The systematic search across multiple electronic databases initially yielded a total of 8,458 studies. When the pre-established inclusion and exclusion criteria were applied, this number was reduced to 6,953 studies. Further refinement through the removal of duplicate entries led to a pool of 5,250 studies. A detailed screening of these studies, focusing on their titles and abstracts, resulted in the exclusion of a significant number – 5,141 studies were deemed not relevant to the scope of this review. Consequently, 109 studies were identified as potentially relevant based on their abstracts. However, upon a full-text review, 92 of these studies were excluded due to various reasons such as not meeting the specific criteria or lacking sufficient depth in MBIs.

During the process of reviewing the full texts, an exhaustive examination of the citations and reference lists from the eligible studies was also conducted. This led to the identification and inclusion of an additional 4 articles that met the review criteria. The entire selection process, following the rigorous standards set by the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, was conducted with a high degree of thoroughness and collaborative discussion amongst the authors. Any uncertainties or ambiguities encountered during the study selection were resolved through these discussions, ensuring a robust and consensus-based approach.

Ultimately, after this comprehensive evaluation, 21 articles were deemed suitable for inclusion in the systematic review. The flow of information through the different phases of this systematic review is graphically represented in Fig. 1. This figure provides a visual summary of the screening process, illustrating the step-by-step reduction and refinement of studies, culminating in the final selection of articles for the review.

PRISMA Flow Diagram Illustrating the Study Selection Process. From: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71. doi: https://doi.org/10.1136/bmj.n71.

Author (year)

Country

Age group

Sample size

Study design

Intervention name

Target beneficiary group(s)

Aspect of mentalization

Main significant findings

Outcomes

Effect size

QA

Bak et al. (2015)

Denmark

adolescents

130

Pre-post pilot study

The Resilience Program

Students, the staff, and the broader community.

Mentalization

3 years after the intervention, 80–100% of staff were still actively using components of the Resilience Program in their work. The program was rated as very valuable by staff. The frequency of violent incidents decreased after the intervention as did staff sick days.

Child violent incidents (p < .05)

Carer sick days (p < .05)

Child violent incidents:

Pre: rate ratio 4.36, 95% CI [2.41–8.56]

Post: rate ratio 2.28, 95% CI [1.37–3.92]

Fair

Bianco et al. (2016)

Italy

8–9 years old

72

Quasi-experimental

ToM conversation-based training

Students

ToM

ToM training group improved significantly more than control group on ToM skills after training and at 2 month follow up.

Child ToM skills

Post: (p < .001) Follow up: (p = .028)

The ToM group outperformed the control one at post-test

Post: d = 0.79

Follow up: d = 0.42

Good

Caputi et al. (2021)

Italy

9–10 years old

210

RCT

ToM conversation-based training

Students

ToM

ToM training group obtained higher ToM and lower loneliness scores at post-test compared to control. However, loneliness effect was not sustained at the 2 month follow up.

Child ToM skills (p < .001)

Child loneliness scores (p = .003)

ToM skills: d = 0.55

Good

Doron E. (2016)

Israel

9–13 years old

150

Quasi-experimental

Creativity and divergent thinking training

Students

Mindfulness, ToM, perspective taking

Children who participated in the program scored significantly higher on the divergent thinking tests after the intervention took place and showed higher creative achievements compared to children in the control groups.

Child creative achievements: fluency (p = .012) and uniqueness (p < .001)

Fluency:

pη2 = 0.05

Uniqueness: pη2 = 0.11

Good

Eppler-Wolff et al. (2020)

USA

7-year-old girl (Marie) and a teacher in her late 50’s (Ms. Henry).

-

Qualitative two case studies

School-Based Mental Health Collaboration (SBMHC)

Students and teachers

Mentalization

Discussed role of nested mentalization in stakeholder support. Did not have control group at follow-up. The cases illustrate SBMHC model can improve reflective functioning and mentalization capacity for both students and teachers through the nested mentalization framework.

NA

Finne et al. (2017)

Norway

11–14 years old

332

Pre-post

Social Perception Training (SPT)

Students

Social cognition, perception, interpretation

Reduction in cognitive distortions and problem behaviours, improvement in social skills and perceived learning environment.

