eyewitness identification lab - university of texas at el paso

Colin G. TredouxUniversity of Cape Town, Rosebank,

South Africa

1 . Introduction2 . The Event3. The Witness4. The Perpetrator5. Obtaining Information from the Witnessb. The Witness Identification7. The Prosecution8. Conclusion

Further Reading

GLOSSARY

blank lineup (target absent lineup) A lineup in which all themembers are known to be innocent .

cognitive interview An interview technique designed toenhance memory and communication of events throughthe use of guided retrieval techniques .

effective size The number of plausible members in a lineup .encoding specificity principle If cues that were present at

encoding are also present at retrieval, recall of storedinformation is more likely .

estimator variables Factors in eyewitness situations whoseinfluence can only be estimated and are not under thecontrol of the justice system.

eyewitness A person who observes an event and can provideinformation about that event .

lineup bias A lineup constructed in a manner that leads thesuspect to stand out from the other lineup members .

Encyclopedia of Applied Psychology,VOLUME 1

Eyewitness Identification

Christian A. Meissner

Roy S. Malpass andLaura A. ZimmermanUniversity of Texas at El Paso,

El Paso, Texas, USA

Florida International University,Miami, Florida, USA

875

mock witness A person who is not a witness to a crime but isasked to identify a perpetrator from a lineup based onanother source of information, such as a verbaldescription.

postevent (mis)information Events that occur after a crimethat influence eyewitness memories of the crime .

sequential lineup A lineup presentation procedure in whichone lineup member is presented to the witness at a time .

showup The presentation of a single suspect to a witness inorder to determine if that suspect is the perpetrator .

simultaneous lineup A lineup presentation procedure inwhich all lineup members are presented to the witness atthe same time (i .e ., in a line) .

system variables Factors that may influence eyewitnessmemory and are under the control of the criminal justicesystem.

verbal overshadowing effect When the act of verballydescribing a perpetrator decreases the witness's ability toaccurately identify a perpetrator from a lineup .

An eyewitness's identification of a perpetrator is aprevalent form of incriminating evidence presented ina criminal case . However, eyewitnesses are frequentlymistaken, and their errors have led to the conviction ofinnocent individuals . For more than a century,psychologists have sought to understand the variety offactors that may influence the accuracy of eyewitnessidentification . This article reviews the current state of

X2004 Elsevier Inc .All rights reserved .

8 76

knowledge in the science of eyewitness identificationaccording to the sequence of events that is likely tounfold from the time that an eyewitness makes anobservation to the prosecution of the perpetrator ofthe crime .

1 . INTRODUCTION

The testimony of eyewitnesses is an important factor inmany criminal cases . Cases that hinge on eyewitnesstestimony can bring perpetrators of crimes to justice,set innocent people free, and enable the police andcourts to carry out their prime societal responsibilitiesof preserving law and order . However, eyewitnesses arefrequently mistaken, leading to the conviction andimprisonment of innocent people for crimes they didnot commit. The extent of this problem is becomingclearer with the exoneration of innocent peoplethrough DNA evidence . According to The InnocenceProject, a New York-based legal team that reviews caserecords of incarcerated individuals who claim theirinnocence, there are currently more than 140 docu-mented DNA-exoneration cases in the United States .Mistaken eyewitness identification is generally themost frequent source of evidence used to convict aninnocent defendant . Of the current database of DNAexonerations, approximately 85% are estimated to haveinvolved faulty eyewitness identification evidence .

Acknowledgment of the dangers of eyewitness iden-tification is not new. In England, for example, theCriminal Law Revision Committee of 1972 gave expli-cit recognition to this fact: "We regard mistakenidentification as by far the greatest cause of actual orpossible wrong convictions." Psychology and lawresearchers have long recognized the vagaries of eye-witness identification, conducting original empiricalresearch from as far back as the late 19th century .This research has made a significant contribution topolice and legal practice, partly because psycholegalresearchers bring an empirical and scientific perspec-tive to a legal problem, which is an approach that isquite unnatural for lawyers and jurists to take . In thecurrent discussion of eyewitness identification, westructure our review, as far as possible, according tothe sequence of events that is likely to unfold whenan eyewitness makes an observation . The processesthat eyewitnesses go through at each stage of theprocess can vary greatly, and the interests of research-ers are quite different at each stage, as are the chal-lenges they face .


1 .1 . Stages in the EyewitnessProcess

There are three main stages of the eyewitness process :(i) perception of an event and the persons central to it,(ii) storage and assimilation of the informationextracted from that experience, and (iii) rememberingaspects of the experience and acting on it. Time delaysbetween the first and third stages can vary greatly.Some information may be requested almost immedi-ately, as when a witness at the crime scene is asked todescribe an event or an offender . Other informationmay be requested much later, such as when a suspect ispresented for identification and the witness is asked tomake a complex set of judgments and decisions .

1 .2 . Categorization of Variables

Pertinent psychological variables differ at the variousstages of the eyewitness process, and different bodies oftheory and research may apply to each variable to under-stand them. Two important categories of variables aresystem variables and estimator variables, distinguishedby Gary Wells in 1978 . Estimator variables arise duringthe first stage of the eyewitness process and are thosefactors over which the criminal justice system has nocontrol, whose degree and magnitude can only be esti-mated after the fact. By definition, these include attri-butes of the witness, the event, and the perpetrator, andthey may involve such things as the opportunity thewitness had to view the event and perpetrator, the stateof mind of the witness at the time, the attentionalconstraints that have impinged upon the witness as aresult of the demands made on the witness's attention, orany person variables (such as race or gender) that mayhave influenced memory performance . System variables,on the other hand, are under the control of the criminaljustice system. These variables focus on the treatment ofthe witness, the manner in which law enforcementofficers attempt to obtain information from the witness,and the interaction of person attributes with these pro-cesses. System variables occur at later stages of theprocess (e .g ., the second and third stages), at the pointswhere information is requested of the witness .

