The Feild-Simulation Pardigm: Examining Witness Perfromance in Lab vs. Real-World Feild Conditions.

Our Field Simulation Experiments

Since 2012 we have collected data on nearly 3000 participants using our field-simulation paradigm. The field-simulation paradigm was designed to combine the control of lab studies with the ecological validity of field experiments by enabling researchers to examine how eyewitnesses perform under highly realistic conditions, while still maintaining a high level of experimental control over the witnessing conditions and the presence of the culprit.  This is accomplished by staging a mock crime and immersing participants in what they are led to believe is an actual police investigation, in which their identification would presumably lead to the arrest and prosecution of the suspect.  Like field experiments, the police are the experimenters and it is assumed that the witnesses’ identifications will lead to the arrest of the suspect. However, unlike field experiments, ground truth of guilt is always known, so researchers can directly examine false identifications of innocent suspects, rather than being limited to examining mistaken filler choices. Additionally, because ground truth is known, investigators can also examine false identifications in showups conducted under field conditions, which could not be accomplished in field-experiments because showups have no innocent fillers. Moreover, since, culprit-presence is controlled and manipulated by the experimenters, this ensures equal numbers of innocent and guilty suspects are present across the conditions being tested.  This level of experimental control enables researchers to make causal claims about which procedures result in better trade-offs between guilty-culprit and innocent-suspect identifications.  This also allows investigators to examine base rate performance for accuracy and false identifications when using different procedures. 

Field-simulations also have a major advantage over tradition field experiments, because, like lab studies, this paradigm permits experimental control over witnessing conditions. Because the researchers control all elements of the protocol from the staged crime to the identification procedures, they can manipulate exposure to the culprit. This allows investigators to systematically examine how certain estimator variables interact with the procedures being used. 


Do witnesses perform the same way in the field versus the lab: Not necessarily.

While there is a growing body of research examining the relatively “cold”, cognitive decision-making components of showups, few attempts have been made to capture the “hot” affective components of showups that are thought to exacerbate the suggestiveness of the procedure. In three simulated-field experiments, we partnered with law enforcement to examine how participants who were led to believe they were involved in an actual criminal investigation (Field-simulation condition) differed from participants who knew they were not part of an actual investigation (Lab-simulation condition). We staged crimes for both conditions, but in the field-simulation condition, law enforcement personnel carried out mock investigations that culminated with a live showup. In Experiment 1 (N=321), which did not include a culprit-present condition, the field-simulation condition increased innocent suspect identifications. The standard showup admonition decreased innocent suspect identifications, but only for dissimilar innocent suspects. Experiment 2 (N=196) added a culprit-present condition and found that the field-simulation condition increased innocent suspect and culprit identifications to a similar extent. Experiment 3 (N=367) replicated the findings of Experiment 2 and examined the impact of admonishing eyewitnesses that if they did not believe the suspect was the culprit, they might have additional opportunities to make an identification. Confidence-accuracy calibration analyses revealed that confidence discriminated accurate from inaccurate identifications in the field, but not in the lab; however, eyewitnesses who made identifications in the field were overconfident and across all levels of confidence were less likely to be correct than eyewitnesses who made identifications in the lab.

Eisen M. L., & Skerrit-Perta, A. Owen, J., Cedre, & Jones, J. (2017). Pre-admonition Suggestion in Live Showups: When witnesses learn the cops believe they caught ‘the’ guy. Applied Cognitive Psychology, 31(5), 520-529.

Eisen M. L., Smith, A., Olaguez, A. & Perta-Skeritt, A. (2017). Showups Conducted by Law Enforcement in the Field: Comparing Witness Performance in the Field Versus the Lab. Psychology, Public Policy and the Law23(1), 1-22.


New Research Administering Lineups on Tablets in the field

Funded by the Arnold Foundation

Showups are primarily used when law enforcement personnel locate a suspect near the scene of the crime, but lack probable cause for arrest. Because law enforcement personnel can only detain the suspect for a short period of time, traditional methods of lineup construction are not practical. Thus, showups are justified in this exigent circumstance in which lineups are not feasible. 

