Autobiographical Memory Impairment in Adolescents in Out-of-Home Care

Abstract

Childhood maltreatment is a major risk factor for emotional problems in adolescence and adulthood and has deleterious effects on cognitive functions such as working memory. A key aspect in the study of the cognitive and affective consequences of maltreatment is autobiographical memory, especially regarding the difficulty retrieving specific memories, known as overgeneral memory. In this study, autobiographical memory tests, working memory, and a depressive symptom assessment were administered to 48 adolescents in care with a history of maltreatment (22 abused and 26 neglected) without mental disorder, who had been removed from their family and were living in residential child care, and to 61 adolescents nonmaltreated who had never been placed in care. The results show that adolescents with a history of maltreatment remember fewer specific events (Cohen d = 1.1–1.3) and that both working memory and depressive symptoms are involved in overgeneral memory (R² = .13–.26). Directions for future research include neuropsychological assessment and implementation of a training program to increase specific memory recall in this population.

Keywords

child maltreatment out-of-home care autobiographical memory working memory depressive symptoms

Acts of physical abuse, sexual abuse, neglect, psychological abuse, all of which are considered maltreatment, constitute profound failures of caregivers to foster normal child development (Goodman et al., 2010). Victims of maltreatment are often deprived of experiences which, according to theories of normal development, are necessary for healthy adaptation. Children who have suffered domestic maltreatment are at risk for problems across several domains of functioning, not only in childhood, but also throughout life especially in adolescence. There is a significant body of research concerning the potential effects of maltreatment on a wide range of cognitive, socio-emotional, and neuropsychological outcomes (Masson et al., 2015).

One of the cognitive domains that has been studied in victims of abuse is autobiographical memory (AM): an aspect of memory related to the recollection of past events that a person has experienced. It is central to human functioning, as it is of fundamental significance for the individual’s sense of self and goal orientation (Williams et al., 2007). Specific autobiographical memories or autobiographical memory specificity (AMS) are very concrete recollections of events that occurred at a specific moment in one’s personal past and include sufficient detail to identify one particular incident (e.g., “the day I got married”). Reduced autobiographical memory specificity (rAMS) is considered a cognitive marker and predictor of the course of depression (Sumner et al., 2010). Research has shown that this phenomenon, also called overgeneral memory (OGM), is more than just a symptom of depression; it functions as a vulnerability factor or trait marker (Williams et al., 2007).

In the context of trauma, a recent meta-analysis by Barry et al. (2018) showed a large effect size of memory specificity (d = 0.77) between participants with and without exposure to trauma was large. This reduced capacity is related to exposure to trauma and may have a significant impact on the development and maintenance of emotional pathologies. A compelling approach to the reduced specificity effect is the affect regulation hypothesis, which proposes that trauma-exposed individuals experiencing trauma-related distress seek to avoid remembering the specific details of their traumatic experiences in an attempt to reduce their distress (Hermans et al., 2008). This avoidant memory style, it is proposed, then generalizes beyond the recollection of the trauma memories until it characterizes much broader domains of autobiographical remembering, resulting in reduced specificity on tests which do not explicitly target trauma memories. As such, OGM would be functional initially following trauma, only becoming maladaptive at a later point, once the style becomes chronic (Hitchcock et al., 2014). This cognitive style has also been associated with decreased effectiveness of executive functions (EFs) and the presence of depressive symptomatology (Williams et al., 2007).

The EFs are a family of top-down mental processes associated with the functioning of the prefrontal cortex, which develop significantly in the preschool years. More specifically, these functions refer to the activation, maintenance, and selection of different courses of action needed to undertake complex behaviors to achieve different goals (Miyake & Friedman, 2012). From a cognitive perspective, it has been suggested that EF plays an important role in the normative development of AM. Drawing on the research with samples of adults in the evaluative review conducted by Sumner (2012), the magnitude of OGM appears to be especially associated with deficits in inhibition, working memory (WM), the ability to update and maintain information in WM, and verbal fluency. Similarly, recent research among school-aged children demonstrated a relationship between OGM and self-reported inhibitory control and revealed that inhibitory control partially mediated the link between children’s depressed mood and OGM (Raes et al., 2010).

Furthermore, the role of depressive symptoms is a key question in the study of the relationship between OGM and trauma. The tendency to recall OGMs, rather than specific past episodes, is consistently observed among individuals who are depressed, compared to those without depression (for review, see Williams et al., 2007). That is, individuals with depression, those in remission, and formerly depressed individuals all have significant difficulties providing specific memories. A number of studies have assessed OGM in adolescent victims of maltreatment in childhood and early adolescence (for a review, see Alley et al., 2015), finding that exposure to maltreatment at early ages leads to greater cognitive vulnerability and higher risk of depression and other mental health problems (Stange et al., 2013). Maltreated children’s memories were more overgeneral and contained more negative self-representations than did those of the nonmaltreated children. Negative self-representations and depression were significantly related to OGM (Valentino et al., 2009).

A recent meta-analytic review of the role of trauma history and mood state in OGM (Ono et al., 2016) showed that AM deficits, seen as negative schemas, may be one of the vulnerable factors that develop and maintain clinical symptoms in trauma-exposed individuals. Furthermore, a longitudinal study with 6,000 adolescents explored the association between childhood trauma exposure and OGM in adolescence. The findings suggested a modest association between exposure to traumatic events and later OGM (Crane et al., 2014).

Previous studies suggest that exposure to traumatic events in childhood is associated with impaired specific memory retrieval. Furthermore, this impairment is partly associated with a greater presence of depressive symptoms and lower WM efficiency. However, these results stem from studies using self-report assessment of trauma experience in either general or psychiatric populations, but little is known about the effects of childhood maltreatment in adolescents in out-of-home without a current mental health diagnosis. We examine whether exposure to maltreatment in childhood has an impact on the ability of adolescents to retrieve specific memories, and on other related variables such as WM and depressive symptoms. This work complements previous studies on trauma experience in nonclinical adolescents in care (residential child care). Residential child care exists to ensure that the needs of children are met when they cannot live with their own family, and is thus a resource that provides security and safety for minors who have been victims of domestic violence. The aim of this study was to analyze OGM in a sample of adolescents in care without mental disorders, and who suffered maltreatment experiences (abuse or neglect) in the family during childhood. In addition, this study analyzes the differences between maltreatment adolescents in residential child care and adolescents nonmaltreated who had never been placed in care.

