Idea Transcript
The Psychological Record, 1999, 49, 137-144
MEMORY DISTORTION: CAN ACCURATE MEMORY BE PRESERVED?
SUSAN M. WINKELSPECHT and ROBERT R. MOWRER Angelo State University
A single experiment is reported which attempts to prevent or attenuate the distortion in memory produced by inaccurate postevent information. Specifically, after viewing a videotape, some participants received misleading information regarding the events depicted in the videotape. Further, some of these participants were given an inoculation procedure intended to educate them about the effects of misleading postevent information prior to receiving the misleading information. This material was presented in written form, videotape, or a combination of written and video form. Contrary to expectations , partiCipants who were misled and received an inoculation procedure evidenced greatest memory distortion . Results are discussed in terms of the wording used in a warning about memory distortion.
It is well established that a memory is not a perfect representation of an event and that new information obtained after the occurrence of an event may alter the way that event is remembered. If the new information is incorrect, the memory may become distorted (Loftus, 1979, 1991). Garry and Loftus (1994) described four categories of studies that represent memory distortion research: (a) the influence of leading questions on eyewitness testimony (e.g., Loftus & Palmer, 1974), (b) the suggestion of the presence of items that were not present during some event (e.g., Loftus, 1975), (c) the manipulation of details about an object present in some event (e.g., Loftus, 1979), and (d) the creation of an enti rely false or fictitious memory (see Loftus, 1997a, or 1997b, for a review). An additional category, in which researchers have looked for the source of memory distortion, is termed source misattribution (e.g., Zaragoza & Lane, 1994). . Much of the memory distortion research has focused on the effect of This research was conducted by Susan M. Winkelspecht in partial fulfillment of the requirements for the M.S. degree at Angelo State University. The authors thank Dr. Stephen F. Davis for his comments on an earlier version of this manuscript. Address all correspondence to Robert R. Mowrer, Department of Psychology and Sociology, Box 10907 ASU Station , Angelo State University, San Angelo , TX 76909. E-mail : Robert. Mowrer@ Angelo.Edu
138
WINKELSPECHT AND MOWRER
presenting new information to a participant after witnessing an event (e.g. , Loftus, Miller, & Burns, 1978). The basic design involves a three-step procedure. Participants watch a slide presentation or a video that is meant to simulate witnessing an event, and then the participants read a narrative or answer questions that are meant to mislead them; the questions or narrative include information not actually contained in the slide presentation or video. Finally, the participants are tested for their memory for details from the slides or video. The number of details the participant reports seeing (or witnessing) that they only read about in the misleading narrative or questionnaire is of interest. Using this procedure many studies have reported inducing participants to recollect a wide variety of items that were never present in the slides or video (e.g., Belli, 1988, 1989; Bowman & Zaragoza, 1989; Ceci, Ross, & Toglia, 1987; Chandler, 1991; Lindsay, 1990; Lindsay & Johnson, 1989; Loftus, 1979; Loftus, Donders, Hoffman, & Schooler, 1989; Loftus & Palmer, 1974; Zaragoza, McCloskey, & Jamis, 1987). W ith respect to the fact that participants can be misled, some research has focused on determining whether memory distortion can be prevented by warning the participants of the possibility of being misled. The logic behind this strategy is that if participants are warned that memory can be easily distorted they may be more vigilant and resistant to suggestion (O'Sullivan & Howe, 1995). However, merely warning participants they would be exposed to misleading information was only marginally successful in one case (Greene, Flynn, & Loftus, 1982) and unsuccessful in another (Zaragoza & Lane, 1994). One successful attempt was repo rted by Saywitz and Moan-Hardie (1994) using children. In this study, children received extensive training that included reducing the power differential between an interviewer and the children, indicating the negative consequences of errors, and showing accurate memory was more important than pleasing the adult interviewer. The children also received practice sessions intended to teach them resistance to suggestion. The children who received the training made fewer incorrect responses to misleading questions. The present study sought to incorporate some of the techniques used by Saywitz and Moan-Hardie (1994) into an inoculation procedure for adults. Unlike previous studies using adults (Greene et aI. , 1982; Zaragoza & Lane, 1994), the inoculation procedure used in the present study included a combination of instruction about memory distortion, a warn ing that the participants may be misled, and an attempt to reduce the power differential between the experimenter and the participant (to reduce demand characteristics). Specifically, after viewing a 12-min defensive driving v ideotape, participants were given either a written inoculation , a videotaped inoculation, a combined written/videotaped inoculation, or no inoculation. Some participants were exposed to misleading information in the form of a questionnaire; the questionnaire contained accurate information for other participants. All participants were then given a 40-item forced choice recognition test. It was expected that misled participants receiving the inoculation would perform similar to
MEMORY DISTORTION
139
