The biggest lie of human memory is that it feels true. Although our recollections seem like literal snapshots of the past, they’re actually deeply flawed reconstructions, a set of stories constantly undergoing rewrites.
Consider our collective memories of 9/11. For the last 10 years, researchers led by William Hirst of the New School and Elizabeth Phelps of New York University have been tracking the steady decay of what people recall about that tragic event. They first quizzed people shortly after the attacks, then after one year, and found that 37% of the details had already changed. Although the most recent data have yet to be published, they’re expected to reveal that the vast majority of remembered “facts” are now make-believe.
If memory flaws only affected our personal past, that would be bad enough. But the problems created by our mistaken recollections affect all of society. More than 75,000 prosecutions every year are based entirely on the recollections of others. While perjury is a felony, the overwhelming majority of eyewitness errors aren’t conscious or intentional. Rather, they’re the inevitable side effects of the remembering process.
In recent years, neuroscientists have documented how these mistakes happen. It turns out that the act of summoning the past to the surface actually changes the memory itself. Although we’ve long imagined our memories as a stable form of information, a data file writ into the circuits of the brain, that persistence is an illusion. In reality, our recollections are always being altered, the details of the past warped by our present feelings and knowledge. The more you remember an event, the less reliable that memory becomes.
And this returns us to the problem of eyewitness testimony. Eyewitnesses are repeatedly asked to recall what they saw, but their answers are inevitably influenced by the questions being asked. The result is more confidence in increasingly less accurate testimony.
Such errors often have tragic consequences. According to the Innocence Project, a legal advocacy group, about 75% of false convictions that are later overturned are based on faulty eyewitness testimony.
Can anything be done to improve the situation? In a new paper by Neil Brewer, a psychologist at Flinders University in Australia, the answer is a resounding yes. Dr. Brewer focused on the police lineup, in which witnesses are asked to pick out a suspect from a collection of similar looking individuals.
Normally, witnesses are encouraged to take their time and carefully consider each possible suspect. But Dr. Brewer knew that strong memory traces are easier to access than weak and mistaken ones, which is why he only gave his witnesses two seconds to make up their minds. He also asked them to estimate how confident they were about the suspects they identified, rather than insisting on a simple yes-no answer.
To test this procedure, Dr. Brewer and his colleagues asked 905 volunteers to watch a series of short films showing such crimes as shoplifting and car theft. The subjects then looked at 12 portraits, only one of which was the actual suspect. According to Dr. Brewer’s data, his version of the lineup led to a large boost in accuracy, with improvements in eyewitness performance ranging from 21% to 66%. Even when subjects were quizzed a week later, those who were forced to choose quickly remained far more trustworthy.
The larger lesson is that, when it comes to human memory, more deliberation is often dangerous. Instead of simply assessing our familiarity with a suspect’s face, we begin searching for clues and guidance. Sometimes this involves picking the person who looks the most suspicious, even if we’ve never seen him before, or being swayed by the subtle hints of police officers and lawyers. As a result, we talk ourselves into having a memory that doesn’t actually exist.
Simple reforms can help to compensate for our mnemonic failings. Unless we are ruthlessly skeptical of the past, we will continue to confuse fact and fiction, and innocents will be sent to jail.