Abducted by a UFO: prevalence information affects young children’s false memories for an implausible event Essay

SUMMARY This study examined whether prevalence information promotes children’s false memories for an implausible event. Forty-four 7–8 and forty-seven 11–12 year old children heard a true narrative about their ï ¬ rst school day and a false narrative about either an implausible event (abducted by a UFO) or a plausible event (almost choking on a candy). Moreover, half of the children in each condition received prevalence information in the form of a false newspaper article while listening to the narratives. Across two interviews, children were asked to report everything they remembered about the events. In both age groups, plausible and implausible events were equally likely to give rise to false memories. Prevalence information increased the number of false memories in 7–8 year olds, but not in 11–12 year olds at Interview 1. Our ï ¬ ndings demonstrate that young children can easily develop false memories of a highly implausible event. Copyright # 2008 John Wiley & Sons, Ltd. Both recent studies (e.g. Pezdek & Hodge, 1999; Strange, Sutherland, & Garry, 2006) and legal cases have demonstrated that children can develop memories of events that never happened, so-called false memories (Loftus, 2004). A well-known legal case is the ‘McMartin Preschool’ trial in which several teachers were accused of ritually abusing hundreds of children across a 10-year period (Garven, Wood, & Malpass, 2000; Garven, Wood, Malpass, & Shaw, 1998; Schreiber et al., 2006). Some of the children recalled extremely bizarre, implausible events such as ï ¬â€šying in helicopters to an isolated farm and watching horses being beaten with baseball bats. The charges against the teachers, however, were eventually dropped; videotapes of the investigative interviews indicated that the children were suggestively interrogated and many experts concluded that the children’s memories were almost certainly false. Controversial cases like the McMartin trial have inspired researchers to investigate how children develop false memories of implausible experiences (Pezdek & Hodge, 1999; Strange et al., 2006), yet the precise antecedents of implausible false memories are still ill-understood. The question we ask here is whether prevalence information—that is, details about the frequency of a false event—is a potential determinant of children’s implausible false memories. *Correspondence to: Henry Otgaar, Faculty of Psychology, Maastricht University, PO Box 616, 6200 MD, Maastricht, The Netherlands. E-mail: henry.otgaar@psychology.unimaas.nl Copyright # 2008 John Wiley & Sons, Ltd. H. Otgaar et al. What do we know about the role of prevalence information in the development of false memories? Mazzoni, Loftus, and Kirsch (2001) describe a three-step process that explains how false memories are formed. According to this model, three conditions must be satisï ¬ ed to create false memories. First, an event has to be considered plausible. Second, the event has to be evaluated as something that genuinely happened. Finally, images and thoughts about the event have to be mistaken as memory details. Consider, now, just the ï ¬ rst stage of Mazzoni et al.’s model (event plausibility) and how prevalence information might affect perceived plausibility. Recent experiments have shown that prevalence information enhances the perceived plausibility of implausible events (Hart & Schooler, 2006; Mazzoni et al., 2001; Pezdek, Blandon-Gitlin, Hart, & Schooler, 2006; Scoboria, Mazzoni, Kirsch, & Jimenez, 2006). Mazzoni et al. (2001) asked undergraduates to read false newspaper articles describing demonic possession. The articles implied, among other things (i.e. a description of what happens in a typical possession experience), that possessions were more common than people previously thought and after reading the articles participants were more likely to believe they had witnessed a demonic possession in the past. Other studies investigating the role of prevalence information in eliciting false beliefs have produced similar striking effects (Hart & Schooler, 2006; Mazzoni et al., 2001; Pezdek et al., 2006; Scoboria et al., 2006). What we do not know, however, is whether prevalence information inï ¬â€šuences the development of false memories (stage 3 of Mazzoni et al.’s model) and not just false beliefs per se. This is an important issue in the false memory literature because several authors have argued that memories and beliefs, although related, are deï ¬ nitely not the same (Scoboria, Mazzoni, Kirsch, & Relyea, 2004; Smeets, Merckelbach, Horselenberg, & Jelicic, 2005). Moreover, the effect of prevalence information has only ever been tested on adults’ beliefs. To date, no study has examined whether prevalence information affects the generation of children’s false memories. What do we know about event plausibility in the development of children’s false memories? In short, research has produced interesting but varied results. Early studies showed that children were more likely to create false memories of plausible than implausible events (Pezdek & Hodge, 1999; Pezdek, Finger, & Hodge, 1997), and researchers suggested that it may be difï ¬ cult to