The demands on research vary depending on thestage of the witnessing process and the type of variableunder study. For early stage estimator variables (suchas the quality of memory encoding), an effective under-standing requires one to simulate them in the labora-tory or to study them in situ, in collaboration with lawenforcement. The latter is rarely done . Law

enforcement generally has a very limited interest inand capability for studying these matters, althoughworldwide there are a few major law enforcementagencies that engage in research in this field. Manyother estimator variables cannot feasibly be studied inlaboratory simulations for ethical or logistical reasons,such as high levels of stress, fear, and violence . Wereview, in general terms, what is known about majorestimator variables in this article .

System variables are quite different from estimatorvariables in that they are implemented in similar waysby law enforcement and researchers, and they act uponpersons who have already been through the first stagesof a criminal event . Thus, witnesses will naturally varyin the amount of information they have encoded aboutthe event and the perpetrator. Witnesses in laboratorystudies of system variables will often participate incrimelike events, but these are not intended as precisesimulations of criminal events . Rather, they are merelyways of giving witnesses a temporal event, with somedegree of complexity, and that includes informationabout an event and a perpetrator about which theymight later be tested . Of most interest is the effect ofvarious system variables (such as the structure of alineup or the instructions provided to a witness) onthe accuracy of information obtained from the witness .

1 .3 . General Research Assumptions

Generally speaking, the following assumptions underlieresearch in this area . First, all individuals functionaccording to the same general set of psychologicalprocesses. For example, perception and memory donot work differently for different people in any funda-mental sense, and the psychological processes that wit-nesses engage in do not fundamentally change indifferent contexts or situations . Second, informationmust be encoded at the time of the witnessed event forit to be reported later. If the witness does not attend tothe event or to information within the event, theninformation about that event will not be available forsubsequent recall or recognition. Any distractions ofattention away from portions of an event will result indecreased encoding and therefore an absence of subse-quent memory for such information . Likewise, attrac-tion of attention to some elements will result in superiorencoding of such information. Finally, the willingness ofthe witnesses to identify someone from a lineup involvesa decision process that can be somewhat controlled bythose administering the identification procedure .Although the specific factors leading to identification

Eyewitness Identification,

may differ from person to person, the decision processesutilized are assumed to be highly consistent .

2 . THE EVENT

An eyewitness is defined by an event: This is somethinghe or she directly observes and will later be asked toprovide information about . In order to study eyewit-nesses, then, we need to study the kind of events towhich they bear witness . However, witnessed eventsrange from the mundane to the terrifying and arepotentially infinite in their diversity . Eyewitnessresearchers are thus faced with the daunting task ofunderstanding how eyewitness memory and relatedprocesses function across a wide range of possibilities .It is important to understand that eyewitness recogni-tion and identification of the criminal offender is based(for the most part) on comparing a face (person orphoto) shown to the witness with an image of a facestored in memory. In general, the lower the quality ofthe image stored in memory, the less likely an eye-witness is to make an accurate recognition decision .Any event or process that degrades perception (i .e .,acquisition of the original image) also reduces the like-lihood of a correct recognition or identification. Thefollowing factors contribute to a witness's opportunityto encode information about an event or an individual .

2 .1 . Time to View the Event/Perpetrator

The duration of the witness's exposure to the offenderis related to later recognition performance, such thatlimiting exposure time generally reduces eyewitnessaccuracy . This has been shown in both laboratory andarchival studies . Although a range of specific times hasbeen employed across studies, it is difficult to calibratespecific time durations to specific levels of identifica-tion accuracy, particularly given the many other factorsinvolved . Furthermore, it is difficult to accuratelyassess the time a witness had to view the perpetratorin situ, as studies have shown that witnesses' post hocrecollection of time estimates can be very inaccurate .

2 .2 . Distance from the Event/Perpetrator

The physical distance between the witness and theoffender is also related to later recognition performance,

8 77

8 78 Eyewitness

with longer distances leading to poorer rates of identifi-cation. Again, it is difficult to calibrate specific distanceswith specific levels of identification accuracy. In addi-tion, witnesses can be very inaccurate at estimating thedistance between themselves and the offender, so theinformation may not be helpful or reliable even if wewere able to calibrate accuracy in a laboratory .

2.3 . Visibility

Light levels (e.g ., time of day) and obscured illumina-tion (e .g., sun shining in the witness's face) have beenstudied by eyewitness researchers and shown to influ-ence both perceptual and identification processes.Specifically, poor lighting and obscured illuminationresult in lower rates of accurate identification .

2 .4. Stress, Fear, and the Presenceof a Weapon

A witness's attention may be impaired or distracted if heor she focuses on the psychological stress or fear accom-panying a criminal or otherwise traumatic event . Even ifa witness tries to be attentive, high fear or stress mayhinder the accuracy of subsequent identifications .Psychologists have often described the effects of stressor fear according to the Yerkesr-Dodson law, whichposits that very low or very high amounts of arousalare most likely to impede encoding and recognition .The presence of a weapon, which may be accompaniedby stress or fear, has also been shown to divert awitness's attention away from the face of the offender .In general, the presence of a weapon reduces both thesubsequent quality of the description provided by thewitness and his or her attempts at perceptual identifica-tion of the perpetrator .

3 . THE WITNESS

Individual eyewitnesses may differ along a number ofdimensions or attributes, and it is natural to askwhether any of these attributes are relevant to an eye-witness' performance. Research has identified severalimportant characteristics .

3 .1 . Witness Age

Although eyewitness memory remains relatively stableacross a wide age range, it is less reliable for individuals

identification

in certain age groups . For example, very early in life,children demonstrate a limited ability to recall informa-tion and show lower accuracy in the identification offaces. They also have a proclivity toward choosing (orguessing) during a lineup identification task, leading toan increased likelihood of mistaken identification of aninnocent person . By early teen years, their performanceis generally indistinguishable, on average, from that ofadults . Although children tend to recall less informationwhen compared with adults, the overall proportion ofcorrect information recalled does not typically differ .Older adults also become somewhat variable, but inthe absence of a disease process (e .g ., dementia), olderadults are indistinguishable, on average, from youngeradults . Sensory changes, such as cataracts or changes inthe contrast or sharpness of vision, may affect an elderlywitness's ability to gather and encode information abouta situation or offender. Elderly adults also appear to bemore susceptible to the effects of suggestive questioningor postevent misinformation than the average adult .Although children may also be susceptible to suggestion,this is presumed to be due to their unwillingness tochallenge an adult's authority . The elderly are morelikely to forget the sources of their information .

