EYEWITNESS MEMORY AN EARLY MODEL OF MEMORY
The capacity and fallibility of human memory was one of the first areas of investigation in psychological research. Through careful experimental work, several distinguished scholars, including Hermann Ebbinghaus (1850–1909), began to unravel some of the fundamental properties of memory functioning (Ebbinghaus, 1885/1994). One model that emerged from this early work described the three memory stages of (i) acquisition (when memories are formed), (ii) retention (holding them in storage) and (iii) retrieval (fetching them from storage).
While memory theory has moved on from this basic model, it is still useful in a discussion of eyewitness memory. [eyewitness testimony the evidence given by witnesses to a crime, typically in the form of a verbal account or person identification] Research into the accuracy of eyewitness testimony has focused on initial observation of the incident (acquisition), the period between seeing and recalling (retention) and, finally, giving testimony (retrieval). Researchers have engaged with a wide range of relevant variables over a long period (Goodman et al., 1999; Ross, Read & Toglia, 1994; Sporer, Malpass & Koehnken, 1996), including:
1 social variables, such as the status of the interrogator;
2 situational variables, such as the type of crime;
3 individual variables, such as witness age; and
4 interrogational variables, such as the type of questioning.
The role of the expert witness
As professional chartered or registered psychologists, forensic psychologists are often called upon to provide reports on particular individuals for court hearings. For example, in a legal context, a forensic evaluation may subsequently be used to assist the court in making an appropriate decision regarding family, civil or criminal matters. A forensic psychologist might be called to give expert evidence in the accuracy of eyewitness memory, or the likelihood of a false confession, or the reliability of children as witnesses when subjected to certain questioning procedures.
The evaluation of the client provided by the forensic psychologist will often involve characterizing the relationship between psychological factors and relevant legal issues. For example, what is the forensic psychologist’s best opinion regarding the possible precipitating factors preceding the crime or civil offence? Findings should be clearly communicated and reflect standard psychological practice, including nationally and internationally accepted psychological instruments and norms. Relevant empirical research that is consistent with the psychologist’s conclusions should be noted. Any recommendations that are made (for example, with respect to rehabilitation) must be legally sound, practical and involve services that are widely available in the individual’s local community. Forensic psychologists should be able to defend their conclusions logically. It is especially important that the psychologist uses explanations that can be understood by non-psychologists, such as the judge, barristers and, of course, members of the jury. The relevant issues should therefore be presented clearly and simply, but without ‘dumbing down’. This takes great skill on the part of the forensic psychologist. The conclusions and recommendations of the forensic psychologist should assist the relevant person or agency in reaching a decision, and should not add unnecessary confusion to that process. In addition to having a relevant training and education background, it is therefore critical for psychologists who undertake forensic evaluations to possess excellent assessment and communication skills. They must also have experience and/or a thorough training in completing psychological evaluations in a legal setting so that they will not be ‘fazed’ by the process. Lawyers engaged in cross-examination can be hostile and seek to undermine the credibility of the psychologists’ professional opinions. ‘Wherever possible, stick to the facts’ is a piece of advice frequently offered to individuals who are presenting in court. Psychologists offeri ng a professional opinion in court are protected by the court and therefore cannot be sued for defamation.
Nevertheless, they should evaluate the core facts of the case in order to reach a professionally informed opinion regarding the psychological issues only. As with any professional, psychologists should not offer opinions outside their area of expertise. For example, they should not speculate on whether a defective mechanism in the workplace may have contributed to the event they have identified; this would be the province of another forensic professional. Egeth, H.E., & McCloskey, M., 1984, ‘Expert testimony about eyewitness behaviour: Is it safe and effective?’ in G.L. Wells & E.F. Loftus (eds), Eyewitness Testimony: Psychological Perspectives, Cambridge: Cambridge University Press.
Acquisition
Research has also considered the effect of particular types of crime. For example, can witnesses to a violent crime be as accurate as witnesses to a non-violent crime? Controlled experimental studies, typically during which witnesses see videotaped crimes of varying degrees of violence, suggest that violence results in poorer witness accuracy (Clifford & Hollin, 1981). But strangely, field studies of real-life witnesses suggest that those who are exposed to highly violent events can give very accurate testimony (Yuille & Cutshall, 1986). Indeed, adult victims of rape usually give a reasonably accurate account of this extreme personal experience of violence (Koss, Tromp & Tharan, 1995). One possible explanation for this apparent contradiction is that, in a stressful situation such as a violent crime, a witness’s attention may narrow to the central (rather than the peripheral) details of the incident. The theory is that the deployment of attention narrows to central details of the event, such as the criminal’s actions, thereby producing less reliable memory for peripheral detail, such as what colour shirt the criminal was wearing (Clifford & Scott, 1978). When the central detail is a life-threatening weapon, witnesses may pay much more attention to the weapon, to the exclusion of other details. This phenomenon is known as ‘weapon focus’ (Loftus, Loftus & Messo, 1987). It is vital to understand the impact on witness memory of factors such as the type of crime. What is encoded during acquisition is critical because it forms the basis for what is stored in memory and eventually retrieved when giving testimony.
