Introduction

Adolescence is a sensitive developmental period between 10 and 18 years characterised by significant physical, neurobiological, and social changes that typically produce beneficial outcomes (Larsen & Luna, 2018; Rapee et al., 2019). However, global estimates indicate that psychopathology onset peaks at approximately 14.5 years (Solmi et al., 2021). Concerningly, younger adolescents experiencing non-specific symptoms (e.g., psychological distress) are at an increased risk of transitioning past subthreshold stages of psychopathology to discrete disorders characterised by increased symptom severity, functional impairment, and neurobiological dysfunction (Cross et al., 2016; Enticott et al., 2022; Hickie et al., 2019; Iorfino et al., 2019). Therefore, research on contributors to emerging psychological distress in younger adolescents is essential to inform evidence-based, preventative interventions for psychopathology. To date, research has investigated various contributors to adolescent psychological distress; two factors proposed are metacognitions and executive functions.

Metacognitions

Metacognitions include beliefs, knowledge, or strategies involved in the interpretation or regulation of one’s thinking (Wells, 2000). According to the self-regulatory executive function (S-REF) model (Wells & Matthews, 1996), cognitive and metacognitive processes are spread across three interacting levels, which include low-level processing, cognitive style, and meta-system. The S-REF model proposes that psychopathology is maintained by the cognitive attentional syndrome (CAS). The CAS involves conceptual components (e.g., perseverative processing in the form of worrying or rumination), attentional components (e.g., fixation of attention on threat-related stimuli), thought control strategies (e.g., thought suppression), and behavioural, cognitive, and emotional avoidance. Together, these processes are part of the individual’s strategy for dealing with threat, self-discrepancies, and any emotion aroused by threats or self-discrepancies (Wells, 2013). However, through these processes, the CAS prolongs biased and repetitive processing of threat-related thought, emotion, and belief, as well as exacerbates emotional distress, and impairs cognitive flexibility, fostering a negative feedback loop that maintains psychopathology. Essentially, the CAS is a perseverative thinking style that consists of worry, rumination, threat monitoring, and typically leads to ineffective coping strategies (e.g., thought suppression or avoidance) that produces maladaptive outcomes and prevents adaptive learning (Wells, 2013).

Research supports the proposition that the CAS can maintain psychosis (Sellers et al., 2018), depression (Normann et al., 2021), and anxiety (Nordahl & Wells, 2019). However, CAS-activation is theoretically contingent upon the endorsement of either positive or negative metacognitive beliefs—herein referred to as positive or negative metacognitions (Wells, 2009). Positive metacognitions involve beliefs that repetitive thinking is beneficial or necessary (e.g., worrying about an upcoming test prevents failure). Negative metacognitions concern the danger or uncontrollability of thoughts or feelings (e.g., worrying is uncontrollable; Wells, 2013). Therefore, based on the assertions of the S-REF model, adolescents who endorse more positive and negative metacognitions are likely to be more psychologically distressed.

Metacognitions and Adolescent Psychological Distress

Metacognitions have been consistently associated with adolescent psychological distress (Campbell et al., 2018; Esbjørn et al., 2015; Köcher et al., 2021). However, discrepant relationships have been reported for the positive metacognition subscale, as assessed by the Metacognitive Questionnaire for Adolescents (MCQ-A; Cartwright-Hatton & Wells, 2004). For example, Debbané et al. (2012) found positive metacognitions were not significantly associated with anxiety in a cross-sectional study of N = 179 Swiss adolescents aged 12–19. Conversely, Ellis and Hudson (2012) reported increased positive metacognitions were associated with increased levels of worry, anxiety, and depression in N = 123 Australians aged 12–17, with no significant sex differences. Small to moderate significant positive associations between positive metacognition and anxiety were also found via meta-analysis (Köcher et al., 2021). Despite discrepant findings, which may be accountable to differences in samples, measures, or uncontrolled factors that increase CAS-vulnerability (e.g., executive deficits), increased positive metacognitions are typically associated with elevated psychological distress in adolescents. Together, this research supports the S-REF model as positive metacognitions require intrusive cognitions to be perceived as helpful, which ultimately increases CAS-activation that subsequently maintains psychological distress (Wells, 2009).

Similarly, negative metacognitions have also been consistently associated with adolescent psychological distress (Köcher et al., 2021; Smith & Hudson, 2013; White & Hudson, 2016). For example, using the MCQ-A Dutch version, Wolters et al. (2016) identified increased negative metacognitions were significantly associated with increased anxiety and depression in N = 317 non-clinical adolescents aged 12 to 18. This research supports the S-REF model, which proposes negative metacognitions directly increase psychological distress and CAS-activation. Overall, there is evidence of theoretical and empirical relationships between positive and negative metacognitions, the CAS, and psychological distress in non-clinical adolescent samples. While metacognitions in adolescents may be influenced by a range of neurodevelopmental factors, including the impacts of early life stress and trauma, which can affect brain circuits involved with emotional and cognitive processing, there appears to be a key role in prefrontal cortex development and the associated executive functions.

