Investigating the Memory Reports of Retractors Regarding Abuse

Introduction

Katrina Fairlie, a 26-year-old young woman, was encouraged by doctors to believe that her father, Jim Fairlie (once a prominent politician and deputy leader of the Scottish Nationalist Party) had abused her as a child. She claimed that her father had not only horribly abused her, but also led a 17-strong pedophile ring. In June 1994, Katrina was admitted to Perth Royal Infirmary due to experiencing severe abdominal pains and sickness. After two operations, her condition did not improve. At the beginning of 1995, she was admitted to the Perth Royal’s psychiatric clinic. Her doctors suspected that she had been sexually abused after which she was sent to recovered memory therapy. During the therapy sessions, Katrina began to recover an increasingly graphic series of memories of her father abusing her almost daily and she was persuaded to make an official statement to the police. Four years later, she said those claims were false and a police investigation revealed no evidence of abuse.1

The case of Katrina illustrates what in the psychological literature has become known as a “retractor”, which can be defined as a person who has repudiated his/her earlier claims of being (sexually) abused (Ost, 2017). The existence of such retractors – while assuming that their retractions are valid – would demonstrate that adults could be induced to falsely believe and remember that they have been exposed to an aversive experience such as being sexually abused during childhood.

Memories of Trauma

In the 1990s, psychologists, therapists, and legal professionals had heated debates about the authenticity of memories recovered in therapy (Howe & Knott, 2015). This debate has been termed the memory wars (Crews, 1995; Otgaar et al., 2019; Patihis et al., 2014). On one side of the debate were oftentimes clinicians who asserted that recovered memories referred to authentic experiences. That is, they argued that these memories were formerly repressed memories of trauma and were stored in an almost inaccessible and unconscious way (Loftus & Davis, 2006; Patihis et al., 2021) and that therapeutic interventions (such as hypnosis) could be deployed to recover such memories (Lindsay & Briere, 1997; Pope & Brown, 1995; Stocks, 1998). On the other side of the debate, memory scholars expressed doubts about the authenticity of these memories. They reasoned that many of these therapeutic interventions were suggestive, possibly leading to the creation of false memories of abuse (i.e., memories of an event that in fact the individual did not experience; Loftus & Davis, 2006).False memories can be evoked spontaneously or due to suggestion (Otgaar et al., 2017). Spontaneous false memories are the result of internal memory mechanisms and arise without any external pressure, such as elicited in the Deese/Roediger-McDermott paradigm (DRM; Deese, 1959; Pardilla-Delgado & Payne, 2017; Roediger & McDermott, 1995). In this paradigm, participants learn word lists (e.g., women, uncle, wife, male, etc.). In subsequent memory tests, they often mistakenly recall or recognize a non-presented but related word (e.g., man) that does not appear in the list but shares a strong associative relationship with the word list. Suggestion-induced false memories occur due to suggestive pressure, which can be illustrated by false memory implantation method (Loftus, 2005; Otgaar et al., 2012; Otgaar, & Howe, 2018). In this paradigm, participants receive false suggestions that they experienced a fabricated event. Such suggestions can lead to about 30% of participants creating false autobiographical memories (Arce et al., 2023; see also Scoboria et al., 2017).

Recovered memories are also divided into two types of memory recovery experiences: one in which memories are recovered in the context of suggestive therapy and one in which spontaneous recovery occurs (Schooler et al., 1997). Based on findings from false memory implantation studies, researchers postulated that such suggested recovered memories are in fact false memories (Chu et al., 1999; Hyman & Pentland, 1996). Importantly, these suggestive false recovered memories may lead to legal proceedings and in its wake false accusations and unjust convictions (Geraerts et al., 2005; Loftus & Davis, 2006). However, there are also examples of people (like in the above-described case of Katrina) who began to doubt the accuracy of their recovered memories of sexual abuse and eventually retracted them. Some retractors even sued their therapist(s) for falsely suggesting abuse to them (Loftus & Davis, 2006; Lynn, McNally, et al., 2023).