Retention
During the retention stage, witness memory may be subject to various influences, such as discussion with other witnesses and exposure to media accounts of the crime, not to mention the fact that memory becomes less accurate over time. So the time interval between acquisition and retrieval is an obvious consideration. Several studies have compared the accuracy of eyewitness face identification over short and long time intervals. Malpass and Devine (1981), for example, chose short (three-day) and long (five-month) intervals. They found, not surprisingly, that after three days there were no false identifications, but after five months the rate of false identifications had risen (table 21.1). Conversely, the rate of correct identifications was initially high but fell significantly at five months. Krafka and Penrod (1985) reported a similar finding with the much shorter time intervals of two hours and 24 hours. The force of the evidence suggests that identification accuracy does decrease with time, although the ra tes for false and correct identifications may be different.
Retrieval
Finally, during the retrieval stage, factors that potentially influence the accuracy of eyewitness testimony include interview style and the use of aids to recall, such as the photofit and identity parades.
Studies of the impact of leading questions show that even subtle changes in question wording can influence testimony. For example, Loftus and Palmer (1974) asked witnesses to a filmed traffic accident to estimate the speed of the cars when ‘they – into eachother’: for different groups of witnesses the blank read ‘contacted’, ‘hit’, ‘bumped’, ‘collided’ or ‘smashed’. The witnesses’ estimates of the speed increased according to the level of force implied by the verb contained in the question. In later questioning, those witnesses who had been asked about the car ‘smash’ were more likely to say – mistakenly – that they had seen broken glass. Additional studies have established that misleading information presented to witnesses is more likely to have an influence on peripheral details than central events (Read & Bruce, 1984). Furthermore, it seems that the effects of leading questions such as those used by Loftus and Palmer (1974) are a direct product of the demands of the questioning procedures, rather than the questions leading to permanent changes in memory (Zar agoza, McCloskey & Jarvis, 1987). This last point emphasizes that witnesses can give incorrect replies to questions even though the memory trace (‘retention’) itself has apparently not been distorted.
THE STRENGTH AND VALIDITY OF THE EVIDENCE
Narby, Cutler and Penrod (1996) have created three categories of witness-related evidence based on reliability and magnitude of effect:
1. reliable and strong factors that show consistent effects on eyewitness memory (e.g. there are differences in memory performance between adult and child witnesses; if a person is wearing a disguise, such as a hat, this influences accuracy of memory; and the length of time, termed ‘exposure duration’, that the witness has to observe an incident);
2. reliable and moderate factors that show effects in some studies but not in others (e.g. the match between the level of confidence a witness has in their memory and how accurate it really is; weapon focus; and crime seriousness); and
3. weak or non-influential factors that have little or no effect on witness accuracy (e.g. witness gender; the personality of the witness; and (within limits) the witness’s level of intelligence).
An issue that is broader than the strength of the evidence concerns its validity when applied to the real world. Do the findings from psychological studies parallel what happens to real crime witnesses? Should research findings be made available to the court to influence real trials? In other words, can psychological studies of eyewitness memory be generalized to real life?
Critics such as Konecni and Ebbesen (1986) and Yuille and Cutshall (1986) note the lack of realism in many experimental studies, such as the use of filmed crimes, and the participants’ awareness of the research aims. The matter boils down to one of control – laboratory studies allow a high degree of control at the expense of realism, while field research is more realistic and ‘ecologically valid’ but prey to a host of influences that reduce control over the variables being measured. This is a problem which has bedevilled many areas of psychological research but its influence is arguably no more profound than in the field of forensic psychology.
The strongest conclusions that can be validly drawn will most likely be derived from a variety of studies (including laboratory studies, case studies, field studies and archival studies) which employ a broad range of different experimental designs and methodologies (see chapter 2 on ‘triangulation’).
