eyewitness memory enhancement in the police - researchgate

Journal of Applied Psychology1985, Vol. 70, No. 2, 401-412

Copyright 1985 by the American Psychological Association, Inc.0021-9010/85/S00.75

Eyewitness Memory Enhancement in the Police Interview:Cognitive Retrieval Mnemonics Versus Hypnosis

R. Edward Geiselman Ronald P. FisherUniversity of California, Los Angeles Florida International University

David P. MacKinnon and Heidi L. HollandUniversity of California, Los Angeles

This research compared the effectiveness of three interview procedures foroptimizing eyewitness memory performance: (a) the "cognitive interview" basedon memory-retrieval mnemonics from current memory theory, (b) the presentlycontroversial hypnosis interview, and (c) the standard (control) police interview.These methods were evaluated empirically in a controlled, yet ecologically valid,laboratory setting. Eighty-nine subjects viewed police training films of simulatedviolent crimes and were questioned individually in interactive interviews 48 hourslater by experienced law-enforcement personnel. Both the cognitive and hypnosisprocedures elicited a significantly greater number of correct items of informationfrom the subjects than did the standard interview. This result, which held evenfor the most critical facts from the films, was most pronounced for crimescenarios in which the density of events was high. The number of incorrect itemsof information generated did not differ across the three interview conditions. Theobserved memory enhancement was interpreted in terms of the memory-guidancetechniques common to both the cognitive and hypnosis interviews. Neitherdifferential questioning time nor heightened subject or interviewer motivationcould explain the results.

The Rand Corporation (1975), in a com-prehensive study of criminal-investigationprocesses, reported that the principal deter-minant of whether or not a case is solved isthe completeness and accuracy of the eyewit-ness account. Nevertheless, eyewitness reportsare known to be incomplete, unreliable, par-tially constructed (confabulated), and malle-able during the questioning procedure (Clif-ford & Hollin, 1983; Loftus, 1975, 1979;Loftus, Miller, & Burns, 1978; Wells, Fergu-son, & Lindsay, 1981). The purpose of thepresent study, therefore, was to suggest meth-ods to enhance the accuracy of eyewitness

This research was supported by a grant from theNational Institute of Justice (USDJ-83-IJ-CX-0025). Wethank the 17 law-enforcement volunteers who served asthe interviewers and Lisa Button, Diane Sindel, KimO'Reilly, Robin Cohen, and Phil Gamier, who playedintegral roles in the conduct and analysis of the experi-ment. We also express our appreciation to the LosAngeles Police Department and the Inglewood PoliceDepartment for their co-operation in this project.

Requests for reprints should be sent to R. EdwardGeiselman, Department of Psychology, University ofCalifornia, Los Angeles, California 90024.

reports and to test these methods empiricallyin a controlled, yet ecologically valid, labo-ratory setting.

Research on eyewitness memory retrievalhas produced few positive suggestions forlaw-enforcement personnel. Two notable ex-ceptions involve the ordering of the questionsto be asked during the interview and thephrasing of the questions. First, the witnessshould be asked to report the incident intheir own words before being asked any spe-cific questions (Geiselman et al., 1984; Hil-gard & Loftus, 1979; Timm, 1983). Thisprocedure reduces the possibility of the in-terviewer leading the witness, and the infor-mation given by a witness during a freereport has been found to be more accurate,although more incomplete, than informationgiven in response to specific questions. Sec-ond, to further avoid leading the witness, thespecific questions should be phrased usingindefinite articles rather than definite articles(Loftus & Zanni, 1975). A third, guided-memory technique was shown to facilitateeyewitness recognition performance in line-up procedures (Malpass & Devine, 1981);

401

402 R. GEISELMAN, R. FISHER, D. MACKINNON, AND H. HOLLAND

but with the exception of Geiselman et al.(1984), little has been done to follow up onsuch memory-enhancement techniques.

Otherwise, as noted by Clifford and Lloyd-Bostock (1983), "The work in the eyewitnessfield [has been] essentially negativistic. . . .In short, the witness [has been] shown to bea somewhat pathetic figure in the face ofextramemorial factors occurring at encoding,during storage or at retrieval" (p. 286). Yuille(1980) has proposed that considerable effortnow be focused on how we can improveeyewitness performance. Wells (1978) madea similar argument with his distinction be-tween variables that can be manipulated toreduce eyewitness fallibility (systern variables)and those that cannot be controlled in actualcrime cases (estimator variables). He con-cluded that system-variable research hasgreater potential for positive contributions tocriminal justice.

One dramatic technique for eyewitnessmemory enhancement is the hypnosis inter-view. Hypnosis has been reported to be usefulin criminal cases (Reiser, 1974, 1976; Reiser& Nielsen, 1980; Schafer & Rubio, 1978;Stratton, 1977), especially when trauma tothe witness is involved. Enhanced memoryunder hypnosis also obtains in some con-trolled laboratory experiments (DePiano &Salzberg, 1981; Griffin, 1980; Stager &Lundy,1985). On the whole, though, the evidenceabout memory under hypnosis is mixed.Many studies find no memory enhancementwith hypnosis (see M. Smith, 1983, for areview). Of greater practical consequence,hypnosis may distort the memory process. Ithas been suggested that hypnotized subjects(a) introduce fabrications into their reportsand exhibit increased error rates (Diamond,1980; Dywan & Bowers, 1984; Orne, 1979),(b) are more susceptible to leading questions(Putnam, 1979; Sanders & Simmons, 1983),and (c) are more likely to view distortedmemories as being accurate (Orne, 1961;Sheehan & Tilden, 1983). In addition, theaccuracy of information generated underhypnosis appears to be unrelated to the wit-nesses' confidence in the information (Zelig& Beidleman, 1981). The case against hyp-nosis also is equivocal, as some researchershave found hypnosis to improve memorywithout showing increased confabulation or

greater susceptibility to misleading questions(Griffin, 1980; Stager & Lundy, 1985). Fur-thermore, even nonhypnotized witnesses arehighly subject to memory alterations (Loftus,1979; Timm, 1981, 1983; Wells et al., 1981),and nonhypnotized witnesses are often inac-curate about the quality of their reports (Def-fenbacher, 1980; Wells & Lindsay, 1983).Nevertheless, as a general safeguard againstthe potential problems encountered withmemory under hypnosis, several United Statesstates have placed some restrictions on theadmissibility of hypnosis recall in a courtof law.

