Introduction

Executive function (EF) helps navigate most daily activities and constitutes a set of cognitive processes allowing self-regulation/self-directed behaviour toward a goal. ER impairments are associated with most forms of psychopathology. Poor EF predicts rumination, and poor use of emotion regulation – risk factors for psychopathologies.

There’s a lot of parallel play between clinical and cognitive approaches to EF, potentially leading to failures to apply theoretical and methodological advances in one field to the other, hindering progress.

Three main goals:

1. Review the current state of knowledge of EF impairments associated with psychopathology and limitations to previous research in light of recent advances in understanding/measuring EF. EF impairments seem to be transdiagnostically related to psychopathology, but limitations of prior research make existing evidence hard to interpret. So specific nature and patterns of impairments are unclear.
2. Offer concrete suggestions for improving assessment of EF, based on conceptual and methodological issues in current research, to advance clinical science. Advocate for better EF assessment. Obtain purer measures, select and analyze tasks minimizing noisiness of EF data.
3. Suggest future directions in EF research and clinical psychological science. Including integrating modern models of EF with hierarchical models of dimensional psychopathology as well as translational implications of EF-informed research on clinical science.

EF Impairments Associated with Psychopathology: Current State of Knowledge

EF is best described consisting of separable but related cognitive processes including unique and shared individual differences, genetic influences, and neural substrates. Aspects of EF that have been heavily studied in clinical psychology include shifting, inhibition, updating, working memory manipulation, verbal fluency, and planning – many of these can be further subdivided.

Previous research on EF impairments associated with psychopathology reviewed in this section has mainly used cross-sectional designs in adult samples, and assessed EF with traditional neuropsychological tasks. But there are limitations in the literature – imposing constraints on the state of knowledge and what can be determined through meta-analysis:

1. Many neuropsychological EF measures tap multiple aspects of EF as well as non-EF abilities. These tasks help screen for severe deficits, but are too broad to answer questions about specific aspects of EF and potential underlying mechanisms.
2. Because they were developed to detect more severe deficits, many traditional neuropsychological tasks may lack sensitivity to detect subtler deficits.
3. These limitations carryover into meta-analyses. In many meta-analyses on EF, tasks are grouped into the processes they are commonly considered to tap. But these categories may be lumping together tasks that may be tapping different and/or multiple processes.

Despite these limitations, meta-analytic evidence indicates that EF deficits are pervasive across disorders and EF tasks.

Impairments on More Specific EF Components: Inhibition, Shifting, Updating, and Working Memory

Definitions:

• Inhibition (I): switching between task sets or response rules.
• Shifting (S): suppressing/resisting an automatic response to make a less automatic but taskrelevant response.
• Updating (U): monitoring and coding incoming information for taskrelevance, and replacing no longer relevant information with newer, more relevant information.
• Working memory (WM): actively maintaining and manipulating information across a short delay.

Impairments across disorders:

• Schizophrenia: largest EF deficits found. Large effect sizes (ES) on measures of S, I, U, visuospatial WM, and verbal manipulation. Medium ES for simple verbal WM maintenance.
• Mood Disorders: same impairments as in schizophrenia, smaller magnitude of deficits.
• Major Depression (MDD): impaired with similar ES to schizophrenia.
• Bipolar Disorders (BD) larger impairments than MDD, but also uniformly impaired across EF domains. Medium ES for S, I, visuospatial WM, and verbal WM manipulation. Small significant ES for verbal WM maintenance.
• Obsessive Compulsive Disorder (OCD): impairments across most domains. Small significant ES for S, I, visuospatial WM, and verbal WM manipulation. Large ES for U. Simple WM unimpaired. Depression often cooccurs with OCD, but deficits in OCD aren’t driven by co-occurring depression.
• Posttraumatic Stress Disorder (PTSD): compared to traumaexposed people not developing PTSD, people with PTSD had worse performance on shifting (medium ES) and visuospatial WM (small ES). PTSD patients do experience inhibition deficits. Unlike OCD, with PTSD, cooccurring depression may account for deficits in PTSD patients.
• Anxiety Disorders: little EF research, mixed findings. Research in nonclinical samples suggest trait anxiety to be associated with impairments in specific aspects of EF, inhibiting competing responses. Little WM research as well. Lastly, evidence of poor EF contributing to attentional bias toward threat in anxious people, involved in anxiety maintenance.
• Attention Deficit Hyperactivity Disorder (ADHD): impairments in S, I, visuospatial WM, and verbal WM manipulation (smallmedium ES) in children and adults. Verbal WM maintenance less impaired. Updating not widely studied. EF also impaired in other externalizing disorders, like oppositional defiant disorder and conduct disorder, but these deficits could be partly accounted for by co-occurring ADHD.
• Substance Use: S, I, and WM impairments across most substance use disorders – generally medium ES. Reviews on EF impairments in substance use are difficult to interpret: 1) inclusion of polysubstance users makes effect of individual drugs difficult to isolate. 2) given the neurotoxic effects of alcohol and other drugs, it isn’t clear to what extent these deficits are a cause or consequence of substance use.

