Science1 publisher2 min readPublished
Re-interviews found 40% of 9/11 memory details changed within the first year
Memory researchers have been re-interviewing the same few thousand Americans since September 2001. About 40% of the details had changed by the one-year interview, the changed versions then stuck, and confidence stayed high.
The Scientist · Science desk
What happened
- Memory researchers began collecting accounts in the days after Sept. 11, 2001, asking a few thousand people across the United States where they were, who they were with and what they were doing.
- The same volunteers were re-interviewed at one year, three years and 10 years, and the team is now collecting their accounts at a quarter century.
- At the one-year interview, about 40% of the details a person gave did not match what that same person had reported a few days after the attacks.
- The volunteers were extremely confident their memories were correct even though so many of the details had moved.
- Whatever a person reported at one year, error included, tended to remain their memory at the follow-ups that came afterwards.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- capability Re-interviewing the same individuals lets a changed detail be dated to a window between two waves. A fresh sample each time would show only that accounts differ from person to person.
- constraint Anyone weighing a first-hand account loses a signal they probably relied on. Certainty ran high in this panel while 40% of details moved, so how sure a person sounds is no guide to which accounts held.
- precedent Because the year-one version persisted, later interviews largely confirm each other. Agreement between a 10-year telling and a 25-year telling is expected even for details that had already changed by 2002.
The 40% is a disagreement rate between two accounts from the same person [6]. The questions were about the circumstances of hearing the news, not the facts of the attack [11]. The baseline is what the volunteer said in the days after Sept. 11, which is itself a memory [6]. What the figure measures is how far an account moved from its own earliest version. About 60% of the details still matched a year on [23].
The changes were big. One man said at the one-year interview that he was on a train when he heard; days after the attacks, he had said he was at work [8]. A woman who first described being with her husband later remembered being with her son [9].
Then the accounts settled. What people said at one year, error included, was still what they said at the later interviews [10]. Follow-ups came at one year, three years and 10 years, and collection at a quarter century is under way [4][5]. Roughly 15 years separate the 10-year wave from the current one [24]. The 40% comes from the one-year comparison; the quarter-century accounts are still being gathered [5][6].
Confidence does usually track accuracy, with exceptions [16]. The researchers' explanation for this being one of them runs through emotion and conversation. Emotional events improve recall of a few important details and worsen recall of many others, while raising confidence [17]. Being heard when you share a memory raises confidence further [18]. The authors think both applied to 9/11, an event that was highly emotional and widely retold [19]. The researchers manipulated neither variable. They interviewed people [1][4].
When they present the findings, the authors wrote, they are usually met with protest and told this may be true of others [21]. "We, too, believe our recollections of learning about 9/11 are accurate," they wrote. "But because we are memory researchers, we know there is a good chance that about 40% of the details we remember are wrong." [20]
Whether an event produces these memories at all depends on the community: flashbulb memories are specific to a community and help bind people to it [25]. A survey from the 1970s found 75% of Black Americans had flashbulb memories of learning of the assassination of Martin Luther King Jr. [22]
What to watch
- Whether the quarter-century wave shows further drift or confirms that accounts froze after year one.
- Whether the researchers publish confidence ratings as numbers or only describe volunteers as extremely confident.
- Whether attrition is reported: who is still answering by the fifth wave, and how they differ from those who left.