Child cognitive distortions (p < .0001)

Child behaviour problems (p < .0001)

Child social skills (p < .0001)

Child perceived learning environment (p = .010)

η2 = 0.11

η2 = 0.09

η2 = 0.08

η2 = 0.02

Fair

Fonagy et al. (2009)

USA

8–11 years old

1345

RCT

Peaceful Schools Program

Students

Mentalization

The CAPSLE intervention significantly reduced peer-reported aggression, victimization, aggressive bystanding, improved empathic mentalizing compared to control. It also reduced disruptive/off-task classroom behaviours.

Child peer-reported aggression (p < .05)

Child peer-reported victimization (p < .01)

Child aggressive bystanding (p < .05) Child empathic mentalizing (p < .01)

Child disruptive/off-task classroom behaviours (p < .05)

Good

Greenberg et al. (1998)

USA

5–12 years old

57

Quasi-experimental

PATHS (Promoting Alternative Thinking Strategies) curriculum

Deaf children

Emotional awareness, understanding others’ perspectives

Intervention group showed significant improvements in students’ social problem-solving skills, emotional recognition skills, and teacher- and parent-rated social competence.

Child social cognition (p < .05)

Child emotional recognition (p < .001)

Child positive social competence (p < .05)

Good

Iuso et al. (2022)

Italy

12–14 years old

191

Pre-post

Psychoeducational program

Students prone to bullying behaviours

Empathy, alexithymia

Psychoeducational program significantly reduced alexithymia levels, increased empathy, and emotion regulation capacities among adolescents.

Child alexithymia (p = .0001)

Child Empathy (p = .0092)

Child Emotional regulation (p < .0001)

Fair

Karray et al. (2020)

Tunisia

8–12 years old

105

Quasi-experimental

“Game of Three Figures” intervention

Students

Empathy, alexithymia

“Game of Three Figures” intervention increased empathy compared to control group.

Child Empathy (p < .05)

Lecce et al. (2014)

Italy

9–10 years old

91

RCT

ToM conversation-based training

Students

ToM

Compared with the control group, children in the experimental group performed significantly better on the ToM task at post-test and follow-up and made greater gains in ToM from pre-test to post-test and from pre-test to follow-up.

Child ToM task performance (p = .001)

pη2 = 0.11

Good

Lombardi et al. (2021)

Italy

8–10 years old

110

Quasi-experimental

ToM conversation-based training

Students

Perspective-taking, ToM, reflective thinking

The training group showed significant improvements in altruistic behaviour and investment decisions compared to the control group.

Child altruistic behaviour (p = .022)

Child investment decisions (p = .041)

η2 = 0.071

η2 = 0.038

Good

Lombardi et al. (2022)

Italy

7–8 years old

56

RCT

Thoughts in Mind - Child (TiM-C)

Students

Emotion regulation, metacognition, ToM

Significant improvements over the training period only in the TiM-C Project group for Metacognition, Emotion Regulation Strategies and a ToM task.

Child metacognition (p < .001)

Child emotion regulation (p = .011)

Child ToM (p < .001)

pη2 = 0.23

pη2 = 0.13

pη2 = 0.29

Good

Oker et al. (2020)

France

9–11 years old

22

Quasi-experimental

Interaction with a virtual tutor agent named “Alice”

Students

Empathy

Students were more accurate and took longer to respond correctly when interacting with the bimodal (facial expressions) agent. They also rated this agent as more empathetic.

Child more accurate response times (p < .04)

Child rated agent as more empathetic (p < .01)

pη2 = 0.20

d = 1.76

Good

Ornaghi et al. (2014)

Italy

6–8 years old

110

Quasi-experimental

Conversational intervention on emotions

Students

Emotion understanding, ToM, empathy

The trained group outperformed the control group on post-test measures of emotion understanding, theory of mind, and empathy (affective domain of empathy showed no difference). The positive effect of the intervention on emotion understanding remained stable 6 months later.

