Early manifestations of externalizing behaviors, such as tantrums, disobedience, and aggression (Peterson et al., Reference Peterson, Dando, D’Souza, Waldie, Carr, Mohal and Morton2018), are observed as early as infancy (Perra et al., Reference Perra, Paine and Hay2021). Recent meta-analysis results in children aged 1–7 years (Vasileva et al., Reference Vasileva, Graf, Reinelt, Petermann and Petermann2021) reveal a 10% prevalence of externalizing psychopathology, including attention-deficit hyperactivity disorder, conduct disorder, and oppositional defiant disorder. Failure to address these maladaptive behaviors in their early stages increases the risk of future issues, including substance abuse and violence (Moffit, Reference Moffit2017), antisocial personality disorder (Diamantopoulou et al., Reference Diamantopoulou, Verhulst and van der Ende2010), and criminal records (Bevilacqua et al., Reference Bevilacqua, Hale, Barker and Viner2018). In fact, early externalizing problems have shown to predict criminal convictions two decades later (Kassing et al., Reference Kassing, Godwin, Lochman and Coie2019). Hence, it is essential to prevent the emergence of externalizing behaviors and/or to avoid their persistence at early ages.
Externalizing behaviors result from a complex interplay of genetics, neurological changes, and environmental factors (Rovira et al., Reference Rovira, Demontis, Sánchez-Mora, Zayats, Klein, Mota, Weber, Garcia-Martínez, Pagerols, Vilar-Ribó, Arribas, Richarte, Corrales, Fadeuilhe, Bosch, Martin, Almos, Doyle, Grevet and Ribasés2020), including biological disposition (Dodge & Pettit, Reference Dodge and Pettit2003), parent–child relationship characteristics (Hewitt-Ramírez & Moreno-Méndez, Reference Hewitt-Ramírez and Moreno-Méndez2018), and the interaction between genetics (e.g., temperament) and environmental (e.g., family) elements (Goodnight et al., Reference Goodnight, Donahue, Waldman, Van Hulle, Rathouz, Lahey and D’Onofrio2016). Since interventions and preventive measures are not applicable to genetic factors, a relevant area of attention and intervention is the realm of the parent–child relationship. Parenting is considered a significant contributing factor of the children’s externalizing behavior (Rothbaum & Weisz, Reference Rothbaum and Weisz1994). Thus, parental abilities become a primary target for interventions. Encouraging the development of specific parenting skills can enhance the bond between parents and children and reduce various adverse effects, particularly externalizing problems (e.g., see Cooke et al.’s Reference Cooke, Deneault, Devereux, Eirich, Fearon and Madigan2022, meta-analysis on parental sensitivity and child behavioral problems). Therefore, the parent–child relationship emerges as an explicative factor worth studying in preventing juvenile crime (Webster-Stratton & Reid, Reference Webster-Stratton, Reid, Weisz and Kazdin2018). While the influence of parenting is well established in the case of both externalizing (Pinquart, Reference Pinquart2017a) and internalizing problems (Lin et al., Reference Lin, Kehoe, Pozzi, Liontos and Whittle2024; Pinquart, Reference Pinquart2017b), the present review focuses on children’s and adolescents’ externalizing behaviors since including both internalizing and externalizing problems is beyond the scope of the present review.
From a developmental perspective, infants’ survival hinges on their caregivers’ capacity to interpret their thoughts, feelings, and behaviors to meet their needs (Lavender et al., Reference Lavender, Waters and Hobson2023). Parental mentalization is closely linked to children’s inner growth, assisting them in managing emotional conflicts and promoting positive psychological development (Caldarera et al., Reference Caldarera, Vitiello, Turcich, Bechis and Baietto2022). In fact, there is evidence showing the links between parental mentalization and positive parenting behaviors (for a systematic review, see Stuhrmann et al., Reference Byrne, Murphy and Connon2022). Specifically, parental mentalization has been robustly related to secure child–parent attachment (Zeegers et al., Reference Zeegers, Colonnesi, Stams and Meins2017), which, in turn, has been strongly associated with a lower incidence of externalizing problems (based on meta-analyses for results with mothers by Fearon et al., Reference Fearon, Bakermans-Kranenburg, Van IJzendoorn, Lapsley and Roisman2010, and with fathers by Deneault et al., Reference Deneault, Bakermans-Kranenburg, Groh, Fearon and Madigan2021).
Despite the existent growing number of studies in the field, there are discrepancies regarding the consistency of the relationship between parental mentalization and children’s externalizing behaviors. Specifically, research findings show varying degrees of such association. For example, research shows that parental mentalization and externalizing problems are positively associated (Meins et al., Reference Meins, Centifanti, Fernyhough and Fishburn2013, Wade et al., Reference Wade, Plamondon and Jenkins2021), weakly negatively related (e.g., Dollberg et al., Reference Dollberg, Hanetz-Gamliel and Levy2023), or strongly inversely correlated (e.g., Khoshroo & Seyed Mousavi, Reference Khoshroo and Seyed Mousavi2022). Such disparities across studies reveal the need of exploring potential reasons behind them. Therefore, the meta-analytic technique, including the use of meta-regression, is instrumental in overcoming inconsistencies in results. The current study represents the first meta-analysis that synthesizes research examining the issue.
