What they saw that night: The problems with eyewitness testimony
The following is the text for my talk for the 2013 Jack the Ripper Conference in London:
To begin I’d like to briefly recap with some of the previous events of the weekend. Yesterday when leaving Shoreditch Town Hall you will have witnessed an argument in the street between a man and a woman. Hopefully at the time it will have attracted your attention and you will have taken notice of the couple involved. Hopefully, you will not have thought of it much since yesterday until this morning when upon your arrival you will have been reminded about the event and asked to fill in questionnaires about what you saw. Many thanks to those of you who did so and returned them during the coffee break, since we have since been collating the answers for the presentation you are about to see.
As you have probably realised, the couple you witnessed on Saturday were in fact actors and the argument scripted and carefully choreographed for us to provide a demonstration for you on the accuracy of eyewitness testimony. Once we have reviewed the results we will look at some of the science behind memory and eyewitness identification and then apply it to the Whitechapel case, to see how far we can rely on some of the key eyewitness evidence and how it affects the evidence supporting certain suspect theories. First though, I’d like to briefly hold our own police line up, and see how Ripperologists would have fared during our own Seaside Home identification. As you may have noticed already from the questionnaires, to save time we are focusing purely on the male (as the Ripper himself was most likely male).
To make it a bit fairer, I have included 3 suspects, out of which I’d like you to pick out our man from yesterday.
Sorry! Wrong slide.
A, B or C.
Hold vote by show of hands.
Thank you everyone who voted. So by popular opinion, Suspect B was the man seen on Saturday afternoon. Now let us take a look at the results of todays questionnaires.
41 People took part (correct answers in bold):
Clean shaven: 14
Full beard: 1
Eastern European: 4
Baseball cap: 4
No hat: 1
So to reveal all – our suspect was actually:
And here is a picture of him. And yes, I cheated. The man was not in the suspect line up, but many of you voted nonetheless.
So the results of the demonstration show us that although some details were correctly remembered by the majority, significant errors were still present and some details the majority had wrong.
But of course, this isn’t definitive evidence about the accuracy of eyewitness testimony. Not all of you took part, you probably aren’t a good cross section of society and we didn’t have very tight scientific control of the demonstration. But fortunately, this is a subject that has been analysed many times by psychologists who have found much scientific evidence, which we will shortly examine.
But is the accuracy of eyewitness testimony important? Yes, Even today with the advent of CCTV, DNA, fingerprinting and other scientific evidence, eyewitness testimony is one of the major sources of evidence for criminal convictions. It is also one of the most unreliable. The Innocence Project estimates that around 75% of convictions where the accused is exonerated by DNA evidence is the result of faulty eyewitness testimony. The fallacy of this area is perhaps best reflected that it is one of the few topics upon which courts will allow psychologists to act as expert witnesses and educate the jury.
To understand how errors can creep into memories of an event or person, it is first important to understand how memory works. Put simply, memory has three stages:
- Acquisition – Where the event is perceived
- Retention – Where the event perceived becomes resident in memory
- Retrieval – Memory searched and pertinent information retrieved and communicated
Errors can occur at any one of these stages and caused by various factors. I’ll briefly go through some of these factors (which are particularly relevant to the Ripper case) and examine some of the evidence for them.
Factors affecting acquisition:
- Exposure – a study by Laughery and others in 1971 showed two groups, four slides of human faces. One group viewed the slides for 2 and a half seconds each, whereas the other group viewed them for 8 seconds. 8 Minutes later they were asked to identify these faces, and the longer exposure group were much more accurate in doing so.
- Detail salience – not all details of an event are equally memorable. Marshall and his colleagues in 1971 showed their subjects a two minute film and then interviewed them about what they saw. Certain details of the film were remembered by nearly everyone, whereas other details were remembered by nearly no one, indicating that some things are simply not memorable.
- Violence of the event – in a 1978 study, Clifford and Scott showed two groups of subject two different films. One film contained two policemen searching for a criminal, being aided by a reluctant third person, whom they persuade verbally to assist them. The second version features a similar plot, but where the third person is coerced by being violently assaulted by one of the police officers into helping. Other than this key event, the films were identical. Those who saw the violent film recalled significantly less details about the film than those who had watched the non violent version, they theorised this may have been because of the stress of witnessing a violent event. Which leads us nicely onto Stress as its own factor.
