We spend a considerable amount of time during the certified investigators' training program teaching procedures that will avoid the contamination of evidence. In other words, what can we do to prevent the possibility that the evidence might change subsequent to the incident itself? Some of that change might occur by natural causes: i.e., liquids evaporate; memories fade.

However, some changes are man-made. To say they're man-made does not mean that they are malicious. In this context we often talk about the need to conduct incident interviews as quickly as possible. The longer it takes to discuss the incident systematically with witnesses, the more likely that conversation with others subsequent to the event might change a witness's memory. (Of course, this is a different phenomenon from the collaboration of witnesses who harbor a malicious intent to tell a false story.)

The authors of this article address the question of whether post event discussion can change memory. They placed 40 subjects into pairs. To one person in each of the pairs they showed a video tape of a crime in which the suspect has an accomplice. The second person in each pair saw virtually the same video, the only difference being that in this latter video there was no accomplice. When each person was asked soon after the viewing whether there was an accomplice, 39 of the 40 subjects answered the question correctly. In other words, in 19 of the 20 pairs, both witnesses remembered the existence or non-existence of the accomplice correctly.

At this point the researchers instructed each of the 19 pairs to discuss what they had viewed. Remember, each person in each of these pairs had initially remembered the video correctly at the start of their discussion. In other words, they would have initially disagreed about the existence of the accomplice. However, at the end of their discussion, 15 of the 19 groups had come to an agreement: one of the witnesses in each of these groups had changed what had been a correct memory for an incorrect one. According to the authors:

For eight of these pairs, no accomplice was reported, and for seven of the pairs an accomplice was reported. Thus, there was no general tendency for the conformity to be towards one direction or the other.

This is only one such study; however, its results are quite powerful. In 15 of 19 cases, two individuals discussing their memories of an event caused one of the pair to remember incorrectly what he or she had only recently remember correctly. The results remind us how important it is to begin an investigation immediately and continue without unnecessary delay.

What is particularly sobering about these findings is that the witnesses were not seeking to concoct a story.