The specific aims of this work were (a) to analyze the differences in AM, WM, and depressive symptomatology between maltreated and nonmaltreated adolescents, (b) to analyze the predictive power of exposure to maltreatment, WM, and depressive symptomatology on adolescent participants’ ability to retrieve specific autobiographical memories. We expected to find a lower ability to access specific memories in adolescents who were victims of abuse or negligence in early childhood. In addition, we expected them to present higher depressive symptomatology and lower WM capacity, while a significant relationship was expected to be found between the difficulty to access specific memories (OGM) and early childhood exposure to maltreatment, depressive symptomatology, and impairments in WM. Finally, we expected childhood exposure to maltreatment, WM, and depressive symptomatology to be significant predictors of OGM.

Method

Participants

Participants were 109 Spanish adolescents, 48 maltreated who were currently living in out-of-home care, and 61 nonmaltreated who had never been placed in care. The maltreated adolescents, who had been victims of abuse or neglect in childhood, were removed from their families and were living in a residental child care. These 48 adolescents are part of a larger longitudinal study, examining the effects of childhood traumatic experiences on subsequent affective, social, and cognitive development.

Maltreated adolescents

A total of 148 adolescents had been victims of childhood trauma and they lived in residential child care in Albacete (Spain) between 2013 and 2015. The inclusion criteria were as follows: adolescents (12–17 years), victims of maltreatment (abuse or neglect) during childhood, no current mental health diagnosis (all adolescents were monitored by clinical psychologists and psychiatrists from the child and juvenile mental health services of the Spanish public health system), and enrolled at school in their corresponding group (secondary education or baccalaureate). Following these criteria, 48 adolescents were selected (32.4% of total) and were divided in two groups: Abuse (physical, psychological, and/or sexual abuse, n = 22) and Neglect (n = 26). Before being placed in out-of-home care, the adolescents lived with their biological families. Table 1 shows the time that adolescents had lived with their biological families, time lived in the current residential child care, and the period of time between removal from the biological family and placement in the current residential child care (itinerant time). For some of the subsequent data analyses, the two groups of maltreated adolescents were put together into one (n = 48).

Table 1.

Background Data on Participants in the Study.

Variable	Nonmaltreated(n = 61)	Abused(n = 22)	Neglected(n = 26)	Statistical Test
Age, M (SD)	14.69 (1.79)	14.36 (1.73)	14.36 (1.73)	F(2, 106) = 0.32
Age in months, M (SD)	181.00 (20.41)	175.64 (22.15)	180.04 (21.35)	F(2, 106) = 0.53
Gender (% male)	55.7%	22.7%	38.5%	χ² = 7.70*
History
Time in Biological Family in months, M (SD)	—	138.32 (34.43)	114.88 (29.16)	t = 2.18 *
Itinerant^a (% of 1 month or more)	—	18.2%	34.6%	χ² = 1.63
Time in current residential child care in months M (SD)	—	35.36 (35.17)	52.12 (39.01)	t = −1.55

Note. In the prior history variables, only the two maltreatment groups are compared.

Itinerant: percentage of adolescents for whom a month or more passed between separation from their biological family and placement in the residential child care where they currently live.

p < .05.

Nonmaltreated adolescents

The group of nonmaltreated adolescents was recruited from a high school in Albacete, with similar socioeconomic characteristics to the schools attended by the maltreated adolescents. The adolescents were selected from among the school’s 700 students ensuring it was proportional in the number of students by school year and age to the sample of adolescents in the maltreated group. We selected minors meeting the following inclusion criteria: adolescents (12–17 years), enrolled at school in their corresponding year group, adolescents living with their biological families or another direct family member and who had never been placed in out-of-home care, and no current mental health diagnosis (with the help of the high school psychologist, any student presenting a mental health problem was excluded from the study). The selection of the nonmaltreated group was randomized between those that followed inclusion criteria. The nonmaltreated group comprised 61 adolescents.

Abused, neglected, and nonmaltreated adolescents did not differ in age, but a significant difference was found between groups regarding gender, χ² (N = 107) = 7.70, p = .021. At the moment of data collection, the percentage of institutionalized female adolescents was higher. Consequently, it was not possible to homogenize both groups by gender. Gender differences were controlled for in subsequent analyses.

Procedure

This research complies with the ethical principles of the Declaration of Helsinki, with the code of ethics of the Spanish Association of Psychologists and with the principles of confidentiality in the treatment of the adolescents were guaranteed in accordance with current Spanish data protection laws (LO 3/2018 of 5 December). All procedures involved with the study were reviewed and approved by the Institutional Review Board. Guardianship regulations are very strict with regard to the use of these adolescents’ personal data. We only have information on the reason why they were removed from their families (abuse or neglect), the time they lived with their biological families, the time of itinerancy and how long they have been in the current custodial care system.

This research was developed in the framework of adolescents in out-of-home care in Albacete (Spain). The research project was presented to the directorate of the Service for the Protection of Minors, to the Directorate-General of Welfare Services, and the coordinators of the all residential child care of Albacete (N = 18). Participation in the research was voluntary, and no adolescents had to participate against their wishes. All the participants gave their signed informed consent. The date and time of the assessment of each adolescent was agreed with the coordinators of the residential child care. Data collection was conducted individually.

In order to assess the nonmaltreated adolescents, we contacted the principal of a high school in Albacete, who agreed to collaborate. Once the high school had agreed to take part in the study, an information letter was sent to students’ parents. The letter informed about the aims of the study and the confidentiality of the results and included a request for informed consent. Furthermore, the minors participated voluntarily and also gave their informed consent. Both parents and participants gave their signed informed consent. The interviews with the minors were conducted individually during tutoring hours.