3 .2 . Alcohol and Other DrugsThere is very little research specifically on the effects ofalcohol and other drugs on eyewitness memory for faces .Although research shows that alcohol somewhat impairsmemory for verbal materials, its influence on face recog-nition or person identification tasks has varied acrossstudies . Whereas some studies have shown no effect ofalcohol on lineup identification, others have demon-strated impairment in both recall and lineup identifica-tion performance, particularly when witnesses consumedalcohol prior to viewing the event and arousal was lim-ited. Of course, drugs that impair or disrupt perceptionnecessarily impair encoding and memory process as well .

3 .3 . Witness Race, Gender,and Occupation

No reliable evidence demonstrates any general advan-tage or disadvantage in eyewitness memory or identifica-tion associated with the race, gender, or occupation ofthe eyewitness [although, it will be noted later that therace(s) or gender(s) of the eyewitness and perpetratormay interact] . However, men and women appear todiffer in the type of information they recall about an

event . For example, women tend to recall more itemssuch as clothing, whereas men focus on items such as thetype of vehicle or weapon .Law enforcement officers are commonly assumed to

be superior at identifying faces and remembering thedetails of events . Although research has generally foundthat the level of experience police officers have, either inyears on the job or through training, does not signifi-cantly enhance their recognition ability, it does have apositive influence on the quality of the descriptions theyprovide . Officers are generally able to provide moredetailed accounts of an event, and they appear to beless susceptible to the effects of postevent (mis)informa-tion when compared to laypersons . Officers also providemore correct descriptive information about the perpe-trator than do laypeople, without an increase in incor-rect information, and tend to elaborate on action detailsmore than laypeople . This elaboration on action details,however, must be weighed against officers' greater ten-dency to misperceive innocent actions as criminal innature. Empirical studies attempting to train individualsto remember events and faces have demonstrated asimilar pattern; namely, individuals' recall accuracycan be improved for details of an event, but it is difficultto improve recognition accuracy for faces .

3 .4. Witness Confidence

Eyewitnesses will often claim to have great confidencein their ability to identify a perpetrator, but for morethan 30 years, research examining the utility of con-fidence as a postdictor of accuracy has generallydemonstrated a weak relationship between degree ofcertainty and identification accuracy . However, thereare some conditions in which a stronger relationshipmay exist. Initial judgments made with very high con-fidence, for instance, have been shown to be quitediagnostic of witness accuracy .

Even though a witness's confidence may initially berelated to the accuracy of the memory, that relationshipcan be changed dramatically by events that occur afterthe eyewitness makes a positive identification . If eye-witnesses are told immediately following a positive iden-tification that they have correctly identified a suspect, notonly does their degree of confidence increase but alsotheir memory for the crime itself may change . For exam-ple, they may become more likely to report that they sawthe criminal longer and under better viewing conditionsthan they had previously reported . In summary, althoughevidence exists for a relationship between confidence andaccuracy under some conditions, confidence has been


shown to be quite malleable . Thus, a witness's statementof confidence, in general, may prove to be a poor indi-cator of identification accuracy .

4. THE PERPETRATOR

Thus far, we have considered only one of the mainactors in the drama that unfolds when an eyewitnessobserves an event, namely the eyewitness himself orherself. The other actor is, of course, the perpetrator(or perpetrators) of the crime, and several factors havebeen shown to influence memory for the perpetrator .

4.1 . Disguise

Disguises are frequently used during the commission ofa crime, and their presence can significantly impairencoding of the perpetrator's face . Disguises may includemasks, sunglasses, or anything that obscures significantportions of the face . In general, the upper portions of theface (e.g ., eyes and hair) provide the most importantinformation for later identification, and disguises thathinder the encoding of these parts of the face are morelikely to prevent identification of the perpetrator .

4.2 . Distinctiveness and Typicalityof the Perpetrator

In general, typical faces are more difficult to distin-guish from other faces in memory, resulting in a higherlikelihood of false identification . In contrast, if theoffender is distinctive in some way, recognition maybe enhanced because the presence of unusual attributes(such as Richard Nixon's chin or Mikhail Gorbachev'sforehead birth mark) can make a face easier to remem-ber and help to distinguish it from other faces inmemory. However, if the lineup identification proce-dure is carried out properly (as discussed later) thedistinctive suspect will be placed among others whoshare the same distinctive attribute, requiring the wit-ness to identify him or her without relying solely onthe memorable attribute . If the witness's attention waspreviously drawn to the distinctive attribute, to theneglect of other attributes or away from a holisticperception of the face, then the witness's identificationaccuracy may be decreased .

Other facial attributes, such as attractiveness or facial"typicality," have been shown to increase the sense thata face has been seen before, resulting in an increase in

879

880

the rate of false identifications . There also appear to be anumber of widely held stereotypes about general facialappearance, which influence the encoding of informa-tion about faces and may therefore influence lineupidentifications by eyewitnesses . For example, faces thatare stereotyped as "criminal" in appearance are morelikely to be identified from a lineup .

4.3 . Familiarity Due to RepeatedViewing

Not surprisingly, if the offender was previouslyunknown to the witness, recognition is less likely thanif the offender was known to the witness . However, evenwhen a perpetrator seems familiar, witnesses may notalways be correct in their perceptions, especially if asense of familiarity develops later in the investigativeprocess. If the suspect was viewed committing thecrime, the witness may have increased feelings of famil-iarity at the time of recognition and identification .However, if the suspect did not commit the crime, it isstill possible that familiarity can develop after the wit-ness views mug shots and photo spreads containingpictures of the innocent suspect . Following repeatedviewings, the appearance of the suspect begins to seemfamiliar to the witness . Errors caused by repeated view-ing have been attributed to errors in source monitoring,or source confusion . According to the source monitoringhypothesis, memory errors occur when a personattempts to identify where the memory (i .e., of theperpetrator) originated following the receipt of poste-vent information that leads the witness to erroneouslyattribute the new information (i .e ., the innocent suspect)to the original witnessed event . For this reason, in-courtidentifications of a defendant provide little real evidencethat the defendant actually committed the crime .Familiarity at this point has either been artificiallystrengthened (when the suspect is the offender) or cre-ated by previous viewing of the suspect's person orphoto (when the suspect is not the offender) .