In light of the legal problems and thetenuous empirical support for the hypnosisinterview, we set out to develop nonhypnoticmnemonics to assist police in interviewingeyewitnesses. Over the course of the last2,000 years, persons interested in memoryenhancement have developed a variety ofmnemonics, ranging from the Greeks' use ofimagery (method of loci) to the more modernnotions of depth of processing and organiza-tion. And, whereas these mnemonics haveproven effective in many learning tasks, theyare inappropriate for police investigation be-cause they must be employed at the encodingor acquisition stage. In the typical crimescenario, the events unfold rapidly underemotionally charged conditions. Eyewitnessattention is most likely narrow in focus, andeyewitness memory is incidental. Therefore,effective memory search procedures are re-quired. Our focus, then, was to develop mne-monics that could be used to facilitate theretrieval stage.

The theoretical underpinnings that guidedour thinking are based on two generally ac-cepted principles of memory. First, the mem-ory trace is composed of several features(Bower, 1967; Underwood, 1969; Wickens,1970), and the effectiveness of a retrieval cueis related to the amount of feature overlapwith the encoded event (Flexser & Tulving,1978). Second, there may be several retrievalpaths to the encoded event, so that informa-tion not accessible with one retrieval cue maybe accessible with a different cue (Tulving,1974). Based on this theoretical framework,Geiselman et al. (1984) developed a memory-retrieval procedure for eyewitnesses calledthe cognitive interview that consists of four

EYEWITNESS MEMORY ENHANCEMENT 403

retrieval mnemonics. Of these, two attemptto increase the feature overlap between en-coding and retrieval contexts: (a) mentallyreinstating the environmental and personalcontext that existed at the time of the crime(Bower, Gilligan, & Monteiro, 1981; Malpass& Devine, 1981; S. Smith, 1979) and (b)reporting everything, even partial information,regardless of the perceived importance of theinformation (M. Smith, 1983). The other twomnemonics encourage using many retrievalpaths: (c) recounting the events in a varietyof orders (Burns, 1981; Whitten & Leonard,1981), and (d) reporting the events from avariety of perspectives (Anderson & Pichert,1978; Firstenberg, 1983).

The cognitive interview was evaluated pos-itively in a preliminary experiment conductedby Geiselman et al. (1984). In that research,actors disrupted a classroom situation andthe students were interviewed subsequentlyas eyewitnesses via a questionnaire. Studentswho were instructed in the four memoryretrieval mnemonics at the time of test re-called more correct information about theincident than did subjects who were toldsimply to keep trying to remember moreinformation. Furthermore, the cognitive in-terview showed none of the drawbacks some-times found with hypnosis: It did not lead tomore incorrect information being generated,nor did it lead to greater eyewitness confidencein the incorrect information. Based on thesepreliminary results, the cognitive interviewrepresents a promising alternative to the hyp-nosis interview.

Although the results from Geiselman et al.(1984) are encouraging, one major limitationin that study was that the conditions of theexperiment were somewhat dissimilar to thosefound in a real crime. The "crime" itself wasa low-arousal innocuous event, and the "in-terview" was an impersonal, standardized,written questionnaire. The present study wasdesigned to maximize the ecological validityof the results: The stimulus materials wereemotionally arousing films of simulatedcrimes; the eyewitness recall protocols werecollected using interactive interviews ratherthan fixed questionnaires; and the interviewswere conducted by experienced law enforce-ment personnel. The present study also ex-tended the earlier work of Geiselman et al.

by comparing the cognitive interview to thehypnosis interview and to the standard (con-trol) police interview. Each of these proce-dures is described in detail in the Methodsection below. The three types of interviewwere compared on (a) the number of correctitems of information elicited, (b) the numberof incorrect items of information elicited,and (c) the number of confabulated items ofinformation generated.

Method

Subjects

The subjects were 89 undergraduate students, 55 malesand 34 females, recruited from three introductory psy-chology classes and one psychology of learning class atthe University of California, Los Angeles. Before agreeingto participate in the study, all subjects were informedthat (a) they would be viewing a film depicting a violentcrime, (b) they would be interviewed about the contentsof the film by an experienced law enforcement profes-sional, and (c) approximately one third of them wouldbe interviewed while under hypnosis.

Subjects from the introductory psychology classes wereoffered 2 hr of credit toward completion of their experi-ment participation requirement. Subjects from the learn-ing course were offered no inducement.

Interviewers

The interviewers were recruited principally through anannouncement placed in the International Journal ofInvestigative and Forensic Hypnosis. Additional partici-pants were obtained from various police departments inSouthern California. The final group of interviewers, 16men and 1 woman, represented a variety of professionswithin the law enforcement domain: police detectives,Central Intelligence Agency investigators, polygraph spe-cialists, and private detectives. To ensure homogeneityamong the interviewers, each interviewer had completeda 40-hr course on forensic hypnosis and had subsequentfield experience on hundreds of cases. Each interviewerwas offered a $70.00 honorarium for their participation.

Each interviewer was randomly assigned to one of thethree interview conditions (cognitive = 6, hypnosis = 7,and standard = 4). The results of the interviews suggestedthat the interviewer population was homogeneous giventhat the effect of interviewer within interview conditionswas not significant (see the Results and Discussion sec-tion).

Three of the 17 interviewers had seen one or two ofthe films described below, but over two years had passedsince that exposure. The five interviews that might havebeen affected by this prior exposure produced dataconsistent with the other interviews in those interviewconditions.

