Table of contents
A woman hears a crash outside her apartment and looks through the window. One car is sitting crooked in the intersection. Another is halfway onto the curb. A man is standing in the street yelling, and somebody already has a phone out recording.
Ten minutes later, three versions of what happened are moving through the building group chat. One person says an aggressive driver caused it. Another says a distracted pedestrian stepped into traffic. A third says the city screwed up when it redesigned the intersection.
Same event. Three stories.
And the weird part is, nobody has to be lying.
Direct Answer
I think one of the biggest mistakes we make when we talk about bias is assuming it happens at the end. We picture somebody getting the facts, thinking them through, and then letting politics, fear, identity, or emotion mess with the conclusion.
But that is usually not where the problem starts.
Bias can shape what information reaches us in the first place, what catches our attention, which sources we trust, how we interpret something that is not completely clear, what we remember later, and which explanation feels right the next time we encounter something similar. Research on selective exposure, source credibility, motivated reasoning, memory distortion, and repetition all points in that direction.[1][2][3][4][6]
Belief formation is not fact in, conclusion out.
There are filters all the way through.
The bias happened before the argument
Go back to the intersection for a second.
One person notices the guy yelling. Another notices the pedestrian. The third person has been angry about that intersection redesign for six months. All three of those things could be true. But whichever piece grabs each person’s attention first has a chance to become the frame for everything else.
That is what makes bias so hard to see in ourselves. It usually does not feel like distortion. It feels like observation.
We think, “I saw what happened.”
Maybe. But what did you see first? What did you miss? What were you already expecting to see? What did somebody else notice that never even registered for you?
Research on motivated reasoning shows that once people commit to an initial judgment, evidence supporting that judgment can carry more influence than evidence pushing against it.[2] So the first conclusion is not just the end of the process. It can start influencing the process that comes after it.
That is the part I keep coming back to.
The belief starts helping decide what counts as evidence for the belief.
We do not all start with the same information
Before we argue about how people interpret information, there is a more basic problem: we are not all receiving the same information.
That sounds obvious when you say it out loud, but I do not think we really act like we understand the consequences of it.
Think about how much of what you know about the world reaches you through a screen. Search results. Facebook. TikTok. YouTube. News sites. Podcasts. Friends sending you links. People you follow. People you blocked three years ago and forgot you blocked. The conversations happening around you. The communities you belong to. Your own habit of clicking certain things and ignoring others.
Research on selective exposure has consistently found that people tend to choose information that supports attitudes and beliefs they already hold.[1] Experimental research involving YouTube has also shown that changing recommendation behavior can change the news people are exposed to and the ideological diversity of what they consume.[5]
That does not mean an algorithm reaches into your brain and installs a belief. I think that explanation is way too simple.
It means the raw material is different.
Two people can sit down to argue about the same issue, both genuinely believing they have looked at the facts, while each has spent months or years being exposed to a different collection of facts, examples, scandals, arguments, experts, clips, and exceptions.
Then we wonder why the other person cannot see what seems completely obvious to us.
They may literally not be looking at the same pile of information.
Then attention cuts the pile down again
Even if two people receive the exact same information, they still are not guaranteed to walk away with the same thing.
Attention is limited. Something grabs us. Something else fades into the background.
The angry guy in the street stands out. The pedestrian does not. Or the pedestrian stands out because you almost got hit by a car last year. Or the intersection itself stands out because you have been complaining about it forever.
We do this constantly.
And again, this does not mean perception is worthless. It means perception is selective.
Once one detail gets promoted in our mind, that detail starts carrying more weight. It becomes part of the story we use to make sense of everything else. By the time we start deciding what happened, some information has already been mentally highlighted and some has already been pushed toward the margins.
We are not working with the event anymore. We are working with the version of the event that survived our attention.
Trust is another filter
Then there is the source.
I have caught myself doing this plenty of times. I see where something came from and I have a reaction before I have even dealt with the actual claim.
Sometimes that reaction is justified. Sources matter. Expertise matters. Track record matters. Somebody who has been consistently dishonest should not get the same automatic credibility as somebody who has demonstrated that they know what they are talking about.