implant false memories of an implausible event (i.e. receiving a rectal enema). In contrast, one recent study shows that children will falsely recall both plausible and implausible events to a similar extent (Strange et al., 2006). Three different explanations might account for these mixed ï ¬ ndings. First, Strange et al. presented children with a doctored photograph of the false event whereas Pezdek and colleagues used false descriptions. Doctored photographs might be considered an extreme form of evidence -one that is very difï ¬ cult for children to refute. It is probable, then, that the doctored photographs skewed the children’s plausibility judgments which in   turn caused them to develop false memories for the plausible and implausible event at a similar rate. Second, Strange et al. compared false events that were either plausible or implausible whereas Pezdek and colleagues (1997, 1999) contrasted false events that differed in terms of script knowledge (i.e. description of what typically occurs in an event). Speciï ¬ cally, they compared a high script knowledge event (i.e. lost in a shopping mall) with a low script knowledge event (i.e. receiving a rectal enema). However, the exact relation between script knowledge and plausibility is not clear (Scoboria et al., 2004). Third, the two false events used in Strange et al.’s and Pezdek et al.’s studies differed with respect to valence. Strange et al.’s events were positive (i.e. taking a hot air balloon ride and drinking a cup of tea with Prince Charles), whereas Pezdek and colleagues implanted false negative events in Copyright # 2008 John Wiley & Sons, Ltd. children’s memory (i.e. lost in a shopping mall and receiving a rectal enema). Studies have shown that valence affects the development of children’s false memories (Ceci, Loftus, Leichtman, & Bruck, 1994; Howe, 2007). Since plausibility, valence and script knowledge seem to play a role in the development of false memories, the false events used in the current study were matched on these factors. To examine whether prevalence information can lead children to develop full-blown false memories of plausible and implausible events, and to examine developmental differences in the development of false memories, we adapted the false narrative procedure (e.g. Garry & Wade, 2005; Loftus & Pickrell, 1995; Pezdek & Hodge, 1999; Pezdek et al., 1997), and exposed some 7–8 year old children and some 11–12 year old children to one true description and one false description of past experiences. Previous studies have shown that these age groups differ developmentally with respect to suggestibility and false memory formation (e.g. Ceci, Ross, & Toglia, 1987). The true description described the child’s ï ¬ rst day at school. The false description was either plausible and described almost choking on a candy, or implausible and described being abducted by a UFO. Half of the children in each group also received prevalence information in the form of a newspaper article. The article suggested that the target false event was much more common than the children probably thought. Our predictions were straightforward: based on the prevalence literature with adults, we predicted that children who heard false prevalence information would be more likely to report false memories than children without false prevalence information. With respect to the role of event plausibility, two predictions can be formulated. Based on studies by Pezdek and colleagues (1997, 1999), we would predict that regardless of prevalence information, plausible events would elicit more false memories than implausible events. However, based on a recent study by Strange et al. (2006), we would expect that plausible and implausible events are equally likely to elicit false memories. Finally, because younger children are more suggestible than older children (for an overview see Bruck & Ceci, 1999), we expected that younger children would be more likely to develop false memories than older children. METHOD Participants The study involved 91 primary school children (48 girls) from two different age groups (n  ¼ 44, 7–8 year olds, M  ¼ 7.68 years, SD  ¼ 0.52; n  ¼ 47, 11–12 year olds, M  ¼ 11.64 years, SD  ¼ 0.53). Children participated after parents and teachers had given informed consent. All children received a small gift in return for their participation. The study was approved by the standing ethical committee of the Faculty of Psychology, Maastricht University. Materials True narratives True narratives described children’s ï ¬ rst day at school. This event was chosen because it was a unique event that had happened to all children at age 4. Children’s parents were contacted by telephone to obtain the following personal details about each child’s ï ¬ rst school day: the family members or friends who escorted the child to school, and the teacher’s and school’s name. These details were incorporated in the true narratives. Copyright # 2008 John Wiley & Sons, Ltd. An example of a true narrative was: Your mother told me that when you were 4 years old, you went for the ï ¬ rst time to the elementary school. The name of the elementary school was Springer and it was located in Maastricht. The name of your teacher was Tom. Your mother took you to