Child emotion understanding (p = .0001)

Child ToM (p = .0002)

Child cognitive empathy ( p < .05 )

pη2 = 0.14

pη2 = 0.12

pη2 = 0.05

Good

Ratcliffe et al. (2014)

Australia

7–13 years old

217

Quasi-experimental

Emotion-Based Social Skills Training (EBSST)

Children with ASD

Emotional understanding, empathy, ToM

EBSST improved teacher-reported emotional competence in students, with a large effect size that was maintained at follow-up. No significant improvements on parent-reported outcomes or more general measures of social skills and mental health.

Carer emotional competence (p < .05)

η2 = 0.18

Good

Sagkal et al. (2012)

Turkey

6th grade elementary school students.

281

Quasi-experimental

Peace Education Program

Students

Empathy

The peace education program was effective in increasing students’ empathy levels compared to the control group. Both boys and girls in the intervention group showed increased empathy.

Child empathy (p < .05)

η2 = 0.17

Good

Twemlow et al. (2005)

USA

elementary/primary school students grade K-5.

3600

Quasi-experimental

Peaceful Schools Program

Students

Mentalization

Compared to control schools, the Peaceful Schools intervention showed significant reductions in peer reported victimization, aggression, aggressive bystanding, disruptive/off-task classroom behaviours over 2 years. Effects on victimization, aggression and aggressive bystanding maintained at 1 year follow-up.

Child peer-reported victimization (P < .01) and aggression (P < .05) and aggressive bystanding (P < .01)

Child disruptive/off-task behaviours (P < .001)

Fair

Twemlow et al. (2001)

USA

5–10 years old

609

Quasi-experimental

Peaceful Schools Program

Students

Mentalization

The experimental school showed significant decreases in disciplinary referrals and suspensions.

The experimental school showed increases in academic achievement test scores.

The control school showed little change.

Child disciplinary referrals (p < .05)

Child suspensions (p < .004)

Child academic achievement test scores (p < .05)

Good

Twemlow et al. (2011)

Jamaica

12–15 years old

2 schools

Quasi-experimental

Peaceful Schools Program

Students

Mentalization, power dynamics, bullying

Improvements across 3 years in academic performance, decreased victimization, increased helpfulness, and the school becoming a place teachers wanted to work, and students wanted to succeed.

Child academic performance (p < .05)

Child victimization (p < .05)

Child helpfulness (p < .05)

Fair

Valle et al. (2016)

Italy

10 years old

46

Quasi-experimental

Thought in Mind (TiM) Project training

Students

Mentalization, ToM

Children of teachers who received TiM training improved more than controls in 3rd order false belief and mentalizing styles.

Child mentalizing (false belief understanding, Strange Stories, Reading the Mind in the Eyes, Mentalizing Task) (p < .05)

pη2 = 0.17–0.46

Fair

Table 1 provides a comprehensive summary of the studies reviewed, detailing specifics such as the authors, countries of origin, years of publication, and demographic data including age range and sample size. It also outlines the methodological attributes like study design, the facets of mentalizing addressed, the nature of the intervention, the target groups, and the principal findings.

The 21 studies included in the review represent a wide geographical spread, conducted across 9 countries. The bulk of these were carried out in Italy (8 studies), followed by the USA (5 studies), and single studies from Australia, Denmark, France, Israel, Jamaica, Norway, Turkey, and Tunisia. Such a broad international representation serves to enhance the generalizability and applicability of the findings across different cultural and educational contexts.

The age groups targeted by these studies varied considerably, ranging from early elementary to early secondary school students. Both mainstream educational settings and special needs populations were included, indicating a broad interest in applying mentalizing interventions across diverse learning environments. A significant majority (18 studies) focused on typically developing students, including some in challenging environments such as violent or under-resourced schools [19,20,21, 37,38,39,40,41]. Additionally, one study specifically targeted children with Autism Spectrum Disorder (ASD) [42], and another was conducted with deaf children [43]. This indicates that while the majority of research on school-based mentalizing interventions has focused on universal prevention for general student bodies, there is a growing interest in targeted interventions for clinical subgroups, suggesting an area ripe for further investigation.