Executive Functions

Executive Functions (EFs), which involve higher-order cognitive skills that regulate goal-directed behaviour (Snyder et al., 2015), are conceptualised by the unity and diversity model (Miyake & Friedman, 2012) into three orthogonal yet interrelated components. These include (1) updating working memory (i.e., constant monitoring and rapid revision of working-memory contents), (2) set-shifting (i.e., moving flexibly between multiple tasks or mental sets), and (3) inhibition (i.e., deliberate overriding of prepotent thought or behavioural responses; Friedman & Miyake, 2017). The interrelated component, termed common EF, involves a general ability to maintain a task set and biases lower-level processing. This basic ability is necessary for all three EFs and is particularly critical for inhibition (Friedman & Miyake, 2017). Throughout adolescence, significant neurobiological changes involving the frontal lobe subserve the development of EFs (Moura et al., 2017). Notably, research indicates EFs are best explained by a single-factor before age 12 and by a three-factor model in late adolescence (Laureys et al., 2022). This supports the unity and diversity model and suggests a gradual differentiation throughout adolescence. Importantly, poor executive functioning (EF) has been shown to be associated with increased psychological distress in adolescents with a range of developmental disorders (Halse et al., 2022).

When considered with the S-REF model, poor common EF appears to conceptually overlap with the maladaptive processing aspects of the CAS, which is theorised to be precipitated by metacognitions, maintain psychopathology, and predominantly manifest as a preservative and biased thinking style. For example, an inability to inhibit intrusive thoughts or emotions (i.e., poor inhibition) in favour of task-relevant behaviour is comparable to maladaptive perseverative processing of intrusions that produces maladaptive outcomes by the CAS (Kraft et al., 2017). Furthermore, research has also identified relationships between EFs and aspects of the CAS, supporting that poorer EFs can predict adolescent worry, rumination, thought suppression, and avoidant coping (Geronimi et al., 2016; Hu et al., 2024; Mennies et al., 2021; Wante et al., 2017). Thus, adolescents with poorer common EF are likely to be psychologically distressed as they may struggle to inhibit the initial intrusive metacognitions that precipitate the CAS, disengage from the perseverative nature of the CAS, or maintain attentional focus on adaptive goal-relevant pursuits. Therefore, poorer common EF may contribute to increased adolescent psychological distress. Furthermore, poorer common EF may strengthen the relationship between metacognitions and adolescent psychological distress. As such, the potential direct and moderating role of EF on the association between psychological distress and metacognitions in adolescents requires investigation.

Executive Functions and Adolescent Psychological Distress

Cross-sectional and longitudinal research utilising child and adolescent samples has identified pervasive deficits in multiple EFs across various psychological disorders (Halse et al., 2022; Romer & Pizzagalli, 2021). For example, psychopathology has been associated with poorer performance on a single EF composite task and inhibition, shifting, and updating subtests in children and adolescents (Bloemen et al., 2018; Kertz et al., 2016; Martel et al., 2017). Research has also found that in older adolescents, the relationship between EF and psychopathology in N = 292 Americans was attributed to poorer common EF (Snyder et al., 2019), which positively correlates with and can be measured using inhibition tasks (Miyake & Friedman, 2012). Conversely, in their study of N = 9,498 community-dwelling adolescents (aged 8–21 years), White et al. (2017) found the relationships between poorer inhibition, common EF, and general psychopathology symptoms were stronger for younger participants. These findings collectively indicate that poor common EF may be a transdiagnostic risk factor for emerging psychopathology in youth. One explanation for this finding may be that adolescents with poorer common EF may struggle to inhibit low-level processing (e.g., intrusive emotions or thoughts) or metacognitions to maintain adaptive behavioural or cognitive regulation. Consequently, this may activate the CAS and heighten psychological distress (Wells, 2009).

Poor common EF has been associated with adolescent psychological distress and as noted above adolescent worry, rumination, and maladaptive coping. Given this, it is conceivable that poorer common EF overlaps with or impacts the maladaptive processing aspects of the CAS, which is theorised to be activated by metacognitions and predominantly manifest as a preservative and biased thinking style. Therefore, it is possible the relationship between metacognitions and psychological distress may be stronger in adolescents with poorer common EF, as they may struggle to inhibit metacognitions and disengage from the CAS (i.e., worrying and rumination). Accordingly, the potential moderating role of executive functioning on the metacognitions and adolescent psychological distress relationship requires empirical investigation.

Impact of Executive Functions on Metacognitions and Psychological Distress

Cross-sectional research with adults has supported an association between increased metacognitions and lower self-reported attentional control and increased shifting errors (Barua et al., 2020; Fernie et al., 2016; Kraft et al., 2017; Nordahl et al., 2019; Spada & Roarty, 2015). However, despite the theoretical and empirical connections between metacognitions and EFs, research investigating their interactive contribution to adolescent psychological distress is limited. Given the lack of evidence in adolescents, the following summary of research in older and clinical populations provides some insight into the potential moderating role of EFs in the link between metacognitions and psychological distress. Firstly, when investigating moderation by self-reported attentional control, Fergus et al. (2012) found the effect of CAS-activation—which is theoretically activated by metacognitions—on anxiety and depressive symptoms, strengthened in N = 456 American undergraduates (Mage = 19.70 years), as attentional control decreased. Furthermore, poorer self-reported EF was found to strengthen the effect of positive—but not negative—metacognitions on post-traumatic stress symptoms in N = 469 trauma-exposed adults (Mage = 35.90; Bardeen & Fergus, 2018). Of note, research has indicated post-traumatic stress symptoms and psychological distress symptoms are significantly and positively correlated (Liang et al., 2022), highlighting the relevance of the findings from Barden and Fergus (2018). Finally, Reinholdt-Dunne et al. (2019) found the positive relationship between total metacognitions and symptoms of anxiety and depression weakened at higher levels of self-reported attentional control in a non-clinical sample of N = 182 Danish children aged 7–14.