Parental mentalization linked to children’s externalizing problems
Overall, mentalization is a dynamic and developmental capacity mainly context- and relationship-specific. Parental mentalization primarily emerges within the context of secure attachment relationships, serving as a key construct in children’s development and beneficial parent–child relationships (Sharp & Fonagy, Reference Sharp and Fonagy2008). Within that framework, parental mentalization is a psychological mechanism that facilitates parental self-regulation and the co-regulation of the child’s emotional states (Stuhrmann et al., Reference Stuhrmann, Göbel, Bindt and Mudra2022). It encompasses parents’ capacity to perceive their child as a psychological entity, differentiating between their children’s mental experiences and their own. Parental ability to comprehend not only their child’s external behaviors but also the internal mental activity (e.g., motives and emotions) that drive those behaviors (Medrea & Benga, Reference Medrea and Benga2021), is closely associated to responding to the child appropriately through affectively attuned parenting behavior (Kelly et al., Reference Kelly, Slade and Grienenberger2005).
In contrast, parents with lower mentalizing capacities may struggle to perceive their children’s mental states, failing to comprehend the motivations behind their children’s actions and to meet their needs, especially in challenging situations. Consequently, this lower mentalizing capacity may result in punitive responses exacerbating the child’s distress and dysregulation (Dollberg et al., Reference Dollberg, Hanetz-Gamliel and Levy2021; Ensink, Bégin et al., Reference Ensink, Bégin, Normandin, Godbout and Fonagy2017). Evidence suggests that poor parental mentalization is correlated with heightened negative emotionality (Smaling et al., Reference Smaling, Huijbregts, van der Heijden, van Goozen and Swaab2016), increased tendencies towards controlling behaviors (Suardi et al., Reference Suardi, Moser, Sancho Rossignol, Manini, Vital, Merminod, Kreis, Ansermet, Serpa and Schechter2020), and with issues in socioemotional development (Nijssens et al., Reference Nijssens, Vliegen and Luyten2020).
In fact, when it comes to the relationship with children’s externalizing behaviors, it has been found that high parental mentalization can contribute to the prevention of the onset of these problems (Suardi et al., Reference Suardi, Moser, Sancho Rossignol, Manini, Vital, Merminod, Kreis, Ansermet, Serpa and Schechter2020), and poor parental mentalization is associated with higher externalizing behaviors (Khoshroo & Seyed Mousavi, Reference Khoshroo and Seyed Mousavi2022). Nevertheless, there are some studies that have reported a null (Zeegers et al., Reference Zeegers, Colonnesi, Noom, Polderman and Stams2020) or even a positive relationship (Kochanska & An, Reference Kochanska and An2023; Wade et al., Reference Wade, Plamondon and Jenkins2021) between parental mentalization and externalizing problems. Therefore, it is evident the need to clarify the association between these variables. In view of the above, we expect a negative relationship. Yet, several circumstances need to be considered in the analysis. A brief account of possible moderators follows.
Parental mentalization: operationalization approaches
Mentalization consists of four primary dimensions: Aspect (automatic vs. controlled), Focus (internal vs. external), Person (self vs. other), and Mode (cognitive vs. affective), being essential that these dimensions maintain a balanced interaction for effective mentalization (Bateman & Fonagy, Reference Bateman, Fonagy, Kealy and Ogrodniczuk2019). Parental mentalization is investigated through three approaches – mind-mindedness (MM; Meins, Reference Meins2013), parental reflective functioning (PRF; Slade et al., Reference Slade, Grienenberger, Bernbach, Levy and Locker2005), and insightfulness (Oppenheim & Koren-Karie, Reference Oppenheim and Koren-Karie2002) – all focusing on parents’ comprehension of their children’s internal states (Zeegers et al., Reference Zeegers, Colonnesi, Stams and Meins2017). Each one operationalizes caregivers’ capacities differently and offers a unique perspective on comprehending parent–child dialogue and interaction (Medrea & Benga, Reference Medrea and Benga2021). A simple way to understand how to operationalize parental mentalization is conceptualizing it, following the terminology of Meins and Fernyhough (Reference Meins and Fernyhough2015), as ‘offline’ (representational) and ‘online’ (interactional) mentalizing abilities.
MM is the only measure that can be evaluated from both online and offline perspectives, while PRF and insightfulness are operationalized in a representational way (offline). Appropriate MM can be measured in: (a) parent–child interactions, where parents provide accurate and spontaneous comments on the child’s mental state, matching the infant’s experience or behavior (online) and (b) via the caregiver’s description of the child in terms of mental attributes (offline; Dollberg, Reference Dollberg2022). PRF can be evaluated by coding an interview (e.g., the Parent Development Interview: PDI; Slade et al., Reference Slade, Grienenberger, Bernbach, Levy and Locker2005; Sleed et al., Reference Sleed, Slade and Fonagy2020), or via self-report using a questionnaire (e.g., The Parental Reflective Functioning Questionnaire: PRFQ; Carlone et al., Reference Carlone, Milan, Decoste, Borelli, McMahon and Suchman2023; Luyten et al., Reference Luyten, Mayes, Nijssens and Fonagy2017). Within the PRFQ, the dimensions include pre-mentalizing (PM), certainty about mental states (CMS), and interest and curiosity (IC), with adequate mentalization characterized by low PM and average CMS and IC scores. The PM subscale, which assesses distortions in mentalizing, reveals particularly detrimental effects of serious mentalization impairments, and is the most closely related dimension to emotional unavailability and insecure attachment (Luyten et al., Reference Luyten, Mayes, Nijssens and Fonagy2017). Insightfulness is assessed through caregivers’ reflections on their relationship with the child in response to questions about the child’s thoughts and feelings (Oppenheim & Koren-Karie, Reference Oppenheim and Koren-Karie2002).