those participants not misled and would differ from those misled but not receiving the inoculation. Further, misled participants receiving both the written and video inoculation were predicted to perform more accurately than those misled participants receiving only the written or video. This prediction was based on the idea that whereas the initial information was presented visually, the misleading information was presented verbally (written). It follows that the most effective means of preventing memory distortion would involve both modes of presentation. Method Participants One hundred and fifty-five college students from a regional southwestern university participated (123 women and 32 men). The participants ranged in age from 18 to 50 years (M = 23.3). Participants were randomly assigned to treatment conditions. Participation was entirely voluntary and some received extra course credit for their assistance. All individuals were treated in accordance the APA ethical principles regarding the use of human participants and the research was approved by the Institutional Review Board (IRB). Materials A 12-min defensive driving instructional video was shown to the participants. The video depicted traffic scenes, including some showing minor collisions and near collisions. There were no injuries depicted in any of the scenes. Some participants were also given an inoculation procedure consisting of a written text and/or a videotaped lecture about the fallibility of memory and a warning that they may be misled . The importance of accuracy rather than pleasing the experimenter (authority figure), and the possible consequences of their responses (Le., causing the wrong person to be punished by giving the wrong answer) were stressed. The written text and video were identical in terms of wording and required no more than 5 minutes to present. A filler written text and videotaped lecture on iconic memory were used for groups not exposed to one or both of the inoculation procedures. The filler activities were used to match all groups on the length of time spent viewing the inoculation video and/or reading the written inoculation text (retention interval). Two versions of a 20-item open-ended questionnaire about details of the defensive driving video were administered (postevent questionnaires). Both versions of the postevent questionnaire contained questions about actual details of the video. For one version, 10 of the questions were misleading (matched with the 10 accurate questions on the other questionnaire). For example, a misleading question was ''When Ben and his passenger were returning from a trip to the country, was his passenger injured when their car
140
WINKELSPECHT AND MOWRER
went nose-first into the ditch:' In the video, the car ran onto the shoulder of the road. The accurate version of this question replaced "nose-first into the ditch" with "went onto the shoulder of the road:' The not misled groups received 20 accurate questions and the misled groups received 10 accurate and 10 inaccurate questions. All questions (misleading and accurate) were chosen to match , as closely as possible, the types of questions used in previous studies. Finally, a 40-item two-alternative forced-choice test was administered. Ten of these items provided a choice between the accurate information (information contained in the video) and the misleading information contained in the postevent questionnaire (critical items). To clarify, the critical question relating to the example given above was 'Where did Ben's car go when he lost control returning from the country?:' The two choices were "a) into a ditch" and "b) onto the shoulder of the road." Ten items provided a choice between accurate information contained in the postevent questionnaire (and video) and information not contained in the video or postevent questionnaire (confirming items) . A confirming question was 'What color was the jacket of the child that Betty almost hit?" The choices were "a) blue" (in video and mentioned on postevent questionnaire) and "b) red" (not in video or on postevent questionnaire). The remaining 20 items provided a choice between accurate information from the video (but not mentioned in the postevent questionnaire) and other information not included in the video or postevent questionnaire (neutral items). A neutral question was "Neal was driving in bad weather conditions. What factor contributed to his accident?" The choices were "a) bald tires" (mentioned in video but not on postevent questionnaire) and "b) no headlights" (not mentioned in video or on postevent questionnaire). As with the postevent questionnaire, items for the forced-choice test were chosen to best match those used in previous studies. Procedure All participants viewed the defensive driving video in groups of 5 to 10. Each group was randomly assigned to a specific condition. After viewing the video, three groups of participants were given a version of the inoculation procedure. One group received the videotaped lecture, a second received the written text, and a third received both the videotaped lecture and the written text. The remaining group did not receive any version of the inoculation procedure. As a control, participants receiving only the videotaped lecture also read the irrelevant text involving iconic memory. Similarly, participants receiving only the written text viewed the irrelevant videotape involving iconic memory. Finally, participants receiving no inoculation received both the irrelevant videotape and written text. Next, all partiCipants were given the 20-item postevent questionnaire about the video. For some participants 10 of the questions were misleading, whereas for the others all of the questions contained accurate details of the video. One week later, all participants were given the recognition test. These arrangements resulted in a 2 x 4 x 3 mixed factorial design [information (misleading or not)
141
MEMORY DISTORTION
x inoculation x item/question type (critical, confirming, or neutral)] with question type being the within subject factor. The dependent variable was the percentage of accuracy for each type of question. Results Means and standard deviations for each condition are presented in Table 1. The data were analyzed using a 2 x 4 x 3 mixed factorial ANOVA. There was no significant main effect of inoculation type, F{3, 147) = 2.15, p= .097, MSE = .015. The analysis revealed a significant main effect of information type, F(1 , 147) = 10.84, P < .01 , MSE = .015. Participants who received misleading information were less accurate than those who did not. There was Table 1 Mean Accuracy Rates and Standard Deviations for Each Condition Item Type Critical
Neutral
Confirming
Misled
Not
Misled
Not
Misled
Not
Mean (SO)
Mean (SO)
Mean (SO)
Mean (SO)
Mean (SO)
Mean (SO)
No Inoculation
.84 (.13) N = 21
.91 (.09) N= 17
.96 (.05)
.94 (.07)
.83 (.09)
.88 (.07)
Video
.76 (.15) N = 24
.92 (.07) N = 24
.97 (.05)
.94 (.07)
.85 (.12)
.86 (.05)
Written
.73 (.15) N= 18
.85 (.11 ) N= 18
