You Did Not Know It All Along. You Just Cannot Remember Not Knowing
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
Once an outcome is known, people remember having expected it, and that reconstruction makes the past look far more predictable than it was when anybody had to act.
How it is measured
The standard design asks people to estimate the probability of an event before it happens, then asks them afterwards to recall what they originally said. Recalled estimates shift systematically towards what actually occurred. A variant tells some participants an outcome and asks them to judge how likely it would have seemed without that knowledge, and those told the outcome consistently rate it more predictable than those not told, even when explicitly instructed to ignore what they know. The instruction does not help, which is the important finding.
The three things it does
The effect operates at several levels at once:
- •Distorting memory of what was actually predicted
- •Inflating judgements of how foreseeable the event was
- •Producing a sense of inevitability about how things unfolded
- •Which together make the past look like a story with one ending
- •And make anybody who failed to foresee it look negligent
- •It is strongest for events that had a clear single outcome
Why it happens
The mechanism appears to be reconstruction rather than lying. Learning an outcome immediately integrates it into everything already known, and memory does not keep a separate record of the state of knowledge before the update, so recalling the earlier state means rebuilding it from a base that now includes the answer. The mind also seeks a coherent account, and knowing the ending makes the preceding events easy to assemble into one, which suppresses the alternative paths that were genuinely open at the time and that felt equally plausible then.
What reduces it
A few techniques measurably weaken the effect where simply trying harder does not. Explaining how the outcome could plausibly have gone the other way forces the alternatives back into view and reliably reduces the distortion, which is the best supported countermeasure. Writing down predictions and reasoning in advance removes the reliance on memory entirely. Evaluating a decision against what was knowable rather than against what happened is the correct standard and requires that the reasoning have been recorded. And considering several possible futures before an event, rather than one, leaves more of them available afterwards.
Where it does real harm
The effect makes fair evaluation of past decisions nearly impossible, which has serious institutional consequences. Medical negligence assessment is affected, since a reviewer knowing the diagnosis finds the symptoms obviously indicative. Inquiries into disasters reliably conclude that warning signs were ignored, when those signs were indistinguishable at the time from thousands of others. Investment decisions are judged by outcome rather than by the reasoning available. The standard remedy is to record the reasoning and the alternatives considered at the moment of decision, so a later review has something other than the outcome to judge against.
The takeaway
Learning an outcome shifts recalled predictions towards it and makes the event seem more foreseeable, and being told to ignore what you know does not help. Memory keeps no separate record of the earlier state of knowledge, so recalling it means rebuilding from a base that already contains the answer. Recording the reasoning at the time is the only reliable defence.