4 .4. Perpetrator Race and Gender

Although the race of the perpetrator alone generally haslittle influence on identification accuracy, when the raceof the perpetrator and the witness are different, theinteraction can impair identification accuracy . Studies ofthe cross-race effect or the own-race bias have shown thatmemory for same-race faces is generally superior tomemory for faces of another, less familiar race . The effect


has been consistently demonstrated over a 35-year periodand has been reliably observed with various ethnic groups(e .g ., Europeans, Southern Africans, Americans, Asians,and Hispanics) . The cross-race effect is of most signifi-cance to the criminal justice system when individualsmistakenly identify a suspect who is not the perpetrator .

Although female faces are generally better recognizedthan male faces, a similar interaction between the genderof the perpetrator and witness has also been noted .Studies of the own-gender effect have demonstratedthat female participants tend to outperform male parti-cipants in remembering female faces . Curiously, though,male and female participants do not differ consistently intheir ability to remember male faces .

5 . OBTAINING INFORMATIONFROM THE WITNESS

Eyewitness accounts of criminal events play a vital partin solving crimes and prosecuting criminals, butresearch has shown that very little of what witnessesactually see ever gets reported to the authorities . Onereason for this limited communication has to do with themanner in which information is elicited during policeinterviews of witnesses . Police interview techniques areoften passed on from veteran officers to rookies orlearned by trial-and-error on the job . Research hasshown that even police officers without formal trainingtend to share similar interviewing styles, and policeinterviews of eyewitnesses are often very loosely struc-tured. After requesting open-ended descriptions, officersfrequently interrupt witnesses by asking specific, direc-tive questions about the crime, typically pertaining toperpetrator height and weight . These questions usuallyelicit brief answers from witnesses and disrupt the flowof information. This directive interview style limits theinformation that witnesses convey and also appears toset up a dynamic between witnesses and officers inwhich witnesses wait passively for officers to direct theinterview. As a result, information not specificallyrequested by the officer may never get mentioned duringthe interview, despite its importance to the case .

Interview methods can seriously influence the qualityof information given by witnesses in other ways, produc-ing information that is less accurate or even fabricated .A classic example of interviewer influence on witnessreports comes from the highly publicized McMartin pre-school case in which accusations of ritual child sexualabuse were made against seven preschool teachers .

Videotaped interviews revealed that the children weresubjected to highly suggestive and leading questions .When these methods were applied in laboratory studies,children's false allegations greatly increased . Experiencessuch as the McMartin case and studies of actual policeinterviews reveal the importance of understanding theeffects of interviewer influence on eyewitness reports .

5 .1 . The Cognitive InterviewOne interview technique developed by psychologistsexplicitly for the purpose of enhancing the retrieval ofeyewitness memory (and limiting the detrimental effectsof interviewer influence) is the cognitive interview .Developed by Geiselman and Fisher in the early 1980s,the cognitive interview consists of four main compo-nents: (i) context reinstatement, which includes men-tally reinstating the environmental and personal contextof the original event; (ii) instruction to "report all"information including partial information, even if itseems unimportant ; (iii) recounting the event in a vari-ety of temporal orders ; and (iv) reporting the eventsfrom a variety of perspectives . A wealth of research hasinvestigated the potential benefits of this technique, bothin laboratory and field settings . A meta-analysis of thesestudies revealed a large increase in the number of correctdetails elicited by the cognitive interview and a smaller,yet significant, increase in the number of incorrectdetails elicited. However, the mesa-analysis also indi-cated that accuracy rates (i .e ., the proportion of correctdetails generated) elicited using the cognitive interviewwere about the same as accuracy rates using traditionalinterview methods (84% for the cognitive interview and82% for standard interviews) .

In subsequent research, Fisher, Geiselman, and col-leagues devised the enhanced cognitive interview, amodified technique that attempted to incorporate stra-tegies such as rapport building to manage the socialdynamics of communicating effectively with a witness .These changes resulted in further increases in theamount of correct information obtained, althoughsome studies also observed significant increases in theamount of incorrect details . Additional research is war-ranted. Training law enforcement officers and others toconsistently employ all the aspects of the cognitive inter-view has proved difficult.

5 .2 . Guided MemoryResearch shows that when various aspects of the cogni-tive interview are broken down, context reinstatement

Eyewitness Identification 88 1

instructions appear to play a vital role in the effective-ness of a cognitive interview . Context reinstatementtakes place when witnesses are provided with contextualcues about the to-be-remembered event . The basis forthe effectiveness of context reinstatement comes fromTulving's encoding specificity principle, which main-tains that retrieval cues will enhance memory when theinformation contained in the retrieval cues matchesinformation contained in the original memory trace . Amesa-analysis of facial recognition studies revealed thatcontext reinstatement strategies produce some of themost substantial benefits to identification accuracy .

The guided memory technique was primarily designedto utilize context reinstatement strategies for enhancingmemory. This technique involves having witnessesvisualize aspects of the event, including physical featuresand traits of the perpetrator, along with various emo-tions and reactions elicited by the surrounding event .Although this method has been effective in enhancingcorrect recognition of perpetrators, little is known aboutthe effects of guided memory on description quality oraccuracy. Nevertheless, the guided memory technique,along with other context reinstatement strategies, hasbeen shown to enhance the retrieval of descriptive infor-mation from witnesses and offers a promising alternativeto current interview techniques.

5 .3 . Generating DescriptionsArchival research has shown that although descriptionsprovided by witnesses are frequently vague and lackingin detail, they are generally quite accurate . The descrip-tors most often reported by witnesses relate to actionevents. Besides action events, witnesses are called on toprovide person descriptions, crime scene descriptions(i .e ., objects and environmental details), and accountsof what was said during the crime . With regard to theperpetrator, research indicates that witnesses are muchmore adept at describing character traits (e .g ., he lookedlike an accountant) and psychological attributes (e .g ., hewas/looked crazy, kind) than physical aspects of a face,and that such global or holistic judgments improvesubsequent attempts at recognition .

