Hindsight Bias
The tendency, after an event, to see the outcome as having been more predictable than it actually was — and to misremember our prior beliefs to fit what eventually happened.
Baruch Fischhoff's 1975 Journal of Experimental Psychology: Human Perception and Performance paper coined the phrase "creeping determinism" for what we now call hindsight bias. Fischhoff gave participants historical scenarios and either told them the outcome or didn't. Those who knew the result rated it as having been far more probable beforehand than the uninformed group did — and, when asked to recall their own earlier estimates, drifted them toward the known answer. The bias has three layers: memory distortion (we misremember what we predicted), inevitability (the actual outcome feels like it had to happen), and foreseeability (we believe we could have called it). It corrupts post-mortems — every failure looks obvious afterward — biases medical and legal judgments of negligence, and inflates our confidence in our own forecasting ability. The strongest counter is a written prediction log: a date-stamped record of what you actually thought before the outcome was known, immune to memory drift.
Frequently Asked Questions
Why is hindsight bias dangerous in post-mortems?
Because once the outcome is known, every preceding decision looks like a step toward it — including the ones that were entirely reasonable given what was known at the time. Post-mortems run under hindsight bias punish the wrong choices and miss the structural causes. The fix: reconstruct what was knowable at each decision point before judging.
How does hindsight bias differ from confirmation bias?
Confirmation bias operates while you are still gathering evidence — you favor what fits your hypothesis. Hindsight bias operates after the outcome is known — you rewrite your prior expectations to fit what actually happened. They often compound: confirmation shapes the prediction, hindsight rewrites the prediction once reality lands.
Can hindsight bias be reduced by simply being aware of it?
Awareness alone is weak. Fischhoff and later researchers found that participants told about the bias still showed it in subsequent judgments. What works better is structural: prediction logs written before outcomes, pre-mortems that imagine multiple endings, and explicit reconstructions of what was actually knowable at the decision point.