Autobiographical memory is a constructive process that concerns both retrospective memory and prospective memory. Previous autobiographical events allow one to imagine and predict future experiences (Schacter and Addis, 2007). During the narration of a memory the spatial and visual information of what they experienced is reactivated, processed and converted into a verbal narrative. In the process of retrieving this information various cognitive functions are involved which allow the activated visual images to be faithful to the original memory, and expressed in a clear, accurate and complete way (Rubin et al., 2003).

The model developed by Conway et al. (2004) and Conway (2009) distinguishes between working self, autobiographical memory, and episodic memory. The working self consists of the conceptual self (Conway et al., 2004) and the goal system. It is the working self that determines which knowledge derived from experience is encoded in autobiographical memory. The authors highlighted that encoding lasts for a limited time and includes sensorial-perceptual information, affects, thoughts, imaginations (Williams et al., 2008). Autobiographical memory contains autobiographical knowledge, the history of our times and episodic memories that derive from lived experience (Conway and Pleydell-Pearce, 2000; Conway, 2005). In the model of Conway (2009), autobiographical memory consists of retrospective autobiographical memory, prospective autobiographical memory, and semantic autobiographical memory.

Autobiographical episodic memory is a constructive process that allows people to flexibly extract and recombine elements of the past. A similar reconstructive process concerns prospective or future memory where previous knowledge and experiences allow us to simulate future ones (Schacter and Addis, 2007). Neuroimaging studies have highlighted how the coding and learning phases of new events affect brain areas different from those involved in the re-enactment phases. Several studies have also highlighted how it is not possible to distinguish the brain areas involved in the recall of a real memory from the production of a fantastic memory (Schacter and Addis, 2007; Paz-Alonso et al., 2008; Van de Ven et al., 2018). As suggested by several studies, the activity of recovering a real memory or imagining a future memory involves the same brain areas. This could explain how at a cognitive level there are no differences between retrospective memory and prospective memory. Indeed, the recall of past events and the imagination of future events involve the same cognitive abilities (Fan et al., 2022). In fact, prospective memory is based on personal events that the people have already experienced, and this makes planning and forecasting what will happen easier and more consistent with one’s sense of identity. According to Nelson (1986), autobiographical memory also has the function of facilitating the prediction of future events, allowing to better anticipate what will happen in the face of recurring events and facilitating defense responses.

According to Wood and Conway (2006), the emotional component also plays an important role in the encoding and reactivation of autobiographical memories in terms of intensity. The positive or negative valence can allow the preservation of vividness, a greater retention over time of the memory and the quality of the narrative structure of the memory.

The ability to tell an autobiographical event is given by the narrative skills that increase with age and allow a description of when, how, where and what happened (Nelson and Fivush, 2004). Indeed, the main narrative skills allow you to give space–time information (Nelson and Fivush, 2004), the description of what happened, who was involved and how the episode occurred (Rubin et al., 2003; Wood and Conway, 2006). According to several studies, narrative skills increase with age and with the development of intellectual and expressive abilities (Nelson and Fivush, 2004; Orbach and Lamb, 2007; Kulkofsky and Klemfuss, 2008). With age, the capacity for retrospective mnemic activation increases in children, even after a long time, which allows them to recall and associate information and provide a story that is even richer in detail (Lamb et al., 2018).

Several studies shown that even children under six had the ability to reliably remember some autobiographical events, even after 1 year (Hudson and Fivush, 1991), but it is necessary to provide a cue to retrieve information. According to Nieto et al. (2017), after the age of 4,5, preschool children show an ability to recall personal events through the cueing. In any case, the accounts of younger children tend to be more fragmented than older children (Newcombe et al., 2007). The youngest children can tell what happened, but the ability to place the event more accurately in time develops after age 10 and becomes complete and adult-like after age 14 (Hudson and Mayhew, 2009; Roberts et al., 2015; Jack et al., 2016).