To re-iterate, the above-mentioned findings were in a range of populations, however, this research indicates increased metacognitions and poorer EF uniquely and interactively predict increased psychological distress in both clinical and non-clinical groups, and across various ages. Importantly, this evidence suggests that poor EF may moderate the relationship between metacognitions and psychological distress. That is, the impact of dysfunctional metacognitions on increased psychological distress may be worsened for individuals with poorer EFs. This interaction extends the S-REF model, potentially indicating individuals with poor EF cannot disengage from the CAS to maintain attentional focus on adaptive goal-pursuits. However, as previous research with adolescents has only assessed metacognitions as a unidimensional construct it remains uncertain whether positive and negative metacognitions uniquely interact with EF in adolescent community samples. Furthermore, self-report EF measures are subject to bias, and their validity in reflecting actual cognitive abilities has been questioned (Løvstad et al., 2016; Williams et al., 2017). Therefore, further research utilising objective EF measures is required. One objective EF measure is the Multi-Tasking Test, a subtest from the Cambridge Neuropsychological Test Automated Battery (CANTAB, 2009), which assesses an individual’s ability to manage conflicting information and inhibit incongruent task-irrelevant information. Specifically, an outcome measure from this subtest, called incongruency cost, measures the difference between median response latency across types of trials. This outcome measure of incongruency cost controls for participants’ baseline processing speed (CANTAB, 2009), ensuring a purer measure of participants’ inhibitory abilities and one that is not influenced by the biases known to impact self-reports.

The Present Study

The CAS conceptually overlaps with poor EFs, particularly poorer common EF. However, despite theoretical connections, current research is limited to individual contributions or relies predominantly upon adult samples or self-report questionnaires.

The rationale for the current study was to examine the relationships between positive metacognitions, negative metacognitions, EF, and psychological distress in early adolescents utilising objective EF measures, to elucidate the nature of their unique and interactive relationships. Specifically, the study aims to extend previous research by investigating the relationships between positive metacognitions, negative metacognitions, EF, and psychological distress in early adolescents aged 12–14 with an objective inhibitory measure representing common EF. Firstly, it was hypothesised that higher positive metacognitions (H1a) and negative metacognitions (H1b) would significantly predict increased psychological distress. Secondly, it was hypothesised that poor common EF, represented by a higher incongruency cost, would significantly predict increased psychological distress (H2). Lastly, it was hypothesised that the relationship between increased psychological distress and increased positive (H3a) and negative metacognitions (H3b) would be further increased by poorer common EF (i.e., higher incongruency cost).

Method

Participants, Procedure, and Design

The present study analysed data collected as part of the Longitudinal Adolescent Brain Study (LABS). LABS is currently being conducted at the Thompson Institute, University of the Sunshine Coast (UniSC), and is an ongoing multimodal, community-based, longitudinal cohort study that aims to investigate adolescent behaviour, mental health, and neurobiological development. Participants, recruited between ages 12 (first year of secondary education, Grade 7) to 15, are tested up to a maximum of 15 times, every four months until they reach age 17 (final year of secondary education, Grade 12). Inclusion criteria required participants to be aged 12 and in the first year of secondary education (Grade 7) to 15 years at intake and proficient in verbal and written English. Exclusion criteria included self-reported head injury involving loss of consciousness (for ≥ 30 min), major medical illness, intellectual disability, or major neurological disorder, as such factors can influence cognitive functioning (Pelicioni et al., 2020; Tsermentseli et al., 2018). The LABS design, recruitment, and procedure is outlined in Beaudequin et al. (2020). Here, only relevant components of the protocol will be discussed.

A correlational, cross-sectional design was implemented in this study. There was one criterion variable (total psychological distress score) and four predictor variables (biological sex, positive metacognitions, negative metacognitions, and EF). The sample size was estimated using a priori power analysis in G*Power 3.1.9.7 (Faul et al., 2007), using a moderate effect size in line with prior literature (Reinholdt et al., 2019). The analysis indicated that 85 participants were required to detect a moderate effect size (f2 = 0.15) with sufficient power of 0.80 (α = 0.05) for linear regression with four predictors.

Data from N = 87 adolescents were analysed. Two cases were removed due to missing data, resulting in a final sample (N = 85) comprised of n = 42 females and n = 43 males aged 12.43 to 14.04 years (M = 13.22, SD = 0.45). Most participants were right-handed (89.4%), 8.2% were left-handed, and 2.4% were ambidextrous. Participants were enrolled in seventh (47.1%), eighth (51.8%), and ninth grade (1.2%) and attended both public (48.2%) and private (51.8%) schools. Participants received a $20 gift voucher as reimbursement for participation following each visit.

Measures

Demographic Questionnaire

Participants completed a demographic and brief medical and mental health history questionnaire. The current study included data on age, biological sex, handedness, and school year (i.e., grade) and type (public/private).