Retractors of Sexual Abuse

Ost et al. (2002) noted that the recovery and withdrawal processes of retractors’ traumatic memories are asymmetric. That is, the time for them to recover the memories of abuse (about 4.5 years) was longer than the time to retract them (about 8.6 weeks). At the same time, when retracting the memories, retractors more often experienced strong social pressure as compared to when they recovered the memories (respectively 84.5% versus 15.5%). The pressure to retract can be attributed to group pressure (e.g., family members, friends, survivors’ group etc.) or to the therapist persuading them to do so. Some retractors indicated that their memories lacked familiarity and context, some began to suspect their memories because they were too clear or too detailed, and some retractors’ memories depended on specific contexts, such as treatment environment (Ost, 2017).

Limited research exists on the characteristics of the memories of retractors. For example, when remembering an event, it is relevant to distinguish between the processes of believing and remembering an event. That is to say, memory is not only the re-experience or recollection of a past experience, but also accompanied by the belief that this experience actually happened. The basic-systems model of episodic memory proposes that recollection and belief are two separate phenomena (Rubin, 2005, 2006). Recollection refers to the mental experience of reliving the event (Scoboria et al., 2014) and is predicted by perceptual, re-experiencing, and emotional characteristics (Rubin et al., 2003; Rubin & Siegler, 2004). Belief, however, refers to appraisals of the event (Otgaar et al., 2018) and is predicted by narrative and contextual characteristics (Rubin et al., 2003; Rubin & Siegler, 2004).

The Nested Model (Scoboria et al., 2004) postulates that before someone can develop a (false) memory of an event, they first need to (falsely) believe that this event has really occurred and that the event is plausible in terms of happening (both generally and personally). However, this is not always the case with memory. A first example relates to childhood amnesia, which refers to the phenomenon that adults have little to no recollection of their infancy and early childhood (Hayne & Jack, 2011). Typically, adults do report some believed-not-remembered events implying that people can believe something has happened to them without having a clear recollection (Scoboria et al., 2004). A second example is that people can have a vivid recollection of an event while they do not believe in the occurrence of the event. This counterintuitive phenomenon is also termed nonbelieved memories, and they are more common than one would expect (Otgaar et al., 2014). To illustrate, in one study, approximately 20% of the 1,593 surveyed people reported having experienced a nonbelieved memory (Mazzoni et al., 2010).

At present, to our knowledge, no research exists that has provided a close look at memory reports of retractors and examined whether they contain any signs of belief and/or recollection of abuse. Examining the memory reports of retractors is in line with recent research recommendations to conduct systematic studies of “retractors” (Lynn, Aksen, et al., 2023, p. 32). Multiple variations can be found regarding the level of belief and recollection when examining retractors’ accounts of traumatic experiences. Although Ost (2017) did not specifically ask retractors about their beliefs and/or recollections concerning the abuse, he did look at former accounts given by retractors and examined himself whether they believed and/or remembered the abuse. Ost found that some retractors seemingly had real visions of the event, thus experienced both strong belief and vivid recollection, despite the bizarre, implausible nature of the memory (Pendergrast, 1995). There are also retractors who neither believe nor have a recollection of abuse-related memories. Their memories of abuse can best be described as an idea they were “playing” with, rather than something they genuinely believed or remembered (de Rivera, 1997). Furthermore, some retractors do not have any recollection of memories of abuse but do have belief in these memories. For example, one retractor was only certain about the age the alleged abuse occurred, but never had any specific memories regarding the spatial and temporal context (Pendergrast, 1995). Finally, some retractors withdrew their belief in the memory of abuse, even though they still have vivid recollections of the negative experience (Ost, 2017), implying that they formed nonbelieved memories. Based on Ost’s (2017) work, and in conjunction with previous research on beliefs and recollections, we further explored the pre- and post-withdrawal beliefs and recollections of the abuse for this particular sample (retractors).