Nevertheless, the strength and direction among these three approaches (MM, PRF, insightfulness) and children’s externalizing problems seems to vary across studies. For instance, the correlation between MM and externalizing problems has been reported as strong (Camisasca et al., Reference Camisasca, Miragoli, Ionio, Milani and Di Blasio2018), weak (Dollberg et al., Reference Dollberg, Hanetz-Gamliel and Levy2023), or nonexistent (Zeegers et al., Reference Zeegers, Colonnesi, Noom, Polderman and Stams2020). Similarly, PRF’s association with externalizing has been observed as strongly negative (Khoshroo & Seyed Mousavi, Reference Khoshroo and Seyed Mousavi2022) or positive (Borelli et al., Reference Borelli, Yates, Hecht, Cervantes, Russo, Arreola, Leal, Torres and Guerra2021). Lastly, analyses of the relationship between insightfulness and externalizing problems have showed positive (Gray et al., Reference Gray, Forbes, Briggs-Gowan and Carter2015) and negative associations (Feniger-Schaal & Koren-Karie, Reference Feniger-Schaal and Koren-Karie2021).
Therefore, in view of the aforementioned discrepancies between studies, further research is required. Thus, the present meta-analysis will explore whether MM, PRF or insightfulness has stronger power to predict the expected negative relationship between parental mentalization and children’s externalizing problems. We anticipate the relationship between parental mentalization and externalizing problems to vary depending on whether it is assessed representationally or through actual interactions. Some parents may possess adequate mentalization skills at the representational level (assessed via PRF), but struggle to demonstrate them in real-time interactions, when measured online through their verbalizations about the child’s mental states during direct play (Shai et al., Reference Shai, Dollberg and Szepsenwol2017). Hence, we expect that, when parental mentalization is analyzed in real-time (online) interactions, the relationship between mentalization and externalizing problems will be stronger than when the measure is a ‘representational’ approach (offline).
Children’s developmental stage
Following normative developmental trajectories, externalizing problems decrease from early childhood to preadolescence, increase during adolescence, and then decrease again from late adolescence into adulthood (Petersen et al., Reference Petersen, Bates, Dodge, Lansford and Pettit2015). Yet the impact of parental mentalization on reducing these behaviors may vary depending on the child’s developmental stage. For example, some studies have found that the relationship between parental mentalization and externalizing problems in adolescents is close to zero (e.g., Benbassat & Priel, Reference Benbassat and Priel2012; Borelli et al., Reference Borelli, Palmer, Vanwoerden and Sharp2019). However, it has been found to be negative during infancy (e.g., Salo et al., Reference Salo, Lipsanen, Sourander, Pajulo and Kalland2022; Smaling et al., Reference Smaling, Huijbregts, van der Heijden, van Goozen and Swaab2016), in young children (e.g., Colonnesi et al., Reference Colonnesi, Zeegers, Majdandžić, van Steensel and Bögels2019; Dollberg et al., Reference Dollberg, Hanetz-Gamliel and Levy2021) and during middle childhood (e.g., Condon et al., Reference Condon, Holland, Slade, Redeker, Mayes and Sadler2019; Ensink et al., Reference Ensink, Bégin, Normandin and Fonagy2016). Moreover, a study comparing both early and middle childhood found a stronger negative effect between parental mentalization and externalizing problems in the older group (Khoshroo & Seyed Mousavi, Reference Khoshroo and Seyed Mousavi2022). Thus, younger children may require higher parental mentalization skills to effectively regulate their emotions and behaviors. Considering the above findings, we may posit a stronger negative relationship between mentalization and externalizing problems during middle childhood and a smaller effect during early childhood and adolescence.
Parenting experience
Given the inconsistency of research evidence, we suggest that parenting experience may be another moderator. While certain studies involving first-time parents revealed a modest association (Salo et al., Reference Salo, Pajulo, Vinzce, Raittila, Sourander and Kalland2021), research involving experienced caregivers uncovered a higher correlation (Lunn et al., Reference Lunn, Lewis and Gannon2019). Hence, we hypothesize that the negative relationship between parental mentalization and children’s externalizing problems will be stronger in experienced parents as compared to first-time parents.
Population type
Given the processes underlying mentalization, psychopathology will invariably entail challenges related to mentalizing skills (Bateman, Reference Bateman2022). As a matter of fact, in a recent systematic review and meta-analysis (Georg et al., Reference Georg, Meyerhöfer, Taubner and Volkert2023), parental depression has been associated to lower parental mentalization. Yet, the relationship between parental mentalization and children’s externalizing problems might differ between clinical and nonclinical populations. In fact, there is research evidence showing a likely stronger effect of parental mentalization in clinical/at risk samples (e.g., Hughes et al., Reference Hughes, Aldercotte and Foley2017). Specifically, maternal mentalization has been found to have a stronger protective role in mitigating externalizing symptoms in challenging situations or with at-risk families, as opposed to general community samples (Meins et al., Reference Meins, Centifanti, Fernyhough and Fishburn2013). Therefore, we expect the strength of the negative association between parental mentalization and externalizing behaviors to be more pronounced in clinical or at-risk samples compared to studies involving the general community.
Parental education
Research exploring various education levels has shown differential associations of parental mentalization with child externalizing problems. MM was a predictor of child externalizing problems at childhood, but only with parents with no secondary education (Meins et al., Reference Meins, Centifanti, Fernyhough and Fishburn2013). However, some studies reported no association between parental mentalization and externalizing behavior when controlling for parental education (e.g., Colonnesi et al., Reference Colonnesi, Zeegers, Majdandžić, van Steensel and Bögels2019; Dollberg et al., Reference Dollberg, Hanetz-Gamliel and Levy2021). Given the mixed nature of the evidence, we refrain from proposing a specific hypothesis regarding this matter.
