Have you ever watched one of your favorite procedural crime shows (there are plenty, so pick your poison) and wondered who designs the techniques used to elicit eyewitness accounts of the crime? Certainly shouting in the face of the witness is likely to influence the quality of recall, but what other factors play a role in eyewitness memory?
If it’s been a while since you saw a line-up on TV, here is a parody to refresh your memory:
[youtube https://www.youtube.com/watch?v=QmW_XCbV0iM]
As a memory researcher and avid fan of CBS’s NCIS, I have noticed that there is a very natural and symbiotic relationship between cognitive psychology and the criminal justice system. Many psychologists have used their experimental findings to inform legal procedures. A prominent example of this is the National Academies of Science 2014 report on police practices. This report has been championed by the Innocence Project and calls in to question many criminal cases that hinge on the results of questionable practices and eyewitness testimonies.
Thus, a very real challenge for the legal system is to elicit eyewitness testimonies in the most unbiased and accurate manner possible. Simultaneously, the challenge for eyewitness researchers is to perform rigorous experiments that produce reliable findings and directly translate to real-world practices. To this end, researchers Jamal Mansour and colleagues recently reported another evaluation of appropriate procedures for assessing eyewitness memory in laboratory settings. In their recent article published in the Psychonomic Society’s journal Behavior Research Methods, they investigated whether multiple-trial viewings influence eyewitness outcomes such as accuracy and confidence of choice.
Although real-world eyewitness scenarios involve viewing a single crime and lineup, adapting this design into a laboratory experiment is costly in time and resources. It also raises issues of having enough statistical power to detect differences. An efficient alternative is to use multiple-trial viewings for each participant to increase statistical power from a smaller sample. However, a very real concern posited by Mansour and colleagues was that if a multiple viewing design per se strongly affects performance, such as yielding high correct identification rates but also high false alarms, consumers of this work (e.g., police, attorneys, etc.) may over-estimate the accuracy of eyewitness testimonies, unaware that performance is an artifact of methodology. Thus, before advocating for this design, the researchers assessed what effect multiple trial viewings had on performance from conventional eyewitness lineup tasks (sequential vs. simultaneous). In sequential lineups, suspects are presented one at a time to the eyewitness, and in simultaneous lineups (very popular on TV) all suspects are shown at the same time.
Mansour and colleagues made several predictions regarding the effect of multiple viewings on eyewitness outcomes. For example, they predicted that viewing multiple trials will have no effect when using simultaneous lineups, given the information available to the eyewitness (e.g. the eyewitness knows the number of suspects they have to choose from), but may affect the rate of correct identifications when the perpetrator is absent with sequential lineups. Mansour and colleagues also predicted that with careful research techniques, perceived success on earlier trials will not result in higher choosing rates (correct or otherwise), on later trials.
To test these predictions, participants were shown 24 video clips of mock crimes ranging from the planning of a burglary/robbery to the planning of a murder. Participants were then asked to identify the perpetrators from either a sequential or simultaneous line-up. Across participants, the videos varied in length and visibility (size of the screen that the video was shown on). This allowed the researchers to assess the role of memory strength. In the videos, the perpetrators’ disguises also varied in style (e.g., hat or sunglasses) and degree (e.g. partially or completely obscured faces) – just like in an episode of NCIS or Law and Order.
Although the results shed light on the role of each forensically relevant factor, here we discuss just a few. First, as to be expected, good memory strength (longer video clips) was associated with better eyewitness outcomes such as the rate of correct identifications, correct rejections, and confidence. This means that the longer an eyewitness spends viewing an event, the better their recall of the perpetrator is likely to be.
Importantly, and more relevant to the focus for this research, there was no real detriment to using multiple trial viewings for assessing eyewitness accuracy, rate of choosing perpetrators, and confidence of choices. The results showed no significant (or at most a negligible) effect of trial number on the eyewitness outcomes, and repetition also did not interact with the other factors of interest. That is, in a nutshell, whether witnesses view a line-up once or multiple times makes no difference to their ultimate accuracy.
As for the sequential vs simultaneous line-up debate, the researchers replicated the standard pattern of more correct identifications from (perpetrator-present) simultaneous line-ups and more correct rejections from (perpetrator-absent) sequential line-ups. Consistent with this is the finding that participants are generally more likely to make a choice (selecting a perpetrator as opposed to selecting ‘not there’) from simultaneous line-ups relative to sequential line-ups.
Interestingly, although people made more correct identifications when viewing simultaneous line-ups, they were more confident in their identification choices when viewing sequential line-ups. Conversely, people made more correct rejections when viewing sequential line-ups, but were more confident in their rejections for simultaneous line-ups (Still scratching your head at that? Me too!). So which lineup type is more superior? Well, the verdict is still out on that.
The remarkable thing about the research by Mansour and colleagues is that the implications of the results are two-fold. First, the finding that multiple-trial viewings have little to no effect on eyewitness outcomes is good news for eyewitness researchers because it informs methodology and it advances research. Secondly, the findings of accuracy and perceived confidence, given the other relevant factors (e.g., types of disguise and memory strength), lends perspective to how real-world eyewitness accounts should be weighted.
Studies like the work of Mansour and colleagues are important for critical scrutiny of criminal justice proceedings. It further highlights the far reaching real-world applications of psychological research, especially for the field of criminal justice. So think about that the next time you sit down to watch an episode of Sherlock or Murder, She Wrote (it’s a classic!).
Article focused on in this post:
Mansour, J. K., Lindsay, R. C. L., & Beaudry, J. L. (2017). Are multiple-trial experiments appropriate for eyewitness identification studies? Accuracy, choosing, and confidence across trials. Behavior Research Methods. DOI: 10.3758/s13428-017-0855-0