school. False narratives False events were selected from a pilot study. In that study, 49 children (M  ¼ 8.02 years, SD  ¼ 1.20, range 6–101) rated the plausibility and valence of 29 events on child-friendly 7-point Smiley scales (anchors:  ¼ implausible/negative,  ¼ plausible/positive) with bigger smiley faces referring to more plausible/more positive events. Speciï ¬ cally, children had to indicate how likely the events were to happen to them (e.g. ‘How likely is it that you almost choke on a candy’?; i.e. personal plausibility; Scoboria et al., 2004) and how pleasant the events were for them (e.g., ‘How pleasant is it that you almost choke on a candy’?). To ensure that they understood the events, all children rated two practice items. Furthermore, 19 children (M  ¼ 8.74 years, SD  ¼ 1.05, range 7–10) were instructed to report everything they knew about each event and the total number of idea units served as our measure of children’s script-knowledge about the events (Scoboria et al., 2004). Based on their ratings, we selected two events, almost choked on a candy and abducted by a UFO. These events were equal in terms of valence (Mchoking  ¼ 1.65, SDchoking  ¼ 1.48, MUFO  ¼ 1.94, SDUFO  ¼ 1.98, t(47) < 1, n.s.) and script knowledge (Mchoking  ¼ 1.11, SDchoking  ¼ 0.99, MUFO  ¼ 0.74, SDUFO  ¼ 1.05, t(18)  ¼ 1.20, n.s.), but differed in terms of plausibility with mean plausibility ratings being higher for the choking event (M  ¼ 5.86, SD  ¼ 2.02) than for the UFO event (M  ¼ 1.63, SD  ¼ 1.75, t(47)  ¼ 10.07, p < .001). Age did not correlate with plausibility, valence and script knowledge for the two events ( ps > .05). Children’s parents conï ¬ rmed that their child had never experienced the false events. The false narratives were: Almost choked on a candy: Your mother told me that you were at a birthday party when you were 4 years old. At this party you received a bag of candies. When you were at home again, you were allowed to have one candy. Your mother saw that you turned blue and she panicked. Then she hit you on the back and the candy came out. Abducted by a UFO: Your mother told me that when you were 4 years old, you were abducted by a UFO. This happened when you were alone outside. You mother was inside the house. Then she suddenly saw through the window that a UFO took you. False newspaper articles For the true and false events a newspaper article was fabricated describing that the event took place quite frequently when participants were age 4. These false newspaper articles were similar in appearance to a local newspaper. Moreover, to personalize the newspaper articles, we included the children’s hometown in the articles. The newspaper articles were 1 Because the age range of our pilot sample did not completely overlap with the age groups of our study, we conducted a 2 (pilot group: younger vs. older children)  2 (event: UFO vs. choking) ANOVA with the latter factor being a within subject factor to examine the effect of age on plausibility judgments. No signiï ¬ cant interaction emerged ( p > .05) indicating that age did not have an impact on the plausibility ratings of our two events. Therefore, the plausibility ratings of our pilot sample can be extended to the older group of our study were randomly assigned to the plausible or implausible event and to the prevalence or no prevalence information condition. Each child was interviewed individually twice over seven days. All interviews were audio taped and transcribed. During the interviews, one true narrative and one false narrative were read aloud, with the latter always being presented in the second position. The procedure of the interviews was similar to that used by Wade, Garry, Read, and Lindsay (2002). At the start of Interview 1, children were told that we were interested in their memories for events that had happened when they were 4 years old. Children were instructed to report everything they remembered about the events. In the prevalence information condition, they were told that to help them remember the events they would be provided with a newspaper article. Subsequently, the interviewer read out the article to the child. Children who did not describe details of the target event were told that ‘many people can’t recall certain events because they haven’t thought about them for such a long time. Please concentrate and try again’. If they still did not recall any details, the interviewer made use of context reinstatement and guided imagery. The purpose of these retrieval techniques was to take the children mentally back to the scene of the event. Speciï ¬ cally, children were told to close their eyes and they were asked to think about their feelings, who was with them, and about the time of the year. After this, children were asked again to recall any details about the event. If they still did not come up with details, the next narrative was presented or the interview was stopped. At the end of Interview 1, children were asked to think about the events every day until the next interview and they were instructed not to talk with others about the events. Parents were asked not to discuss these events with their children. Interview 2 was similar to Interview 1. At the end of Interview 2, they were debriefed using ethical