Recently, an interesting review paper was published by Lacy and Stark (2013) in Nature Reviews Neuroscience. It discusses how the scientific ignorance of law enforcement personnel, judges and jurors about memory and how it works has detrimental impact on the efficacy of legal system and human lives.
Common misconceptions about memory
There is a lot of common misunderstandings about memory. It is not like a video camera, perfectly recording and storing memory for later retrieval. It can change over time, adding and subtracting details, filling in blanks and become distorted. A popular misconception is that the accuracy of a memory is proportional to the confidence assigned to it by the eyewitness. In reality, the correlation between confidence and accuracy is weak. Hypnosis is not considered to be effective for accurate recall, despite the fact that over half of the general public sampled believe it to be (among the memory expert sample, 0% agreed or strongly agreed).
One striking finding is that people who work within law enforcement and the criminal justice system display only a little less ignorance about memory as do college students, who in turn are not that much better than the general public. This has profound implications for the court system.
How memory can get distorted
There are many different ways that human memory can become distorted. In principle, distortions can occur during perception, encoding, storage and retrieval. Here are some of the main causes of distortion mentioned by the paper.
Memory confidence tends to increase if exposed to positive feedback by law enforcement personnel after the identification or decrease if exposed to negative feedback. Because they are often perceived as authorities, their feedback is taken into account by eyewitnesses in making an estimate of accuracy. This kind of effect can occur even if the post-identification feedback is non-verbal e. g. facial expression and body language if the person conducting the line-up is not blinded to who the police suspects for the crime. Even repeated questioning can increase the confidence in the accuracy of a memory.
Distortions of memory can occur after being exposed to misleading information about that memory. Just subtle variations in the wording of a question (e. g. “smashed” versus “hit” related to a car crash) can cause memory distortions, not just about the speed at impact, but also about peripheral details of the scene, such as presence of broken glass.
Passage of time
Memories can be distorted by merely the passing of time. The classic study demonstrating this asked people were when they first heard about the events of 9/11. Years later, when asked again, around 40% of people had different memories of the situation. In line with previous research showing a weak correlation between confidence and accuracy, the confidence of the study participants was still high.
Filling in the gap
People have a tendency to fill in the gaps i. e. remembering details that did not actually occur, but fits the general expectations of a certain crime situations.
Something similar occurs when lawyers ask leading questions. The information provided by the lawyer can bleed over into becoming a memory of what was said by the eyewitness testimony. This effect still occurs even when jurors are told to focus on what the eyewitness is saying and not the lawyer.
The neuroscience of memory distortion
Memories can be strengthened by a process called long-term potentiation, where the signaling between two neurons is strengthened because they fire together. When certain neurons are wired together do not fire in a coordinated fashion, this can led to a weakened synaptic connection (called long-term depression), and thus presumably a weakened memory. When a person attempts to retrieve a memory, the associated synapses become labile. If the memory is not updated through reconsolidation it may be forgotten. During this process, memory distortions could occur. It is also clear that new and old memories of a situation can interfere with each other. If these are similar in some respects, the attempt to retrieve the old memory can make details of the new memory bleed over. The paper goes into addition details about the neuroscience background.
The consequences of ignorance about memory
The paper discusses several legal consequences of the fact that law enforcement personnel, judges and jurors tend to not be aware about the science of memory. Two startling examples relates to common decisions not to prosecute sexual assaults and how DNA evidence has exonerated hundreds of people who were convicted based on eyewitness testimony.
Prosecution of sexual assaults
According to the review article, around 86% of sexual assaults reported to the police are not prosecuted because the law enforcement personnel consider the testimony given by the putative victim to be unreliable. For sure, some proportion of those testimonies may actually be of low accuracy, but many testimonies might be seen as unreliable because the victims does not display a high degree of confidence. As we have seen, such an inference is not warranted because the correlation between confidence and accuracy is weak.
