Chapter 3 - Eyewitness Identification and Testimony

I. Introduction

Eyewitness identification and testimony is a critical component of the criminal justice system. It is used to guide police investigations and prosecutions and can be compelling evidence in a courtroom. However, even the most well-meaning and confident eyewitness can be mistaken. The U.S. Innocence Project estimates that eyewitness error was a contributing cause in 70 percent of the 356 wrongfully convicted accused who have been exonerated by DNA evidence in that country.^{Footnote 54} The American Psychological Association estimates that one in three eyewitnesses make an erroneous identification.^{Footnote 55} As Mr. Justice David Doherty of the Ontario Court of Appeal famously put it:

[T]he spectre of erroneous convictions based on honest and convincing, but mistaken, eyewitness identification haunts the criminal law.^{Footnote 56}

As described in the 2011 Report, there is considerable research that has documented that eyewitness misidentification has long been regarded as the leading cause of wrongful convictions. Eyewitnesses can make mistakes and their memories can be impacted by a variety of factors, including being inadvertently influenced or biased by police officers and police procedures. Eyewitness misidentification, especially in the early stages of an investigation, can set in motion a chain of errors causing an innocent person to be charged and leading the trier of fact toward a guilty verdict. It is the criminal justice system’s responsibility to help eyewitnesses make the most accurate identification possible. As stated in the 2011 Report, eyewitnesses, law enforcement and the public at large will benefit from identification procedures that are designed according to scientific research and conducted consistently nationwide. This has not changed.

Eyewitness error has been studied by social scientists for more than 100 years and there is a vast body of research on the subject.^{Footnote 57} Environmental factors such as distance and poor lighting, as well as physical aspects such as witness vision, hearing, attention and intoxication, all have evident effects on eyewitness accuracy. The memory recall process itself has also been identified as another important factor to consider in this discussion.

Although it is common to think of memory as a series of photos or recordings in someone’s mind, psychology tells us it is actually a reconstructive, malleable process, which occurs in three stages. First, when an individual is exposed to a piece of information or an event, the memory is encoded; it is at this stage when environmental factors and physical aspects of the witness can affect the accuracy of the memory. The memory then goes through the second stage, the retention stage, where it is stored before reaching the third phase, memory retrieval. Although a witness can cycle back and forth between the retention and retrieval phase, a memory is only encoded once, and it decays the longer it is retained. During this time, the witness may also be exposed to post-event information which may not be accurate, but may distort the memory nonetheless. When this occurs, the witness is unaware that the memory has been altered, and it is difficult if not impossible to restore it to its “original” version.^{Footnote 58}

Despite such challenges, there are a number of best practices and legal safeguards in place across the criminal justice system to minimize the effects of eyewitness error when it occurs.

II. 2011 Recommendations

While the 2005 Report’s recommendations remain valid, the Subcommittee recommended slight refinements as follows:

