Wired reports on a fascinating new study showing how eyewitness memory can be influenced by video. Participants in the study were paired with a partner (who was actually part of the research team) to play a gambling-based computer game. The participants bet money on their own ability to answer multiple choice questions. The game relied on participants to honestly pay back money they'd earned when they got a question wrong.
After they had finished with the questions, the participants were told that their partner had cheated, even though the partner hadn't. One group of students were then shown a video that had been digitally altered to make the partner look as if he had actually cheated. Though they were told to only report their partner if they were 100 percent sure he had cheated, and that the partner would be punished based on their decision, about half the participants who viewed the video still reported their partner for cheating. The second group wassn't shown the video, but only told of its existence. Just 10 percent of them still reported their partner for cheating.
The article may overstate the study's lessons a bit, at least in their relevance to the criminal justice world. It seems unlikely that many criminal cases are tainted by video digitally manipulated to show obvious guilt—even if you're cynical enough to believe law enforcement officials would try it, I'd think it would be pretty easy for someone with expertise in video editing to detect. But the broader point—that eyewitness memory is highly susceptible to suggestion—is worth heeding, and has been confirmed in numerous other studies.