.95 (.07)
.94 (.07)
.84 (.11 )
.85 (.07)
Video/ Written
.76 (.14) N=17
.87 (.11 ) N= 16
.93 (.09)
.93 (.09)
.86 (.09)
.81 (.09)
also a significant main effect of item type, F{2, 294) = 91.16, p< .01, MSE= .007. Participants were more accurate on confirming items than either critical or neutral items. There was also a significant two-factor interaction between information type and item type, F(2, 294) = 26.36, P < .01, MSE = .007. Misled participants were less accurate on critical items than participants who were not misled . There were no significant differences in performance between misled and not misled participants on either confirming or neutral items. The remaining two-factor interactions, information type x inoculation and inoculation x item type, failed to reach significance; F{3, 147) = .23, MSE = .015 and F(6 , 294) = 1.77, MSE = .007, respectively. There was a significant three-way interaction, F{6, 294) =2.19, P < .05, MSE =.007. Analyses of the simple main effects identified the locus of the threefactor interaction as involving the critical questions. That is, these analyses revealed that there were no significant differences between any groups on
WINKELSPECHT AND MOWRER
142
either the confirming or neutral items. Of particular interest then were the analyses of the critical items. These analyses indicated that there was a significant main effect of the inoculation procedure for the misled conditions, F(3, 147) = 3.73, P < .05, but not for the not misled conditions, F(3, 147) = 1.33, P > .05, MSE = .015. Multiple t tests, using a Bonferroni correction (alpha = .008), revealed that the three groups that were misled and received an inoculation procedure were significantly less accurate than their corresponding misled, not inoculated control, but did not differ significantly from each other. Table 2 presents the tvalues with their associated p values for these pairwise comparisons on the critical items. Table 2 Multiple Comparisons for Misled Conditions on Critical Items None None
- - --
Video
Written
VideolWritten
t= -2.88 df= 36 p= .004
t= -4 .04 df= 39 p= .0001
t= -3 .00 df= 35 p= .003
t= 1.13 df= 43 p=0.26
t= 0.23 df= 37 p= 0.82
Video
t= 0.85 df=40 p= 0.40
Written
Discussion Contrary to our expectations, it appears the inoculation procedure, rather than preventing or attenuating memory distortion, actually enhanced it. This finding presents a challenge in terms of an explanation . One possibility involves the wording used in the inoculation procedure. In Zaragoza and Lane's (1994) study, which was unsuccessful, the warning used terminology that indicated to the participants that they would definitely be exposed to material that contained misleading information. In our study, we used terminology that indicated to the participants that they "may" be exposed to material that contained misleading ·information. Specifically, the last paragraph of the inoculation included the following statement: "In a few minutes you will be given a questionnaire about the video you just viewed. Some of these questions may [underline added] be misleading. It is possible that you may assume that the questions were written by somebody who knows the facts about the video, and therefore accept the misleading information as facts. Keep in mind that if [underline added] there are misleading questions, they may be only guesses, and it is very important that you accurately report your memory of the event." The uncertainty created by the words "may" and "if" might have resulted in the misled participants becoming more susceptible to the inaccurate information. According to Zaragoza and Lane (1994), the act of answering a misleading question leads participants to replay the original event and if the question was misleading, incorporate some of the inaccurate information
MEMORY DISTORTION
143
into a memorial representation. Zaragoza and Lane's participants, knowing that they were to be presented with inaccuracies, may have been more certain about which details were inaccurate than the participants in the present study who may have been uncertain as to whether or not there were misleading questions. Perhaps the combination of having received an inoculation procedure and the subsequent uncertainty among participants in our study led them to replay the original event to a greater degree than those who did not receive an inoculation .. This activity would allow more opportunity for the misleading information to become incorporated into the memory for the original event. It would follow then, that the not misled yet inoculated participants should also engage in greater rehearsal. However, because they were not exposed to misleading information they should not show a decrement in performance. Although not assessed in this study, the idea that inoculated participants spent more time evaluating the information (misleading or not) is consistent with the finding of Greene et al. (1982) that a warning caused participants to take longer to read the postevent narrative. It is difficult to compare our results to those of Greene et al. (1982) because they used a misleading narrative rather than a questionnaire. In addition, the exact wording used for their warning is unclear. However, their minimal success at preventing distortion may have been caused by the weaker misleading effect of a narrative relative to a questionnaire (Zaragoza & Lane, 1994). According to Zaragoza and Lane, misleading questions result in greater processing of the information relative to a narrative and thus more opportunity to incorporate the inaccurate information into a memorial representation. In the absence of a strong misinformation effect, the warning neither increased nor decreased memory distortion. It is unclear as to the particular reason our inoculation exacerbated the memory distortion effect. Our results suggest there may be subtle (or perhaps not so subtle) aspects critical to consider when developing a procedure des igned to prevent inaccuracies in memory because of postevent processing. Finding a procedure that can prevent, or at the minimum reduce, memory distortion is an important goal. Although the differences between our misled and not misled groups may only involve two or three questions, these may be the questions that end up having a significant impact on someone's life. Although this impact may be positive or negative, its inaccuracy is important.
