Normalizing forgetting of childhood events
1
Running Head: NORMALIZING FORGETTING OF CHILDHOOD EVENTS
So that is why I don’t remember: Normalizing forgetting of childhood events influences false autobiographical beliefs but not memories
Alan Scoboria Department of Psychology, University of Windsor, 401 Sunset, Windsor, ON, Canada, N9B3P4 Steven Jay Lynn Binghamton University, Binghamton, NY, 13902, United States Joanna Hessen Department of Psychology, University of Windsor, 401 Sunset, Windsor, ON, Canada, N9B3P4 Stephanie Fisico Department of Psychology, University of Windsor, 401 Sunset, Windsor, ON, Canada, N9B3P4
Memory, 15(8), 801-813 Corresponding author: Alan Scoboria, Ph.D., C.Psych. 282 Chrysler Hall South 401 Sunset Avenue Windsor, Ontario N9B 3P4 519-253-3000, x4090 fax 519-973-7021
[email protected]
Keywords: autobiographical memory, false memory, belief, plausibility
Normalizing forgetting of childhood events
Abstract Changes in autobiographical belief and memory ratings for childhood events, after informing individuals that forgetting childhood events is common, were investigated. Participants received false prevalence information (indicating that a particular childhood event occurred frequently in the population) plus a rationale normalizing the forgetting of childhood events, false prevalence information alone, or no manipulation, for one (Study 1) or two (Study 2) unlikely childhood events. Results demonstrated that combining prevalence information and the “forgetting rationale” substantially influenced autobiographical belief ratings, whereas prevalence information alone had no impact (Study 1) or a significantly lesser impact (Study 2) upon belief ratings. Prevalence information consistently impacted plausibility ratings. No changes in memory ratings were observed. These results provide further support for a nested relationship between judgments of plausibility, belief, and memory in evaluating the occurrence of autobiographical events. Furthermore, the results suggest that some purported false memory phenomena may instead reflect the development of autobiographical false beliefs in the absence of memory.
2
Normalizing forgetting of childhood events
3
So that is why I don’t remember: Normalizing forgetting of childhood events influences false autobiographical beliefs but not memories. Researchers have lavished considerable attention on how false memories develop, and have demonstrated that false memory formation is a robust phenomenon (Loftus & Bernstein, 2005). In a survey of the literature, Lindsay, Hagen, Read, Wade and Garry (2004) reported that false memories have been created on average in 37% of participants across studies. Numerous procedures, such as social pressure, repeated recall efforts, false feedback that events are likely to have occurred, imagination, dream interpretation, doctoring photographs, and the inclusion of self-relevant details in false narratives have been associated with false memory creation (Loftus & Pickrell, 1995; Hyman & Pentland, 1996; Mazzoni Lombardi, Malvagia, & Loftus, 1999; Wade, Garry, Read, & Lindsay, 2002; Garry & Wade, 2005; Desjardins & Scoboria, in press). Other research has begun to dismantle the components of these complex false memory paradigms, to examine how different manipulations influence memory creation (eg., Bernstein, Whittlesea, & Loftus, 2002; Mazzoni & Memon, 2003; Sharman, Manning & Garry, 2005). Based on this burgeoning literature, researchers (Ost, 2003; Rubin, Schrauf & Greenberg, 2003; Scoboria, Mazzoni, Kirsch, & Relyea, 2004; DePrince, Allard, Oh, & Freyd, 2004; Smeets, Merckelbach, Horselenberg, & Jelicic, 2006) have made a distinction between the concepts of autobiographical belief (believing that an event occurred, whether or not one has an accompanying memory; eg. most individuals believe they were born, but do not remember the event), and autobiographical memory (remembering the event, with accompanying perceptual detail). This distinction has important implications for understanding false autobiographical remembering. Whereas work in this area has focused on the development of memories, it is equally important to understand how people come to believe that events did occur, whether or
Normalizing forgetting of childhood events
4
not they report a memory for the event. Various published sources indicate that people make decisions based upon autobiographical beliefs that are held in the absence of memory. Bernstein, Laney, Morris, and Loftus (2005) reported that participants who received a suggestion about having become sick eating a particular food during childhood, and subsequently reported reduced preference for the food, did so largely because they believed that the event occurred; few participants reported a memory for the suggested event. Ost (2003) discusses individuals who “remembered” and then later retracted claims of childhood abuse, who report that they never actually remembered the abuse, but rather came to strongly believe that the abuse must have occurred. Kassin (1997), in drawing parallels between the false memory and false confession literatures, observes that some individuals who confess to an act that they have not committed internalize the act (indicating they believe they have done it, analogous to a false autobiographical belief); however, only a small subset also provide elaborative accounts of what they had done (analogous to a false memory). Finally, Gudjonsson (2003, pp. 223-224) discusses the case of a man who had spent the night at a woman’s apartment, and was accused by her the next day of having stolen her purse. When accused, he was unable to state that he did not take the item, and in light of absence of the purse and his view of his generally poor memory skills, reported believing that he must have taken the item (that was later found to have fallen behind a piece of furniture). This individual came, fortunately for a brief time, to believe he had committed a crime, despite lacking memory for the behaviours involved. We will return to this concept – that someone may come to believe that an event occurred based upon expectations about the function of memory. Few studies have empirically examined the distinction between belief and memory. Most studies that examine the formation of false autobiographical memories do not assess belief.