guidelines for false memory research with children (Goodman, Quas, & Redlich, 1998). RESULTS AND DISCUSSION An extensive number of children were extremely surprised during the debrieï ¬ ng when they were told that the false event did not happen to them. For example, one 8-year old child responded ‘It really did happen’ where another one said ‘I really can remember seeing the UFO’. After the debrieï ¬ ng, 39% (n  ¼ 13) of the children remained absolutely conï ¬ dent that they experienced the false events. We debriefed these children until they understood the events were false. Together, these ï ¬ ndings suggest that the false memories in this study were not the result of children falsely assenting or trying to please the interviewer. True events True memories were categorized as either remembered or not remembered. To be categorized as remembered, children had to report at least two of the three personal details correctly. Children’s true recall was near ceiling. They remembered 88 (97%) events at Interview 1 and 89 (98%) events during Interview 2, x2(1)  ¼ .07, n.s. False events For the false events, two independent judges classiï ¬ ed each memory report as no false memory, images but not memories or false memory according to criteria used by Lindsay, Hagen, Read, Wade, and Garry (2004). If a child attempted to recall the false event, but did Copyright # 2008 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 23: 115–125 (2009) DOI: 10.1002/acp Prevalence information, plausibility, and children’s false memories not have any memory of the event or did not report any details that were beyond the false description, the report was categorized as no false memory. A report was judged as an image when children speculated about details and described images related to the false events. For example, one child reported: ‘I think I almost choked on a candy on the birthday of Mauk. I am not sure. It was not a pleasant feeling’. To be classiï ¬ ed as a false memory, children had to indicate that they remembered the event and provide details beyond those mentioned in the narrative, but related to the narrative. To give an example of a detail, one child stated that he remembered being taken to the UFO through a blue beam of light. If children stated that they thought the event and/or certain details could have happened, then this was not scored as a false memory. Furthermore, to minimize the effect of demand characteristics, direct responses to interviewer prompts were not classiï ¬ ed as a false memory. The following dialogue from Interview 2 illustrates a child’s false memory of the UFO abduction. Child: ‘I saw cameras and ï ¬â€šashes and some people in the UFO’. Interviewer: ‘How many people did you see’? Child: ‘Approximately nine or ten’. Interviewer: ‘What kind of people’? Child: ‘People like me, children’. Interviewer: ‘What else did you see’? Child: ‘I saw some people and also some blue/green puppets were passing’. Inter-rater agreement for classiï ¬ cation of the memory reports was high; k  ¼ 0.92 for Interview 1 and k  ¼ 0.94 for Interview 2. Collapsing across the conditions, at Interview 1, 33% (n  ¼ 30) of the children developed a false memory. Thirty per cent (n  ¼ 9) of these children assented to the false events immediately, that is prior to guided imagery and context reinstatement. Thirty-six per cent of the children (n  ¼ 33), with 67% (n  ¼ 20) immediately assenting, ‘remembered’ the false events at Interview 2, x2(1)  ¼ 26.61, p < .001, Cramer’s V  ¼ 0.54. Some of the children who rejected the false events at Interview 2 indicated, despite the explicit instruction at Interview 1, that they had discussed the false events with their parents. The increase in false memories over time is in line with previous studies with adults and children (e.g. Lindsay et al., 2004; Strange et al., 2006; Wade et al., 2002). Furthermore, 10% (n  ¼ 9) of the children were classiï ¬ ed as having an image of the false events at Interview 1. At Interview 2, this percentage decreased to 7% (n  ¼ 6), x2(1)  ¼ 58.53, p < .001, Cramer’s V  ¼ 0.80. Recall that the primary question in this study was whether prevalence information boosts the likelihood of plausible and implausible false memories. Table 1 shows the percentage and number of children who reported false memories as a function of interview and condition. To examine the role of age, event type, and prevalence information in the development of false memories, we conducted a logistic regression analysis with the dependent variable being false memory (0  ¼ no false memory/images, 1  ¼ false memory). In this analysis, we only focused on ‘genuine’ false memories and did not collapse across false memories and images. Although non-parametric methods, such as logistic regression, often lack the statistical power to detect interactions (Sawilowsky, 1990), there are four important points to note about these data. First, the only signiï ¬ cant interaction found was an Age  Prevalence information interaction at Interview 1. Prevalence information enhanced the development of 7–8 year old children’s false memories but not 11–12 year old children’s false memories, and this effect occurred at Interview 1 (B  ¼ 2.16, SE  ¼ 0.96, Copyright # 2008 John Wiley & Sons, Ltd.