Reliability of Memory

We have now studied two models explaining how information can be processed in short-term memory, eventually becoming lodged in long-term memory.  But once information becomes stored in long-term memory, is it always reliable?  Are our memories accurate records of the past, or can they become contaminated?  In this lesson, we'll be exploring the fascinating topic of false and distorted memories.
Think Critically

In the United States (and many other countries), a criminal trial is decided by jury.  Twelve men and women are chosen from the community, and these twelve individuals must decide whether the accused criminal is innocent or guilty.

Imagine you are one member of a jury.  You hear the testimony of a young woman, who was brutally raped by a man who broke into her house in the middle of the night.  "Do you see the man who raped you?", asks the lawyer.  "Yes, I do" says the woman.  "It is the defendant, the man sitting over there."  The lawyer probes the witness further.  "Are you absolutely certain it is that man?" the lawyer asks.  "Yes," the woman replies.  "While he was raping me, I studied his face carefully.  I wanted to be certain that if I lived, he would pay for his crime."

After hearing this testimony, would you vote for the defendant being "guilty" or "not guilty"?  Why?
Reconstructive Memory

A large number of criminal trials are decided based on eyewitness testimony.  When a victim points to the defendant and says, with absolute certainty, "This is the man who raped me", most juries tend to be convinced.  This suggests that we tend to trust people's memories.  We might regard memory as being much like a camera or tape recorder, which accurately stores a record of the past.  We imagine that a victim can mentally rewind to the moment of the crime, and form an accurate memory of the person who attacked her.

In the last few decades, however, research has challenged the idea that memory works anything like a tape recorder.  Most psychologists now agree that memory is reconstructive.  In other words, we must consciously rebuild our memories every time we try to remember something.  And when we try to rebuild our memories, we may be influenced by other factors - such as our beliefs or feelings about the past, or unrelated memories that get misplaced and mixed together.

Can you remember the last conversation you had on the telephone?  You probably remember who you spoke to, the main topic of conversation, perhaps a few memorable phrases that you or the other person spoke.  But it is highly unlikely that you remember, word for word, the entire conversation.  Memory doesn't work much like a video camera at all.  Rather, we only store memory traces of the past - brief fragments of memory, rather than an entire, complete record.

An analogy might be useful here.  Imagine you are a scientist, on the verge of discovering a new species of dinosaur.  You have discovered a skeleton of this dinosaur buried in the ground.  But the bones, of course, are not all in their proper places.  Rather, the bones are haphazardly scattered all over, some bones may be missing, and the bones of other animals may be mixed in.  When trying to reconstruct a complete skeleton of this dinosaur, there is a good chance that mistakes will be made.  You might arrange the bones in the wrong order, or mistakenly fill in the missing gaps with the bones of another animal.  This is quite similar to the process of reconstructive memory.  When we try to remember something, we must rebuild the past out of memory traces, much like a scientist rebuilds a skeleton out of bones.  And since our memories are rarely complete, we might fill in the missing gaps in our memory with the wrong information.

We have already seen how schemas can lead to false or distorted memories, which supports the reconstructive view of memory.  To recap:

  • Information that is consistent with your schemas will be remembered well, but information which is inconsistent with your schemas may be forgotten or distorted to "fit" your schemas

  • When you can't remember all the details of an event, you may "fill in" the missing details with your best guess, based on your expectations and beliefs about the past

Video Activity

  • Watch the video below, following the instructions
  • Can you explain the findings discussed in the video?

  • How can schemas explain the findings discussed in the video?

Memory & biased questioning

Another reason why memories can become contaminated is biased questioning.  Since our memories aren't perfect, we often fill in missing information with our "best guess".  And our "best guess" can be influenced by a multitude of factors, including the expectations and suggestions of people around us.

For instance, it is common for police to conduct a photo lineup in order to confirm the identity of a crime suspect.  During a photo lineup, a witness will be shown a series of photographs, and asked to identify the criminal.  An example of a photo lineup is shown below.  Can you think of a reason why a photo lineup might lead to distortions in memory?  Remember that the witness might only have seen the criminal once!

A significant problem with photo lineups is that the witness may assume that one of the men in the lineup must be the criminal.  In such circumstances, the witness may choose the photograph that most closely resembles the actual criminal, believing that this must be the person who committed the crime.  The problem is that the actual criminal might not be in the photo lineup at all.  Perhaps the police suspected an innocent person who closely resembles the criminal, and the witness may confirm the police's mistaken suspicion by choosing that person in the photo lineup.  Then, the witness's memory of the man in the lineup might get mixed in with the memory of the man who actually committed the crime, until they are one and the same.