Normalizing forgetting of childhood events
5
There are several exceptions. Mazzoni and Memon (2003) demonstrated that imagination can lead to the creation of false memories that are accompanied by false autobiographical beliefs. Desjardins and Scoboria (in press) reported that individuals exposed to self-relevant information in false event narratives endorsed more false memories as compared with those not exposed to self-relevant information, but that these groups did not differ in levels of belief. However, both of these studies were primarily interested in false memory formation, and neither study was designed to examine factors that differentially influence belief and memory ratings. Other research suggests that autobiographical belief can be distinguished from other processes thought to be involved in the formation of false memories. Mazzoni, Loftus, and Kirsch (2001) demonstrated that plausibility and belief are influenced by different manipulations; plausibility by providing information about the frequency and nature of events, and belief by providing false feedback that events likely occurred. However, this work did not assess memory. The “imagination inflation” literature has explored the effect of imagination on changes in likelihood ratings for childhood events using the Life Events Inventory (LEI; Garry, Manning, Loftus & Sherman, 1996). Although this and similar studies (eg. Heaps & Nash, 1998; Pezdek & Eddy, 2001; Sharman, Manning & Garry, 2005) demonstrate that briefly imagining events results in enhanced likelihood ratings (analogous to autobiographical belief), these studies do not discriminate belief and memory, rendering the interpretation of results ambiguous. Furthermore, belief and memory have been distinguished from a third construct relevant to false memory formation: event plausibility. The first study examining plausibility (Pezdek, Finger & Hodge, 1997) demonstrated that participants were more likely to endorse as a memory a suggested event that was high in plausibility (lost in a mall) than an event low in plausibility. Subsequent work has demonstrated that the plausibility of events is not fixed, and that greater
Normalizing forgetting of childhood events
6
increases in likelihood ratings are associated with accompanying increases in plausibility ratings (Mazzoni, Loftus, and Kirsch, 2001). Other studies have shown that providing false prevalence information results in increased Life Events Inventory likelihood ratings (Hart & Schooler, 2007; Pezdek, Blandon-Gitlin, Lam, Hart & Schooler, 2006). Furthermore, Hart and Schooler, (2007; Study 2), reported that prevalence information, while impacting likelihood judgments, does not influence memory ratings. However, in these studies plausibility and likelihood were not measured separately; thus it is unclear whether belief, plausibility, or both were affected. The one study that examined the effect of prevalence information on plausibility, belief and memory simultaneously (Scoboria et al, 2006) found that prevalence information consistently influenced plausibility ratings, whereas autobiographical belief ratings were impacted to a lesser degree for one event (bone density screening) and not at all for another event (enema). A recent model (Scoboria et al., 2004) articulates a conceptual framework that encompasses plausibility, belief, and memory for autobiographical events. This model discusses the constructs of autobiographical memory (remembering an event), autobiographical belief (believing an event did occur, whether or not it is remembered), personal plausibility (judging that an event could have occurred in relation to oneself, whether or not it is believed or remembered) and general plausibility (judging that an event could have occurred in relation to someone). These concepts are generally in line with models of false memory formation (Hyman & Loftus, 1998; Hyman & Kleinknecht, 1999; Mazzoni & Kirsch, 2002). Data collected using the Autobiographical Beliefs and Memory Questionnaire (ABMQ) supports the distinction between plausibility, belief, and memory (Scoboria et al., 2004; Scoboria et al., 2006), as well as the hypothesis that these constructs are “nested.” This means that
Normalizing forgetting of childhood events
7
memory for an event necessarily implies belief and plausibility, and belief implies plausibility. However, viewing an event as plausible does not necessarily indicate belief, and believing that an event occurred does not necessarily indicate memory. In both of the aforementioned studies, the nested relationship was upheld in over 95% of cases (exceptions are discussed in Scoboria et al., 2004). Scoboria, et al., (2004) also reported that these constructs covary, but not to such a degree that they can be considered to tap identical processes. Thus it should be possible to manipulate plausibility, belief and memory partially independently. Specifically, the model predicts manipulations that impact plausibility should have little impact on belief ratings. However, manipulations that impact belief ratings should also impact plausibility, but should not notably impact memory ratings. Demonstrating that some procedures have a greater impact on plausibility alone, whereas others affect both autobiographical belief and plausibility, would provide further evidence for the validity of the nested model, as well as expand understanding of the nature of belief. The nested model suggests that in order to enhance belief, judgments of an event must shift from “could have occurred” to “did occur.” We propose that this shift can be facilitated by normalizing the forgetting of events – in the case of the present study, childhood events. Many of the aforementioned false memory formation studies informed participants that forgetting childhood events is common, and that events can often be remembered at a later time. For example, in Wade et al. (2002), participants unable to remember an event were informed that, “many people can’t recall certain childhood events at first because they haven’t thought about them for such a long time.” (p. 599); and in Hyman and Pentland (1996), participants were told that it is, “normal for people to remember more as they worked at recalling events” (p 105). We hypothesized that such instructions may alter the decision criteria for evaluating the occurrence of unremembered
Normalizing forgetting of childhood events
8
childhood events, by providing an explanation as to why events may have been forgotten. However, such a “forgetting rationale” alone is unlikely to lead to a change in autobiographical belief for any specific event, as there is no reason to think that any particular unremembered event is likely to have occurred without specific reason to consider that event. Rather, a combination of prevalence information about a specific event, that has been shown to enhance plausibility (Scoboria et al., 2006), in combination with a rationale as to why the event is not recalled, may lead to increases in belief. This argument is consistent with Mazzoni and Kirsch’s (2002) metacognitive model regarding decision processes involved in determining whether an unremembered event did or did not occur (see also Strack & Bless, 1994). They argue that individuals use many sources of information when evaluating their lack of memory for the event. When an event is not remembered, the degree to which the lack of memory is diagnostic of the non-occurrence of the event is evaluated. If the lack of memory is determined to be diagnostic, then the event is rejected as not having occurred. If the lack of memory is not determined to be diagnostic, then other information is used to infer whether or not the event occurred. Expectations about the forgetting of childhood events are one such source of information. If forgetting is viewed as common, then one may be less likely to interpret a lack of memory as indicating that something did not occur (at least in relation to other evidence that it may have occurred). Thus an expectation that memory is unreliable may shift the balance of evidence regarding the occurrence of the event away from internal processes, in favor of compelling external evidence. The external evidence may then be used to infer that the event in fact occurred. We conducted two studies to investigate the effect of providing a rationale for forgetting childhood events on ratings of general plausibility, personal plausibility, autobiographical belief