The collective sample size of over 7,500 children and adolescents across these 21 studies underscores the significant reach and potential impact of mentalizing interventions delivered through school settings. Such interventions demonstrate the capacity to engage large numbers of youth, particularly those who may not traditionally have access to additional counselling or psychological services. This extensive reach is critical for fostering widespread improvements in mentalizing abilities and related psychosocial outcomes among diverse student populations.

The aspects of mentalizing targeted within the selected studies were varied, exemplifying the multifaceted nature of this concept. The studies endeavored to bolster several components of mentalizing, including empathy, Theory of Mind (ToM), emotional comprehension, perspective-taking, and mindfulness. A notable number of these studies primarily focused on enhancing ToM [37, 40]. This diversity in focus underscores the comprehensive approach of the interventions in addressing mentalizing.

Each intervention was uniquely designed to cater to the specific mentalizing components relevant to its target population. This tailored approach ensures that the interventions are appropriately aligned with the distinct needs and developmental stages of the students involved. For example, in younger children, where foundational aspects of ToM are still developing, interventions may concentrate more on fostering basic understanding of emotions and perspectives of others. In contrast, with older children or adolescents, the emphasis might shift towards more complex aspects of mentalizing, such as understanding nuanced emotional states and advanced perspective-taking.

The broad spectrum of mentalizing aspects addressed in these studies reflects an understanding of the complex interplay of cognitive and emotional processes in social interactions and personal development. By focusing on diverse components of mentalizing, these interventions aim not only to enhance the immediate social and emotional competencies of students but also to lay a groundwork for their long-term psychological well-being and interpersonal success. This approach is particularly relevant in educational settings, where fostering a range of mentalizing skills can significantly impact students’ academic and social experiences.

Of the 21 studies identified by the review, four were randomized controlled trials (RCTs) [19, 40, 44, 45], whilst 13 had a quasi-experimental design. Although these quasi-experimental studies can provide valuable insights, the absence of an experimental design means that potential for confounding cannot be ruled out, which limits the inferences that may be drawn from the findings. Among the RCTs, one study [19] was notable for its relatively high-quality evidence, demonstrating significant reductions in peer-reported aggression and victimization, improved empathic mentalizing, and decreased disruptive/off-task classroom behaviours. This study, along with another RCT designs [44], which showed significant improvements in Theory of Mind (ToM) task performance, underscores the potential for mentalization-based interventions to provide robust evidence of positive impact.

Of the non-controlled studies, one paper reported results of a small qualitative evaluation [39], while most were simple pre-post intervention cohort studies, without a comparison group. These studies often demonstrated significant improvements in targeted outcomes, such as reductions in violent incidents [33] and alexithymia levels [41], although the lack of control groups means that natural fluctuations and external factors could not be discounted as explanation of observed benefit.

The assessment of study quality involved an examination of the study design, randomisation, sample size, and duration of follow-up. Although some studies incorporated control groups, the use of semi-random or non-random assignment to groups makes it challenging to establish causal links between interventions and outcome. A further recurrent issue was the underpowered designs with small sample sizes [21, 46], which affects the ability to generalise findings and highlights the necessity for larger-scale studies. Follow-up durations were frequently brief [44, 47], which leaves open the possibility of rapid reversal of observed benefits. Certain researchers have pointed out the provisional character of their findings [33] and highlighted the importance of replication, the need for recruitment of more diverse samples [19] and the ongoing refinement and evaluation of bespoke programmes. No study was compliant with Open Science guidelines with fully prespecified statistical modelling approaches and outcomes expectations.

Across the 21 evaluation studies, a wide range of outcomes were assessed. The results for each study are presented in detail in Table 1. A noteworthy finding from this review was the consistent significant improvement in children’s mentalizing capacities across several studies, although most of them did not set out measures of mentalizing as primary outcomes per se. Significant enhancements ToM Skills Improvement was presented in 5 studies post-intervention, demonstrating the impact of MBI on cognitive and affective ToM development [37, 40, 44, 45, 47]. In 3 studies, interventions led to a reduction in peer-reported aggression and victimization [11, 19, 20], and 2 studies reported improvement in emotional regulation capacities among children [41, 45].

