In 1974, Elizabeth Loftus and John Palmer from the University of Washington conducted two experiments  investigating the effects of the words used in a question about an event on the memory the participants had of this event.
In the first experiment, the participants were shown several car crashes and were afterwards asked to estimate at which speed the cars collided. Previous studies indicated that humans are surprisingly bad at estimating the speed of cars. After watching films of cars colliding, the participants were given a questionnaire concerning the accident. One question was “About how fast were the cars going when they X?” X was the variable in this case. The participants were divided in five groups, all having a different ‘X’ in their question: (1) contacted, (2) hit, (3) bumped, (4) collided and (5) smashed. The hypothesis was that the choice of verb can have an influence on the speed estimated.
Their results indicate that this was the case; the choice of word had a statistically significant effect on the speed estimate. The largest difference was found for ‘contacted’ vs. ‘smashed’ (about 9mph). While these results confirm their hypothesis, it is unclear whether this was merely due to response bias or due to a real modification of the memory the participants had of the accidents.
Therefore, a second study was conducted: Participants watched a film of a multiple car accident and filled out a questionnaire, again including a speed estimate, afterwards. This time, the participants were divided in three groups: (1) One group being asked about the speed during the accident using the verb ‘smashed’, (2) one group being asked about the speed during the accident using the verb ‘hit’ and (3) a control group not being asked about the speed during the accident at all. The speed estimates for ‘smashed’ vs. ‘hit’ were again significantly different, as expected on the basis of experiment one.
One week later, participants were asked to fill out another questionnaire including several questions concerning the accident. One of these questions was: “Did you see any broken glass?” There was no broken glass on the film, but the hypothesis was that the ‘smashed’ group might report having seen broken glass because they estimated the accident to occur at a higher speed – making the involvement of broken glass more likely. Their results confirmed this: People in the ‘smashed’ group were significantly more likely to report having seen broken glass than people in the ‘hit’ and ‘control’ groups.
This suggests that a memory of an event can be modified by things as simple as different wordings in questions about the event. While this particular study suggests high flexibility of memory, it is difficult to draw conclusions from it for real life events because it occurred in a laboratory study, including only films of accidents. As noted here , there was a study concerning eyewitness testimony showing relatively constant and accurate memory of an emotional event over time. This casts doubt on whether memory is as flexible as the Loftus and Palmer experiments suggested if important real life events are concerned.
This blog post is part of the series “Interesting psychology experiments”.
Interesting psychology experiments 1: The marshmallow experiment
Interesting psychology experiments 3: Contagious yawning works with dogs
 Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of auto-mobile destruction: An example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behavior, 13, 585-589
 McLeod, S. A. (2010). Loftus and Palmer