An important question related to the issue of retractors is why people withdraw their belief in memories of events that happened in the past. Scoboria et al. (2015) mentioned eight reasons for withdrawing belief in autobiographical memories: social feedback (i.e., being told by others that the event did not happen), event plausibility (i.e., realizing that the event is impossible, meaningless, or it is inconsistent with actual scientific knowledge), alternative attributions (i.e., the memory may have resulted from a dream or nightmare), general memory beliefs (i.e., the belief that childhood memories cannot be retained when being very young or are unreliable), internal event features (i.e., some feature of the internal representation leading people to question the memory, like feelings that the memory was odd or unusual), consistency with external evidence (i.e., “If I injured myself, I would have a scar”), views of self and others (i.e., “I am a neat freak so there is no way”), and personal motivation (i.e., “I convinced myself it wasn’t real”). Of all these reasons, social feedback (42.2%, n = 158), and event plausibility (19.5%, n = 73) were found to be the main motives for decreasing people’s beliefs.

Ost (2017) noted that these reasons were also relevant for the withdrawal of beliefs in retractors. Different from what had been found in the Scoboria et al. (2015) study, inconsistency with external evidence (27.2%, n = 22) appeared to be the main reason why people withdraw their beliefs in memories of sexual abuse. However, retractors seldom took the initiative to look for evidence, and more often, they found the evidence by accident. For example, after media were emphasizing misremembering information about sexual abuse, some retractors began to think more critically. This finding matches the passive role in memory verification strategies (Scoboria et al., 2004). Furthermore, Ost found that social feedback was also an important category (19.6%, n = 22). In most cases, this kind of feedback was provided by a therapist, but also other people who directly told them that their memories of the events might be inaccurate.

The Current Study

Research regarding retractors’ experiences is limited. Furthermore, some crucial gaps can be detected in this line of research. First, past research has essentially focused on retractors’ experiences of recovering their abuse. Studies that did examine the actual retraction itself were limited to specific circumstances (e.g., treatment settings) and the reasons that initially led to the withdrawals (Chu et al., 1999; Davis, 2000; Goodyear-Smith et al., 1997). Second, previous research did not investigate to what extent retractors still believe in or have recollections of the child sexual abuse experiences (Ost, 2017). In other words: past researchers have not specifically asked retractors themselves about their belief and recollection concerning the experience.

Hence, the current research tested a new group of retractors and specifically examined their belief and recollection before and after retraction. We aimed to establish to what extent their responses could be labeled as nonbelieved memories. In addition, we also investigated the reasons for the withdrawal of the belief, the outcomes of recovered and retracted memories, and other details about the memory of the abuse. Given that the retractors have withdrawn their beliefs about having been abused in memory, we expected that these belief scores will decrease after withdrawal, and that a substantial proportion of retractors may exhibit nonbelieved memories. That is, a significant amount of the retractors will still have recollections about these traumatic memories, although they no longer believe in the authenticity of these memories. Moreover, based on previous findings of individual differences (Ost, 2017), we also predicted that multiple combinations of belief and recollection will be reported. For example, recollection scores will be greater than or equal to belief scores.

Method

Participants

Participants were selected via a Facebook group called ‘False Memory Syndrome Action Network’ (FMSAN). This group is – amongst others – composed of individuals who have undergone therapeutic interventions leading to false memories of abuse and later retracted their earlier claims of abuse. The participants from the Facebook group also contacted other retractors who were willing to contribute. A total of 63 participants completed the questionnaire, of which 7 participants were excluded because they completed the questionnaire within a very short period of time (less than 5 minutes), questioning the validity of their responses. Therefore, we used the data of 56 self-reported retractors (M_age = 38.30, SD = 15.20, range 16-72; 41 females, 13 males, 2 participants did not report their gender). With regard to marital status, most participants reported to be single (n = 25); others were married (n = 14), divorced (n = 8), living with partner (n = 7), and engaged (n = 2). In terms of educational levels, the majority had finished high school (n = 15) or a bachelor’s degree (n = 24); other participants had master’s degrees (n = 8), doctoral degrees (n = 4) or received junior high school education (n = 4). Just one participant finished elementary school. Some participants completed all the questionnaire items, but some participants only completed part of the questionnaire. The specific number of people who completed the specific parts of the questionnaire will be described in detail in the Results section.

