Source: Matthew Sharps
In previous posts of the Forensic View and elsewhere (e.g., Sharps, 2022, 2024), we have seen that eyewitness memories of crimes and suspects can be altered after the fact as time goes by and as witnesses encounter new information. Many people are not aware of the malleability of the memory process; this lack of awareness can lead to the idea that it doesn’t really matter if a given witness sees repeated depictions of the suspect in investigative procedures. People may believe that if a witness has identified a given suspect, then no further exposure to that suspect will change the given memory.
They couldn’t be more wrong.
A witness to a given crime may see several depictions of suspects putatively involved in that crime, or even the suspects themselves, as the relevant investigation proceeds. I recall a case in which a suspect in a particular crime was inadvertently brought into a police station at exactly the wrong moment; he was observed in custody by an eyewitness to the crime in question. In a horrible coincidence, at the very moment of encountering the suspect, the witness was in the process of attempting to make an identification from a “mug book,” a catalog of pictures of possible suspects.
Not surprisingly, the witness made a positive identification of the suspect under arrest.
In many trials, a witness is asked to identify the defendant seated beside the defense attorney in the courtroom; such dramatic identifications may appear quite compelling to jurors. Yet the witness may have seen one or more pictures of the defendant in a lineup prior to making the courtroom identification. How can we be certain that witnesses are making such identifications from the crime scene itself, or merely from a previous investigative test of their eyewitness memory?
According to important work by Wixted, Wells, Loftus, and Garrett (2021), we can’t.
Eyewitness memory of a human face is classified as recognition memory, essentially the simplest type of visual memory. The reason for this relative simplicity has to do with the mental effort involved. If you have to remember something from scratch, in free recall with no relevant external memory support, the mental effort involved can be formidable, and the resultant memory report may not be very accurate. A less effortful type of memory lies in cued recall, in which we are provided with cues constituting partial support for the memory to be recalled, perhaps something like, “Did you see the person wearing the blue shirt?” or “Who in the group was wearing gloves like these?” Cued recall is generally easier than free recall because there is at least some external support, the relevant cues, for the recall process.
But recognition memory for a face is even easier. We saw the face at the crime scene, and we’re seeing it again in the retrieval phase of the task as law enforcement officers or attorneys ask us about specific pictures in, for example, a lineup. We have virtually complete external support for our memory of the given face.
But wait a minute: If recognition memory is the simplest kind of memory, doesn’t that mean that eyewitness identification is the easiest type of task?
Absolutely not. As Wixted et al. explain, a lineup yields memory traces of all the faces observed in the procedure, including, of course, that of the suspect. So, that trace would be expected to be stronger on any later eyewitness procedure involving the same witness, whether the suspect was guilty or not.
And, since recognition memory is less mentally effortful, we might expect that later trace to be readily recoverable. Furthermore, the more opportunities the witness has to see the suspect’s face, in real life or pictures, the more we would expect to see this memory strengthened, whether that memory actually came from the crime scene or from a subsequent lineup.
But wait. Given that eyewitness memory is frequently poor, why would we expect the additional eyewitness traces caused by the memory tests themselves to contribute to false identifications? Wouldn’t these new traces be poor as well?
Not necessarily. At actual crime scenes, there are many factors influencing eyewitness processing, and very few of them operate for the better. Relative darkness, speed of action resulting in diminished encoding times, heightened arousal resulting from fear or anger, dissociative processes—any or all can result in diminished memory (e.g., Sharps, 2022, 2024).
However, if you’re looking at lineup pictures in relative comfort, with good lighting and plenty of time, your memory processes under these conditions may be a lot better than we might expect. This has been known for half a century (e.g., Nelson, Metzler, and Reed, 1974), but given the time interval between then and now, it might be worth resurrecting these venerable findings here. Among the important results of this study was the fact that respondents yielded better than 85 percent correct recognition of photographs and maintained this high recognition rate for seven weeks!
Context matters; this was demonstrated in other research (e.g., Loftus and Bell, 1975), which resulted in quite inferior picture memory performance making use of very brief exposure times of one second or less. The Nelson et al. study made use of 10-second exposure times, much more representative of the “take all the time you need” conditions of a typical law enforcement identification procedure. Clearly, the conditions of observation make a huge difference in memory retention.
Given all of this, we might expect that the memory traces derived from a lineup might be appreciably stronger, at least in many cases and in purely cognitive terms, than those derived from the typically less optimal realm of the original crime scene.
The findings of Wixstead et al. are extremely important in criminal justice. We are less likely to convict the innocent, and more likely to convict the guilty, if the memory of a given eyewitness is only tested once.