1. The following are reasonable standards and practices that should be implemented and integrated by all police agencies:
   1. If possible, an officer who is independent of the investigation should be in charge of the lineup or photo-pack presentation. This officer should not know who the suspect is, avoiding the possibility of inadvertent hints or reactions that could lead the witness before the identification takes place, or increase the witness’s degree of confidence afterward.
   2. The witness should be advised that the actual perpetrator may not be in the lineup or photo-pack, and therefore the witness should not feel that they must make an identification.
   3. The suspect should not stand out in the lineup or photo-pack as being different from the others, based on the eyewitness’s previous description of the perpetrator, or based on other factors that would draw extra attention to the suspect.
   4. All of the witness’s comments and statements made during the lineup or photo-pack viewing should be recorded verbatim, either in writing or if feasible and practicable, by audio or videotaping.
   5. If the identification process occurs on police premises, reasonable steps should be taken to remove the witness on completion of the lineup to prevent any potential feedback by other officers involved in the investigation and cross contamination by contact with other witnesses.
   6. Show-ups should be used only in rare circumstances, such as when the suspect is apprehended near the crime scene shortly after the event.
   7. A photo-pack should be provided sequentially, and not as a package, thus preventing ‘relative judgments’.
2. For prosecutors, the following practical suggestions should be considered:
   1. Assume the identity of the accused is always at issue unless the defence specifically admits it on the record. Timely preparation and a critical review of all of the available identification evidence, including the manner in which it was obtained, is required as it will affect the conduct and quality of the trial.
   2. Allow the witness a reasonable opportunity to review all previously given statements and confirm that the statements were accurate and a true reflection of their observations at the time. Carefully canvass the full range of the indicia of the identification, including any distinguishing features that augment this evidence. Remember that it is the collective impact of all of the evidence that will be considered in support of a conviction. Defects in one witness’s identification can be overcome by the consideration of other evidence.
   3. Never interview witnesses collectively. Never prompt or coach a witness by offering clues or hints about the identity of the accused in court. Do not condone or participate in a “show-up” lineup. Never show a witness an isolated photograph or image of an accused during the interview.
   4. When meeting with witnesses in serious cases, it is wise, if it is feasible and practical, to have a third party present to ensure there is no later disagreement about what took place at the meeting.
   5. Never tell a witness that lihey are right or wrong in their identification.
   6. Remember that dpsclosure is a continuing obligation. All inculpatory and exculpatory evidence must be disclosed to the defence in a timely fashion. In the event that a witness materially changes their original statement, by offering more or recanting previously given information during an interview, the defence must be told. In these circumstances, it would be prudent to enlist the services of a police officer to record a further statement in writing setting out these material changes.
   7. Always lead evidence of the history of the identification. It is vitally important that the trier of fact not only be told of the identification but also all the circumstances involved in obtaining it, e.g. the composition of the photo-pack. Be wary of prosecutions based on weak single-witness identifications. While not required by law to secure a conviction, ascertain whether there is any corroboration of an eyewitness’s identification in order to overcome any deficiencies in the quality of that evidence.
3. The use of expert evidence on the frailties of eyewitness identification is redundant and unnecessary in the fact-finding process. A proper charge and caution by the trial judge can best deal with the inherent dangers of identification evidence.
4. Workshops on proper interviewing should be incorporated in regular and ongoing training sessions for police and prosecutors.
5. Presentations on the perils of eyewitness misidentifications, including presentations by experts in the field of memory and eyewitness identification, should be incorporated in regular and ongoing training sessions for police and prosecutors.

III. Status of Recommendations

Chapter 7 of this Report highlights a recent survey which sought to determine if police and prosecutor training across Canada contained wrongful conviction course content, and eyewitness identification content specifically. Eighty-five percent of responding police agencies and 50 percent of responding police training institutes indicated that their courses contained information about eyewitness identification. Sixty-three percent of responding prosecution services indicated their training contained this information. This suggests that while the majority of the responding agencies are providing information about the role of eyewitness error in wrongful convictions, further strides can be taken in disseminating this important information. Reviewing the standards and practices listed above is useful information for police and prosecutors, and is even more effective when incorporated into early training initiatives.

IV. Court Commentary on Eyewitness Identification and Testimony Since 2011 Report

A review of eyewitness evidence and commentary by Canadian courts since 2011 suggests that judges are certainly aware of the dangers associated with this kind of testimony, and the need for the court to seek strong confirmatory evidence in such cases to prevent a wrongful conviction. While the cases and commentary cited below deal with different aspects of eyewitness identification evidence, one similarity is evident: witnesses are using social media and their smartphones to help identify the alleged perpetrator - sometimes prior to viewing a police lineup or photo array. Furthermore, this behavior appears to be “tainting” or “contaminating” eyewitness descriptions over time, resulting in the eyewitness providing a description of the alleged perpetrator at trial that is far more detailed than the description that was originally given to police, immediately following the incident.