Photo lineups are an example of forced choice questions - questions in which there are a fixed set of possible answers, and it is implied that one of those possible answers is the correct one.  If a parent asks a child, "Do you like eating spinach, or do you love eating spinach?", then the forced nature of the question becomes obvious.  But photo lineups are a less obvious example of a forced choice question, which makes them even more dangerous.  A witness may easily assume that one of the photos in the lineup must be of the criminal, and make a choice based on that false assumption.  

Another example of biased questions is leading questions.  ​For an example of leading questions, watch the video below, which is based on a famous research study by Loftus and Palmer.

Research Study: Loftus and Palmer

Aim:  Investigate how leading questions can influence eyewitness memory


  • The study was carried out on American students, who were shown a video of a car crash

  • The students were randomly divided into groups, and each group was asked a slightly different question regarding the speed of the cars at the time of the accident.  One group was asked "How fast were the cars going when they smashed each other?", while for other groups, the word smashed was replaced by either hitcollided, or bumped

  • In a follow-up experiment, participants were also asked if they had seen broken glass


  • Participants estimate that the cars were travelling at a significantly higher speed when the question involved the word smashed. 

  • Significantly more participants also reported seeing broken glass when the question involved the word smashed  (in reality, there was none)


  • Leading questions can alter memory of an event

  • This may be because the word "smashed" is associated with severe accidents, suggesting higher speeds and broken glass


  • A well-controlled laboratory experiment, demonstrating a causal relationship between the independent variable (the verb in the leading question) and the dependent variables (speed estimate and whether broken glass was reported)

  • Participants were all American university students, so the findings may not apply to other cultures or age groups.  In particular, the students who took part in this experiment may have not had much driving experience. Perhaps more experienced drivers would be less susceptible to leading questions regarding car speed

  • This experiment took place in a laboratory, and the speed estimates carried no real-life significance.  Therefore, participants may not have been particularly motivated to be accurate.  It is possible that someone who had seen an accident in real life would take greater care to report an accurate accident speed to the police

IB Psych Matters

The Innocence Project is an organization dedicated to helping people wrongfully convicted for crimes which they did not commit.  The project has led to the release of over 350 innocent people, whose innocence was established by DNA evidence.  In over 75% of those cases, it was faulty eyewitness testimony that sent the wrong person to jail.  Time and time again, a witness points an accusing finger at the wrong person.

  • Read this article about a famous case of witness misidentification.  Why do you think the witness, Jeniffer, misidentified her rapist?  Think of the role of forced choice or leading questions

  • Imagine you are a psychologist asked to testify as a memory witness in a criminal trial.  Prepare a short speech in which you will explain to the jury the extent to which eyewitness testimony is reliable

  • I can explain why memory is a reconstructive process

  • I can explain how schemas can lead to false or distorted memories

  • I can describe the role of forced choice and leading questions in creating false memories

  • I can describe the Aim, Procedure, Findings, and Conclusion of the research study carried out by Loftus & Palmer, and evaluate the study
Quiz Yourself!

​1.  "Every time you remember something, you must rebuild your memory of the past".  Which view of memory is this most related to?

(a) Schema theory

(b) Multi-store model

(c) Reconstructive memory

(d) Memory traces

2.  When hearing a list of words like bed, snore, drowsy, and snooze, many people also (falsely) recall hearing the word sleep.  Which view of memory best explains this?

(a) Schema theory

(b) Multi-store model

(c) Reconstructive memory

(d) Memory traces

3.  A witness is shown a photo lineup of 10 men, and asked to identify which man stole her wallet.  What kind of question is this, and why?

(a) Leading question, because it assumes that her wallet was stolen

(b) Forced choice question, because it assumes that her wallet was stolen

(c) Leading question, because it assumes one of the 10 men is the actual criminal

(d) Forced choice question, because it assumes that one of the 10 men is the actual criminal

4.  Some have criticized Loftus and Palmer's study for having low ecological validity. Why?

(a) Participants knew they were in an experiment, so motivation to be accurate was low

(b) Participants were college students, and so might not have had much driving experience

(c) Participants were all Americans, so results might not reflect other cultures

(d) Participants did not know whether the car crash was real or staged

5.  What assumption is evident in the work performed by the Innocence Project?

(a) Over 75% of eyewitness testimony is false

(b) DNA evidence is more reliable than eyewitness testimony in determining guilt

(c) Witnesses often lie while they are giving testimony

(d) Every wrongfully convicted criminal can be exonerated by DNA evidence

​1 - C, 2- A, 3 - D, 4 - A, 5 - B