Materials and Apparatus

Films. The four films used in this experiment wereborrowed from the training academy of the Los Angeles


Police Department (LAPD). The academy uses thesefilms as part of a computerized training process in whichpolice officers are exposed to simulated, life-threateningsituations (Decision Evaluation Firearms Trainer). Eachfilm presents an audiovisual scenario of a violent crimeor crime situation that lasts approximately 4 min. Thescenarios of the four films include a bank robbery, aliquor store holdup, a family dispute, and a searchthrough a warehouse. In each film, at least one individualis shot and killed. The scenarios are realistic in thatmonitored physiological reactions of officers in traininghave been found to be comparable to reactions thatwould be expected in similar street situations (LAPD).The films are rich in quantifiable information includingperson descriptions, mannerisms, weapons, and sequencesof events.

The films were projected onto a 9 ft-by-9 ft (2.7 m X2.7 m) mm projector equipped with 4-track nonopticalsound. All films were shown in the same large lecturehall.

Interview environment. The interviews were conductedat the Center for Computer-Based Behavioral Studies(CCBS) in the Department of Psychology at the Universityof California, Los Angeles. Among the facilities at CCBSare separate cubicles (approximately 6 ft-by-6 ft; 1.8 m X1.8 m) such that several interviews can be carried outsimultaneously in an undisturbed fashion.

All interviews were audio recorded on standard cassetteplayer/recorders, and the subjects wore lapel microphones.In addition to the audio recordings, subjects in thehypnosis condition were monitored using video camerasthat were mounted in every room, regardless of theinterview condition. A graduate student trained in hyp-nosis from the Clinical Psychology program at UCLAobserved the ongoing hypnosis sessions on monitors in acontrol room.

Interview Conditions

Three weeks prior to the interviews, each interviewerreceived instructions for one, and only one, of the followingthree interview procedures:

Standard interview. These interviewers were told touse the questioning procedures that they normally woulduse without a hypnotic induction procedure. The onlyrestriction was that each "witness" was to be asked firstto describe in their own words what they remembered(open-ended report). Then, and only then, were theinterviewers to ask any specific questions about the filmbased on the witnesses's account. The practice of askingfor an open-ended report first is commonly followed bymost investigators that we have interviewed, and it issupported in basic research reported by Geiselman et al.(1984), Hilgard and Loftus (1979), and Timm (1983).That is, information given during the open-ended reporttypically is more accurate.

Hypnosis interview. In accordance with the guidelinesof Orne, Soskis, Dinges, and Orne (1984) for conductingan hypnosis interview, the subjects in this condition firstwere to be asked to describe the film in their own wordsprior to any hypnosis induction. The interviewer thenwas to perform an hypnosis induction, and subsequentlyask the witness to restate what he or she rememberedfrom the film, followed in turn by any specific questions

about the film based on the witnesses's report. Onlyverbal responses were to be permitted; that is, no finger-movement responses were allowed.

To preserve ecological validity, the interviewers werefree to use whatever techniques they wanted to performthe hypnosis induction.

Cognitive interview. In this condition, the interviewerswere to describe four general memory-retrieval techniquesto the subjects before the questioning began. A four-itemlist of the techniques was placed in full view of thewitness during the entire interview as a reference guide.Otherwise, the format of this interview was the same asthat for the standard interview. The following descriptionsof the techniques were read by the interviewer to thesubject verbatim at the beginning of the interview:

(a) Reinstate the Context: Try to reinstate in yourmind the context surrounding the incident Think aboutwhat the surrounding environment looked like at thescene, such as rooms, the weather, any nearby people orobjects. Also think about how you were feeling at thetime and think about your reactions to the incident.

(b) Report Everything: Some people hold back infor-mation because they are not quite sure that the infor-mation is important. Please do not edit anything out ofyour report, even things you think may not be important.

(c) Recall the Events in Different Orders: It is naturalto go through the incident from beginning to end.However, you also should try to go through the events inreverse order. Or, try starting with the thing that impressedyou the most in the incident and then go from there,working both forward in time and backward.

(d) Change Perspectives: Try to recall the incidentfrom different perspectives that you may have had, oradopt the perspectives of others that were present duringthe incident. For example, try to place yourself in therole of a prominent character in the incident and thinkabout what he or she must have seen.

All interviewers were informed that it was preferableto err in the direction of terminating an interviewprematurely if that should become necessary. In particular,the interviewers were asked to terminate an interview ifthe subject appeared to become more anxious about theinterview as the session progressed. None of the interviewswere terminated prematurely.

Procedure

Each subject participated in two sessions. During thefirst session, groups of 8 to 12 subjects each saw one ofthe four films. The subjects were asked to refrain fromdiscussing the film among themselves. After the film, agraduate student trained in hypnosis from the Departmentof Psychology at UCLA informed all subjects aboutmisconceptions concerning hypnosis and answered anyquestions. This presentation was based on our observationsof presentations made by hypno-investigators in the fieldand on suggestions made by Reiser (1980) in his handbookon investigative hypnosis.

Approximately 48 hr after viewing the film, the subjectswere interviewed by the law-enforcement personnel. Uponarrival at this second session, the subjects were assignedrandomly to one of the three interview conditions (cog-nitive = 33, hypnosis = 30, and standard = 26). Thesubjects were given the eye-roll test for hypnotic suscep-


tibility (Spiegel, 1972), and the subjects were interviewedindividually in separate rooms. Each interviewer ques-tioned approximately five subjects during the course ofthe day, and each interviewer interviewed at least onewitness of each crime. Before each interview, the inter-viewer was told only the title of the crime scenario thathad been witnessed by the subject (e.g., bank robbery).