Materials

Based on the survey used by Ost et al. (2017), a questionnaire containing 120 questions divided into seven sections was developed. The first section of the questionnaire involved demographic details of the participants, such as questions on gender, marital status, and education level. Section two was about how memories were recovered. For example, participants were asked to mention significant events that led them to recover (or suspect) memories of childhood sexual abuse. The third section included questions about the details of the therapy/counselling. Examples included information about the participant’s motivation for receiving therapy, the year they started therapy, and the duration and frequency of therapy. Section four contained 33 questions on the details of the recovered memories before the retraction. For example, questions were asked concerning the timing of the recovery, the quality of the memory, and the participants’ beliefs, recollections, and rationalizations about the event. The fifth section involved questions about the outcome of recovered memories, for example the consequences for the relationship with the alleged abuser or handling reactions of family and friends. Section six contained 33 questions on the occurrence of the retraction. For example, questions were included on the duration of the retraction, the reason(s) for withdrawing belief, and any pressure experienced to retract. Lastly, section seven was about the outcomes of retracting the recovered memories. More specifically, questions were asked about the costs and benefits of the retraction, such as possible reconciliation with the alleged abuser(s), relationships with family/friends, and legal actions.

Procedure

Most participants completed the entire questionnaire online using Qualtrics. One participant completed the questionnaire together with one of the researchers during an online meeting. Participants completed the questionnaire after providing informed consent. Completing all the seven sections of the study took more than one hour. Once the participant finished the questionnaire, they were thanked for their participation and were offered help, should they be experiencing any distress or negative emotions. They also received a link guiding them to further information on the research. The study was approved by the Social and Societal Ethics Committee (SMEC) and Privacy and Ethics (PRET), Catholic University of Leuven. Data and materials are available at https://osf.io/euvpy/. The data was processed in accordance with the Belgian personal data protection rules (Data Protection Authority, 2018; European Council, 2016).

Results

The Time to Believe and Retract the Abuse-related Memories

With regard to how long it took to initially become convinced the memories were valid, six of the 24 participants who completed this question noted that they reached this point right away (e.g., in the first session with their therapist). Six participants gave inconclusive answers, for example, one could not give a time indication, whereas another one mentioned that it took “weeks to months”. The shortest time mentioned was one day, while the longest was nine years, which means that the time ranged from one day to several years. Descriptive statistics on the time were only calculated using the exact estimates and are expressed in days (M = 1,134.85, SD = 2,576.4).

When looking at the time it took the participants to become convinced that the memories were not true, the shortest time someone reported was two months. However, this was an extremely low value, as all other participants expressed their response in terms of years. Eight participants could not give an exact number (i.e., “unsure” or “years”). The time thus ranged from two months to 24 years. Vague answers were excluded, and the average time was expressed in days (M = 1,689.09, SD = 2,450.5). One participant even stated that “I’ve had times that I had a hard time believing, but when I really looked at it in context, with a timeline, I was all of a sudden convinced it couldn’t be true.” A Wilcoxon matched pairs test was then conducted to compare the time taken to become convinced the recovered memories were true and to become convinced they were not true. Again, only the answers with an exact estimate were included in the test, which means that this test did not include all participants (n = 24). The statistical analysis (Z = -2.20, df = 10, p = .028, Cramer’s V = 2.24) indicated that the time taken for these participants to recover abuse memories were significantly shorter than the time before they became convinced the memories were not true.

Twenty-four participants completed the question regarding the age at which they stopped believing in the memory of the abuse. The reported ages ranged from 3 to 33 years (M = 12.74, SD = 8.98). Three participants were unsure when the belief in the memory of abuse ended. Three participants who were relatively old believed in the alleged abuse for a very long time, respectively 25, 31, and 33 years. Two participants even thought that the belief in the memory of the abuse never completely ended. In contrast, the shortest belief duration was as long as one year. The average duration of believing they had been abused was 10.28 years (SD = 8.66).

