Executive function (EF) processes help us respond to the environment and regulate our thoughts and behaviours towards our goals. Meta-analytic evidence shows that EF deficits are pervasive across psychopathologies and EF tasks, proposing that they may be transdiagnostic risk factors for psychopathology. Four key limitations in the majority of prior research hinder progress in testing this hypothesis.

1. Most studies use standard neuropsychological test approaches that confound different aspects of EF, leading to seemingly undifferentiated nature of impairments across disorders.
2. Vast majority of research has investigated individual disorders without taking the pattern across disorders into account – ignoring comorbidity between disorders.
3. Most research has focused on adults, missing the key adolescent/emerging adult phase for EF development and psychopathology risk.
4. Few studies have examined how (mediation) and for whom (moderation) EF and psychopathology are related.

This study aims to address these limitations by linkages between latent dimensions of EF and latent psychopathology dimensions in a sample of adolescents and emerging adults.

Unity/Diversity Model of EF

EF is best characterized as separable but related cognitive processes, with unique and shared differences, genetic influences, and neural substrates. The unity/diversity model focuses on three aspects of EF.

1. Shifting, defined as switching between task sets or response rules (e.g. classification by shape or colour).
2. Inhibition, defined as suppressing or resisting automatic responses in order to make less automatic but task-relevant responses (e.g. Stroop task).
3. Updating, defined as monitoring for and adding task relevant information in working memory, and deleting irrelevant information.

These abilities correlate, suggesting there’s a common EF ability involved in all three aspects. Common EF is posited to be the ability to monitor for and maintain goal and context information and use that information to bias ongoing processing.

Each ability can be decomposed into what’s common across all three EF’s (unity) and what is unique to that particular ability (diversity). Two large, independent youth and adult samples support a model with common EF factor and updating – and shifting- specific factors’ common EF accounts for individual differences in inhibition. The different components of EF identified in this model differentially predict individual differences in clinically relevant behaviours, with recent evidence finding that the common EF is the primary predictor, relating to behavioural disinhibition, attentional problems, and transdiagnostic psychopathology. Meta-analytic evidence shows generally similar effect sizes across core EF domains, consistent with the theory that individuals with multiple forms of psychopathologies have impairments in the unitary component of EF.

Bifactor Models of Psychopathology and Links to EF

Psychopathology has also been shown to consist of both common and specific factors. As opposed to historical conceptualizations of disorders as categorical conditions they’re now conceptualized as continuous symptom dimensions.

There’s a long history of modeling internalizing and externalizing dimensions of psychopathology liability, recently expanded to include a common factor (p factor). P factor model has been replicated in multiple samples. Critical question to ask of such models is whether they’re related to theoretical and practically important risk factors and outcomes and can help advance clinical science (are they useful?). Significant evidence has validated p factor in relation to a wide variety of psychopathology risks and outcomes in developmental continuity.

These models thus hold promise for clarifying the many patterns between risk factors and categorically defined disorders. Recent conceptual models have proposed that executive dysfunction is a risk factor for common psychopathology. Several studies have found the p factor to be associated with poorer performance on EF tasks, including working memory and a single EF composite in children, working memory, flexibility, response inhibition, and updating tasks in adolescents, and working memory and shifting tasks in adults. P factor has also been linked to structure and function of prefrontal areas involved in EF in youth.

To note, these studies used manifest EF variables or single, unitary EF factors, and so didn’t directly test the hypothesis that common psychopathology liability is linked to poorer common EF. Only one study so far has tested links between the bifactor model of psychopathology and the unity/diversity model of EF. This study found that the common EF factor assessed at age 17 was predicted by p factor assessed across childhood and adolescence only for male participants and based on teacher, not parent, ratings. When modeled separately, internalizing related to poorer common EF across genders and raters, and externalizing related to better shifting-specific EF. So associations between the p factor and poor EF are consistently found across studies, but results have somewhat varied.

Mechanisms Linking EF and Psychopathology

There’s a lack of research investigating why EF is associated with psychopathology dimensions. Hankin et al (2016) posited that stress could mediate risk between EF and expression of common and internalizing psychopathology. Dependent stressors are negative life events partly influenced by a person’s behaviour (e.g. failing an exam). The stress generation model maintains that individual difference vulnerabilities related to psychopathology impair functioning and increase risk for dependent stressful life events. Poor EF has been proposed as a factor that can contribute to stress generation. Poor EF may contribute to functional impairments resulting in dependent stressors (e.g. failing an exam because of failure to plan). It was found that the link between EF task performance and internalizing symptoms increase with age due to increased association between EF and dependent stressors. Possibly because adult caretakers compensate for younger adolescents’ poor EF (e.g. reminders to complete homework), preventing poor EF from being translated into behaviours that lead to stressful life events.