Analysis of Protocols

Each tape recorded interview was transcribed by twoof four different research assistants trained by the authors.The second listener filled in any information missed inthe original transcription. The transcriptions of the tapesthen were given to another member of the research teamwho categorized the information into"three exhaustivelists for each film: persons, objects, and events. Thepersons category included physical appearance, clothing,mannerisms, and speech characteristics. The objects cat-egory included guns, knives, cars, and carried articles.The events category included movements, number ofshots, interperson contacts, conversation, and generalsequencing. These exhaustive lists were compiled andmatched against the information contained in the fourfilms for accuracy. Opinionated responses, such as "thesuspect was nervous," were not scored and were deletedfrom the lists.

This catalogue of information then was used to scoreeach subject's transcribed report for (a) the number ofcorrect bits of information recalled, (b) the number ofincorrect bits of information generated (e.g., the wronghair color of a suspect), and (c) the number of confabulatedbits of information generated (e.g., a description of asuspect's face when the face was not shown in the film).In the few cases where a subject changed a responseduring the interview, only the final response was scored.This scoring was carried out by five members of theresearch team. Each person worked independently andwas randomly assigned at least one transcription fromeach interview condition and each film. To evaluate thereliability of this process across the interview conditions,a two-way analysis of variance (ANOVA) was performedon each of the three memory-performance variables withinterview condition and scorer as the factors. Althoughthe main effect of scorer was significant for both thenumber of correct items f\4, 74) = 4.59, p < .003, andthe number of incorrect items F(4, 74) = 2.48, p < .05,the Scorer X Interview Condition interaction was notsignificant in any of the three analyses (all Fs < 1.21,p > .30). Thus, although some scorers may have beenmore liberal in counting a response as correct or incorrect,these differences did not appear differentially for any oneof the three interview conditions.

Design and Statistical Analysis

A 3 (interview condition) by 4 (type of crime/film) by2 (hypnotic susceptibility: low = 1-6, high = 7-11) be-tween-subjects ANOVA design was used. There were threememoryrperformance dependent variables (number ofitems correct, number incorrect, and number confabu-lated). In addition to the three memory variables, furthermeasures of interest were questioning time (total interviewtime excluding casual conversation, the hypnosis induc-

Table 1Performance Measures for the ThreeInterview Procedures

Variable Cognitive Hypnosis Standard

No. correct (C)No. incorrectNo. confabulatedQuestion time

(QT): minC:QT

(covariate)No. questions

asked

41.157.300.70

39.70

39.46

54.90

38.005.901.00

28.20

38.77

34.82

29.406.100.40

32.10

29.56

68.90

tions, and the retrieval methods training) and the numberof questions asked. Subject gender differences in perfor-mance also were examined.

Results and Discussion

Memory-Performance Measures

Table 1 presents five performance measuresas a function of the type of interview proce-dure. Prior to the substantive statistical anal-yses of the memory-performance data, threenested, random-effects ANOVAS were con-ducted on the number correct, number in-correct, and number confabulated dependentvariables to determine whether the scoresdiffered reliably between interviewers withinthe interview conditions. The univariate Fwith 14 and 72 degrees of freedom wascomputed to be 1.55 (p > .12) for numbercorrect, 1.51 (p > .13) for number incorrect,and 1.23 (p > .27) for number confabulated.Thus, the pooled error terms (the variancebetween subjects across interviewers withineach interview condition) were used in thebetween-interview comparisons (Winer, 1962,p. 207) to maximize the potential power ofthe tests.

The main effect for the number of correctitems generated was significant, f\2, 77) =5.27, p < .008. A Tukey's posttest showedthat both the cognitive and hypnosis inter-views elicited a greater number of correctitems of information than the control inter-view (ps < .05) but that the cognitive andhypnosis interviews did not differ. The maineffect for the number of incorrect items ofinformation generated was not significant,

, 77) = 1.99, p > . 14. This result is unlikely


to be a floor effect because the average errorrate was 16%. Furthermore, as will be de-scribed shortly, the incorrect information dataare sensitive enough to show some reliabledifferences—between males and females. Insum then, the enhanced recall with the cog-nitive and hypnosis interviews reflects moreeffective memory retrieval and cannot beinterpreted as a shift in response criterion(Dywan & Bowers, 1983).1 The main effectfor the number of confabulated items alsowas not significant, F(2, 77) = 2.48, p > .09.As can be seen in Table 1, given our definitionof a confabulated item of information, fewsubjects confabulated in any of the interviewconditions.

The results for the cognitive interviewclosely replicate those obtained by Geiselmanet al. (1984), in which subjects were inter-viewed about a classroom intrusion using astructured questionnaire. In both experi-ments, a greater number of correct items ofinformation were generated with the cognitiveinterview than with the control interview,and without an increase in the number ofincorrect items. Thus, the cognitive interviewhas been shown to be useful for the enhance-ment of eyewitness memory performanceboth under conditions of experimental controland under conditions of high ecologicalvalidity.

Whereas the present study showed en-hanced memory with the hypnosis interview,this is not the most frequently reported out-come of laboratory experiments designed toevaluate the effects of hypnosis on memoryperformance. In studies where emphasis hasbeen placed on experimental control, hypnosisprocedures most often have been shown eitherto not affect memory performance or theyhave been found to lead to more incorrectinformation (Orne et al., 1984). There aremany differences between the present designand those of previous studies, and furtherresearch is required to specify the factorsresponsible for the differences in outcome.We believe the principal candidate factors tobe: the nature of the materials, the interactivenature of the interviews, and the populationof interviewers. The present equality ofperformance observed with the cognitiveand hypnosis procedures is consistent withTimm's (1983) speculation that any memory-

enhancement effects of the hypnosis interviewlie in its memory-guidance components.