The Recovered Memories

Thirty-three (58.93%, 95% CI [46.0%, 71.8%]) participants considered their recovery of abuse-related memories to be caused during therapy. Twenty-four of the 56 participants completed section two on the reason(s) why people recovered their memories (Figure 1). Note that for this part, participants could indicate multiple response options. “Flashbacks or images in your mind” (n = 17, 70.83%, 95% CI [52.6%, 0.89%]) and “feelings that something awful happened” (n = 9, 37.5%, 95% CI [18.1%, 0.56.9%]) were the two main reasons why people began to recover memories of abuse.

The remembered categories of abuse included sexual, physical, emotional abuse, neglect, and even satanic ritual abuse. Thirty-nine participants completed this part. For this question, participants could provide multiple responses. Sexual abuse was the most frequently endorsed category (n = 35, 89.74%, 95% CI [80.2%, 99.3%]; see Figure 2). A substantial proportion of participants (n = 22, 56.41%, 95% CI [40.8%, 72%]) reported to remember more than one type of abuse.

Twenty-two participants provided specific content of the recovered memories. Similarly, participants could endorse multiple options. Not surprisingly, given that the sexual abuse category was most common, the memory content was often concerned with fondling (n = 16, 72.73%, 95% CI [54.1%, 91.3%]), touching (n = 16, 72.73%, 95% CI [54.1%, 91.3%]), followed by intercourse (63.64%, 95% CI [43.5%, 83.7%]), multiple perpetrators (n = 12, 54.55%, 95% CI [54.1%, 91.3%]), and kissing (n = 8, 36.37%, 95% CI [33.7%, 75.4%]). Other content was mentioned as well (n = 10, 45.46%, 95% CI [24.6%, 66.3%]), and included ritual abuse, bestiality, aliens, and church abuse. One participant gave more insight into the memory and remembered being photographed and that the father had threatened to set the participant and her sister on fire.

Twenty-five participants elaborated on the people that they accused of the abuse. Participants could endorse multiple response options for this question (Figure 3). The accused persons were family members, and most frequently the parents. There was no notable difference between father (n = 15, 60%, 95% CI [40.8%, 79.2%]) and mother (n = 10, 40%, 95% CI [20.8%, 59.2%]) as the alleged abuser. Another 16 participants (64%, 95% CI [45.2%, 82.8%]) chose “Others”, and identified grandparents, neighbors, or strangers as the abusers.

The Retracted Memories

Seventeen participants indicated one or more reasons for why they retracted their memories. Participants provided multiple responses according to their own situation (Figure 4). External evidence (n = 8, 47.06%, 95% CI [23.3%, 70.8%]) was most prominent. Three participants said that reading a magazine article was the main reason why they realized that their memories may not be true and one of them stated: “I realized my draw toward the narrative, its implausibility, the process of splitting that allowed me to idealize my therapist and demonize my parents, the secondary gain from being a victim, the convenience and absolution of the blame for my own issues that the victim narrative provided, the power of suggestion, and the negative effects on my life.”

Recollection and Belief before and after Retraction

Twenty-three participants completed ratings of beliefs and recollections before and after retraction of memories. A paired-samples t-test was conducted to compare the memory reports before and after retraction (see Table 1). The belief score2 before retraction significantly decreased after the retraction, t(20) = 19.33, p < .001, M_diff = 7.94, 95% CI [7.08, 8.79], Hedges’ g = 2.60. The recollection score before retraction was slightly higher than the score after retraction, but this difference was not statistically significant, t(22) = 0.89, p = .382, M_diff = 0.49, 95% CI [-0.65, 1.64], Hedges’ g = 0.14. Hence, participants still had some memories of the event in question, but they did no longer strongly believed them. The fact that the average belief score was higher than the recollection score before the retraction but lower after the retraction is an important finding in the context of nonbelieved memories. When the recollection score is at least one point higher than the belief score, the memory can be categorized as a nonbelieved memory (Scoboria et al., 2004). Using this criterion, we calculated the proportion of participants that formed nonbelieved memories, which was the case in 30.43% (n = 7, 95% CI [11.6%, 49.2%]) of the participants. There were 43.48% (n = 10, 95% CI [23.2%, 63.7%]) of participants who had higher belief than recollection scores after the retraction, and another 39.13% (n = 9, 95% CI [19.2%, 59.1%]) who reported equal belief and recollection scores.

