Plain-language answer
Hindsight bias is the tendency, once you know how something turned out, to overestimate how predictable that outcome really was beforehand — and to misremember your own earlier uncertainty as having been smaller than it actually was. The classic experimental demonstration gave people historical vignettes and asked some of them to judge, in advance, how likely each of several possible outcomes was; people who were later told which outcome actually happened rated it as having been more likely all along, even though nothing about the evidence available beforehand had changed.(Fischhoff, 1975)
Why it matters
Once you know a coincidence already happened, it is very hard to reconstruct how surprising it should have seemed before you knew the outcome — which is exactly the number that matters for judging how unusual it really was. Hindsight bias pushes that reconstructed estimate in one direction only: toward “this was bound to happen,” which makes genuinely unlikely coincidences feel merely confirmed, and makes merely unremarkable ones feel retrospectively special.
Worked example: rewriting the prediction
Imagine being asked, before a coin toss, how likely heads is: most people say 50%, correctly. After the toss lands heads, ask people to recall what they would have predicted, and many will recall — sincerely, not dishonestly — a number above 50%. Nothing about the coin changed; what changed is that the actual outcome is now available to colour the memory of the prediction. Coincidence stories are usually retold only after the outcome is already known, which means the “how surprising was this really” step is being reconstructed under exactly the condition most likely to bias it toward “very.”
Common misconception
“I had a feeling this would happen” is not necessarily a false memory invented after the fact — but hindsight bias means it also can’t be taken at face value without some way of checking it. The only way to tell the difference between a genuine advance prediction and a retrospectively inflated one is to have a record made before the outcome was known. This is exactly why how to evaluate a coincidence claim puts “was this specified before or after the events” as an early, load-bearing question, not an afterthought.
Limits and open questions
Fischhoff’s original work distinguished two related effects: people exaggerating how much foresight they themselves would have had (retrospective hindsight bias), and people exaggerating how much foresight someone else should reasonably have had (a judgment often applied, fairly or not, when assessing decisions after the fact). Both are well replicated, but the exact size of the effect varies across situations and individuals, and isn’t something this site treats as a fixed number to plug into a calculation — it’s a reason for caution about any “I knew it” claim made after the fact, not a quantity with one settled value.
Related
- Confirmation bias covers the related tendency to notice and remember confirming evidence more readily than disconfirming evidence.
- How to evaluate a coincidence claim — step 4 asks directly whether a matching rule was set before or after the outcome was known, which is the practical defence against hindsight bias.
Key takeaways
- Knowing an outcome changes how predictable it feels in retrospect, even though the actual evidence available beforehand hasn’t changed.
- A recalled “advance feeling” about an outcome can’t be distinguished from a hindsight-inflated memory without a record made before the outcome was known.
- This is one of the main reasons this site insists on asking when a prediction or matching rule was actually made, not just trusting how confident someone sounds about having expected it.