Plain-language answer
Confirmation bias is the tendency to search for, interpret, and recall information in ways that fit what you already believe or expect, while giving less attention to information that doesn’t. It shows up in how evidence is gathered (looking for confirming cases more than disconfirming ones), how it’s interpreted (reading ambiguous evidence as supportive), and how it’s remembered (confirming instances simply stick better than disconfirming ones).(Nickerson, 1998)
Why it matters
Coincidence-spotting runs on exactly this asymmetry. If you believe a particular date, name, or person is “significant” for you, you will notice every time it turns up and barely register the much larger number of times it doesn’t — or the times something equally striking happened involving a date, name, or person you’d attached no significance to at all. The believed pattern accumulates supporting instances because it’s the one being watched for; the unbelieved near-misses never get counted because nobody’s counting them.
Worked example: the unkept tally
Suppose you notice you keep thinking of a particular old friend right before they call. Each time it happens, it registers — it’s the kind of thing you’d tell someone about. But this only looks like a pattern because the other three relevant counts are never kept: the times you thought of them and they didn’t call; the times they called and you hadn’t been thinking of them; and the times you thought of any of the dozens of other people in your life shortly before hearing from them, which would be just as “meaningful” by the same logic but don’t get noticed because no belief is primed to notice them.
Without all four counts — thought-then-called, thought-then-not, called-without-thought, and neither — there’s no way to tell whether thinking of someone actually predicts their call more than chance would. Confirmation bias is precisely what stops those other three counts from ever being kept in the first place.
Common misconception
Confirmation bias isn’t a failure that only happens to people who are careless or credulous — it’s a default feature of ordinary memory and attention, and research reviewing decades of work on it finds it operating across experts and novices, skeptics and believers alike.(Nickerson, 1998) Knowing about it intellectually reduces its effect only partially; actually counting the disconfirming cases, not just remembering to “watch out for bias,” is what catches it in practice.
Limits and open questions
Confirmation-seeking isn’t purely a flaw, either: in many everyday situations, testing a hypothesis by looking for instances that fit it is a reasonably efficient strategy, and a full, balanced search for disconfirmation on every belief would be exhausting and often unnecessary. The trouble is specifically with low-stakes but emotionally compelling beliefs — like “this kind of coincidence follows me around” — where the biased sample never gets corrected because nobody was ever trying to test it rigorously in the first place.
Related
- How to evaluate a coincidence claim builds in a structural defence against this: step 7 (“count the opportunities”) explicitly requires counting the misses, not just the hits.
- Pattern perception and the clustering illusion covers the related tendency to see structure in sequences that have none.
Key takeaways
- Confirmation bias shapes what evidence you look for, how you interpret ambiguous evidence, and what you remember — not just what you conclude.
- A “pattern” built entirely from noticed hits, with no record of the misses, can’t be distinguished from chance no matter how many hits there are.
- This is a normal feature of memory and attention, not a personal failing — which is exactly why it needs a deliberate counting habit, not just awareness, to correct for.