Examination of the interview recordingsproduced no new insight on the issue ofwhether witnesses are more susceptible toleading questions under hypnosis. This isbecause virtually no leading questions (ques-tions containing "given" information thatwas not provided by the witness) were askedby the present interviewers in any of theconditions. Even though most interviewersquestioned more than one witness from atleast one of the crime scenarios, only onequestion in the 89 interviews was identifiedas clearly leading the witness. Given that, toour knowledge, the present study is the firstto record and analyze the interviews of ex-perienced law-enforcement investigators, thisoutcome itself is an important normativeresult. In contrast, Yuille (1984) reported theresults of a survey in which a significantpercentage of Canadian police personnelagreed that "direct (often leading) questionsmust be asked so that the witness is remindedof relevant facts" (p. 20). It is possible, there-fore, that the present population of interview-ers exhibited exceptional interviewing skillsand/or that the college subjects who servedas the eyewitnesses showed atypical memory

1 The reader may note that the percentage of correctresponses (number correct/number correct •+• number in-correct + number confabulated) is approximately equalacross the interview conditions: .84 for cognitive, .85 forhypnosis, .82 for standard). This equivalence does notimply that memory is equally good because the threegroups differed in terms of the total number of responses:The cognitive (49.5) and the hypnosis (44.9) groups mademore responses than did the standard (35.9) group. Ithas been shown in other recall studies (Roediger, 1973)that there is a tradeoff between accuracy and number ofresponses made: As the subject makes more responses,the percentage of incorrect responses (intrusions) increases.Thus, one must take into account both accuracy andextent of recall to measure memory properly. In thepresent study, the cognitive and hypnosis interviewselicited more responses without a drop in accuracy rate,testifying to their superiority over the standard interview.Without taking into account both measures (accuracyand number of responses), one might be led to theconclusion that an interview that elicited only one, butcorrect, response (100% accuracy) is as effective as onethat elicits one hundred, all correct, responses (100%accuracy). The problem is similar to the speed-accuracytradeoff: One must examine both speed and accuracysimultaneously. Equivalent accuracy rates reflect equalperformance only if speed is constant across conditions.


Table 2Performance Measures for the Three Interview Procedures as a Function of Crime Scenario

Bank robbery

Variable

No. correctNo. incorrectNo. confabulated

C

40.47.80.7

H

35.45.90.4

S

28.77.00.7

Family dispute

C

40.66.80.6

H

35.55.90.8

S

33.36.80.2

Liquor store

C

57.38.00.6

H

50.88.61.6

S

29.93.10.7

Warehouse

C

24.35.50.9

H

28.93.41.0

S

21.75.50.0

Note. C = cognitive, H = hypnosis, S = standard.

performance. In either case, leading questionswould not have been necessary to generatereasonably complete reports. There is someindication from a recent replication experi-ment in our laboratory (Geiselman, Fisher,MacKinnon, & Holland, 1985) that a greaternumber of leading questions are asked whennonstudents serve as the witnesses (20 leadingquestions in 53 interviews). Nevertheless, eventhis incidence can be considered to be infre-quent. A final possibility is that the interview-ers were exceedingly careful in conductingthe interviews because they were aware ofbeing observed. Although this possibilitywould be difficult to test empirically, it doesnot appeal to the authors because such con-servatism would have suppressed differencesbetween the interview conditions. There wereno obvious indications that the interviewswere stilted.

Questioning Time

Table 1 also presents the average total timethat the interviewers spent questioning thewitnesses. As noted in the Method section,these times exclude any intervals spent incasual conversation or in the hypnosis induc-tions or cognitive retrieval methods training.The main effect for questioning time wassignificant, F(2, 77) = 5.49, p < .006. A Tu-key's posttest showed that the interviewerswho used the cognitive procedure spent moretime asking questions than did interviewerswho used either the hypnosis or standardprocedures, which did not differ. Perhaps thesuperior recall in the cognitive over the stan-dard interview occurred because of the greatertime spent in questioning the witnesses. Be-cause of this possibility, the number-correctdata were reanalyzed with questioning time

as a covariate. The adjusted means from thisanalysis are presented in the bottom row ofTable 1. As before, the main effect was sig-nificant, F(2, 76) = 4.65, p < .02; and a Tu-key's posttest showed that both the cognitiveand hypnosis interviews led to more cor-rect information than the standard interview(ps < .05).

Number of Questions Asked

The average number of questions asked ineach interview condition is presented in thebottom row of Table 1. The main effect fortype of interview was significant, f\2, 77) =5.87, p < .006, with significantly fewer ques-tions asked in both the cognitive and hypnosisconditions than in the standard condition(p < .05 and p < .01, respectively). Thus, thememory enhancement achieved with the cog-nitive and hypnosis procedures cannot beexplained in terms of the interviewers askingmore questions. To the contrary, the cognitiveand hypnosis techniques were more efficient(0.75 and 1.09 items correct per question,respectively, versus 0.42 items correct perquestion in the standard condition).

Type of Crime Scenario

The three memory-performance measuresare presented in Table 2 as a function of thetype of interview and crime scenario (film).The crime scenario interacted with the typeof interview only for the number of correctitems of information generated, F(3, 77) =8.80, p < .001. As can be seen in Table 2,the superiority of the cognitive and hypnosisinterviews is most evident with the bankrobbery and liquor store holdup scenarios.This result was not due to a ceiling effect on


Table 3Performance Measures for the Three Interview Procedures as a Function of Eye-Roll Score

Variable

Cognitive Hypnosis

Low High Low High

Standard

Low

Note. Low: eyeroll = 1-6, High: eyeroll = 7-11. (Eyeroll scores were not recorded for 3 subjects.)

High

No. correctNo. incorrectNo. confabulated

39.97.10.7

39.36.80.8

37.25.60.8

37.66.41.1

28.36.20.3

28.25.40.4

the number of correct items possible in theother two films because, with each film, thehighest subject score was at least three stan-dard errors above the mean for either thecognitive or hypnosis conditions.

The authors had no a priori hypothesesregarding this interaction. However, an after-the-fact examination of the four films revealedone striking difference between the two crimescenarios that showed a significant effect ofinterview method and the two that did not.In both the bank robbery and liquor storeholdup, several actions occurred simulta-neously, and the number of events per unittime was high. In contrast, in both the familydispute and warehouse search scenarios, theevents took place in a sequential fashion at arelatively slower pace. Perhaps the guidedmemory-search procedures common to boththe cognitive and hypnosis interviews weremore beneficial when processing of the to-be-remembered information was restricted atencoding by the density of events.

