Our Faulty Cognitive Instruments
Eyewitness identification is the leading cause of false conviction. Why? Because memory is not as reliable as we think.
Last weekend, I attended a Forensic Science Society conference entitled: “Are you open to Suggestion? The Role of Subjectivity & Influence in Forensic Evaluation/Interpretation.” I have learned a great many interesting things.
In his opening address, The Hon. Sir Nigel Sweeney, High Court judge, warned of “the risk of the fleeting glimpse.” The most honest and reliable witness may be mistaken. They may convince themselves. As Hannah Fawcett, Teeside University Psychology lecturer, pointed out in her presentation, the truth may not be fact. An eyewitness may be telling the truth. They may completely believe what they think they saw. They are not lying but their memory is inaccurate.
As well as the risk of the fleeting glimpse, other factors can affect our accuracy in recalling events. Hannah Fawcett talked about these estimator variables. Viewing conditions can affect what we remember. If we can’t see properly, if it’s dark or our glimpse is fleeting or obstructed, then our ability to identify a person is affected. Intoxication has an effect too. It seems obvious to suggest that the more intoxicated we are, the less accurate our memory becomes. Research confirms this. Being confronted by violence or the fear of violence can affect our powers of recall. The ability of our brains to generalize is also a factor in our ability to recall events. Our brains have an implicit tendency to make associations. This is how we cope with the world. University of Sheffield Professor of philosophy, Jennifer Saul, spoke about this in her presentation. Our brains categorize objects. Such schemas, or collections of associations are automatic and stem from our existing knowledge of those objects. As our knowledge grows, the more elaborate our collections of associations become. The problem with schemas in terms of recall, is that they may lead us to add associated information to our memories in terms of details that didn’t actually happen. We see what we expect to see, based on our existing knowledge of such situations, and we do it unknowingly.
Co-witness discussion can also lead individual witnesses to add details to their accounts, based on what other witnesses said they saw. Again, if those additional pieces of information that we hear from other witnesses fit with our schema, we may think we saw them too because they meet our expectations. This highlights the need for police to interview witnesses quickly, before they have chance to discuss the event with other witnesses and unknowingly alter their memory of it.
Hannah Fawcett went on to speak about the misinformation paradigm. How, in experiments, subjects were shown photoshopped childhood photographs of themselves in situations in which they had actually never been. At first, the subjects rejected the situation, claiming to have no memory of it, but by the second or third session viewing the photos, some subjects believed that they remembered these false events. The fact that false memories can be implanted in this way is disturbing. What affect might suggestion have on eyewitness accounts of a crime over time?
A biased instruction or question may invoke confirmation bias, leading to an unknowingly false witness account. Professor Saul invited us to consider the question, “Is Jane friendly?” When asked the question, a subject will consider all the instances when they have experienced Jane being friendly, and assuming there are some instances, will conclude that, yes, Jane is indeed friendly. However, if asked, “Is Jane unfriendly?”, the subject will recall a completely different set of circumstances and may reach a conclusion that contradicts their original answer. So, the phrasing of the question affects the answer. Modern policing methods use cognitive interviewing techniques which increase the frequency of correct witness information and decrease the inclusion of fabrications. Mental context reinstatement puts the witness back at the scene and asks them to think about, not only what they saw, but how they felt, what they thought, how did it smell or sound. This helps to overcome some presumptions. Witnesses are asked to report every detail, no matter how trivial. They are also asked to change their perspective by viewing the scene from a different position, or to describe events in a different temporal order. These techniques allow witnesses to challenge some of the blanks their brain might have unknowingly filled in for them, based on schemas.
Facial recognition is a particular problem for eyewitnesses. Our ability to describe an offender is limited by our lack of vocabulary to describe facial features. The problem is compounded by our brain’s preference to process the whole face, rather than the individual facial features. The issue of recognition versus recall is also a factor leading to false identification. It’s quite common to think, “I know that face, but where from..?” There have been cases of false conviction based on eyewitness identification of a suspect who is recognised by the witness from somewhere, but, as it turns out, not from the crime scene, despite what the witness recalls. The way police line ups are conducted can be a factor in false identification. Ideally, line ups should be double blind: neither the witness nor the officer conducting the line up is aware of the identity of the suspect, or if they are even in the line up. Therefore, the officer is unable to give any unconscious cues to the witness. The case of Ronald Cotton gives us an example of what can happen.
Hannah Fawcett pointed out that erroneous experts are even more dangerous than erroneous witnesses. Whilst judges and lawyers might now be more aware of the problems of implicit bias, jury members are unaware. Because of factors such as the so called ‘CSI effect’, an expert perceived to be confident in their testimony, is more likely to be believed by a jury, even if their evidence is flawed.