Measures

Sentence Completion for Events from the Past Test (SCEPT) (Raes et al., 2007)

Autobiographical memories were recorded using 11 short incomplete sentences as cues and participants were asked to supply the stem of the sentence in written form. A sample item is “When I think back to/of . . . ” or “I will never forget . . . ” The Spanish version of the SCEPT (Ricarte et al., 2016) was previously validated in adolescents (64.7% females). Inter-rater agreement calculated on 15% of the total obtained sentences was 92%, Kappa = .89. Two independent researchers categorized responses on the SCEPT. The 11 possible responses were classified by both participants and examiner as: extended memories, categoric memories, specific memories, or omission. Sentences describing unique events lasting less than 24 hours were categorized as specific memories (e.g., “The day my little brother was born”); sentences describing repeated summaries of events were categorized as categoric memories (e.g., “My birthdays”); sentences describing events lasting longer than 24 hours or periods of life were categorized as extended memories (e.g., “When I was a child”). In the case of any discrepancy, they were blind-coded by a third examiner experienced in correcting the test. Interjudge agreement was calculated for 25% of the sentences, yielding an 87% of concordance. In addition, for the purpose of this study, specific memories were categorized by positive and negative valence. Studies such as that by Ono et al. (2016) have shown differences in the use of positively versus negatively valenced cues. Although the SCEPT does not use cue words, we thought it was appropriate to include this difference in the analyses when classifying the specific memories. The affective valence of the specific memories was conducted by two experts, who independently coded the valence of the memories (positive or negative). In the case of any disagreement, the opinion of a third expert was available. In this case, an inter-rater agreement of 91% was obtained.

Moreover, the almost total lack of specific memories was considered to indicate rAMS or OGM, following the same criteria used in the study by Crane et al. (2014) for the classification. The results were coded, and participants were classified into two groups: those who retrieved one or no specific memory (OGM = 1) and those who recalled two or more (OGM = 0).

Wechsler Intelligence Scale for Children (WISC-IV, Wechsler, 2003)

A study by Benson et al. (2013) showed that several subtests of the WISC-IV can be used to measure WM ability. We used the following subtests from the WISC-IV, which were all administered individually:

Digits: The test includes two tasks: Digit Span Forward and Digit Span Backward. In the first part, the examiner reads a list of numbers and the participant is required to repeat them in the same order, while in the second part, the numbers must be repeated in the reverse order to the spoken sequence.

Arithmetic: This task is used in the WISC as a complementary WM test. The child is given a limited time to solve a series of orally administered arithmetic questions. The test involves mental manipulation of information, concentration, attention, short- and long-term memory, numerical reasoning ability, and mental alertness. It may also motivate the use of fluid reasoning, the ability to identify sequences, and logical reasoning.

Children’s Depression Inventory

The Children’s Depression Inventory (CDI; Kovacs, 1992) is a downward extension of the Beck Depression Inventory (Beck, 1967), with language and format changes to accommodate youth aged 7 and older. For each of the 27 items, participants select one of three response alternatives that vary in severity. In this study, we used the Spanish version created by Del Barrio et al. (1999), which has an internal consistency of .82 and test–retest reliability of .70. These authors propose a two-factor model for a population of Spanish adolescents: dysphoria (depression symptoms) and negative self-esteem.

The CDI and SCEPT were self-administered in the presence of a researcher specialized in assessment who was able to resolve any problems of understanding or doubts about the questions included in the tests. The same researcher also administered the WISC.

Data Analysis

To analyze the differences between maltreated and nonmaltreated adolescents, differences in categories of retrieval as a function of group status were assessed using multivariate analysis of covariance (MANCOVA). Broadly speaking, when multiple dependent variables (DVs) are present, conducting a multivariate analysis of variance (MANOVA) is beneficial because it reduces the likelihood of Type-I error (Haase & Ellis, 1987; Huberty & Morris, 1989). In addition, as DVs are moderately interrelated (between .3 and .7), and we were interested in evaluating the linear combination effect of DVs, the best choice is the MANOVA (Maxwell, 2001). In addition, covariates are incorporated into the MANOVA, thus serving to reduce the error term in the model. Bearing in mind that the groups were not equal in gender, we included gender (male/female) as a covariate and “group” (nonmaltreated, abused, or neglected) was entered as the between-subjects factors, whereas the different categories of retrieval (specific, categoric, extended, and omission) were entered as DVs. Post-hoc analyses were performed using the Bonferroni adjustment included in the GLM SPSS procedure for each category of retrieval. Another MANCOVA was performed, group as the independent variable and specific positive and negative memories as DVs. Finally, two new MANCOVAs were used to measure the differences by group in the two WISC subtests and the two CDI subtest, respectively.

To analyze the capacity of maltreatment, WM, and depression symptoms to predict OGM, relations between group, categories of retrieval, cognitive functioning, and depression symptoms were analyzed using Pearson product–moment correlations. Two logistic regression models were used to determine the capacity of the variables of group, WM and depression symptoms to predict OGM. In the first, WM and depressive symptomatology were included as prognostic variables on OGM. In the first, WM and depressive symptomatology were included as prognostic variables on OGM. Given that logistic regression does not allow us to test for change in the total amount of variance explained, a second logistic regression was tested including maltreatment group variable as a predictor. Data analysis was conducted using SPSS, version 20, and STATA, version 14.

Sample Power

G*Power (version 3.1.9.3) was used to calculate the power of the sample. With the calculation of the F test and MANCOVA, taking an effect size value of 0.77 obtained from the work by Barry et al. (2018), an α value of 0.05, with n = 109, divided in three groups, and with five measure variables (SCEPT, WISC, and CDI), we obtained a power of 1.00 (1−β err prob), critical F = 1.87, numerator df = 10,0, denominator df = 206, noncentrality parameter λ = 167.86, and Pillai V = 1.06. For the regression analysis, we used the two tailed t-test, with H1 ρ² = 0.3, H0 ρ² = 0, α err prob = 0.05, total sample size = 109, and number of predictors = 5. We obtained a power of 0.99 (1−β err prob), lower critical R² = .008 and upper critical R² = .115.

Results

Differences Between Maltreated and Nonmaltreated Groups

AM

To analyze the differences in AM across the three groups, we conducted a MANCOVA, taking group (abuse, neglect, and nonmaltreatment) as the independent variable and the number of memories in each of the different categories of the SCEPT: specific, categoric, extended, and omission, as DVs. We included gender (male/female) as the control variable. The Bonferroni correction test was used to evaluate post hoc differences. The multivariate analysis was statistically significant (Wilks = .598, F(8, 204) = 7.47, p < .001, η² = .227). As can be seen in Table 2, there was a significant effect of group (p < .01) in all the SCEPT categories, and the value of η² was large for specific memories (.28). In the post hoc comparisons, there were differences (p < .05) between the group of nonmaltreated adolescents and each of the other two groups, except in the category of omissions, where the nonmaltreatment group differed only from the neglect group. Broadly speaking, there was a higher prevalence of specific memories in the group of nonmaltreated adolescents, while in the other two groups extended memories were more prevalent (see Figure 1). The effect size was d = 1.30 when we compared the number of specific memories in the control group and the neglect group, and d = 1.07 when compared with the abused group.