Measuring crime - A research issue
A great deal of research in forensic psychology relies on a measure of crime as key outcome measure. For example, the evaluation of crime prevention initiatives, innovative police procedures or offender treatment strategies all rely on measuring their impact on crime in order to estimate their effectiveness and make decisions regarding their continued funding.
There are several ways of measuring crime (for example, conducting victim surveys to gain knowledge of local or national estimates of levels of crime, asking known criminals to give self-reports about their offending, or looking to official reconviction records). We know that there will be differences across measurement of crime according to the use of victim reports, self-reports or official figures. But what can we say about variation within a type of measurement? For example, can it be taken for granted that there will be consistency in official records of crime?
A study by Friendship, Thornton, Erikson and Beech (2001) looked at the two main sources of criminal history information held in England and Wales. It is from these sources that researchers take official reconviction figures.
Design and procedure:- The research was concerned with two sources of criminal history data:
1. The Offenders Index (OI) is a computerized database containing criminal histories, based on court appearances, of all those convicted of standard offences in England and Wales since 1963.
2. The National Identification System (NIS) is based on police records held both on microfiche and a computer database.
In order to compare these data sources, the researchers took a sample of 134 sexual offenders and compared the data for offence history and reconvictions for this group as recorded on the OI and the NIS.
Results and implications:- There were variations between the two data sources in their recording of criminal history variables of the sexual offenders.
This variation, in turn, indicated that the reconviction rates derived from the two data sources differed. Based on the OI, the reconviction rate for the sample was 22 per cent reconviction for general offences and 10 per cent for further sexual offences, but for the NIS, the comparable rates were 25 per cent and 12 per cent, respectively. Also, when a composite measure of reconviction was derived by combining OI and NIS data to give the ‘best’ estimate, the reconviction rates, again for general and sexual offences, were 32 and 13 per cent respectively. Friendship et al. are correct in stating that their findings will be of great benefit to researchers. The use of composite measures of reconviction, based on both the OI and NIS data sources, gives a more complete indication of reconviction than either source used alone. This is essential information, based on empirical study, for researchers whose work may inform both policy-makers and practitioners.
Friendship, C., Thornton, D., Erikson, M., & Beech, A., 200 1, ‘Reconviction: A critique and comparison of two main data sources in England and Wales’, Legal and Criminological Psychology, 6, 121–9.
The capacity and fallibility of human memory was one of the first areas of investigation in psychological research. Through careful experimental work, several distinguished scholars, including Hermann Ebbinghaus (1850–1909), began to unravel some of the fundamental properties of memory functioning (Ebbinghaus, 1885/1994). One model that emerged from this early work described the three memory stages of (i) acquisition (when memories are formed), (ii) retention (holding them in storage) and (iii) retrieval (fetching them from storage).
While memory theory has moved on from this basic model, it is still useful in a discussion of eyewitness memory. [eyewitness testimony the evidence given by witnesses to a crime, typically in the form of a verbal account or person identification] Research into the accuracy of eyewitness testimony has focused on initial observation of the incident (acquisition), the period between seeing and recalling (retention) and, finally, giving testimony (retrieval). Researchers have engaged with a wide range of relevant variables over a long period (Goodman et al., 1999; Ross, Read & Toglia, 1994; Sporer, Malpass & Koehnken, 1996), including:
1 social variables, such as the status of the interrogator;
2 situational variables, such as the type of crime;
3 individual variables, such as witness age; and
4 interrogational variables, such as the type of questioning.
The role of the expert witness
As professional chartered or registered psychologists, forensic psychologists are often called upon to provide reports on particular individuals for court hearings. For example, in a legal context, a forensic evaluation may subsequently be used to assist the court in making an appropriate decision regarding family, civil or criminal matters. A forensic psychologist might be called to give expert evidence in the accuracy of eyewitness memory, or the likelihood of a false confession, or the reliability of children as witnesses when subjected to certain questioning procedures.