The justice system will increasingly have to wrestle with the issues posed by witnesses who conduct post-offence research via social media in order to identify suspects. Witnesses searching for and identifying suspects through the internet are doing so without the safeguards employed by police agencies, and their searches are not being video-recorded, making the trier of fact unable to review the process. When this evidence is presented to police agencies, it is incumbent upon them to take a statement from the witness documenting how the identification was made, and not simply accept it at face value. Though this identification evidence is obtained in circumstances that lack the markers of reliability identified in the Sophonow Inquiry, it can still be pertinent and weighed by judges.

a) R. v. Leeds, 2013 NSSC 364^{Footnote 59}

The accused in this case was charged with manslaughter following an assault in an alleyway after a New Year’s Eve party. The identity of the perpetrator was the sole issue at trial, and the only identification evidence was that of one eyewitness, the girlfriend of the deceased. This decision is particularly illuminating as it deals with eyewitness identification evidence at length, outlining variables such as the environmental factors of the crime scene (e.g., poor lighting) and physical aspects of the eyewitness (e.g., level of intoxication) as well as how the photo arrays were assembled and presented. The commentary offered below is related to the possible contamination of the eyewitness by exposure to members of the deceased’s family (both in person and electronically), immediately following the assault. The accused was found not guilty.

[14] Eye witness identification evidence must be based on the independent recollection of the witness and not recollection arising as the result of discussions with and amongst various people. Such evidence may be compromised where an eye witness has discussed with others his or her recollection of the person’s appearance before making an identification: R. v. Holden (2001), 2001 CanLII 14562 (ON CA), 56 O.R. (3d) 119 (C.A.) at p.136-137. In some cases, the failure to mention distinctive characteristics of a suspect in an initial description to the police may be quite material to the reliability of the identification.

[39] I do not accept that [the witness’s] identification of the accused was truly her independent recollection of the person she saw. The different descriptions she gave to the first responding police officers: the fact that she was in a room at the hospital with family and friends of the deceased for several hours and discussed the identity of the assailant with them; the fact that the accused was unknown to her at the time of the incident but that she was nonetheless able to give the police a nickname and that she and another person who had been at the hospital attended the police station together and were kept together before viewing the photo lineup. These factors all speak to the reliability and independence of [the witness’s] recollection.

[40] My conclusion regarding the reliability and independence of [the witness’s] identification evidence is further buttressed by the fact that while viewing the photographs at the police station [the witness] had her cell phone in hand and appeared to be looking at it while being shown the photographs. This was noted by [the detective] who was showing her the photographs and at one point he asked her to put the phone away.

[60] The fact that [the witness’s] description became more detailed over time is of concern. It speaks of the tainting which occurred as a result of her discussions with others about the identity of the perpetrator.

b) R. v. Mohamed, 2014 ABCA 398^{Footnote 60}

The Alberta Court of Appeal discussed the identification evidence of the key witness, specifically the viewing of Facebook photographs shown to him by an acquaintance, where he was led to believe by the acquaintance that the suspect would be present in the photographs. The Court of Appeal distinguished this process from the process of a police-conducted photo-lineup, commenting:

[30] It is better compared to a situation where an eyewitness who does not know a shooter at the time of the offence later identifies him on the street after a third party has drawn attention to him. Such evidence is not entirely without weight simply because it was not obtained in the context of the safeguards provided in a police lineup. Rather, “it is what it is”, and the trial judge has the obligation to assess it in the context of all the evidence to determine its degree of reliability.

c) R. v. Jobe, 2016 NSSC 282^{Footnote 61}

This trial involved three co-accused who were charged in an armed robbery at a hotel. The Court found that eyewitness identification was the foundation of the Crown’s case and thereby exercised “special caution” in considering it. The three accused were found not guilty.

[20] At this point in her direct testimony, [the witness] began to describe the three men…I found that throughout her direct examination [the witness’s] description of the men varied. The difficulty with these varying descriptions became compounded when [the witness] allowed that some of her recollections were based upon social media and other exchanges in the roughly one year since the robbery. In this regard, I accept Defence counsels’ submissions that this aspect of [the witness’s] testimony, which they term crowd-sourced, is full of hearsay and must be disregarded.