- Stress – Previous factors have been more focused on the event, but this is focused on the witness and their mindset. There is much scientific evidence focusing on the effect of stress on various tasks in animals (and a few more ethically dubious ones doing the same to humans), but due to scientific ethics it is difficult to morally justify inflicting extreme stress on a human in order to measure how it effects their memory. One notable study from 1976 devised by Johnson and Scott did so by holding an experiment where they either had the subject hear an innocuous conversation about equipment failure, followed by an actor entering the room with greasy hands holding a pen, or another scenario where they witnessed a violent argument followed by an actor entering the room with bloody hands and holding a knife. In the weapon scenario, subjects were less likely to be able to identify details or identify the actor.
- Expectations – another internal factor that shows that a persons expectations about an event can effect how they perceive it.
- Perceptual Activity – When an eyewitness witnesses an event, are they actually paying attention to relevant details or do other activities have their attention. For example, if someone is witnessing an armed robbery are they focusing on the robbers face or are they too busy trying to think of a way to escape unharmed.
Next we move on to:
Factors Affecting Retention
- Time delay – Marshall conducted a study in 1966 where he showed subjects a short film and interviewed them either straight away or after a week. As expected, a week later the subjects correctly remembered far less details than those who were interviewed straight away.
- Post event information – information we are later given about an event we have witnessed can effect our memory of it. An example of this comes from a paper from 1927 by Bird. A university lecture was held, where the lecturer discussed the results of a series of experiments. A reporter was present, who wrote an article on the lecture for the local newspaper that was riddled with errors and inaccuracies. A week later an exam was given on the information from the lecture, but those who had read the newspaper article answered with the inaccuracies they had read in the article having misremembered the information given in the lecture as a result. The same could quite easily happen should a witness to an event read a newspaper report containing an error or something they had not seen.
- Leading and misleading questions – When questions asked by an interviewer lead the rememberer to an answer, or mislead them into misremembering a detail about the event. We will focus on this in more detail later.
Finally we have:
Factors affecting retrieval
- Types of queries – Witnesses tend to be more accurate, but less complete when asked to produce a narrative of an event when compared to when they are answering direct questions. A wide range of studies including Cady in 1924, Marquis and colleagues in 1972, Lipton in 1977 and Snee and Lush in 1944 all found their subjects produced more errors when asked direct questions about an event they witnessed than compared to when they simply narrated what happened. These studies indicate that a witness should be allowed to freely narrate an event, and then answer direct questions, lest the questions interfere or contaminate the memory and cause misremembering (for example, if a person is asked if there was a gun during a robbery when none was present, they may subsequently remember “something about a gun” when later free recalling the event).
- Wording of a question – The way a question is worded can greatly influence how the question is answered. For example, a study by Harris in 1973 asked people differently phrased questions such as “How tall was the basketball player?” and “How short was the basketball player?” or “How long was the movie?” and “How short was the movie?”. The way these seemingly contextually identical questions are worded had an effect on the answers, with those who were asked the second question estimating significantly lower the height of the basketball player or the length of a movie. A study by Loftus and Palmer in 1974 showed participants a video of an automobile accident. Participants were asked to estimate the speed the car was going when it was involved in the collision, with different groups being asked a slightly different worded question of “About how fast were the cars going when they smashed/collided/bumped/hit/contacted each other”, with those given the word “smashed” estimating higher than those given the word contacted – on average about 10 mph.
Both of these factors can lead to misremembering due to questions being misleading (and often quite innocently). We briefly mentioned this possibility when looking at factors affecting retention, but we will examine it in detail now.
Misremembering and the misinformation effect
We have already discussed how memory is malleable and how when retrieving a memory we reconstruct a series of events. Misinformation can be placed in memory due to leading (or misleading) or poorly worded questions and this can later be present in subsequent recalls, indistinguishable from the actual event that occured.