Table 2.

Means, Standard Deviations, and MANCOVA Results by Maltreatment Subtype Group Controlled for Gender.

Dependent Variable	Abused(n = 22)		Neglected(n = 26)		Nonmaltreated(n = 61)		F (2, 106)	p	η²	Post Hoc
Dependent Variable	M	SD	M	SD	M	SD	F (2, 106)	p	η²	Post Hoc
SCEPT
Specific	2.05	1.98	1.54	2.00	4.44	2.32	20.71	.000	.283	A B
Categoric	0.27	0.55	0.31	0.61	1.15	1.27	7.16	.001	.120	A B
Extended	5.86	2.88	5.62	3.08	3.61	2.39	8.45	.000	.139	A B
Omission	2.82	2.36	3.46	3.29	1.74	1.99	4.61	.012	.081	B
Specific memories
Sp. Positive	0.95	1.58	0.69	1.12	3.13	2.20	20.03	.000	.276	A B
Sp. Negative	0.59	0.79	0.54	0.94	1.03	1.34	1.93	.149	.036	N/A
WISC
Digit total	10.68	3.01	12.08	3.51	14.56	4.11	9.08	.000	.147	A B
Arithmetic	3.68	1.93	3.96	1.64	5.26	1.53	8.95	.000	.146	A B
CDI
Dysphoria	5.36	6.38	6.54	6.36	3.30	3.35	4.13	.019	.073	B
Self-esteem	5.41	4.98	5.92	4.04	7.28	3.54	2.23	.112	.041	N/A

Note. Italic p < .001, Underlined p < .01, Normal type p < .05, N/A not applicable. The values of F and η² correspond to the tests on between-subjects effects from the MANCOVAs. The covariate variable (gender) showed no significance in any of the MANCOVAs conducted. MANCOVAs = multivariate analysis of covariance; Post hoc: A = nonmaltreated vs. abused; B = nonmaltreated vs. neglected; C = abused vs. neglected; SCEPT = Sentence Completion for Events from the Past Test; WISC = Wechsler Intelligence Scale for Children; CDI = Children’s Depression Inventory.

Figure 1.

Mean number of memories (±SE) assigned to the different SCEPT response categories for the three groups (abused, neglected, and nonmaltreated).

Another similar MANCOVA was performed, but this time taking specific positive and negative memories as DVs and group as the independent variable. The multivariate analysis was statistically significant (Wilks = .781, F(4, 208) = 6.82, p < .001, η² = .122). The number of specific positive memories (SPMs) was higher in the nonmaltreatment group than in the other two (η² = .28). However, there were no differences between groups for specific negative memories (Table 2). The nonmaltreatment group retrieved significantly more SPMs than negative ones (t(60df) = 6.09, p < .001, CI 95% = 1.41–2.78, d = 1.57), while in the abuse group (t(21df) = 1.05, p = .304, CI 95% = −.35 to 1.08, d = 0.45) and the neglect group (t(25df) = 0.75, p = .461, CI 95% = −.26 to .57, d = 0.30), there were no differences in the means (see Figure 2).

Figure 2.

Mean number of specific positive and negative memories (±SE) on the SCEPT for the three groups.

Finally, given that the absence of autobiographical memories has been considered an indicator of OGM, participants were classified into two groups: those who retrieved one or no memory (OGM = 1) and those who retrieved two or more (OGM = 0). A total of 34 adolescents (31.19%) presented OGM. The differences between groups were considerable (χ²(1df) = 29.44, p < .001); only 9.84% (6 of 61) of nonmaltreated adolescents presented OGM, while in the maltreatment group (abuse and neglect) the proportion was 58.33% (28 of 48).

WM

To analyze the differences in WM, we performed a MANCOVA, taking group (abuse, neglect, and nonmaltreatment) as the independent variable, and the total digit span score and the arithmetic WISC-IV subtest scores as DVs. We included gender (male/female) as the control variable. We used the Bonferroni correction test to evaluate post hoc differences. The multivariate analysis were statistically significant (Wilks = .801, F(4, 208) = 6.02, p < .001, η² = .104). As can be seen in Table 1, there was a significant effect of group (p < .001) in both WISC-IV categories. The post hoc comparisons found differences (p < .01) between the nonmaltreatment group and the other two groups in both total digit span and arithmetic scores. Broadly speaking, we can see that the nonmaltreatment group scored higher in both tests (see Figure 3).

Figure 3.

Mean direct scores for the two working memory subscales from the WISC for the three groups.

Depression symptoms

To analyze the differences in depressive symptomatology among the three groups, we performed a MANCOVA, taking group (abuse, neglect, and nonmaltreatment) as the independent variables, and the scores on the CDI dysphoria and self-esteem subtest as the DVs. We included gender (male/female) as the control variable. We used the Bonferroni correction test to evaluate post hoc differences. The multivariate analyses were statistically significant (Wilks = .755, F(4, 208) = 7.58, p < .001, η² = .131). As can be seen in Table 2, in dysphoria there was a significant effect of group (p < .05), although the value of η² was low (.07). The post hoc comparisons only showed differences (p < .05) between the nonmaltreatment group and the neglect group. No differences were found in the self-esteem subscale (see Figure 4).

Figure 4.

Mean direct scores for the two CDI subscales for the three groups.

Influence of Maltreatment, WM, and Depression Symptoms in OGM

Our previous analyses were based on the exploratory hypothesis that there could be differences among the maltreatment children (abused and neglected) in the variables tested, taking into account the type of abuse experienced. However, no differences were found between these two groups in the post hoc analysis. Consequently, we decided to consider them together in the subsequent analyses. In the correlational analyses and the logistic regression, we used the dichotomized group, where previously we had included three levels.