The Peaceful Schools project [20, 25] included children’s self-reported experiences of and beliefs about aggression and victimization using various measures. One study explicitly attempted to measure the impact of the intervention on children’s mentalizing capacity [21], assessing this outcome using multiple measures (false belief understanding, Strange Stories, Reading the mind in the Eyes, Mentalizing Task). The results were mixed, showing positive intervention effects on only a few of these measures.

Most studies assessed outcomes for children in one or more domains. Common measures included child behaviour, wellbeing, and social skills. Almost all studies that assessed children’s general wellbeing found at least some positive changes over the intervention period.

It should be noted that the outcomes relating to child well-being were often carer-, clinician-, or teacher-reported. This reliance on indirect reporting can introduce bias and may not fully capture the child’s perspective. Two studies [25, 33] included children’s self-reported experiences of the beliefs about aggression and victimization using the Peer Experiences Questionnaire [48], which showed robust intervention effects.

Some studies aimed to support children’s wellbeing indirectly by targeting parents, carers, teachers, or staff. For these approaches, outcomes for the caregivers themselves, alongside carer-reported outcomes for the children, were often assessed. Several studies assessed caregivers’ mentalizing capacity using various measures. Only a few studies showed clear evidence of improved caregiver mentalizing from pre- to post-intervention [39, 41], with others finding no strong evidence for intervention effects in this domain. Additional outcomes for caregivers included measures of parenting stress and caregiver self-efficacy. Improvements in these areas were reported in a few studies, highlighting the potential secondary benefits of mentalization-based interventions.

In term of effect sizes, findings reveal that mentalizing-based interventions generally lead to medium to large effect sizes across various psychological and social outcomes. ToM conversation-based training consistently showed medium to large effect sizes, highlighting its efficacy in enhancing cognitive and social abilities. Notably, some studies reported very large effects for specific outcomes, such as perceived empathy (d = 1.76; 46). However, it’s important to note that not all studies reported standardized effect sizes, and the metrics used varied across studies, making direct comparisons challenging. Other mentalizing, school and social cognition programs, though less consistent in reporting effect sizes, demonstrated significant reductions in aggressive behaviours and improvements in mentalizing capacities.

Overall, the reported effect sizes suggest that MBIs are effective in fostering mentalization, empathy, social competence, emotional regulation, and academic performance among students. The reported effect sizes underline the substantial impact these programs can have, reinforcing their importance in educational and developmental contexts. Notably, structured trainings reveal relatively moderate to large effect sizes, demonstrating the robust potential of these interventions to significantly enhance various developmental outcomes in educational settings.

A number of studies have focused on implementing ToM training programmes, which are anchored in discussions about mental states, and was shown to significantly improve ToM skills among students. These programmes have been introduced in various educational settings, primarily targeting primary school children. For example, a ToM training programme consisting of four sessions was carried out, facilitated by primary school teachers for 72 children aged 8–9 years [37]. This programme combined individual work and group discussions centred around mental states, drawing upon short stories as a basis. The findings indicated that the group receiving ToM training showed significantly more improvement in ToM task performance compared to a control group, both immediately following the training and at a follow-up two months later. They reported that the ToM group outperformed the control one with a moderate effect size for ToM skills in post-intervention (d = 0.79, 95% CI [0.96, 3.16]) for the differences.

In a similar vein, a ToM training that involved 91 children aged 9–10 years [44], included conversations about mental states, using stories and activities focused on mental state verbs across four sessions, each lasting 50 min. The study included a control group that engaged in discussions about physical events. The children in the ToM group demonstrated notably greater progress in advanced ToM tasks, with a moderate effect size (η2 = 0.11, p = .001), both from the start to the end of the training and in the two-month follow-up.

Another study that was conducted in Italy involved 210 students aged 9–10 years in a similar ToM conversation-based training programme [40]. The outcomes measured were ToM abilities and feelings of loneliness. Post-training, the ToM group not only achieved higher scores in ToM but also reported lower levels of loneliness compared to the control group. However, these improvements were not observed in the follow-up after two months.

The scope of their conversational training programme was extended to include perspective-taking and reflective thinking, alongside ToM, for 110 primary school children in Italy [49]. The results of this train