Eyewitness Identification - Semantic Scholar are unaware of the problems with eyewitness ... 2 People v , Guzman 47 Cal . App 3rd 380 121 Rptr 69 ... affect not only the likelihood

Eyewitness IdentificationWhat Can a Psychologist Tell a Jury?

Michael McCloskey Johns Hopkins UniversityHoward E. Egeth Johns Hopkins University

ABSTRACT: Psychologists have long been concernedabout the use of eyewitness testimony in the court-room. Recently, it has been suggested that experi-mental psychologists should testify as expert wit-nesses in cases involving eyewitnesses to inform thejury about problems with eyewitness testimony. Inthis article we examine the arguments offered in fa-vor of the use of expert testimony about eyewitnesses.We suggest that contrary to strong claims made re-cently by several psychologists and lawyers, it is byno means clear that expert psychological testimonyabout eyewitnesses would improve jurors' ability toevaluate eyewitness testimony. In fact, it is even pos-sible that this sort of expert testimony would havedetrimental effects. We suggest that experimentalpsychologists should carefully consider the issuesraised in this article when deciding whether to offerexpert testimony.

Imagine the following scene. An experimental psy-chologist is sitting in his or her office, lost in thought,when the phone rings. An attorney is calling.

Psychologist: Hello.Attorney: Hello, Dr. Smith, this is Joe Doaks;

I'm an attorney. I've been retained by a man whohas been charged with armed robbery. The prose-cution's case does not look very impressive to meexcept for one point—at a lineup my client waspositively identified by an eyewitness to the crime.

What I'm calling about is to ask if you'd bewilling to serve as an expert witness for the defenseto explain to the jury the problems with eyewitnesstestimony.

Psychologist: Well, I don't know. I don't thinkI'd be comfortable going into a courtroom and im-pugning the testimony of a witness.

Attorney: No, no, of course not. I would notwant you to try to tell the jury whether or not theeyewitness is right. The judge would never allow suchtestimony anyway. All I would like you to do is talkin fairly general terms about the problems in per-ception and memory that can occur when a personobserves a crime. I mean, the whole thing took placevery quickly, the witness was very frightened, and

the lineup didn't take place until 10 days after thecrime. And on top of that, my client is black andthe witness is white. These factors all can affect theaccuracy of an eyewitness identification.

Psychologist: You seem to be quite knowl-edgeable about eyewitness testimony yourself; whatdo you need me for?

Attorney: Basically, I can only ask questions.I'm not permitted to lecture the jury about psy-chology. But you can. You would not be the firstpsychologist to give expert testimony about prob-lems with eyewitnesses. Many of your colleagueshave testified as experts in cases involving eyewit-nesses, and many more probably will in the nearfuture. Judges sometimes don't allow this sort ofexpert psychological testimony, but in many casesthey do admit it.

Psychologist: This sounds very interesting, butI don't want to rush into it. Let me think about itfor a few days and do some reading.

Attorney: Very good. I'll get back to you in afew days. Goodbye.

Psychologist: Goodbye.

What should the psychologist do? What con-siderations should inform the decision to testify ornot to testify? In the following pages, we discuss sev-eral basic questions that we think the psychologistshould consider in the process of reaching his or herdecision. It should be made clear at the outset thatthis article has been written by and (largely) for ex-perimental psychologists. The content is psycholog-ical. We have made no effort to consider legal aspectsof the problem, such as laws affecting the admissi-bility of psychological testimony in various jurisdic-tions (such discussions are provided by Ellison &Buckout, 1981; Gass, 1979; Loftus & Monahan,1980; Woodier, 1977).

During the preparation of this article, the authors were supportedin part by National Science Foundation Grant BNS 80-22670.We would like to thank John Jonides, Judith McKenna, AllysonWashburn, Elizabeth Webb, Stephen Grossman, Rod Lindsay,Elizabeth Loftus, and Gary Wells for their helpful comments.

Requests for reprints should be sent to Michael McCloskey,Department of Psychology, Johns Hopkins University, Baltimore,Maryland 21218.

550 May 1983 • American PsychologistCopyright 1983 by the American Psychological Association, Inc.

A basic premise of our discussion is that in-tervention in the workings of the justice systemshould not be undertaken lightly. In particular, wetake the position that expert psychological testimonyabout perception and memory in eyewitnesses'should be offered only if there is clear evidence thatsuch testimony has salutary effects. As we discussin greater detail below, the use of expert psycholog-ical testimony in the absence of clear evidence of itsutility would carry substantial risks both for the sys-tem of justice and for the psychological profession.Consequently, in the following evaluation of argu-ments that have been offered in favor of the use ofexpert psychological testimony, we ask not, Doesthis claim seem plausible? or Might this assumptionbe valid? but rather, What does the available evi-dence say about this argument?

The final decision about whether to testify isup to the individual psychologist. However, we sug-gest that contrary to the claims of several psychol-ogists and lawyers (e.g., Addison, 1978; Ellison &Buckhout, 1981; Loftus, 1979; Loftus & Monahan,1980; Lower, 1978;Starkman, 1979; Woocher, 1977),the available evidence fails to demonstrate the gen-eral utility of expert psychological testimony and infact does not even rule out the possibility that suchtestimony may have detrimental effects.

Rationales for the Use of ExpertPsychological TestimonyTwo major rationales have been offered for the useof expert psychological testimony. First, the discrim-ination rationale asserts that jurors cannot discrim-inate adequately between accurate and inaccurateeyewitnesses (Lindsay, Wells, & Rumpel, 1981; Lof-tus, 1979; Loftus & Monahan, 1980; Wells, Lindsay,& Ferguson, 1979; Wells, Lindsay, & Tousignant,1980). Consequently, the argument continues, jurorsoften disbelieve accurate witnesses and believe in-accurate witnesses. According to this view, expertpsychological testimony could improve juror dis-crimination by informing jurors about factors knownto influence witness accuracy and by cautioningagainst reliance on irrelevant factors.

The second rationale for the use of expert psy-chological testimony asserts that jurors are in gen-eral too willing to believe eyewitness testimony (El-lison & Buckhout, 1981; Lindsay et al., 1981; Loftus,1974, 1979; Loftus & Monahan, 1980; Wall, 1965;Wells et al., 1980). According to this overbelief ra-tionale, an expert witness could increase juror skep-ticism to a more appropriate level by discussing re-search demonstrating the unreliability of eyewitnesstestimony and by pointing out aspects of the caseat hand (e.g., stress experienced by the witness) thatmight have led to witness inaccuracy. Loftus (1979)provides a strong statement of the overbelief claim:

"Since jurors rarely regard eyewitness testimonywith any skepticism, the expert testimony will in-crease the likelihood of this happening. This is itsvalue" (p. 197).

Both the discrimination and the overbelief ra-tionales make two fundamental claims: (a) Jurorsneed help in evaluating eyewitness testimony and(b) expert psychological testimony can provide thishelp. In the following sections we examine theseclaims.

Do Jurors Need Help?

Overbelief. Consider first the claim that jurors aretoo willing to believe eyewitnesses. Several argu-ments have been advanced in support of this claim.One argument that is frequently implicit in discus-sions of juror evaluation of eyewitness testimony(e.g., Ellison & Buckhout, 1981; Loftus, 1979;Woocher, 1977) is that the conclusion of juror over-belief follows from research showing that eyewitnesstestimony is often unreliable. An important but un-stated assumption here is that jurors are not awareof the unreliability of eyewitnesses and consequentlyare too willing to believe eyewitness testimony. How-ever, there is virtually no empirical evidence thatpeople are unaware of the problems with eyewitnesstestimony. Further, there appears to be no reasonto assume a priori that people are not cognizant ofthese problems. Cases of mistaken identification areoften widely publicized and wrongful conviction onthe basis of mistaken or perjured eyewitness testi-mony is a rather common theme in fiction. In ad-dition, there is no consensus within the legal com-munity that jurors are unaware of the unreliabilityof eyewitnesses and consequently give too much cre-dence to eyewitness testimony. For example, in rul-ing against the admission of expert psychologicaltestimony, the trial judge in the case of People v.Guzman2 stated: "It is something that everyoneknows about, the problems of identification. Thejurors here were well questioned regarding their ex-perience . . . with having mistakenly identified peo-ple. Everyone knows these things happen." Thus, inthe absence of evidence that jurors are unaware ofthe unreliability of eyewitness testimony, the con-clusion that jurors are too willing to believe eyewit-nesses cannot legitimately be drawn from researchdemonstrating that eyewitnesses are often inaccu-rate.