Reporting figure
From a methodological perspective, the reporting figure may also affect the relationship between parental mentalization and behavior. There may be differences between parents’ and teachers’ reports of children’s externalizing problems, as these informants rely on distinct experiences (Deneault et al., Reference Deneault, Hammond and Madigan2023). In fact, a meta-analysis reported that parents often perceive more externalizing behaviors than teachers (Carneiro et al., Reference Carneiro, Soares, Rescorla and Dias2021). In addition, when the same observer evaluates two different variables, the correlation between them may increase due to covariance bias (Hoyt, Reference Hoyt2000). Thus, it is reasonable to think that the relationship between parental mentalization and children’s externalizing problems would be stronger when the primary caregivers are the reporting figure of the latter variable (as opposed to teachers or other reporting figures).
Aims of the meta-analysis
In summary, the evidence that parental mentalization is associated with a decrease in children’s externalizing problems is inconclusive. Therefore, one of the objectives of this meta-analysis is to confirm the type of relationship between these variables. We also seek to determine the magnitude of this association, as well as to investigate factors that may moderate this relationship. The present meta-analysis aims at examining the relationship among infants, children, and adolescents. It will consider the following moderators: (a) different approaches for the operationalization of parental mentalization (MM, PRF or insightfulness), (b) children’s developmental stage (early childhood, middle childhood or adolescence), (c) parenting experience (first-time vs. experienced parenting), (d) sample type (clinical/at-risk vs. community), (e) parental education, and (f) reporting figure (primary caregiver vs. others).
The main hypothesis is that higher parental mentalization will be associated with fewer externalizing problems in children and adolescents. Furthermore, we expect a stronger negative association between higher levels of parental mentalization and externalizing problems in real time interactions (vs. representational approaches), in middle childhood (vs. younger children and adolescents), in experienced (vs. first-time) parenting family situations, in clinical/at-risk (vs. community), and in primary caregiver (vs. others’ reports). Lastly, based on conflicting evidence when examining the possible moderating effects of parental education, we will take an exploratory approach regarding this question.
Methods
Inclusion and exclusion criteria
This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl, Brennan, Chou, Glanville, Grimshaw, Hróbjartsson, Lalu, Li, Loder, Mayo-Wilson and Moher2021). For their inclusion, studies had to meet the following criteria: (a) written in English or Spanish; (b) empirical in nature; (c) reported in peer-reviewed journal articles; and (d) included children aged between 0 and 18 years in the sample. Studies were excluded if: (a) they provided insufficient data for effect size analysis, and additional data could not be obtained from the respective authors; (b) they did not evaluate parental mentalization;Footnote 1 (c) they did not assess children’s externalizing behavior in the age range under analysis (0-18 years); (d) they did not investigate the relationship between parental mentalization and child externalizing symptoms; or (e) were dissertations or book chapters.
Search strategy criteria and selection of studies
The initial search was performed in June 2021; it was updated in January 2024 to retrieve documents included in databases until December 2023. First, six electronic databases were searched to identify records related to parental mentalization and externalizing problems. These databases included APA PsycArticles, APA PsycInfo, PsycBooks, PSICODOC, Scopus, and Web of Science. The following combination of terms was used in the abstract, keywords, and subjects: Mentali* OR reflective function* OR mind-mind* OR mind-related OR insightful AND external* OR internal* OR behav* disorder* OR behav* difficult* OR behav* problem* OR conduct disorder* OR conduct* difficult* OR conduct problem* OR disruptive behav* OR child* problem OR antisocial behav* OR ADHD OR CBCL. Initially, we retrieved 5,442 records; eventually, we included 42 studies (N = 7,761 participants) with 52 effect sizes in the meta-analysis (see Figure 1).
Flow diagram.
To ensure consistence and replicability of the meta-analysis, a random sample of every fifth screened paper was independently coded by the second author; disagreements were resolved by having the first author and the second author review the article in question and jointly come to a consensus regarding the appropriate coding. Subsequently, the first and the second author independently reviewed all full-text level records. A high agreement was obtained (Kappa = 0.98). The third author resolved discrepancies.
The first and the third authors independently assessed the risk of bias of each of the included studies and discussed their assessments to achieve consensus. An adapted version of the Newcastle–Ottawa Scale for cross-sectional studies (NOS; Herzog et al., Reference Herzog, Álvarez-Pasquin, Díaz, Del Barrio, Estrada and Gil2013; Wells et al., Reference Wells, Shea, O’Connell, Peterson, Welch, Losos and Tugwell2000) was used to evaluate the quality of the methodology used in the included studies. The NOS is a checklist that assesses the appropriateness of research design, the recruitment strategy, the representativeness of sample, the objectivity/reliability of outcome determination, the power calculation provided, and appropriate statistical analyses. Score disagreements were resolved by consensus and a final agreed-upon rating was assigned to each study (see S1 of Supplementary Online Materials).
Finally, the data extraction process involved two stages. The first author retrieved the necessary data for analysis (see Table 1), and the third author independently checked the values. Any disagreements were resolved by all authors.
Information of the studies included in the meta-analysis
Notes. BASC − 2 = Behavior Assessment Scale for Children−Second Edition; BITSEA = Brief Infant−Toddler Social and Emotional Assessment; CBCL = Child Behavior Checklist; ECBI = The Eyberg Child Behavior Inventory; FMSS = Five-Minute Speech Sample; ICQ = Infant’s Characteristics Questionnaire; ITSEA = Infant−Toddler Social and Emotional Assessment; nd = No data; Ohio = Ohio Youth Problem Functioning, and Satisfaction Scales; PDI = Parent Development Interview; SCBE − 30 = Social Competence and Behavior Evaluation − 30; SDQ = Strengths and Difficulties Questionnaire; TRF = Teacher Report Form; YSR = Youth Self−Report.