Psychological Distress

The 10-item Kessler Psychological Distress Scale (K10; Kessler & Mroczek, 1992), which measures non-specific anxiety and depressive symptoms, was used to measure participants’ psychological distress. Participants self-estimated the intensity of various emotional experiences over the past 30-days (e.g., “During the last 30 days, about how often did you feel nervous?”) on a 5-point Likert scale ranging from 1 (none of the time) to 5 (all of the time). Scores were summated and valid scores ranged from 10 to 50. Higher scores indicated increased psychological distress. Overall scores ≥ 20 indicated a psychological disorder classified by differing levels of severity (i.e., 20–24 = mild, 25–29 = moderate, 30–50 = severe; Victorian Government Department of Human Services, 2002). The K10 has demonstrated high sensitivity (85%), high specificity (83%), good internal consistency (α = 0.86) within adolescent samples (Chan & Fung, 2014), and high convergent validity with anxiety and depression measures (Kessler et al., 2002; Lace et al., 2019).

Metacognitive Beliefs

The 30-item Metacognitions Questionnaire for Adolescents (MCQ-A; Cartwright-Hatton & Wells, 2004) was used to measure participants’ positive and negative metacognitions. Participants rated their agreement with 30 statements on a 4-point Likert scale ranging from 1 (do not agree) to 4 (agree very much). Five subscales are generated which include (1) positive beliefs about worry, (2) negative beliefs about worry, (3) cognitive confidence, (4) beliefs about the need to control thoughts, and (5) cognitive self-consciousness. This study only used the positive (MCQ-PBW) and negative beliefs about worry (MCQ-NBW) subscales (i.e., positive and negative metacognitions). The MCQ-PBW (six items, ‘positive metacognitions’) measures the extent to which an individual believes that perseverative thinking is useful (e.g., “worrying helps me cope”). The MCQ-NBW (six items, ‘negative metacognitions’) measures the extent to which an individual believes that perseverative thinking is uncontrollable and dangerous (e.g., “my worrying is bad for me”). Scores were summated and could range from 6 to 24 (per subscale). Higher scores indicated higher endorsement of metacognitions. The MCQ-A has received the most extensive psychometric support as an adolescent measure of metacognitions (Myers et al., 2019), demonstrating robust internal consistency overall (α = 0.91) and good internal consistency for both positive (α = 0.88) and negative (α = 0.84) metacognition subscales. Moreover, the MCQ-A has correlated with anxiety and depression measures and accurately differentiated between clinical and non-clinical samples of anxious adolescents (Cartwright-Hatton et al., 2004), demonstrating good convergent and discriminant validity.

The Multi-Tasking Test

The Multi-Tasking Test (MTT), which is a computerised subtest from the Cambridge Neuropsychological Test Automated Battery (CANTAB, 2009), was used to measure participants’ common EF. The MTT measures a participant’s ability to manage conflicting information and inhibit incongruent task-irrelevant information. A singular black arrow pointed to the left or right, was presented on the left or right side of the screen. Each trial displayed a cue in the upper screen centre that indicated whether to identify the arrow’s direction or side. Participants were required to respond to the cue correctly and quickly by selecting the left or right button at the bottom of the tablet screen. The task, which lasted 8-min and included 40 practice and 120 assessment trials, was separated into three blocks. Voice over provided instructions followed by a practice session with feedback (i.e., voiceover reporting correct or incorrect for each response) preceded each block to ensure participants understood the task. In the first two blocks, the rule was consistent across trials (e.g., only cued “direction” or “side”). In the third block, the rule randomly switched across trials. Across all blocks, some trials displayed congruent stimuli (e.g., right-directed, right-side) whereas other trials displayed incongruent stimuli (e.g., right-directed, left-side).

This study used the outcome measure of incongruency cost (i.e., the difference between median response latency, from stimulus presentation to button press, on the trials that were incongruent versus the trials that were congruent). Two scores are calculated for each participant across trials. One score represents the median time from stimulus presentation to correct response button press for incongruent trials, another represents the median time from stimulus to correct response button press for congruent trials. Incorrect responses did not contribute to this outcome measure. The difference between these two median scores is then calculated to determine a new score (i.e., incongruency cost). This calculation controlled for participants’ baseline processing speed, ensuring a purer measure of participants’ inhibitory abilities (CANTAB, 2009). A positive score indicates faster performance on congruent trials and a negative score indicates faster performance on incongruent trials. A higher incongruency cost, measured in milliseconds (ms), indicates the participant takes longer to process conflicting information. Scores can range between −1900 ms to 1900 ms (CANTAB, 2009). The underlying premise of this task is that the greater the delay in responding (time from stimulus presentation to accurate response), the more cognitive effort was required, thus reflecting poor executive control. The MTT has demonstrated reasonable sensitivity (82.80%) and specificity (74.50%) for detecting mild cognitive impairment and good internal consistency (α = 0.85) as a Stroop-measure within adolescent samples (Martínez-Loredo et al., 2017; Wang et al., 2022).

Statistical Analyses

The IBM Statistical Package for Social Sciences (SPSS Version 28.0) was used for all analyses. Results were considered significant at an alpha of 0.05. To assess whether positive metacognitions, negative metacognitions, and EF would positively predict psychological distress, a multiple hierarchical regression was conducted. Sex was dummy coded (male = 0, female = 1) and entered into the first block, to control for its influence on the criterion (Van Droogenbroeck et al., 2018). The second and third block included predictor variables relevant to the hypotheses, entered into the model in their theoretical order of importance for predicting the criterion (Field, 2018). Thus, positive metacognitions and negative metacognitions were entered into the second block to assess H1. EF was entered into the third block to assess H2.

