Published on 2026-07-25
Why are we so sure of memories that turn out to be wrong?
A 10-year study tracked how people remembered learning about 9/11. Their accounts kept changing. Their confidence in them never dropped one bit.
Ask anyone old enough where they were when they heard about a huge, shocking piece of news, and watch what happens to their face. It softens into something specific: the kitchen, the radio, the exact sentence a friend used to tell them. People don't hedge these memories. They don't say 'I think'. They say 'I remember it like it was yesterday', and they mean it completely. Which makes it a little unsettling that this exact kind of memory is one of the ones psychologists have caught changing the most, while the person holding it stayed just as certain the whole time.
A name from 1977
Psychologists Roger Brown and James Kulik gave this kind of memory its name in 1977: a flashbulb memory, the vivid, detailed snapshot of the moment you first learned about a surprising and consequential event, not the event itself, but the scene around learning of it. Their original research looked at how people recalled hearing the news of assassinations, including that of Martin Luther King Jr. People described the room, the company, the exact words, all with a confidence that felt closer to a photograph than an ordinary memory.
Ten years of checking the photograph
A photograph, though, is exactly what a flashbulb memory isn't, and the clearest demonstration of that came from a study that tracked more than 3,000 people across seven American cities after the September 11 attacks. Psychologist William Hirst, Elizabeth Phelps and a large team of co-authors published their findings in 2015, in the Journal of Experimental Psychology: General, after surveying the same participants one week after the attacks, then again at 11 months, then again at 35 months, asking each time for the same details: where they were, who told them, what they were doing. Answers that matched what someone had said the very first week counted as consistent. By 11 months, consistency across the group was only 63 percent. By 35 months, it had slipped to 57 percent. Nearly half of the specifics people gave a week after the attacks had shifted by the time three years had passed.
Here's the part that should give anyone pause: none of that erosion showed up in how sure people felt. Confidence ratings stayed high across every survey, averaging above 4 on a 5 point scale, even for the exact details that had already drifted. People weren't performing certainty for the researchers. They genuinely felt they remembered correctly, at the same moment their own account contradicted what they'd said a year earlier. One detail fared worst of all: how people remembered feeling in the moment. Emotional recall matched the original report only about 40 percent of the time, even though the emotion is usually the part people describe with the most conviction.
A vivid memory and an accurate one are not the same claim, even when they arrive in the same confident sentence.
What the vividness is actually made of
None of this means the researchers caught anyone lying, or that flashbulb memories are somehow fake. The surprise and the emotional jolt of a huge piece of news are real, and they do carve a memory that feels unusually sharp. What the studies keep finding is that the sharpness is a feeling about the memory, not a measurement of it. The vividness comes partly from how often a story gets replayed, discussed, compared with other people's versions, folded into the news coverage that followed it, and every one of those retellings is a chance for a small detail to shift while the overall feeling of clarity stays exactly as strong.
That's a strange kind of consistency: the story changes, and the sense of certainty about the story doesn't move an inch. It means confidence, on its own, tells you almost nothing about whether a specific memory holds up, whether it's your memory of a global event or a smaller one closer to home, like the exact words in an old family argument or the precise order events happened in during a stressful week.
This also explains why arguing with someone over a shared memory rarely goes anywhere. Both people can describe their version with equal conviction, equal detail, equal insistence that they'd never forget a moment like that, and both can still be partly wrong, because conviction was never the thing being measured in the first place. The researchers weren't testing who felt most sure. They were quietly checking each account against the same person's account from a year or two earlier, which is a comparison almost nobody makes in everyday life.
So the next time a memory arrives dressed in total certainty, that certainty is worth noticing rather than trusting outright. It's not a sign the memory is false. It's simply not proof that it's true, and after a decade of researchers checking people's stories against their own earlier stories, that gap between feeling sure and being right is the one detail that never seems to fade.
Sources
- How Accurate Are Memories of 9/11? (Scientific American) (scientificamerican.com)
- Long-term memory for the terrorist attack of September 11 (PMC, National Institutes of Health) (pmc.ncbi.nlm.nih.gov)
- Flashbulb Memory In Psychology: Definition and Examples (Simply Psychology) (simplypsychology.org)