Abstract
Amid the rising prevalence of autism spectrum disorder (ASD), understanding its comorbidities, especially the frequent and severe co-occurrence of depression, has never been more critical. This study investigates the underlying mechanisms linking depressive symptoms and autistic traits by examining the roles of mentalization, attachment, social support, and psychological flexibility. Utilising a large sample (N=2409) including individuals both diagnosed with autism (N=133) and from the general population, we employed a spectrum approach to analyse the relationship between autistic traits (AQ) and depressive symptom severity (BDI). Our findings reveal that mentalization, perceived social support, and psychological flexibility act as significant mediators, indicating complete mediation of the relationship between AQ and BDI, while the direct effect of AQ on BDI was not statistically significant. Psychological flexibility emerged as the strongest mediator across both ASD and non-ASD groups. Despite significant differences in scores between the ASD and non-ASD groups, the mediating effects of mentalization, social support, and psychological flexibility were consistent, underscoring the complex interplay between autistic traits and mental health. The study highlights the high propensity for depressive symptoms in individuals with high AQ scores, coupled with mentalization deficits, reduced perception of social support, and psychological inflexibility. These findings align with existing literature and emphasise the clinical relevance of addressing mentalization and psychological flexibility in therapeutic strategies for individuals with autism. Future research should aim to develop and validate ASD-specific mentalization-based interventions to improve reflective function and emotional resilience, ultimately enhancing the quality of life for individuals across the autism spectrum.
Lay abstract This study explores whether autism itself increases the risk of depression or if other factors contribute to this susceptibility. We conducted an online survey with people diagnosed with autism and people from the general population. The analysis showed that the ability to understand different mental states (mentalization), the recognition of help from others (perceived social support), and the capacity to cope with difficult experiences (psychological resilience) fully explained the relationship between autistic traits and depressive symptoms. There was no direct link between autistic traits and depression. Psychological resilience was the most influential factor in both groups. The study found that individuals with more autistic traits are more prone to depressive symptoms due to difficulties in mentalization, reduced perception of social support, and lower psychological flexibility. These findings highlight the need for targeted interventions to improve mentalization, social skills, and emotional resilience, which could enhance the quality of life for people with autism.
Introduction
1. What is the relevance of studying depression in autism?
Autism spectrum disorder (ASD) is a neurodevelopmental condition characterised by difficulties in social communication and interaction, restricted or repetitive behaviours or interests, and sensory processing (American Psychiatric Association, 2013, 2022). The rising prevalence of ASD (Zeidan et al., 2022) makes it increasingly crucial to understand its comorbidities, particularly the frequent and severe co-occurrence of depression (Bougeard et al., 2021). A number of studies have examined the relationship between ASD and depression, suggesting that individuals with autism have a higher rate of depression than the population average (e.g. (Hollocks et al., 2019; Lai et al., 2019; Wigham et al., 2017). While the prevalence of depressive disorders is estimated to be up to 7-11% in the general population (Lim et al., 2018; Wittchen et al., 2011; Wittchen & Jacobi, 2005), it is approximately 10-15%, ranging from 2.5% to 47.1% in ASD (Hossain et al., 2020), and individuals with ASD are four times more likely to experience depression during their lifetime compared to neurotypical individuals (Hudson et al., 2019) according to recent meta-analyses. Despite the high prevalence of depressive symptoms in ASD patients, the nature of the relationship between the two conditions remains uncertain.
As depression is a multifactorial disorder and ASD is a heterogeneous spectrum condition, it is difficult to single out one causal factor. It has been recognised that one in two individuals with autism has alexithymia (Albantakis et al., 2020); loneliness (Schiltz et al., 2021; Umagami et al., 2022), emotion regulation deficits, and intolerance of uncertainty have also been found to be associated with symptoms of anxiety and depression in autistic populations (Cai et al., 2018). Furthermore, depressive symptoms severely impact the quality of life (Thiel et al., 2024).
In this article, we explore the interconnected phenomena of mentalization, attachment, and peer support and psychological flexibility as potential mediators to understand their combined impact on the relationship between ASD and depressive symptoms. Our inquiry aligns with the mentalization theory proposed by Fonagy, Luyten, and Bateman, subsequently extended to illuminate diverse psychopathological phenomena (Fonagy & Allison, 2014; Fonagy & Bateman, 2016; Fonagy & Luyten, 2009; Luyten & Fonagy, 2015, 2018). Through this exploration, we endeavour to shed light on the intricate web connecting ASD and depressive symptomatology.