1 We will henceforth use the phrase "expert psychologicaltestimony" as a convenient shorthand for the more cumbersome"expert psychological testimony about perception and memoryin eyewitnesses." It should be understood that we are referringonly to expert testimony about eyewitnesses and not to othersorts of expert psychological testimony.

2 People v, Guzman, 47 Cal. App. 3rd 380, 121 Cal. Rptr.69(1975).

May 1983 • American Psychologist 551

A second argument asserts that juror overbeliefis demonstrated by the existence of documentedcases in which defendants were wrongfully convictedon the basis of eyewitness testimony later shown tobe mistaken. The implicit assumption here seemsto be that if jurors were appropriately skeptical,wrongful convictions based upon erroneous identi-fications would never occur. But this is not a tenableposition. It must be borne in mind that the degreeof skepticism jurors exhibit toward eyewitnesses willaffect not only the likelihood that an innocent de-fendant will be convicted but also the likelihood thata guilty defendant will be convicted. Thus, an in-crease in juror skepticism toward eyewitness testi-mony would decrease convictions of the guilty aswell as convictions of the innocent, and a degree ofskepticism that eliminated wrongful conviction onthe basis of eyewitness testimony would also elim-inate any role of eyewitnesses in the conviction ofthe guilty. In signal-detection terms, it is unfortunatebut true that except in situations involving very highsignal-to-noise ratios, one cannot eliminate falsealarms without also eliminating hits merely by shift-ing one's decision criterion. Consequently, somewrongful convictions must be accepted as the un-avoidable price for convicting guilty defendants. AsRembar (1980) puts it, "A system of justice that hasno miscarriages of justice is not a workable systemof justice" (p. 95).

To demonstrate juror overbelief in eyewit-nesses, one must show not merely that erroneousconvictions based on eyewitness testimony some-times occur but that the ratio of conviction of theinnocent to conviction of the guilty is unacceptablyhigh. However, documented cases of wrongful con-viction resulting from mistaken eyewitness testi-mony obviously represent only a small fraction of1% of the cases in which defendants were convictedat least in part on the basis of eyewitness testimony.Thus, although we cannot say what should be con-sidered an acceptable ratio of conviction of the in-nocent to conviction of the guilty, it would seem tobe difficult to argue that documented cases of wrong-ful conviction establish that the ratio is unacceptablyhigh. Consequently, the known cases of erroneousconvictions fail to demonstrate that jurors are toowilling to believe eyewitness testimony (Pachella,1981).

Our point here is not that the frequency ofwrongful conviction is acceptably low but merelythat known cases of erroneous conviction fail toestablish that the frequency is unacceptably high.Thus, our argument is not affected by the possibilitythat documented cases of wrongful conviction rep-resent only the tip of the iceberg. In the absence ofa means of estimating the number of undocumentedcases of wrongful conviction, these undocumented

cases cannot be used as evidence that erroneous con-viction on the basis of eyewitness testimony occurstoo often.

A third argument offered in favor of the overbe-lief claim is based on an experiment by Loftus (1974,1979; see also Cavoukian, Note 1). In this experi-ment, university students read a very brief summaryof evidence presented at a robbery-murder trial andvoted individually for conviction or acquittal. In theno-eyewitness condition, the trial description men-tioned only physical evidence against the defendant(e.g., money found in the defendant's room). Only18% of the subjects in this condition voted for con-viction. In the eyewitness condition, the trial de-scription mentioned the physical evidence and alsoindicated that an eyewitness had identified the de-fendant as the robber. In this condition, 72% of thestudents voted to convict. Finally, in the discredited-eyewitness condition, subjects were told about thephysical evidence and the eyewitness identification.However, they were also informed that the defenseattorney "claimed the witness had not been wearinghis glasses the day of the robbery, and since he hadvision poorer than 20/400, he could not possiblyhave seen the face of the robber from where hestood" (Loftus, 1979, p. 117). In this condition, 68%of the students voted to convict. The high percentageof subjects voting for conviction in the eyewitnesscondition and the lack of difference between theeyewitness and discredited-eyewitness conditions, itis argued, indicate that people give too much cre-dence to eyewitness testimony.

Several recent studies, however, cast doubt onthis conclusion. First, an experiment was recentlyconducted in our laboratory in which subjects readdetailed summaries (4,000-6,500 words) of a ficti-tious bank robbery trial and voted individually forconviction or acquittal (McKenna, Mellott, & Webb,Note 2). The bank teller who was robbed chose thedefendant from a lineup two days after the robberyand positively identified him during the trial. In ad-dition, the prosecution demonstrated that an amountof money closely matching that stolen from the bankwas found in the defendant's possession.

The defense consisted of the testimony of thedefendant's mother. She stated that the money foundin the defendant's possession was a loan from herso that the defendant could buy a car and that thedefendant had followed his normal routine the dayof the robbery, coming home from his night job inthe morning, going to sleep, and getting up at 5 p.m.On cross-examination, the defendant's mother ad-mitted that she could not be sure that the defendantwas at home at the time of the robbery, as his doorwas closed.

Part of our original purpose in conducting thisstudy was to examine the impact of expert psycho-

552 May 1983 • American Psychologist

logical testimony. Hence, the experiment includeda no-expert-testimony condition, in which only thetestimony described above was presented, and anexpert-testimony condition, in which the trial sum-mary included testimony of an experimental psy-chologist concerning factors (e.g., stress) that maylead to inaccurate eyewitness identifications.

In spite of the teller's positive identification ofthe defendant, guilty verdicts were obtained fromonly 2 of the 24 subjects (8%) in the no-expert-tes-timony condition and 3 of the 48 subjects (6%) inthe condition involving expert testimony. Exami-nation of the subjects' explanations for their verdictsrevealed that many, including those in the no-expert-testimony condition, felt that although the defen-dant may well have been guilty, it was possible thatthe teller had made an erroneous identification.Consequently, they were not certain enough of thedefendant's guilt to vote for conviction. In subse-quent experiments, we have replicated these resultsusing adults from the Baltimore community as sub-jects and with a trial scenario in which the defensecase in a robbery-murder trial consisted solely ofthe defendant's testimony that he was at home aloneat the time of the crime.

Similar findings have been obtained by Hosch,Beck, and Mclntyre (1980). In their study, subjectsserving in eight six-person juries viewed a trial inwhich an eyewitness positively identified the defen-dant. Four of the juries heard expert psychologicaltestimony; the other four juries did not. After de-liberating, all eight juries voted unanimously foracquittal. These results are somewhat difficult toreconcile with Loftus's claim that "jurors rarely re-gard eyewitness testimony with any skepticism."

Other studies have examined the claim that ju-rors will believe even a discredited eyewitness. In anexperiment in which subjects made individual guilty/not guilty decisions after reading a detailed sum-mary of a robbery-murder trial, we found that thesubjects disregarded a prosecution eyewitness whowas convincingly discredited (McCloskey, Egeth,Webb, Washburn, & McKenna, Note 3). Hatvanyand Strack (1980) and Weinberg and Baron (1982)have obtained similar results.

Thus, studies using methodologies similar tothat of the Loftus (1974, 1979) experiment haveshown that (a) a high percentage of subjects do notroutinely vote guilty when an eyewitness has posi-tively identified the defendant and (b) when a witnessis convincingly discredited, his or her testimony isdisregarded. Although definite conclusions aboutthe behavior of jurors in actual trials cannot readilybe drawn from these studies, the results clearly sug-gest that the Loftus study should not be taken asstrong support for the juror overbelief argument.3

A final argument in favor of the claim that ju-

rors overbelieve eyewitnesses stems from a recentseries of experiments by Wells, Lindsay, and theircolleagues (Lindsay et al, 1981; Wells et al., 1980).In these experiments, subjects serving as witnessesviewed a staged crime and then attempted to identifythe criminal from an array of photographs. Wit-nesses who made accurate identifications as well aswitnesses who identified the wrong person were thenvideotaped as they answered questions about theviewing conditions, the appearance of the criminal,and so forth. Additional subjects serving as jurorswatched videotapes of witnesses and judged for eachwhether the witness had made an accurate identi-fication. Under some witnessing conditions, the per-centage of jurors believing a witness was higher thanthe percentage of witnesses who made accurate iden-tifications. For example, in one situation 50% of thewitnesses made an accurate identification. However,jurors viewing videotapes believed witnesses fromthat condition 66% of the time. Lindsay et al. (1981)and Wells et al. (1980) argue on the basis of theseresults that jurors are too willing to believe eyewit-ness testimony.

