Your memories are less accurate than you think

Earlier this year, the Air Force Office of Special Investigations reported that a cold case of nearly 14 years had finally been cracked. In 2015, a woman who was attacked by her Air Force instructor in 2000 had been able to describe a family portrait she noticed in his home. The instructor denied that it had ever hung on his wall—until the prosecution projected a photo of his family sitting on their living room sofa with the portrait visible behind them.

To close the case, investigators used a technique called cognitive interviewing, which was designed to help police question witnesses without blurring their memories.

If you question someone the right way, you can extract memories of events that took place years ago. And in this case, those memories were accurate. Yet our memories can also be contaminated by outside forces, internal biases, and even our own thoughts. Not every detail we recall is trustworthy.

Psychologists have spent decades studying why our memories become distorted—and how best to help eyewitnesses remember what they saw accurately. Meanwhile, neuroscientists are probing the ways our brains recollect visual information, and what that means for eyewitness testimony.

It’s not possible to interview eyewitnesses in such a way that promises pristine accuracy. But there are ways to keep false memories to a minimum, and even give ourselves a better shot at enhancing our own recollections.


There are many ways that our recollections can be distorted. That’s why Ronald Fisher, a psychologist at Florida International University in Miami, developed the cognitive interview. “If the interview is done properly, witness descriptions are generally pretty accurate,” he says.

But Elizabeth Loftus, a psychologist at the University of California, Irvine, isn’t so sure. She studies the forces that taint people’s memories after an experience is over, and has consulted on hundreds of criminal cases. “Just because somebody tells you something, and they say it with a lot of detail and a lot of confidence and a lot of emotion, doesn’t mean it really happened,” she says.

Both researchers, however, have seen how malleable our recollections are, and how easy it is for false memories to take root.

One problem Fisher has observed is that investigators have a habit of dominating their interviews with witnesses. “When you ask questions as opposed to encouraging people to give you a free narrative, that’s where you get errors,” Fisher says. “Your responses to open-ended questions tend to be much more accurate than responses to very closed questions, as in, ‘what color eyes does he have?’”

It’s also better to avoid pressing the witness to provide details that they feel uncertain about. “You might get less information like that, but the information you do get is more likely to be correct,” Fisher says.

When witnesses learn new information about the crime, they can fold those details into their understanding of the event—whether or not those details are correct. This sometimes happens when the interviewer asks suggestive questions that clue the witness into what the investigator believes. “If I ask you a question like, ‘wasn’t he wearing a green shirt?’, well, obviously I believe he was wearing a green shirt,” Fisher says. “You think that I know, so you might incorporate that into your memory.”

Loftus once consulted on a case where a recording featured a police officer trying to prompt a witness to identify the suspect in a set of photos. The witness said they did not recognize any of the pictures, but the interviewer was not satisfied. “The officer said, ‘I see your eyes drifting down to number six, what’s going on?’” Loftus says. “It was one of the more highly suggestive things I’ve ever seen in an actual interview.”

Misleading information can be conveyed in subtle ways. In some of Loftus’s early experiments in the 1970s, she asked what people recalled after viewing videos of car accidents. When Loftus asked how fast the two cars were traveling when they “smashed” into each other, people estimated that the vehicles were moving 7 miles per hour faster than they did when she used the word “hit.” A week later, those people were also more likely to recall broken glass in the video, although none had actually been visible.

Witnesses can absorb new information that contaminates their memories by talking to each other or seeing media coverage of the crime as well. Sometimes, a person can even unwittingly corrupt their own memories by speculating about how the event might have transpired. Loftus has spoken to people who witnessed car wrecks and were certain about which direction both vehicles were traveling when the accident happened. But investigators discovered that based on the routes both drivers were taking, the witness’s version of events was impossible. It could be that the witness made assumptions based on where the cars came to rest after colliding, she says. “Your own thoughts can act like external information and you can distort your own memory that way.”

Best-case scenario

If a person is sharing what they think is an accurate account, is there any way to tell what part of that is real and what’s not? Not at the moment, Loftus says. “You can’t just go by the verbal report and know whether something is a genuine memory or whether it’s a product of some other process—external information, suggestion, inference—unless you have corroboration.”

When Loftus gets involved in court cases, she tries to find out what outside influences might be responsible for false memories. While there’s no ironclad way to declare that some piece of a person’s memory is inaccurate, there are a few circumstances under which recollections are more trustworthy, she says.

Imagine that two men of the same race spend 45 minutes together before one decides to rob the other. The next day, the victim walks into a pizza parlor and sees a man he identifies as the perpetrator sitting at a table. If the accused man’s lawyer were to give her a call, Loftus says, she would not have much advice.

“You had a long exposure time, you had a short passage of time, there was no bad police procedure involved in the identification, you don’t have a cross-racial identification,” she says. A memory scientist would likely conclude that this was a fairly reliable recollection.