2. What is the role of mediating factors in the relationship between autism and depression?
Mentalization
Social communication deficits are core symptoms of ASD. Mentalizing deficits occur early but remain observable throughout life (Baron-Cohen, 2000; Chung et al., 2014). Research in recent years on ASD has consistently shown deficits in theory of mind (ToM; one specific aspect of mentalization) (Yirmiya et al., 1998), emotion perception and processing (Brewer et al., 2017; Velikonja et al., 2019), empathy, affective mentalizing, and cognitive mentalizing and related altered neural functioning (Arioli et al., 2021). ASD is, however, a heterogenous spectrum; it is important to note that within the condition, variability in mentalizing abilities is not negligible (Lombardo et al., 2016).
Mentalization deficits are associated with depressive symptoms as well. Identifying and interpreting mental states of the self and others have been found to be impaired among patients with depression, and the degree of mentalizing impairment was associated with the severity of depression and worse clinical outcome (Fischer-Kern et al., 2013; Lee et al., 2005; Lombardo et al., 2016). Deficits in theory of mind were shown to be a predictor of relapse and impairment in social functioning in depression (Inoue et al., 2006) and were associated with cognitive deficits as well (Y. Wang et al., 2008). Fischer-Kern et al. also showed that, besides mentalizing deficits, attachment insecurity and unresolved loss are also important predictors of depression (Fischer-Kern et al., 2022). A current integrative model of depression, the stress-reward-mentalizing model suggests a developmental cascade. This sequence involves elevated stress levels and persistent heightened arousal, potentially coupled with genetic predisposition, culminating in impaired reward sensitivity within attachment relationships and the realm of agency and self-governance. Consequently, this progression impacts the capacity for mentalizing and social cognitive functions. (Luyten & Fonagy, 2018).
Attachment
Emotion recognition, reciprocal social communication, theory of mind, and other aspects of mentalization are essential to form secure attachment and maintain a balanced parent-child relationship (Fonagy et al., 1991; Keenan et al., 2017). Children with ASD show impairments in social, and emotional capacities from an early age (American Psychiatric Association, 2022; Baron-Cohen et al., 1985; Cassel et al., 2007; Dawson et al., 2004; Nuske et al., 2013). They exhibit heightened attachment insecurity (Naber et al., 2007), as well as decreased sensitivity in their relationships with caregivers (Rutgers et al., 2004; Van IJzendoorn et al., 2007), including more contact resistance and less contact-seeking behaviours (Rogers et al., 1993), fewer pro-social responses to caregivers and difficulty being soothed (Grzadzinski et al., 2014), and less frequent looking, showing, smiling, and mutual play behaviours (Akdemir et al., 2009; Dissanayake & Crossley, 1997). According to a recent review, only 47% of children with ASD demonstrated secure attachments (Teague et al., 2017). The notion that individuals with ASD have difficulties to form attachments, however, does not appear to be supported by the observation that they express a preference for the caregiver over a stranger, and experience distress when unexpectedly separated from them (Capps et al., 1994; Dissanayake & Crossley, 1997; Koren-Karie et al., 2009; Rogers et al., 1991). By employing questionnaire-based measures, instead of observation-based paradigms like the Strange Situation (Ainsworth et al., 1978), during middle childhood, no differences in attachment security were found between autistic and neurotypical participants (Bauminger et al., 2010; Chandler & Dissanayake, 2014; Wu et al., 2015). Upon reaching adulthood, attachment style remains relatively stable (Chris Fraley, 2002), but the object of attachment shifts from parents to peers, including friends and romantic partners (Doherty & Feeney, 2004; Hazan & Zeifman, 1994; Nickerson & Nagle, 2005). Insecure attachment, especially avoidant attachment style, is more prevalent in ASD groups than in the general population (Taylor et al., 2008), and associated with autistic traits in non-clinical samples (Gallitto & Leth-Steensen, 2015), and it is associated with lower relationship satisfaction (Beffel et al., 2021). A recent study found that adult attachment styles are important predictors of mental health: higher anxious attachment predicted higher depression and anxiety, but interestingly, higher avoidant attachment predicted lower anxiety and stress in autistic adults (Lee et al., 2022).
Attachment-related distress and avoidance have been associated with psychopathology (Mikulincer & Shaver, 2012, 2017), including anxiety (Bosmans et al., 2010) and depression (Beatson & Taryan, 2003; Besser & Priel, 2003; Cantazaro & Wei, 2010; Carnelley et al., 1994; Dagan et al., 2018; Malik et al., 2015; Wei et al., 2006). Individuals reporting low attachment anxiety and avoidance have the ability to maintain interpersonal relationships through secure attachment and report higher levels of adaptive emotion regulation, positive self- and other perceptions, and greater psychological well-being (Mikulincer & Shaver, 2019; Raque-Bogdan et al., 2011). Attachment style has an impact on emotional, behavioural, and relational patterns (Mikulincer & Shaver, 2017), resilience to stressful life events (Mikulincer & Shaver, 2012), and the quality of interpersonal relationships (Collins, 1996; Shaver et al., 2000; Torquati & Vazsonyi, 1999).