5 .4 . Verbal OvershadowingAlthough generating a description of the perpetrator isregarded as a benign activity, studies have indicated thatverbally describing a target face may have a detrimentalinfluence on subsequent accuracy at lineup identification .This phenomenon is referred to as verbal overshadowing .

882

Witnesses who provide a description of a target faceperform worse on a subsequent identification task com-pared to witnesses who did not provide a description .The effect occurs regardless of whether the perpetrator ispresent or absent from the lineup or whether the lineup isadministered sequentially or simultaneously . The nega-tive effect is most pronounced when witnesses areencouraged to adopt a liberal response criterion and toreport their memory in great detail .

5.5 . Postevent (Mis)Information

Many things can happen to eyewitnesses between thetime they give their report to the police and the timethey subsequently attempt a lineup or in-court identifi-cation. Some intervening events can adversely (or posi-tively) affect their identification accuracy . Eyewitnessesmay read newspaper reports about the event they wit-nessed, they may talk to other eyewitnesses about thesame event, or they may be exposed to additionalinformation during the course of the police investigationinto the crime . This is known as postevent information .Its (potentially) strongly biasing effects on eyewitnesstestimony have been extensively investigated and docu-mented by Elizabeth Loftus and colleagues .

5.6. Mug Shot Collectionsand Composite Reproductions

Following the initial interview with the witness, policemay seek to obtain a potential physical likeness of theperpetrator by having the witness attempt to recognizethe perpetrator from among a library of criminal mugshots or by having the witness create a composite imagebased on his or her memory.

5.6.1 . Searching Mug Shot Collections

Mug shot collections are maintained by many law enfor-cement agencies for the simple reason that criminalrecidivism rates are high, and many perpetrators havebeen photographed by the police on a previous occasion .Researchers have investigated the influence of searchingmug shot collections on later identification accuracy . Thisquestion has very important legal ramifications since eye-witnesses who have been exposed to a mug shot collec-tion prior to making an identification of a suspect from alineup may be more susceptible to a source monitoringerror, mistaking the intervening mug shot for theperpetrator. Some studies demonstrated that mug shot

Eyewitness Identification.

exposure can "contaminate" an eyewitness's memory,but others did not. Legally, the fact that an eyewitnesshas been subjected to a potentially biasing and avoidableprocedure may be enough to discredit that witness .

A second, often neglected, research question concernsthe cumbersome manner in which mug shot searches areconducted . Typically, large collections of photographsare stored in albums (printed or electronic), leaving theeyewitness to page through several thousand photo-graphs in a vain attempt to "spot" the perpetrator . Someresearchers have tried to improve mug shot search pro-cedures. For example, English researchers devised adescription-driven system in the late 1980s calledFRAME, which appeared to work quite well but waslater criticized for depending on a limited database . Forinstance, there were very few mug shots of old men in thedatabase ; thus, it was comparatively easy to search thedatabase for old men, inflating the accuracy rate . Asimilar, description-based system has been developed inCanada, and although it appears promising, there is notenough research to draw a definitive conclusion .Researchers in Hong Kong are exploring ways of struc-turing mug shot searches according to the physical simi-larity of faces, but initial results have been disappointing .

5.6.2. Making and Utilizing CompositePortraits

Law enforcement agencies in many parts of the worldrely on practical technologies to help crime witnessesreconstruct likenesses of faces . These technologies rangefrom sketch artistry to proprietary computerized com-posite systems such as Identikit, Photofit, E-Fit, Mac-a-Mug, Faces, and Comp-U-Sketch . Unfortunately, thesetechnologies have not performed well under empiricalexamination . In most instances, they appear to producepoor-quality composites, which are difficult to match totarget faces, even when the target is in full view of thewitness. There was some hope that the move to compu-terized composite software would result in systems con-taining much larger libraries of features that would beeasier to search and graphically "post process" (e .g .,allowing the easy addition of specific facial featuressuch as scars or alteration of the aspect ratio of a face) .There is no evidence that computerized systems (e.g .,Faces and E-Fit) lead to better reconstructions thanmanual" systems (e.g., Identikit and Photofit), except

when the target is in full view of the witness during thereconstruction, in which case there is a clear advantagefor computerized systems .

Recent improvements in computing power have pro-duced more versatile, mathematically sophisticatedcomposite systems . Several groups of researchers areexperimenting with "eigenface" systems that potentiallyallow witnesses to search limitless populations of faceswith the assistance of genetic algorithms . Results arepromising when witnesses are allowed to reconstructfaces while they are in full view but not from memory .There is evidence that verbal descriptions may be ofmore assistance to eyewitnesses than visual likenessesor reconstructions .

6 . THE WITNESS IDENTIFICATION

6.1 . Measuring Lineup Fairness

There are two key events for an eyewitness, as far as thelaw is concerned . The first is the event they witnessed,and the second is the identification of the personwhom they saw commits the crime . When and how thesecond event unfolds is just as critical as the firstbecause this identification will constitute the evidenceagainst the defendant . Historically, eyewitnesses wereprimarily asked to identify perpetrators in court, butthis practice is frowned upon nowadays since it is verysuggestive (indeed, the witness would look foolish ifhe or she pointed anywhere but at the defendant) . Analternative approach is for police to arrange an encoun-ter between the eyewitness and the accused and tohope for a spontaneous identification (i .e ., a "show-up"identification) . This slightly less suggestible techniquestill jeopardizes the liberty of innocent people .

The most widely used alternative to in-court identi-fications in the United States, and in many othercountries, is the "lineup," in which the suspect isplaced alongside a number of men (or women) ofreasonably similar physical appearance and demeanor,and the witness is asked to choose the perpetrator if heor she is present in the array . Police lineups originatedin English criminal law and procedure, and it is clearthat the notion of "fairness" is their raison d'etre .Lineups are intended to secure an identification thatcan potentially incriminate the perpetrator, but onethat is fair to innocent people who might be suspects .Lineups are not invariably fair in everyday practicesince innocent people have been convicted on thebasis of eyewitness identification from a lineup . Oneof the strongest strands of eyewitness research is thatinvestigating all aspects of the fairness of lineups, fromtheir construction to their administration. An array of

Eyewitness Identification 883

methods and measures that can be applied to policelineups has been developed .In a method known as mock witness evaluation,

researchers ask people who did not see the crime eventand are "blind" to the identity of the perpetrator (orinnocent suspect) to try and identify the suspect in thelineup. This is usually achieved by giving mock wit-nesses a brief description of the perpetrator (preferablythe very description the eyewitness gave to the police)and asking them to indicate the lineup member who bestmatches this description. If mock witnesses are able toidentify the suspect at a rate greater than l/k (where k isthe number of lineup members), the lineup is said to bebiased. The proportion of mock witnesses choosing thesuspect is a measure of "lineup bias ."