The Innocence Project
There are over 300 well-documented cases were DNA evidence has exonerated individuals who were wrongfully convicted of a crime, often (~75%) based on eyewitness testimony. The average time that these individuals spent incarcerated was a little over 13 years. On the outside, the financial compensation is typically low and their lives are in ruins. Friends and family have disowned them and they have a hard time finding work.
So what can be done to mitigate and reduce these problems in a legal context?
The review paper recommends a number of suggestions for reducing the problems associated with misconceptions and distortions related to memory. These suggestions apply to a number of areas, such as the interviewing of eyewitnesses, suspect identification and jurors.
Because of potential for distortions caused by e. g. post-identification feedback, any confidence statements should be recorded right away. The interviewer should use the enhanced cognitive interview technique: giving over the control of the conversation to the witness (to avoid unintentional manipulation by individuals with perceived authority), reminiscing back to the context of situation (sight, smells, sounds feelings, thoughts etc.), a free recall segment to remember as much as possible (even if it is not in chronological order) and a segment with open questions.
The paper even makes the suggestion that several eyewitness accounts that show some degree of consistency should not automatically be believed to have more accuracy, as they may not be independent and all individuals are susceptible to the same kinds of memory biases and errors. Although the paper does not discuss this, a key case study of this is the murder of former Swedish Minister of Foreign Affairs Anna Lindh. She was stabbed in a major department store in Stockholm in late 2003. The police managed to identify and isolate witnesses. Unfortunately, they were all isolated in the same room and started talking to each other. Furthermore, the police held some initial eyewitnesses interviews in that room in the presence of all witnesses, some of whom interjected and “corrected” the testimony of other individuals.
These processes ended up distorting and contaminating the memory of the witnesses, leading the police to release a lot of incorrect details about the suspect. They even initially arrested the wrong person. Around a week later, DNA analysis of items left behind by the murderer cleared this person and another individual, Mijailo Mijailović, was arrested. Mijailović was later sentenced to life imprisonment. Without DNA evidence, this could have been a miscarriage of justice. For more details about the social influence on eyewitnesses in the Lindh case, see Granhag, Memon and Hjelmsäter (2010).
The non-suspects selected to take part in a line-up should resemble the suspect. The law enforcement personnel should remind the witness that the suspect might not be in the line-up and the individuals that are part of the line-up should be presented sequentially instead of simultaneously. Both of these guidelines has been shown to strongly reduce false positives (and slightly reduce true positives). The line-up should also be double-blind i.e. done by a law enforcement personnel who does not know who the suspect is. If short on staff, this can be done by a computer displaying images of the individuals taking part in the line-up.
They should be allowed to take notes during court proceedings as the information presented is vast and it might be difficult to find relevant information in long court transcripts. Taking notes also seems to improve memory. They should also be informed about the fact that memory is not immune to changes, the potential for leading and misleading questions to distort memory, that confidence is not an accurate guide to accuracy and that people tend to fill in the gaps in their memory by inventing things that may or may not have happened.
The paper offers three main conclusions. First, there are a lot of misconceptions about memory even by law enforcement personnel, judges and even jurors and this has adverse effects in the courtroom. More education and awareness about the science of memory is needed, both for the general public and for people who work in association to the legal system. Second, the legal system should reexamine the evidential value of eyewitness testimony and they even point to suggestions that individuals should not be convicted based on eyewitness testimony alone. Finally, more research should be done so that we can know more about its limitations, how we can detect errors and improve it.
Granhag, Pär Anders, Memon, Amina, & Hjelmsäter, Emma Roos af. (2010). Social influence on eyewitness memory. In P. A. Granhag (Ed.), Forensic Psychology in Context: Nordic and international approaches (pp. 139-140). Portland, Oregon: Willian Publishing.
Lacy J. W., & Stark C. E. (2013). The neuroscience of memory: implications for the courtroom. Nature reviews. Neuroscience, 14 (9), 649-58 PMID: 23942467