Although the Wells and Lindsay argumentseems plausible, it is not entirely valid. The logic ofthe argument appears to be as follows: The findingthat juror belief rates exceed witness accuracy ratesimplies that jurors overestimate the probability thatan eyewitness is accurate, and this in turn impliesthat jurors are too willing to believe eyewitnesses.

There are some difficulties with the first stepin this argument, because the finding that the per-centage of jurors believing a witness was higher thanthe percentage of witnesses who were accurate doesnot necessarily imply that the jurors overestimatedthe probability that the witness was accurate.4 How-ever, even if we ignore this problem and assume thatin some situations jurors overestimate the proba-bility that an eyewitness is accurate, the conclusionthat jurors are too willing to believe eyewitnesses

3 One question of interest at this point is, What are thereasons for the differences in results between the Loftus experi-ment and later studies? Results we have obtained (McCloskey etal., Note 3) suggest that Loftus obtained high percentages of guiltyverdicts whereas subjects in our study and in that of Hosch etal. (1980) rarely voted guilty because subjects in the latter twostudies, but not in the Loftus experiment, received judges' in-structions on the beyond a reasonable doubt criterion for votingguilty. However, the reasons for the discrepancies between theLoftus study and other experiments in regard to the effects ofdiscrediting manipulations are not clear (see Weinberg & Baron,1982).

4 A simple example serves to make this point. Consider asituation in which 90% of witnesses make an accurate identifi-cation. If jurors accurately estimate the probability that a witnesswas accurate at .9, all jurors will probably make believe decisions,and the juror belief rate (100%) will exceed the witness accuracyrate (90%).


does not follow. As we discussed earlier, to say thatjurors are too willing to believe eyewitnesses meansthat jurors are too willing to convict on the basis ofeyewitness testimony (or, more technically, that theweight given by jurors to eyewitness testimony re-sults in an unacceptably high ratio of number ofinnocent defendants convicted to number of guiltydefendants convicted).

Although we would expect the likelihood ofconviction to increase monotonically with jurors'degree of belief in a prosecution eyewitness, datasuggesting that jurors overestimate the probabilitythat eyewitnesses are accurate do not necessarily im-ply that jurors are overly willing to convict on thebasis of eyewitness testimony. The reasonable doubtcriterion, among other things, intervenes betweenjudging the likelihood that a witness is accurate andvoting to convict or to acquit. In our research, wehave frequently seen subjects, who read trial sum-maries and arrived at verdicts, vote not guilty inspite of a stated belief that an eyewitness who iden-tified the defendant was probably correct. These sub-jects generally say that although they believe thedefendant is probably guilty, they are not certainbeyond a reasonable doubt. Thus, if the criterion forconvict decisions is sufficiently stringent, juror over-estimates of witness accuracy need not result in anoverwillingness to convict on the basis of eyewitnesstestimony. More generally, our point here is that inthe absence of data concerning jurors' criteria forconvict/acquit decisions, we cannot determine fromjuror estimates of witness accuracy (or, more spe-cifically, from the believe/disbelieve judgments col-lected by Wells et al. and Lindsay et al.) whetherjurors are insufficiently likely, just likely enough, ortoo likely to convict on the basis of eyewitness tes-timony.

One other point should be made regarding theWells et al. and Lindsay et al. results. In consideringwhether jurors overbelieve eyewitnesses, we have fo-cused on the question of whether juror evaluationof eyewitness testimony results in an acceptable ratioof number of innocent defendants convicted tonumber of guilty defendants convicted. However,one may also ask whether jurors give eyewitness tes-timony appropriate weight relative to other sorts ofevidence (e.g., the defendant's fingerprints found atthe scene of the crime, money found in the defen-dant's possession, alibi evidence). If it could be de-termined that jurors give eyewitness testimony toomuch weight relative to some other types of evi-dence, it might be argued that at least in some sensejurors overbelieve eyewitness testimony. (Of course,if in this situation juror evaluation of eyewitnessesproduced appropriate conviction rates, it mightmore reasonably be concluded that the other evi-dence was underbelieved.)

Unfortunately, the question of whether jurorsgive eyewitness testimony too much weight relativeto other sorts of evidence is difficult to answer, forat least two reasons. First, there are few studies com-paring the weight given to eyewitness testimony withthat accorded other types of evidence. Second, andmost important, it is difficult to determine howmuch weight various sorts of evidence should begiven relative to each other. Thus, we have little tosay about juror evaluation of eyewitnesses relativeto other types of evidence. However, we should pointout that the Wells et al. and Lindsay et al. findingssuggesting that subject-jurors may sometimes over-estimate the probability that an eyewitness identi-fication was accurate do not imply that jurors over-value eyewitness testimony relative to other sorts ofevidence. It is certainly conceivable that studies us-ing the Wells-Lindsay paradigm, with other kindsof evidence substituted for eyewitness testimony,might show that jurors "overbelieve" (in the Wells-Lindsay sense) those other kinds of evidence as well.(Although, as we mentioned earlier, this sort of over-belief would not necessarily imply an overwilling-ness to convict.) Thus, the Wells et al. and Lindsayet al. findings demonstrate neither that jurors areoverly willing to convict on the basis of eyewitnesstestimony nor that eyewitness testimony is overval-ued relative to other sorts of evidence.

Finally, in discussing juror overbelief we shouldmention two recent studies of actual trial outcomesthat cast doubt on the claim that jurors rarely regardeyewitness testimony with any skepticism. Chen(1981) tabulated the outcomes of all criminal casesin the Los Angeles County system from July 1977through December 1978. Other factors partialedout, the ratio of convictions in cases with at leastone eyewitness identification of the defendant toconvictions in cases without identification was 1.1to 1. Similarly, Myers (1979) examined the 201criminal cases tried by jury in Marion County, In-diana, between January 1974 and June 1976. Shefound that convictions were no more likely in casesinvolving identification of the defendant by a victimor other eyewitness(es) than in cases where there wasno eyewitness identification. These sorts of resultsare, of course, not definitive. For example, the pros-ecution may have brought otherwise weaker casesto trial when an eyewitness was available than whenthere was no eyewitness. Nevertheless, the Chen andMyers findings cast doubt on the claim that jurorsrarely regard eyewitness testimony with any skep-ticism. A dramatic illustration of this point is pro-vided by the case of a man who was arrested 13times and tried 5 times in an 18-month period fora series of crimes that were later confessed to byanother man. What is noteworthy about this caseis that he was acquitted in all five trials, even though


one or more eyewitnesses testified against him ineach (Shoemaker, 1980).

In summary, the available evidence fails toshow that jurors are overly willing to believe eyewit-ness testimony. This does not mean that jurors ex-hibit an appropriate degree of skepticism towardeyewitness testimony. Our point is simply that con-trary to the claims of many psychologists and lawyers(e.g., Ellison & Buckhout, 1981; Lindsay etal., 1981;Loftus, 1974, 1979;Loftus&Monahan, 1980; Wellset al., 1980; Woocher, 1977), juror overbelief ineyewitnesses has not been demonstrated. Conse-quently, it is by no means clear that there is a needfor expert psychological testimony to make jurorsmore skeptical.

Discrimination. The discrimination rationaleasserts that regardless of whether jurors are generallyskeptical or generally credulous of eyewitness testi-mony, they cannot distinguish well between accurateand inaccurate eyewitnesses. According to this view,expert psychological testimony could improve jurordiscrimination.

The claim that jurors cannot readily discrim-inate accurate from inaccurate eyewitnesses appearsto be well founded. Cases of wrongful convictionbased on mistaken eyewitness testimony demon-strate that juror discrimination is not perfect, andthe recent studies of Wells, Lindsay, and their col-leagues suggest that jurors' ability to distinguish ac-curate and inaccurate eyewitnesses may indeed bequite poor. As we mentioned above, these research-ers conducted a series of studies in which subjectsserving as jurors judged whether witnesses to stagedcrimes made an accurate or inaccurate identificationof the perpetrator. They found that within a givencrime situation, jurors were as likely to believe in-accurate witnesses as they were to believe accuratewitnesses (Lindsay et al., 1981; Wells et al., 1979,1980).