In light of modern technology, it is more possible now than ever that law enforcement personnel could construct a lineup in the field in a matter of minutes and administer the lineup with an iPad. Yet, it is not a foregone conclusion that a photo-array lineup carried out with an iPad will outperform the live presentation of a suspect in a showup procedure. Indeed, it might simply be the case that the benefit of live presentation in a showup procedure outweighs the benefit of fillers in lineup procedures. In the present research, we compare six-person iPad-based lineups with showups in a simulated-field procedure in which law enforcement personnel administer identification procedures to eyewitnesses who believe they are involved in an actual criminal case. 

We recently collected data on over 1,000 participants using this method. 

Manuscript in preparation.

I Remember the Tattoo: Misidentifying an Innocent Suspect Can Alter a Witness’s Memory for the Perpetrator’s Face 

Can mistakenly identifying an innocent suspect change your memory for the culprit’s face? Case studies of people who were falsely identified, convicted, and then later exonerated with exculpatory evidence have revealed that when witnesses become convinced of the suspect’s guilt, their memory for the culprit is often altered, such that when they remembered the crime, they recalled the face of the innocent suspect as the perpetrator.  Although anecdotal cases studies of this phenomenon are relatively common, no extant controlled experiments have clearly demonstrated this effect.  Findings from the current study demonstrated that misidentifying an innocent suspect resulted in a specific alteration to many witnesses’ memory for the perpetrator, and that post-identification feedback enhanced this effect. Participants viewed a video of a simulated and were led to misidentify an innocent suspect from a suggestive culprit-absent photo-array in which the suspect was the only viable choice.  The innocent suspect either had a tattoo photoshopped onto his face, or, like the culprit, had no visible facial markings or tattoos.  After making a selection, half the participants received post-identification feedback confirming their mistaken identification.  Finally, participant-witnesses were asked to describe the perpetrator in their own words.  Results revealed that 44.6% of participant-witnesses who misidentified the tattooed suspect mistakenly recalled that the carjacker had a tattoo on his face.  As predicted, receiving confirming feedback significantly boosted erroneous reports of recalling the tattoo.  

We conducted this experiment live, and also online.  We found that the effect was more powerful live compared to MTurk. Indeed, we have made this comparison in other studies and found that the face-to-face interaction appears to enhance suggestive influence from the lineup administrator.

The live version of this experiment is currently under review.

What happens when a witness either does not choose or selects a filler from the lineup and the officer prompts them to continue their search?

“Does Anyone Else Look Familiar?”: The Effects of Subtle Disconfirming Feedback on Confidence Judgments and Lineup Decisions

This study was designed to examine the effect of subtle disconfirming feedback on witness confidence judgments and lineup decision-making. Subtle disconfirming feedback took two forms in this study. Participants’ lineup decisions were either questioned by asking them in a neutral tone if they were sure about their choice and if anyone else looked familiar, or this same questioning was done in concert with skeptical gestures (e.g., shaking one’s head from side-to-side and squinting). It was hypothesized that subtle disconfirming feedback would lead to a decrease in witness confidence and the selection of additional photos, and that the effect would be strongest when the skeptical gestures were employed. In addition, we examined the relationship between confidence change following post-identification feedback and both memory for the perpetrator and suggestibility as measured by errors on misleading questions related to the event. Results revealed that subtle disconfirming feedback led to a significant decrease in confidence. Also, when participants were asked if anyone else looked familiar, more than 40% selected an additional picture; and of that group, the second choice was made with equal or greater confidence than their original decision 30% of the time. Moreover, suggestibility as measured by errors on misleading question was related to confidence change.

Eisen, M. L., Cedre, G., Williams, T, & Jones. J. M. (2018) Does anyone else look familiar? Influencing identification decisions by asking witnesses to reexamine the lineup. Law and Human Behavior42 (4), 306-320

What happens when one witness hears another say that the perpetrator had a distinctive tattoo, and only one guy in the lineup has that type of tattoo; but he is not the actual culprit?

