As detectives and the viewers of crime dramas know, eyewitnesses are notoriously unreliable. This is not necessarily intentional on their part, but a result of 3 things: conditions of poor visibility, human inaccuracy at facial identification, and biases introduced into identification procedures like line-ups and photo-identifications. Visual quality also decreases with distance. Mistaken identification, confirmed by DNA testing, is responsible for a large percentage of overturned convictions. University of Washington psychologist Elizabeth Loftus calls human memory "a living thing that changes shape, expands, shrinks and expands again." But expert witness Marc Green points out, "Although eyewitness identification is highly fallible, it still carries great weight with jurors." Eyewitness expert Paul Michel concurs, "An eyewitness's testimony can be very persuasive. The witness can have an emotional connection to the jury that physical evidence can't possess. But human vision is fallible. An eyewitness testimony should be evaluated to ascertain if the witness could have seen what they claim to have seen."
Police artists base their sketches of suspects on fragmentary memories that may admittedly miss the mark. The gap between what an eyewitness sees and what he or she remembers seeing was narrowed with computer programs like Identi-Kit that put together a composite image based on remembered facial features. But now researchers have developed new software that helps witnesses recreate and recognize suspects using principles borrowed from the fields of optics and genetics. The EFIT-V system was developed at the University of Kent by Christopher Solomon (photo above next to a composite sketch generated by the EFIT-V system). The program generates its own faces that progressively evolve to match the witness’ memories. The witness starts with a general description such as "I remember a young white male with dark hair." Nine different computer-generated faces that roughly fit the description are generated, and the witness identifies the best and worst matches. The software uses the best fit as a template to automatically generate nine new faces with slightly tweaked features, based on what it learned from the rejected faces. As the features change, the witness' selections guide the evolution of the face. By tapping into recognition instead of recall, the EFIT-V system proves very effective even when witnesses say they can't describe a person, and has led to twice as many identifications of suspects as traditional methods. A very similar program called FacePrints has been developed and patented by Victor S. Johnston of New Mexico State University. As screen after screen of revised faces appears, the witness can keep features that seem correct by locking them in, explains graduate student Craig Caldwell. Within 10 generations of 20 faces, he says, FacePrints can match or beat any other composite method.
But whether a witnesses memories are fed through a sketch artist directly or the latest technology, here are some tips to being a better eyewitness:
- Look at the whole face and note how the features relate to each other.
- Make a mental note about other people the suspect may resemble, an immediate comparison that may make a face easier to recall later.
- Remember anything that makes the suspect unique, such as a scar or tattoo.
- Rather than estimating, remember height in relation to a doorway or sign.
- Avoid the instinct to focus on the victim in favor of getting a good look at the criminal.
- Resist the tendency to look for expected details, such as those publicized in media reports.