Hypnotic Susceptibility

Table 3 presents the three memory-perfor-mance measures as a function of the type ofinterview and level of hypnotic susceptibility(as indexed by the Spiegel eye-roll score).Neither the main effect of hypnotic suscepti-bility nor the interaction between type ofinterview and hypnotic susceptibility was sig-nificant for any of the three memory-perfor-mance measures (all Fs< 1). The lack ofsignificant differences, especially for the hyp-nosis interview, was surprising but does notaffect the above interpretation of the results.In fact, this outcome is consistent with thehypothesis that the facilitative effects of thehypnosis interview lie in the memory-guid-

ance procedures (Timm, 1983) and not inthe induction.2

Gender of Eyewitness

The gender of the eyewitness was found tobe unrelated to: (a) the number of correctitems generated, F(\, 65) < 1; (b) the numberof confabulated items generated, F(\, 65) <1; and (c) questioning time, F(l, 47) = 1.64,p > .20. The only significant difference wasfound in the number of incorrect items gen-erated, F(l, 65) = 8.50, p < .005: Males gen-erated a greater number of incorrect itemsthan females (7.12 versus 4.92). Given thatthis result was not accompanied by an in-crease in correct information, the conclusionis that the females exhibited superior memoryperformance. This difference did not interactwith other factors in the experiment, andtherefore, the conclusions drawn here aboutinterview methods hold for both male andfemale eyewitnesses.

Recall of Critical Facts

The preceding analyses of the memoryperformance data were carried out irrespectiveof the relative importance of the informationthat was generated across the interview con-ditions. Therefore, 20 facts from both thebank robbery and liquor-store holdup films,where differences in overall memory perfor-

2 It is interesting to note that some of the interviewersstated in an informal debriefing session that they disagreedwith the belief that only a small percentage of individualsare highly susceptible to hypnosis. If one inductionmethod fails with a given person, another method mightprove effective. Perhaps with their approach, any singleindex of susceptibility will not reflect the level of hypnosisultimately achieved.


Table 4Performance Measures for Twenty Critical Facts

Variable

No. correctNo. incorrect

C

11.21.3

Bank robbery

H S

11.9 8.51.5 1.8

C

12.70.9

Liquor store

H

12.82.0

S

9.91.0

Note. C = Cognitive, H = Hypnosis, S = Standard.

mance were observed, were chosen for selec-tive scoring as the most important items ofinformation from those crime scenarios. Thelists of critical facts were generated indepen-dently by five members of the research staff,and these lists were discussed and merged ina subsequent meeting of the entire group.Then, the protocols from the 46 subjects whowitnessed either of the two target films werescored for the 20 critical facts. Each protocolwas scored by one member of the staff.Neither the main effect of scorer nor theScorer X Interview Procedure interaction wassignificant for either the number corrector number incorrect memory variables(Fs < 1.03).

The average memory-performance scoresare presented in Table 4. The main effect forinterview condition was significant only forthe number of correct items of informationgenerated, F(2, 40) = 3.64, p < .04. As inthe overall analysis, both the cognitive andhypnosis procedures led to the recall of morecorrect items than did the standard interview(/TS < .05). Thus, the cognitive and hypnosisinterviews were successful in the enhancementof eyewitness memory for the most criticalfacts, not simply for ancillary facts. Theseeffects did not depend on the crime scenario,the gender of the witness, or the level of eye-roll score (all Fs < I).

Conclusions

The major finding of this study is thatboth cognitive-retrieval mnemonics and tech-niques inherent in the forensic use of hypnosisare effective for the enhancement of eyewit-ness memory retrieval in the police interview.This was observed to be true especially forcrime scenarios in which the density of eventswas high. We believe these effects to lie in

the guided memory-search components ofthe cognitive and hypnosis interviews. Bothof these procedures encourage the eyewitnessto mentally reinstate the contextual elementsthat were present at the time of the crime.In addition, the hypnosis procedure frequentlydraws upon a videotape replay analogy with"fixed-frame" and "zoom-in" capabilities(Reiser, 1980). It is plausible that this tech-nique, in effect, simulates components of theno-edit and varied retrieval perspectives mne-monics from the cognitive interview. In con-trast, the standard interview as observed hereconsists mainly of repeated attempts to recallthe target information, each time in the sameway without supplemental memory-retrievalguidance.

Further research will delineate which ofthe retrieval mnemonics are most effectivefor the recovery of specific kinds of infor-mation. At present, it is instructive to notethat the cognitive interview can be learnedand applied with little training. The inter-viewers who carried out the cognitive inter-views in the present research, for example,studied a two-page description of the cognitivemethods and participated in a 15-min dis-cussion prior to conducting the interviews.In addition to the savings in training time,the present study showed that less time isrequired on average to instruct a witness inthe general cognitive mnemonics (6.7 min)than to perform an hypnosis induction (27.1min). One further advantage of the cognitiveinterview is that it circumvents the presentlegal problems that surround the forensic useof hypnosis. However, the effectiveness of thecognitive interview relative to the hypnosisinterview in cases of severe trauma to avictim or witness remains to be evaluated.

The present study evaluated the cognitiveinterview in a more ecologically valid setting


than that employed by Geiselman et al.(1984). The stimulus materials were selectedand presented to enhance the arousal of thewitness; an interactive questioning formatwas followed; and the interviews were carriedout by experienced law-enforcement person-nel. The importance of validating laboratorydata on eyewitness phenomena under morenatural conditions has been stressed by otherauthors (Malpass & Devine, 1980; Monahan& Loftus, 1982; Reiser, 1980). The descre-pancy between the memory-enhancementqualities of forensic hypnosis observed hereand results typically obtained under moreartificial conditions underscores the impor-tance of this validation. There still are majordifferences between the present laboratorysetting and a real-world crime. For example,the element of personal involvement can neverbe achieved completely in studies of this type.However, it is interesting to note that themajority of the present subjects responded tothe questions using personal pronouns, in arole-playing manner, as if they had actuallyexperienced the crime. Another potentiallyimportant factor is the element of surprise(Murray & Wells, 1982). The present subjectsknew that they eventually would be ques-tioned about what they saw in the films.Nevertheless, the results that were obtainedhere with the cognitive interview are consis-tent with those reported by Geiselman et al.(1984) in which the subjects' memories wereincidental.