The personal plausibility score before the retraction was statistical significantly lower after retraction, t(22) = 3.89, p = .001, M_diff = 1.65, 95% CI [0.77, 2.53], Hedges’ g = 0.51, which indicates that participants considered the abuse as less plausible to have occurred to them. For scoring the memory characteristics, we looked at visual details, sound, vividness, and event details. The memory characteristics score was significantly higher before the retraction as compared to after retraction, t(22) = 2.40, p = .025, M_diff = 0.94, 95% CI [0.13, 1.74], Hedges’ g = 0.35.

Outcomes before and after Retraction

Twenty-three participants elaborated on the outcome of the retraction. Before retraction, almost all participants claimed that there were some costs or negative outcomes for them by believing that they had been abused. The most important ones were feeling depressed and suffering from a range of psychological problems (n = 16, 69.57%, 95% CI [50.8%, 88.4%]) and losing their job and social relationships with other people (n = 9, 39.13%, 95% CI [19.2%, 59.1%]). Some of them (n = 14, 60.87%, 95% CI [40.9%, 80.8%]) also mentioned benefits, for example, that their psychological and health problems seemed to make sense.

After retraction, 43.48% (n = 10, 95% CI [23.2%, 63.7%]) of the participants felt that the largest cost was the loss of friends and family. Five participants noted that they did no longer believe in psychotherapy. Compared with before retraction, more participants (73.91%, 95% CI [56.0%, 91.9%], n = 17) believed that withdrawing the memories about the abuse had some benefits, for example, giving them a new chance to try to build their lives up from the ground, establishing new and intimate relationships with family and friends, and no longer wrongly blaming another person.

Participants were also asked whether they confronted their alleged abuser at the time of recovering the memories about abuse and/or reconciliated with that person after the retraction. There was only a minority (n = 4, 17.39%, 95% CI [1.9%, 32.9%]) who did this before the retraction, while the majority (n = 17, 73.91%, 95% CI [56.0%, 91.9%]) reconciliated after the retraction (face-to-face or over the phone). Following confrontation, the alleged abusers of the two participants reacted by denying the accusation, which at that time only served as more evidence in favor of the memories. One of the participants wrote a letter to the alleged accuser and obtained forgiveness. The other participants did not seek confrontation to avoid drama or harm (e.g., “I was warned that my abusers would kill me”), or because the person in question could not be reached and it felt like a bad idea. The reconciliation on the other hand involved some feelings of guilt and embarrassment, but was overall a positive experience, as they could apologize and in most cases restore the relationship. All of them were still satisfied about doing this, even though some of the relationships were never truly restored. The ones who chose not to reconciliate did this because there never really was a solid relationship.

Furthermore, the participants were asked whether they undertook any legal action against their alleged abuser at the time of recovering their memories or against their therapist after the retraction. Six participants (26.09%, 95% CI [8.1%, 44.0%]) took legal action against their alleged abuser. Other participants were unwilling to take legal action against their alleged abusers, because their therapist told them that they did not have any other evidence except their memories to prove that they had been abused and the police would not believe them anyway. One participant who testified against his alleged abuser succeeded in court. The accused individual received a 20-year prison sentence and was registered as a sex offender. The participant had mixed emotions doing this, as this person was his father, but at the same time other people were praiseful about this action. At present, he still feels terrible about it. After retraction four participants sued their therapist. They were glad to undertake this legal action, even though the process made them nervous and distressed. They indicated that they were not just standing up for themselves but also for other people in the same position. The others faced some struggles; they explained that too much time had passed, the therapist had died, or the type of therapy was no longer being practiced.