Many studies (Mostly by the researcher Elizabeth Loftus who performed the car crash study I just described) have demonstrated that witnesses will falsely remember details thanks to misleading questions.
- A “Yield” sign not a stop sign
- Nonexistent broken glass during a car accident
- Nonexistent objects such as tape measures.
- Straight hair as being curly
- Moustaches on clean shaven men
- A barn being present in an unspoilt countryside location with no buildings.
In February 1978 David Webb was released after serving 10 months of a 50 year sentence for rape, attempted rape and attempted robbery.
Webb was arrested because he resembled a composite picture of the suspect, and was positively identified by eyewitnesses at his trial. However, there were some inconsistencies in the testimony of the eyewitnesses and Webb also had alibi witnesses. Despite this, he was found guilty and imprisoned for 50 years.
Several months later, another man confessed to the crimes for which Webb was found guilty and when placed in a line up together, witnesses no longer could positively identify Webb as the culprit.
The police themselves acknowledged that the two men looked similar, but Webb himself did not think the true culprit looked anything like him.
This case is just the tip of the iceberg of mistaken identities in criminal cases. A large scale study conducted by Buckhout in 1975 had tv viewers view a short film of a mugging and then had viewers call in with their choices from a suspect line up. Out of over 2000 callers, over 1800 picked the wrong suspect. 14% picked the correct man – which is slightly lower than the probability of guessing.
But why is recognising faces so difficult? What factors in addition to the ones we have already discussed relating to memory play a part?
- Race – race is a major factor in being able to successfully identity the face of a suspect. Research has found we are far less successful at identifying the faces of members of other races or ethnic groups. In 2001 Meissner and Brigham conducted a meta analysis of 30 years worth of research into cross racial identification. They analysed the results of 39 research articles, containing 91 sample groups and nearly 5000 participants. They found that there is a significant own race bias when it comes to facial identification – that you are more likely to be able to successfully identify a person of the same race as you and less likely to make a false identification, when compared to someone of a different race. The evidence seems to suggest that this isn’t based on any discriminatory views the person may hold, and exposure to faces of people from different races may slightly improve a persons accuracy, but on the most part it seems to be biological.
- Unconscious transference – if a person is already familiar to a witness, they mistaking lay recognise them as being the perpetrator. For example, Wall (1965) presented a case where a railway ticket seller identified a sailor as being an armed robber who had held him up. The sailor however had an iron clad alibi, and it later transpired t he was based nearby and had bought tickets from the ticket agent on 3 occasions, so presumably the ticket agent seeing him in a line up and recognising him, assumed he must have been the armed robber. Another example is a study by Buckhout in 1974 where 40% participants who had witnessed a staged assault on a professor mistakenly picked an innocent bystander as being the perpetrator from a photo line up.
- Lineups – often the witness will assume that the perpetrator is present (after all, why would police hold a lineup if they weren’t fairly certain they had the criminal), so guess based on a “best match”. This can obviously be even more dangerous if the lineup is not fair (ie one person particularly stands out) or only one suspect is shown.
What is most interesting is that the research showing the fallibility of eyewitness memory and the potential to misremember events has been highly consistent across the 100 or so years psychology has examined it. In 1908 Hugo Munsterberg detailed an experiment that took place in Berlin in 1902. I’ll now quote from his at the time controversial work “The memory of the witness”:
“A few years ago a painful scene occurred in Berlin, in the University Seminary of Professor von Liszt, the famous criminologist. The Professor had spoken about a book. One of the older students suddenly shouts, “I wanted to throw light on the matter from the standpoint of Christian morality!” Another student throws in, “I cannot stand that!” The first starts up, exclaiming, “You have insulted me!” The second clenches his fist and cries, “If you say another word –” The first draws a revolver. The second rushes madly upon him. The Professor steps between them and, as he grasps the man’s arm, the revolver goes off. General uproar. In that moment Professor Liszt secures order and asks a part of the students to write an exact account of all that has happened. The whole had been a comedy, carefully planned and rehearsed by the three actors for the purpose of studying the exactitude of observation and recollection. Those who did not write the report at once were, part of them, asked to write it the next day or a week later; and others had to depose their observations under cross-examination. The whole objective performance was cut up into fourteen little parts which referred partly to actions, partly to words. As mistakes there were counted the omissions, the wrong additions and the alterations. The smallest number of mistakes gave twenty-six per cent. of erroneous statements; the largest was eighty per cent. The reports with reference to the second half of the performance, which was more strongly emotional, gave an average of fifteen per cent. more mistakes than those of the first half. Words were put into the mouths of men who had been silent spectators during the whole short episode; actions were attributed to the chief participants of which not the slightest trace existed; and essential parts of the tragi-comedy were completely eliminated from the memory of a number of witnesses.”
