Faculty Mentor: Kara Moore (Psychology, University of Utah)

Introduction

Eyewitness memory is a fascinating subject due to the psychological and legal implications it holds. Eyewitness accounts have historically been highly revered by jurors and judges alike. However, in recent research by the innocence project, it was found that nearly 70% of exonerations involved eyewitness misidentification (The Innocence Project, 2020). Likewise, an examination of three wrongful convictions done by Carla Stenzel (2017) found that in each of the cases, an eyewitness was “sure” they had picked out the correct suspect from the lineup (Stenzel, 2017). The connection between this sense of sureness, or confidence, and accuracy is known as the confidence-accuracy relationship. Simply put, many believe that the more confidence a witness has, the more accurate their recollection must be despite other confounding factors. This claim was supported in work done by Key et al., (2023), utilizing mock jurors. Researchers observed that mock jurors were more likely to vote guilty when the eyewitness’s confidence was high, regardless of the timing of the confidence judgment, or if there were any inconsistencies in the confidence levels (Key et al., 2023). With the number of exoneration cases continuing to rise, practitioners and scholars alike have taken interest in the ability confidence has to predict accuracy. This study aims to combine current research on the confidence-accuracy relationship with newly developed factors such as the age of the witness as well as how long they viewed the culprit for.

Literature Review

One particular goal of this study is to test the pristine conditions hypothesis (Wixted and Wells, 2017) in children and adults with both suboptimal and optimal viewing conditions. The pristine conditions hypothesis is the idea that a highly confident witness will be equally accurate, given pristine testing conditions are used. There are five pristine conditions. The first condition is to only include one suspect per lineup, with the rest of the individuals in the lineup being made up of known-innocent fillers. This ensures the risk of mistaken identification is minimized, as it does not overwhelm the witness in scenarios with multiple suspects. The second is to create a lineup in which a suspect (both innocent or guilty) does not stand out. The fillers, or other individuals for the lineup should be relatively similar to the description of the suspect to prevent bias and misidentification. As shown by the “dud effect” (Charman, Wells, & Joy, 2011), witnesses have increased confidence when choosing an innocent person who resembles the perpetrator in a mix of fillers who appear completely different from their recollection. The third condition is to alert the witness to the possibility of the offender potentially not being in the lineup. This grants witnesses the peace of mind that they do not have to pick a suspect unless they are confident the suspect is one of the individuals in the lineup. There have been many cases in which the perpetrator was not involved in the lineup and an innocent person was convicted due to the pressure of the witness needing to pick someone (Wixted and Wells, 2017). The fourth is to use double-blind testing when administering a lineup. This simply means that the individual who is performing the lineup should not be made aware of the suspect to reduce the likelihood of any indicators – conscious or not – being given to the witness by the administrator. Lastly, the fifth is to collect a confidence statement at the time of the identification. Again, this should be collected utilizing the double-blind method. This confidence statement can be used as an indicator of reliability, under the pretense that the other conditions were met.

Additionally, the present research aims to discover the mediating factor of viewing conditions on the pristine condition’s hypothesis. Likewise, the current study also seeks to determine if age is a mediator in accuracy in addition to duration, which is an expansion on previous literature. One of viewing conditions is the length of time an eyewitness saw the perpetrator. Bornstein et al., (2012) found that longer exposures to perpetrators face did result in greater accuracy when compared to shorter exposures in adults. Keast et al’s 2007 article Children’s metacognitive judgments in an eyewitness identification task, delved into the confidence-accuracy relationship and how correlated they were across different age groups. They found that children were extremely overconfident in their answers, with this result remaining steady across two different experiments. The first experiment tested the confidence-accuracy relationship using a biased and an unbiased lineup. The biased lineup was created by omitting the mention that the suspect may not be in the photos, making the child choose a person regardless of whether the suspect was there or not. Across accurate identifications, it was shown that a child’s confidence was remarkably higher compared to what was seen in the adult groups. For inaccurate identifications or rejections, adults and children were closer in similarity. The second experiment used the same unbiased lineup method with all participants. In this case, half of the participants were given control questions in between their lineup choice and confidence scale, whereas the other half were given hypothesis disconfirmation questions instead. This second experiment upheld the findings that overconfidence is high among children, and the confidence-accuracy relationship being poorly calibrated. Though this research suggests that children may be overconfident, it unfortunately did not use the new cutting-edge analysis used by Wixted and Wells (2017). The analysis used by Keast et al., (2017) is actually known to find weaker confidence-accuracy relationships than what was used by Wixted and Wells (2017). The current study worked to expand on this research, utilizing the pristine conditions hypothesis as well as the same statistical analyses as Wixted and Wells (2017) to further determine the confidence-accuracy relationship in children.

One unique component of the current study is the usage of virtual reality (VR) technology. The advantage of utilizing VR technology is that it has “the ability to create recognizable, three-dimensional facsimiles of real objects in space” (Wilson & Soranzo, 2015). This is imperative to our research, as it allows the participant to be fully immersed in the environment, granting the ability to elicit responses similar to a real scenario. As further discussed by Wilson and Soranzo (2015), VR grants researchers the ability to create a very specific scenario that will be viewed relatively similarly by each participant. Likewise, VR technology gives us the opportunity to replicate a scenario that cannot otherwise be replicated in the real world due to ethical or an assortment of other issues, such as crime. The current study will use VR to portray the scenario to each participant, fully immersing them into the environment and allowing us to measure their real-life responses.

