Memory Reconstruction: Why Eyewitness Accounts Are Unreliable

Since the 1990s, when DNA testing was first introduced into the courtroom, researchers have reported that 73% of the 239 convictions overturned through DNA testing were solely based on mistaken eyewitness testimony.

How could so many eyewitnesses be wrong?

First, let’s talk about the concept of memory.

Many people think that human memory is like watching a recorded video– the mind “records” events in real-time, and then is able to replay the exact same video on demand whenever we want to remember that event.

This is probably because we tend to experience our memories as small movie clips that are played in our minds.

However, memory is not played back, but rather reconstructed when we recall them: the act of remembering, is more akin to putting puzzle pieces together than retrieving a video recording, says memory researcher and psychologist Elizabeth F. Loftus (who specializes in studying eyewitness testimonies and whether language can affect memory).

Memories are not stored in entirety; they are neural pathways that re-fire every time we attempt to recall a memory, and they are susceptible to change.

In a courtroom setting, even procedural questioning by a lawyer can alter the witness’s testimony, because fragments of the memory may unknowingly be combined with information provided by the questioner, leading to inaccurate recall. The power of suggestion!

What affects memory recall?

Gaps in memory: memory gaps can be subconsciously filled with “made-up” sequences of events over time to make the memory coherent and make sense as a whole; these false sequences will later seem as real to you as the original event.

You can’t remember who else was at the party, but your girlfriend Cindy was usually present at these events with you– so over time your memory of the event might include Cindy being present, even though she really wasn’t.

Subsequent events: events that occur after an event can change the memory of the original event.

For example, let’s say that you and your close friend were at a party.  At the party, you have a huge argument which results in a falling-out that lasts for years.  Your memory of the original event, the party, may include your friend being rude and cold to you– even if it is not true.  The subsequent event, the argument, altered the memory of the original.

Diction – In her research, Dr. Loftus would show participants a video clip of an automobile accident. She then asked a series of questions about the accident, changing the way she worded the questions for different groups.

For example, some questions would be alternatively phrased as:

“How fast would you estimate the car was going when it smashed into the other vehicle?” or

“How fast would you estimate the car was going when it hit the other vehicle?”

The first question is known as a leading question (one that suggests or leads to the desired answer).

Then, she proceeded to ask participants in the study the estimated speed of the vehicle at the time of collision and if they remembered seeing broken glass in the video clip.

As you might have guessed, when Dr. Loftus used the word ‘smashed‘, the estimated speed was reported as higher than when she used the word hit.  Three times as many participants ‘remembered’ seeing broken glass if the word ‘smashed’ was used rather than the word ‘hit’.

(There was no broken glass in the video.)

One week after the initial study, a follow-up questionnaire was given to participants with the question “did you see broken glass? Y/N”.

More than twice as many participants in the “smashed” diction group reported seeing broken glass a week before than the “hit” diction group.

In another study conducted by Dr. Loftus, false memory construction was tested.

Subjects were given written accounts of four events, three of which they had actually experienced and the fourth being completely false; this event centered on the participant being lost in a public place when he or she was younger.  An assisting relative provided realistic details for the false story, such as a description of the location at which the subject was lost.

After reading each story, subjects were asked to write down what they remembered/did not remember about the incident (or to indicate that they did not remember it at all).

Amazingly enough, one-third of the participants reported fully remembering the false event. In two follow-up interviews, 25% of the subjects still claimed that they remembered the untrue story, a figure consistent with the findings of similar studies.

This research suggests that memory is easily compromised and distorted by questioning and external information acquired after the event merging with original memories to cause inaccurate recall and/or reconstructed memory.

The results from the second study suggest that this effect is not just due to a response-bias, because leading questions actually altered the memory a participant had for the event completely.

Addition of false details to a memory of an event is called confabulation.  This has significant implications for the questions used in police interviews of eyewitnesses following crimes.

Not only is memory recall unreliable, but so is memory formation: from just 10 feet away, a witness is not able to see a person’s eyelashes. At 200 feet, the eyes themselves are a blur– and it is very likely that no facial features can be clearly distinguished.

This may seem logical, but criminal cases have been decided based on these highly unreliable and subjective testimonies from eyewitnesses who were more than 400 feet away from the perpetrator.

(Stress and other confounding variables can also affect memory formation and recall.)

Significance? Since memories are reconstructed when being recalled, here are some things to keep in mind:

• Diction is crucial! Word choice can influence and affect people’s memories– including your own.

• Self-reports of past behavior are unreliable. No one, including yourself, will not accurately remember what others said and did.

• Similarly, take what individuals say after an event with a grain of salt. It’s being reconstructed and is highly susceptible to change.

Eyewitness Identification: Issues in Common Knowledge and Generalization. Gary L. Wells and Lisa E. Hasel in Beyond Common Sense: Psychological Science in the Courtroom. Edited by E. Borgida and S. T. Fiske. Wiley-Blackwell, 2007.

Elizabeth F. Loftus and John C. Palmer, Reconstruction of Automobile Destruction:
An Example of the Interaction Between Language and Memory, Journal of Verbal Learning and Verbal Behavior, 13, 585-589 (1974).

Eyewitness Testimony: Civil and Criminal. Fourth edition. Elizabeth F. Loftus. LexisNexis, 2007.

Psychological Science in the Courtroom: Consensus and Controversy. Edited by Jennifer L. Skeem, Kevin S. Douglas and Scott O. Lilienfeld. Guilford Press, 2009.