A second aspect of lineup fairness concerns the num-ber of plausible foils it contains . There are records ofpolice lineups in the United States and other countrieswhere the suspect was the only representative of aparticular race or ethnic group in the lineup . In suchlineups, the suspect's identity is immediately suggestedto the identifying witness . The number of plausiblelineup members is referred to as the effective size andis distinguished from the nominal (or actual) size of thelineup. There are currently two measures of effective sizein use : (i) a descriptive measure formulated by Malpassin 1981 and (ii) a closely related inferential measureformulated by Tredoux in 1998 . For a lineup to beconsidered fair, it should receive favorable scores onmeasures of both lineup bias and lineup size .

The validity of lineup fairness measures has beenevaluated in a small number of studies, with positiveresults . Lineup fairness is frequently assessed byresearchers examining police lineups in specific legalcases . A small literature exists regarding proactive useof researchers in constructing lineups that meet fairnessstandards on a priori grounds. A substantial discussionof lineup fairness appeared in a special edition of thejournal Applied Cognitive Psychology in 1999 .

6.2 . Lineup Instruction Effects

Some legal and criminal jurisdictions prescribe theinstructions, or admonitions, that officers are to provideto eyewitnesses when presenting them with a lineup .The key question is whether these warnings have aneffect on the rate of incorrect identification of innocentsuspects, which is the main scourge of police lineups .Empirical studies have explored the effect of biasedinstructions, and several staged-crime experimentshave shown that instructions that presuppose the

884

presence of the criminal in the lineup ("Point out theperson who committed the offence") lead to muchhigher error rates than instructions that do not ("Notethat the criminal may not be present . If the criminal ispresent, point him out") . The implication is that theinstructions given to witnesses should be temperedquite carefully with a warning indicating the perpetra-tor's possible absence . Research comparing a set ofclearly biased instructions with instructions usuallygiven by the Los Angeles Police Department and a setof "more balanced" instructions showed an increase inmistaken identifications for the biased instructions butno such increase with either of the other two sets ofinstructions .

6.3 . Alternative Strategies : BlankLineups, Sequential Lineups,

and Relative Judgments

That lineups produce high rates of false identificationhas been known for a long time, and researchers havepuzzled over the causes . The cause was long thought tobe that witnesses are led to frame their decision task aschoosing the offender from among the people in thelineup, thereby selecting the best choice from amongthe alternatives presented. The witness enters the identi-fication task presuming that the police would not con-duct a lineup if they did not have a suspect firmly inmind and thus interprets the task as requiring him or herto identify the lineup member that the police suspect .A number of additional strategies to reduce erro-

neous identifications, apart from the lineup instruc-tions and admonitions discussed previously, havebeen proposed. One solution, suggested approximately50 years ago by Glanville Williams, the famous Englishlegal authority, is to present the witness with a blanklineup (a lineup known not to contain the offender/suspect) to "trap" people who feel they must make anidentification into making a harmless one . Witnesseswho choose someone from this blank lineup are clearlymistaken and are dismissed as too unreliable to com-plete the main identification task . Only witnesses whodo not pick from the blank lineup are presented with asuspect-present lineup . Studies have employed blanklineups in staged crime experiments and discoveredthat witnesses who choose a member of the blanklineup are almost twice as likely to make an incorrectidentification from a subsequent perpetrator-presentlineup and likewise only about half as likely to cor-rectly identify the perpetrator .


There are a number of problems introduced by such anapproach. First, blank lineups are not feasible in practicesince the public would quickly learn that a lineup consistsof two parts, only the second of which involves the realsuspect, rendering the blank lineup ineffectual . Second, itburdens law enforcement with constructing twice as manylineups as they currently construct. (The use of moderncomputing systems and large mug shot databases maysolve this problem in the near future.) Third, law enforce-ment resists "throwing away" a witness who has not yetbeen given a chance to view the suspect on grounds that awitness may make a choice from the blank lineup andsubsequently, when presented with the suspect-presentlineup, enthusiastically identify the suspect, claimingthat he or she was mistaken before but is now sure this isthe offender . Expecting police officers to forego the poten-tial for this "evidence" might be too much to ask .

Another structural manipulation or alteration thathas attracted a great deal of attention from psychologicalresearchers is known as the sequential lineup . Instead ofpresenting multiple lineups in sequence, or members ofone lineup simultaneously (exposing the witness to thesuspect and a number of foils at the same time), theindividual lineup members are presented one at a timeand the witness is required to identify the perpetratorfrom the members of this sequence . Witnesses are led tobelieve that there are more photos to be seen than theyare actually shown and are instructed to decide if eachphoto is or is not the perpetrator . This procedure isbased on an analysis of the lineup task articulated byGlanville Williams in 1963 ("The witness may . . . beinclined to pick out someone, and that someone will bethe one member of the parade who comes closest to hisown recollection of the criminal") and psychologicallyinterpreted by Rod Lindsay and Gary Wells in 1985 .They suggested that the structure of the conventionalpolice lineup (all members presented at once) inviteswitnesses to compare the members of the lineupto one another and identify the person that "bestmatches" the witness's memory, leading to a decisionbased not on the virtual correspondence of the witness'smemory of the perpetrator to one of the lineup membersbut on the relative correspondence of the lineup mem-bers to the witness's memory, a relative judgment strat-egy. Presenting lineup photographs sequentially is said toinhibit eyewitnesses from relying on relative judgmentsand, instead, fosters absolute judgments made by com-paring their memory of the perpetrator with each lineupmember. Empirical studies have demonstrated that theuse of sequential lineups can reduce false identificationsand increase correct rejections of lineups that do not

contain the perpetrator, although a significant reductionin the rate of correct identifications in lineups that docontain the perpetrator has also been noted .