But there is a difference between considering the source and letting the source make the decision for you.
Controlled experiments have found that source credibility can change how plausible people judge information to be while they are reading it.[3] That makes sense. If I trust the messenger, I am more likely to give the message a chance. If I distrust the messenger, I may reject it before I really examine it.
The problem is that true information can come from somebody we dislike, and bad information can come from somebody we trust.
That is uncomfortable because it means “Who said it?” cannot replace “Is it true?”
Both questions matter.
Sometimes reasoning becomes the defense attorney
This is where I think it gets even more interesting.
We like to believe that better reasoning automatically protects us from bias. Sometimes it does. Being able to slow down, examine evidence, spot contradictions, and question assumptions obviously helps.
But reasoning can also become the defense attorney for something we already want to believe.
In studies of contested political issues, people have rated arguments supporting their existing attitudes as stronger, pushed harder against opposing evidence, and sought out information that supported what they already believed.[2]
That does not mean people are stupid.
Actually, that is part of the point.
A smart person who is motivated to protect a conclusion may simply be better at constructing a sophisticated argument for keeping it.
I have started thinking about this less as, “Am I reasoning well?” and more as, “What job is my reasoning doing right now?”
Am I testing the belief?
Or am I defending it?
Those are not the same thing.
And if there is no evidence that could change my mind, I am probably not investigating anymore. I am building a case.
Memory does not save us either
Even if we somehow got through all of that cleanly, there is still memory.
Most of us know memory is imperfect, but I think we still treat our own memories like video files. We remember something clearly, so it feels like we are replaying what happened.
That is not how memory works.
Classic research by Elizabeth Loftus and colleagues showed that information introduced after an event can change what people later report remembering about the original event.[4]
Think about what that means in normal life.
Something happens. You see part of it. Then somebody tells you what they saw. You read a post about it. You talk about it three times. Somebody adds a detail. You hear the same explanation over and over. A month later, you are not necessarily pulling the untouched original event out of storage. The event has been living inside everything that happened after it.
Then repetition adds another problem.
Research on the illusory truth effect has found that repeated statements can feel more truthful simply because they are familiar, even when people have knowledge that contradicts the repeated claim.[6]
That one should bother all of us a little bit.
Because repetition is everywhere.
A claim gets repeated because people believe it. Then the repetition makes the claim feel more believable. Then more people repeat it.
At some point familiarity starts wearing the clothes of evidence.
The loop is the part that matters most
If bias only affected one decision at one moment, this would be a much smaller problem.
But beliefs change what happens next.
Once I believe something, that belief can influence what I search for, who I listen to, which headline I click, what I notice, which source I trust, and what I dismiss without spending much time on it. Research on selective exposure shows that people often prefer information that fits their existing attitudes, especially when those attitudes matter to them.[1]
So the process is not really a straight line.
It loops.
Reality or event → information available to me → what I actually receive → what I notice → what I trust → how I interpret it → what I remember → what I believe → what information I choose next.
Yesterday’s conclusion becomes part of tomorrow’s filter.
That is how two people can start with a relatively small difference and end up living inside completely different understandings of the world. Not because one woke up and decided to reject reality, but because each trip through the loop changes the conditions of the next trip.
A little difference can compound.
So what do we actually do about it?
I do not think the answer is trying to become unbiased.
Good luck with that.
We are human beings. We are going to have prior experiences, emotions, loyalties, fears, assumptions, blind spots, and preferences. Pretending we can delete all of that is probably just another way to become blind to it.
The better goal is to build some friction into the process.
When I feel absolutely certain about something, especially something that fits neatly into what I already believe, I want to get better at asking where the information came from before it reached me.
What did I not see?
What would the strongest version of the other explanation be, not the dumbest version I can easily knock down?
Am I rejecting a claim because the evidence is bad, or because I hate the source?
Am I remembering the original event, or the story I have told about the event twenty times since then?
And the biggest one: what evidence would actually make me change my mind?