Social support
Developmental research on mentalization suggests that environmental influences play a significant role in the development of mentalization. In recent years, there has been a notable shift in this respect: while earlier theories (Fonagy et al., 1991) emphasised the specific role of dyadic (parent-child) attachment in facilitating or inhibiting the development of mentalization, recent views have taken a more comprehensive approach. According to this, in addition to the family environment, peers, and wider sociocultural factors are in close interaction with attachment, mentalization and social learning, through the capacity to rely on others to provide social information (epistemic trust) and the ability to benefit from the positive effects of the environment (salutogenesis) (Fonagy & Allison, 2014; Luyten et al., 2020). As ASD is linked to difficulties in social communication and interaction, people with ASD are less able to form social relationships and access and incorporate the benefits of such peer support. The effect of perceived social support has been measured mainly among the parents of children with ASD (Alon, 2019; He et al., 2022; Lu et al., 2021). Studies examining social support in adult ASD found that social support was associated with psychological, social, and environmental well-being (Bishop-Fitzpatrick et al., 2018; Leader et al., 2021; Tsermentseli, 2022).
Social support is a protective factor against depression, with parental support being the most influential during childhood and adolescence, whereas spouses, family, and friends become more important in adulthood and senior age (Gariépy et al., 2016). Perceived social support mediated loneliness and depression in the elderly (Liu et al., 2016), lower quality support predicted a sixfold risk of depressive symptoms among college students (Hefner & Eisenberg, 2009), and elevated the risk of postpartum depression (Akbari et al., 2020), and more importantly showed to be a moderator between negative life events and depression in adolescence (Miloseva et al., 2017). On the contrary, higher levels of perceived social support were linked to positive affect, overall life satisfaction, and favourable outcomes of mental disorders (Siedlecki et al., 2014; Wang et al., 2018).
Psychological flexibility and ASD
Psychological resilience and flexibility are also important factors in mental health (Kashdan & Rottenberg, 2010; Rutten et al., 2013; Yasinski et al., 2020). Psychological inflexibility is crucial in understanding the relationship between depression and autism due to its effects on coping strategies, emotional regulation, and adaptive functioning. Individuals with autism often struggle to adapt to changing circumstances. Inflexibility can appear as rigid thinking patterns, resistance to change, and difficulty shifting focus from negative thoughts or experiences, which can intensify their depressive symptoms Although ASD is characterised by behavioural inflexibility (American Psychiatric Association, 2022), the role of psychological inflexibility was measured only once among autistic adults. A recent study reported that psychological flexibility (measured by AAQ-II) and attachment anxiety mediated the association between childhood experiences of parenting and mental health in later life in individuals with autism (Lee et al., 2022). We argue that psychological flexibility can be instrumental in several ways when considering its reciprocal relationship with attachment, mentalizing, and perceived social support in individuals with autism. It might foster considering alternative perspectives, managing conflicting emotions, adapting their responses in interpersonal challenges, and being open to seeking and receiving support.
3. Hypotheses and aims of the current paper
In this study, we aim to validate the observed higher prevalence of depression within the autism context, initially through a comparative analysis of depressive symptom severity. Our first objective involves examining potential correlations between autistic traits, and depressive symptoms, spanning both autistic and neurotypical populations. Given the recognized significance of attachment, mentalization, and perceived social support in mental well-being, both in neurotypicals and in autism, our investigation delves into their mediating role in the relationship between autistic traits and depressive symptoms. Furthermore, we aim to assess the impact of psychological flexibility. Last, we plan to discern whether the hypothesised effects exhibit diagnosis-specific disparities. Specifically, we aim to ascertain whether an ASD diagnosis inherently increases depression risk, or each domain exerts equivalent importance in both neurotypicals and individuals with autism.
Methods
Recruitment procedure, ethical concerns
Participants were requested to complete an online survey between September 2021 and April 2022. Initially, we reached out to outpatient units, institutions, and counsellors specialising in aiding individuals diagnosed with autism spectrum disorder, with the purpose of facilitating questionnaire accessibility for those with an established ASD diagnosis.
The language of the questionnaires was Hungarian and the wording was designed to be as clear as possible, but experts from these organisations also checked the questionnaire and made some suggestions and this was used to prepare the final version of the survey. Subsequently, we also recruited non-ASD participants through media and social media advertisements.
Participants provided informed consent online. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration. The study was approved by the Semmelweis University Regional and Institutional Committee of Science and Research Ethics, SE RKEB: 159/2021.