References BELLI R. F. (1988) . Color blend retrievals: Compromise memories or deliberate compromise responses. Memory and Cognition, 16,314-326. BELLI, R. F. (1989). Influences of misleading postevent information: Misinformation interference and acceptance. Journal of Experimental Psychology: General, 118, 72-85.
144
WINKELSPECHT AND MOWRER
BOWMAN, L., & ZARAGOZA, M. (1989) . Similarity of encoding context does not influence resistance to memory impairment following misinformation. American Journal of Psychology, 102, 249-264. CECI , S. J., ROSS, D. F. , & TOGLlA, M. P. (1987). Suggestibility of children's memory: Psycho legal implications. Journal of Experimental Psychology: General, 116, 38-49. CHANDLER, C. C. (1991). How memory for an event is influenced by related events: Interference in modified recognition tests. Journal of Experimental Psychology: Learning, Memory and Cognition, 17, 115-125. GARRY, M., & LOFTUS, E. F. (1994). Pseudomemories without hypnosis. The International Journal of Clinical and Experimental Hypnosis, 42, 363-378. GREENE, E. , FLYNN , M. S. , & LOFTUS, E. F. (1982). Inducing resistance to misleading information. Journal of Vemal Learning and Vemal Behavior, 21 , 207-219. LINDSAY, D. S. (1990). Misleading suggestions can impair eyewitnesses' ability to remember event details. Journal of Experimental Psychology: Learning, Memory and Cognition, 16, 1077-1083. LINDSAY, D. S., & JOHNSON, M. K. (1989). The eyewitness suggestibility effect and memory for source. Memory and Cognition, 17, 349-358. LOFTUS, E. F. (1975) . Leading questions and the eyewitness report. Cognitive Psychology, 7, 560-572. LOFTUS, E. F. (1979). Reactions to blatantly contradictory information. Memory and Cognition, 7, 368-374. LOFTUS, E. F. (1991). Made in memory: Distortions in recollection after misleading information. In G. Bower (Ed.), The psychology of learning and motivation (Vol. 27, pp. 187-215). Orlando, FL: Academic. LOFTUS, E. F. (1997a). Creating false memories. Scientific American, 277(3) , 70-75. LOFTUS, E. F. (1997b). Memory for a past that never was. Current Directions in Psychological Science, 6(3), 60-65. LOFTUS, E. F. , DONDERS, K., HOFFMAN, H. G., & SCHOOLER, J. W. (1989) . Creating new memories that are quickly assessed and confidently held. Memory and Cognition, 17, 607-616. LOFTUS, E. F. , MILLER, D. G., & BURNS, H. J. (1978). Semantic integration of verbal information into a visual memory. Journal of Experimental Psychology: Human Learning and Memory, 4, 19-31. LOFTUS, E. F. , & PALMER, J. C. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of Vemal Learning and Vemal Behavior, 13, 585-589. O'SULLIVAN, J. T. , & HOWE, M. L. (1995). Metamemory and memory construction. Consciousness and Cognition, 4, 104-110. SAYWITZ, K. J., & MOAN-HARDIE, S. (1994). Reducing the potential for distortion of childhood memories. Consciousness and Cognition, 3, 408-425. ZARAGOZA , M . S., & LANE, S. M. (1994) . Source misattributions and the suggestibility of eyewitness memory. Journal of Experimental Psychology: Learning, Memory and Cognition, 20, 934-945. ZARAGOZA, M. S. , MCCLOSKEY, M., & JAMIS, M. (1987). Misleading postevent information and recall of the original event: Further evidence against the memory impai rment hypothesis. Journal of Experimental Psychology: Learning, Memory and Cognition, 13, 36-44.