Dependent stressful life events strongly predict rumination, a pattern of repetitive thought in response to an emotional state. Rumination is proposed to act as a transdiagnostic risk factor by amplifying current mood states and impairing problem solving. Predicts that rumination is associated with the p factor, but hasn’t been directly tested.

Mediation by stress generation and rumination could potentially explain why EF impairments are broadly associated with psychopathology. Study found that dependent stressful life events predicted internalizing symptoms both directly and via increased rumination – both factors transdiagnostic risks, stress predicts p factor.

Current Study

Recent research using latent dimensional models of psychopathology have begun to disentangle sources of the broad EF impairments observed across disorders. Research suggests that associations may be mainly driven by shared psychopathology liability (p factor). Most research on EF relating to p factor models hasn’t employed models that can differentiate between EF components. Lastly, mediating mechanisms of poor EF conferring risk for psychopathology remain speculative.

The current study tests associations between the unity/diversity model of EF and the latent dimensions of psychopathology liability in a community sample of youth during key adolescent to emergent adult period of enhanced psychopathology risk and continuing EF development.

This study is cross-sectional, but allows preliminary tests of potential mediating mechanisms which future longitudinal studies can investigate. Study aims to clarify potential risk pathways between EF impairments and forms of psychopathology liability, and accelerate progress in understanding how these impairments may contribute to co-occurrence across psychopathologies. This study extends previous research demonstrating that poor EF predicted anxiety and depression symptoms via stress generation and rumination, and that these effects were stronger in older youth.

They hypothesized that seemingly broad deficits on EF tasks associated with forms of psychopathology are best explained by poorer common EF associated with common psychopathology liability. Because of mixed and limited prior research, there’s no a priori hypotheses regarding externalizing or internalizing- specific liability factors and conduct exploratory analyses. If, as they predicted, the common EF was associated with the p factor, they hypothesized that this association would be partially mediated by the indirect path through dependent stressful life events and rumination, and that the EF-stress path would be stronger for older youth.

Method

Participants

292 participants aged 13-22. Selected to maximize racial and economic diversity.

Procedure

Youth participated in hour lab visits, with breaks. Participants gave written informed consent (18-22) or parental consent (13-17). Youth completed three EF tasks assessing updating, shifting, and inhibition. Participants asked to complete self-report questionnaires online before their visit. All studies approved by IRB.

Measures

EF Tasks

EF tasks from Friedman et al. (2016) except for stop signal task. The Friedman et al. tasks have been found to have good internal and test-retest reliability, and convergent validity including significant factor loadings on common EF and, for updating and shifting tasks, the updating- and shifting-specific factors. Stop signal task validated in reference to mathematical models and neural indices of response inhibition.

Updating: for all updating tasks, the performance measure is the proportion of correct responses across all trials. Tasks: Keep track, Letter memory, Spatial 2-back twelve

Shifting: for all tasks, participants first practiced a block of each sub-task separately, followed by mixed-task blocks. Performance measure for shifting tasks is the switch cost: difference in mean response time between correct tasks switch trials and task repeat trials in mixed blocks. Participants instructed to respond as quickly as possible without making mistakes, indicated by an error beep. Tasks: Number-Letter, Colour-shape, Category switch

Inhibition: Tasks: Antisaccade, Stop signal, Stroop

Questionnaires

• Children’s Depression Inventory – CDI
• Penn State Worry Questionnaire for Children – PSWQC
• Manifest Anxiety Scale for Children – MASC
• Child Behaviour Checklist Youth SelfReport – YSR
• Multisite Multimodal Treatment Study of Children with ADHD – MTA
• Strengths and Difficulties Questionnaire – SDQ
• Adolescent Life Events Questionnaire Revised – ALEQR
• Children’s Response Styles Questionnaire – CRSQ

Discussion

This study aimed to understand possible risk pathways between poorer EF and internalizing and common psychopathology (p factor) liability. For older youth, poorer common EF, was associated with higher internalizing liability in a one-factor internalizing model, and higher common psychopathology (p factor) in the bifactor model. Poorer common EF was also associated with more dependent stressful life events in older youth, which in turn associated with higher rumination for all youth. For all youth, rumination and dependent stressful life events were associated with higher internalizing liability in one-factor internalizing model, and with higher levels of the p factor and internalizing-specific factor in bifactor model. Overall results suggest that the many-to-many relations between different forms of psychopathology and measures of different aspects of EF may be more parsimoniously explained by a link between common EF and both common psychopathology liability (p factor) and internalizing-specific liability in older youth – suggesting that stress and rumination may serve as mediators of these relations.

