased assimilation." But really, what happened is this: we don't process information. We *sort* it. Into two piles. Stuff that proves I'm right... stuff that must be flawed somehow.
And we do it so fast, so automatically, that we experience the sorting *as* thinking.
So here's the turn. Why do we do this? Why would evolution build a brain that's so bad at changing its mind?
Here's one answer: because changing your mind used to be dangerous.
Think about your ancestors fifty thousand years ago. You're part of a small tribe. Maybe thirty people. Your survival depends on cooperation, shared beliefs, group cohesion.
Now imagine you start questioning the tribe's core assumptions. Maybe about which animals are safe to hunt. Or which neighboring group can be trusted. Or how the spirits work.
You're not just being intellectually curious. You're threatening the social fabric. You're making yourself unreliable.
In a world where exile means death, being right matters less than being aligned.
Jonathan Haidt argues this in *The Righteous Mind* — that reasoning didn't evolve to find truth. It evolved to win arguments and maintain relationships. Your brain is a lawyer, not a scientist. It's building a case, not testing a hypothesis.
But there's another layer.
Leon Festinger had an answer in nineteen fifty-four, and it's not the answer you'd expect. He called it cognitive dissonance. The idea is simple: holding contradictory beliefs creates psychological discomfort. Real discomfort. Measurable.
In later studies, researchers found that cognitive dissonance activates the anterior cingulate cortex — the same region that lights up during physical pain. Like mental nausea.
And we'll do almost anything to make that feeling stop. Including — and this is the key — reinterpreting evidence until it no longer contradicts what we believe.
It's not that we're lying to ourselves. It's that our brains are running a background program: minimize discomfort, maximize coherence.
Confirmation bias is a feature of that program. Not a bug.
Think about that for a second.
Your brain isn't trying to find truth. It's trying to find *peace*.
Daniel Kahneman and Amos Tversky spent decades mapping this territory. Their work in the seventies and beyond showed that we have two systems of thinking.
System One is fast, intuitive, runs on pattern recognition. System Two is slow, analytical, effortful.
And here's the problem: System One is always on. It's making snap judgments, filtering information, deciding what's important before System Two even wakes up.
Confirmation bias lives in System One. By the time you're consciously thinking about evidence, the sorting has already happened.
Kahneman tells this story — not about politics or beliefs, but about doctors. Diagnostic errors.
A doctor sees symptoms, forms a hypothesis, and then, without meaning to, starts noticing all the evidence that fits that diagnosis and downplaying what doesn't.
One study tracked emergency room physicians and found that when they made an initial diagnosis in the first three minutes, they stuck with it ninety-five percent of the time. Even when contradictory evidence emerged.
It's not malice. It's momentum. The first impression creates a gravity well, and everything else bends toward it.
This shows up everywhere.
In courtrooms, where a study of mock juries found that eighty percent of jurors decide guilt or innocence during opening statements... and spend the rest of the trial justifying that decision.
In laboratories, where scientists can spend years trying to prove a theory instead of trying to break it.
Karl Popper built an entire philosophy of science around this problem in the nineteen forties. He said: a theory isn't scientific unless it's falsifiable. Unless you can imagine evidence that would prove it wrong.
Because if you can't, you're not doing science. You're just collecting confirmations.
And here's the beautiful, terrible thing about Popper's insight: he was writing this while watching totalitarian regimes turn ideology into religion. Fascism. Stalinism. Systems that couldn't be questioned because they'd defined themselves as unfalsifiable.
Every contradiction became proof of enemies, saboteurs, insufficient faith.
Popper wasn't just describing how science should work. He was describing how thinking dies.
But that raises a question. If this bias is universal... if it affects scientists and doctors and judges and all of us... can we do anything about it?
There's been work in the last few years on something called "debiasing interventions." One technique is beautifully simple: consider the opposite.
Before you make a decision, force yourself to argue for the other side. Generate reasons why you might be wrong.