More than 100 years later psychologists are still examining the reliability of eyewitnesses, providing techniques to improve this and raising questions about whether we can trust the memory of what people have seen – or think they have seen. So I ask the question now – can we truly trust any of the eyewitnesses who claim to have seen Jack the Ripper and trust any identification they may have made?
Elizabeth Long claims to have seen Annie Chapman with a man shortly before her murder. She doesn’t seem to be able to describe much more than his clothes, but she describes him as looking “foreign” despite claiming not to have seen his face. Perhaps her memory of the event had been influenced by the newspaper gossip that Jack the Ripper must be a foreigner or a Jew?
Israel Schwartz witnessed a woman being assaulted at the entrance to Dutfields Yard shortly before the murder of Elizabeth Stride.
- Schwartz is walking South down Berner Street from Commercial Street at 12.45am
- Street would have been dark, but not pitch black.
- He see’s a man assaulting a woman at entrance to Dutfields Yard. This is “broad shouldered man”
- Altercation obviously caught his attention as he crossed road to avoid it.
- Turned back to look as he stepped off the curb on other side of the road and then he saw “Pipeman”.
- Pipeman also caught his attention, by shouting “Lipski” and pursuing Shwartz who ran. Some sources report Schwartz saying Pipeman had a knife, others a pipe.
- Schwartz would only have had a few seconds to get a look at the man with the woman assumed to be Stride and also Pipeman.
- With Pipeman, if there was a weapon there may have been weapon focus
- Would Schwartz have had enough time to see relevant details and encode them? Could encoding or recall be affected by other factors? Stress, weapon focus, etc.
- Schwartz needed a translator and gave different story to The Star than to police. This could affect the reliability of his statement.
- Stride’s murder is also the “odd one out” in the Ripper case – what if her murder on the night of the double event is just a coincidence and she is not a Ripper victim? Does what Schwartz saw possibly support this interpretation?
Later that same night Joseph Lawende, Joseph Levy and Harry Harries see a woman believed to be Catherine Eddowes at the entrance to Church Passage with a man. She had her hand on the man’s chest, but not to push him away.
- Lawende, Levy and Harries were across the street from where Eddowes and the man stood in the entrance to Church Passage.
- As they passed, something about the couple evidently drew attention to the men as Levy stated “I don’t like going home by myself when I see these sorts of characters about.” But he did not pay any further attention to them or provide a description. Harry Harries saw nothing at all.
- Lawende on the other hand apparently had a fairly good look at the couple and later identified Eddowes clothes at the mortuary.
- He provided a description of the man, but was adamant he would not recognise him again.
- Where the couple were standing would have been fairly well lit as they were standing directly underneath the lamp at the entrance to Church Passage.
- Other than Levy’s comments about how distasteful the couple were, there is little to draw Lawende’s attention to them and he would only have seen them for moments across the street (Lawende estimates the distance as about 15 or 16 feet) as he walked by.
- Despite saying he would not recognise the man again, Lawende is often described as being the most reliable witness and was called to identify James Sadler as being Jack the Ripper during the Frances Coles investigation in 1891
George Hutchinson claims to have seen Mary Kelly with a man outside Millers Court. He gives a very detailed description to the police several days after the murders.
- Hutchinson is speaking to Mary Kelly on Commercial Street, she leaves heading towards Thrawl street when she is approached by a man.
- They speak for a few moments while Hutchinson watches from further down the street, standing underneath a lamp outside The Queens Head.