Participants
Overall 2409 participants were recruited and started the survey. Completing the whole survey took about 40 minutes, and 1783 participants answered all the items. Participants were allowed to stop at any point and resume filling out the questionnaires later. We have constructed four selection criteria to ensure high quality of our data. First, we have excluded those participants who did not fill out the BDI, or the AQ questionnaires which were intended to measure the depressive and autistic symptom severity, respectively. Participants between the ages of 18 and 65 were included. Further, we have confirmed that all our participants have completed at least 8 years of elementary education or have provided realistic data regarding the years they had spent in education: we have excluded all participants who claimed to have spent more years in education than their age minus 5 years. Finally, we have also rigorously screened our sample based on the time spent filling out the questionnaires that were of interest to us. As participants were allowed to take a break and resume filling out the questionnaires later at their convenience, there were some extremely high values in the time spent filling out many of the questionnaires. Thus, we decided to assess the time it took participants to fill out the respective questionnaires by calculating measures that are more robust to outliers. We have calculated the median filling time and median absolute deviation of filling time as a measure of spread for each questionnaire. We have excluded those participants who have spent less than the median minus one and a half median absolute deviation time on the following questionnaires: AQ, BDI, MZQ, AAS, MSPSS, and AAQ-II.
Questionnaires
Demographic data collection
Demographic information was collected using a structured questionnaire administered at the beginning of the study. Participants were asked about their age, gender, education level, socioeconomic and marital status.
Autism Spectrum Quotient (AQ)
Autism Spectrum Quotient is a 50-item self-report questionnaire to measure the degree of autistic traits in adults with normal intelligence. The AQ assesses five different areas: social skill, attention switching, attention to detail, communication, and imagination; 10 items cover each area, on a 4-point Likert scale with values of ‘definitely agree’, ‘slightly agree’, ‘slightly disagree’ or ‘definitely disagree’. At the end of the questionnaire, each participant was given feedback on their score on the AQ scale and whether they should seek professional advice. The internal consistency of items in each of the five domains and Cronbach’s alpha coefficients were all moderate to high (Baron-Cohen et al., 2001). In the present study, the overall reliability was good (total α = 0.89, ASD α = 0.86, NTP α = 0.86).
Beck Depression Inventory (BDI) is a widely used 21-item, self-report inventory measuring symptoms of depression with good internal consistency (α = 0.86) (Beck et al., 1988). In our study, this scale had good internal consistency (total α = 0.90, ASD α = 0.90, NTP α = 0.90).
Mentalization Questionnaire (MZQ)
The Mentalization Questionnaire is a 15-item self-report measure of mentalization (Hausberg et al., 2012). The items of the MZQ cover four domains of mentalization: ‘Refusing self-reflection’, ‘Emotional awareness’, ‘Psychic equivalence mode’, and ‘Regulation of affect’. The structural validity of the Hungarian version of the MZQ was assessed among psychotic patients and the study confirmed the four-factor structure which differs from healthy samples (Fekete et al., 2019). In our study, however, we used the overall MZQ score, which had good reliability in the sample (total α = 0.86, ASD α = 0.80, NTP α = 0.85).
Adult attachment scale (AAS)
Adult attachment style was measured by the Adult Attachment Scale (AAS), which consists of 18 Likert items that can be grouped into two subscales (Collins, 1996). Psychometric properties of the AAS were evaluated in a Hungarian sample recently (Őri et al., 2021). In our study, we used the scores reflecting anxious and avoidant attachment styles, which had good internal consistency (total α (anxious) = 0.88 and α (avoidant) = 0.86, ASD α (anxious) = 0.87 and α (avoidant) = 0.83, NTP α (anxious) = 0.88 and α (avoidant) = 0.86).
Multidimensional Scale of Perceived Social Support (MSPSS) during adolescence
Social support during adolescence was measured with the Multidimensional Scale of Perceived Social Support (MSPSS), which has 12 Likert items (Canty-Mitchell & Zimet, 2000; Papp-Zipernovszky et al., 2017). Participants were asked to focus on their adolescence. In our study, this scale had excellent internal consistency (total α = 0.95, ASD α = 0.91, NTP α = 0.95).
Acceptance and Action Questionnaire-II (AAQ-II)
AAQ-II is a measure of psychological flexibility, which has 7 Likert items (7-point scale, from “never true” to “always true”). Higher total scores mean less flexibility. The tool has satisfactory psychometric properties (α = 0.84) (Bond et al., 2011), and in a Hungarian sample the internal consistency was good (α = 0.90) (Eisenbeck & Szabó-Bartha, 2018). In our study, this scale had good internal consistency (total α = 0.93, ASD α = 0.91, NTP α = 0.93).