Current Research

As defined above, the pristine conditions hypothesis is the idea that a highly confident witness will be equally accurate, given pristine testing conditions are used. This is theorized to remain true even if the viewing conditions are suboptimal. However, some research has found that this hypothesis does not hold under suboptimal conditions. For example, findings from Lockamyeir et al’s (2020) paper indicated that across multiple experiments, confidence did not predict accuracy at a distance of 20 meters, where it did predict accuracy at 3 and 10. This is where the current study steps in, to further bridge the gap between this knowledge and how other mediating factors affect it. The claim has never been tested with children, which will allow our findings to be the first on the topic – potentially not holding the hypothesis. Likewise, the suboptimal viewing conditions are a shorter length of exposure rather than a longer distance, introducing a different way of testing the concept. Utilizing virtual reality (VR), participants are immersed into a fictional scenario via a 360-degree real video, where they are exposed to the perpetrators for different times, and then asked to recognize the men via lineups and their confidence. It is with this methodology that we hope to test whether the confidence- accuracy relationship maintains itself across diverse ages and two different viewing conditions.

Methods

Design

The study will use a 2 (exposure length: short [6 seconds] x long [34 seconds]) by 2 (culprits’ presence in lineup: present or absent) mixed design. The presence of the culprit is a within participants variable, meaning the participants will experience a lineup with and without the suspect. Exposure duration will be randomized between participants, meaning the participant will only be given one condition.
Lineups consist of six individuals, each appearing simultaneously alongside a red X to indicate the absence of the culprit. Both child and adult participants are given the exact same procedure to reduce any confounding variables.

Participants

Participants in this study are recruited from children’s museums as well as the university lab (adults only.) Participants have to be at least five years of age with no maximum age. Participants will be screened for any potential medical conditions that could harm or hinder their VR experience. Participants who have these conditions, and those who do not finish the study will not be included in the final sample.

Procedure

Before beginning the study, participants are briefed on the basics of our research as well as the risks associated with using VR. Participants then sign a consent and/or assent form to indicate their permission to participate. After explaining the VR equipment to the participant, trained research assistants will put the technology on the individual, play a practice video (to ensure the participant will not get motion sick), and then the subsequent crime video. Two versions of the video were created, containing either a long (total video length 5 minutes) or short (total video length 3 minutes) exposure duration of the suspect. In both videos, the other events remained the same. The video starts with the main character Ambrosia greeting the viewer and an on-screen friend before discussing various objects she had bought at a shopping mall. She shows the participants the objects and proceeds to put them in her bags on the bench at either side of her. From there, two men appear on screen and approach her on both sides, in a friendly manner. They briefly speak with her before proceeding to grab her bags and run off screen in opposite directions. Ambrosia is then approached by a nearby bystander, who asks if she is okay and proceeds to tell the participant they can help Ambrosia find the men.

At the conclusion of the video, the participant alerts the assistant, who will then remove everything. After the headset and accessories have been removed, the research assistant will begin administering the survey. Practice questions for both the line up and confidence questions are given to best ensure comprehension by the participant, regardless of their age.

Bibliography

Bornstein, B. H., Deffenbacher, K. A., Penrod, S. D., & McGorty, E. K. (2012). Effects of exposure time and cognitive operations on facial identification accuracy: a meta-analysis of two variables associated with initial memory strength. Psychology, Crime & Law, 18(5), 473–490. https://doi.org/10.1080/1068316X.2010.508458

Hanczakowski, M., Zawadzka, K. & Higham, P.A. The dud-alternative effect in memory for associations: Putting confidence into local context. Psychon Bull Rev 21, 543–548 (2014). https://doi.org/10.3758/s13423-013-0497-x

Innocence Staff. (2023, April 23). How eyewitness misidentification can send innocent people to prison. Innocence Project. https://innocenceproject.org/how-eyewitness-misidentification-can- send-innocent-people- to-prison/

Keast, A., Brewer, N., & Wells, G.L. (2007). Children’s metacognitive judgments in an eyewitness identification task. Journal of experimental child psychology, 97 4, 286-314 .

Key, K. N., Neuschatz, J. S., Gronlund, S. D., Deloach, D., Wetmore, S. A., McAdoo, R. M., & McCollum, D. (2023). High eyewitness confidence is always compelling: That’s a problem.

Psychology, Crime & Law, 29(1), 120–141. https://doi.org/10.1080/1068316X.2021.2007912

Lockamyeir, R. F., Carlson, C. A., Jones, A. R., Carlson, M. A., & Weatherford, D.R. (2020). The effect of viewing distance on empirical discriminability and the confidence-accuracy relationship for eyewitness identification. Applied Cognitive Psychology, 34(5), 1047–1060. https:// doi. org/10.1002/acp. 3683

Stenzel, Carla. (2017). Eyewitness misidentification: mistake that blinds investigations, sways juries, and locks innocent people behind bars. Creighton Law Review, 50(3), 515-532.

Wilson, C.J., & Soranzo, A. (2015). The Use of Virtual Reality in Psychology: A Case Study in Visual Perception. Computational and Mathematical Methods in Medicine, 2015.

Wixted, J. T., & Wells, G. L. (2017). The Relationship Between Eyewitness Confidence and Identification Accuracy: A New Synthesis. Psychological Science in the Public Interest, 18(1), 10-65. https://doi.org/10.1177/1529100616686966