Narrating repeated events can promote confidence in terms of narrative skills, to the detriment of details relating to individual episodes if they present similar characteristics. The narration of specific events can be associated with more vivid and detailed descriptions, while common events tend to be narrated for the most central elements due to a certain rigidity of the mnemic scheme (Wang et al., 2011; Roberts et al., 2015).

The knowledge and semantic understanding of what is happened facilitates its narration. For this reason, studies on children’s autobiographical memory used frequent cue words belonging to children’s language. From the age of 8, autobiographical narration appears to increase in relation to the increase in children’s semantic, linguistic and cognitive skills (Brubacher et al., 2012).

According to Kulkofsky and Klemfuss (2008), both social and cognitive processes are involved in developing adequate narrative skills. By sharing and narrating their experiences to others, children learn to attribute meaning to events and to organize the main markers of the narrative which are linguistic-semantic and temporal (not only of when it happened, but also in terms of sequentiality before and after, so as to say a linear succession of what happened; Nelson and Fivush, 2004; Kulkofsky and Klemfuss, 2008).

Security and trust in one’s memory can represent protective factors compared to factors of distortion and suggestion. Distrust Memory Syndrome is in fact one of the main factors of vulnerability to misleading information and social pressures during a forensic interrogation (Gudjonsson, 2018).

Several studies have highlighted gender differences in the production of episodic events and in the description of details. Women tend to show a greater ability to retrieve specific autobiographical events than men (Herlitz and Rehnman, 2008; Wang et al., 2011). Gender differences in episodic memory would tend to emerge early. Research has also found that, compared to men, women show greater accessibility in recalling personal experiences from early childhood (Rubin, 2000), and are more capable of recalling memories of specific events, events unique or happened only once.

Lamb et al. (2018) found that judicial interviews with children occur using lots of suggestive questions that risk to altering the reconstruction of what happened and the memory of the event. The model of suggestibility that has found wide validity in the forensic field is that Interrogative or Immediate Suggestibility of Gudjonsson and Clark (1986). When witnesses are exposed to unanswerable or misleading questions, they may show weakness due to their uncertainty, vulnerability toward authority figures and high expectations of success. Their vulnerability may increase if exposed to negative criticism, leading them to modify their responses and be more suggested (Gudjonsson, 2018; Vagni et al., 2018). Younger children generally present greater vulnerability and a lower ability to provide resistant responses and use adequate coping strategies (Maiorano and Vagni, 2020; Gudjonsson et al., 2021, 2022; Vagni et al., 2021, 2023). Males and females show the same vulnerability to interrogative suggestibility. In fact, several studies have not found gender differences in answering leading questions (Gudjonsson, 2003; Klemfuss and Olaguez, 2020).

Loftus’ studies have found that, despite the presence of a very accurate autobiographical memory, both children and adults can be induced to modify their memory trace in a suggestive way (Berkowitz and Loftus, 2018). Other studies, on the contrary, showed that having high narrative skills had a protective function with regard to external sources of suggestion. Kulkofsky and Klemfuss (2008) have showed that having high narrative skills is associated with a greater source monitoring capacity which allows children to evaluate the information being acquired and compare it with their own knowledge to carry out assessments of discrimination, coherence, plausibility in a more plastic and flexible way. High autobiographical narrative skills thus seem to be related to reduced levels of yielding to suggestive questions.

The ability to handle memory and cognitive patterns more flexibly increases with age and this seems to ensure that children are more competent in processing and managing suggestibility factors (Gudjonsson et al., 2016). In line with this, Gudjonsson et al. (2021, 2022) proposed that trust in one’s memory and the high source monitoring skill are protective and resistant factors to suggestive vulnerability.