Although these results are clearly very disturb-ing, the situation is perhaps not as bleak as it firstappears. In the Lindsay et al. (1981) study, the wit-nesses viewed a staged crime under either good,moderate, or poor viewing conditions (in the poo/-condition, for example, the criminal was visible foronly 12 seconds and wore a hat that completely cov-ered his hair, whereas in the good condition the crim-inal's head was uncovered and he was visible for 20seconds). Accurate identification of the criminal wasmade by 33%, 50%, and 74% of the witnesses in thepoor, moderate, and good viewing conditions, re-spectively. Jurors in the study (see also Wells et al.,1980) were just as likely to believe inaccurate asaccurate witnesses within a given viewing condition.(For example, inaccurate witnesses from the goodviewing condition were believed as often as accuratewitnesses from the same condition.) However, jurors

were less likely to believe witnesses, the less favorablethe viewing conditions. Specifically, juror belief rateswere 77%, 66%, and 62%, for witnesses in the good,moderate, and poor viewing conditions, respectively.These results suggest that jurors weighing eyewitnesstestimony are able to take into account at least somefactors that influence witness accuracy. Neverthe-less, it seems clear that jurors' ability to discriminateaccurate from inaccurate witnesses is far from per-fect.

Can Psychologists Provide the Needed Help?

Our conclusions above are that (a) it is by no meansclear that jurors need expert testimony to makethem appropriately skeptical but that (b) there isroom for improvement in the ability of jurors todiscriminate accurate from inaccurate witnesses. Inthis section, we consider the possible effects of expertpsychological testimony on jurors.

Three recent studies suggest that expert psy-chological testimony may serve to make jurors moreskeptical of eyewitness testimony (Hosch et al.,1980; Loftus, 1980; Wells et al., 1980). For example,Wells et al. found that expert psychological testi-mony reduced the likelihood that a subject wouldbelieve that an eyewitness to a staged crime madean accurate identification, and Loftus reported thatsubjects who read brief trial summaries were lesslikely to vote guilty when the summaries includedexpert psychological testimony than when no suchtestimony was included. It is not a straightforwardmatter to extrapolate the results of these studies tothe verdicts of real juries. However, even if we acceptthe findings at face value, we are left with the fol-lowing question: Given the absence of clear evidencethat jurors overbelieve eyewitnesses, is it really ap-propriate for psychologists to offer expert testimonythat serves to reduce jurors' overall level of belief ineyewitnesses?

If we turn now to the possible effects of expertpsychological testimony on jurors' ability to discrim-inate accurate from inaccurate eyewitnesses, we findonly one relevant study (Wells et al., 1980). Thisstudy employed the basic Wells-Lindsay paradigmdescribed above, in which subjects serving as jurorsjudged whether or not witnesses to a staged crimeaccurately identified the perpetrator. In the Wells etal. (1980) study, half of the "jurors" received expertpsychological advice before judging the credibilityof witnesses, and the remaining "jurors" receivedno expert advice. The expert testimony emphasizedtwo general points. The first was that eyewitnessidentification in criminal cases is quite differentfrom recognizing one's friends and associates, in thatresearch using staged crimes has shown that, de-pending on conditions, 15% to 85% of eyewitnessesmay choose the wrong person from a lineup. The


second major point was that there is considerableevidence to show that witness confidence may havelittle or no relationship to witness accuracy. Thepsychologist also mentioned that jurors should payattention to situational factors that might affect wit-ness accuracy.

"Jurors" in the expert-advice and no-expert-advice conditions viewed videotapes of witnessesand made believe/disbelieve judgments. The vid-eotapes were taken from the Lindsay et al. (1981)study in which witnesses observed a staged crimeunder poor, moderate, and good viewing conditions(resulting in 33%, 50%, and 74% accurate identifi-cations, respectively).

As mentioned above, the expert psychologicaltestimony reduced the jurors' overall willingness tobelieve eyewitnesses. However, the expert testimonyhad absolutely no effect on jurors' ability to dis-criminate accurate from inaccurate witnesses.

In summary, the available evidence suggeststhat there may be a rather ironic mismatch betweenthe type of help needed by jurors and the possibleeffects of expert psychological testimony. Specifi-cally, jurors clearly need help in discriminating ac-curate from inaccurate eyewitnesses but may notneed to be made more skeptical overall. Expert tes-timony, on the other hand, may serve to increasejuror skepticism but not to improve juror discrim-ination.

Of course, firm conclusions at this point wouldbe premature. This is especially true with regard tothe effects of expert psychological testimony on jurordiscrimination. As we have seen, only one study rel-evant to this issue has been conducted (Wells et al.,1980). Furthermore, the expert testimony used inthat study may not have been optimal for improvingjuror discrimination. Aside from the negative adviceto ignore witness confidence, jurors were told onlyto examine "situational factors." It is certainly con-ceivable that expert testimony that provided a de-tailed discussion of specific factors that affect witnessaccuracy would result in better juror discrimination.Nevertheless, we must conclude that at present thereis no evidence that expert psychological advice im-proves juror evaluation of eyewitness testimony.

Expert Psychological Testimony and theFactors Affecting Witness AccuracyAt this point, an advocate of the use of expert psy-chological testimony might argue as follows: "Al-though the Wells et al. (1980) study failed to showan improvement in juror discrimination of accurateand inaccurate witnesses as a result of expert advice,this failure probably reflects, as mentioned above,the vagueness of the expert's remarks. There is everyreason to believe that more specific expert testimonywould improve juror discrimination. Empirical re-

search has identified many variables that affect wit-ness accuracy in ways that are not obvious to thelay juror. Expert testimony that discusses these fac-tors in detail would, and in practice does, increasejurors' ability to distinguish accurate from inaccu-rate eyewitnesses."

The validity of this argument can be assessedonly through additional research. We suggest, how-ever, that there is less reason than might be supposedfor optimism about the effects of expert psycholog-ical testimony on juror discrimination. The claimthat detailed expert testimony would improve jurordiscrimination rests on the assumption that thereare many variables for which both of the followingare true: (a) The relationship between the variableand eyewitness accuracy is known to psychologistsas the result of empirical research, and (b) jurors donot understand how the variable is related to witnessaccuracy. However, it turns out to be surprisinglydifficult to find variables of this sort. In other words,for many (if not most) variables that have been listedas suitable topics for expert testimony, either theeffects of the variable on witness accuracy are notwell documented, or these effects are probably ob-vious to the juror. A few examples will illustrate thispoint.

For a variable such as exposure duration, thewell-documented effects are probably obvious to ju-rors. It is difficult to imagine that jurors are notaware of the fact that longer exposures lead to in-creased witness accuracy.

For retention interval, the situation is slightlymore complex. Since the time of Ebbinghaus (1885/1913), the verbal-learning literature has quite con-sistently shown that retention declines as a functionof the delay between the learning experience and thesubsequent test. For face recognition, there are fewerstudies, and the available data are not entirely con-sistent. Many studies show run-of-the-mill retentionlosses. For example, Shepherd and Ellis (1973) mea-sured recognition performance a few minutes, 6days, or 35 days after exposure. Performance de-clined from 87% correct to 71% over that period.However, others have failed to find performance de-clining over time. For example, Goldstein andChance (1971) found accuracy to be unaffected bydelay over the range 0-48 hours. Similarly, Laugh-ery, Fessler, Lenorovitz, and Yoblick (1974) foundno difference in recognition performance among thesix retention intervals they studied (4 minutes, 30minutes, 1 hour, 4 hours, 1 day, and 1 week). Finally,it is worth noting that Carr, Deffenbacher, and Leu(Note 4; see also Deffenbacher, Carr, & Leu, 1981)actually found a reminiscence effect—recognitionof faces was slightly better 2 weeks after originalviewing than 2 minutes after. This pattern of resultsdid not obtain for the other classes of stimuli that


were tested (concrete nouns, pictures of commonobjects, and pictures of landscapes).

There are two ways in which this situation canbe assessed. First, one could conclude that at presentthe effects of retention interval on face recognitionare not sufficiently well understood to be discussedin expert testimony. Alternatively, one could arguethat the available evidence on memory in generaloverwhelmingly supports the generalization that re-tention declines with delay between acquisition andtest. According to this view, the face recognitionstudies showing no effect of retention interval, orreminiscence, would be said to fail to reflect the truestate of affairs. If this latter position is taken, how-ever, it follows that the true effects of retention in-terval on face recognition probably match jurors'beliefs about these effects and consequently that ex-pert testimony about retention interval may be un-necessary.

