It wouldn't surprise me if confessions are so convincing that even some of the police who push for false confessions end up believing that the confessions are true.
False confessions seem so illogical, especially for someone like Joseph Dick of the Norfolk Four, who got a double life sentence after confessing. Why do people confess to crimes they didn’t commit? Some do it for the chance at fame (more than 200 people confessed to kidnapping Charles Lindbergh’s baby), but many more do it for reasons that are far more puzzling to the average person. In the November 2004 issue of Psychological Science in the Public Interest, APS Fellow Saul Kassin looked at the body of research and described how the police are able to interrogate suspects until they confess to a crime they didn’t commit.
Generally, it starts because people give up their Miranda rights. In fact, Richard A. Leo found that a majority of people give up the right to remain silent and the right to an attorney. In fact, according to self-report data, innocent suspects gave up their rights more often than guilty suspects (most told Leo either that this was because they felt that they didn’t have anything to hide because they were innocent or that they thought it would make them look guilty).
Once a suspect starts talking, the police can use a variety of techniques to make the accused feel as though they are better off confessing than continuing to deny (these include promises of leniency and threats of harsher interrogation or sentences). If a suspect feels like a conviction is inevitable not matter what he or she says, confessing may seem like a good idea.
But, in some cases, the accused comes to believe that he or she actually did commit the crime. It’s been shown repeatedly that memory is quite malleable and unreliable. Elizabeth Loftus has repeatedly shown that the human brain can create memories out of thin air with some prompting. In a famous series of experiments, Loftus, APS Past President, was able to help people create memories for events that never happened in their lives simply through prompting. She helped them “remember” being lost in a shopping mall when they were children, and the longer the experiment went on, the more details they “remembered.” The longer police interrogate a suspect, emphatic about his guilt and peppering their interrogation with details of the crime, the more likely a suspect is to become convinced himself.
I've never heard of any research on whether some people have more stable memories than others, but I bet there's a large amount of variation.
The results show that confessions can have a powerful effect on other evidence. Of the people who had identified a subject from the original lineup, 60 percent changed their identification when told that someone else had confessed. Plus, 44 percent of the people who originally determined that none of the suspects in the lineup committed the crime changed their mind when told that someone had confessed (and 50 percent changed when told that a specific person had confessed). When asked about their decision, “about half of the people seemed to say, ‘Well, the investigator told me there was a confession, so that must be true.’ So they were just believing the investigator,” Hasel said. “But the other half really seemed to be changing their memory. So that memory can never really be regained once it’s been tainted.” What’s more, people who were told that the person they wrongly pinpointed as the culprit had confessed saw their confidence levels soar. After that confirmation, they remembered the crime better and were more sure about details. The implications for inside the courtroom are obvious if eyewitnesses who incorrectly picked someone out of a lineup can become so sure of their choice after learning that the person confessed. “It is noteworthy that whereas physical evidence is immutable (once collected and preserved, it can always be retested), an eyewitness’s identification decision cannot later be revisited without contamination,” Kassin and Hasel write.
Link thanks to Less Wrong.