- They turn and walk towards him, and as they walk past, Hutchinson deliberately stoops down to look at his face, which is returned by a stern look.
- Hutchinson then follows them down Dorset Street and watches them entering the court. He stands by the entrance of the court for 45 minutes before leaving.
- Hutchinson gave his statement to Abberline who seemed to believe him. But it raises so many questions of why!
- Why did Hutchinson watch them? What drew his attention? Why did he try and look a the mans face? Why did he follow them? Why did he stand outside Millers Court for three quarters of an hour?
- Hutchinson’s statement is incredibly detailed – so is it accurate? Evidence suggests that the most detailed witness descriptions are rarely the most accurate. There is also delay. Hutchinson did not come forward until the 12th November – after the press coverage and inquest. Could that have influenced his description?
- We must also consider that maybe it is not even true. Other witnesses state they saw a man standing watching the entrance to Millers Court – was this Hutchinson and he is just providing himself an alibi? Had he been “dining out” telling people he had seen the Ripper that night for several days and now felt obliged to go to the police to avoid suspicion?
- Regardless, there is something very fishy about Hutchinson – how he acted, why would he bother to acquire and retain the information, the detail provided and the delay involved.
Before we move on to the Seaside Home Identification, I’d like to briefly move away from the Ripper case for a moment and look at a modern real life example where eyewitness testimony has raised questions about accuracy – particularly in regards to subsequent identification of the suspect.
In 2001 Abdelbaset al-Megrahi was convicted of the bombing of Pan Am Flight 103 in 1988. The Lockerbie Bomber (as he has become known) was convicted mainly based on the evidence of a Maltese clothing shop owner named Tony Gauci who linked Megrahi to the improvised explosive device used in the bombing, having apparently sold the clothes to Megrahi that the bomb was wrapped in. Gauci testified to having sold the clothes to a man of Libyan appearance, but over 23 interviews gave contradictory evidence as to the purchasers age, appearance, and the date of purchase. According to one official report, four days before identifying Megrahi for the first time, Gauci was apparently shown a magazine containing a picture of Megrahi and linking him with the bombing. In court he was shown the same magazine picture and asked if he had identified the man in the picture as being the man who purchased the clothes and was he in court. He answered: “He is the man on this side. He resembles him a lot”.
Before Megrahi was transfered to Libya on compassionate grounds in 2009, he was in the process of preparing an appeal against his conviction. As part of his appeal was a report prepared by Professor David Canter analysing the eyewitness testimony of Tony Gauci.
In the summary of his report, Canter states the following opinions:
- There are two distinct aspects of Mr Gauci’s memories that can be considered in the light of what is now known of the psychology of memory. One is the recall of the alleged event in his shop and the details of what happened then. The other is his alleged recognition of Mr Megrahi as a person who had come into his shop. These are rather different aspects of memory, but in both cases I consider there are grounds for doubting that his recall or his recognition are of the events or person as Mr Gauci claims.
- Mr Gauci’s reported memory is of one particular customer and sales event out of many others, recalled, initially, at a point ten months after the alleged event had occurred. In the absence of any particular features, at the time the event happened, that would render the event highly distinctive or emotionally significant to Mr Gauci, the scientific understanding of the rapid decay of human memory soon after the event being remembered would suggest that it is extremely unlikely the details of such a memory are accurate. The decay of memory is also inconsistent with any claims that in court, ten years after the incident, Mr Gauci could have a more accurate memory than closer to the time of the event.
- As well as memory for the event being unlikely, recognition of the individual alleged to be involved and the accurate identification, in such circumstances, of a strange individual more than 2 years later, is highly unlikely.
- The uninformed view that memory is open to improvement after the initial event by simple, effortful thinking on subsequent occasions is not supported by scientific research. Enrichment of the details of what is remembered is only possible by intensive psychological techniques including context re- instatement, which were not utilised, as far as I can tell, in relation to the central features of this case. Furthermore, such improvements tend towards providing additional detail rather than contradicting earlier accounts. Therefore the inconsistencies in Mr Gauci’s testimony, that include a number of changes in details of alleged fact, across the seventeen statements I have examined, are more in accord with a vague memory that Mr Gauci is keen to make clearer than the developing of a more accurate memory of an actual incident.