Normalizing forgetting of childhood events
9
and memory for unlikely childhood events. Undergraduate participants provided baseline Autobiographical Belief and Memory Questionnaire data. One to two weeks later they received prevalence information and a rationale normalizing the forgetting of childhood events, prevalence information alone, or no manipulation, following which they then provided postmanipulation data. We predicted that the combination of the prevalence information and forgetting rationale would enhance both plausibility and autobiographical belief ratings. Furthermore, we predicted that prevalence information would primarily affect plausibility ratings, whereas belief ratings be influenced to a much lesser extent, if at all. Finally, as the manipulations are not directed toward enhancing episodic memory-like characteristics of events, we predicted that neither manipulation would influence memory ratings. Study 1 Method Participants Participants were 62 undergraduate students enrolled in psychology courses at the University of Windsor, (48 female; age range 17 – 45, M = 20.22, SD = 2.91). Four participants outside the target age range (17-30), and two who did not complete Time 2 until 21 days after Time 1, were excluded from analyses. Participants received course credit for participation. Measures ABMQ. The Autobiographical Beliefs and Memory Questionnaire (ABMQ; Scoboria et al., 2004) assesses (in order) general plausibility, personal plausibility, autobiographical belief, and memory. Participants read instructions to aid in discriminating among constructs, following which they rated each variable for 10 events. Each variable is rated on a 1 to 8 point Likert-like scale, anchored ‘Not at all plausible’ and ‘Extremely plausible’ for general and personal plausibility; ‘Definitely did not happen’ and ‘Definitely happened’ for autobiographical belief;
Normalizing forgetting of childhood events 10
and ‘No memory for event at all’ and ‘Clear and complete memory for event’ for memory. Baseline and post-manipulation measures consisted of the ABMQ applied to 10 events. The events were selected from recent work using the ABMQ to represent a range of plausibility (Scoboria et al., 2004), and included (in order of presentation): losing a toy (high plausibility), having a tooth extracted (moderate plausibility), seeing a ghost (low plausibility), bone density screening (moderate), receiving an enema (moderate), having a skin sample taken by a school nurse (moderate), choking on a small object (high), getting lost in a shopping mall (high), witnessing a demonic possession (low), and seeing parents naked together (moderate). Receiving a bone density screening as a child was selected as the target event to include a medical event highly unlikely to have occurred in our sample, because prior estimates have indicated it is rated moderate in plausibility on average, and because this target event has been used in prior research on the nested model (Scoboria et al., 2006). The target event was presented fourth in the series of events. This placement allows for an event of each general level of plausibility (high, low, moderate) to precede the target event. The purpose of this is to assist subjects in calibrating their responses on the ABMQ. Procedures Procedures were adapted from Scoboria et al. (2006), and were approved by the University of Windsor Research Ethics Board. Participants were contacted via email and invited to participate in a two-part experiment. They then completed informed consent and the Time 1 ABMQ on-line within 24 hours. Two weeks later (M = 14.13 days, SD = 2.44), participants attended an in-lab session, where they were randomized to conditions, and completed the procedures at a computer terminal. Participants in the prevalence plus forgetting rationale group viewed 5 web pages. The first page introduced the study. The second page provided false
Normalizing forgetting of childhood events 11
prevalence information indicating that bone density screenings were common for children in Ontario between 1978 and 1988, following which participants completed questions about the material. These questions consisted of a manipulation check, that asked participants to indicate about what they had read, and a series of questions about the article to encourage encoding of the information (i.e., whether the information was previously known, whether parents should have permitted the procedure, whether the information was surprising, if the information was uncomfortable, whether the information was helpful, the quality of the writing, and the convincingness of the writing). The third page provided a medical narrative (adapted from Hart & Schooler, 2007; and used in Scoboria et al., 2006), that was included as a filler task and which was followed by a similar series of questions. The fourth page provided the following rationale regarding forgetting of childhood events: Information about Forgetting and Remembering Many people forget events that happened to them in their past. Psychoanalysts and therapists have discussed forgetting in terms like "repression," or "dissociation." However, these two terms that generally refer to the forgetting of traumatic events are really too narrow. We have all had the experience of forgetting everyday events that have happened, only to recall them later. In some instances, we may even be surprised that we forgot a particular occurrence in the first place. Remembering and forgetting occur all of the time. Even events that were emotionally positive or negative at the time are often forgotten. They can then be recovered at a later time by thinking carefully and trying to feel if the event really did occur. Some of the information you have just read provides evidence that you very likely underwent a bone density screening during your childhood. We are going to ask you to rate childhood events another time. Please think very carefully about each event when you rate it and whether it might have occurred.
Participants then filled out the post-manipulation ABMQ and were debriefed. Participants in the prevalence only group viewed the same pages, absent the fourth page (the forgetting rationale). Participants in the control group did not view the second (prevalence)
Normalizing forgetting of childhood events 12
or fourth (forgetting rationale) pages. Study 1 Results Baseline and post-manipulation scores for the target event are reported in Table 1. As expected, the target event was rated as moderate in general plausibility, and low in personal plausibility, belief, and memory. These baseline ratings were similar to those reported in Scoboria et al. (2006). No differences were found among the treatment groups at baseline. A series of planned comparisons were conducted to assess change between baseline and post-manipulation scores, with Bonferroni corrections for multiple comparisons. Separate analyses were conducted controlling for and not controlling for differences in Time 1 responding; results did not differ by analytic method. Due to concerns about the reliability of change scores (Keppel & Zedeck, 1989), and the fact that the covariate was significant in all analyses (all p < .01), the analyses including the covariate are presented below. Mean adjusted change scores for all variables are presented in Figure 1. Autobiographical belief The main prediction was that participants in the prevalence plus forgetting rationale group would show greater increase in belief ratings than the other groups. This was the case. The prevalence plus forgetting group increased significantly on belief as compared with the other groups, t (1,54) = 4.81, p < .01, d = 1.21. A final contrast confirmed that the prevalence alone and control groups did not differ on this variable, t (1,34) = 1.67, ns. Frequencies of increases of two or more points on the belief scale were also analyzed by group, which demonstrated significant differences in cell frequencies, Χ2 (2) = 18.43, p < .01. A significantly greater proportion of participants in the prevalence plus forgetting group increased on this variable (50%), as compared with the remaining groups (5% in each).