Difficulty in retrieving specific memories has consistently been associated with childhood maltreatment, depressive symptomatology, and with impaired performance on executive functioning tasks, especially WM capacity. In this section, we analyze the relative importance of these variables in OGM. Correlation analyses and two logistic regressions were performed. In the first regression, we took OGM (yes/no) as the DV and depressive symptomatology and WM as predictor variables. This first analysis did not include the maltreated group as a predictor variable. In the second regression, to assess the impact of early childhood trauma experience on OGM, the maltreated group was included as a predictor variable.

First, as shown in Table 3, we calculated the correlation indices between maltreatment group, the WM variables (digit total and arithmetic), depression symptoms (dysphoria and negative self-esteem) and OGM group. We also included the correlations of gender, used as the control variable. In subsequent analyses, we included the variables that significantly correlated with OGM as predictor variables.

Table 3.

Correlations Between Predictive Variables and OGM.

Variable	2	3	4	5	6	7
1. Gender	.244*	−.174	−.170	.069	−.026	−.091
2. Group	—	−.292**	−.412**	.267*	−.197*	.520**
3. WM: WISC digits		—	.430**	−.057	.035	.194*
4. WM: WISC arithmetic			—	−.157	.062	.307**
5. Dysphoria				—	.564**	−.309**
6. Self-esteem					—	.141
7. OGM						—

Note. Gender was coded: 0 male, 1 female. Group was coded: 0 Nonmaltreatment, 1 Maltreatment. OGM was coded 0 Absent, 1 Present. OGM = overgeneral memory; WM = working memory; WISC = Wechsler Intelligence Scale for Children.

p < .01. **p < .001.

Second, we performed a logistic regression analysis, taking WM and depressive symptomatology as prognostic variables on OGM. The results revealed that depressive symptomatology (Wald = 6.86, p = .009, odds ratio [OR] = 0.89, CI 95% = .81–.97) and arithmetic (Wald = 6.37, p = .012, OR = 1.41, CI 95% = 1.08–1.85) are significant predictors of OGM. The complete model presents a log likelihood value of −59.16 (LR χ² = 16.97, p < .001, Pseudo R² = .13, Nagelkerke’s R² = .20).

Third, we included a new prognostic dichotomous variable of the effect of maltreatment (abuse-neglect vs. nonmaltreatment), with the complete model increasing its prognostic capacity (Log likelihood = −49.88 (LR χ² = 35.53, p < .001, Pseudo R² = .26, Nagelkerke’s R² = .39). In this case, the only significant effect was that of maltreatment. The effect of depressive symptomatology and the effect of the arithmetic disappeared (Table 4).

Table 4.

Summary of the Logistic Regression Analysis Predicting OGM.

Variable	B	E.T	Wald	Exp(B)	CI 95% for Exp(B)		Nagelkerke^a
Variable	B	E.T	Wald	Exp(B)	LL	UL	R ²
Maltreatment group	2.219	.554	16.016***	9.199	3.103	27.272	.347
WISC arithmetic	−0.137	.149	0.851	0.872	0.651	1.167	.361
Depression symptoms	0.090	.050	3.277	1.094	0.993	1.206	.396
Constant	−1.854	.884	4.402	0.157

Note. ^aHierarchical regression, R² value when each variable is introduced. ***p < .001. OGM = overgeneral memory; CI = confidence interval; LL = lower limit; UL = upper limit; WISC = Wechsler Intelligence Scale for Children; ET = SE Estimate standard error.

Discussion

The main aim of this study was to assess specific memory retrieval in a population of adolescents, who, having been victims of abuse or neglect in childhood, were removed from their families, lived in a residental child care, were attending school regularly, and had no mental health diagnosis. The data showed that the maltreated adolescents have difficulties in retrieving specific memories, regardless of the type of abuse they have suffered. Besides exposure to abuse, depressive symptomology and WM are associated with the deficit in specific memory retrieval.

AM

The main finding of this study is that maltreated adolescents recall significantly fewer specific events than their nonmaltreated peers living with their biological families. This difference exists regardless of whether they are victims of abuse (d = 1.07) or neglect (d = 1.30). Broadly speaking, it can be seen that specific memories are more prevalent in the nonmaltreatment group, while in the other two groups, extended memories are more prevalent. The same is true when considering the OGM variable. While almost 6 of every 10 adolescent victims of abuse exhibit OGM (understood as the inability to retrieve any specific memories or the retrieval of just one), 9 of every 10 nonmaltreated adolescents are able to recall two or more specific memories.

A central finding of this research is that abused and neglected adolescents demonstrated more OGM than did nonmaltreated adolescents, which is similar to results found by previous cumulative research in the field (Alley et al., 2015; Barry et al., 2018; Crane et al., 2014; Griffith et al., 2016; Hitchcock et al., 2014; Ono et al., 2016; Valentino et al., 2009). This investigation provides further evidence of OGM as a function of abuse among a sample of adolescents with documented histories of maltreatment. Independent corroboration of self-reported maltreatment experiences is particularly critical for research on the impact of trauma on memory. The finding of this level of OGM in adolescents with no current mental health disorder, but who have been exposed to abuse in the past, is in line with the understanding of OGM as a cognitive style associated with greater vulnerability in traumatic situations. The results are also similar to those of Kleim and Ehlers (2008), who showed that low memory specificity predicts the exacerbation of depressive mood after a stressful event, and previous research showing OGM in depressed people with history of trauma. The current research advances in the study of AM in adolescents with a documented history of abuse and neglect, expanding on the range of maltreatment in previous works such as that by Ogle et al. (2013), in which adolescents with documented histories of child sexual abuse (CSA) reported fewer specific autobiographical memories than adolescents without histories of CSA. These findings are consistent with previous literature and with predictions derived from the CaRFAX theory of rAMS (Williams et al., 2007). According to this theory of AM development, individuals who suffer early abuse may exhibit difficulty advancing beyond the level of generic memory retrieval characteristic of young children and may develop and maintain a pervasive OGM retrieval style into adolescence.