[24] The fact that [the witness’s] description of these individuals became more detailed over time is of concern. It speaks of the tainting which occurred as a result of her social media and other exchanges about the possible identities of the robbers…

d) R. v. R. P., 2017 ONCJ 743^{Footnote 62}

This case concerned an underage complainant/witness who was sexually propositioned by a man who was unknown to her, but lived in her community. The Court found that the complainant’s identification evidence, which largely resulted from a Facebook search several weeks later, had insufficient reliability to resolve the reasonable doubt within the trial judge’s mind. The accused was acquitted.

[8] The complainant told the police that she found the picture on Facebook three or four weeks earlier but did not report it because “I didn’t know if it was super important”. Eventually, she told her father and he took her to the police station. Before this happened, the complainant saw the man again. She explained she was in her father’s car and they passed him on the street. She pointed him out to her father and he said, “That’s who I thought it was”.

[9] The complainant acknowledged her encounter with the man lasted “one minute or so” and that she had never seen him before. She searched Facebook for him after she overheard her father tell her step-mother that the culprit was [R.P]. The discovery of his picture happened 11 to 18 days after the encounter. The complainant testified she was certain the culprit is the man in the photograph.

[11] The Crown concedes the dangers inherent in “stranger identification” cases and that the interaction between the parties was brief. Counsel points out, however, that there was sufficient opportunity for the complainant to observe and later relate a detailed description of the man and his clothing as well as a residential address. On overhearing the name of the suspect, she immediately recognized him on Facebook. Counsel submits that other than this suggestion by her father, there is no other “tainting” of her identification.

[12] The Defence submits that it is problematic that the identification in this case is based upon a single picture found on social media by the complainant after searching a name she overheard her father say was the culprit. Counsel argues that the latter point is especially significant given the close relationship between father and daughter. That the complainant trusts her father is evident in the video record of the police statement. Counsel also points out that the interaction between the complainant and the culprit was brief and the evidence of the defendant’s address is not sufficiently compelling.

e) R. v. Assefa, 2018 ABCA 6^{Footnote 63}

This was the appeal of a man convicted of assault at a house party. The morning after the house party, the victim used Instagram to identify the accused, who had attended the party with someone the victim knew. The victim then sent the Instagram photos to police, which were entered as exhibits at trial. The Court of Appeal upheld the conviction, as the trial judge ruled the identification by the victim was recognition evidence, rather than the proverbial “fleeting glance.”

[22] As the Crown points out, there is a significant difference between the identification of a stranger and identification of someone the witness is able to recognize. The New Brunswick Court of Appeal has recently commented that “where an eyewitness has no previous knowledge of the identity of a suspect, the circumstances surrounding the encounter take on much more meaning”: R. v. Arsenault, 2016 NBCA 47 (CanLII) at para 30, 342 CCC (3d) 322.

[23] The trial judge here took all of the circumstances surrounding the encounter and Gardner’s identification of the assailant into account. When Gardner reviewed Edgar’s Instagram photos, he was looking for the face of a person he had seen and recognized throughout the evening in the company of Edgar, a man he knew and had spoken to on previous occasions. The trial judge found that his was not a fleeting glance case, but one of recognition. She noted that Gardner’s identification is not precisely the same as that of a witness who has known a suspect over a lengthy period of time, but found that Gardner’s eyewitness account nevertheless did not suffer from all the frailties of the identification of a fleeting glance of a stranger. It was the task of the trial judge to take the circumstances of the identification into account and assess the weight to be accorded to it. We see no reviewable error in the trial judge’s assessment of Gardner’s identification evidence and in her reliance on that evidence to convict the appellant.