Normalizing forgetting of childhood events 13 General and Personal plausibility The second prediction was that the prevalence groups would demonstrate greater change in general and personal plausibility than the control group, but would not differ from one another. Both predictions were upheld. Analysis of group means using independent samples ttests demonstrated that the two groups receiving prevalence information increased significantly as compared with the control group for general plausibility, t (1,54) = 2.24, p < .05, d = .60; and personal plausibility, t (1,54) = 3.46, p < .01, d = .94; whereas the two prevalence groups did not differ on either variable (both p > .10). Memory The prediction for memory ratings was for no group differences, and no notable change. A one-way ANOVA failed to demonstrate significant differences between groups on memory, F (1,54) = .83, ns. Furthermore, a one group t-test failed to demonstrate a difference between the average level of change on memory and zero, t (54) = .73, ns. Non-target events Finally, we examined change on ABMQ variables for the nine non-target events using mixed ANCOVA and ANOVA analyses, resulting in a 9 (events within subjects) by 3 (treatment groups between subjects) design. No significant main effect or interactions were observed (all p > .10). Thus, the manipulations appeared to have systematically influenced only the target event. Study 1 Discussion As predicted, the provision of false prevalence information resulted in changes in general and personal plausibility ratings. However, it was only the addition of a rationale normalizing the forgetting of childhood events that influenced autobiographical belief ratings. Furthermore, none of the manipulations influenced memory ratings. These results provide further support for the
Normalizing forgetting of childhood events 14
distinctions articulated in the nested model, and the argument that plausibility, belief, and memory represent partially independent judgments involved in evaluating the occurrence of unremembered childhood events. We conducted a second study to address several methodological limitations of Study 1. First, to assess the generalizability of the findings, we added a second target event. Having a skin sample taken by a school nurse was selected as it was deemed similar to the density event in being unlikely to have occurred in childhood, contains similar medical content, and has previously shown ABMQ ratings similar to those of bone density screening (Scoboria et al., 2004). Secondly, we partially controlled for order effects by counterbalancing the presentation of target events. Whereas in Study 1 the target event appeared fourth in the list of events, in Study 2, the target event randomly appeared early (fourth) or late (seventh) on the ABMQ. We predicted that the results of Study 1 would be replicated, and that no statistical differences would emerge as a function of target event or order. Study 2 Method Participants Participants were 164 undergraduate students enrolled in psychology courses at the University of Windsor, (88% female; age range 17 – 46, M = 21.84, SD = 4.58). Eight female participants who were outside the target age range (17-30) were excluded from analyses. Participants received course credit for participation. Measures The ABMQ, described above, was modified so that the target event randomly appeared in either the fourth or seventh position on the list of ten events. Procedures
Normalizing forgetting of childhood events 15
Procedures were identical to Study 1, with the exception that just over one week elapsed on average between Time 1 and Time 2 (mean = 7.63 days, SD = 2.13). Furthermore, the prevalence information described either bone density screening or skin sample taken by a school nurse as having been common for children in Ontario between 1978 and 1988. Participants were randomly assigned to one of five conditions (prevalence plus forgetting rationale for density or skin; prevalence for density or skin; or no manipulation), within which half of the participants encountered the target event fourth, and half encountered the target event seventh on the ABMQ (events randomly appeared fourth or seventh for the control group as well). Study 2 Results Baseline ratings for the two target events are presented in Table 2. We note that baseline ratings were highly similar for the two events. As in Study 1, both events were moderate in plausibility, and low in belief and memory. No differences were found among the treatment groups at baseline (all p > .10). We first examined whether results differed as a function of event type and/or location of target event on the ABMQ. Analyses indicated that neither factor influenced the results. The full model was tested using 3x2x2 between subjects ANCOVAs (controlling for baseline responding) and ANOVAs (group: prevalence + forgetting rationale, prevalence alone, control; event: density, skin; event order: fourth, seventh) on change scores for the four ABMQ variables. No significant main effects or interactions were observed for event type or order of target event placement for any of the variables (all p > .10) using either analytic approach. We thus opted to collapse across order and event type, and conducted a series of planned comparisons with Bonferroni corrections on baseline adjusted change scores to determine whether the findings of Study 1 had been replicated. Baseline and post-manipulation scores for target events are reported
Normalizing forgetting of childhood events 16
in Table 3. Mean adjusted change scores are presented in Figure 2. Autobiographical belief In Study 1, the prevalence plus forgetting rationale group demonstrated greater change in belief than did the prevalence or control groups, which did not differ from one another. This finding was replicated. Prevalence plus forgetting resulted in greater change in belief than in the other groups, however the prevalence only and control groups also differed significantly. Comparison of the prevalence plus forgetting and other groups revealed that the former increased significantly on belief relative to the latter, t (1,153) = 8.32, p < .01, d = .66. Furthermore, the prevalence only group increased more than the control group, t (1,94) = 4.62, p < .01, d = .80. A final contrast confirmed that the prevalence plus forgetting rationale group demonstrated greater change in belief than the prevalence alone group, t (1,153) = 2.96, p < .01, d = .43. Thus, prevalence information resulted in change in belief ratings relative to the control group; and adding the forgetting rationale resulted in greater change than prevalence alone. In Study 1, frequencies of increases on belief were also greater in the prevalence plus forgetting rationale group than in the other groups. This result was also replicated. Frequencies of increasing 2 or more points on belief demonstrated significantly different cell frequencies, Χ2 (2) = 18.47, p < .01. A greater proportion of participants in the prevalence plus forgetting group increased (45%), as compared with the prevalence (24%) and Control (6%) groups. General and Personal plausibility In Study 1, the experimental groups showed greater change in both general and personal plausibility than the control group, but did not differ from one another. This result was replicated, with the exception that in Study 2 the prevalence plus forgetting rationale group increased more in personal plausibility than the prevalence group. Analyses demonstrated that