Expert witnesses are people too! Our cognitive instruments must also be faulty, which is frightening because we use them every day to reach apparently objective conclusions. How does implicit bias affect a forensic science interpretation?
The objective nature of fingerprint evidence has been questioned recently in the light of the McKie case. Such is the perceived conclusive nature of a fingerprint match, former detective, Shirley McKie, was tried for perjury in 1997 when she denied leaving a fingerprint at a murder scene. However, the fingerprint match reported by Strathclyde Police was later found to be erroneous. Her appeal and exoneration in 2006 led to a public enquiry chaired by Sir Anthony Campbell. His report, issued last December, gives 86 recommendations, three of which relate to cognitive bias:
- The Scottish Police Service Authority should review their procedures to reduce cognitive bias.
- Fingerprint examiners should be trained to be conscious of cognitive bias.
- The SPSA should consider what case information is required for fingerprint examiners to perform their work and the information they are party to should be recorded.
In her presentation, Joanne Tierney of the SPSA explained the measures that have been implemented in response to the enquiry. She acknowledged expert opinion in defining a fingerprint match as a subjective discipline. Prior to the enquiry, experts had become complacent and the boundaries between fact and opinion had become blurred. Historically, a fingerprint match was verified by an independent examiner who was given full knowledge of the first examiner’s opinion of the match. The enquiry suggested that it’s entirely possible that a fingerprint expert verifying a match might be subject to confirmation bias, especially if the first examiner were a more senior colleague. Presentations by Dr David Charlton, Senior Fingerprint Officer of Surrey and Sussex Police and Luke McGarr, secretary of The Fingerprint Society, suggested that examiners might also be unknowingly influenced by other factors, such as the seriousness of the offence or the urgency with which a result was required. It was clear from Joanne Tierney’s presentation, that the SPSA fingerprint experts have undergone a culture change. Now, verification of a match is carried out without access to case information or knowledge of other expert opinion, and in the case of complex comparisons, where the first and second examiners can’t agree, an independent panel caries out a further blind comparison. It remains to be seen whether procedures have changed in the UK fingerprint bureaux, but I suspect that if not, change is imminent.
If the problem of cognitive bias is to be considered, then it should be considered across all areas of forensic science. This conference has certainly made me more aware of the problems. In his introductory address, Sir Nigel Sweeney explained that the level of information required by an expert will depend on the expert’s discipline. I am reassured by this acknowledgement. In my expert area, detailed case information is required in order to make complex interpretations of body fluid distribution and DNA profiling results. These interpretations do not just seek to address who the DNA is from, but also how the body fluid was deposited, and whether that relates to a specific activity. Information surrounding timescales and the condition of the crime scene are important factors to be taken into account when addressing these issues. In order to avoid implicit bias, we have numerous checks and balances built into our procedures. However, the fact remains that as forensic experts, we are still exposed to other pieces of case information, often relating to the victim and suspect, that do not condition our interpretations. The problem with that is, that once you know something, you can’t ‘unknow’ it.
Sir Nigel was very clear in his suggestions for avoiding cognitive bias. He explained that the expert:
- Should seek to regulate the information they are exposed to, and to avoid emotive information.
- Must set out the substance of all the facts which are material to their opinion.
- Must be aware of the risks and continually challenge themselves and their colleagues.
- Must record full reasons for their judgement contemporaneously.
Points 2 – 4 are routinely carried out and form part of our regular checking procedures, but point one is problematic, as I have already outlined. As forensic experts, we often need to trawl through pages of suspect and victim interview transcript, for example, to find the relevant conditioning information. It’s very difficult to avoid that extra information, especially the emotive information surrounding a serious sexual assault or murder case. That information almost certainly stirs the emotions.
So what? We’re experts, so we can act objectively regardless of the extra information, right? Wrong! Professor Saul was very clear in her analysis of cognitive bias and emphasised that expertise is no protection. People routinely overestimate their own objectivity. In fact, assuming that you have superior objectivity makes your implicit bias worse.
What then, is the solution? Firstly, these biases are unconscious. We should not be blamed for them. However, this knowledge challenges our ability to function as experts. Professor Saul reassures that action is possible if we are aware of the problems of cognitive bias, although she warns that trying not to be biased is usually counter productive, only resulting in more bias. Our first action should be to avoid the information that might cause the bias in the first place. We don’t need to give up on ideas of knowledge and expertise, but we do need to accept that we are human beings who have faulty cognitive instruments.
For more information on Professor Saul’s collaborative project on bias, investigate www.biasproject.org
Filed under: Forensic Science