The ability to recall autobiographical information varies significantly in both children and adults. This variability appears to be primarily linked to individual characteristics that influence the vividness and accuracy of reactivating visuo-spatial information. In children, the development of autobiographical abilities appears to increase with age. Emotional factors also contribute significantly to this process, as lower emotional arousal tends to be associated with a weaker recall compared to situations with higher emotional arousal. Most studies on autobiographical memory have employed the cueing technique (Crovitz and Schiffman, 1974; Rubin and Schulkind, 1997; Conway et al., 1999; Congleton and Berntsen, 2018). However, it is essential to note that the cueing technique used in these studies comes with recognized limitations in the scientific community. Firstly, it is important to emphasize that the stimulus words used often lead to an excessive degree of freedom in memory retrieval, rendering the recollections less comparable to each other (Conway, 2005; Griffith et al., 2012). Some studies have attempted to address this limitation through the assignment of scores typically ranging from 0 to 3. According to Griffith et al. (2012), the criteria for inclusion or exclusion of responses are different in different studies. Nevertheless, it should be noted that these numerical values are associated with qualitative criteria rather than a purely metric approach. Consequently, this opens the possibility for evaluative discretion on the part of the examiner when assigning a specific score.

Clinical and forensic practice lacks a standardized measuring tool for autobiographical narrative skills. This was the main objective of our study aimed at validating a metric tool for measuring the individual narrative abilities of the autobiographical memory. In fact, many studies have dealt with children’s memory by evaluating the quality of the story or considering the presence/absence of some information (Goodman et al., 2002, 2010).

Most studies on autobiographical memory have used the Autobiographical Memory Test (AMT; Griffith et al., 2012), based on the original methodology developed by Williams and Broadbent (1986). In the AMT, participants are presented with a series of guide words, and are asked to produce a relevant personal memory. The memories provided are then coded based on level of specificity (Griffith et al., 2012). Most of these studies divided the cue words on the basis of positive vs. negative valence, and only a few studies also included neutral valence words (Griffith et al., 2012; Ros et al., 2018). Generally between 10 and 20 keywords were used (Liu et al., 2014). Generally studies that used AMT obtained a single factor solution. Age is an important factor in determining the ability to recall specific events from autobiographical memory. The AMT test was also applied to pre-school children demonstrating how children above the age of 4 can present an adequate autobiographical function (Nieto et al., 2017). The cue-words used by Nieto et al. (2017) concerned basic emotions that children may already know, such as joy, anger, and sadness.

There are also other techniques to evaluate autobiographical memory in children, such as those based on standardized open memory questions about the recent and more distant past concerning common autobiographical events in young children’s lives (Wang, 2004, 2008).

In the AMT procedure, participants are given few restrictions on the retrieval process and are told that the recalled event could have happened either recently or a long time ago as well as being told that the narrated event could be important or trivial (Griffith et al., 2012). The answers are coded on the basis of the following categories: (1) specific memories, (2) extended memories or those that refer to one or more days (for example “my wedding”), (3) categorical memories (school memories, or those linked to my family), (4) semantic associates, or (5) omissions or non-response.

However, no study has used a standardized metric tool to compare between children of different ages. Furthermore, studies on children’s autobiographical memory are based on their ability to recall memories associated with cue-words, but not on the measurement of the main narrative skills (What, When, Who, Where, and How) in terms of numerical assignment of their presence.

At same time, several studies have analyzed the suggestive vulnerability of children using metric and standardized tools (Gudjonsson et al., 2016, 2020, 2021, 2022), but not in association with objective measures of autobiographical memory. In forensic practice it is important to understand the relationship between autobiographical capacity and how children manage the suggestibility factors inherent in a forensic interview. For example, an earlier study by Kulkofsky and Klemfuss (2008) investigated the effect of narrative skills in reducing suggestibility, but without using standardized tools and metric measurements.

The Interrogative Suggestibility model is closely linked with the Gudjonsson Suggestibility Scale (Gudjonsson, 1997), a tool designed to gauge an individual’s susceptibility to suggestive influences in the context of learned and non-autobiographical memories (Vagni et al., 2015, 2017, 2022; Gudjonsson et al., 2020). To date, there are no studies that have verified the impact of autobiographical skills on interrogative suggestibility using metric and standardized measures Therefore, we aim to investigate whether, and to what extent, children of varying ages with strong autobiographical memory skills exhibit enhanced resistance to suggestibility factors. However, prior to embarking on this investigation, it is essential to address the existing gap in the literature pertaining to the absence of a valid and reliable metric tool for assessing autobiographical and narrative skills in children.