Discussion

The debate on the authenticity of memories recovered in therapy lingers on (i.e., Lynn, McNally, et al., 2023; Otgaar et al., 2019; Patihis et al., 2014; Patihis et al., 2023; Patihis et al., 2019; Pope et al., 2022). This observation underscores the importance of examining people who have retracted memories that were once recovered. The main purpose of this study was to examine memory reports of retractors of abuse. More specifically, we examined details, plausibility, beliefs, and recollections of the abuse before and after the retraction, the reasons for the withdrawal of the belief, and the outcomes of recovered and retracted memories.

First, our results showed that the average duration of participants having a memory about the abuse was about 12 years and the time that it took to recover the memory of the abuse was quite long and ranged from 5 to 56 years. The most common alleged abuser was the father or the mother, which was also in line with results from previous studies (Andrews et al., 1999; Goodyear-Smith et al., 1997; Gudjonsson, 1997; Lief & Fetkewicz, 1995; Van Koppen & Merckelbach, 1999).

As expected, the role of therapy was also very apparent in the retractors’ experiences (see also Ost et al., 2002). Our study found that therapy was the most important way for recovering memories. A substantial proportion of the 56 retractors (59%) indicated that they started to remember the abuse during the course of psychological treatment. This is consistent with the findings of Ost et al. (2002), who noted that 80% (n = 36) of their participants recovered their memory during or after treatment. The techniques used in certain forms of therapy are sometimes highly suggestive (Loftus & Davis, 2006). According to the Source Monitoring Framework (Johnson et al., 1993), which refers to the cognitive processes of attributing the source of memory to a mental experience, people can easily misattribute a mental experience as a memory. The chance that this happens is higher when suggestion is involved (Belli, 2012). Due to external suggestive pressure, such as misinformation, implantation, or memory integration (Otgaar & Howe, 2018), suggestion-induced false memories can appear during therapy.

Seventeen participants in the present study mentioned the feeling of terrible things happening, flashbacks or images, and reading a book as the main reasons for recovering memories. This is also consistent with the findings of the study by Ost et al. (2002). In addition, this study also found that the therapies or other people were the main reason for recovering memories. Sexual abuse was the most frequently mentioned type of abuse. Dodier and Patihis (2021) and Patihis and Pendergrast (2019) also found that the most commonly reported type of recovered memories of child abuse was sexual in nature.

More than half of the participants indicated that they had faced multiple abusive events, which seems to be consistent with other studies. Lief and Fetkewicz (1995), for example, even reported that 95% (n = 95) of the participants in their study remembered more than one type of abuse. The participants took significantly longer to retract the memories (1,689.09 days, range between 2 months and 24 years) than to recover them (1,135 days, range between one day and 9 years). Ost et al. (2002) reported an equal average time of 9 weeks and a range between one day and 24 weeks to recover. These researchers reported a somewhat shorter time period for their participants to retract, with an average time of 5 years and a range between one week to 12 years. Some participants seemed to be unable to give accurate estimates.

With regard to the reasons why people retracted their memories, external evidence appeared as the main reason. Participants realized that their memories were likely wrong after reading some magazine articles and looked for other relevant external evidence, such as photos of abuse sites to support their ideas. The latter finding is in contrast with the research of Scoboria et al. (2015), who noted that social feedback was the main reason for participants to withdraw their belief in memory. In the current research, social feedback was less important as a reason for why participants withdrew their belief in abusive memories. This may be due to the fact that of the 17 people who specified the reason for retraction recovered these traumatic memories during their treatment session. They indicated that the therapist used suggestive interventions to make the participant produce their false memories of abuse. As such, participants might be more suspicious about social feedback, especially when given by therapists. Subsequent investigation of the outcomes of the retracted memory also showed that the participants’ trust in others had also decreased. Memories of abuse created due to psychotherapy left them traumatized for many years, and some of them even sued their therapist for this reason. Looking for some seemingly reliable external evidence (such as photographs or reports of physical injuries) seemed more reliable to them.