The apparent advantages of sequential lineups hasmade them attractive to law enforcement authorities insome jurisdictions, and they have been promoted veryactively to law enforcement by some members of theeyewitness research community . Research on sequentiallineups has been very important, but there is also con-troversy over whether the research has been extensiveand careful enough to warrant promotion of sequentiallineups as a replacement for traditional police lineups .

6.4. Lineups Involving OtherSensory Modes

The lineup is usually a test of visual memory, but mostlegal systems also recognize that additional cues toidentity may reside in the voice, mannerisms, and gaitof the perpetrator .

6.4.1 . Earwitness Identification

The case of voice identification has attracted muchattention in the research literature . The most pertinentfinding has been that familiarity of the voice plays a keyrole in identification, and that identification of unfami-liar voices is prone to yield alarmingly high false-alarmrates. Unfamiliar voices are very difficult to recognizeand especially easy to confuse . Similar to the own-racebias in facial identification, several studies have observedthat voices of another, less familiar race are also moredifficult to recognize . In addition, a witness's memoryfor the voice of the perpetrator is equally susceptible tothe many estimator variables discussed previously,including opportunity to hear the voice clearly andvarious factors that may draw the witness's focus ofattention away from the auditory stimulus .

6.4.2. Multimode Lineups

Lineups typically used in Western countries restrict thepresentation of the lineup to one mode, most frequentlyinvolving visual presentation . However, cues from asingle (sensory) domain may be degraded with respectto the original context (e.g., physical disguise), and it iseasier to identify the perpetrator with the assistance ofcues from other domains. Researchers have begun toexplore the utility of multimode lineups on witnessidentification performance . Although there is little


published research, the idea of allowing witnesses accessto cues that are not solely visual in nature is already inpractice in Sweden. The lineup is formed in a room inwhich lineup members are allowed to sit down, smoke,communicate, and behave in general as they ordinarilywould over an extended period of time . These individ-uals are then viewed by the witness from behind a one-way mirror, and members of the lineup are not awarewhen the witness is present. Unfortunately, no systema-tic evaluation of this strategy has been performed .

6.4.3 . Live vs Photographic Lineups

Lineups currently differ within single sensory modes,such as the practice of eliciting identifications fromphotographic lineups as an alternative to corporeal(live person) lineups. In many areas of the UnitedStates, for example, photographic lineups are favoredover corporeal lineups . In many foreign jurisdictions,however, identifications from corporeal lineups arerequired if eyewitness identification evidence is tobe presented in court. Intuitively, corporeal lineupswould seem to have an advantage: The detail to beobtained from a live inspection of the suspect mustsurely exceed the detail present in a photograph .However, the consensus from several studies is thatthere is either very little difference between thecapacity of photographic and corporeal lineups toelicit identifications or no difference at all . An exten-sive review of the literature reported no notabledifference between live, video, or photographic line-ups. There is no evidence to suggest that media thatembellish cues will aid identification accuracy orreduce false alarms .

7 . THE PROSECUTION

After the police obtain a positive identification fromthe eyewitness that their suspect is indeed the perpe-trator, they are ready to hand the case over for prosecu-tion. The district attorney (or attorney general ordirector of public prosecutions) may decide not toprosecute, but in most jurisdictions eyewitness identi-fication is considered strong evidence against the per-petrator and the case will proceed to court . A numberof factors in the court process can influence an eye-witness's accuracy and can change the impact that aneyewitness's testimony will have at trial . For instance,the prosecution (or defense) can pose leading ques-tions to the witness, eliciting false recollections, or an

886

Eyewitness

expert can be called to testify on the vagaries of eye-witness identification. In the following sections, someof the research on two potentially important factors issummarized . First, we briefly discuss a debate thatconcerns the role and value of expert testimony byeyewitness researchers in legal trials . Second, we dis-cuss research that has examined the effectiveness ofjurors in assessing the reliability of eyewitness testi-mony at trial and whether juries benefit from expertassistance on eyewitness memory .

7 .1 . Expert Testimony

Many researchers have offered expert testimony to thecourts in cases in which eyewitness evidence has beenat issue, although this undertaking has been somewhatcontroversial within the discipline of psychology. Avariety of arguments have surfaced on this issue,including disputes over the consistency of findingsand the adequacy of current knowledge for presenta-tion to the courts . Several surveys of researchers andexperts in the eyewitness literature by Saul Kassin andcolleagues have attempted to address these concerns byidentifying areas of agreement. In 1989, this surveyrevealed that the following areas were considered "reli-able" by the majority of respondents : limited exposuretime, lineup instruction effects, prior expectations onthe part of the witness, postevent misinformation, andthe weak relationship between confidence and accu-racy. An update to this survey of experts in 2001demonstrated some additions to this listing thatincluded confidence malleability, mug shot-inducedbiases, child witness suggestibility, the effects of alco-hol intoxication, the cross-race effect, the weaponfocus effect, forgetting over time, and lineup presenta-tion manipulations (simultaneous vs sequential line-ups) . Given the developing nature of the science ofeyewitness identification, it is reasonable to assumesome shifts in experts' opinions regarding variousphenomena . Nevertheless, Kassin and colleaguesnoted the remarkable consistency in ratings acrossthe two surveys for the majority of topics covered .

7 .2 . Jurors' "Common Knowledge"of Problems with Eyewitness

Identification

One basis for rejecting an eyewitness expert's testimonyin court is the belief that factors affecting the reliability ofeyewitness identification are common knowledge to the

Identification

lay juror . Researchers have sought to determine exactlywhat laypersons know about factors affecting eyewit-nesses. Three basic methodologies have been used toinvestigate this information: (i) surveying jury-eligiblecitizens as to their knowledge and beliefs, (ii) assessingjurors' ability to predict the outcome in an eyewitnessidentification experiment, and (iii) using mock trials toassess the influence of trial techniques (such as cross-examination of the eyewitness or the presence/absenceof expert evidence on eyewitness memory) .