Data Analysis
R, RStudio 2023.03.1, JASP 0.16.1 and IBM SPSS Statistics 22.0 statistical software were used to perform the analyses. First, the Wilcoxon rank sum test and Fisher’s exact test statistics were used to compare the ASD and non-ASD samples (Tables 1 and 3). Second, receiver operating characteristic (ROC) curve analysis was performed to examine the sensitivity and specificity of the AQ to determine a scoring cutoff value that discriminates between autistic and non-autistic cases. In the third step, Spearman correlation was used to investigate the relationships between the tested variables (autism quotient, depression, attachment style, mentalization, perceived social support, and psychological flexibility) in the ASD and NTP groups separately. We compared the correlation coefficients between ASD and NTP groups by z-test (Eid et al., 2013). The normal distribution was assumed based on the central limit theorem (Fischer, 2011). Finally, linear regression analysis was used with a robust bootstrapping method to test the proposed moderated mediation model. In the moderated mediation model autism quotient (AQ score) was the independent variable, attachment style (AAS scores), mentalization (MZQ total score), perceived social support (MSPSS total score), and psychological flexibility (AAQ-II total score) were the mediators, and the depression severity index (BDI score) was the outcome variable. In addition, the model also tested the moderator role of diagnostic groups (autism spectrum group or non-ASD group). As there were significant differences between groups regarding gender, age, and education, we included these variables as covariates in our statistical model.
Results
Descriptive statistics
The basic demographic data of the participants are shown in Table 1. The gender distribution in the non-ASD group reflects the generally higher female response rate (female: 66.1% vs. male: 32.3%), 1.6% of respondents identified themselves as other gender. In the ASD group, by contrast, the proportion of women and men was roughly evenly balanced (female: 48.1% vs. male: 42.9%; as men are more likely to be diagnosed with ASD, but women tend to be more willing to answer questionnaires), and there was a significantly higher proportion of those who defined themselves as other gender (9.0%). For completeness, we note that we also received responses from people who thought they themselves are on the autism spectrum or are thought to be on the spectrum by friends or relatives, but who did not have a formal diagnosis at the time of completing the survey. Their AQ scores (See Supplementary material Figure S1 and S2 & ROC analysis) suggest that there is a real possibility that their assumption is correct, however, as it was not possible to investigate and diagnose them, this group was not included in the further analysis. Their basic demographic data are shown in column “Self-dg ASD” of Table 1. Detailed demographic data for the two study groups are presented in Table 2.
The ASD and non-ASD groups scored differently on all questionnaires, with the ASD group scoring significantly higher on AQ, BDI, MZQ, AAS (both insecure attachment subscale), and AAQ-II, and lower on the MSPSS scale, where this indicates lower perceived peer support during adolescence. Detailed data and between-group differences are shown in Table 3. and Figures 1 and 2.
Sensitivity and specificity of the AQ
We assessed the ability of the AQ to discriminate between ASD and NTP cases by interpreting a ROC curve. In the first step, we tested the sensitivity and specificity in all samples, including the NTP group and participants with autism, both clinically diagnosed and self-diagnosed. Then, we excluded the cases of self-diagnosed autism, and we used the analysis again in our sample. First, the area under the ROC curve was 0.85 (95% confidence interval 0.83–0.88), indicating an excellent level of discrimination (Supplementary material, Figure S1.) (Hosmer, 2013). The results show that when a threshold of 26 was used, the AQ correctly identified more than 75% of cases (Supplementary material, Figure S1 and Table S1.) and showed high levels of both sensitivity and specificity. In the second analysis (without self-diagnosed ASD cases), the area under the ROC curve was 0.82 (95% confidence interval 0.78-0.86), indicating a similarly excellent level of discrimination. In this analysis, when the cutoff is 26, the AQ can identify almost 75% of cases (Supplementary material, Figure S2 and Table S2.). Despite these findings, we decided to exclude self-diagnosed autistic participants (N=100) from the further analysis, as there was no possibility to confirm or refute their diagnosis.
Correlation analysis
We used Spearman correlation to examine the relationships between the AQ, depression (BDI), mentalization (MZQ), attachment style (anxious or avoidant), perceived social support (MSPSS), and acceptance and action (AAQ-II) and age in two groups (autism and non-ASD), separately. We found various degrees of correlation between the variables after correcting for multiple comparisons using the FDR (false discovery rate) method (see Figure 3). To assess the significance of the difference between correlation coefficients in ASD and non-ASD participants, we used the z-test. This shows the significant difference between ASD and non-ASD in terms of correlation coefficients (hence the mirror-like pattern). E.g. Avoidant attachment (AAS avo) and perceived social support (MSPSS) score is significantly differently correlated in non-ASD compared to ASD group. (Figure 3.)
Moderated mediation
A multivariate moderated mediation model was conducted to test the moderating effect of the grouping variable (ASD or NTP) on the relationship between autism quotient and depression via mediation factors, such as attachment styles (anxious or avoidant), mentalization, perceived social support, and psychological flexibility (Figure 4 and Table 5). To control for potential confounding effects, gender, age and years of education were included as covariates.