Normalizing forgetting of childhood events 17
groups receiving prevalence information increased significantly as compared with the control group for general plausibility, t (1,151) = 9.27, p < .01, d = .88, and personal plausibility, t (1,151) = 4.61, p < .01, d = .57. The prevalence plus forgetting and prevalence groups did not differ significantly in general plausibility, t (1,126) = 1.06, ns, but did differ in personal plausibility, t (1,126) = 4.31, p < .01, d = .37, favoring the prevalence plus forgetting group. Memory In Study 1, groups did not differ in change in memory ratings, and change in memory was found to be statistically equal to zero. This finding was replicated. Statistical tests failed to demonstrate differences between groups in memory, F (2,152) = 1.50, ns; and change in memory did not differ from zero, t (155) = 1.27, ns. Non-target events As in Study 1, we examined average levels of change on ABMQ variables for non-target events. The treatment groups were contrasted using analyses as described for Study 1. No significant main effects or interactions were observed (all p > .10). Study 2 Discussion Study 2 replicated the main findings of Study 1. In both studies, prevalence information resulted in significant change in plausibility ratings, relative to the control group. Adding the forgetting rationale to prevalence information resulted in significant change in autobiographical belief ratings, relative to both plausibility alone and controls. Furthermore, none of the manipulations impacted memory ratings. Several differences did emerge between the two studies. First, change in personal plausibility ratings was greater in the prevalence plus forgetting rationale group than the prevalence alone group in Study 2, whereas these groups were statistically equal in Study 1.
Normalizing forgetting of childhood events 18
Examination of Figure 1 suggests that this pattern may have been present in Study 1 as well, although not reaching statistical significance. Secondly, the prevalence manipulation did not influence belief ratings in Study 1, but resulted in belief change in Study 2. As personal plausibility and belief are conceptually related with a degree of covariance, such inconsistencies are not surprising. Prior work (Scoboria et al., 2006) has found prevalence information to influence personal plausibility and to a lesser degree belief for bone density screening, whereas it affected only personal plausibility for rectal enema. Thus, prevalence information sometimes impacts only personal plausibility, and other times influences both personal plausibility and belief to a lesser extent. Further work is needed to determine if this is attributable to the relationship between variables, the method of measurement, or other factors. General Discussion These findings are generally consistent with the nested model proposed by Scoboria, et al. (2004), which predicts that manipulations which target belief will influence belief but not memory ratings; and manipulations that target personal plausibility will influence general plausibility but not belief or memory. Models that articulate the development of false memories can be revised to include autobiographical belief. For example, Hyman and Kleinknecht (1999) discuss false memory formation as a three-step process: 1) accepting the event as plausible; 2) constructing a mental representation that takes the form of a memory; and, 3) committing a source monitoring error, by which the representation is attributed to being a memory, rather than an imagination. The emergence of belief as an important construct suggests that an additional step is needed to refine this model: following enhancement of plausibility, belief in the occurrence of the event must develop prior to the misattribution of mental representations as a
Normalizing forgetting of childhood events 19
memory. This leads to interesting questions about which processes differentially impact belief and memory, as well as the degree to which belief must be present before memories are created. We now turn our attention to the finding that prevalence information impacted plausibility ratings, but prevalence plus a rationale for forgetting led to substantially increased belief ratings. What might explain these findings? The false prevalence information likely performs a number of functions. First, it establishes that the event commonly occurred within the culture. This is the aspect of the manipulation that likely targets general plausibility: by providing information that the event occurs, it is not surprising to find that ratings that the event did occur in relation to others increased. The prevalence information also makes the event relevant to the individual themselves, by indicating that the event occurred in relation to a high proportion of a segment of the population that participants were members, in a particular geographic locale, and at a time in childhood that the individual knows they were present. Scoboria et al. (2006) have recently argued that what distinguishes general from personal plausibility is a shift to self-relevance. Once individuals come to see an event as potentially having occurred to them, judgments of personal plausibility will increase. To develop an autobiographical belief, one must move from “could have occurred” to “did occur.” This requires either reliable evidence that the event did in fact occur (eg. from a family member, or viewing a photograph), and/or a rationale as to why one does not remember the occurrence of the event. By normalizing the forgetting of childhood events, individuals may view childhood memory as a less reliable source of information for determining the occurrence of the event. Thus, individuals might shift to rely more upon external sources of information to decide whether an event did or did not occur. Memories, however, do not develop until an episodic mental representation is created that is attributed to a past event.