The evaluation of the client provided by the forensic psychologist will often involve characterizing the relationship between psychological factors and relevant legal issues. For example, what is the forensic psychologist’s best opinion regarding the possible precipitating factors preceding the crime or civil offence? Findings should be clearly communicated and reflect standard psychological practice, including nationally and internationally accepted psychological instruments and norms. Relevant empirical research that is consistent with the psychologist’s conclusions should be noted. Any recommendations that are made (for example, with respect to rehabilitation) must be legally sound, practical and involve services that are widely available in the individual’s local community. Forensic psychologists should be able to defend their conclusions logically. It is especially important that the psychologist uses explanations that can be understood by non-psychologists, such as the judge, barristers and, of course, members of the jury. The relevant issues should therefore be presented clearly and simply, but without ‘dumbing down’. This takes great skill on the part of the forensic psychologist. The conclusions and recommendations of the forensic psychologist should assist the relevant person or agency in reaching a decision, and should not add unnecessary confusion to that process. In addition to having a relevant training and education background, it is therefore critical for psychologists who undertake forensic evaluations to possess excellent assessment and communication skills. They must also have experience and/or a thorough training in completing psychological evaluations in a legal setting so that they will not be ‘fazed’ by the process. Lawyers engaged in cross-examination can be hostile and seek to undermine the credibility of the psychologists’ professional opinions. ‘Wherever possible, stick to the facts’ is a piece of advice frequently offered to individuals who are presenting in court. Psychologists offeri ng a professional opinion in court are protected by the court and therefore cannot be sued for defamation.
Nevertheless, they should evaluate the core facts of the case in order to reach a professionally informed opinion regarding the psychological issues only. As with any professional, psychologists should not offer opinions outside their area of expertise. For example, they should not speculate on whether a defective mechanism in the workplace may have contributed to the event they have identified; this would be the province of another forensic professional. Egeth, H.E., & McCloskey, M., 1984, ‘Expert testimony about eyewitness behaviour: Is it safe and effective?’ in G.L. Wells & E.F. Loftus (eds), Eyewitness Testimony: Psychological Perspectives, Cambridge: Cambridge University Press.
Acquisition
Research has also considered the effect of particular types of crime. For example, can witnesses to a violent crime be as accurate as witnesses to a non-violent crime? Controlled experimental studies, typically during which witnesses see videotaped crimes of varying degrees of violence, suggest that violence results in poorer witness accuracy (Clifford & Hollin, 1981). But strangely, field studies of real-life witnesses suggest that those who are exposed to highly violent events can give very accurate testimony (Yuille & Cutshall, 1986). Indeed, adult victims of rape usually give a reasonably accurate account of this extreme personal experience of violence (Koss, Tromp & Tharan, 1995). One possible explanation for this apparent contradiction is that, in a stressful situation such as a violent crime, a witness’s attention may narrow to the central (rather than the peripheral) details of the incident. The theory is that the deployment of attention narrows to central details of the event, such as the criminal’s actions, thereby producing less reliable memory for peripheral detail, such as what colour shirt the criminal was wearing (Clifford & Scott, 1978). When the central detail is a life-threatening weapon, witnesses may pay much more attention to the weapon, to the exclusion of other details. This phenomenon is known as ‘weapon focus’ (Loftus, Loftus & Messo, 1987). It is vital to understand the impact on witness memory of factors such as the type of crime. What is encoded during acquisition is critical because it forms the basis for what is stored in memory and eventually retrieved when giving testimony.
Retention
During the retention stage, witness memory may be subject to various influences, such as discussion with other witnesses and exposure to media accounts of the crime, not to mention the fact that memory becomes less accurate over time. So the time interval between acquisition and retrieval is an obvious consideration. Several studies have compared the accuracy of eyewitness face identification over short and long time intervals. Malpass and Devine (1981), for example, chose short (three-day) and long (five-month) intervals. They found, not surprisingly, that after three days there were no false identifications, but after five months the rate of false identifications had risen (table 21.1). Conversely, the rate of correct identifications was initially high but fell significantly at five months. Krafka and Penrod (1985) reported a similar finding with the much shorter time intervals of two hours and 24 hours. The force of the evidence suggests that identification accuracy does decrease with time, although the ra tes for false and correct identifications may be different.
Retrieval
Finally, during the retrieval stage, factors that potentially influence the accuracy of eyewitness testimony include interview style and the use of aids to recall, such as the photofit and identity parades.
Studies of the impact of leading questions show that even subtle changes in question wording can influence testimony. For example, Loftus and Palmer (1974) asked witnesses to a filmed traffic accident to estimate the speed of the cars when ‘they – into eachother’: for different groups of witnesses the blank read ‘contacted’, ‘hit’, ‘bumped’, ‘collided’ or ‘smashed’. The witnesses’ estimates of the speed increased according to the level of force implied by the verb contained in the question. In later questioning, those witnesses who had been asked about the car ‘smash’ were more likely to say – mistakenly – that they had seen broken glass. Additional studies have established that misleading information presented to witnesses is more likely to have an influence on peripheral details than central events (Read & Bruce, 1984). Furthermore, it seems that the effects of leading questions such as those used by Loftus and Palmer (1974) are a direct product of the demands of the questioning procedures, rather than the questions leading to permanent changes in memory (Zar agoza, McCloskey & Jarvis, 1987). This last point emphasizes that witnesses can give incorrect replies to questions even though the memory trace (‘retention’) itself has apparently not been distorted.