As shown in Table 5, the main effects of autism quotient (AQ) on different mediators were significant and positive, except for perceived social support (MSPSS), which was significant and negative. The main effect of groups (NTP or ASD) was significant in the case of AAS anxiety score, AAQ-II and BDI, corresponding to higher levels of attachment anxiety, psychological inflexibility, and depressive symptomatology in the ASD group. Additionally, mentalizing difficulties (MZQ) and psychological inflexibility (AAQ-II) showed significant positive, while perceived social support (MSPSS) showed significant negative associations with the outcome variable. More importantly, we found a significant indirect effect of autistic traits on depressive symptomatology via mentalizing, perceived social support, and psychological flexibility. It is worth highlighting that there was full mediation via these factors between autistic traits and depressive symptoms, as the direct effect between AQ and BDI remained insignificant. Moderating effects of the grouping variable (NTP or ASD) did not prove significant on any of the indirect pathways.
Discussion
The present study is dedicated to investigate the underlying mechanisms connecting depressive symptoms and autistic traits by extensively examining the impacts of mentalization, attachment, perceived social support, and psychological flexibility. Our large sample comprises participants diagnosed with autism as well as individuals from the general population, thus integrating both the autism spectrum perspective and categorical diagnostic approach. Our investigation employed between-group comparisons and a spectrum approach, utilising a moderated mediation method to analyse the relationship between autistic traits and depressive symptom severity. We found that mentalizing, perceived social support and psychological flexibility play a mediating role in the relationship between autistic traits and depressive symptoms. Despite significant differences in all scores between the ASD and non-ASD groups, psychological flexibility emerged as the strongest mediator across both groups regardless of diagnosis. The results revealed that factors commonly associated with autism had a mediatory effect between the extent of autistic traits and the score indicating the severity of depression. In fact, this is not a partial mediation: upon including the indirect effects, the residual direct effect loses significance. Mentalizing, perceived social support, and psychological flexibility, fully mediate the influence of autistic traits on depressive symptoms.
Depression in autism presents a notable burden, not solely due to its high frequency, but also due to challenges in its recognition in clinical practice: the onset of symptoms may be atypical or presumed by professionals to be symptoms of ASD itself (Chandrasekhar & Sikich, 2015). The observed propensity for depressive symptoms in individuals with high AQ scores, coupled with mentalization deficit, reduced perception of social support and psychological inflexibility, underscores the complex interplay between autistic traits and mental health. This aligns with existing literature indicating that ASD individuals with high AQ scores tend to experience elevated rates of depression (Hollocks et al., 2019; Lai et al., 2019; Wigham et al., 2017). Furthermore, the study reveals that neurotypical individuals with high AQ scores are also prone to depressive symptoms, with difficulties in mentalization, psychological inflexibility, and lower levels of perceived social support elevating the risk of experiencing more severe depressive symptoms (Beatson & Taryan, 2003; Besser & Priel, 2003; Cantazaro & Wei, 2010; Carnelley et al., 1994; Dagan et al., 2018; Fischer-Kern et al., 2013, 2022; Fischer-Kern & Tmej, 2019; Inoue et al., 2006; L. Lee et al., 2005; Malik et al., 2015; Wei et al., 2006). We can additionally assume that these aforementioned factors may hinder individuals from expressing their difficulties and reaching out for support.
Our results confirm that high AQ scores, coupled with a deficit in mentalization, intensify the severity of depressive symptoms. Given the established associations between ASD and impaired mentalization (Arioli et al., 2021; Baron-Cohen, 2000; Brewer et al., 2017; Chung et al., 2014; Velikonja et al., 2019; Yirmiya et al., 1998), as well as the impact of mentalization deficits on depressive symptoms (Fischer-Kern et al., 2013; Fischer-Kern & Tmej, 2019; L. Lee et al., 2005) it is imperative to explore how these factors interact and potentially influence therapeutic outcomes. Although they have different profiles, the severity of mentalization deficits in ASD is comparable to that of schizophrenia (Bliksted et al., 2016; Fernandes et al., 2018; Gilleen et al., 2017; Oliver et al., 2021), and research suggests that individuals with schizophrenia often struggle with social cognition, including difficulties inferring others’ intentions (Braak et al., 2022; Kronbichler et al., 2017). Encouragingly, Debbané and colleagues have already suggested how a therapeutic approach targeting mentalizing, mentalization-based therapy (MBT) can be used effectively to treat young people at risk of psychosis, opening up perspectives for the ASD population as well (Debbané et al., 2016). In this study we used the overall mentalizing score measured via a short and self-report questionnaire. Future research should measure mentalizing skills in a more nuanced way (e.g., along affective-cognitive, implicit-explicit, self-other, external-internal mentalizing), particularly in autism, where atypicalities may exist within these dimensions.
