Malcolm Gladwell, the author of The Tipping Point and Blink, makes a good point about the identifications by the accuser in the Duke lacrosse case:
[T]he Duke case is an example of another, even more problematic aspect of eyewitness identifications, and that is that we aren’t particular good at making them across races. There is a huge amount of psychological research in their area, pioneered by Roy Malpass at the University of Texas at El Paso. A few years ago, John Brigham and Christian Meissner did a big meta-analysis of all of the cross-racial identification studies and concluded that given the task of picking someone out of a lineup, the average person is something like 1.4 times more likely to correctly identify an own-race face than a different-race face, and 1.6 times more likely to incorrectly identify a different race face. These are not trivial error rates. Clearly we need to treat cross-racial identifications with a special level of caution. (Here’s the link to the UTEP eyewitness laboratory: eyewitness.utep.edu/race.html)
The problem seems to be that when we encounter someone from a different group we process them at the group level. We code the face in our memory under the category black or white, and not under the category of someone with, say, an oval face and brown eyes and a prominent chin. Race, in other words, trumps other visual features that would be more helpful in distinguishing one person from another. Why do we do this? One idea is simply that it’s a result of lack of familiarity: that the more we “know” a racial type, the more sophisticated our encoding becomes. Another idea is that it’s a manifestation of in-group/out-group bias. The thing about coding by group and not by facial feature is that it’s a lot faster. And from an evolutionary standpoint, you’d want to use quicker processing methodologies in dealing with those who come from unfamiliar—and potentially unfriendly—groups. The bottom line is that the adage that “all blacks look the same” to whites (and all whites look the same blacks) has some real foundation.
This has been a huge issue for years in white identifications of black suspects. I would venture to guess that there are thousands of African Americans in prison right now for crimes they didn’t commit, largely because whites have far too much faith in their ability to tell one black face from another. Now, in the Duke case, we have a black identification of white suspects. The shoe is on the other foot. It will be interesting to see whether the legal system is any more willing to acknowledge the real limitations of eye-witness identifications when it is suspects from the racial majority who are on the receiving end of the bias, not the other way around.