Similarly, in R. v. Delorme, 2017 SKCA 3, the Saskatchewan Court of Appeal upheld a conviction where the key identification witness was shown a photograph of the suspect on Facebook and told the suspect was “wanted,” while he waited at the police station to give his statement. The Court of Appeal found the trial judge had looked for markers of reliability in the witness’s testimony, and had compared it to the corroboratory evidence, and that his identification evidence was not tainted by his exposure to Facebook.^{Footnote 64}

When post-offence identification evidence is offered by a witness, it is incumbent upon police and prosecutors to establish the circumstances of the identification, so that the trier of fact has an accurate understanding of the reliability of the identification evidence. The Ontario Court of Appeal’s comments in R. v. Smerciak, (1946) 87 CCC 175 remain relevant:

The conditions under which an observation is made, the care with which it is made, and the ability of the observer, affect the weight of the evidence. In addition to such matters, and of the utmost importance, is the method used to recall or refresh the recollections of a witness who is to be relied upon to identify a person suspected of wrongdoing or who is under arrest. If a witness has no previous knowledge of the accused person so as to make him familiar with that person’s appearance, the greatest care ought to be used to ensure the absolute independence and freedom of judgment of the witness. His recognition ought to proceed without suggestion, assistance or bias created directly or indirectly. Conversely, if the means employed to obtain evidence of identification involve any acts which might reasonably prejudice the accused, the value of the evidence may be partially or wholly destroyed. Anything which tends to convey to a witness that a person is suspected by the authorities, or is charged with an offence, is obviously prejudicial and wrongful. Submitting a prisoner alone for scrutiny after arrest is unfair and unjust. Likewise, permitting a witness to see a single photograph of a suspected person or of a prisoner, after arrest and before scrutiny, can have no other effect, in my opinion, than one of prejudice to such a person.

In 2017, over 70 percent of regular internet users in Canada were reported to be social media users,^{Footnote 65} which suggests that social media platforms will continue to be used by eyewitnesses and victims as a means of identifying accused persons. Police and prosecutors must be aware of the inherent issues of post-offence identification, how it affects witnesses’ reliability, and the inherent dangers to identification and recognition evidence generated in an unrecorded and uncontrolled setting. As a result, it is more important than ever for police to meticulously record all comments or statements made by a witness during a lineup or photo-pack viewing via video, audio, or in writing. Although it is recognized that not every police agency has the necessary technology to video record eyewitness identification procedures, the American National Academy of Sciences has recommended that this become a standard practice for law enforcement to follow.^{Footnote 66}

V. Social Media Use and Memory Research

As discussed above, post-event information supplied by police, prosecutors, other eyewitnesses and social media sources has been clearly shown to alter people’s memories of a person or an event. Researchers have found that altered photos, subtle suggestions and leading questions can result in witnesses experiencing vivid and yet completely false memories of an event.^{Footnote 67} For instance, arbitrarily being labeled a “good” or “bad” witness can significantly affect eyewitness observation scores.^{Footnote 68} Even the hand gestures of interviewing police officers have been shown to alter witness accounts of a crime.^{Footnote 69} The potential for post-event information to contaminate, “taint” or otherwise alter eyewitness recall, and the importance of preventing this from occurring, is what informed many of the reasonable standards and practices for police and prosecutors in the recommendations in this chapter. However, when eyewitnesses post their firsthand accounts of a crime on social media, is it possible that the act of simply “posting” this information could contaminate their memory of the event in question?

New research on memory and social media has found that people remember experiences better when they post them on sites like Facebook.^{Footnote 70} However, the nature of social media sites encourages, as one 2010 court decision put it, “a certain amount of puffery” in one’s postings.^{Footnote 71} A 2012 study on social media psychology found that users selectively screen photos and information they post on social media platforms to develop online personae that may not necessarily reflect their daily life or true emotional state.^{Footnote 72} A paper published in the Connecticut Law Review argues that the informal nature of social media encourages exaggerations and falsifications among its users.^{Footnote 73} Additionally, the memory and social media study cited above acknowledges that when we post online, we are “creating a sense of self”, and that by shaping the way we remember our experiences, it is also shaping who we are.^{Footnote 74} This information suggests that posting firsthand eyewitness accounts of a crime on social media may influence the person’s memory and recall of that event. More research is required in this area.