THE STRENGTH AND VALIDITY OF THE EVIDENCE
Narby, Cutler and Penrod (1996) have created three categories of witness-related evidence based on reliability and magnitude of effect:
1. reliable and strong factors that show consistent effects on eyewitness memory (e.g. there are differences in memory performance between adult and child witnesses; if a person is wearing a disguise, such as a hat, this influences accuracy of memory; and the length of time, termed ‘exposure duration’, that the witness has to observe an incident);
2. reliable and moderate factors that show effects in some studies but not in others (e.g. the match between the level of confidence a witness has in their memory and how accurate it really is; weapon focus; and crime seriousness); and
3. weak or non-influential factors that have little or no effect on witness accuracy (e.g. witness gender; the personality of the witness; and (within limits) the witness’s level of intelligence).
An issue that is broader than the strength of the evidence concerns its validity when applied to the real world. Do the findings from psychological studies parallel what happens to real crime witnesses? Should research findings be made available to the court to influence real trials? In other words, can psychological studies of eyewitness memory be generalized to real life?
Critics such as Konecni and Ebbesen (1986) and Yuille and Cutshall (1986) note the lack of realism in many experimental studies, such as the use of filmed crimes, and the participants’ awareness of the research aims. The matter boils down to one of control – laboratory studies allow a high degree of control at the expense of realism, while field research is more realistic and ‘ecologically valid’ but prey to a host of influences that reduce control over the variables being measured. This is a problem which has bedevilled many areas of psychological research but its influence is arguably no more profound than in the field of forensic psychology.
The strongest conclusions that can be validly drawn will most likely be derived from a variety of studies (including laboratory studies, case studies, field studies and archival studies) which employ a broad range of different experimental designs and methodologies (see chapter 2 on ‘triangulation’).
Measuring crime - A research issue
A great deal of research in forensic psychology relies on a measure of crime as key outcome measure. For example, the evaluation of crime prevention initiatives, innovative police procedures or offender treatment strategies all rely on measuring their impact on crime in order to estimate their effectiveness and make decisions regarding their continued funding.
There are several ways of measuring crime (for example, conducting victim surveys to gain knowledge of local or national estimates of levels of crime, asking known criminals to give self-reports about their offending, or looking to official reconviction records). We know that there will be differences across measurement of crime according to the use of victim reports, self-reports or official figures. But what can we say about variation within a type of measurement? For example, can it be taken for granted that there will be consistency in official records of crime?
A study by Friendship, Thornton, Erikson and Beech (2001) looked at the two main sources of criminal history information held in England and Wales. It is from these sources that researchers take official reconviction figures.
Design and procedure:- The research was concerned with two sources of criminal history data:
1. The Offenders Index (OI) is a computerized database containing criminal histories, based on court appearances, of all those convicted of standard offences in England and Wales since 1963.
2. The National Identification System (NIS) is based on police records held both on microfiche and a computer database.
In order to compare these data sources, the researchers took a sample of 134 sexual offenders and compared the data for offence history and reconvictions for this group as recorded on the OI and the NIS.
Results and implications:- There were variations between the two data sources in their recording of criminal history variables of the sexual offenders.
This variation, in turn, indicated that the reconviction rates derived from the two data sources differed. Based on the OI, the reconviction rate for the sample was 22 per cent reconviction for general offences and 10 per cent for further sexual offences, but for the NIS, the comparable rates were 25 per cent and 12 per cent, respectively. Also, when a composite measure of reconviction was derived by combining OI and NIS data to give the ‘best’ estimate, the reconviction rates, again for general and sexual offences, were 32 and 13 per cent respectively. Friendship et al. are correct in stating that their findings will be of great benefit to researchers. The use of composite measures of reconviction, based on both the OI and NIS data sources, gives a more complete indication of reconviction than either source used alone. This is essential information, based on empirical study, for researchers whose work may inform both policy-makers and practitioners.
Friendship, C., Thornton, D., Erikson, M., & Beech, A., 200 1, ‘Reconviction: A critique and comparison of two main data sources in England and Wales’, Legal and Criminological Psychology, 6, 121–9.
No comments:
Post a Comment