Normalizing forgetting of childhood events 20
There are several limitations of this work which should be noted. We note that the hierarchical ordering of items on the ABMQ makes ratings dependent upon one another. We do not know how the fact that individuals know the ratings they have made, as well as which ratings they are about to make, may have affected the results. This is a problem inherent in any method in which several constructs are systematically measured – it is challenging to prevent ratings from influencing one another. Thus, at present it is unknown to what degree plausibility, belief, and memory judgments are in fact reasonably independent, and to what degree the pattern of results is an artifact of measurement. As all responses are selected in relation to other responses, the evidence for independence of ratings is tentative. Even with dependent ratings, our results provide interesting information about how individuals judge the plausibility, belief, and memory for unlikely events. Though participants in these studies were aware of their multiple ratings, they were not aware of the between-groups nature of the design. Hence, the theoretically consistent differences between the groups, and the fact that these differences were replicated across studies, cannot be explained in terms of the judgments being dependent upon one another. Several research strategies might assess the dependence of judgments. First, the structure of the nested model can be tested. Rather than being made aware that all four variables will be rated prior to making ratings, individuals could be instructed to rate a series of events for one construct at a time. By systematically varying item order, responses when raters are unaware of the next construct to be rated can be obtained. Thus, estimates uninfluenced by previously rating other variables can be evaluated. Presumably, if these constructs are reasonably independent, the structure of the model would hold under different item orders (see Scoboria et al., 2004, for further discussion of the model structure). A second step would then be needed to demonstrate
Normalizing forgetting of childhood events 21
that manipulations such as those employed in the present studies continue to impact variables as predicted. One approach could involve the use of multiple target events per participant in a similar design, counterbalancing order of items, and blocking ratings such that each of the items is rated first for each target event. A complementary approach would have participants in different groups make different clusters of ratings: some making only belief ratings, some making only plausibility ratings, others making both ratings, and so forth. A related concern is the use of multiple ratings of multiple events. As the same items are rated event after event, a similar concern arises that ratings of preceding events and anticipation of rating future events may impact results. This criticism is common to much of the published false memory literature. As with many methodological decisions, there are trade-offs between measuring multiple versus few events. Providing multiple events can enhance the reliability of findings: if results hold despite the presence of multiple distracters, the findings are more convincing. If only target events and critical items are presented, a study can be criticized as leading participants to produce desired outcomes. Furthermore, strategies such as completely randomizing the presentation of events and ABMQ items are not desirable, as this would place undue expectations upon raters to shift mental set from item to item. This being said, future work could systematically examine the impact of the number of target events, and the order of presentation of items and events to explore the impact of number of ratings and ordering of items on ABMQ based results. This would also help to address the issue of dependent ratings. A further concern is whether plausibility, belief, and memory represent distinct processes, or are better conceptualized as representing judgments of increasing intensity along the same continuum. This is similar to the debate as to whether remember / know judgments for items are best construed as a single strength dimension, or are better understood as reflecting separate
Normalizing forgetting of childhood events 22
processes (see Macmillan & Rotello, 2006, for a review). A thorough discussion is beyond the scope of this paper, but we will make several comments. It does appear that the ABMQ model and the current results, which suggest that belief is influenced by semantically based inferential processes whereas memory ratings are not, are consistent with Tulving’s (1985) articulation of remember judgments as reflecting episodic information, and know judgments as reflecting semantic information. We also note that the evaluation of whole autobiographical events is more complex than old/new and remember/know judgments to items on lists, and thus more processes are likely involved when remembering past events. Thus it does not seem appropriate to assert that the presence of an autobiographical belief in the absence of memory is simply a memory held with lower confidence. People frequently hold autobiographical beliefs with confidence, in the absence of any memory for the event. To illustrate, consider the question “were you born?” from above. It would be surprising to find much variability for belief ratings for this event. It is also difficult to imagine the situation in which someone who believes that they were born and holds no memory for the event could be argued to hold this fact in lower confidence than does someone who holds an equal level of belief and also has a memory for being born (which does occur, for example via therapeutic techniques directed towards reliving the “birth trauma” [eg., Janov, 1983]; though the nature of this type of “memory” as being accurate is certainly debatable). Future work which distinguishes instances in which memory and belief are simultaneously present, in contrast with instances in which only belief is present in the absence of memory, are needed to better understand these complex relationships. Further work which demonstrates how false memories develop out of false beliefs is also needed. Such research might first produce beliefs in the absence of memory, after which procedures which enhance the episodic details for events are employed (eg., via imagination).
Normalizing forgetting of childhood events 23
Procedures to facilitate memory development might be borrowed from false memory studies which have successfully planted memories by using multiple imagination sessions occurring over a period of time (eg. Hyman & Pentland, 1996; Goff & Roediger, 1998), thus allowing for source monitoring errors to occur (Johnson, Hashtroudi & Lindsay, 1993). Whether plausibility and autobiographical belief are in fact distinct, or represent points on a single dimension, is a more complicated question. Prior work has discussed the two types of plausibility as a continuum (Scoboria, 2004), and other types of plausibility have been proposed (such as “cultural plausibility”; Mazzoni, Loftus & Kirsch, 2001). General and personal plausibility are operationalized on the ABMQ to represent two conceptually useful points on the plausibility dimension: perception of the plausibility of an event for the general population, and in relation to one’s self. These two judgments seem intuitively valuable in understanding the process by which an event can shift from “did not occur to me” or “don’t know if it occurred to me”, to “could have occurred to others” and “could have occurred to me.” Plausibility is thus conceptualized as an ontological judgment – to what degree do events exist in an individual’s reality, and to what degree do events exist for them personally. This is conceptually distinguished from autobiographical belief, which is the judgment as to whether an event did in fact occur. Put another way, plausibility judgments are estimates of possible occurrence, whereas belief judgments are estimates of actual occurrence. What leads individuals to confuse “could occur” with “did occur” is a very interesting question, worthy of further empirical attention. We also note that the target events use in the present study were purposely selected to be ones which were unlikely to have occurred, which we consider a strength of the design. These events also happen to be ones that are relatively unknown to participants. This is not a trivial point, as the knowledge one holds about events may matter. Though some research suggests that
Normalizing forgetting of childhood events 24
script knowledge for events is not associated with plausibility ratings (Scoboria et al, 2004), and that providing script information does not impact personal plausibility or belief (Scoboria et al, 2006), or Life Events Inventory likelihood ratings (Hart & Schooler, 2007; Pezdek, et al., 2007, Study 1), other recent work has suggested that presentation of script information within a familiar context does impact Life Events Inventory ratings (Pezdek, et al., 2007, Study 2). Certainly, a basic level of script knowledge is required in order to form judgments about events in an informed manner. The prevalence information provided in the present study did supply basic details which participants could use to evaluate events. However, it is unclear how the present findings may relate to events for which participants have a greater degree of prior knowledge, or events of greater personal significance (such as the types of events at the centre of false memory research, eg., abuse). Further research of the type described herein, using events which are more familiar or which are of greater personal significance, will produce interesting findings. Further work should examine these factors with events of lesser initial plausibility. In conclusion, these results suggest that some aspects of the false memory literature may be more appropriately understood as being about the development of false beliefs for the autobiographical past. Thus, the potential consequences of holding false autobiographical beliefs for childhood events require further examination. Furthermore, processes which lead to the development of false beliefs in the absence of false memory, the ways in which memory informs beliefs, and factors which facilitate the development of beliefs into memories, warrant further work. This is important in light of empirical results which suggest that false beliefs are easier to produce than are false memories. Hence, confident false beliefs held in the absence of memory may be a more prevalent concern.