Specificity to EF Dimensions

Links with psychopathology liability were specific to the common EF factor, rather than updating- or shifting-specific EF. Consistent with a previous study testing associations between the p factor and the unity/diversity model and with most previous studies with more specific symptom dimensions as well as broad impairments. But one study found that depression symptoms were associated with lower common EF cross-sectionally, but prospectively predicted lower updating-specific EF, suggesting that associations may differ in some cases.

Why may common EF factor be linked to psychopathology liability? It’s thought to capture the ability to actively maintain and manage goals and use them to control ongoing processing, a demand shared by all EF tasks. On the other hand, updating- and shifting-specific demands are likely to occur only intermittently in daily life. So poor common EF may be likely to impair functioning, including in ways that increase stress. This relation is likely transactional, with stress leading to EF impairments.

Age Moderation

Common EF in this study significantly interacted with age, such that poorer common EF only associated with dependent stressful life events in older youth. This may be due to the role adults play in compensating for younger adolescents’ poor EF, buffering against stress generation.

Though age moderation is predicted based on past findings, other studies found associations between EF and the p factor in younger individuals. However, they also used different methods of assessing psychopathology which could account for the differences between those studies and the current one.

Stress and Rumination Effects

Primary focus of this study was to better understand EF-psychopathology liability links, but also provided new insights into how stressful life events and rumination relate to the bifactor model of psychopathology liability.

1. Dependent stressful life events were associated with all three psychopathology liability dimensions. Association with p factor is consistent with the model of stress as a transdiagnostic risk factor, but associations with internalizing- and externalizing-specific factors also suggest that stress further confers specific risk for internalizing and externalizing dimensions. Dependent stressful life events are a strong risk factor for depression and anxiety, so may be more strongly associated with the internalizing-specific factor than other types of stress.
2. Rumination hasn’t been previously examined in relation to the bifactor model. This study found rumination to be strongly associated with the p factor. It was also associated with higher internalizing-specific liability, but much weaker association – supports the view that rumination is a broad transdiagnostic risk factor rather than specific to depression or internalizing psychopathology. However, this additional association suggests it may confer risk for anxiety and depression via both common and specific mechanisms.
3. Youth who engage in higher rumination had lower externalizing-specific liability, after variance shared with internalizing was accounted for.

Limitations and Future Directions

There were several limitations in this study:

• Study was crosssectional, precluding conclusions about the temporal ordering of variables in the models.
• Tested a hypothesized process model based on previous research and theory, identifying possible mediating mechanisms as a promising start for future longitudinal studies.
• More longitudinal research is needed with EF and psychopathology assessed at multiple time points to see if EF is a risk or consequence of common psychopathology, or both (transactional relations).
• This study focused on psychopathology and EF in adolescence and emerging adulthood in a nonselected community sample. Future research is needed to determine if the model generalizes to other age groups, and to high-risk or clinical populations. More research is needed to test whether EF-stress generation links are specific to the late adolescence/emerging adulthood developmental period.

Translational Implications

More research is needed in order to translate these findings to practice. But better understanding the mechanisms by which EF may serve as a transdiagnostic risk factor for psychopathology has the potential to inform new targets for intervention. There’s a lot of interest in EF training as a potential prevention/treatment strategy, but until now there’s little evidence that training transfer to real-world functioning. Interventions aimed at disrupting the link between poor EF and stress generation could be a promising approach, rather than attempting to train EF.  This study found that the link between EF and dependent stressful life events was specific to common EF, suggesting that training in compensatory strategies for goal-management could potentially mitigate effects of poor common EF to reduce stress and potentially reduce psychopathology risk.

Conclusions

Broad patterns of poorer performance on EF tasks associated with forms of psychopathology symptoms may be best explained by associations between common EF and common psychopathology liability, though there are also internalizing-specific associations. Poorer common EF associated with internalizing and the pf factor via dependent stressful life vents and rumination in older youth. This developmental period of increased demands for independence may be a critical window for risk associated with poor EF. Interventions aimed at disrupting the link between poor goal-management and stress have potential for reducing dimensional psychopathology, particularly in emerging adults.