To assess whether EF would strengthen the positive association between metacognitions and psychological distress (H3), two separate multiple regression analyses were conducted using PROCESS Macro 4, Model 1 (Hayes, 2022). PROCESS has several advantages. Firstly, PROCESS can centre predictor scores to create standardised coefficients and compute the interaction term. Secondly, PROCESS provides simple slopes analysis data to examine the conditional effect of the focal predictor on the criterion at varying values of the moderator. Lastly, PROCESS indicates the level of the moderator at which the focal predictor–criterion relationship becomes non-significant via the Johnson-Neyman conditional effects (Hayes, 2022).

Moderation regression tests the conditional effect of X on Y, where X is the predictor, W is the moderator that exerts a conditional effect on the relationship between X and Y, and Y is the criterion (Hayes, 2022). In this study, psychological distress was entered as the criterion (Y), EF was entered as the moderator (W), positive or negative metacognitions (X) were entered as the predictor, and sex and positive or negative metacognitions were entered as covariates. A conceptual model of the hypothesised relationship is provided in Fig. 1.

Fig. 1
figure 1

Hypothesised model of metacognitions, executive function, and psychological distress. Figure adapted from Igartua and Hayes (2021)

Full size image

Results

Assumption Testing

Data (N = 87) was screened for errors and missing values. Two cases were removed due to non-completion of the MTT. An analysis of standard residuals detected three univariate outliers outside ± Z = 3.29. Significance of the Shapiro-Wilks test (p < 0.05), visual examination of histograms and Q-Q Plots, and positively skewed K10 (Zskew = 6.18), MCQ-PBW (Zskew = 5.95), and MCQ-NBW (Zskew = 4.51) indicated the assumption of univariate normality was violated. Although positive skew is expected in a community sample (Butterworth et al., 2020), a Log10 transformation was applied (Tabachnick & Fidell, 2012). Two separate analyses were conducted, one with transformed and the other with untransformed data, to determine if transformation impacted the model and improved assumption violations. Transformation improved normality slightly for the K10 (Zskew = 3.57) and MCQ-PBW (Zskew = 3.46), largely for MCQ-NBW (Zskew = 1.95), and the histograms appeared normally distributed. Transformation did not appear to impact the model significance or direction of effects. A visual examination of scatterplots indicated the assumption of linearity between the criterion and each predictor was met. Therefore, to maximise power given the small sample size (Field, 2018), univariate outliers were retained, and analyses were conducted with data transformed for N = 85.

Assessment of multivariate outliers indicated no Mahalanobis’ distance scores exceeded the critical χ2 of 18.47 (p < 0.001) for a regression model with four predictors (Tabachnick & Fidell, 2012). Cook’s distance was below 1 for all cases, indicating no cases exhibited undue influence on the estimates within the regression model, also supporting the decision to retain the univariate outliers (Cook & Weisberg, 1982). The data met the assumption of independent errors, as the Durbin-Watson value of 1.98 was within the required range of 1.5–2.5 (Field, 2018). There was no presence of multicollinearity, with VIF < 10, Tolerance > 0.20, and bivariate correlations between predictor variables acceptable (r ≤ 0.85; Cohen, 2003; Field, 2018), with non-zero variances (MCQ-PBW = 13.09, MCQ-NBW = 16.99, MTT-EF = 2202.22, K10 = 46.50). A visual inspection of residual histogram and ZRED against ZRESID scatterplots indicated residuals were distributed non-symmetrically, curved, and cone-shaped, violating the assumptions of normality, linearity, and homoscedasticity of residuals. However, this was also improved by transformation, with all these residual assumptions considered met.

Preliminary Analysis

Mean scores, standard deviations, the range of scores, and Cronbach’s alpha for each variable are presented in Table 1. All measures demonstrated good psychometric properties for use with adolescent populations. Mean psychological distress, negative metacognitions, and incongruency cost scores were slightly but not significantly above normative values for adolescents (i.e., listed in CANTAB, 2009; Cartwright & Wells, 2004; Kessler & Mroczek, 1992; Slade et al., 2011; Slykerman et al., 2018). However, mean positive metacognitions scores were significantly below normative values for adolescents, t(84) = −4.75, p =  < 0.001. Together, this indicates that the sample was predominantly psychologically healthy and reported a lower endorsement of maladaptive beliefs that worrying is helpful (i.e., positive metacognitions) relatives to norms.

Table 1 Descriptive statistics, Cronbach’s alpha, and zero-order correlations between study variables
Full size table

Correlations between study variables are also presented in Table 1. Results indicated moderate to strong, significant, positive correlations between positive metacognitions, negative metacognitions, and psychological distress. EF was not significantly correlated with psychological distress. Biological sex had small to moderate, significant positive correlations with positive metacognitions, negative metacognitions, EF, and psychological distress.

The impact of sex was assessed using a multivariate analysis of variance (MANOVA). The MANOVA demonstrated significant differences between males and females on the combined variables [V = 0.188, F(4, 80) = 4.64, p = 0.002, η2 = 0.19, 1 – β = 0.94]. Of the univariate tests, sex impacted levels of psychological distress, [F(1, 83) = 9.00, p < 0.004, η2 = 0.10, 1 – β = 0.84] and negative metacognitions, [F(1, 83) = 18.74, p < 0.001, η2 = 0.18, 1 − β = 0.99]. Females scored higher on both variables.