Analysis plan
Our primary analysis included 52 effect sizes assessing parental mentalization and externalizing problems (see Table 1 for overall effect sizes and study characteristics). The following moderators were examined: approaches for the operationalization of parental mentalization (MM, PRF, or insightfulness), children’s developmental stage (early childhood, middle childhood, or adolescence), parenting experience (first-time vs. experienced), the sample type (clinical/at-risk vs. community), and reporting figure (primary caregiver vs. others). Only categorical moderator variables that had at least five contrasts in the categories were used (Bakermans-Kranenburg et al., Reference Bakermans-Kranenburg, Van Ijzendoorn and Juffer2003); thus, we were unable to analyze parental insightfulness (k = 3). The data from parental education was noncomparable among the studies to perform moderation analyses (i.e., studies reporting the percentages of participants’ educational attainment). When studies reported both maternal and paternal correlations, we chose the maternal correlation for consistency, as more studies focused on maternal mentalization. In longitudinal studies and interventions, we selected the effect size reported at baseline. In order to maintain a comparable number of studies, the operationalization of PRF was prioritized when both PRF and MM were reported, aiming to balance moderation analyses.
Taking into account the three dimensions of PRFQ, we have not considered the CMS or the IC subscales. The reason is that the optimal levels of both subscales are at the midpoint; consequently, a curvilinear analysis might be more appropriate, but it complicates the linear regression required for meta-analytic techniques. Therefore, we have chosen the PM correlation in order to perform the analyses. Accordingly, correlation scores were reverted because it assesses a non-mentalizing stance, indicating that higher scores correspond to lower mentalizing abilities. Moreover, to test whether the relationship between mentalization and externalizing problems differed across the various approaches for the operationalization of mentalization, we compared effect sizes between offline vs. online measures. To define age groups, we have categorized individuals into early childhood (0-5 years), middle childhood (6-12 years), and adolescence (13-18 years) (see S2 of Supplementary Online Materials).
The analysis was conducted using R v4.2.2 (R Core Team, 2022) and the metafor package v4.0.0. Random effects model was chosen due to its ability to account for the excess variance (Viechtbauer, Reference Viechtbauer2010). Effect sizes (Pearson’s r) were computed to have the same statistic depicting the relationship between parental mentalization (MM or PRF) and externalizing problems. Standard formulas allowed for the transformation of diverse effect sizes, such as group differences (Wilson, Reference Wilson2023) and regression coefficients (Peterson & Brown, Reference Peterson and Brown2005). In order to reduce variability, pooled effect sizes were computed with corresponding 95% confidence intervals (Borenstein & Hedges, Reference Borenstein, Hedges, Cooper and Hedges2019).
Results
Characteristics of the studies
The following lines provide a summary of some noteworthy aspects of the included research in the meta-analyses. A total of 42 studies with 52 effect sizes were incorporated in the meta-analysis. The investigations were conducted in various countries, including Australia, Belgium, Canada, Finland, Israel, Italy, the Netherlands, Switzerland, Türkiye, the UK, and the USA. The meta-analysis included 7,761 caregivers and their children. Most studies (k = 38) included birth parents, while a few involved other caregivers (k = 4), such as adoptive and foster parents. Twenty-six studies included participants from the community, nine studies used clinical or risk samples (i.e., children who experienced abuse or low socioeconomic families), and seven studies compared both sample types. Intervention programs were included in seven research papers. Studies analyzing parental mentalization as a mediator between parent/child mental health (i.e., parental depression or psychopathological impact of sexual abuse in children) on externalizing problems found higher levels of this ability related to fewer externalizing problems. This suggests both direct and indirect effects of parental mentalization on externalizing behaviors (see S3 of Supplementary Online Materials for details of variables analyzed in the studies).
Meta-analytic effects
Effect sizes (Pearson’s r) were computed to examine the relationship between parental mentalization and externalizing problems. The coefficient based on the random-effects model of r was −.19 (95% [CI: −.24, −.13]), ranging from −.56 to .23, with most estimates being negative (83%). As expected, there was a negative relationship between parental mentalization and externalizing problems. This indicates a small to moderate negative relationship between the variables, implying that higher parental mentalization is associated with fewer externalizing problems observed in children. Figure 2 displays a forest plot illustrating observed outcomes and estimates.
Forest plot.
The Q-test suggests heterogeneity among the true outcomes (Q(51) = 383.760, p = .001, τ2 = .029, I 2 = 83.750), denoting that a significant proportion of the variability in effect sizes is due to differences between studies rather than sampling error. The high I2 value suggests that approximately 84% of the total variation across studies is attributable to heterogeneity. Consequently, despite the average outcome being estimated as negative, some studies may indicate a positive true outcome (e.g., higher mentalization, elevated externalizing problems). Figure 3 presents a funnel plot of the estimates. Neither the rank correlation (p = .894) nor the regression (p = .849) test indicated any funnel plot asymmetry, suggesting no publication bias.
Funnel plot.
We conducted further exploration of the possible sources of heterogeneity through moderation analyses. The mixed-effects model indicated that the association between parental mentalization and externalizing behaviors varied significantly across different developmental stages. As hypothesized, it was stronger for the middle childhood category (r = −.328, p < .001) than for younger children (r = −.112, p = .002), but was nonsignificant in the adolescence group (r = −.129, p = .065). Also, in line with our expectations, parenting experience also turned out to be a significant moderator, such that the inverse relationship between parental mentalization and children’s externalizing problems was greater in the case of experienced parents (r = −.208, p < .001). In the rest of the moderation analyses – (a) different approaches for the operationalization of parental mentalization (MM or PRF), (b) sample type (clinical/at-risk or community), and (c) reporting figure (primary caregiver vs. others) –, there were no differences between the levels of the moderators (see Table 2). Therefore, our hypotheses in relation to these variables were not confirmed.