Emotion recognition, reciprocal social communication, theory of mind, and other aspects of mentalization are essential to form secure attachment and maintain a balanced parent-child relationship. Conversely, secure attachment can foster the development of mentalizing skills (Fonagy et al., 1991; Keenan et al., 2017). Although our results did not confirm a significant mediating effect of insecure (avoidant and/or anxious) attachment on depressive symptoms, higher AQ scores were associated with higher scores reflecting both avoidant and anxious attachment styles. Against the background of the former, we hypothesise that mentalization, as a mediator known to be closely related to attachment, suppressed the effect of insecure attachment on depressive symptoms. Other studies have shown that, in addition to mentalization deficits, attachment insecurity and unprocessed loss are important predictors of depression (Fischer-Kern et al., 2022), and that adult attachment representations influence the effectiveness of psychotherapeutic treatment in clinically depressed patients (Reiner et al., 2016). We can speculate that due to the factors mentioned above, regarding attachment, although it may not be a causal factor or therapeutic target in itself, its close relationship with other symptoms and constructs suggests that its role may still be worth considering. The patient’s attachment style could potentially influence the effectiveness, pace, or even the therapist’s exposure during the therapeutic process. Considering the significant interplay between attachment, mentalization deficits, and depressive symptoms, it becomes evident that assessing and improving patients’ mentalizing capacity is crucial for tailoring effective therapeutic interventions.
In addition to impaired mentalization, the distorted perception of social support might be crucial in understanding the relationship between autistic traits and depressive symptoms. Adjusting behaviour based on social norms is a common practice, however it can be particularly challenging for individuals with ASD due to their mentalization deficit, which necessitates considerable cognitive effort and can be mentally exhausting. As a result, it might increase the tendency to develop anxiety and depression and can negatively affect the development of self-identity (Hull et al., 2017). The phenomenon of camouflage, an explicit effort to “mask” or “compensate” for autistic features, has been observed in ASD (Hull et al., 2017; Lai et al., 2019), which also may contribute to the challenges of diagnosing ASD, especially in women (Howe et al., 2015; Lai et al., 2017). A study examining this issue found that increased levels of camouflaging correlate with higher levels of depression and anxiety (Bernardin et al., 2021). Contrary to the misconception that individuals with ASD avoid social interactions due to disinterest, this avoidance is often a maladaptive coping strategy driven by heightened anxiety. In summary, the complex relationship between impaired perception of social support, camouflaging behaviours, and maladaptive coping mechanisms underscores the fundamental need for social connection. Furthermore, an insecure attachment style can hinder the development of healthy relationships and further distort the perception of social support, leading to a cycle of increased mentalization difficulties and depressive symptoms.
One of the most notable findings of our study was that psychological flexibility was found to be the strongest mediator of the association between autistic traits and depressive symptoms in both ASD and non-ASD populations. However, ASD is characterised by behavioural inflexibility (American Psychiatric Association, 2022), yet the role of psychological inflexibility was measured only once among autistic adults. A recent study reported that psychological flexibility (measured by AAQ-II) and attachment anxiety mediated the association between childhood experiences of parenting and mental health in later life (Lee et al., 2022). This study underscores the key role of psychological flexibility in preventing and treating depression, directing focus towards enhancing it. Previous investigations into the relationship between attachment and psychological flexibility, as well as trauma, open a window into the common risk factors that overlap between ASD and non-ASD populations (Zilberstein, 2014). Painful memories, emotion regulation difficulties, and experiences of trauma emerge as shared risk factors contributing to vulnerability for depressive symptoms in both groups (Carr et al., 2013; Kerns et al., 2015; Kessler & Magee, 1994; Taylor & Gotham, 2016; Tennant, 2002). We argue that interventions targeting psychological flexibility may be particularly effective in addressing depressive symptoms in both ASD and non-ASD populations.
Clinical relevance and therapeutic implications
A significant number of individuals with autism spectrum disorder seek therapy for depression, presenting unique challenges for effective treatment. Managing (both recognising and treating) depression in this population is particularly difficult due to the complex interplay of social, cognitive, and emotional factors inherent to ASD. However, recent findings suggest promising opportunities for intervention. In the following section, we explore these possibilities, drawing on the research results presented earlier to identify effective therapeutic approaches tailored to the needs of individuals with autism and depression.
In our study, the depressive symptoms assessed using the Beck Depression Inventory (BDI) correlated well with observations in clinical practice, thus the clinical relevance of our results can be supported by the experience that questionnaire-based assessment tools are well suited for screening depression in the autistic population, where verbal exploration attempts are often hampered by the specificities of people with ASD: e.g. difficulties in recognising and expressing emotions or establishing interpersonal relationships.
