September 10, 2021
Most people who are old enough to remember Sept. 11, 2001, can quickly recall where they were when they learned that America was under attack. They can recall how they heard the news and who they were with. They probably have some recollection of their feelings in the moment.
People tend to be extremely confident in the accuracy of these details, too. The same is true for other historical events, like the JFK assassination, the Challenger explosion and the fall of the Berlin Wall. These highly emotional — and often shocking — events create what scientists call flashbulb memories — seemingly unforgettable imprints of what people were doing when they received word.
"We all have an idea — you hear this all the time — that memory for bigger, more emotional, more impactful, more personally meaningful events is better," said Karen Mitchell, assistant chairperson of West Chester University's psychology department. "How many times do you hear people say, 'I'll never forget it?'"
But science suggests that people mistakenly place more confidence in the accuracy of these memories than they should. In reality, these memories often include inaccuracies. And it's not simply because people forget details. Sometimes, false memories are inserted.
Memories are not like archived video recordings that can be played back directly, experts say. Rather, the brain stores portions of lived experiences in different places and must reconstruct them when commanded. But every time they're accessed, information can be added or deleted. Similar experiences can even get confused into one memory.
Vishnu P. Murty, the principal investigator at Temple University's Adaptive Memory Lab, likens a memory to a note written on an iPad. That document might get edited later to include new experiences or correct mistaken information. Some sentences might get erased. The note can be accessed at any time, but it's no longer in its original form.
"In that way, memories are these living things," Murty said. "Each time we reactivate a memory, we are providing an opportunity to alter that memory."
Historic events, like 9/11, offer a glimpse into how that process plays out and why many memories contain inadvertent inaccuracies. They also help explain why people are so confident in their most vidid memories.
In the days immediately after 9/11, Mitchell and other researchers asked thousands of people how they learned of the terroristic attacks and the ways they reacted to the news. They also asked a series of questions about the event itself, including the number of planes involved and the crash sites.
Nearly one year later, they asked the same people the same questions. They did so again ahead of the third and 10th anniversaries. Their goal was to trace the consistency of their memories over time.
The study is among the most ambitious research conducted on flashbulb memories. Its findings are remarkable.
Within a year of 9/11, most people had significant inconsistencies in their flashbulb memories — the details involving the ways they learned of the attacks. On average, only about 63% of their responses matched what they had reported just days after. But their confidence in the accuracy of those memories was sky-high.
I think the broader idea is that our memories are not just a replay of prior events where sometimes the replay gets fuzzy," Wolk said. "There's also creation or construction of events that may not have occurred that can create these very strong feelings of prior experience of something. — David Wolk, Penn Memory Center
In the ensuing years, their memory consistency and confidence levels slightly declined before stabilizing. The inconsistencies reported at the first anniversary tended to be repeated over time with little correction, suggesting that a "stable, but not necessarily accurate, memory begins to form by the first year," the researchers wrote.
By contrast, people were much better at recalling the specific details of the attacks, known as event memories. They recounted the number of planes and crash sites with overwhelming accuracy throughout the 10-year period. Media coverage appeared to reinforce — and sometimes correct — these factual details.
The researchers did not conduct a follow-up survey prior to the 20th anniversary due to various factors, including the difficulty in tracking people over such a prolonged period of time. But Mitchell suspects gradual declines in flashbulb and event memory retention would have continued.
"If you're asking me personally what I would expect, I think forgetting would continue at a slow rate," Mitchell said. "It's not that there's no further forgetting, it's just that it slows down."
Similar test-retest studies have been done following all sorts of events. None of them have spanned the length of time involved in this 9/11 study. But a major theme runs through them: There is a considerable disparity between the confidence people place in their memories and the accuracies of them.
Consider one noteworthy example from a memory study on the Challenger explosion. Just 24 hours after the tragedy, an Emory University student wrote that she learned of the news when people walked into her religion class and began talking about it.
But 2 1/2 years later, her story was radically different: She claimed she was in her dorm room with her roommate when a news alert came on TV; she then rushed upstairs to inform a friend and called her parents. And just like the participants in the 9/11 study, she was extremely confident that what she remembered was accurate.
"I think if it teaches us nothing else," Mitchell said, "it's that some of these lay intuitions we have about how memory works really isn't how it works."
One of the primary reasons that the brain does not store memories as if they were video recordings is due to a lack of storage space, said Dr. David Wolk, co-director of the Penn Memory Center. For instance, if the brain added a new recording every time people saw their living room chairs, there would be a lot of unnecessary recordings piled up.
Instead, the brain saves space by separating the components of a memory and storing them with similar components from other memories. But it links those components together for future retrieval of the memory.
"If you had to store every event like a film, you'd have lots of copies of that chair," Wolk said. "In this case, you could just have this memory of that chair and that can be linked to a variety of different memories. It requires less storage space to have that flexibility."
But this setup occasionally can lead people astray, because the brain sometimes makes sourcing errors and other mistakes in the reconstruction process. When asked to retrieve one memory, the brain mistakenly may include details from a related experience, too.
"What happens is there's lots of other related memories that you have that have overlapping features to the one that you previously experienced," Wolk said. "So when you try to retrieve one event, you sometimes may retrieve aspects of another event that overlaps with it."