A witness might remember a crime more accurately if they have outside expertise in some aspect of what they are seeing, Fisher says. So if you are a hair stylist, you might be able to remember the perpetrator’s hair color and style more accurately than another observer would.

Sometimes, verification can come from other witnesses. Fisher joined detectives in the Miami-Dade Police Department as they interviewed witnesses to robberies in the 1980s. He reported that 94 percent of all witness statements were corroborated by other witnesses. That is not a measure of accuracy, he says, because in most investigations no one knows exactly what happened. Still, when two witnesses describe an event the same way, it lends credibility to their recollections.

The details that a person is most confident about are also most likely to be correct. There are no guarantees, however. “Certain people just exude more confidence than others, so confidence is not a perfect predictor,” Fisher says. And people can still feel confident about a memory that’s been contaminated by misleading or flat-out wrong outside information. “As you pile on these problematic factors…then you can erode any relationship between confidence and accuracy,” Loftus cautions.

One way to avoid contaminating a witness’s memories is for police to send someone unfamiliar with the suspect to run lineups or photo identifications. “A person who does not know who the suspect is cannot inadvertently cue the witness,” Loftus says.

Sometimes, witnesses later remember details that were not in their initial account. Research also indicates that if one detail of someone’s recollections is wrong, it doesn’t meant that the whole account is riddled with inaccuracies.

The forest before the trees

To understand how mistakes creep into our memories, researchers are probing into the details of how the brain recollects what we’ve witnessed.

Whenever you look at a scene, your brain processes the details like lines and color first, and then assigns higher-level categories, like whether you are in a dining room or a hospital room. For a long time, scientists assumed that your brain recalls visual information the same way it was encoded: details first.

But a recent experiment suggests that your brain summons general concepts first, and then fills in the minutiae later. Researchers asked people to view a single tilted line and then recreate it from memory. Later, the participants viewed two lines and then had to estimate both, one after the other. The participants’ accuracy in estimating the single tilted line’s orientation did not predict how they performed when they had to recreate two lines. People also tended to overestimate the space between the lines, but were adept at remembering which line was counterclockwise relative to the other.

Instead of recalling each line individually, the team realized, people were recalling the lines’ overall relationship—which one was angled farther from horizontal than the other—and using it to estimate the positions of both lines.

It’s easier for our brains to handle categories than minute details, says coauthor Ning Qian, a neuroscientist at Columbia University’s Zuckerman Institute. “Whether the second line is clockwise or counterclockwise from the first, that’s just one bit of information, it’s kind of like black versus white, yes versus no,” he says. “The lower-level feature, which is each line’s individual orientation…can vary continuously over 0 to 180 degrees, and that’s much harder to store.”

His findings imply that if you try to remember visual information, your brain first decides what kind of object you were looking at (such as a dining room table) before it sweats the small stuff (like how big the table was or what kind of wood it was carved from). This constrains what details can be filled in, says Qian, who reported the findings October 9 in the journal Proceedings of the National Academy of Sciences.

So what does this have to do with eyewitnesses? It’s possible that we might misremember what we’ve seen because our brains recreate the big picture first. Say you saw a man in a suit running away from the scene of a crime. When asked to describe him, your brain will first provide “man in suit”. So you might describe him as wearing a tie, because that’s what you would expect to see. But the man may not have actually worn a tie. Or you might know that you saw a man running in the street, and when you remember “man in street,” you exaggerate and recall him running right down the middle of the road instead of near the sidewalk.

In future, Qian would like to figure out a way to design experiments that would help interviewers discern which parts of a witness’s visual recollections are distorted.

Paving memory lane

So, is there anything you can do to improve your own recollections? Whether you’ve been in a car accident and know you’ll have to testify later or just want to fix a special night in your memory, there are a few ways to give yourself a better shot at recalling the experience accurately.

Recounting the events helps cement them in your memory. The fresher the recollection is, the better. Loftus recommends writing down everything you remember before you share it with anyone else, and before anyone else can give you more information about what happened.

It also helps to act out the event or make a simple drawing, Fisher says. “I would try to expand my ability to describe things by not forcing myself to use only words,” he says. “Event sketches preserve a lot of information, especially visual or spatial experiences that might be difficult to describe or remember only using a verbal approach.”

Later on, you can jog your memory by returning to the spot where the event occurred. “One thing I would try to do is put myself back in the same frame of mind, and even in the same location,” Fisher says. “I would sit by myself, I would have no other sources of distraction. I would probably close my eyes so I could concentrate more intensely.” And, of course, you can peruse cellphone messages, calendars, and other external information to remind yourself what happened.

There’s no way to make our memories completely infallible, though. “We do these demonstrations all the time where you have somebody come into a classroom and commit some outrageous act and then leave,” Loftus says. “And right away you’ll get all kinds of different versions from people about what the guy looked like and what he was wearing and what he said.”

And there’s no way to measure whether most eyewitness statements are truly accurate, either, Loftus says. It all depends on how swiftly the person was interviewed and many other conditions.

Ultimately, our minds are not tape recorders, and memory can only take us so far.