Finally, the present results are not consis-tent with an interpretation that would attrib-ute the enhancement of memory performanceto heightened subject or interviewer motiva-tion. First, with such an interpretation, per-formance in the hypnosis condition logicallyshould have exceeded that with the cognitiveinterview because all subjects volunteered forthe experiment only after being informed ofthe possibility that they would be hypnotized.Immediately before the interviews, the sub-jects in the cognitive condition, as well as thesubjects in the standard condition, were toldthat they would not be hypnotized. Averageperformance in the hypnosis condition wasnot limited by a ceiling effect on the numberof correct items possible. Second, it was ourimpression that the subjects in all conditionswere well motivated in the experiment. The

majority of the subjects in all conditions roleplayed in answering the questions. Third, theinterviewers were given a description only ofthe interview condition in which they wereto participate. Furthermore, the questioningtime was shortest of all for the hypnosisinterview where memory performance wasrelatively good, and the average number ofquestions asked was smaller in both the cog-nitive and hypnosis conditions than in thestandard condition. These results would ap-pear to contradict any interpretation wherethe quality of the interviews is hypothesizedto have been inadvertently manipulated bythe interviewers. Fourth, the superiority ofthe cognitive and hypnosis interviews wasobserved for only two of the four crimescenarios, those with rapidly and simulta-neously occurring events. If subjects weresimply more motivated in the cognitive andhypnosis groups, we could reasonably expectthe superiority to extend across all four sce-narios. Fifth, there is no evidence that mem-ory retrieval performance is improved withgreater motivation in any case (Weiner, 1966).Sixth, and most important, the effects of thecognitive and hypnosis interviews were spe-cific to the generation of correct items ofinformation. That is, the results cannot becouched in terms of a simple shift in responsecriterion resulting in greater overall produc-tivity or a willingness to please the interviewerwith embellished reports in reaction to de-mand characteristics.

ReferencesAnderson, R. C, & Pichert, J. W. (1978). Recall of

previously unrecallable information following a shiftin perspective. Journal of Verbal Learning and VerbalBehavior, 17, 1-12.

Bower, G. (1967). A multicomponent theory of thememory trace. In K. W. Spence & J. T. Spence (Eds.),The psychology of learning and motivation (Vol. 1, pp.230-325). New York: Academic Press.

Bower, G. H., Gilligan, S. C., & Monteiro, K. P. (1981).Selectivity of learning caused by affective states. Journalof Experimental Psychology: General, 110, 451-472.

Burns, M. J. (1981). The mental retracing of prioractivities: Evidence for reminiscence in ordered re-trieval. (Doctoral dissertation, University of California,Los Angeles, 1981). Dissertation Abstracts Interna-tional, 42, 2108B.

Clifford, B. R., & Hollin, C. R. (1983). The effects ofdiscussion on recall accuracy and agreement. Journalof Applied Social Psychology, 13, 234-244.

Clifford, B. R., & Lloyd-Bostock, S. M. A. (1983).


Witness Evidence: Conclusion and prospect. InS. M. A. Lloyd-Bostock & B. R. Clifford (Eds.),Evaluating witness evidence (pp. 285-290). New York:Wiley.

Deffenbacher, K. A. (1980). Eyewitness accuracy andconfidence: Can we infer anything about their relation-ship? Law and Human Behavior, 4, 243-260.

DePiano, F. A., & Salzberg, H. C. (1981). Hypnosis asan aid to recall of meaningful information presentedunder three types of arousal. International Journal ofClinical and Experimental Hypnosis, 29, 383-400.

Diamond, B. L. (1980). Inherent problems in the use ofpretrial hypnosis in a prospective witness. CaliforniaLaw Review, 68, 313-349.

Dywan, J., & Bowers, K. S. (1983). The use of hypnosisto enhance recall. Science, 222, 184-185.

Firstenberg, I. (1983). The role of retrieval variability inthe interrogation of human memory. (Doctoral disser-tation, University of California, Los Angeles, 1983).Dissertation Abstracts International, 44, 1623B.

Flexser, A., & Tulving, E. (1978). Retrieval independencein recognition and recall. Psychological Review, 85,153-171.

Geiselman, R. E., Fisher, R. P., Firstenberg, I., Hutton,L. A., Sullivan, S., Avetissian, I., & Prosk, A. (1984).Enhancement of eyewitness memory: An empiricalevaluation of the cognitive interview. Journal of PoliceScience and Administration, 12, 74-80.

Geiselman, R. E., Fisher, R. P., MacKinnon, D. P., &Holland, H. L. (1985). Eyewitness memory enhance-ment with the cognitive interview. Manuscript submit-ted for publication.

Griffin, G. R. (1980). Hypnosis: Towards a logical ap-proach in using hypnosis in law enforcement agencies.Journal of Police Science and Administration, 8, 385-389.

Hilgard, E. R., & Loftus, E. F. (1979). Effective interro-gation of the eyewitness. International Journal of Clin-ical & Experimental Hypnosis, 27, 342-357.

Loftus, E. F. (1975). Leading questions and eyewitnessreport. Cognitive Psychology, 7, 560-572.

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

Loftus, E. F., Miller, D. G., & Burns, H. G. (1978).Semantic integration of verbal information into visualmemory. Journal of Experimental Psychology: HumanLearning and Memory, 4, 19-31.