Normalizing forgetting of childhood events 25
References Bernstein, D.M., Whittlesea, B.W.A., & Loftus, E.F. (2002). Increasing confidence in remote autobiographical memory and general knowledge: Extensions of the revelation effect. Memory & Cognition, 30(3), 432-438. Bernstein, D.M., Laney, C., Morris, E.K., & Loftus, E.F. (2005). False memories about food can lead to food avoidance. Social Cognition, 23(1), 11-34. DePrince, A.P., Allard, C.B., Oh, H., & Freyd, J.J. (2004). What's in a Name for Memory Errors? Implications and Ethical Issues Arising From the Use of the Term "False Memory" for Errors in Memory for Details. Ethics & Behavior, 14(3), 201-233. Desjardins, T., & Scoboria, A. (in press). ‘You and your best friend Suzy put Slime in Ms. Smollett’s desk’: Producing false memories with self-relevant details. Psychonomic Bulletin and Review. Garry, M., Manning, C., Loftus, E.F., & Sherman, S.J. (1996). Imagination inflation. Psychonomic Bulletin and Review, 3, 208-214. Garry, M., & Wade, K.A. (2005). Actually, a picture is worth less than 45 words: Narratives produce more false memories than photographs do. Psychonomic Bulletin and Review, 12, 359-366. Goff, L. M., & Roediger, H. L., III (1998). Imagination inflation for action events: Repeated imaginings lead to illusory recollections. Memory & Cognition, 26, 20-33. Gudjonsson, G.H. (2003). The psychology of interrogations and confessions. Chichester, UK: Wiley. Hart, R. & Schooler, J. (2006). Increasing belief in the experience of an invasive procedure that never happened: The role of plausibility and schematicity . Applied Cognitive
Normalizing forgetting of childhood events 26 Psychology, 20(5), 661-669. Heaps, C. & Nash, M. (1998). Individual differences in imagination inflation. Psychonomic Bulletin & Review, 6(2), 313-318. Hyman, I. E., Jr., & Kleinknecht, E. (1999). False childhood memories: Research, theory, and applications. In L. M. Williams & V. L. Banyard (Eds.) Trauma and memory (pp. 175188). Thousand Oaks, CA: Sage. Hyman, I.E., & Loftus, E.F. (1998). Errors in autobiographical memory. Clinical Psychology Review, 18(8), 933–947. Hyman, I.E., & Pentland, J. (1996). The role of mental imagery in the creation of false childhood memories. Journal of Memory and Language, 35, 101-117. Janov, A. (1983). Imprints: The lifelong effects of the birth experience. Coward-McCann: New York. Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 3-28. Kassin, S.M. (1997). False memories turned against the self. Psychological Inquiry, 8, 300-302. Keppel, G., & Zedeck, Z. (1989). Data analysis for research designs. W.H. Freeman and Company: New York. Lindsay, D.S., Hagen, L., Read, J.D., Wade, K.A., & Garry, M. (2004). True photographs and false memories. Psychological Science, 5, 149-154. Loftus E.F., & Benstein, D.M. (2005). Rich false memories: The royal road to success. In Healy, A.F. (Ed), Experimental cognitive psychology and its applications. Decade of behavior. (pp. 101-113). Washington, DC, US: American Psychological Association. Loftus, E.F., & Pickrell, J.E. (1995). The formation of false memories. Psychiatric Annals, 25(12),
Normalizing forgetting of childhood events 27
720-725. Macmillan, N. A., & Rotello, C. M. (2006). Deciding about decision models of remember and know judgments: A reply to Murdock. Psychological Review,113, 657-665.