First, survey studies have been conducted by admin-istering questionnaires, such as the Knowledge ofEyewitness Behavior Questionnaire, in order to assessbeliefs about factors that affect the accuracy and relia-bility of eyewitness identification . The results of suchstudies have demonstrated that respondents appearedinsensitive to such effects as the age of the witness(young or old) and the retention interval prior toidentification, and that respondents tended to believe(contrary to research findings) that training couldimprove identification accuracy. Second, studies haveasked participants to read written summaries of identi-fication experiments and then to "postdict" the accu-racy of participant-witnesses . Results in such studieshave indicated that participants overestimate accuracyrates, suggesting that individuals often believe wit-nesses to be much more accurate in their judgmentsthan they truly are . Finally, research employing themock trial" as a method for assessing jurors' common-sense knowledge has manipulated different factorsknown to influence, or not to influence, identificationaccuracy and then assessed whether participantjurorsare sensitive to these factors in their verdicts . A com-mon outcome is that witness confidence is given toomuch weight, with jurors believing confident witnessesto be more accurate, which may not necessarily be thecase. In contrast, jurors failed to assign sufficientweight to other factors, such as opportunity to observethe perpetrator or the presence of a weapon, whichshould have been considered . Taken together, thisresearch appears to demonstrate that lay jurors lackthe requisite knowledge to appropriately evaluate eye-witness identification evidence, and that expert wit-nesses may be useful for providing such informationat trial . Several studies have also attempted to assessthe benefits of expert testimony within the mock trialparadigm. Their results have generally shown thatexposure to an eyewitness expert leads jurors to moreappropriately weight identification evidence based onfactors known to influence the reliability of anidentification .

8 . CONCLUSION

In a 1978 paper that has been extraordinarily influen-tial in the eyewitness literature, Gary Wells argued thatthe quintessence of eyewitness research is its applica-tion to a legal problem and advised that it is incumbenton eyewitness researchers to show the practical utilityof their research . This approach has indeed yieldedgreat rewards: One of the most important pieces inthe eyewitness literature is the coconstruction ofguidelines by psychologists, lawyers, and criminal jus-tice professionals for collecting evidence from eyewit-nesses, sponsored by the National Institute for Justice .

There is a cost, though, to the near-exclusive appliedresearch focus taken by eyewitness researchers . Anapplied focus typically leads to research that is interestedin improving practice (e.g., lowering the rate of false-positive identifications) but is less concerned withexplaining how and why a particular intervention works .For example, we know that the sequential lineup returns alower rate of false positives than the traditional policelineup, but we do not have a good answer for why this isthe case . In general, eyewitness research has not appre-ciated the value of theoretical explanations and models .This is not surprising given the explicit applied orienta-tion of researchers in the area, but it may now be time forresearchers to redress the imbalance . A key reason forpromoting such a redress is the belief that the applicationsthat stem from applied eyewitness research could be morepowerful if they were based on an explanatory model. Ifwe understood the mechanism, for instance, that makesconservative instructions reduce the false-positive rate,then we might be able to design alternative forms of lineupthat maximally reduce this rate .

We do not wish to propose that eyewitness researchjettison its applied orientation: On the contrary, wewish to strengthen that orientation . The question ishow this should be done . Our contention is that areturn to the laboratory might, paradoxically, create astronger applied foundation for the discipline .

See Also the Following Articles

Forensic Mental Health Assessment ® Interrogation andInterviewing


Further Reading

Fisher, R . P ., & Geiselman, R. E. (1992) . Memory enhancingtechniques for investigative interviewing . Springfield, IL :Thomas .

Kassin, S . M., Tubb, V. A., Hosch, H . M., & Memon, A .(2001). On the "general acceptance" of eyewitness testi-mony research: A new survey of the experts . AmericanPsychologist, 56, 405-416 .

Loftus, E . F . (1979) . Eyewitness testimony . Cambridge, MA :Harvard University Press .

Malpass, R. S. (1981). Effective size and defendant bias ineyewitness identification lineups . Law & Human Behavior,5, 299-309 .

Malpass, R. S., & Devine, P . G. (1980). Realism and eye-witness identification research . Law & Human Behavior, 4,347-358 .

Malpass, R . S., & Lindsay, R. (1999). Measuring line-upfairness . Applied Cognitive Psychology, 13, S1-S7 .

Meissner, C . A., & Brigham, J. C. (2001) . Thirty years ofinvestigating the own-race bias in memory for faces : Amesa-analytic review . Psychology, Public Policy, & Law, 7,3-35 .

Ross, D., Read, J . D., & Toglia, M . (1994) . Adult eyewitnesstestimony: Current trends and developments . London :Cambridge University Press .

Shapiro, P . N., & Penrod, S. (1986) . Meta-analysis of facialidentification studies . Psychological Bulletin, 100, 139-156.

Sporer, S. L., Malpass, R. S., & Koehnken, G. (1996) .Psychological issues in eyewitness identification . Hillsdale,NJ: Erlbaum.

Steblay, N . M., Dysart, J ., Fulero, S., & Lindsay, R . C. L .(2001). Eyewitness accuracy rates in sequential and simul-taneous lineup presentations : A mesa-analytic comparison .Law & Human Behavior, 25, 459-473 .

Technical Working Group for Eyewitness Evidence . (1999) .Eyewitness evidence: A guidefor law enforcement [Booklet] .Washington, DC : U .S . Department of Justice, Office ofJustice Programs, National Institute of Justice .

Tredoux, C . (1998). Statistical inference on measures oflineup fairness . Law & Human Behavior, 22, 217-237 .

Wells, G . L. (1978) . Applied eyewitness testimony research :System variables and estimator variables . Journal ofPersonality & Social Psychology, 36,1546-1557 .

Wells, G . L., & Olson, E. A . (2003) . Eyewitness testimony .Annual Review of Psychology, 54, 277-295 .

Wells, G . L., Small, M ., Penrod, S ., Malpass, R . S ., Fulero, S . M .,& Brimacombe, C . A. E. (1998) . Eyewitness identifi-cation procedures : Recommendations for lineups andphotospreads. Law & Human Behavior, 22, 603-647.

eyewitness identification lab - university of texas at el paso

Documents