Eyewitness Identification: The State of the Field and Best Practices

What about witnesses who say: "I will never forget that face"- does trauma cause an imprint on people or not? What about perpetrators who either hide parts of their faces or who rapidly shave, grow facial hair, use wigs, make up? (anecdote- during Covid, I was interested in how easily I could identify people known to me, even if most of their faces were covered- even people I had not seen in years.)

Great questions, all of which could yield long responses, but I’ll try to keep it concise. It is a common misconception that there are “flashbulb memories” that become immediately indelible due to stress or trauma. The research on stress and eyewitness memory indicates that this is not an accurate picture. Instead, eyewitnesses can feel particularly confident (i.e., overconfident) in these memories, but stress and trauma are much more likely to harm than to help memory. This question speaks to two somewhat separate issues: disguise and appearance change. Fortunately there is eyewitness research on both issues. Regarding disguise, of course it is more challenging for eyewitnesses to remember a part of the face compared to the whole face, and it begs the question of how police should construct the lineup. However, there is research to show that, for example, if the perpetrator wore a hoodie covering part of his face, police should have everyone in the lineup do the same. As for appearance change, research shows that it can harm eyewitness accuracy quite severely, in at least three ways. First, if the suspect in a lineup is guilty but no longer matches the original description perfectly (e.g., he shaved his beard, or grew a beard), participants will have a difficult time correctly identifying him. Second, if the suspect in the lineup is innocent but happens to resemble the perpetrator in some ways (e.g., perp described as having a beard and the innocent guy happens to have a similar beard), he can certainly get falsely identified, which makes it particularly critical that police make sure that everyone in the lineup matches the perp’s description (e.g., they all have a beard). Third, eyewitnesses making such a false ID can still be confident that the innocent man is the perpetrator, violating the confidence-accuracy relationship. This situation of course is doubly problematic: not only has an innocent person been identified, but the eyewitness can be confident he is the perpetrator, due largely to that feature that matches their memory (e.g., the beard). As for police driving the eyewitness around the neighborhood of the crime scene, looking for the perpetrator, that is not as problematic as a showup, but it is still problematic. Think of a lineup or photo array as a recognition memory test akin to a multiple choice test on a student exam (though with the option to reject all items). You want this test to be a fair way of testing their memory, so that if they do choose one of the options (i.e., identifying a suspect) quickly and with high confidence, you can also be fairly confident that the option they chose is correct (i.e., he is guilty). Driving around looking for the perp is akin to the student scanning the textbook for the right answer. This works OK if they know exactly what the correct answer is, but this would be like an eyewitness having a perfect memory for the perpetrator (not just what he was wearing, as innocent individuals in the neighborhood could certainly be wearing similar clothing, and/or the real perp could have shed that clothing right away). As it is unlikely that the eyewitness has a perfect memory for the perpetrator, it is best to set up the strongest and fairest memory test in the form of a fair properly administered photo array, not driving around looking for the perp. Moreover, there could be interference created by looking at several individuals in the neighborhood, akin to looking through a mugbook or online, and research shows that these methods can contaminate memory.