Regarding the valence of the memory, the number of SPMs is larger in the nonmaltreatment group than in the other two. However, there are no differences between groups in the number of specific negative memories. The same differences in SPM affect both the abuse and neglect groups. In short, the differences are found in positive memories but not in negative ones, regardless of the type of maltreatment. This finding is consistent with the results of the meta-analysis conducted by Ono et al. (2016), who located more homogeneous studies with a larger effect size in OGM for positive memories, indicating that difficulty recalling SPMs may be a more typical form of OGM among trauma and depression populations than other types of OGM. Another study also provided evidence that child maltreatment is associated with impaired WM for positive emotions (Cromheeke et al., 2014). It should also be emphasized that OGM is a vulnerability factor for future clinical symptoms among trauma-exposed individuals (Bryant et al., 2007).

WM

Generally speaking, it can be seen that the nonmaltreatment group scored higher in all the WM tests administered. This finding coincides with other studies showing that maltreated children perform worse on a wide range of cognitive tests, especially those related to EFs and WM (Masson et al., 2015). This study sought to account for the persistent influence of childhood maltreatment on neuropsychological functions in later life (Cromheeke et al., 2014). Although limited, previous work in individuals with experience of early stress (ES) has documented impairments in cognitive control such as inhibition, response shifting, and WM (Mueller et al., 2010).

Furthermore, the results of our work are consistent with other studies in both clinical (Dalgleish et al., 2007) and nonclinical populations (Ros et al., 2010), on the relation between EFs, especially WM, and access to specific memories. Our findings support the theory that executive functioning is one of the factors involved in OGM. Consistent with the notion that executive control is associated with AMS, regardless of the presence of psychopathology (Williams et al., 2007), higher WM scores were associated with higher AMS in both groups. These results are similar to those of Sumner (2012) in her study based on the verbal fluency test.

Depressive Symptoms

Despite adolescents with a current mental health disorder being excluded from the study, participants from the maltreatment group exhibited more depressive symptoms than the controls. This is consistent with other studies suggesting that exposure to maltreatment implies a greater presence of depressive symptoms (Stange et al., 2013). Although the scores are not excessively high, the data supports maltreated adolescents’ increased vulnerability to depression. Prior works have also indicated that sub-threshold depression is associated with substantial functional impairment among adolescents, and sub-clinical depressive symptoms in adolescence predict the onset of major depressive disorder (MDD) in adulthood (van Lang et al., 2007).

Traumatic events and experience of ES during childhood or adolescence, such as parental loss, witnessing violence, or sexual or physical abuse have been linked to a variety of adverse health outcomes during the life span (McCrory et al., 2010). More specifically, these experiences have been shown to be an important risk factor for the development of later psychopathologies, such as depression, post-traumatic stress disorder, anxiety, and substance abuse (Green et al., 2010). The appearance of depressive symptoms in adolescents previously exposed to maltreatment, even in the absence of clinically diagnosed depression, implies reduced quality of life in the present and greater likelihood of clinical problems in the near future.

Apart from a greater presence of depressive symptomology, the adolescents in the maltreatment group exhibited a higher frequency of OGM. These aspects are related. Our results coincide with those of previous studies suggesting that in clinically depressed children and adolescents, OGM is a stable cognitive marker of vulnerability to depression (Vrielynck et al., 2007). This study supports the idea that the factor of vulnerability is also found in adolescents with depressive symptoms. In short, we have found an association between maltreatment, OGM, and depressive symptomology in a group of adolescents exposed to abuse in childhood.

Using multivariate analyses, we found that the number of specific memories is associated with both the arithmetic test and the depressive symptoms score. It is possible that the greater presence of depressive symptoms in the maltreated sample might, in some way, have influenced performance on WM tasks. Although mood state at the time of test administration might have affected performance, current evidence from neurobiology and epidemiology suggests that early life stressors such as child maltreatment and related traumatic stressors may alter brain structure and function and result with possible long-term consequences, representing a latent neurobiological risk factor for later psychopathology and heightened risk taking (De Brito et al., 2013). Some studies show that the effects of maltreatment on emotional processing occur in individuals both with and without mental health disorder (Masten et al., 2008). In short, the effects of abuse and neglect, above all at preschool age, are associated with impaired emotional functioning in later ages. In this respect, especially relevant is a recent longitudinal study (Luby et al., 2016) showing a significant effect of early childhood maternal support on hippocampal volume growth across school age and early adolescence and suggesting an early childhood sensitive period for these effects (Nelson et al., 2008). They also show that this growth trajectory is associated with later emotional functioning.

In sum, experiencing multiple forms of child maltreatment and related traumatic stressors is associated with an increased likelihood of AM disturbances related to childhood (Brown et al., 2007) and OGM, a psychological variable that has been linked to vulnerability to clinical depression (Crane et al., 2014).

This study has certain limitations. The first lies with the observational, transversal nature of the study, which means the results are limited as regards the causal direction of the relationships. A strategy to overcome this limitation would be a longitudinal study, with the first data collection being conducted when children enter the residential child care, with subsequent follow-ups. In any event, a strategy of this type would have to resolve the potential ethical dilemmas associated with confidentiality and the administration of tutelage, which is no simple matter. The second limitation regards the assessment of WM and EFs. Although the WISC subtests have shown a high degree of validity for the measurement of WM, other EFs, such as inhibitory control (Raes et al., 2010), should be studied. Finally, an important limitation to take into account in this work is the small sample size: on one hand, due to the specificity of such a sample and, on the other, the inclusion criteria used, considering that although residential child care in Spain is for minors aged 7 and over, we only worked with participants aged 12 and above (adolescents).

This study suggests two future lines of research. The first is the measurement of variables that complement those already included in this study: voluntary and involuntary autobiographical memories, neuropsychological assessment, and measurement of variables related to communication and interaction patterns in the biological families. The second possibility regards the implementation of an intervention aimed at reducing OGM in adolescents exposed to abuse in childhood. Memory specificity training might be an element to be considered in clinical practice with these types of adolescents.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

María Verónica Jimeno

Jose Miguel Latorre

Author Biographies

María Verónica Jimeno is a doctor in psychology and associate professor in criminology at the University of Castilla-La Mancha. Since 2016, she is a member of the Center of Research of Criminology of this University. Her research interests have always centered on the victimology of childhood and adolescence.