In this paper, we propose a new instrument called the Children Recalling Autobiographical Memory test (CRAM). This tool has its theoretical framework in the Conway’s model of autobiographical memory (Conway et al., 2004; Conway, 2009). The selection of cue-words derives from some words with positive and negative valence used in previous studies with children (Griffith et al., 2012; Nieto et al., 2017; Ros et al., 2018) and from the work of Pierucci et al. (2023). It is a tool composed of 20 cue words, which measures the main narrative skills with respect to retrospective and prospective autobiographical memory using the cueing technique. The selection of cue words occurred following the criteria used by Conway (2001), according to which autobiographical memories are organized in both a more general and specific sense, both for retrospective and prospective memory. Conway (2001) also used the following criteria: words that refer to commonly used objects; clichés and emotions. In choosing the cue words we also tried to identify terms that are also frequently used by children, and pertinent to children’s lives (such as: “birthday” and “school”). Indeed, autobiographical episodic memory is therefore linked to “temporally” significant events, but also to events repeated several times over time and linked to objects/people of everyday life (Conway, 2001). Several authors (Jones and Pipe, 2002; Wood and Conway, 2006; Conway, 2009) also highlighted how in addition to emotional intensity, personal memory is influenced by the emotional value experienced. The CRAM assesses children autobiographical and narrative skills in five main domains: Where, What, When, Who and How, giving rise to a simple and interpretable score from 0 to 5 for each item, according to the numbers of domains reported. In this study, we conducted the validation and reliability analysis on this new tool, and then we used it to study the relationship between autobiographical memory capacity and suggestibility in a sample of children of different ages.

Based on the reviewed literature, we formulated the following set of hypotheses:

The sample included 321 participants (47.4% female; 52.6% male) recruited in several Italian schools. Participant ages ranged from 7 to 16 years old (M = 11.03 and SD = 2.49). Ninety-nine children (30.8%) were aged 7–9 years; 125 children (38.8%) were aged 10–12 years, and 97 children (30.2%) were 13–16 years. Data collection took place between February and May 2023 and the participants were met after their parents/guardians had submitted signed consent forms. The researchers administered the instruments in classrooms and verified the inclusion criteria for the study: (a) understanding of the Italian language; (b) absence of serious or severe intellectual pathologies; (c) absence of auditory and visual deficits. Study procedures were implemented during class time and anonymity of the participants was ensured. The study involved an assessment of suggestibility and autobiographical narrative skills.

The instruments were administered following the same procedure with all the participants: the Gudjonsson Suggestibility Scale 2 (GSS2) was administered first, while the autobiographical memory instrument was administered during the latency phase between the memory task and the suggestive interview of the GSS2.

Immediate recall of the GSS2 and autobiographical memory were collected in written form by the participants themselves; while the suggestive interview was administered orally and individually as required by the standard test procedure (Gudjonsson, 1997).

The study was conducted following and respecting the ethical principles in accordance with ethical research involving children. The study conformed to all ethical guidelines for research with human participants and followed the Declaration of Helsinki. The informed consent was signed before the inclusion of the children in the study, and it contained information on the objective of the study, methods of conduct, anonymity, and information on the conservation of sensitive data. The study was approved by the institutional ethics committee.

The CRAM items were checked for EFA assumptions. The KMO test revealed an overall value of 0.94, and the Bartlett’s test of sphericity was significant, χ²(190) = 2,071, p < 0.001, indicating that assumptions for the EFA analysis were met. The analysis of the eigenvalues and of the scree plot (see Figure 1) support a single-factor solution. In particular, following the Kaiser–Guttman’s criterion, only the first factor had an eigenvalue higher than 1.0 (6.55), whereas the other had lower eigenvalues (<0.82). Therefore, we further conducted the EFA with a single factor.