- It is now well established by psychologists that what a witness genuinely claims to remember, with confidence, is open to influence from various sources that will lead to distortions in the actual memory, such as implicitly leading questioning and prior exposure in relation to material that could influence identification of a suspect. So, although working with limited material that provides little information on the police interview process, I do consider that the variations and inconsistencies in Mr Gauci’s accounts of what he remembers are in accord with the recognised processes of influence on memory.
- The processes at play in the present case are probably more powerful because of Mr Gauci’s apparent desire to please and his recognition of the importance of his testimony. This could give rise to a phenomenon known as the reduction of ‘cognitive dissonance’ in which a person adjusts his cognitions to accord with his actions. In this case unconsciously adjusting his confidence in his opinion to make it coherent with his commitment to helping the police enquiry.
The report by Canter illustrates quite nicely some of the theoretical concerns with eyewitness memory we have already examined and also links us nicely on to:
Seaside Home Police Identification
According to the Swanson Marinalia – that is the copy of The Lighter Side of my Official Life by Sir Robert Anderson, owned by former Superindendent Donald Swanson where he made extensive notes in the margins – Aaron Kosminski was identified as being Jack the Ripper by a witness in an identity parade held at the “Seaside Home”. When the witness discovered that the suspect was a fellow Jew he reportedly refused to testify against him in court.
The two main potential witnesses are Israel Schwartz and Joseph Lawende – both of whose eyewitness reports raise some serious questions as we have previously seen, and at least one of whom claimed they would not recognise the man if they saw him again – if Lawende was the witness could it be that he did not refuse to testify because the man was a Jew, but because he wasn’t certain in his own mind that this was definitely the man he saw, so did not want to hang an innocent man?
Was this line up good practice? In short no. By giving the witness only one suspect to pick out as Jack the Ripper the police were inadvertently telling the witness that this was Jack the Ripper and merely asking them to confirm the fact. Research has shown that witnesses automatically assume the guilty party will be in a line up.
In short, the Seaside Home Police Identification is very flimsy evidence, and even if Schwartz or Lawende did positively identify Aaron Kosminski it does not automatically make him Jack the Ripper.
The Jack the Ripper Photofit
Finally, quite a few recent theories have used the e-fit based on eyewitness descriptions of the killer that was produced for a 2005 television documentary as part of the evidence that their favoured suspect was Jack the Ripper.
I am now doing the same. 125 years since the final Jack the Ripper murder and today I will exclusively reveal who the killer was using the e-fit of Jack the Ripper as conclusive proof. Are you ready?
Here was a joke where the photo of the e-fit merged into Freddie Mercury, while Queen’s “I want to break free” played.
Yes, that is correct. Jack the Ripper stopped killing in November 1888 as he took up song writing instead and 82 years later would go on to form the rock band Queen.
In all seriousness, I don’t think any of the eyewitness descriptions contain detailed enough facial descriptions to produce that accurate a likeness. Also in a real police case the witness themselves will choose the features for the photofit, not simply an artist working using a vague statement. In short, I don’t think that this photo takes us anywhere with the case and certainly shouldn’t be considered to be evidence promoting or exonerating a suspect.
- Eyewitness memory is not always reliable. Memory can be adversely effected at any stage.
- The main eyewitness testimony we rely on in the case should be treated with caution. As should the seaside home identification.
- Freddie Mercury was probably not actually Jack the Ripper.
We do not know for sure how this I D at the Seaside home was conducted. A line up? Doesn’t seem like it to me as the word ‘confronted’ was used by Anderson (“the moment he was confronted with him”). This was a confrontation I. D.
Now this serves, primarily, as a means of unsettling a suspect rather than securing an identity. In other words this act was more to try and gain more information from the suspect, or even a confession, rather than ascertaining that the witness saw the suspect.
This can be very effective and was common for the period, so in terms of ‘good practice”, it depends on what you are referring to. Identification or ascertaining guilt/confession?