Jose Miguel Latorre is bachelor in psychology and doctor in experimental psychology. He is full professor with the Universidad de Castilla-La Mancha, Albacete, Spain, in the area of Basic Psychology. He has experience in Basic and Applied Experimental Psychology in the study of aging and various transdiagnostic factors in psychopathology, mainly autobiographical memory. Together with others, he has developed psychotherapy and cognitive training techniques for older people with cognitive impairment or depression and patients with schizophrenia. Currently, he is working on the study of neural and psychological factors involved in the induction of positive emotions in older people.

María José Cantero is a doctor in psychology and associate professor at the developmental and educational psychology department of the University of Valencia, Spain. Her teaching and research activity is focused in the area of Developmental Psychology, mainly in the field of affective-social development. She teaches several subjects related to Childhood Development, Emotional Bonds, and Affective Education in undergraduate and postgraduate programs at the University of Valencia. She has also supervised more than 80 master’s thesis/final degree projects and six PhD theses. She is author of more than 30 research articles in national and international journals, and several book chapters for leading Spanish publishers. She has participated in several funded research projects and has attended more than 50 scientific conference and congresses. She has been a reviewer of several JCR journals: Annals of Psychology, Acta Colombiana de Psicologia, and Aggression and Violent Behavior. Her main line of research is Developmental Psychology, especially the fields of attachment and parenting. She has designed and validated a questionnaire of adult attachment in the Spanish population and has adapted several instruments for assessing the quality of child attachment.

References

Alley

Chae

Cordon

Kalomiris

Goodman

G. S.

(2015). Child maltreatment and autobiographical memory development: Emotion regulation and trauma-related psychopathology. In Watson

L. A.

Bernsten

(Eds.), Clinical perspectives on autobiographical memory (pp. 85–106). Cambridge University Press.

Barry

T. J.

Lenaert

Hermans

Raes

Griffith

J. W.

(2018). Meta-analysis of the association between autobiographical memory specificity and exposure to trauma. Journal of Traumatic Stress, 31, 35–46. https://doi.org/10.1002/jts.22263

Beck

A. T.

(1967). Depression: Clinical, experimental, and theoretical aspects. Harper & Row.

Benson

Hulac

D. M.

Bernstein

J. D.

(2013). An independent confirmatory factor analysis of the Wechsler Intelligence Scale for Children Fourth Edition (WISC-IV) Integrated: What do the process approach subtests measure? Psychological Assessment, 25, 692–705. https://doi.org/10.1027/a0032298

Brown

D. W.

Anda

R. F.

Edwards

V. J.

Felitti

V. J.

Dube

S. R.

Giles

W. H.

(2007). Adverse childhood experiences and childhood autobiographical memory disturbance. Child Abuse & Neglect, 31, 961–969.

Bryant

R. A.

Sutherland

Guthrie

R. M.

(2007). Impaired specific autobiographical memory as a risk factor for posttraumatic stress after trauma. Journal of Abnormal Psychology, 116, 837–841. https://doi.org/10.1037/0021-843X.116.4.837

Crane

Heron

Gunnell

Lewis

Evans

Williams

J. M. G.

(2014). Childhood traumatic events and adolescent overgeneral autobiographical memory: Findings in a UK cohort. Journal of Behavior Therapy and Experimental Psychiatry, 45, 330–338. https://doi.org/10.1016/j.jbtep.2014.02.004

Cromheeke

Herpoel

L. A.

Mueller

S. C.

(2014). Childhood abuse is related to working memory impairment for positive emotion in female university students. Child Maltreatment, 19, 38–48. https://doi.org/10.1177/1077559513511522

Dalgleish

Williams

Golden

Perkins

Barrett

Barnard

(2007). Reduced specificity of autobiographical memory and depression: The role of executive control. Journal of Experimental Psychology: General, 136, 23–42. https://doi.org/10.1037/0096-3445.136.1.23

10.

De Brito

S. A.

Viding

Sebastian

C. L.

Kelly

P. A.

Mechelli

Maris

McCrory

E. J.

(2013). Reduced orbitofrontal and temporal grey matter in a community sample of maltreated children. Journal of Child Psychology and Psychiatry, 54, 105–112. https://doi.org/10.1111/j.1469-7610.2012.02597.x

11.

Del Barrio

Moreno

López

(1999). El Children’s Depression Inventory, CDI. Su aplicación a la población española [The Children’s Depression Inventory, CDI. It’s application to spanish population]. Clínica y Salud, 10, 393–416.

12.

Goodman

G. S.

Quas

J. A.

Ogle

C. M.

(2010). Child maltreatment and memory. Annual Review of Psychology, 61, 325–351. https://doi.org/10.1146/annurev.psych.093008.100403

13.

Green

J. G.

McLaughlin

K. A.

Berglund

P. A.

Gruber

M. J.

Sampson

N. A.

Zaslavsky

A. M.

, et al. (2010). Childhood adversities and adult psychiatric disorders in the National Comorbidity Survey Replication I: Associations with first onset of DSM-IV disorders. Archives of General Psychiatry, 67, 113–123.

14.

Griffith

J. W.

Claes

Hompes

Vrieze

Vermote

Debeer

Lenaert

Raes

Hermans

(2016). Effects of childhood abuse on autobiographical memory in current major depressive disorder. Cognitive Therapy and Research, 40, 774–782. https://doi.org/10.1007/s10608-016-9784-y

15.

Haase

R. F.

Ellis

M. V.

(1987). Multivariate analysis of variance. Journal of Counseling Psychology, 34, 404–413. https://doi.org/10.1037/0022-0167.34.4.404

16.

Hermans

de Decker

de Peuter

Raes

Eelen

Williams

J. M. G.

(2008). Autobiographical memory specificity and affect regulation: Coping with a negative life event. Depression and Anxiety, 25, 787–792. https://doi.org/10.1002/da.20326

17.

Hitchcock

Nixon

R. D. V.

Weber

(2014). A longitudinal examination of overgeneral memory and psychopathology in children following recent trauma exposure. Applied Cognitive Psychology, 28, 531–538. https://doi.org/10.1002/acp.3027

18.

Huberty

C. J.

Morris

J. D.

(1989). Multivariate analysis versus multiple univariate analyses. Psychological Bulletin, 105, 302–308. https://doi.org/10.1037/0033-2909.105.2.302

19.