Complex Tasks: Verbal Fluency and Planning

Complex tasks may tap multiple aspects of EF – problematic if goal is to understand which specific processes are impaired. Nonetheless, these tasks still used in clinical studies of EF.

Deficits in verbal fluency are widespread across disorders. Meta-analyses show largest deficit for adults with schizophrenia and depression to be semantic verbal fluency. Semantic verbal fluency also impaired in people with BD, OCD, and ADHD, but inconsistent evidence for verbal fluency in PTSD patients. Little research on verbal fluency in anxiety disorder, mixed results.

Verbal fluency tasks impose multiple EF demands. One possibility of why semantic verbal fluency is more impaired in schizophrenic, BD, and depression patients is that it may place heavier demands on shifting. Another possibility is that semantic memory retrieval deficits could contribute to semantic verbal fluency impairment, especially in schizophrenia. Conversely, larger effect for phonemic verbal fluency in ADHD patients could be due to deficits in phonological processing, since ADHD and reading disabilities frequently co-occur. Deficits in verbal fluency could come from various sources –  difficult to interpret results from complex tasks.

Planning is less studied. Depression and BD patients have significant impairments in planning. Meta-analyses found mixed results for planning tasks in people with ADHD and OCD. There’s inconsistent evidence for planning deficits in PTSD. In theory, planning tasks tap multiple EF aspects, but standard measures of planning may be less sensitive than other tasks in detecting subtler deficits in some disorders.

Summary of Previous Findings

• Evidence shows deficits on various EF tasks to be associated with numerous psychopathologies. Most disorders are associated with fairly uniform deficits EF tasks, but there are variations in effects.  Results seem consistent with broad, transdiagnostic, impairment in EF. Exception is verbal WM maintenance – smaller deficits. Findings support that Wm deficits in these disorders are because of impairment in the central executive of working memory rather than content-specific maintenance systems – consistent with view that there are broad EF impairments associated with psychopathology, rather than in specific aspects of EF.

Limitations of Previous Research and Suggestions for Future Research

EF is hard to study, define, and measure. Now will be outlined the limitations of how EF has been defined, conceptualized, and measured in previous research, and concrete suggestions will be presented to address these limitations.      

Conceptual Issues: Models of EF

Many previous clinical studies of EF have treated it as either unitary, or a list of separate, specific abilities. Seeing it as unitary over-lumps diverse tasks into a single construct, seeing it as diverse over-splits, treating a list of tasks as if they were assessing separate abilities rather than a common set of component processes supporting completion of more complex tasks.

The best evidence indicates individual difference in EF to show unity and diversity. Different components of EF correlate with each other, tapping some common underlying ability (unity), while showing separability (diversity). General structure of common and specific elements is shared by different models of EF, focusing on different components and levels of analysis.

• Behavioural level of analysis – different models have focused on partly overlapping sets of EF components – e.g. Baddeley’s’ central executive system containing subsystems, others proposing a twofactor EF model.
• Neural level – models have proposed that distinct, interconnected prefrontal regions support functions like setting tasks goals, initiating responses etc.

Though these models differ, they have points of convergence, often agreeing on core cognitive and neural mechanisms involved in EF.