Statistical test for heterogeneity and moderators included in the meta-analysis
Notes. CI = confidence Interval, QE = Q-statistic for residual heterogeneity, QM = Q-statistic for moderators.
Discussion
The purpose of this study was to review the literature on the relationship between parental mentalization and children’s externalizing problems, analyzing results from a quantitative perspective. To the best of our knowledge, this is the first attempt at summarizing the empirical results on the topic. The included studies were conducted in different countries, predominantly involved biological parents with their children, and were drawn from a combination of community and clinical/at-risk samples. We observed that the expected negative association between parental mentalization and child externalizing problems was predominantly consistent across studies. The magnitude of the effect size may be considered small to moderate (Funder & Ozer, Reference Funder and Ozer2019), being slightly stronger than similar meta-analyses in the field, such as those reviewing the predictive power of parental sensitivity on children’s behavior problems – e.g., r = −.14 in Cooke et al.’s (Reference Cooke, Deneault, Devereux, Eirich, Fearon and Madigan2022) and r = −.08 in Rodrigues et al.’s (Reference Rodrigues, Sokolovic, Madigan, Luo, Silva, Misra and Jenkins2021).
Thus, it seems that parental ability to understand a child’s mental states allows caregivers to better respond to the child’s needs, likely prompting a decrease in externalizing problems and confirming the relevance of paying special attention to parenting mentalization. This finding reveals substantive effects that underline the importance of intervening with parents on mentalization as a useful strategy that can contribute to prevent later children’s externalizing problems.
Beyond the overall negative association between parental mentalization and externalizing problems in offspring, the significant heterogeneity in this relationship suggests the need to consider potential moderators. With respect to the possible moderators examined, the different approaches and measures to parental mentalization did not reveal significant effects. Children’s developmental stage (early and middle childhood) and parenting experience (experienced parenting vs. first-time parents) were found to strengthen the negative relationship between parental mentalization and externalizing problems. Population type (clinical/at risk vs. community), and reporting figure (primary caregivers vs. others) failed to reach statistical significance. Insightfulness and parental education could not be analyzed due to lack of data. Below we outline some reflections on the possible meaning of our results.
Contrary to what expected, different operationalization of parental mentalization (offline-representational vs. online-interactional) did not moderate the relationship between parental mentalization and externalizing problems. Although different approaches to parental mentalization are reported in the literature (e.g., Medrea & Benga, Reference Medrea and Benga2021), our results suggest that these may be closer to each other. This may indicate that parental mentalization might be a singular construct that can be operationalized in various ways. In line with our results, Yatziv et al. (Reference Yatziv, Kessler and Atzaba-Poria2020) found PRF (offline-representational) and MM (online-interactional) to be positively correlated, that is, mothers who tended to spontaneously describe their child in a mentalistic manner during interactions (MM) were more likely to have more elaborated representations of their child’s mind (PRF). Hence, both operationalizations appear to explore the common domain of the parent–child relationship through the lenses of parent’s mentalizing abilities.
In line with our expectations, children’s developmental stage moderated the negative relationship between parental mentalization and externalizing behaviors. These results show that parental mentalization is inversely associated with externalizing problems both in the early and middle childhood age groups as compared to adolescence. In addition, the relationship was stronger during middle childhood as compared to early childhood. Differences in developmental competencies and in the relative weight of contextual influences among these three age groups may explain these results.
Even though in the context of younger children there is a significant reduction in heterogeneity, child’s temperament may play an important role in predicting externalizing symptoms (Frick & Morris, Reference Frick and Morris2004; Muris & Ollendick, Reference Muris and Ollendick2005; Putnam et al., Reference Putnam, Ellis, Rothbart, Eliasz and Angleitner2001), and temperament influences seem to be stronger in the early years (Bates et al., Reference Bates, Schermerhorn, Petersen, Zentner, Shiner, Zentner and Shiner2012). Thus, the effect of child temperament might be associated with increased attenuation of the relationship between parental mentalization and less externalizing behaviors in the younger children group. Furthermore, greater competences in different domains of development in middle childhood (vs. early childhood) may be responsible for the stronger negative relationship between parental mentalization and externalizing problems in this age group. For example, some characteristic advances in cognitive competences in middle childhood underlie the increase in socio-cognitive skills typical of this stage (Collins et al., Reference Collins, Madsen, Susman-Stillman and Bornstein2013), such as children’s mentalization capacities. Middle childhood can be considered a key phase in the development of lifelong mentalization because it can only confidently be stated that a normally developing child has a full and explicit mentalization capacity from the age of six onwards (Frith & Frith, Reference Frith and Frith2003). Since children’s and adolescents’ own mentalization has been associated with externalizing problems (Midgley et al., Reference Midgley, Ensink, Lindqvist, Malberg, Muller, Lindqvist, Malberg and Muller2017), better mentalizing equipment in middle childhood (as compared to the earlier stage) may support a stronger association between parental mentalization and children’s externalizing behaviors.
In addition to this tentative explanation, and closely related to mentalization, deficits in self-regulation have also been linked to externalizing problems (Eisenberg et al., Reference Eisenberg, Hernández and Spinrad2017, Perry et al., Reference Perry, Calkins, Dollar, Keane and Shanahan2018). There is longitudinal evidence of age-related increases in self-regulation, especially from early to middle childhood. Specifically, in Raffaelli et al.’s (Reference Raffaelli, Crockett and Shen2005) study, a significant increase in self-regulation only was observed between early and middle childhood, whereas the difference between middle childhood and early adolescence was not significant. Therefore, developmental differences in self-regulation may contribute to the relationship between parental mentalization and fewer externalizing problems being stronger in middle than in early childhood.
