• Heather D. Flowe, PhD

In the news: how media reports impact eyewitness memory reporting


“Your eyes can deceive you, don’t trust them.”


Quoting Star Wars is an unusual way for me to start a blog. But in the context of a recent paper of mine, the words spoken by Obi-Wan Kenobi to a young Luke Skywalker, said quote is rather relevant.


In the case of eyewitness testimony, we are increasingly exposed to news articles and media that highlight inaccuracies in the memory of a witness leading to wrongful convictions. An interesting question to ask is how this news impacts memory reporting in criminal justice contexts. Does this news make us less likely to trust our eyes?


For instance, in addition to widely-publicised news reports such as the Frontline story about Ronald Cotton in 1997 there have been television programs (The Good Wife), movies (Making of a Monster), and songs (Time, by Fabulous) that have drawn the public’s attention to wrongful convictions and eyewitness memory errors (Norris & Mullinix, 2019).


Drawing attention to wrongful convictions is valid, and valuable, but per the U.S. National Registry of Exonerations there have been 2,566 exonerations to date, with mistaken eyewitness identification named as a cause of the wrongful conviction in 741 of the cases. They are always damaging, but perhaps not as widespread as we might think, considering that over 44 million criminal* cases are filed every year in U.S. state trial courts alone.


The goal of eyewitness researcher is to improve eyewitness memory accuracy (i.e., increasing the number of correct responses and simultaneously decreasing the number of incorrect responses that witnesses provide, also known as discrimination accuracy). Doing so is beneficial in both reducing the number of wrongful convictions and supporting more rightful convictions of perpetrators who might otherwise walk free.


To that end, myself and colleagues and investigated whether the nature of these stories and media might affect eyewitness memory reporting. We published our results in memory of Dr James Ost, who was honoured for his contributions to the field of applied memory in a recent special issue in Memory.


We conducted two experiments involving mock crimes and fictitious news reports.


In the first, participants were shown a short video clip of a perpetrator breaking into a car and running away. They were then randomly assigned one of three false BBC News reports; a report on the accuracy of eyewitness memory, a report on the inaccuracy of eyewitness memory, and a control condition (a report on the accuracy of ballistics evidence).


Then we asked them to recall the events of the crime. Those who read that eyewitness memory is inaccurate were less confident in their memory accuracy and reported less information about the mock crime compared to those in the other conditions.


However, the specificity and accuracy of recall did not vary across conditions.


In the second experiment, participants watched a different mock crime in which a male perpetrator in his twenties spiked a woman’s drink. The same false BBC News reports from the first experiment were again employed.


They were then asked to identify the perpetrator from a six-person photo lineup. The lineup members did not have distinguishing features, were all wearing the same clothing, and were the same age, race, build, and had the same hair colour. Some participants were filtered into viewing a lineup where the perpetrator was present, and some into a group where the perpetrator was absent.


Participants who had read that eyewitness memory is inaccurate evaluated their ability to recall the mock crime as being relatively poorer, but their lineup decisions did not differ compared to other participants.


What does this all mean?


Well, ours is not the only study to investigate the relationship between information reported in the media and memory reporting. James Ost was also fascinated by this topic (e.g., Ost et al., 2008).


Our findings are consistent with those that have gone before in finding that there is a relationship, but go further and suggest that news stories (and other media) on the subject of eyewitness memory inaccuracy affects how people evaluate their memory capability, and that this can subsequently influences how people report their memory to the police.


Police interviews are normally comprised of free-recall prompts (e.g., “What happened?”) and cued-recall questions (e.g., “What colour was the perpetrator’s shirt?”). In a free-report scenario, an eyewitness can determine whether to report or withhold information, and regulate what we refer to as the ‘grain size’ of their answers.


A fine-grained answer is richer in detail than a coarse-grained answer. In simple terms, it is the difference between a reporting that a perpetrator wore dark shirt and describing a ‘short-sleeved dark blue shirt with a collar’.


Goldsmith et al. (2002) found evidence that people first try to retrieve a relevant fine-grained answer from their memory. They then assess the likely accuracy of the answer based on their confidence in it. If they do not consider themselves to be confident in that fine-grained memory they will answer at a coarser level.


Now, if the witness’ confidence in their memory is modulated by news reports or a movie they have recently seen, it stands to reason that their threshold for reporting fine-grained answers is heightened and that they will adopt a more conservative approach to their responses.


That was certainly the case in our first experiment, but not the second.


Perhaps memory output was affected by the news story in the first experiment because it was more specifically applicable to participants’ memory test. The news was about eyewitness memory in general and only briefly mentioned identification performance. Indeed, research comparing general versus specific warnings on misinformation reporting has found warnings must be specific to the task at hand to have impact (Higham et al., 2017). If the news had specifically been about mistaken identification, it may have impacted lineup decisions.


Nevertheless, our findings add to the ongoing debate about the extent to which eyewitness memory should be characterised as inherently ‘reliable’ or ‘unreliable’, where ‘reliable’ means that witnesses know when their memories are accurate or inaccurate and report information accordingly, accurately assigning confidence that reflects their accuracy.


Research shows that confidence collected at the time of initial ID and recall is informative about accuracy. Confident eyewitnesses are seen to be more reliable. However, if confidence and memory reporting strategies have been influenced by external factors, such as media reports about eyewitness inaccuracy, it has the potential to lead a reliable witness to underestimate their memory accuracy.


This highlights the importance of collecting confidence under conditions where participants report their memories and assign confidence after reflecting specifically upon the quality of their own memory for the event in question.


Doing so will improve the reliability of memory evidence and, in turn, benefit the justice system in achieving more accurate convictions.



Eyewitness memory in the news can affect the strategic regulation of memory reporting is available here.


With thanks to my colleagues Muhammad Mussaffa Butt, Melissa Colloff and Elizabeth Magner.


*This figure includes felonies and misdemeanors. One criticism of combining both categories of crime to consider alongside the number of wrongful conviction cases, is that perhaps eyewitness testimony may be more crucial in felony than misdemeanor cases. Afterall, the wrongful conviction cases in the media are felonies for which defendants spent 14 years (on average) in prison. However, even if we limit the numbers to only felonies, the numbers are still staggering. For instance, the National Judicial Reporting Program provides figures we can extrapolate from, as it includes nationally representative data from 300 counties in 2006 (the last year for which data were available). There were 494,055 felony cases that year in just those counties alone (there are 3006 counties in the US). It seems reasonable to assume that witnesses also provide correct IDs and statements, and without this information, cases would not have been prosecuted. Witness memory reports provide evidentiary leads that enable investigator to uncover other evidence, and witnesses can make accurate identifications as we know from the laboratory studies. We don't know how often they do in real world cases. Nevertheless, it is important to keep in mind that wrongful conviction cases with erroneous eyewitness statements and testimony are one side of the coin, while lawful convictions with accurate statements and testimony are on the other. Special thanks to @Arpita_Ghosh_ for pointing me in the direction of these stats.