If I cannot answer that last question, that should tell me something.
It does not automatically mean I am wrong. Some beliefs have overwhelming evidence behind them. But if every possible piece of conflicting information gets instantly reclassified as fake, corrupt, biased, irrelevant, misunderstood, or proof of the conspiracy itself, then the belief has become almost impossible to test.
At that point it is not only a conclusion anymore.
It is a filter protecting itself.
Back at the intersection
The intersection still happened one way.
That matters.
Saying people experience information through filters does not mean there is no objective reality or that every interpretation is equally valid. The car was somewhere. The pedestrian was somewhere. Somebody had the right of way. Somebody may have made a mistake. There are facts underneath the stories.
The three people in the group chat may each have one legitimate piece of what happened. The mistake is when one piece quietly becomes the whole thing.
That is where I think intellectual humility actually becomes useful. Not the fake version where we shrug and say nobody can know anything. The useful version is being confident enough to have a position while leaving enough room to discover that something important never made it through our filters.
Maybe I am right.
But what am I missing?
That question is a lot more useful than assuming the only biased people are the ones who disagree with me.
What this looks like at scale
Now take the apartment group chat and make it the entire internet.
Millions of people are moving through different search results, feeds, clips, headlines, podcasts, recommendation systems, influencers, institutions, social circles, and increasingly AI-generated summaries. Research suggests both personal choices and platform design can affect which information people encounter.[1][5]
Then every person brings a different history into whatever reaches them.
Different trust. Different fear. Different identity. Different memories. Different experiences. Different things they have heard repeated a thousand times. Different beliefs already sitting there waiting to interpret the next piece of information.
We call it “the public conversation,” but I am not sure there is one public conversation anymore. There are millions of overlapping versions of it.
That does not mean changing somebody’s feed automatically changes their politics or worldview. Research on information environments is more complicated than that, and effects are not uniform or deterministic.[5]
But it does mean we should be careful when something feels so obvious that we cannot understand how another person could possibly see it differently.
Maybe they are ignoring reality.
That happens.
Maybe we are.
That happens too.
Or maybe each of us is holding a different piece of the same intersection and mistaking our piece for the entire scene.
Understanding bias is not a trick for proving everyone else wrong. If anything, using it that way completely misses the point.
The real value is learning to inspect our own path to certainty.
What reached me? What did I notice? What did I trust? What did I dismiss? What did I remember? What did I already believe? And what would I have to see to change my mind?
The closer we can get to answering those questions honestly, the better chance we have of separating what actually happened from the story our filters built around it.
References
-
Hart W, Albarracín D, Eagly AH, Brechan I, Lindberg MJ, Merrill L. Feeling Validated Versus Being Correct: A Meta-Analysis of Selective Exposure to Information. Psychological Bulletin. 2009;135(4):555-588. https://pmc.ncbi.nlm.nih.gov/articles/PMC4797953/
-
Taber CS, Lodge M. Motivated Skepticism in the Evaluation of Political Beliefs. American Journal of Political Science. 2006;50(3):755-769. https://doi.org/10.1111/j.1540-5907.2006.00214.x
-
Wertgen AG, Richter T. Source credibility modulates the validation of implausible information. Memory & Cognition. 2020;48(8):1359-1375. https://doi.org/10.3758/s13421-020-01067-9
-
Loftus EF, Miller DG, Burns HJ. Semantic integration of verbal information into a visual memory. Journal of Experimental Psychology: Human Learning and Memory. 1978;4(1):19-31. https://doi.org/10.1037/0278-7393.4.1.19
-
Yu X, Haroon M, Menchen-Trevino E, Wojcieszak M. Nudging recommendation algorithms increases news consumption and diversity on YouTube. PNAS Nexus. 2024;3(12):pgae518. https://doi.org/10.1093/pnasnexus/pgae518
-
Fazio LK, Brashier NM, Payne BK, Marsh EJ. Knowledge does not protect against illusory truth. Journal of Experimental Psychology: General. 2015;144(5):993-1002. https://doi.org/10.1037/xge0000098