Consider now the cross-racial identification ef-fect. Several studies have shown that cross-racialidentification (e.g., white witness-black defendant)is more difficult than within-racial identification(e.g., black witness-black defendant; Malpass &Kravitz, 1969). This result is often discussed (e.g.,Loftus, 1979) as if it were not obvious to the layjuror. However, the claim that jurors are unawareof the difficulty of cross-racial identification is ques-tionable at best. For example, the cliche "they alllook alike to me" used in reference to members ofanother race suggests that there may be a generalawareness of the difficulty of cross-racial identifi-cation. In fact, the Devlin (1976) report describesstudies of cross-racial identification as "support forwhat is widely accepted on the basis of commonintuition" (p. 73).

Loftus (1979; see also Deffenbacher & Loftus,1982; Yarmey & Jones, 1983), in a study of beliefsabout factors affecting witness accuracy, found thatonly 55% of the subjects correctly answered a four-alternative multiple-choice question concerningcross-racial identification. However, this result prob-ably should not be taken too seriously as evidencethat people do not understand the difficulty of cross-racial identification, because Loftus's question wasextremely complex and difficult to understand:

Two women are walking to school one morning, oneof them an Asian and the other white. Suddenly, two men,one black and one white, jump into their path and attemptto grab their purses. Later, the women are shown photo-graphs of known purse snatchers in the area. Which state-ment best describes your view of the women's ability toidentify the purse snatchers?(a) Both the Asian and the white woman will find the

white man harder to identify than the black man.(b) The white woman will find the black man more dif-

ficult to identify than the white man.

(c) The Asian woman will have an easier time than thewhite woman making an accurate identification ofboth men.

(d) The white woman will find the black man easier toidentify than the white man. (p. 172)

In a similar study, Deffenbacher and Loftus(1982) asked subjects their opinion of the cliche"they all look alike to me." The answer Deffenbacherand Loftus deemed correct, "it is true," was chosenby only about 20% of the subjects. However, thisresult should not be taken to indicate that peopleare unaware of the difficulty of cross-racial identi-fication because "they all look alike to me" is ob-viously a gross overstatement. Thus, many peoplemay judge this statement to be technically false eventhough they feel that there is a kernel of truth in it.

At present, then, it is by no means clear thatjurors are unaware of the difficulty of cross-racialidentification.

Finally, for several variables that purportedlyhave nonobvious effects on accuracy and conse-quently are cited as appropriate topics for experttestimony, there is in fact little empirical evidenceabout how (or even if) these variables affect eyewit-ness performance. Consider, for example, weaponfocus, which is the alleged tendency of a personthreatened with a weapon to focus on the weaponand consequently pay little attention to the appear-ance of his or her assailant. Although weapon focusis frequently cited as an important factor in assessingeyewitness accuracy and has been discussed in ex-pert testimony (see Loftus, 1979, pp. 223-224),there is virtually no evidence that the phenomenonactually occurs. The single unpublished experimentcited as a demonstration of weapon focus (Johnson& Scott, Note 5) is "suggestive. . . but it is far fromconclusive" (Loftus, 1979, p. 36). As another ex-ample, it is widely claimed (e.g., Loftus, 1979;Woocher, 1977) that stress or arousal experiencedby a witness during an event has detrimental effectson the accuracy of the witness's testimony, and thisclaim is frequently a prominent feature of experttestimony. Unfortunately, there is little basis for thisclaim. Deffenbacher (1982), in a review of researchconcerning the effects of arousal on the reliabilityof eyewitness testimony, lists 19 relevant studies. Hereports that 10 of these studies found decreases ineyewitness accuracy with increased arousal, whereasthe remaining 9 found that increases in arousal im-proved eyewitness performance or had no effect.

Deffenbacher claims that these seemingly dis-parate results conform to the Yerkes-Dodson law,which states that the function relating stress orarousal to performance is an inverted U, such thatperformance is poorer at very high or very low levelsof arousal than at intermediate levels. This claim isunwarranted, however, because Deffenbacher fits the


data to the Yerkes-Dodson function simply by as-suming that studies showing that performance in-creased with arousal involved levels of arousal belowthe optimal arousal level, whereas studies showingimpairment of performance with increasing arousalinvolved arousal levels above the optimal level. Def-fenbacher also claims that actual crime situationsusually involve arousal levels higher than the opti-mal level and so concludes that eyewitness perfor-mance in these situations is adversely affected.Again, however, the claim that stress in crime situ-ations is above the optimal level is merely an as-sumption. Deffenbacher's claims about arousal andeyewitness performance may well be correct, but atpresent there is little empirical basis for these claims.

There may be variables that have well-docu-mented effects that are not obvious to jurors. Biasesin identification procedures provide one possibleexample. However, it is by no means the case thatthere are a large number of variables with well-doc-umented nonobvious effects. Thus, the argumentthat expert psychological testimony could almostcertainly improve juror discrimination does not ap-pear to be well founded. Testimony that asserts asfact effects that have not been demonstrated (e.g.,effects of stress or weapon focus) is clearly inappro-priate (and in any event there is no reason to believethat the introduction of undocumented assertionswould improve juror performance); testimony lim-ited to documented phenomena, however, may telljurors little that they don't already know.5

It might be argued that expert testimony aboutobvious variables such as exposure duration, light-ing, retention interval, and so forth could be bene-ficial even if jurors understand the effects of thesevariables, because jurors might not spontaneouslythink about such variables when evaluating a wit-ness's testimony. However, this argument ignores thefact that the defense attorney in a case involving aneyewitness identification will (in opening and closingstatements and in the examination of witnesses) cer-tainly call to the jury's attention any factors (e.g.,poor lighting) suggesting that the identification maybe inaccurate. Similarly, the prosecutor will pointout factors (e.g., long exposure duration) suggestingthat the identification is accurate. Thus, expert psy-chological testimony does not appear to be neededto call the jury's attention to obvious variables.

The Risks of Premature InterventionWe have argued that the available evidence fails todemonstrate that expert psychological testimonywill routinely improve jurors' ability to evaluateeyewitness testimony. However, neither do the datarule out the possibility that expert testimony couldhave beneficial effects.

Clearly, what is needed is additional researchconcerning eyewitness testimony, juror evaluationof eyewitness testimony, and the effects on jurors ofvarious sorts of expert psychological testimony.6 Ifthis research establishes that jurors are too willingto convict on the basis of eyewitness testimony orthat jurors give disproportionate weight to eyewit-nesses relative to other sorts of evidence, then expertpsychological testimony should be considered as oneof several possible methods of improving juror per-formance. Similarly, if research demonstrates thatexpert testimony can improve jurors' ability to dis-criminate accurate from inaccurate witnesses with-out producing undesirable side effects, such testi-mony clearly should be employed in the courtroom.

In the meantime, however, what should a psy-chologist do when asked to testify? When we havediscussed our misgivings about expert psychologicaltestimony with our colleagues, the reaction has oftenbeen something like this: "Well, maybe you're rightwhen you say that it hasn't been demonstrated thatexpert psychological testimony helps the jury. How-ever, it might help, and at least it can't hurt, so whynot use it?"

We strongly disagree with this argument forseveral reasons. First, contrary to the claim that "atleast it can't hurt," the possibility that psychologicaltestimony has detrimental effects cannot be ruledout. For example, if jurors are already appropriatelyskeptical about eyewitness testimony, expert psy-chological testimony might make jurors too skep-tical.7 In addition, in discussing phenomena that areincompletely understood, an expert might give

5 It is worth pointing out here that even if there were a largenumber of variables with well-documented nonobvious effects,expert psychological testimony concerning these variables wouldnot necessarily improve juror evaluation of eyewitness testimony.The information provided by the psychologist would be proba-bilistic in nature (Loftus & Monahan, 1980; Pachella, 1981) andwould not provide the jury with a basis for deciding that aneyewitness was definitely accurate or definitely inaccurate. Fur-thermore, the expert testimony generally would not give the juryany basis for deciding how large an effect a particular factor wouldhave in the situation at hand. Thus, the expert psychological tes-timony might be of little benefit to the jurors and could evenhave detrimental effects if, for example, jurors grossly overesti-mated the importance of one or more of the factors discussedby the psychologist.

