Martin Conway: There have been some studies of what is sometimes called flashbulb memories. These are memories for major public events, but they’re curious memories because they’re memories about where you were and what you were doing and who you were with when you first heard about the public event.
The classic one being the assassination of the American President John F Kennedy, but more recently people have looked at memory for the space shuttle disaster, testing people’s memories within a few days of that being shown on television, and then testing their memories at delays of two, three and even four years. The most recent event that this has been done for is the people’s memories for learning about 9/11. The main finding, here’s really quite an interesting one, and that is that there is a fairly rapid decline in the information you can recall. But after a few months that stabilises and what’s left seems to stay there for a very very long time.
Now if we extrapolated from those sorts of findings to, for example, someone witnessing a crime, we might expect exactly the same sort of pattern; that is that immediately after the crime you’re starting to forget details of it. But after a period of weeks and perhaps months you’re going to reach a point where what’s left stays there.
Okay, and so now you’re in a curious situation, you’ve got the fragments of the event that you can remember quite well and then bits which you’ve forgotten, and so when you’re remembering the event perhaps you unconsciously, or even consciously, infer what must have been the case, and so now you can see how a person might start to build up in their account a memory that’s partially true and partially false.
And they may not even know it’s partially false, and it could be completely wrong in many respects because you’ve forgotten them now you’ve inferred them, and if you keep inferring them perhaps over a series of interviews with for example the police, then you might start to have your inferences become integrated in your memory of giving an account of the event and now you've got something which is seriously misleading as a memory.