Truth, Lies and Police Lineups

Print More
lineup

Photo by Evolution Labs via Flickr

It’s often the most dramatic moment in a TV crime procedural: An eyewitness is brought into the station house to identify a crime suspect from a lineup. Tension builds as detectives await the definitive truth that will lead to a conviction.

In real life, however, things are rarely as cut and dry. In one meta study of 6,734 lineups conducted in police precincts around the U.S., nearly 40 percent of the eyewitnesses fingered someone who was added as a “filler”—that is, known to be innocent.

Another meta-analysis found that as many as 50 percent of eyewitness identifications –based on police lineups or mixing photos of suspects with “fillers” —may be inaccurate. Such misidentifications account for a large portion of the wrongful convictions overturned by the Innocence Project.

Most troubling of all: the proliferating number of such misidentifications means that countless culprits have escaped detection.

A recent research paper published in the Journal of Research in Memory and Cognition argues it’s long past time to replace this “antiquated” approach to eyewitness identification.

The authors, Neil Brewer and James Doyle, propose a “screening” method in which eyewitnesses are asked to grade the probability that one of an array of persons presented to them in a lineup matches the perpetrator of a crime.

The alternative approach replaces the “categorical traditional identification decision with a procedure whereby the witness simply rates how confident they are that each person in the lineup is the culprit,” they explain.

Brewer is an eyewitness memory researcher at the College of Education, Psychology and Social Work at Finders University in Australia. Doyle is a veteran Boston criminal defense attorney, as well as a legal affairs columnist for The Crime Report.

The authors argue that the confidence-comparison technique that Brewer played a leading role in developing should trigger a serious collaborative effort by researchers and practitioners to replace the error-prone traditional lineup.

The approach builds on the work of a number of eyewitness specialists who have examined various “markers” for judging the confidence of eyewitnesses, in the face of a number of reforms that are already in wide use around the U.S.— such as ensuring that the police who oversee the lineup are themselves unaware of the identity of the suspect, so they cannot unconsciously transmit any signals.

Despite these reforms, they note, “it is clear from research that mistaken identification of innocent suspects or fillers, and failures to recognize the culprits when in the lineup, will occur too frequently.”

Ending the “categorical” approach demanded of eyewitnesses — a “yes” or “no” – would make room for more nuanced decision-making that allows police and prosecutors to use additional information the eyewitness might have that can assist in identification, the authors write.

Using the example of an eyewitness who picks an innocent “filler” over the real culprit, they argue that there may be other signals that have obscured memory for individuals who aren’t confident of their choice to begin with.

“What if the witness—despite having picked a filler—had access to valuable information in memory that is simply not accessed within the framework of the lineup’s categorical decision? Perhaps, for example, the culprit is in the lineup and is a good match to the witness’s memory, but a filler appears to be a better match.

“…It seems plausible that the witness may have access to invaluable memorial information that could be tapped―if we could find a way of doing so.”

The alternative approach suggested in the paper involves giving the eyewitness a screen in which he or she can click on 11 “confidence buttons” when looking at an array of individuals in a lineup or in a set of photos. The buttons provide the option of rating between zero percent and 100 percent their confidence of whether the person they are looking at matches their memory of the suspect.

“The confidence ratings…provide a probabilistic guide as to whether the culprit is in the lineup,” the paper says. “And this, of course is the fundamental question being asked…because in single-suspect lineups all other lineup members are known to be innocent.”

The approach is aimed at extracting the maximum information the eyewitness holds in memory about the suspect—information that might otherwise be lost if an unconfident witness is just asked to affirm or reject the choices she has been presented with.

The authors admit that a “probability” rating is not conducive to the demands of judges and prosecutors who are seeking a clear case for conviction. But they argue that it offers a more practical approach to conducting unbiased investigations.

“The procedure offers critical information for investigators and the courts that the traditional lineup cannot provide,” they wrote.

They added that having some informative evidence against the suspect might “be more attractive than having the case weakened by the witness’ failure to make a categorical identification.”

The authors acknowledge that adapting this approach would be a challenging shift for traditional players in the justice system, but said they hoped their idea could provoke researchers and practitioners into in-depth discussions about other potential alternatives..

“Under the (current) system, witnesses, not investigators or prosecutors, have essentially shouldered the responsibility for the case proceeding or being abandoned,” the paper said.

“With the confidence ratings procedure, the witness is basically ceding that responsibility to the investigator and the prosecutor who would have to use the data to make that choice.”

While that may not be the stuff of crime dramas, the result is likely to be “fewer mistaken identifications and fewer offenders avoiding justice,” the authors wrote.

The full paper can be downloaded here.

Stephen Handelman is editor of The Crime Report.

Leave a Reply

Your email address will not be published. Required fields are marked *