VI. Mitigating Strategies

To mitigate potential memory contamination from social media use, as well as from other sources, research suggests that when police conduct a non-leading interview soon after an event (and preferably before a Facebook posting), they can help consolidate witness memories, leading to stronger memory performance over time.^{Footnote 75} Investigative interviews that maximize narrative detail and minimize specific prompting have been shown to elicit more accurate and detailed responses from witnesses.^{Footnote 76} Furthermore, allowing witnesses to provide details at their own pace promotes more elaborate memory retrieval and results in a more coherent account of what occurred.^{Footnote 77} Child witnesses, for example, who were allowed to talk about something innocuous (e.g., what they did for summer vacation) at their own direction and pace, prior to remembering a target event, remembered two and a half times as many details about the target event as children who answered questions from an interviewer-directed script.^{Footnote 78} Powell and Snow propose that the effectiveness of interview questions should be judged by their adherence to the acronym SAFE for:

• Simple language;
• Absence of specific details (not previously raised) or coercive techniques;
• Flexibility in allowing the interviewee to choose what information will be reported; and
• Encourages an elaborate response (such as tell me everything that happened at Joe’s house).^{Footnote 79}

Promoting deep memory processing among eyewitnesses through non-leading interview techniques reduces the chance of eyewitness identification error and promotes accurate eyewitness testimony in court. To this end, organizations in Australia are pooling their resources to develop a national interview framework, which encompasses best practices and guidelines for training.^{Footnote 80}

In Canada, many police agencies, including the RCMP, use Cognitive Interviewing, which has been shown to successfully enhance the memory retrieval process.^{Footnote 81} In this phased approach, the officer begins the interview by establishing trust and strong communication with the witness by making them feel valued and an equal partner in the process.^{Footnote 82} Developing a strong rapport with the witness has been shown to improve both the quality and quantity of the information that is recalled in the interview and is intended to reduce witness anxiety, which can affect memory retrieval. It is also intended to make the witness more resilient and less likely to comply with leading questions, which are often precursors to eyewitness error.

Cognitive Interview practices include:^{Footnote 83}

• Beginning an interview with open-ended questions. This primes the witness to give elaborate and detailed responses and helps to consolidate memories. Even important demographic information (such as a contact phone number) should be collected from the witness at the end of the interview, as providing short, closed answers has been shown to negatively affect the ensuing recall process.
• Memory retrieval at the most detailed level requires intense focus and concentration. It is important for the witness to feel there is adequate time to search through their memory and to report everything that is recalled, even details that might not seem important to the witness. The witness should be encouraged to conduct as many retrieval attempts as possible during this time.
• Approach the interviewer role with a facilitator mindset. To ensure the witness remembers and consolidates as much information about the event as possible, they should do most of the mental work and talking throughout the interview. The officer should use active listening techniques and keep interruptions to a minimum.
• Context has a powerful effect on memory. Mentally reconstructing an event in a witness’s mind can enhance overall recall.
• At the end of the interview, the officer should summarize the account, using the witness’s own words. This process helps to ensure the officer understands all the information that has been shared, and also functions as a final memory retrieval process for the witness, who should be encouraged to add any new information that may be recalled during this time. If closed-ended questions are required for further clarification, they should be asked during the final phase of the interview.

The recommendations in the 2005 and 2011 Reports related to police photo-pack presentations remain valid and sound. Indeed, the state of Louisiana, for example, recently introduced a bill (which is before the legislature at the time of this writing) to adopt statewide procedures for how eyewitness identifications are handled by law enforcement agencies which closely mirror these recommendations.^{Footnote 84}

Incorporating Cognitive Interviewing into standard practice enhances these practices and will not only result in the witness’s retrieval of more information for police to use in their investigation, but will also act as an additional safeguard against memory “tainting,” eyewitness error, and ultimately wrongful convictions. Cognitive Interviewing techniques can also be used by prosecutors when they meet with a witness before a case.