6 It is beyond the scope of this article to discuss what formfuture research should take. However, we note that due to a lackof external validity, many of the studies we have discussed maybe limited in the extent to which their results can be generalizedto actual situations. We have chosen not to dwell on problemsof external validity in our discussion of previous research, butthe external validity issue should be given careful considerationin future work (see, e.g., KoneCni & Ebbesen, 1979).

7 It is worth mentioning here that the expert testimony inthe Wells et al. (1980) study apparently caused subject-jurors inat least some conditions to underestimate the probability that aneyewitness had made an accurate identification. In a conditionyielding 50% witness accuracy, subject-jurors who heard expert


groundless information to the jury (see, e.g., dis-cussion of stress and weapon focus above). Thus, itis even conceivable that expert testimony could de-crease the ability of jurors to discriminate accuratefrom inaccurate witnesses.

A second difficulty with the argument that ex-pert psychological testimony should be used becauseit might help is that a trial judge would not admitexpert testimony on this basis. For expert testimonyto be admitted, stronger arguments in its favor wouldhave to be offered. Thus, a psychologist who decidedto testify on the basis of the "it might help-at leastit can't hurt" argument would in some sense haveto misrepresent his or her testimony to the court.

Finally, the use of expert psychological testi-mony in the absence of clear evidence that it benefitsthe jury carries risks to the psychological professionas a whole. Because the effects of expert testimonyare currently unclear, it is inevitable that psychol-ogists will disagree with one another about its use.This disagreement is likely to lead ultimately to a"battle of experts" in at least some cases where at-tempts are made to introduce expert psychologicaltestimony. The battle of experts could take severaldifferent forms. For example, when the defense at-tempted to introduce expert .psychological testi-mony, the prosecution could use its own expert toargue that the defense psychologist should not beallowed to testify in front of the jury. The prose-cution expert could state that many of the assertionsthat would be made by the defense expert (e.g., abouteffects of stress or weapon focus) are not firmly es-tablished by psychological research and that the ex-pert testimony could conceivably have detrimentaleffects on the jurors' evaluation of eyewitness testi-mony. If the judge nevertheless decided to admit thetestimony of the defense expert, the prosecutionpsychologist could also testify in front of the jury,arguing that support was lacking for many of thedefense psychologist's claims. The prosecution ex-pert might also attempt to counter the defense ex-pert's arguments about the unreliability of eyewit-ness testimony by pointing out factors in the caseat hand that would facilitate eyewitness performance(e.g., long exposure duration, good lighting).

A battle of experts could also take more subtleforms. For example, a psychologist could simply

testimony believed witnesses only about 32% of the time. Sim-ilarly, in a condition producing 74% witness accuracy, the jurorbelief rate was only about 53%. If we assume that jurors makebelieve decisions when they estimate the probability of witnessaccuracy to be greater than .5, these results suggest substantialunderestimation of the probability that a witness was accurate.Of course, such underestimation does not necessarily imply un-derwillingness to convict. Nevertheless, as the Wells et al. resultssuggest, the possibility that expert psychological testimony maymake jurors too skeptical deserves careful consideration.

8 People v. Guzman, supra, p. 72, emphasis added.

serve as an advisor to the prosecutor, helping himor her to prepare an effective cross-examination ofthe defense expert. In any event, regardless of theform taken by a battle of experts, courtroom con-frontation between defense and prosecution psy-chologists would almost certainly work to the de-triment of the psychological profession, creating (orsustaining) the impression of psychology as a sub-jective, unscientific discipline and of the psychologistas a "gun for hire." The current situation with regardto psychiatric and psychological expert testimonyconcerning insanity, dangerousness, and the likeshould serve as food for thought to experimentalpsychologists.

It may have occurred to the reader that court-room battles involving members of other professions(e.g., chemists, engineers, physicians) occur quitefrequently but that these professions do not seemto have suffered serious loss of credibility. Perhaps,then, there is little reason to be concerned about theconsequences of battles involving experimental psy-chologists.

Unfortunately, the comparison to professionslike medicine, chemistry, and the like may be mis-leading. We strongly suspect that courtroom battlesin which opposing experts positively assert contra-dictory propositions do decrease public respect fora profession. Nevertheless, professions such as med-icine and physics may escape serious loss of credi-bility because of clear records of past accomplish-ments and frequent reports of new achievements.Unfortunately, psychology probably does not havethe sort of strong public reputation needed to endurebattles of experts without significant damage. Thismay be especially true of the reputation of psy-chology among members of the legal profession.Consider, for example, the following excerpt froman appeals court decision8 in which it was ruled thata trial judge had not erred in excluding expert psy-chological testimony about eyewitnesses:How far should the trial judge go in allowing so-calledscientific testimony, such as that of polygraph operators,hypnotists, "truth drug" administrants, as well as purve-yors of general psychological theories, to substitute for thecommon sense of the jury? Surely the answer is "not inall cases, or even in the ordinary or usual case."

Apart from the possible effects of a battle ofexperts on the reputation of the field, there is alsoa question of what the effects of such a battle mightbe on the outcome of a trial. A battle of expertscould conceivably improve juror decision making.On the other hand, both experts might end up beingignored. As with many of the other issues discussedin this article, this one probably should be decidedempirically rather than by guesswork.

To illustrate our points about the possible con-sequences of a battle of experts, we conclude with


an example of one of the milder forms that such abattle might take. Specifically, we present a hypo-thetical cross-examination of a defense psychologistby a prosecutor who has been thoroughly briefed byhis or her own expert. We ask the reader to considerwhether this sort of occurrence would benefit eitherthe psychological profession or the justice system.

Prosecutor: Are you suggesting that it is im-possible for an eyewitness to accurately identify acriminal?

Psychologist: No, but accurate identification isquite difficult.

Prosecutor: In the studies conducted by your-self and your colleagues, do any of the participantsdo very well at identifying people?

Psychologist: Yes, some subjects do quite well,but others do very poorly.

Prosecutor: Dr. Smith, are you aware of studiesshowing that subjects made accurate identificationsover 90% of the time?

Psychologist: Yes, there are such studies. Usu-ally, however, the performance of witnesses is worsethan that.

Prosecutor: Isn't it true that the accuracy ratesin identification studies depend heavily on the con-ditions of the experiment, such as how many faceseach subject sees, how long each face is seen, andso forth?

Psychologist: Yes, that is obviously true of anyexperiment.

Prosecutor: Isn't it also the case that conditionsin most experiments are deliberately arranged sothat accuracy is low?

Psychologist: Well, yes, in a way that's true. Ifnone of the subjects make any errors we don't learnanything from an experiment. For example, if wewanted to see whether poor lighting makes identi-fication harder, we would do an experiment wheresubjects see people under good and poor lightingconditions. We would then look to see whether ac-curacy was lower with poor lighting. If the task weretoo easy, most or all of the subjects might makeaccurate identifications and we wouldn't learn any-thing about the effects of lighting. To find outwhether lighting is important, we have to have asituation in which subjects make some errors. So toensure that errors occur, we might let the subjectsview people for only a short period of time, fromsome distance away, and so forth.

Prosecutor: Isn't it true, Dr. Smith, that evenin tests giving low overall accuracy, some witnessesidentify the right person?

Psychologist: Yes.Prosecutor: Is there any way you can tell be-

forehand which witnesses will make an accurateidentification and which witnesses will be inaccu-rate?

Psychologist: No, at the present time we haveno good way of telling in a particular situation whichwitnesses will be accurate and which ones will beinaccurate. All we can say is that certain conditionsyield lower accuracy than others.

Prosecutor: You have testified, Dr. Smith, thatin your psychological tests, the accuracy of witnessesvaries widely according to the conditions of the test.You have also stated that in conducting tests, psy-chologists deliberately create situations that producelow accuracy. Finally, you have said that there is noway you can tell whether a particular individual willmake an accurate or inaccurate identification in aparticular situation. How, then, can your tests beapplied to the present case, in which a bank tellerlooked at a single bank robber for a much longertime than in most research studies? How can theresults of experiments be used to suggest that thewitness is inaccurate in his identification?

Psychologist: I cannot comment on the accu-racy of any particular witness. All I can do is explainwhat sorts of conditions lead to a good or pooreyewitness performance.

Prosecutor: I am very interested in what youhave called "weapon focus," Dr. Smith. Can you tellus about some of the experiments that demonstratethis effect?