It sounds obvious, but studies show it works. A twenty seventeen experiment found that people who spent just two minutes arguing against their own position were thirty percent more accurate in their final judgments.
Not perfectly rational. But measurably better.
It creates just enough friction to slow down System One.
Another approach: pre-commitment. Decide in advance what evidence would change your mind. Write it down.
Because once you've formed an opinion, that question becomes almost impossible to answer honestly. The goalposts move. The standards shift.
Philip Tetlock studied this with expert forecasters — people whose job is predicting geopolitical events. The best forecasters, the ones he called "superforecasters," all did the same thing: they specified in advance what would prove them wrong.
They gave themselves no room to wiggle.
And they were right twice as often as everyone else.
And this is where it gets really uncomfortable.
Because the internet was supposed to help with this, right? Access to infinite information. Every perspective at your fingertips. The marketplace of ideas, finally realized.
Except that's not what happened.
The nineteen nineties brought us filter bubbles. Algorithms that learn what you click, what you linger on, what makes you feel validated. And then they show you more of that.
Not because there's a conspiracy. Because that's what keeps you scrolling.
YouTube's recommendation algorithm, according to a twenty nineteen analysis, doesn't show you what's true. It shows you what people who watched your last video watched next.
Which means if you watch one video questioning climate change, the algorithm interprets that as "this person likes climate skepticism" and serves you ten more.
Confirmation bias, which used to be a personal cognitive quirk, became structural. Built into how information flows.
By twenty sixteen, Oxford Dictionaries declared "post-truth" the word of the year.
And here's the finding that should genuinely worry us: a twenty twenty-one study by Christopher Bail took Democrats and Republicans and paid them to follow opposing political accounts on Twitter for a month.
The hypothesis was that exposure would increase understanding.
Instead? Republicans who followed liberal accounts became *more* conservative. Democrats who followed conservative accounts became *more* liberal.
People didn't think, "Huh, interesting perspective." They thought, "See? That's how crazy the other side is."
So the very thing we thought would fix the problem — more exposure to different views — can make it worse.
Because confirmation bias doesn't just filter information. It filters how we interpret disagreement. Every opposing argument becomes more proof that we're right and they're deluded.
I want to sit with that for a second. Because it's bleak.
If more information doesn't help... if exposure to other views backfires... if even scientists with training and incentives to be objective fall into the same traps... what's left?
Maybe the answer isn't in the information. Maybe it's in the question we bring to it.
Here's what I mean.
Bacon's insight from sixteen twenty wasn't just that we seek confirmation. It was that we do it unconsciously. The bias is invisible to the person experiencing it.
So the fix isn't to try harder to be objective. The fix is to assume you're biased and work backward from there.
There's a practice I've been trying.
When I feel certain about something — really certain, the kind of certain that makes me want to argue — I pause and ask: what would I need to see to change my mind?
And if the answer is "nothing" or "I don't know," that's a red flag. Not that I'm wrong. But that I've stopped thinking and started defending.
Because here's the thing about confirmation bias. It's not going away. Four hundred years of knowing about it hasn't made it disappear. Evolution spent millions of years building this machinery. You're not going to think your way out of it in an afternoon.
But knowing about it changes what you can do with it.
You can't eliminate the bias. But you can notice when it's running. You can catch yourself sorting instead of thinking.
And in that gap between the automatic response and the conscious choice, there's room to move.
So here's what you can try.
Next time you encounter an idea that makes you want to immediately argue... don't. Just for thirty seconds, hold it.
Ask yourself: why does this threaten me? What am I protecting?
And then — this is the hard part — ask what the smartest version of that opposing view would look like. Not the straw man. The steel man. The version that would make you think, "Okay, I can see why someone might believe that."
You might still disagree. Probably will.
But you'll disagree with something real instead of something you invented to be wrong.
That's not nothing. That's the difference between defending a position and actually thinking. That's the difference between being right by accident... and being right on purpose.
That's how you turn Bacon's warning into a practice.
Four hundred years late. But better late than never.