VII. Identification and Eyewitness Confidence

Since the release of the 2011 Report, further research has been conducted on the relationship between eyewitness confidence and accuracy. Much of the research has described the critical role of law enforcement in ensuring proper protocols are followed to avoid or mitigate bias or witness contamination in light of the frailties of human memory.

In October 2014, the National Research Council of the National Academy of Sciences (USA) released its report Identifying the Culprit; Assessing Eyewitness Identification.^{Footnote 85} This report outlines research on vision, memory and the variables that may impact a witness’s ability to accurately identify an alleged offender. Amongst its many recommendations, the report recommends investigators document a witness’s self-assessment of confidence in the identification at the time the witness first identifies a suspect. Evidence indicates that while self-reported confidence at the time of trial is not a reliable predictor of eyewitness accuracy (the witness having been influenced by external factors such as recall bias and opinions voiced by others), greater self-confidence at the moment of initial identification is generally reflective of greater accuracy of identification. That said, the strength of the confidence-accuracy relationship is dependent upon factors such as environmental conditions, persons involved and individual emotional states.^{Footnote 86}

The research conducted indicates that low-confidence initial identifications will almost always signal low accuracy, regardless of whether the identification procedures were pristine. However, high-confidence identification on an initial test generally signals high accuracy when pristine testing conditions were used by law enforcement. Ultimately, the report recommends collecting a confidence statement at the time of the initial identification.

In 2017, John Wixted and Gary Wells authored The Relationship Between Eyewitness Confidence and Identification Accuracy: A New Synthesis. Wixted and Wells’s research confirms that the greatest threat regarding the ability to rely on confidence in eyewitness identification occurs when witnesses receive post-identification feedback that suggest they made an accurate identification. It appears in the U.S. there is a growing trend within the legal system to disregard eyewitness confidence, with no distinction drawn as to whether the eyewitness-identification procedures were appropriate or not and with no distinction drawn between witness confidence at the time of the initial identification versus witness confidence at a later time. It is suggested that the legal system should draw a distinction between initial confidence that was obtained using pristine testing procedures and confidence obtained later or under conditions known to compromise the confidence-accuracy relationship.^{Footnote 87}

Wixted and Wells also describe various factors known to affect eyewitness memory and the importance of these factors being taken into consideration. Many of these factors are what they call estimator variables - variables that affect memory but are outside of the control of the legal system. Some common estimator variables they mention include:

1. Race (cross-race identifications are less accurate than same race identifications)
2. Exposure duration (brief exposure to the subject results in worse memory by the witness than longer exposure)
3. Lighting (poor lighting during the crime results in worse memory for the witness than good lighting)
4. Retention interval (a longer duration between the witnessed crime and the first lineup test may result in worse memory for the witness than a shorter duration).
5. Stress (high stress can lead to worse memory by the witness than low stress)
6. Weapon focus (witness memory is worse when a weapon is present than when no weapon is present).

The research conducted indicates that low-confidence initial identifications will almost always signal low accuracy, regardless of whether the identification procedures were pristine or not. High-confidence identification on an initial test generally signals high accuracy when pristine testing conditions were used by law enforcement. They recommend collecting a confidence statement at the time of the initial identification; when an eyewitness makes an identification, a statement should be obtained from that eyewitness indicating how confident they are that the person identified is the offender.

In summary, the biggest obstacle in relying heavily on confidence in eyewitness identification occurs when witnesses receive post-identification feedback that suggests they made an accurate identification. As per previous recommendations, this type of threat is mitigated by having an officer who is independent of the investigation conduct the photo-pack presentation without knowing who the suspect is, thus avoiding the possibility of inadvertent hints or reactions that could lead the witness before the identification takes place, or increase the witness’s degree of confidence afterward.