Kleim

Ehlers

(2008). Reduced autobiographical memory specificity predicts depression after posttraumatic stress disorder after recent trauma. Journal of Consulting and Clinical Psychology, 76, 231–242. https://doi.org/10.1037/0022-006X.76.2.231

20.

Kovacs

(1992). Children’s Depression Inventory manual. Multi-Health Systems.

21.

Luby

J. L.

Belden

Harms

M. P.

Tillman

Barch

D. M.

(2016). Preschool is a sensitive period for the influence of maternal support on the trajectory of hippocampal development. Proceedings of the National Academy of Sciences, 113, 5742–5747. https://doi.org/10.1073/pnas.1601443113

22.

Masson

Bussières

East-Richard

R-Mercier

Cellard

(2015). Neuropsychological profile of children, adolescents and adults experiencing maltreatment: A meta-analysis. The Clinical Neuropsychologist, 29, 573–594. https://doi.org/10.1080/13854046.2015.1061057

23.

Masten

C. L.

Guyer

A. E.

Hodgdon

McClure

E. B.

Charney

D. S.

Ernst

Kaufman

Pine

D. S.

Monk

C. S.

(2008). Recognition of facial emotions among maltreated children with high rates of post-traumatic stress disorder. Child Abuse & Neglect, 32, 139–153. https://doi.org/10.1016/j.chiabu.2007.09.006

24.

Maxwell

(2001). When to use MANOVA and significant MANOVAs and insignificant ANOVAs or vice versa. Journal of Consumer Psychology, 10, 1–22930.

25.

McCrory

De Brito

S. A.

Viding

(2010). Research review: The neurobiology and genetics of maltreatment and adversity. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 51, 1079–1095. https://doi.org/10.1111/j.1469-7610.2010.0227.x

26.

Miyake

Friedman

N. P.

(2012). The nature and organization of individual differences in executive functions: Four general conclusions. Current Directions in Psychological Science, 21, 8–14. https://doi.org/10.1177/0963721411429458

27.

Mueller

S. C.

Maheu

F. S.

Dozier

Peloso

Mandell

Leibenluft

… Ernst

(2010). Early-life stress is associated with impairment in cognitive control in adolescence: An fMRI study. Neuropsychologia, 48, 3037–3044. https://doi.org/S0028-3932(10)00242-3

28.

Nelson

C. A.

Zeanah

C. H.

Fox

N. A.

Marshall

P. J.

Smyke

A. T.

Guthrie

(2008). Cognitive recovery in socially deprived young children: The Bucharest early intervention Project. Science, 318, 1937–1940. https://doi.org/10.1126/science.1143921

29.

Ogle

C. M.

Block

S. D.

Harris

L. S.

Goodman

G. S.

Pineda

Timmer

… Saywitz

K. J.

(2013). Autobiographical memory specificity in child sexual abuse victims. Developmental Psychopathology, 25, 321–332. https://doi.org/10.1017/S0954579412001083

30.

Ono

Devilly

G. J.

Shum

D. H. K.

(2016). A meta-analytic review of overgeneral memory: The role of trauma history, mood, and the presence of posttraumatic stress disorder. Psychological Trauma: Theory, Research, Practice, and Policy, 8, 157–164. https://doi.org/10.1037/tra0000027

31.

Raes

Hermans

Williams

J. M.

Eelen

(2007). A sentence completion procedure as an alternative to the Autobiographical Memory Test for assessing overgeneral memory in non-clinical populations. Memory, 15, 495–507. https://doi.org/10.1080/09658210701390982

32.

Raes

Verstraeten

Bijttebier

Vasey

Dalgleish

(2010). Inhibitory control mediates the relationship between depressed mood and overgeneral memory recall in children. Journal of Clinical Child and Adolescent Psychology, 39, 276–281. https://doi.org/10.1080/15374410903532684

33.

Ricarte

J. J.

Navarro

Latorre

J. M.

Ros

Watkins

(2016). Age and gender differences in emotion regulation strategies: Autobiographical memory, rumination, problem solving and distraction. Spanish Journal of Psychology, 19, E43. https://doi.org/10.1017/sjp.2016.46

34.

Ros

Latorre

J. M.

Serrano

J. P.

(2010). Working memory capacity and overgeneral autobiographical memory in young and older adults. Aging, Neuropsychology, and Cognition, 17, 89–107. https://doi.org/10.1080/13825580903042650

35.

Stange

J. P.

Hamlat

E. J.

Hamilton

J. L.

Abramson

L. Y.

Alloy

L. B.

(2013). Overgeneral autobiographical memory, emotional maltreatment, and depressive symptoms in adolescence: Evidence of a cognitive vulnerability–stress interaction. Journal of Adolescence, 36, 201–208. https://doi.org/10.1016/j.adolescence.2012.11.001

36.

Sumner

J. A.

(2012). The mechanisms underlying overgeneral autobiographical memory: An evaluative review of evidence for the CaR-FA-X model. Clinical Psychology Review, 32, 34–48. https://doi.org/10.1016/j.cpr.2011.10.003

37.

Sumner

J. A.

Griffith

J. W.

Mineka

(2010). Overgeneral autobiographical memory as a predictor of the course of depression: A meta-analysis. Behaviour Research and Therapy, 48, 614–625. https://doi.org/10.1016/j.brat.2010.03.013

38.

Valentino

Toth

S. L.

Cicchetti

(2009). Autobiographical memory functioning among abuse, neglected, and nonmaltreated children: The overgeneral memory effect. Journal of Child Psychology and Psychiatry, 50, 1029–1038. https://doi.org/10.1111/j.1469-7610.2009.02072.x

39.

van Lang

N. D.

Ferdinand

R. F.

Verhulst

F. C.

(2007). Predictors of future depression in early and late adolescence. Journal of Affective Disorders, 97, 137–144. https://doi.org/10.1016/j.jad.2006.06.007

40.

Vrielynck

Deplus

Philippot

(2007). Overgeneral autobiographical memory and depressive disorder in children. Journal of Clinical Child and Adolescent Psychology, 36, 95–105.

41.

Wechsler

(2003). WISC-IV: Technical and interpretative manual. Pearson.

42.

Williams

J. M. G.

Barnhofer

Crane

Hermans

Raes

Watkins

Dalgleish

(2007). Autobiographical memory specificity and emotional disorder. Psychological Bulletin, 133, 122–148.