The study of Frolli et al. hypothesised that early intervention using behavioural interventions such as Applied Behaviour Analysis for children with autism and reflexive functions training on parents can improve the development of joint attention, a cognitive precursor to the theory of mind and mentalization processes: they showed that parents who were able to make sense of both their own mental state and that of their child can serve as a protective factor for the child’s development even in atypical developmental situations such as in ASD (Frolli et al., 2021). In our view, it is not only parents of children with ASD who may benefit from this type of training, but also the adult development (reflective functioning, mentalizing) of the individuals themselves, if this was not done earlier in their lives. Utilising psychotherapeutic techniques aimed at enhancing psychological flexibility could also be considered as an additional intervention strategy such as acceptance and commitment therapy (ACT), which belongs to the third generation of the cognitive-behavioural approach. The point is to help patients to improve their psychological flexibility in order to accept unavoidable private events (Lee et al., 2022b; Suarez et al., 2022). A systematic review on the components of ACT highlighted the long-term effectiveness of mindfulness-based interventions (Suarez et al., 2022). The exploration of mindfulness and its role in self-mentalization introduces a potential avenue for intervention, highlighting the importance of cultivating mindfulness skills to enhance emotional regulation and reduce the impact of traumatic experiences.
Mentalization based therapy (MBT) can help individuals develop a more nuanced awareness of their thoughts, feelings and motivations, leading to increased self-reflection and self-understanding, thus it is not unreasonable to assume that mentalisation-based interventions can be beneficial for individuals on the autism spectrum. Furthermore, by enhancing their own and others’ mentalisation skills, MBT can promote better interpersonal understanding and communication (Roncone et al., 2002; Soto-Icaza et al., 2019). There is currently little research to support the effectiveness of MBT in the ASD population. The difficulty in designing an ASD-specific MBT protocol lies in the fact that this is a heterogeneous group with complex, multiple profiles in both social and cognitive symptoms, as well as a heterogeneous biological background (Frye et al., 2019; Lord et al., 2020; Pinto et al., 2014; Sanders et al., 2015; J. Wang et al., 2013). Neural mechanisms may be involved, with previous research suggesting a reduced neural response in key regions of the mentalisation brain network (temporoparietal junction, medial prefrontal cortex, precuneus/posterior cingulate cortex, temporal poles, superior temporal sulcus) (Castelli et al., 2002; Happé & Frith, 2014; Kana et al., 2014; Kennedy & Adolphs, 2012; Lombardo et al., 2011; O’Nions et al., 2014; Saxe et al., 2009; Soto-Icaza et al., 2019), including areas associated with social information processing early in development (temporoparietal junction and superior temporal sulcus) (Kosakowski et al., 2022; Patel et al., 2019). The core symptoms of ASD include social dysfunction, which is thought to affect caregiver relationships, and may be further complicated by the fact that relatives may show subclinical symptoms. However, despite these barriers, attempts are being made with encouraging results (Costa-Cordella et al., 2023; Krämer et al., 2021; Thevenet et al., 2018). In the long term, a mentalization-based approach is promising, but autism-specific adaptation in method development will presumably be essential.
In summary these findings emphasise the need for targeted interventions focusing on mentalization, social skills and support for individuals with ASD, especially those with high AQ scores. In our study, psychological flexibility was identified as the most influential mediator, emphasising the significance of interventions aimed at fostering emotional resilience and improving flexibility in both ASD and non-ASD groups.
Limitations
The strength of our research is that the self-report questionnaires, which could be completed online, allowed for a relatively large number of completions. However, the validity of the responses to the self-report questionnaires may be questionable, and there is a possibility, especially in the case of mentalizing skills, that self-reflection skills of the respondents may bias answers. However, there are interesting findings to suggest that individuals’ subjective perceptions may be more determinative of the development of psychopathology than objective maltreatment per se (Danese & Widom, 2020). Although the ASD group was recruited from institutions working with individuals diagnosed with ASD, the non-ASD sample included individuals who were suspected - by themselves or by others - being affected by autism spectrum disorder. As their diagnostic status is unknown, we ultimately decided to exclude them from the analysis. In addition, it cannot be ruled out that there may be others in the non-ASD sample who are on the spectrum but are not aware of their condition. A further limitation is the cross-sectional nature of the study, necessitating future longitudinal studies to confirm these findings and explore causative relationships.
Conclusions
In summary, this comprehensive study integrates findings related to mentalization, attachment, perceived social support, and psychological flexibility: risk factors associated with depression in ASD and non-ASD populations. The nuanced understanding of these factors provides a foundation for targeted interventions and underscores the importance of personalised approaches in addressing depressive symptoms in individuals across the autism spectrum. Further research is needed to enhance and broaden the insights discussed, with the ultimate goal of contributing to the development of effective therapeutic strategies based on mentalization. These strategies are intended to improve reflective function and psychological flexibility for the vulnerable population of people with autism.
Data availability statement
The data that support the findings of this study are available on request from the corresponding author.
Conflict of interest statement
The authors declare that there is no conflict of interest.