Regarding the absence of a link between parental mentalization on adolescents’ externalizing problems, peer influences may partly account for this finding. Although there is an increase of the influence of peers and other adults outside the family sphere on children’s development in middle childhood, it is not until early adolescence that contact with peers dominate their social networks (Steinberg & Silk, Reference Steinberg, Silk and Bornstein2013). In fact, during adolescence peer influence extends particularly to externalizing behaviors (Giletta et al., Reference Giletta, Choukas-Bradley, Maes, Linthicum, Card and Prinstein2021). Adolescents tend to influence each other’s externalizing problems within the same social circle, often mirroring and adopting behaviors over time through mechanisms such as encouragement, emulation, and peer pressure (Fortuin et al., Reference Fortuin, van Geel and Vedder2015). Specifically, mild externalizing problems significantly affect adolescents’ likability within their peer groups, leading them to align their behaviors with those of their peers. Hence, in the case of adolescence, the influence of peers may be more significant, thus diminishing the importance of the negative relationship between parental mentalization and externalizing problems. Other influences may refer to other parental skills, where harsh and psychological control are associated with more externalizing problems over time (Pinquart, Reference Pinquart2017a), while parental monitoring predicted lower externalizing behaviors (Bailey et al., Reference Bailey, Hill, Oesterle and Hawkins2009; Lopez-Tamayo et al., Reference Lopez-Tamayo, LaVome Robinson, Lambert, Jason and Ialongo2016). Such findings may indicate that, during adolescence, both positive and negative parenting may exert a stronger effect on externalizing behaviors than parental mentalization.
Overall, other factors (i.e., temperament during early childhood, increased developmental competences in middle childhood, and peer influence and/or parental monitoring during adolescence) may diminish the effect of parental mentalization on children’s externalizing problems. These third variables may explain the stronger decrease of externalizing problems in middle childhood as compared to the other two age groups. In support of this idea, Lo and Wong’s (Reference Lo and Wong2022) meta-analysis demonstrates that interventions designed to enhance PRF were more effective for parents of school-aged children compared to parents of infants. Thus, interventions to enhance parental mentalization in order to prevent or reduce children’s externalizing problems, should be precisely directed to the middle childhood age group. In this regard, given that low-quality studies seem to prevail in middle childhood research (Midgley et al., Reference Midgley, Sprecher and Sleed2021), it is imperative to attain a solid foundation for mentalization-based interventions at this stage of development.
In line with our assumption, parenting experience also moderated the association between parental mentalization and children’s externalizing problems, indicating that in the case of experienced parents the negative relationship is stronger. In other words, with the experience gained in raising previous children (Whiteman et al., Reference Whiteman, McHale and Crouter2003), parental mentalization exerts a stronger effect in diminishing externalizing problems. Moreover, other parental characteristics such as parental stress (Mak et al., Reference Mak, Yin, Li, Cheung and Oon2020), low parental self-efficacy (Bodalski et al., Reference Bodalski, Joshua Bradley, Neger, Shi, Bridges and Flory2023), and low marital satisfaction (Vaez et al., Reference Vaez, Indran, Abdollahi, Juhari and Mansor2015) are associated with children’s externalizing symptoms and may influence the relationship between the target variables of this meta-analysis. In fact, some studies reveal the above factors to be typically present in first-time parents (i.e., low parental self-efficacy in Amin et al., Reference Amin, Tam and Shorey2018; high parental stress in Lévesque et al., Reference Lévesque, Bisson, Charton and Fernet2020, and low marital satisfaction in Bogdan et al., Reference Bogdan, Turliuc and Candel2022). Hence, these parental characteristics may affect the relationship between parental mentalization and externalizing problems. Consequently, in order to reduce externalizing problems through improving parental mentalization skills, practitioners may deliver their interventions focusing on the experienced parents.
We hypothesized that sample type would influence the relationship between parental mentalization and children’s externalizing problems. Surprisingly, our analysis revealed no significant impact of the population type on this association. This aligns with recent meta-analyses investigating parental sensitivity and behavior problems, which similarly found no substantial moderating effect when comparing clinical/at-risk samples against the general population (Cooke et al., Reference Cooke, Deneault, Devereux, Eirich, Fearon and Madigan2022; Rodrigues et al., Reference Rodrigues, Sokolovic, Madigan, Luo, Silva, Misra and Jenkins2021). Despite empirical studies suggesting a potentially stronger influence of parental mentalization in clinical/at-risk samples (e.g., Hughes et al., Reference Hughes, Aldercotte and Foley2017), our meta-analysis did not find such association.
The vast diversity (i.e., from low socioeconomic status to inpatient psychiatric facility) and the broad severity (i.e., COVID vs. childhood sexual abuse) within the clinical/at-risk sample might explain the unexpected absence of moderation effects. Such variability within these populations could lead to differing correlations. For instance, Suardi et al. (Reference Suardi, Moser, Sancho Rossignol, Manini, Vital, Merminod, Kreis, Ansermet, Serpa and Schechter2020) reported a more robust relationship between parental mentalization and externalizing problems in a posttraumatic stress disordered parents’ sample, while Walker et al. (Reference Walker, Wheatcroft and Camic2012) found a stronger relationship in the general population. These findings underscore the complexity of controlling for severity sources within clinical/at-risk samples and emphasize the need for further research to delineate the nuanced influences of different risk factors within diverse samples on the extent of moderation effects.