It is also important to always be mindful of cases such as R. v. Hanemaayer^{Footnote 88}, where a witness can be confident but also inaccurate, even at the early stages of the process. Evidence suggests that confidence statements can be a helpful tool to law enforcement, but they must be used carefully.

VIII. Updated Recommendations

1. The following are reasonable standards and practices that should be implemented and integrated by all police agencies:
   1. If possible, an officer who is independent of the investigation should be in charge of photo-pack presentation. This officer should not know who the suspect is, avoiding the possibility of inadvertent hints or reactions that could lead the witness before the identification takes place, or increase the witness’s degree of confidence afterward.
   2. The witness should be advised that the actual perpetrator may not be in the photo-pack, and therefore the witness should not feel that they must make an identification.
   3. The suspect should not stand out in the photo-pack as being different from the others, based on the eyewitness’s previous description of the perpetrator, or based on other factors that would draw extra attention to the suspect.
   4. All of the witness’s comments and statements made during photo-pack viewing should be recorded verbatim, by video and audio recording, or if that is not feasible, in writing. When an eyewitness makes an identification, a statement should be obtained from that eyewitness indicating how confident they are that the person identified is the perpetrator.
   5. If the identification process occurs on police premises, reasonable steps should be taken to remove the witness as soon as possible upon completion of the photo-pack presentation to prevent any potential feedback by other officers involved in the investigation and cross contamination by contact with other witnesses.
   6. Show-ups should be used only in rare circumstances, such as when the suspect is apprehended near the crime scene shortly after the event.
   7. A photo-pack should be provided sequentially, and not as a package, thus preventing ‘relative judgments’.
2. For prosecutors, the following practical suggestions should be considered:
   1. Assume the identity of the accused is always at issue unless the defence specifically admits it on the record. Timely preparation and a critical review of all of the available identification evidence, including the manner in which it was obtained, is required as it will affect the conduct and quality of the trial.
   2. Allow the witness a reasonable opportunity to review all previously provided statements and confirm that the statements were accurate and a true reflection of their observations at the time. Carefully canvass the full range of the indicia of the identification, including any distinguishing features that augment this evidence. Remember that it is the collective impact of all of the evidence that will be considered in support of a conviction. Defects in one witness’s identification can be overcome by the consideration of other evidence.
   3. Never iliterview witnesses collectively. Never prompt or coach a witness by offering clues or hints about the identity of the accused in court. Do not participate in a “show-up” lineup. Never show a witness an isolated photograph or image of an accused during the interview.
   4. When meeting with witnesses in serious cases, it is wise, if it is feasible and practical, to have a third party present to ensure there is no later disagreement about what took place at the meeting.
   5. Never tell a witness that they are right or wrong in their identification.
   6. Remember that disclosure is a continuing obligation. All inculpatory and exculpatory evidence must be disclosed to the defence in a timely fashion. In the event that a witness materially changes their original statement, by offering more or recanting previously given information during an interview, the defence must be told. In these circumstances, it would be prudent to enlist the services of a police officer to record a further statement in writing setting out these material changes.
   7. Always lead evidence of the history of the identification. It is vitally important that the trier of fact not only be told of the identification but also all the circumstances involved in obtaining it, e.g., the composition of the photo-pack.
   8. Be wary of prosecutions based on weak single-witness identifications. While not required by law to secure a conviction, ascertain whether there is any corroboration of an eyewitness’s identification in order to overcome any deficiencies in the quality of that evidence.
3. The use of expert evidence on the frailties of eyewitness identification is redundant and unnecessary in the fact-finding process. A proper charge and caution by the trial judge can best deal with the inherent dangers of identification evidence.
4. Workshops on Cognitive Interviewing should be incorporated in regular and ongoing training sessions for police and prosecutors.
5. Presentations on the perils of eyewitness misidentifications, including the academic research and results of experts in the field of memory and eyewitness identification, should be incorporated in regular and ongoing training sessions for police and prosecutors.