The fallibility of memory

Product & Experience

Although flashbulb memories are vivid, they are also full of errors. In 1986 the space shuttle Challenger exploded. If you try to recall that event, you probably remember it vividly. The day after this tragic event, Ulric Neisser, a professor who researches memories like these, had his students write down their memories of what had happened. Three years later he asked them to write their memory of the event again (Neisser, 1992). Over 90 percent of the later reports differed from the originals. Half of them were inaccurate in two-thirds of the details. One person, when shown the description she had written three years earlier, said, “I know that’s my handwriting, but I couldn’t possibly have written that.” Similar research has been conducted on individuals with memories of the 9/11 attacks, with similar results.

At 10.07 the customer call ends. At 10.12, sales writes: “Strong appetite for the premium plan.” Product records: “Pricing caused visible hesitation.” Research adds: “Customer wants the export feature before discussing price.” All three attended the same forty minutes.

At noon, they use the recording to settle two details and complicate the rest. The customer did praise the premium plan, then paused for six seconds after hearing the price. The export question came earlier and sounded casual. Its ambiguous tone lets everyone retain their preferred account.

On Friday, a decision paper quotes the praise and omits the pause. Three months later, the premium plan misses its target. The retrospective says customers had warned the team about exporting. By then, everyone describes the casual question as a warning, complete with urgency nobody heard at the time.

So the review uses the materials created before the result was known: recording, decision paper, forecasts and confidence ranges. The team marks each statement as observed, inferred or remembered. Several beautifully polished inventions appear in that final category.

Nobody is accused of lying. That would be easier. A liar can be challenged; an honest person can hold an inaccurate recollection with full conviction.

At 4.20, the group reaches the sentence that approved the original launch. Two people remember objecting to it. Both their names appear under “agreed”. Someone opens the recording again.