Finally, we expected that the relationship would be stronger when primary caregivers reported on both mentalization and externalizing behaviors; yet, the results show no influence of the reporting figure on this relation. This finding may be explained due to children behaving differently at home or at school, allowing both teachers and caregivers to provide unique responses. It might also be due to the unique way parents perceive their child’s behavior depending on gender. For example, Mowlem et al. (Reference Mowlem, Agnew-Blais, Taylor and Asherson2019) found that parents overrated externalizing symptoms (e.g., hyperactivity/impulsivity) in boys but not in girls. Nevertheless, other authors reported similar results on externalizing problems both on parents and teachers after the pandemic (Watts & Pattnaik, Reference Watts and Pattnaik2023). In any case, further consideration of both parental and teacher perspective in reporting may provide valuable insights given the distinct experiences and observations of these informants (Deneault et al., Reference Deneault, Hammond and Madigan2023).
Limitations and future directions
The current meta-analytic review is not without limitations. Firstly, although certain possible moderators may be influential, eventually they could not be examined due to the limited number of studies. This was the case of factors such as both the caregivers’ and children’s gender. Upon data availability in the future, these questions may be explored. In fact, parents often respond differently to the child based on their children’s gender, which included variations in displays of affection (Morawska, Reference Morawska2020). Boys tend to display more anger and aggression (Perra et al., Reference Perra, Paine and Hay2021), while girls are perceived as being more cuddly (Sechi et al., Reference Sechi, Vismara, Rollè, Prino and Lucarelli2020). Research also suggests that parents typically display higher PM tendencies with daughters compared to sons, suggesting greater challenges in understanding daughters’ mental experiences (Pazzagli et al., Reference Pazzagli, Delvecchio, Raspa, Mazzeschi and Luyten2018). Moreover, bearing in mind that the significance of a child’s temperament in relation to parenting and developmental aspects is widely acknowledged (e.g., Belsky & van IJzendoorn, Reference Belsky and van IJzendoorn2017; Clark et al., Reference Clark, Kochanska and Ready2000), its examination as a possible moderator should be warranted. Despite its importance, this variable was not included in the meta-analysis due to the limited existing studies focusing on a child’s temperament as a possible moderator. Nonetheless, future research endeavors should investigate the potential moderating role of a child’s temperament in the relationships between parental mentalization and child externalizing problems.
Secondly, we should acknowledge that the conceptualization of ‘externalizing problems’ as a unitary construct is troublesome. As pointed out by Bongers et al. (Reference Bongers, Koot, Van Der Ende and Verhulst2004), there are various manifestations within the spectrum of ‘externalizing behaviors’ – opposition, aggression, property violations, and status violations, according to Frick et al.’s (Reference Frick, Lahey, Loeber, Tannenbaum, Van Horn, Christ, Hart and Hanson1993) taxonomy. These various manifestations of symptoms may follow distinct developmental paths, and the influence of parental mentalization may vary across each trajectory.
A further possible limitation concerns the transactional nature of the relationship between parenting and externalizing problems in offspring. As the meta-analysis by Yan et al. (Reference Yan, Ansari and Peng2021) shows, not only do parental behaviors predict externalizing problems, but such problems have comparable effects on subsequent parental functioning. Given that there are not enough longitudinal studies to fully understand these dynamics over time, more prospective research is needed to gain deeper insights into how these transactions evolve and inform more effective intervention strategies. Finally, since we only analyzed externalizing problems, future research may benefit from the comparison of effects on the relationship between parental mentalization and both externalizing and internalizing behaviors.
Final conclusions
In the context of practical applications, equipping caregivers with appropriate mentalization skills could enhance their ability to regulate their own and their children’s emotions, and responding with attuned behaviors to address appropriately the needs of their sons or daughters. Evidence shows that improvement in parental mentalization is associated with lower levels of children’s externalizing problems (e.g., Camoirano, Reference Camoirano2017). Our results suggest that parental programs aimed at improving parental mentalization skills to prevent and/or mitigate children’s externalizing behaviors are particularly recommended in the case of experienced parents and during the middle childhood developmental stage. Beyond the reduction of externalizing problems, mentalization-based interventions are suggested to be beneficial for both parents and children. Notably, improvements have been observed in caregivers’ psychological distress and reflective function (Byrne et al., Reference Byrne, Murphy and Connon2020) and in children’s general well-being (Midgley et al., Reference Midgley, Sprecher and Sleed2021). All in all, our findings pave the way for practitioners to design parental mentalization interventions that may prevent potential future criminal trajectories.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0954579424001391.
Acknowledgments
We would like to thank researchers who promptly answered our requests for additional data and/or analyses from their studies.
Authors contributions
M.N.R. – Conceptualized and designed the study, funding acquisition, literature search, data coding, data extraction, statistical analyses, interpretation of data, and drafting and first write-up of the manuscript.
B.T.G. – Conceptualized and designed the study, funding acquisition, interpretation of data, overall supervision, and write-up of the manuscript.
I.A.A. – Conceptualized and designed the study, funding acquisition, interpretation of data, overall supervision, and write-up of the manuscript.
All authors discussed the results and contributed to the final version of the manuscript.
Funding statement
This research was supported by a pre-doctoral grant from the Education Department of the Basque Government (PRE_2021_2_0050) awarded to the first author under second and third author’s supervision, a grant from the Spanish Ministry of Science and Innovation (PID2020-115738 GB-I00, funded by MCIN/AEI/10.13039/501100011033/) and a grant from the Basque Government to Research Groups (‘Culture, Cognition, and Emotion’ Consolidated Group; IT1598-22). Open Access funding provided by the University of the Basque Country, UPV/EHU.
Competing interests
None.
Open access