Mazzoni, G.A.L., & Kirsch, I. (2002). Autobiographical memories and beliefs: A preliminary metacognitive model. In Perfect, T. J. & Schwartz, B.L. (Eds). Applied metacognition, (pp. 121-145). Cambridge University Press, New York. Mazzoni, G.A.L., Loftus, E.F., & Kirsch, I. (2001). Changing beliefs about implausible autobiographical events: A little plausibility goes a long way. Journal of Experimental Psychology: Applied, 7, 51-59. Mazzoni, G.A., Lombardi, P., Malvagia, S. & Loftus, E.F. (1999). Dream interpretation and false beliefs. Professional Psychology: Research and Practice, 30, 45-50. Mazzoni, G., Memon, A. (2003). Imagination can create false memories. Psychological Science, 14(2), 186-188. Ost, J. (2003). Seeking the middle ground in the ‘memory wars’. British Journal of Psychology, 94, 125-139. Pezdek, K., Blandon-Gitlin, I., Lam, S., Hart, R.E., & Schooler, J.W. (2006). Is knowing believing? The role of event plausibility and background knowledge in planting false beliefs about the personal past. Memory and Cognition, 34, 1628-1635. Pezdek, K. & Eddy, R.M. (2001). Imagination inflation: A statistical artifact of regression toward the mean. Memory & Cognition, 29, 707-718. Pezdek, K., Finger, K., Hodge, D. (1997). Planting false childhood memories: the role of event plausibility. Psychological Science, 8(6), 437–441. Rubin, D.C., Schrauf, R.W., & Greenberg D.L. (2003). Belief and recollection of autobiographical
Normalizing forgetting of childhood events 28 memories. Memory & Cognition, 31, 887-901. Scoboria, A., Mazzoni, G., Kirsch, I. & Jimenez, S. (2006). The effects of prevalence and script information on plausibility, belief, and memory of autobiographical events. Applied Cognitive Psychology, 20, 1049–1064. Scoboria, A., Mazzoni, G., Kirsch, I. & Relya, M. (2004). Plausibility and belief in autobiographical memory. Applied Cognitive Psychology, 18, 791-807. Sharman, S.J., Manning, C.G., Garry, M. (2005). Explain this: Explaining childhood events inflates confidence for those events. Applied Cognitive Psychology, 19, 67-74. Smeets, T., Merckelbach, H., Horselenberg, R. & Jelicic, M. (2005). Trying to recollect past events: Confidence, beliefs and memories. Clinical Psychology Review, 25(7), 917-934. Strack, F., & Bless, H. (1994). Memory for non-occurrence. Metacognitive and presuppositional strategies. Journal of Memory and Language, 33, 203-217. Tulving, E. (1985). How many memory systems are there? American Psychologist, 40, 385-398. Wade, K.A., Garry, M., Read, J.D., & Lindsay, D.S. (2002). A picture is worth a thousand lies: Using false photographs to create false childhood memories. Psychonomic Bulletin & Review, 9, 597-603.
Normalizing forgetting of childhood events 29
Author Note Address correspondence to Alan Scoboria, Department of Psychology, 282 Chrysler Hall South, 401 Sunset, Windsor, ON, N9B3P4, Canada;
[email protected]. We thank Patricia Bauer for editorial guidance, and the valuable comments of three anonymous reviewers. We also thank Sarah Jesso, Amy Kossert and Amedina Simone for assistance with data collection.
Normalizing forgetting of childhood events 30
Table 1. Study 1, Mean Time 1 and Time 2 Autobiographical Belief and Memory Questionnaire scores by treatment group. General plausibility Mean
SD
Personal plausibility Mean
SD
Autobiographical belief Mean
SD
Memory Mean
SD
Prevalence + Forgetting rationale Time 1
4.67
0.46
2.39
0.39
1.33
0.14
1.11
0.08
Time 2
5.89
0.43
4.89
0.54
3.50
0.51
1.22
0.17
Time 1
4.94
0.43
2.94
0.53
1.65
0.34
1.18
0.18
Time 2
6.18
0.43
4.18
0.56
1.76
0.29
1.06
0.06
Time 1
5.00
0.44
3.00
0.48
2.11
0.38
1.21
0.21
Time 2
4.74
0.51
2.58
0.41
1.42
0.23
1.00
0.00
Prevalence
Control
1-8 point scale.
Normalizing forgetting of childhood events 31
Table 2. Study 2, Mean Time 1 Autobiographical Belief and Memory Questionnaire scores by event. General plausibility Mean
SD
Personal plausibility Mean
SD
Autobiographical belief Mean
SD
Memory Mean
SD
Density
5.09
2.26
3.18
2.23
1.82
1.28
1.07
0.34
Skin
4.99
2.15
3.41
2.31
2.23
1.57
1.21
0.63
1-8 point scale.
Normalizing forgetting of childhood events 32
Table 3. Study 2, Mean Time 1 and Time 2 Autobiographical Belief and Memory Questionnaire scores by treatment group. General plausibility Mean
SD
Personal plausibility Mean
SD
Autobiographical belief Mean
SD
Memory Mean
SD
Prevalence + Forgetting rationale Time 1
5.20
1.93
3.49
2.46
1.95
1.44
1.08
0.34
Time 2
6.78
1.30
5.53
2.36
3.56
2.18
1.15
0.48
Time 1
4.83
2.28
2.90
2.06
1.86
1.09
1.08
0.27
Time 2
6.06
1.74
4.16
2.27
2.73
1.80
1.22
0.73
Time 1
4.85
2.10
3.43
2.21
2.35
1.54
1.21
0.54
Time 2
5.24
2.00
3.90
2.09
2.19
1.44
1.12
0.22
Prevalence
Control
1-8 point scale.
Normalizing forgetting of childhood events 33
Figure Captions Figure 1. Study 1, Change in Autobiographical Belief and Memory Questionnaire variables by group, adjusted for baseline responses. Figure 2. Study 2, Change in Autobiographical Belief and Memory Questionnaire variables by group, adjusted for baseline responses.
Normalizing forgetting of childhood events 34
Figure 1. General plausibility
Personal plausibility
Belief
Memory
Baseline adjusted change score
3.00 2.50 2.00 1.50 1.00 0.50 0.00 -0.50 -1.00
Prevalence + Rationale
Prevalence
Treatment group
Error bars show standard errors.
Control
Normalizing forgetting of childhood events 35
Figure 2. General plausibility
Personal plausibility
Belief
Memory
Baseline adjusted change score
3.00 2.50 2.00 1.50 1.00 0.50 0.00 -0.50 -1.00
Prevalence + Rationale
Prevalence
Treatment group
Error bars show standard errors.
Control