Main Analyses

Hypothesis 1 and 2: Metacognitions and Executive Function on Psychological Distress

A hierarchical multiple regression analysis was conducted to assess whether biological sex, positive metacognitions, negative metacognitions, and EF positively predicted psychological distress. Results are presented in Table 2. In the first block, sex significantly accounted for increased variability in psychological distress, [R2 = 0.10, F(1, 83) = 9.00, p = 0.004]. The addition of positive metacognitions and negative metacognitions in the second block significantly accounted for an additional 37% of variance in psychological distress scores, [ΔR2 = 0.37, ΔF(2, 81) = 27.67, p < 0.001]. The addition of EF in the third block did not significantly contribute to variance in psychological distress, [ΔR2 = 0.00, ΔF(1, 80) = 0.24, p = 0.629]. In combination, the four variables significantly and positively predicted psychological distress, [R2 = 0.47, F(4, 80) = 17.42, p < 0.001]. According to Cohen’s (1988) conventions, a combined effect of this magnitude is considered large (f2 < 0.35). Negative metacognitions were the strongest predictor and accounted for 19.36% of significant unique variance, while positive metacognitions predicted approximately 3.06% of significant unique variance in psychological distress. Therefore, increased positive and negative metacognition scores predicted increased levels of adolescent psychological distress, supporting H1a and H1b. However, higher incongruency cost (i.e., poorer EF) did not significantly predict increased psychological distress, failing to support H2.

Table 2 Hierarchical multiple regression analysis predicting psychological distress from predictors
Full size table

Hypothesis 3: Moderating Role of Executive Function

The results for the PROCESS multiple regression analysis are summarised in Table 3. To assess whether EF strengthened the positive effect of positive metacognitions on psychological distress (H3a), positive metacognitions was entered as the predictor and psychological distress was entered as the criterion. To account for their influence as identified in the preliminary analysis, sex and negative metacognitions were entered as covariates. The model indicated a significant direct positive effect of positive metacognitions on psychological distress. The direct effect of EF on psychological distress was not significant. The interaction effect of EF on positive metacognitions and psychological distress was not significant. Therefore, the relationship between positive metacognitions and psychological distress was not significantly moderated by EF.

Table 3 Multiple regression analyses for moderation by executive function
Full size table

To assess whether EF strengthened the positive effect of negative metacognitions on psychological distress (H3b), negative metacognitions was entered as the predictor and psychological distress was entered as the criterion. To control for their influence, sex and positive metacognitions were entered as covariates. The model indicated a significant direct positive effect of negative metacognitions on psychological distress. The direct effect of EF on psychological distress was not significant. The interactive effect of EF on negative metacognitions and psychological distress was positive and significant, accounting for approximately 6% of the variance in psychological distress scores, [ΔR2 = 0.06, F(1, 79) = 11.68, p = 0.001]. Therefore, the positive effect of negative metacognition on psychological distress was significantly strengthened by EF.

As presented in Fig. 2, the simple slopes analysis tested the significance of the interaction effect at varying levels of EF. Negative metacognitions and psychological distress were strongly related at high incongruency costs (indicating poorer EF, 1 SD above the mean) but not significantly related at low incongruency costs (indicating better EF, 1 SD below the mean). Thus, participants who demonstrated higher incongruency costs reported increased levels of psychological distress when negative metacognitions scores were high, compared to when negative metacognitions scores were low. Participants with average incongruency costs also reported higher psychological distress scores when negative metacognitions scores were high, compared to when negative metacognitions scores were low. However, participants with lower incongruency costs (i.e., better EF) displayed similar levels of psychological distress at low and high levels of negative metacognitions. The Johnson-Neyman technique (Hayes, 2012) indicated the relationship between negative metacognitions and psychological distress became unconditional when incongruency costs were below −44.71. Therefore, negative metacognitions did not directly impact psychological distress scores for participants who scored below −44.71.

Fig. 2
figure 2

The moderating impact of executive function on negative metacognitions and psychological distress. Regression lines demonstrating the negative metacognition–psychological distress relationship at low (−1 SD) and high (+ 1 SD) EF incongruency costs. Y-axis has restricted range to better illustrate the magnitude of difference

Full size image

Discussion

This study revealed that higher endorsement of positive and negative metacognitions significantly predicted increased psychological distress, supporting H1a and H1b. Furthermore, increased incongruency costs (i.e., poorer EF) significantly strengthened the relationship between negative metacognitions and psychological distress, supporting H3b. However, poorer EF did not significantly predict increased psychological distress nor strengthen the positive metacognitions–psychological distress relationship, failing to support H2 and H3a. Overall, these findings demonstrate positive and negative metacognitions significantly and positively predict adolescent psychological distress. Furthermore, the positive effect of negative metacognitions on psychological distress is significantly exacerbated in adolescents with poorer EF.

Preliminary Findings

The presence of significant, positive bivariate relationships between psychological distress, positive metacognitions, and negative metacognitions found in this study is consistent with the established relationships between these variables (Köcher et al., 2021). Furthermore, the higher levels of psychological distress self-reported by females here is consistent with prior literature (Van Droogenbroeck et al., 2018). Although research has predominantly suggested that there are no significant sex differences in metacognitions (Ellis & Hudson, 2011; Smith & Hudson, 2013; White & Hudson, 2016), this study supports emerging literature indicating that females report significantly more negative metacognitions (Shakeshaft et al., 2020). Overall, these results highlight the importance of controlling for the effect of sex when investigating relationships between metacognitions and psychological distress in future research.