Psychologist: Well, there is one experiment thatgives some support for the weapon focus idea, butit isn't really conclusive. Actually . . .

Prosecutor: I am very surprised, Dr. Smith, thatyou are willing to testify about a theory for whichthere is no experimental evidence.

Psychologist: Well, the idea of weapon focuswas developed not so much from experiments butbecause people threatened with weapons often re-port having seen the weapon very clearly and oftencan describe it in great detail.

Prosecutor: That's very interesting. Are yousaying that people who are able to give a clear de-scription of, say, a gun, and who claim to have seenit clearly, would be said to have focused on the gun?

Psychologist: Yes.Prosecutor: Mr. Robinson, the eyewitness in

this case, has testified that he saw the robber's faceclearly, and he gave a clear description of the robber.Would you say, then, Dr. Smith, that Mr. Robinsonmust have focused on the robber's face?

Psychologist: Well . . .Prosecutor: I would like to ask you about your

testimony concerning cross-racial identifications.Are you saying that a white person could never iden-tify a black person?

Psychologist: No. I merely said that it is moredifficult for a white person to identify a black personthan a white person.


Prosecutor: Are you suggesting that most cross-racial identifications are wrong?

Psychologist: No, I did not say that.Prosecutor: Well, then, are cross-racial identi-

fications often incorrect?Psychologist: I cannot say exactly how often

cross-racial identifications are erroneous, only thatcross-racial identifications are less likely to be cor-rect than within-racial identifications,

Prosecutor: How much less likely?Psychologist: It is difficult for me to answer that

question without having the studies in front of me.However, I can say that several studies have foundthat cross-racial identifications were significantly lesslikely to be correct than within-racial identifications.

Prosecutor: Does that mean that there mightbe a difference of about 80% in the accuracy ofwithin- versus cross-racial identifications?

Psychologist: No, the difference in accuracy isnot that large.

Prosecutor: Would 50% be a more reasonablefigure?

Psychologist: No, the difference is somewhatsmaller than that.

Prosecutor: Well, Dr. Smith, can you estimatefor the jury just how much less likely a cross-racialidentification is to be correct than a within-racialidentification?

Psychologist: I can't be sure of the exact figures,but I believe that most studies show about a 10%difference in accuracy between within- and cross-racial identifications.

Prosecutor: That's very interesting. I'm sur-prised that such a small difference could be consid-ered significant. Would every subject show this ef-fect? In other words, would everyone be very slightlyless likely to correctly identify a person of anotherrace?

Psychologist: Not necessarily. In most studies,the results are not exactly the same for every subject.The 10% difference between within- and cross-racialidentifications would represent the average perfor-mance of a group of subjects.

Prosecutor: So it is probably the case that somepeople in these studies did just as well at cross-racialidentification as within-racial identification?

Psychologist: That's possible.Prosecutor: So you can't say for any individual

that you haven't tested whether that individual isless likely to be correct in a cross-racial or a within-racial identification?

Psychologist: All I can say is that, in general,cross-racial identifications are more difficult thanwithin-racial identifications.

Prosecutor: I would like to ask you a few ques-tions about your testimony on the effects of stresson performance. You testified, I believe, that people

under a moderate amount of stress are better at re-membering and perceiving than people under veryhigh or very low stress.

Psychologist: That's right. As I said before, therelationship between stress and performance is ex-pressed by what is called the Yerkes-Dodson law,which is a well-known principle of psychology.

Prosecutor: How much stress, Dr. Smith, ismoderate stress? That is, what level of stress leadsto the best performance, and how much stress musta person be under before his or her performancedeteriorates?

Psychologist: That depends on the type of taskinvolved. Some tasks can be performed well undera fair amount of stress, and in other tasks the sameamount of stress would impair performance. In gen-eral, the more complex the task, the lower the levelof stress that gives the best performance.

Prosecutor: I assume that since you have tes-tified about the effects of stress on eyewitness iden-tification, psychologists must have studied this issueextensively.

Psychologist: Yes, there have been a numberof studies on stress and identification.

Prosecutor: Do all of these studies show thatpeople do poorly when they are under stress?

Psychologist: Well, many of the studies showa detrimental effect of stress.

Prosecutor: Are there also studies in which peo-ple under stress did as well as or even better thanpeople who were not under stress?

Psychologist: Yes, there are such studies, butthey generally used rather low levels of stress. Thestudies showing improved performance under stressprobably involved stress levels below the point ofoptimum stress, whereas studies finding impairedperformance probably involved stress above the op-timal level. So all of the studies fit the Yerkes-Dod-son law I described earlier.

Prosecutor: I see. So I guess you are saying thatthere is some method psychologists use to measurethe stress people experience in an experiment, andthese measurements show that stress levels werelower in studies where stress helped than in studieswhere it hurt.

Psychologist: Well, no, not exactly. No singlemeasure of stress was used in all of the studies. Butif we look at the procedures that were used, it ap-pears that studies showing improved performanceunder stress involved lower stress levels than studiesshowing impairment.

Prosecutor: Are you aware of a study by Clif-ford and Hollin [Note 6] in which people who werestressed by loud noise did more poorly at identifyingfaces than people who were not exposed to noise?

Psychologist: Yes, I know of that study.Prosecutor: So according to what you have


said, this study probably involved levels of stressabove the optimal level. Is that right?

Psychologist: Yes, that is correct.Prosecutor: Are you aware of a study by

Majcher [1974] in which people exposed to loudnoise did better at recognizing faces than people whowere not stressed?

Psychologist: Yes.Prosecutor: So you would say that this study

involved stress below the optimal level. Is that right?Psychologist: Well, yes, I guess so.Prosecutor: I must admit I am confused, Dr.

Smith. Isn't it true that the stress-inducing noise wasactually louder in the Majcher study, which you saidinvolved below-optimal stress levels, than in theClifford and Hollin study, which you said involvedabove-optimal stress?

Psychologist: Well, yes, I believe that's right.Prosecutor: Do you know some other details

of these studies that lead you to believe that inducedstress was higher in the Clifford and Hollin studyeven though the noise was louder in the Majcherexperiment, or are you simply making whatever as-sumptions are needed to make these studies fit yourYerkes-Dodson law?

Psychologist: I do not know enough of the de-tails of these particular studies to comment on themfurther.

Prosecutor: Are you aware of a study by John-son and Scott [Note 5] in which people sitting in awaiting room heard a violent altercation in a nearbyroom and then saw a person carrying a bloody letteropener come out of that room into the waitingroom?

Psychologist: Yes.Prosecutor: How would you compare the stress

experienced by someone facing an apparently vio-lent person who has a bloody letter opener with thestress induced by loud noise?

Psychologist: Well, it's hard to say exactly, butthe stress would probably be greater in the situationinvolving the person with the letter opener.

Prosecutor: So you would say that the stress inthis situation was probably above the optimal level?

Psychologist: Again, it's hard to say for sure,but I would say that is likely.

Prosecutor: Isn't it true that men who experi-enced this stressful situation in the Johnson andScott experiment did better on several memory tests,including an identification test, than men who wereexposed to a nonstressful situation?

Psychologist: Well . . . yes, I believe that iscorrect. However, I believe that for female subjectsin that study, the stressful situation led to worse per-formance on some tests, including the identificationtest.

Prosecutor: Wouldn't you agree, Dr. Smith,

that the picture emerging from psychological studiesof the effects of stress on eyewitness identificationis somewhat less than crystal clear?

Psychologist: In any set of studies there arebound to be a few inconsistencies. In general, how-ever, the research shows that high stress impairseyewitness identification.

Prosecutor: Let me ask you one other thingabout stress, Dr. Smith. Would the effects of stressin a task be the same for everyone? In other words,would the level of stress at which performance be-gins to be impaired be the same for all individuals?

Psychologist: Not necessarily. It is quite pos-sible that a level of stress that impaired performancefor one person might have little effect on another.

Prosecutor: According to your testimony, thelevel of stress at which performance begins to beimpaired is different for different situations and fordifferent people. Would you agree, then, that for aparticular person in a particular situation, it wouldbe impossible to tell how much stress would be nec-essary to impair perception and memory withouttesting that person directly in that situation?

Psychologist: Yes, but as I have said, in generalhigh stress impairs performance.

Prosecutor: How much time have you spenttesting Mr. Robinson, the eyewitness in this case?

Psychologist: I have not tested him at all. I havenever even met him.