In line with previous research (Ost, 2017), memories change in terms of belief and recollection during the process of retraction. More precisely, following the retraction, significant decreases in belief scores were documented, whereas the recollection score did not change significantly. This means that after the retractors withdrew the abuse-related memory, they were inclined to believe less in this memory, although they still retained a vivid recollection of what they initially thought had happened. This is in line with research on nonbelieved memories: people may have vivid autobiographical memories of an event but stop believing or withdraw the belief for that event (Mazzoni et al., 2010; Otgaar et al., 2014), and this appeared to be the case in 43% of the participants in the present study. There were even 30% (n = 23) who developed nonbelieved memories, which is in line with Ost’s (2017) research on retractors.

The personal plausibility score before the retraction was significantly higher than after retraction, which means that the abuse is thought of as less likely to have occurred to them. These results mimic the above-described findings on the belief of the retracted memory and are consistent with the metacognitive model of memory (Mazzoni & Kirsch, 2002) and the Nested Model (Scoboria et al., 2004) in which lower beliefs about memory events are accompanied by lower plausibility scores. We also found a significant decrease in memory characteristics scores before and after withdrawal. Although this could be the result of the retraction itself, it could also be due to the gradual fading of memory clarity over time (Vallet et al., 2016).

Before the withdrawal of memories, many participants (70%) felt that their downside of the recovered sexual abuse memory was becoming depressed and suffering from a range of psychological problems. After withdrawal, the vast majority of participants (74%) indicated that withdrawing abuse-related memories had some clear benefits, for example, giving them a new chance to re-build their lives. It can be seen that withdrawing the memory of abuse can make more people courageously face the previous harm and start a new life. This may be a good phenomenon. Moreover, our findings also showed that before retraction, two participants took legal action against their alleged abuser, while most participants were unwilling to do so. Hence, memories of victims of abuse should be treated cautiously, because if retractors believed the (false) memory of their abuse during the recovery and prosecute the alleged abuser, then it means that innocent people can be sent to prison.

The current study is subject to a number of limitations. The first shortcoming has to do with the quite limited sample size of 56 participants, although it should be kept in mind that retractors represent a very special population which are difficult to recruit). As is typically the case with small samples, the statistical power to detect true findings may be reduced and so it remains unclear to what extent the results are representative for all retractors. Another important limitation pertains to the fact that the present findings are only based on the personal and subjective accounts of the retractors. Although the veracity of their claims was never intended to be questioned, it is still true that they are reconstructing their experiences, and for some participants the presumed events took place many years ago. A final drawback relates to the method, namely the use of an online questionnaire. It is possible that some questions were not fully understood in the way they were intended to be. Further, as the nature of the study concerns a highly sensitive topic, it may well be that some responses were reluctant to share the experiences and so their responses may not have been as in-depth as they could have been. However, the participants were given the opportunity to complete the questionnaire in the presence of another person to help them through the process, and one of the participants did in fact choose to do so.

Despite these shortcomings, our study also offers some new and crucial insights. Our findings of retractors’ accounts of withdrawing belief in their recollections provide evidence for the existence of nonbelieved memories for traumatic and personally relevant autobiographical events. We made a comparison for retractors’ beliefs, recollections, and personal plausibility scores before and after challenge. A significant reduction was found for the beliefs and personal plausibility after retraction, which adds new evidence of nonbelieved memories of trauma in a ‘naturalistic’ setting.

In summary, this study investigated the memory reports of retractors of abuse. The central question was whether there were notable differences in these reports before and after retraction. Compared with the time it took to withdraw the memory, retractors needed a considerable time to come to the conclusion that their childhood abuse memories were false. Furthermore, belief and personal plausibility scores significantly decreased following retraction. The process of retraction occurred gradually and was more of a personal choice once they realized that the recovered memories were not true. In most cases, the decision to retract was beneficial. It took retractors multiple years to withdraw their memory, which means that innocent people could be falsely accused for years. Therefore, the authenticity of retractors’ memories is particularly important. More extensive research is needed with other retractor samples, which could possibly focus on other important topics such as the reliability of retractors’ memories.