Loftus, E. F., & Zanni, G. (1975). Eyewitness testimony:The influence of the wording of a question. Bulletinof the Psychonomic Society, 5, 86-88.

Malpass, R. S., & Devine, P. G. (1980). Realism andeyewitness identification research. Law and HumanBehavior, 4, 347-358.

Malpass, R. S., & Devine, P. G. (1981). Guided memoryin eyewitness identification. Journal of Applied Psy-chology, 66, 343-350.

Monahan, J., & Loftus, E. F. (1982). The psychology oflaw. Annual Review of Psychology, 33, 441-475.

Murray, D. M., & Wells, G. L. (1982). Does knowledgethat a crime was staged affect eyewitness performance.Journal of Applied Social Psychology, 12, 42-53.

Orne, M. T. (1961). The potential uses of hypnosis ininterrogation. In A. D. Biderman & H. Zimmer (Eds.),The manipulation of human behavior (pp. 169-215).New York: Wiley.

Orne, M. T. (1979). The use and misuse of hypnosis incourt. The International Journal of Clinical and Ex-perimental Hypnosis, 27, 311-341.

Orne, M. T, Soskis, D. A., Dinges, D. F., & Orne, E. C.(1984). Hypnotically induced testimony. In G. L. Wells& E. F. Loftus (Eds.), Eyewitness testimony: Psycho-logical perspectives (pp. 171-213). Cambridge Univer-sity Press.

Putnam, W. H. (1979). Hypnosis and distortions ineyewitness memory. International Journal of Clinicaland Experimental Hypnosis, 27, 437-448.

Rand Corporation. (1975, October). The criminal inves-tigation process (Vol. 1-3). Rand Corporation TechnicalReport R-1776-DOJ, R-1777-DOJ, Santa Monica, CA.

Reiser, M. (1974). Hypnosis as an aid in homicideinvestigation. American Journal of Clinical Hypnosis,17, 84-87.

Reiser, M. (1976). Hypnosis as a tool in criminal inves-tigation. The Police Chief, 46, 39-40.

Reiser, M. (1980). Handbook of investigative hypnosis.Los Angeles: LEHI.

Reiser, M., & Nielsen, M. (1980). Investigative hypnosis:A developing specialty. American Journal of ClinicalHypnosis, 23, 75-84.

Roediger, H. L. (1973). Inhibition in recall from cuingwith recall targets. Journal of Verbal Learning andVerbal Behavior, 12, 644-657.

Sanders, G. S., & Simmons, W. L. (1983). Use ofhypnosis to enhance eyewitness accuracy: Does itwork? Journal of Applied Psychology, 68, 70-77.

Schaefer, D. W, & Rubio, R. (1978). Hypnosis to aidthe recall of witnesses. International Journal of Clinicaland Experimental Hypnosis, 26, 81-91.

Sheehan, P. W, & Tilden, J. (1983). Effects of suggestibilityand hypnosis on accurate and distorted retrieval frommemory. Journal of Experimental Psychology: Learning,Memory, & Cognition, 9, 283-293.

Smith, M. (1983). Hypnotic memory enhancement ofwitnesses: Does it work? Psychological Bulletin, 94,387-407.

Smith, S. (1979). Remembering in and out of context.Journal of Experimental Psychology: Human Learningand Memory, 5, 460-471.

Spiegel, H. (1972). An eye-roll test for hypnotizability.The American Journal of Clinical Hypnosis, 15, 25-28.

Stager, G. L., & Lundy, R. M. (1985). Hypnosis and thelearning and recall of visually presented material.International Journal of Clinical and ExperimentalHypnosis, 33, 27-39.

Stratton, J. G. (1977). The use of hypnosis in lawenforcement criminal investigations: A pilot program.Journal of Police Science and Administration, April,399-406.

Timm, H. W. (1981). The effects of forensic hypnosistechniques on eyewitness recall and recognition. Journalof Police Science and Administration, 9, 188-194.

Timm, H. W. (1983). The factors theoretically affectingthe impact of forensic hypnosis techniques on eyewit-ness recall. Journal of Police Science and Administra-tion, 11, 442-450.

Tulving, E. (1974). Cue-dependent forgetting. AmericanScientist, 62, 74-82.

Underwood, B. J. (1969). Attributes of memory. Psycho-logical Review, 76, 559-573.


Weiner, B. (1966). The effects of motivation on theavailability and retrieval of memory traces. Psycholog-ical Bulletin, 65, 24-37.

Wells, G. L. (1978). Applied eyewitness testimony re-search: System variables and estimator variables. Jour-nal of Personality and Social Psychology, 36, 1546-1557.

Wells, G. L., Ferguson, T. J., & Lindsay, C. C. L. (1981).The tractability of eyewitness confidence and its im-plications for triers of fact. Journal of Applied Psy-chology, 66, 688-696.

Wells, G. L., & Lindsay, C. C. L. (1983). How do peopleinfer accuracy? In S. M. A. Lloyd-Bostock & B. R.Clifford (Eds.), Evaluating witness memory (pp. 41-56). New York: Wiley.

Whitten, W, & Leonard, J. (1981). Directed searchthrough autobiographical memory. Memory & Cogni-tion, 9, 566-579.

Wickens, D. (1970). Encoding categories of words: Anempirical approach to meaning. Psychological Review,77, 1-15.

Winer, B. J. (1962). Statistical priniciples in experimentaldesign. New York: McGraw-Hill.

Yuille, J. C. (1980). A critical examination of the psy-chological and practical implications of eyewitnessresearch. Law and Human Behavior, 4, 335-345.

Yuille, J. C. (1984). Research and teaching with police:A Canadian example. International Review of AppliedPsychology, 33, 5-23.

Zelig, M., & Beidleman, W. B. (1981). The investigativeuse of hypnosis: A word of caution. InternationalJournal of Clinical and Experimental Hypnosis, 29,401-412.

Received July 17, 1984Revision received September 21, 1984 •

eyewitness memory enhancement in the police - researchgate

Documents