Metacognitions on Psychological Distress (H 1)

As predicted, increased positive and negative metacognition scores significantly predicted increased levels of psychological distress, beyond the effect of sex. The effect size for their collective contribution was large, suggesting that metacognitions are meaningfully and clinically significant in understanding adolescent psychological distress. These findings suggest that adolescents who endorse more positive and negative metacognitions are more likely to experience symptoms of psychological distress. This is consistent with previous studies (Ellis & Hudson, 2012; Köcher et al., 2021; Lachat Shakeshaft et al., 2020; Schultz et al., 2023; Şen Demirdöğen et al., 2021; White & Hudson, 2016), but is partially consistent with Debbané et al. (2012), who only found negative metacognitions were associated with adolescent anxiety. The current results are congruent with the S-REF model, which proposes that as metacognitions are experienced as intrusions, they are appraised in a perseverative and biased manner which activates rumination and worry (Wells, 2009). Specifically, this self-directed, biased, and negative processing causes adolescents to attend to evidence that supports the initial intrusion, thereby exacerbating their psychological distress (Wells, 2013). Overall, these findings provide support for the S-REF model, validating the relationships between metacognitions and psychological distress in adolescents.

Executive Functioning on Psychological Distress (H 2)

Contrary to the hypothesis, poorer EF (i.e., higher incongruency costs) did not significantly predict increased psychological distress. This indicates that adolescents with poorer EF were not more likely to be psychological distressed. These results differ from literature that has found poor EF predicted psychological distress in younger adolescents (Kertz et al., 2016; White et al., 2017), but is consistent with research that has only observed an association between EF performance and internalising symptoms in older adolescents (Snyder et al., 2019). The study sample may provide some explanation for this discrepancy. Firstly, a larger proportion attended private schools (51.80%) compared to the typical percentage enrolled in private schools (34.90%; Australian Bureau of Statistics, 2021). Research suggests that adolescents who attend private schools are often from well-educated, higher income families, with well-organised parental involvement (Park & Bills, 2015), which all predict improved academic, social, and emotional outcomes in adolescents (Pianta & Ansari, 2018). Secondly, the sample comprised of younger adolescents. It has been speculated that at younger ages (i.e., 13–16), adult caretakers can compensate for an adolescent’s poor EF (e.g., correcting maladaptive social behaviours), preventing EF deficits from translating into behaviours that generate psychological distress (e.g., interpersonal conflict; Snyder & Hankin, 2016). Therefore, parental support or resources may have compensated for any deleterious effects of poorer EF on psychological distress in this sample. In addition, the measurement of psychological distress may explain the null findings. Research investigating relationships between EF and adolescent psychopathology (Halse et al., 2022; Romer et al., 2021) has found that poorer EF was associated with adolescent psychopathology in community samples, when measured as a latent construct (i.e., the “P factor”) representing shared psychopathological mechanisms across multiple mental disorders (e.g., fear, distress, and externalising). Therefore, in community samples, future research should investigate the relationships between EF and psychological distress as measured by a latent factor of psychopathology.

Alternatively, the measurement of common EF used here may explain why poor EF did not predict psychological distress. Although the MTT theoretically overlaps with common EF (Friedman & Miyake, 2017), Synder et al. (2015) recommends that multiple measures corresponding to each component of the unity and diversity model—aggregated into a latent variable—is required for effective measurement. Furthermore, the CANTAB subtests were originally devised for detecting EF impairments in those with dementia (Fray & Robbins, 1996). Therefore, future research should incorporate multiple objective EF measures that are aggregated, targeted, and sensitive enough to detect the magnitude of deficits expected in non-clinical adolescent populations. For example, Miyake et al. (2000) has developed multiple tasks to assess inhibition, shifting, and updating which could be aggregated to improve clinical utility. Overall, further research using additional objective measures is needed to elucidate the relationship between common EF and psychological distress in younger adolescents.

Moderating Role of Executive Functioning (H 3)

The results for the third hypothesis will be discussed in order of relative value, as determined by the significance of the moderating impact of EF.

Negative Metacognitions on Psychological Distress

The current finding revealed that increased incongruency costs (i.e., poorer EF) significantly strengthened the effect of negative metacognitions on psychological distress. That is, adolescents who endorsed more beliefs that worry is uncontrollable or damaging combined with poorer EF were more likely to report higher levels of psychological distress. This finding is consistent with prior research that demonstrates better self-reported EF weakened the association between psychological distress symptoms and metacognitions (Reinholdt-Dunne et al., 2019) yet inconsistent with others that found poorer self-reported EF did not strengthen the negative metacognitions–post-traumatic stress symptoms relationship (Bardeen & Fergus, 2018). This discrepancy may highlight the differential impact of EF abilities on the psychological distress–metacognitions relationship relative to their impact on the post-traumatic stress symptoms–metacognitions relationship. Together, this interaction supports an integration of the S-REF model and unity and diversity model (Miyake & Friedman, 2012; Wells, 2009). Specifically, adolescents with emerging inhibitory difficulties—difficulties ignoring or overriding thought or behaviour—may be more vulnerable to the CAS (i.e., perseverative thinking style) and the perpetual cycle it facilitates, once activated by negative metacognitions. This is possibly because negative metacognitions are highly intrusive (Wells, 2009), rendering their activation particularly difficult to ignore. The inability to ignore negative metacognitions (e.g., belief that worrying about assessments is uncontrollable), may interfere with task-relevant, adaptive behaviour (e.g., studying). The outcome of this interference (e.g., inadequate study) may exacerbate psychological distress and the CAS (e.g., increased worrying), depleting cognitive resources to support inhibition. Together, this process may be appraised as supporting initial negative metacognitions, cultivating a perpetual cycle that maintains distress and inhibitory deficits.