The result of the EFA analysis was reported in Table 1, along with items’ scores (mean and SD) and factor loadings. The one-factor solution adopted explained about 33% of the total variance and all items showed loadings >0.429. The one-factor model revealed also good fit indexes, with relative χ² = 1.93, CFI = 0.92, TLI = 0.91, and RMSEA = 0.05 [0.04, 0.06]. Based on this analysis, we computed an overall score for autobiographical memory to be used in the subsequent analyses.

First, we addressed the internal consistency of this score by conducting a reliability analysis. The CRAM score showed good internal reliability, with Cronbach’s alpha of 0.905 and McDonald’s omega was 0.906. Item-score correlations ranged from 0.412 and 0.622. Overall, this analysis supported a good level of reliability for the CRAM score.

We then move to the analysis of the relationship between the CRAM score, the GSS2 scores, and the demographic variables, i.e., age and gender. First, we tested the effect of gender on the psychological variables in a series of independent samples t-tests. As reported in Table 2, we obtained a significant effect for the effect of gender on CRAM score, with mean score for male = 35.30 (SD = 12.00) and mean score for females = 41.60 (SD = 14.50), whereas gender had no significant effect on any GSS2 scores, all ps > 0.219.

Afterward, we moved to test the correlations among age and the psychological variables, as reported in Table 3. Age was strongly and positively correlated with CRAM score (as depicted in Figure 2), and negatively correlated in a weaker manner with suggestibility scores, i.e., total GSS2 score, YELD1 and YELD2. In line with our hypothesis, CRAM score was negatively and mildly correlated with some suggestibility scores, such as YELD1 and total GSS2 score. Lastly, all the GSS2 scores were strongly and positively correlated with each other. Overall, this analysis indicated that the relationship between autobiographic skills and suggestibility was supported in our sample, and that age seemed related to both these variables. In a more refined analysis on the relationship between age and CRAM score, we decomposed the total CRAM score into the five aspects as related to What, Where, When, Who, and How, and correlated each of these sub-scores with the age. We found that age positively correlated with the scores When (r = 0.217, p < 0.001) and Who (r = 0.130, p < 0.05), and less strongly with Where (r = 0.108, p < 0.05) and How (r = 0.108, p < 0.05) scores, while it did not significantly correlate in our data with the What score (r = 0.071, p = 0.206).

Lastly, we investigate the hypothesis that the reduction of suggestibility observed with age is mainly due to an increase in autobiographical skills. Based on the previously reported results, we than conducted a hierarchical linear regression model, in which we included in the first step CRAM score and age as antecedent variables, gender as a covariate, and GSS2 total score as the dependent variable, while in the second step we added the interaction term CRAM score * age. On note, removing the covariate gender did not change the pattern of results obtained for the other variables. The first step model was significant, F(3, 309) = 4.92, R = 0.214, R² = 0.046, p < 0.01. Model estimates and statics are reported in Table 4, including both regression coefficients (as b) and standardized coefficients (as β). As reported, both age and CRAM score had a significant negative relationship with suggestibility, while age had no relationship with the dependent variable. The second step model also included the interaction term CRAM * age. Adding this term significantly increased model fit, ΔR² = 0.016, p < 0.022. In fact, as reported in Table 4, the interaction term was significant. To further probe this interaction, we ran a simple slope analysis on the effect of age on suggestibility at three levels of autobiographical skills, i.e., low (−1 SD from the mean), medium (mean value), or high (+1 SD from the mean). We obtained a non-significant relationship of age and suggestibility when CRAM score was low, b = 0.016, SE = 0.168, p = 0.925, whereas such a relationship was significant for average, b = −0.253, SE = 0.121, p = 0.037, or high, b = −0.522, SE = 0.174, p = 0.003, level of CRAM score. The slopes are also reported in Figure 3. Overall, this analysis indicates that autobiographical skills are related to suggestibility irrespective of age and that they could also moderate the effect of age on suggestibility, meaning that only children with medium or high levels of those skills are less likely to be susceptible to external influences.