Unity/diversity model – captures several features of what’s believed to be the key components of EF, practical to use for understanding EF at the behavioural level, and has potential to shed light on commonalities and differences in impairments across populations by differentiating common and specific components of EF. Focuses on three aspects of EF: updating WM, shifting, and inhibition. This unity/diversity pattern has been consistently found over other samples. Each EF ability can be decomposed into what is common across all three (unity (common EF)), and what’s unique to each ability (diversity).

The unity/diversity model suggests decomposing task performance into common and specific abilities that could better map the underlying cognitive processes. New approach that has produced significant findings:

1. No unique variance left for inhibition after accounting for common EF – individual differences in EF fully account for individual differences in inhibition.
2. Common EF and shifting-specific components sometimes show opposing patterns of correlations with other measures, possible trade-offs between stability and flexibility. Specific deficits in stability or flexibility will only be apparent when performance on shifting tasks is decomposed into common EF and shifting-specific factors.
3. Different components of EF identified by this model differentially predict individual differences in clinically important behaviours. Evidence shows common EF as the primary source of this predictive power. Similarity of effect sizes on EF domains in disorders suggest that psychopathology more may be more broadly associated with impairment in common EF, suggesting that decomposing it may not have important implications for understanding EF deficits associated with psychopathology.

Methodological Issues

Multiple Measures

Biggest problem in measuring EF -> task-impurity problem. All tasks necessarily include systematic variance attributable to non-EF processes associated with that task context, making it difficult to cleanly measure the variance of interest. Since most clinical EF studies have used a single task to assess EF processes of interest, results are nearly always a mixture of non-, common-, and specific-EF component effects, making interpretation difficult.

This problem can be alleviated by using multiple measures of each component under investigation. If the chosen tasks share little systematic non-EF variance, one can see what’s common across tasks and use the resulting ‘purer’ variable as a measure of EF. E.g. measures of each EF component should be used then aggregated to measure common EF.

Simplest way to combine data from measures is to calculate a z-mean across tasks. Advantage – z-mean across tasks instead of individual tasks, variance in the scores not related to the constructs of interest no longer drive the effects. Disadvantage – merely combines scores, error variance is still there and can be a source of reduced power. So, if the sample is large enough, it’s preferred to use latent variable approaches for extracting the variance shared across tasks while removing error variance, - e.g. factor analysis, structural equation modeling.

Task Selection

It’s also important to carefully pick tasks. Many clinical studies now use traditional neuropsychological measures tapping multiple aspects of (non-)EF abilities. Useful for screening for severe deficits but too broad to answer questions about specific aspects of EF that may be implicated in psychopathology. Complex neuropsychological tests tap a variety of cognitive processes, making interpretation difficult. This can be addressed by using tasks designed to specifically place demands on individual aspects. Important to include specific tasks to identify what processes account for impairment on broad neuropsychological tasks.

Many studies also use questionnaires or self-report measures. These correlate poorly with task-based measures of EF, and shouldn’t be assumed to be measuring the same constructs. Questionnaire-based measures have ecological validity as they as about real-world situations. But they pose interpretational problems because of the multiple executive and non-executive functions involved in real world settings/contextual influence. Specific questions about EF are best addressed using targeted tasks.

Sensitivity and reliability of tasks are important. Tasks should be sensitive to the magnitude of deficits expected form the sample being tested. Tasks with low reliability have poor correlations with other measures. Reliability is sample specific. Unfortunately, complex EF tasks tend to have low internal/test-retest reliability due to different strategies used when completing them.

Problems with sensitivity and reliability are problematic because they could lead to false negatives resulting in not being published, or being published with the conclusion that EF isn’t impaired in the clinical group.

There are a number of commercially available task batteries including tasks assessing EF.

• Advantages – often they have more extensive psychometric evaluations and norms, their standardization allows for clear comparison.
• Disadvantages – they generally don’t provide comprehensive coverage of different components aligned with current models, and don’t provide multiple measures of each construct needed for latent variable approaches. Also less likely to yield new insights because they’ve been heavily used in most clinical populations.