The effects of plausibility and retention interval on eyewitness conformity

One hundred and seventy seven participants watched a video of a simulated carjacking. Right after the video witness statements were taken from each participant; as they were asked to describe what they just witnessed.  All sessions involved five participants, consisting of two confederates and three subjects. The confederates’ each made three errors in their reports (6 in total). These errors included included three pieces of plausible and three pieces of implausible information about the culprit. These suggestions involved distinctive and forensically relevant features of the perpetrator; including the suggestion that the perpetrator had a tattoo on his neck. After a 5-minute, 50-minute, or 48-hour delay, free recall was obtained from each participant individually, and they were asked to pick the perpetrator from a target present or absent photo array.  In each of these photo arrays a filler with a tattoo on his neck was included. Results revealed a main effect for plausibility, the plausible misinformation was more likely to be included in the participants memory reports at each retention interval. Also, participants also selected the filler with the tattoo on his neck 38% of the time at the 5 and 50-minute retention intervals, and 36% of the time at 48 hours.

Eisen, M. L., Gabbert, F., Ying, R., & Williams, J. (2017). “I Think He Had A Tattoo On His Neck”: How Co-Witness Discussions About A Perpetrator’s Description Can Affect Eyewitness Identification Decisions. Journal of Applied Memory and Cognition6(3) 274-282.


In this series of studies we examined how the introduction of bias by depicting the defendant as a gang member can affect jurors memory for the evidence and their decision-making. In four different studies over 1000 participants viewed three different abbreviated simulated trials where biased was systematically varied, as the defendant was either described as being gang affiliated or not. Consistently jurors voted guilty more often when the gang bias was induced.  Also, after deliberating in groups, many jurors continued to voted guilty in gang bias conditions even when reasonable doubt was clearly established, as evidence by no guilty verdicts in the no bias condition.

Ripple Effects: How Biased Evidence Can Cause Distortion in Jurors’ Memory

Pizzaro et al, (2006) proposed the theory of Ripple effects to explain how judgments of moral responsibility can have biasing effects on people’s memory for details of a crime. This research highlights the importance of understanding how potentially prejudicial information about a defendant can affect jurors’ perception of, and memory for, the evidence presented.  This is issue come up often in gang cases. In most US states, if the accused party is on record as having any association with gangs, it is common for that crime to be charged as  ‘gang-related’. Eisen et al, (2013) noted that labeling a defendant as a ‘gang-member” can induce a confirmation bias, leading jurors to filter the evidence through a negative stereotype, resulting in greater attention to information that confirms the established bias. However, there is also good reason to believe that this type of negative stereotype can have a ripple effect that extends well beyond how jurors filter the information, and also affects their memory for the evidence. This study was designed to examine how Ripple effects can bias jurors’ retrospective memory for the evidence presented in a trail where the defendant is charged with a gang-related offense.

Two experiments were conducted. In each experiment, participants viewed a simulated trial designed to establish reasonable doubt where the presence of gang evidence was manipulated. In both trials, the only evidence presented was a tentative eyewitness identification.  It was hypothesized that when the defendant was identified as a gang member, jurors would be more likely to vote guilty, even when reasonable doubt was established. In experiment 1, participants (n=212) viewed a trial involving robbery; in experiment 2, participants (n=235) viewed a more serious crime involving a murder. In the filmed trials, the defense attorney and prosecutor were played by judges, the victim/eyewitness was played by an actor, the investigating officer was played by a local police official, and the people’s gang expert was played by an officer who frequently provides gang testimony in court. Participants provided verdicts before and after deliberations.

Results demonstrated that when there was no mention of gang evidence, only one person out of 249 participants voted guilty across trials after deliberations. The fact that only one juror in the no-gang condition voted guilty shows that reasonable doubt was clearly established.  However, a significant minority of jurors continued to vote guilty when the defendant was described as a dangerous gang member (10% for robbery vs. 19% for murder). Qualitative analyses of the mock jurors post deliberation responses show evidence of a ripple effect. When asked why the participants voted guilty, in each experiment, the majority of participants in the gang condition noted that the defendant’s prior criminal history played a role in their decision. However, no evidence of this was offered in the trial.  It appears that hearing expert testimony on the operation of the gang lead jurors to incorrectly recall that evidence of prior criminal conduct was actually presented at trial.

Other studies on point

Eisen, M. L. & Dodson, B, & Olaguez., A (2015). How Prejudicial is Gang Evidence? Washington College of Law, Criminal Law Practitioner.

Eisen, M. L. Dodson, B. & Dohi, G. (2014). Probative or Prejudicial: Can Gang Evidence Trump Reasonable Doubt? UCLA Law Review, Discourse