Prosecutor: Then how can you testify about theeffects of stress on his ability to identify the personwho robbed him?

Psychologist: I cannot make any judgmentabout whether Mr. Robinson as an individual is anaccurate or inaccurate witness. I can only describethe principles concerning eyewitness identificationsthat have been discovered through psychological re-search.

Prosecutor: How can these vague principles beof help to the jury, Dr. Smith, when you, with allyour knowledge and experience, cannot use themto tell whether a witness was accurate or not?

Psychologist: It is not my function to decidethat.9

9 It is worth reiterating here that the points we have madein this article are meant to apply only to expert testimony. Wedo not intend to imply that experimental psychologists have nouseful role to play in the judicial system. On the contrary, thereare probably many useful functions experimental psychologistscould serve (e.g., assisting a defense attorney in determiningwhether there are possible sources of bias in a lineup identifi-cation of a defendant). In fact, there may even be special circum-stances in which expert psychological testimony would be justi-fied. In particular, expert testimony might be warranted in sit-uations where the psychologist could assert positively that awitness could not have seen what he or she claimed to have seen.Suppose, for example, that a crucial element in a case waswhether or not an eyewitness could have noted the color of asweater worn by a defendant on a clear, moonless night (in the


REFERENCE NOTES

1. Cavoukian, A. Eyewitness testimony: The ineffectiveness ofdiscrediting information. Paper presented at the meeting ofthe American Psychological Association, Montreal, August1980.

2. McKenna, J., Mellott, A., & Webb, E. Juror evaluation ofeyewitness testimony. Paper presented at the meeting of theEastern Psychological Association, New York, April 1981.

3. McCloskey, M., Egeth, H., Webb, E., Washburn, A., &McKenna, J. Eyewitnesses, jurors and the issue of overbelief.Unpublished manuscript, Johns Hopkins University, 1981.

4. Carr, T. H., Deffenbacher, K. A., & Leu, J. R. Is there lessinterference in memory for faces? Paper presented at the meet-ing of the Psychonomic Society, Phoenix, November 1979.

5. Johnson, C., & Scott, B. Eyewitness testimony and suspectidentification as a function of arousal, sex of witness and sched-uling of interrogation. Paper presented at the meeting of theAmerican Psychological Association, Washington, D.C., Sep-tember 1976.

6. Clifford, B. R., & Hollin, C. R. Experimentally manipulatedarousal and eyewitness testimony. Unpublished manuscript,North East London Polytechnic, 1978.

REFERENCES

Addison, B. M. Expert testimony on eyewitness perception. Dick-inson Law Review, 1978, 82, 464-485.

Chen, H. T. Disposition of felony arrests: A sequential analysisof the judicial decision making process. Unpublished doctoraldissertation, University of Massachusetts, 1981.

Deffenbacher, K. A. The influence of arousal on reliability oftestimony. In B. R. Clifford & S. Lloyd-Bostock (Eds.), Eval-uating witness evidence: Recent psychological research and newperspectives. Chichester, England: Wiley, 1983.

Deffenbacher, K. A., Carr, T. H., & Leu, J. R. Memory for words,pictures and faces: Retroactive interference, forgetting and rem-iniscence. Journal of Experimental Psychology: Human Learn-ing and Memory, 1981, 7, 299-305.

Deffenbacher, K. A., & Loftus, E. F. Do jurors share a commonunderstanding concerning eyewitness behavior? Law and Hu-man Behavior, 1982, 6, 15-30.

Devlin, Hon. Lord P. (Chair). Report to the secretary of stale forthe home department of the departmental committee on evi-dence of identification in criminal cases. London: Her Majesty'sStationery Office, 1976.

Ebbinghaus, H. [Memory: A contribution to experimental psy-chology] (H. A. Ruger & C. E. Bussenues, trans.). New York:Teacher's College, Columbia University, 1913. (Originally pub-lished, 1885.)

Ellison, K. W., & Buckhout, R. Psychology and criminal justice.New York: Harper & Row, 1981.

Gass, R. S. The psychologist as expert witness: Science in thecourtroom. Maryland Law Review, 1979, 38, 539-621.

Goldstein, A. G., & Chance, J. Visual recognition memory forcomplex configurations. Perception and Psychophysics, 1971,9, 237-241.

absence of any artificial source of illumination). Given that mea-surements indicated that the illumination was far below thethreshold of photopic vision, a psychologist could testify that thewitness simply could not have correctly identified the color ofthe sweater.

Hatvany, N., & Strack, F. The impact of a discredited key witness.Journal of Applied Social Psychology, 1980, 10, 490-509.

Hosch, H. M., Beck, E. L., & Mclntyre, P. Influence of experttestimony regarding eyewitness accuracy on jury decisions.Law and Human Behavior, 1980, 4, 287-296.

Konecni, V. J., & Ebbesen, E. B. External validity of research inlegal psychology. Law and Human Behavior, 1979, 3, 39-70.

Laughery, K. R., Fessler, P. K., Lenorovitz, D. R., & Yoblick,D. A. Time delay and similarity effects in facial recognition.Journal of Applied Psychology, 1974, 59, 490-496.

Lindsay, R. C. L., Wells, G. L., & Rumpel, C. M. Can peopledetect eyewitness-identification accuracy within and across sit-uations? Journal of Applied Psychology, 1981, 66, 79-89.

Loftus, E. F. Reconstructing memory: The incredible eyewitness.Psychology Today, 1974, 8, 116-119.

Loftus, E. F. Eyewitness testimony. Cambridge, Mass.: HarvardUniversity Press, 1979.

Loftus, E. F. Impact of expert psychological testimony on theunreliability of eyewitness identification. Journal of AppliedPsychology, 1980, 65, 9-15.

Loftus, E. F., & Monahan, J. Trial by data: Psychological researchas legal evidence. American Psychologist, 1980, 35, 270-283.

Lower, J. S. Psychologists as expert witnesses. Law and PsychologyReview, 1978, 4, 127-139.

Majcher, L. L. Facial recognition as a function of arousal level,exposure duration and delay interval. Unpublished master'sthesis, University of Missouri, 1974.

Malpass, R. S., & Kravitz, J. Recognition for faces of own andother race. Journal of Personality and Social Psychology, 1969,13, 330-334.

Myers, M. A. Rule departures and making law: Juries and theirverdicts. Law and Society, 1979, 13, 781-797.

Pachella, R. G. The truth and nothing but the truth (Review ofEyewitness testimony by E. F. Loftus and The psychology ofeyewitness testimony by A. D. Yarmey). Contemporary Psy-chology, 1981, 26, 85-87.

Rembar, C. The law of the land: The evolution of our legal system.New York: Simon & Schuster, 1980.

Shepherd, J. W., & Ellis, H. O. The effect of attractiveness onrecognition memory for faces. American Journal of Psychology,1973, 86, 627-634.

Shoemaker, J. 18 months of hell for rapist's lookalike. ChicagoTribune, December 28, 1980.

Starkman, D. The use of eyewitness identification evidence incriminal trials. Criminal Law Quarterly, 1979, 21, 361-386.

Wall, P. M. Eyewitness identification in criminal cases. Spring-field, 111.: Charles C Thomas, 1965.

Weinberg, H. I., & Baron, R. S. The discredible eyewitness. Per-sonality and Social Psychology Bulletin, 1982, 8, 60-67.

Wells, G. L., Lindsay, R. C. L., & Ferguson, T. J. Accuracy, con-fidence and juror perceptions in eyewitness testimony. Journalof Applied Psychology, 1979, 64, 440-448.

Wells, G. L., Lindsay, R. C. L., & Tousignant, J. P. Effects ofexpert psychological advice on human performance in judgingthe validity of eyewitness testimony. Law and Human Behavior,1980, 4, 275-285.

Woocher, F. D. Did your eyes deceive you? Expert psychologicaltestimony on the unreliability of eyewitness identification.Stanford Law Review, 1977, 29, 960-1030.

Yarmey, A. D., & Jones, H. P. T. Is the psychology of eyewitnessidentification a matter of common sense? In B. R. Clifford& S. Lloyd-Bostock (Eds.), Evaluating witness evidence: Recentpsychological research and new perspectives. Chichester, En-gland: Wiley, 1983.


Documents

Eyewitness Identification - Semantic Scholar are unaware of the problems with eyewitness ... 2 People v , Guzman 47 Cal . App 3rd 380 121 Rptr 69 ... affect not only the likelihood