Other Methodological Considerations

How the data is collected and analyzed is also important. When the total individual variance in EF task performance in broken into EF, task-specific and error components, the ‘noise’ of non-EF task-specific variance and error variance can be large, while the ‘signal’ of EF-specific variance may be small. So to detect the most important signal for the inquiry, it’s important to minimize error variance and maximize power. First, strong need to increase sample size to improve power (underpowered studies lead to file drawer problem and lack of replicability). Second, once data’s collected, reliability and validity of the measures depend on how they are screened and analyzed. Also important to screen for and address the presence of outliers.

Future Directions

Given the discussed limitations to previous research and the goal of understanding links between EF and psychopathology at a level of specificity that can support translational research, we propose two broad directions for future research.

Testing Models of Unity/Diversity Across Both EF and Psychopathology

First direction suggested is that the problem of understanding the undifferentiated nature of EF impairments across disorders may be made more tractable by testing models that include both unity and diversity in psychopathology and EF.

What gives rise to broad patterns of impairment in EF across disorders? These deficits can’t easily be explained by non-specific factors. In most cases effect sizes are similar across core EF domains. This pattern of broad impairment is consistent with the theory that people with multiple forms of psychopathology have impairments in the unitary component of EF, posited to be the ability to actively maintain task goals.

Latent variable models of psychopathology find that there’s a common factor spanning all aspects of common psychopathologies in addition to more specific aspects (the ‘p factor’). Transdiagnostic impairments in EF might be explained by a link between the p Factor and common EF. But, cognitive factors that appear as transdiagnostic at one level of analysis may not be when more detailed measures at multiple levels of analysis are considered.

Causal Models

Second direction suggested is that research needs to move beyond cross-sectional case-control designs to test different possible causal links between EF and psychopathology.

A general shortcoming of the broad field of cognitive risks in psychopathology across the lifespan is the frequent lack of consideration of possible models of how cognitive impairments and psychopathology may be causally related. It’s unknown if EF deficits (a) precede and are a potential causal risk factor for developing psychopathology, (b) follow, and are a consequence of psychopathology, or (c) are a correlate of psychopathology without playing a causal role.

Many studies assume a particular causal model, but there have been fewer attempts to try to rule out/in particular models based on evidence. Cross-sectional case-control studies aren’t able to differentiate between these possible models.

• In sum, the causal links between EF and psychopathology haven’t been well established, and the mechanisms connecting them are unknown and in need of theoretical and empirical investigation. These questions have important implications for prevention and treatment.

Treatment Implications

Current evidence suggests that approaches aimed at teaching compensatory strategies may be the most promising direction for future translational research. There’s little evidence supporting direct training of EF (targeting the weakness instead of compensatory strategies), little evidence that it generalizes to real world functions or improves clinical symptoms. There’s still the possibility that types of training that better target areas of weakness might provide better transfer.

Treatment and prevention programs involving compensatory strategies may therefore be a more promising direction for translational research e.g. goal management techniques, - shown to improve functional outcomes in schizophrenic individuals.

In addition to this there may be a need to adapt and personalize current treatment approaches to match client’s EF abilities – tailoring treatment approaches through better understanding of their profile.

Executive function deficits also have important implications for psychopharmacological treatments. For behavioural therapies, pre-treatment EF has been shown to predict drug treatment response.  Better understanding of EF deficits can enhance targeting of medications that affect the neurotransmitter systems known to be involved in those EF processes.

Conclusion

Cognitive and clinical psychology have followed largely independent paths. It’s argued that it’s necessary to move past the ‘parallel play’ of these fields to push clinical psychological science toward a better understanding of how/why EF is broadly compromised across disorders.

It’s recommended to apply validated models of EF to clinical research using multiple tasks to obtain purer measures, and also select/analyze tasks that minimize noisiness of EF data.

To address the task impurity problem/improve reliability, we recommend carefully choosing EF components to focus on and using multiple measures of each component of interest and combining them to composite scores or latent variable analysis.

We advise using more specific EF measures rather than traditional, broad neuropsychological tests. It’s the hope that combining current theoretical and methodological advances of clinical and cognitive science can advance the field towards understanding underlying processes involved in EF impairments at a level that enables translational research to improve treatment.