Positive Metacognitions on Psychological Distress

Unexpectedly, increased incongruency costs (i.e., poorer EF) did not significantly strengthen the association between positive metacognitions and psychological distress. That is, adolescents with poorer EF and increased beliefs that worrying is beneficial were not more likely to report higher levels of psychological distress. This finding contradicts earlier research which found EF moderated the association between psychological distress and metacognitions (Reinholdt-Dunne et al., 2019) and positive metacognitions and post-traumatic stress symptoms (Bardeen & Fergus, 2018). This discrepancy may be explained by these studies use of self-report EF measures, which may not reflect actual executive abilities. Furthermore, as previously discussed, while the current study removes the bias of self-reported EF, the objective measurement of common EF used here may require additional measures to accurately capture this skill in early adolescence. Therefore, future research must investigate whether common EF, assessed as a latent factor, strengthens the association between positive metacognitions and psychological distress.

The current finding provides additional support for the S-REF model (Wells, 2009). Specifically, positive metacognitions are theoretically experienced as less intrusive relative to negative metacognitions (Wells, 2009), which indicates that better inhibitory abilities may not be required to ignore them. Therefore, other potential moderators need to be explored. Set-shifting deficits are one potential moderator which have been associated with adolescent psychological distress (Bloemen et al., 2018; Kertz et al., 2016). Potentially, adolescents with increased positive metacognitions and shifting deficits may struggle to change coping strategies (e.g., seeking support) or mental sets (e.g., worrying is not helpful) despite contradictory outcomes challenging their positive metacognitions (e.g., failing a test despite using worry to prepare). As this shifting deficit may prolong CAS-activation, theoretically adolescents’ psychological distress should increase.

Other potential moderators include the simultaneous endorsement of other metacognitions or coping strategies. For example, Wells (2000) has proposed the co-occurrence of negative and positive metacognitions predicts greater psychopathology. Research has also demonstrated adaptive metacognitions (e.g., I can refrain from worrying) predicts adaptive coping (Beer & Moneta, 2012), which can weaken the depressive symptoms–rumination relationship (Thompson et al., 2010). Together, this research suggests that lack of negative metacognitions or the presence of adaptive metacognitions and coping may weaken the influence of positive metacognitions on psychological distress. This outcome may result from an increased likelihood of adaptively resolving discrepancies, which terminates the CAS (Wells, 2009). Therefore, these potential moderators require investigation.

Implications

Collectively, the current findings indicate increased positive and negative metacognitions can increase adolescent psychological distress, particularly for those with poorer EF. These results provide theoretical support and extend the S-REF model by validating the individual and interactive relationships between EF, metacognitions, and psychological distress. The current findings also have practical implications for clinicians and continued research. Metacognitions and EF of young adolescent clients can be assessed and targeted in therapy. Specifically, metacognitive therapy may equip at-risk adolescents with strategies to modify metacognitions or directly improve executive abilities required to disengage from the CAS, so that psychological distress and its associated negative clinical consequences (e.g., progression into psychopathology) are less likely, as previously found (Thorslund et al., 2020; Walczak et al., 2021). However, given the interaction between EF and negative metacognitions only increased the prediction of psychological distress by 6%, continued research is required regarding the influence of other moderators.

Limitations

This study has several limitations, firstly, metacognitions and psychological distress were assessed by self-report measures. Although the MCQ-A and K10 are well-validated, due to their retrospective nature and emphasis on negative internal states, they can be subject to error and bias (e.g., negativity bias; Sato & Kawahara, 2011; Shrout et al., 2018). This study also used a cross-sectional design, meaning that causation could not be confirmed. In addition, results of interaction effects should be interpreted with caution as the power analyses conducted did not account for this. Relevant demographic information about participants’ racial/ethnic identification and culture/geographic background were also not collected and, in the LABS, a proxy measure of socioeconomic status was used. Therefore, future studies should include other, more objective measures of socioeconomic status. Furthermore, participants were recruited from a high socio-economic region of Australia, limiting the generalisability of the results. Next, although there is evidence that stage of puberty can influence adolescent psychological distress (Van Rijn et al., 2023), this was not measured in the current study. However, the constrained age range in this cohort reduces the likelihood of difference due to puberty. A puberty measure has since been added to the LABS research program which can be included in future longitudinal follow-up studies. Lastly, confounding variables that also influence adolescent psychological distress, such as socioeconomic status, personality traits, adaptive metacognition, and environmental factors (e.g., parental or peer conflict or participant hunger), were not all controlled for (Beer & Moneta, 2012; Benzi et al., 2019; Reiss et al., 2019). However, there are numerous study strengths. Specifically, this study utilised an objective EF test that controlled for baseline processing speed. Furthermore, the findings extended and provided further support for the associations between the S-REF model, EF, and psychological distress in a relatively large and appropriately powered community sample of Australian adolescents.

Conclusion and Future Directions

Our findings suggest that positive and negative metacognitions increase adolescent psychological distress. Further, adolescents with poorer EF may be especially susceptible to the deleterious impact of negative metacognitions on psychological distress, consistent with the S-REF model (Wells & Matthew, 1996). Future research could investigate the longitudinal associations between metacognitions and psychological distress throughout adolescence.