News Release

NSF awards $670,000 to use virtual reality to evaluate eyewitness accuracy

The study will examine how physiological measurements and eye tracking interact with accuracy and confidence when witnessing crimes in a virtual environment

Grant and Award Announcement

University of Arkansas

James Lampinen

image: James Lampinen view more 

Credit: University Relations, University of Arkansas

The National Science Foundation awarded just under $670,000 to a team of researchers, including one at the U of A, to study the relationship between eyewitness confidence and accuracy across a range of variables using virtual reality.

Distinguished Professor of psychological sciences at the U of A James Lampinen will serve as principal investigator for the three-year award. Assistant professor of psychology at Oklahoma State University Kara Moore will serve as co-PI.

“Eyewitness identification is crucial evidence in thousands of criminal cases per year,” Lampinen explained. “For years now, I have been working with law enforcement on optimizing the procedures that can be used to collect eyewitness identification evidence. The research funded by the National Science Foundation will provide an exciting opportunity to better understand when identifications are likely to be accurate or inaccurate. Ultimately, that will lead to a more just criminal justice system.”

The primary purpose of the three-year award is to probe the pristine conditions hypothesis, which suggests that high-confidence witness identifications will be "remarkably accurate" when identification procedures are optimal. Optimal procedures would include things like ensuring people in police lineups are plausible alternatives to the suspect; using double blind administration, in which neither the administrator nor the eyewitness know who the suspect is; and receiving unbiased lineup instructions, such as letting witnesses know that there’s no obligation to choose a suspect in any given lineup.

The pristine condition hypothesis further suggests that so-called “estimator variables,” which include things like the presence of a weapon, quality of viewing conditions and distance the eyewitness is from the suspect, are not of significant importance when confidence is high, even when estimator variables are suboptimal.

Ultimately, researchers are interested in determining if there are exceptions to the hypothesis, and under what circumstances those exceptions are likely to be present.

Another area of interest is the degree of a witness’ “meta-cognition knowledge.” This is the idea that once witnesses are aware of how estimator variables influence the accuracy of eyewitness identifications – theoretically, the more suboptimal the variable, the lower the accuracy of identification - they will take that into consideration when evaluating their own confidence level – potentially revising it downward.

To evaluate these estimator variables, investigators will use virtual reality equipment to simulate different scenarios, including varying the distance and viewing time of a scripted incident, as well as the ages and races of the actors involved, incorporating cross-race and same-race situations.

Why virtual reality? According to Lampinen, because it “can mimic emotional experiences similar to those encountered in real environments.” Essentially, it provides a more immersive experience. VR can also be used to track eye movements and measure physiological responses (perhaps triggered by the presence of a weapon) to see what may be happening on an unconscious level.

The award provides for the purchase of a high-definition virtual reality camera that can record in 360 degree, 3-D video, a virtual reality developer, employment of theater students to act in scripted videos, graduate students and a post-doctoral researcher, among other things.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.