The inconsistencies observed in flashbulb memories illustrate this.
A man might say he learned of the Sept. 11 attacks while watching TV. In reality, he may have first learned of the news from a friend and later seen the images on TV. But at some point, his brain made an error. Then that memory, complete with its inaccuracies, gets stored, complicating the matter.
Such errors aren't uncommon. Wolk doubts he actually watched the second plane hit the South Tower when it happened — but that's mostly because he's spent his career studying the way the brain works. Otherwise, he suspects he might mistakenly believe the TV images he saw later that day were live.
"You remember the images on the TV and you remember sitting at the TV and watching it; it's linked with that day," Wolk said. "It becomes very hard to disentangle that experience of that day, which is already going to be tightly associated with how you did actually hear about the event. If you rehearse it the wrong way, even once, that memory is going to start to take over as the vivid and salient one."
In the laboratory, scientists have found ways to lure the brain into making such a sourcing error. In one technique, known as the DRM paradigm, researchers read people a list of words highly associated with a word that is not stated. When asked to recall as many words as possible, people frequently include the unstated word — perhaps because they thought of the word as the list was read and their brains filed that experience away, too. And they often are very confident it was indeed read to them.
"I think the broader idea is that our memories are not just a replay of prior events where sometimes the replay gets fuzzy," Wolk said. "There's also creation or construction of events that may not have occurred that can create these very strong feelings of prior experience of something."
But in the case of the 9/11 study, why did the study participants' flashbulb accounts change so significantly by the first anniversary and then essentially stabilize over the ensuing years, with the same inconsistencies often repeated?
The frequent retrieval of those memories over that first year likely played a big role, Murty said.
Our confidence in our memories is a direct result of the contents of the memories. By that I don't mean the facts of it. I mean the features of it. — Karen Mitchell, West Chester University
"Especially with emotional memories, like 9/11 or the Challenger explosion, ... I think we want to turn them into narratives," Murty said. "We're trying to figure out the story that we tell ourselves, or the story that we tell other people. We want confidence in that story also."
Consider an engagement story, he said. The first time the person tells it, the story contains a bunch of rich details. The next few times, a new detail may be included and others may get left out. But by the 15th time it's told, the story has been refined into a consistent narrative — even if it contains inaccuracies unintentionally inserted along the way.
"If you're looking at consistency, it's going to be pretty inconsistent in the first year as you're still working your way through that memory," Murty said. "But once you're at year one, the story is down. And good luck trying to re-angle that."
Though the study participants were not very good at accurately recalling the details of how they learned of the terroristic attacks, they had little problem remembering the facts of the event itself.
In this realm, the consistent media coverage and frequent discussion of 9/11 both reinforced information and provided opportunities for correction, Mitchell said.
Notably, her research group found that the ability to recall the location of President George W. Bush at the time of the attacks — a non-critical detail — significantly improved following the release of the documentary "Fahrenheit 9/11," regardless of whether people actually saw it.
The 2004 film had generated considerable discussion over Bush's reaction to the news. Seven years later, the participants' retention of Bush's location still held steady. A similar memory correction was observed following the release of the 2006 movie "United 93."
But there's no such answer key for people's flashbulb memories.
If "I tell you exactly where I was when I heard about that — I was in my car. I was on College Street in New Haven, Connecticut, coming back from the fMRI scanner. It was sunny," Mitchell said, "What are you going to say? No, you weren't?"
And so inaccuracies can continue uncorrected.
Because people generally are unable to fact-check their personal memories against any kind of established record, they must find other ways to determine their accuracy.
"Our confidence in our memories is a direct result of the contents of the memories," Mitchell said. "By that I don't mean the facts of it. I mean the features of it. Can I remember a lot of detail or a little bit of detail? Are those details — even if there's only one or two of them — really vivid?
"So confidence is sort of a derived feeling. It's a derived experience. I'm taking all of these details and I'm making some judgment. Is this a real memory or not? If I have a lot of that stuff, that's what makes me feel like I really experienced this myself in the past."
But sometimes people mistakenly take the vividness of a specific feature and apply confidence to their entire memories, Murty said.
"We just make a single confidence based off of the thing that we feel is really vivid," Murty said. "That one portion of it might be really accurate, but that doesn't mean your whole memory is accurate. And we might falsely assign confidence to the single parts of it."
And vivid emotions have their own impacts on memory — particularly negatively-associated emotions like stress, anxiety and fear. If someone feels shocked or threatened, visual attention becomes very restricted, Mitchell said. Content sharpens, but context fades.
If a woman is held up at gunpoint, she likely will be able to describe the weapon in exquisite detail, but have no clue whether the perpetrator was wearing glasses, Mitchell said. That's because the gun is a direct threat to her life; the perpetrator's appearance has little importance in the moment.
"There's both physiological and behavioral reasons why emotion has these differential effects," Mitchell said. "We're still trying to understand why it can improve memory for content and disrupt memory for context."
But if there's one lesson that can be applied from the research on flashbulb memories, Mitchell said, it's this: The idea that people will never forget or confuse memories from life's most emotionally impactful events is a myth.