Event-related potential measures of executive functioning from preschool to adolescence

Executive functions are a collection of cognitive processes that help us to regulate our thoughts and behaviours to make plans, solve problems, and attain goals.1, 2 These skills are important throughout the lifespan, contributing to school readiness and academic achievement3 and to later career success. Major subcomponents of executive functioning have been described as attention, inhibition, self-regulation, working memory, cognitive flexibility, planning, organization, problem-solving, and performance-monitoring.4 Basic executive functions, including the inhibition of an inappropriate motor response, emerge early in life and subsequently lay down the foundations for later development of higher-order executive functions, including planning and problem-solving. Whether the subcomponents of executive functioning are already differentiated in the first few years of life or emerge from a more undifferentiated system with development is still debated. Although it is widely agreed that, from around 7 years of age, the overarching structure of executive functions is relatively stable,5, 6 the structure of executive functions may be more unitary and less differentiated earlier in life.7, 8

Executive functions are compromised in different ways across a range of developmental disorders and in acquired brain injury,9, 10 and are susceptible to disease and poorer environments.11, 12 While the prolonged period of development makes executive functions particularly vulnerable, their higher malleability may also provide a window of opportunity to improve executive functions through interventions.13

Neuroimaging techniques have shed some light onto the development of the neural systems underlying executive functions.14, 15 A parallel has been drawn between the gradual integration of executive functions and the prolonged development of the prefrontal cortex,16-18 but it is also clear that executive functions depend upon a wider neural system.19 In addition to magnetic resonance imaging (MRI) studies, event-related potential (ERP) measures have contributed to our increasing comprehension of the developing neural substrates underlying this cognitive domain.20-25 This method has several advantages over MRI, including being relatively easy, practical, and cost-effective to use with younger children, and providing more precise information about the timing of brain events underlying behavioural performance.26, 27 Its high temporal resolution affords a closer look into various processing stages that lead to a single behavioural response, and provides another source of information in the investigation of the developmental differentiation of executive functions.

Here, we provide an overview of ERP studies relevant to executive function development from preschool to adolescence. ERP research from this period of development has not previously been reviewed, despite the growing number of studies and the substantial changes observed in ERP components related to executive functions during this developmental stage. This review aims to provide a summary of developmental changes observed in key ERP components throughout this period, collating studies that look at different domains of executive functioning, and providing a useful reference and overview for researchers and clinicians new to the area of ERP research in developmental populations, as well as an overview of the field for those currently engaged in work in this field. We will focus on four of the most extensively studied areas of executive functioning in neurophysiological research: inhibitory control, working memory, information processing, and performance monitoring and their associated ERP components (as listed in Table SI, online supporting information). As previously described, executive functions are often compromised by acquired brain injury and in various developmental disorders. Thus, following our overview of typical development, we will discuss one application of ERP methods in a neurodevelopmental disorder that has been the most widely investigated disorder using these methods: attention-deficit–hyperactivity disorder (ADHD).28

The N200: inhibitory control and inference suppression

The N200 component of ERP is believed to reflect the cognitive control necessary for successful inhibitory control and interference suppression.29 Inhibitory control is the ability to control a dominant, pre-potent motor or cognitive response, but it also involves processes such as interference suppression, emotional control, and directed forgetting, where a participant is explicitly told to remember and forget specific stimuli.30 The ability to control interference from irrelevant stimuli and to inhibit a pre-potent response to selectively attend to task-related events is important in the development of behavioural–emotional control and for academic attainment.31 Inhibition grows increasingly pertinent in the transition from early childhood into adolescence as young people gain more independence.

The N200 is a negative wave produced after successful inhibition with a peak latency of approximately 200 to 300 milliseconds after stimulus onset. Its neural generators include the frontal and superior temporal cortex, and the anterior cingulate cortex.32 The N200 can sometimes be referred to as the N2a, the N2b, or the N2c, depending on the particular paradigm used and thus the brain areas that are recruited. The degree to which specific brain areas are recruited can vary according to factors such as the demand for other executive skills, such as working memory, in a specific paradigm, the response modality used, and the history of previous responses.32

A larger peak in overt response inhibition tasks supports the association between the N200 and inhibitory control. For example, in the Go/No-go paradigm where the participant responds to a ‘Go’ stimulus but ignores the ‘No-go’ stimulus, a larger peak is seen when ‘No-go’ stimuli have some similar dimensions as the ‘Go’ stimuli, or when there is increased pressure to respond faster.33, 34 The N200 can also be observed in other paradigms, the most common of which are the Go/No-go task, the Stop-signal task, the Stroop task, and the Flanker task (see Table SII, online supporting information, for a detailed description of these tasks).

The N200 response may also vary according to the type of inhibitory control required, with some evidence suggesting a dissociation of interference suppression and response inhibition.35 In a combined Go/No-go–Flanker task with 14 young adults, the incongruous flanker, which requires the suppression of distracting information, elicited a more central topography and a more delayed N200 peak than the No-go condition, which requires inhibition of a prepotent response.36 This later peak is seen in tasks with distractors that need to be suppressed for successful task completion and is often referred to as the N2pc. However, some recent findings in three studies of children and adults (n=10–37) do not support the idea that the N2pc exclusively reflects distractor-suppression processes. It has been proposed that the N2pc may instead reflect a combination of attention selection and distractor suppression.37, 38 No firm conclusion about the precise relationship between N2pc and behaviour can be drawn, owing to the limited sample sizes and varying age ranges across the available literature.

There has been some debate in the literature about the interpretation of the N200. Some studies have suggested that it may reflect the monitoring of conflict instead of response inhibition.29, 39 The N200 has also been compared with the error-related negativity (ERN), which is elicited in trials where commissions are made. Some research suggests that the ERN may reflect error detection or inhibition. The ERN and the N200 were shown to have different scalp topographies in a Go/No-go task, which implies that different mechanisms and generators subsume these two components.40 One study showed that distinct cortical areas were associated with response inhibition, commission errors, and behavioural correction using electroencephalography (EEG) and functional MRI during a Go/No-go task.41 Error detection was correlated with activation of the anterior cingulate and pre-supplementary motor area, whereas behavioural correction was related to the anterior cingulate as well as the left prefrontal cortex.

The P300: information processing

The P300 is a positive waveform that appears at approximately 300 milliseconds in auditory ERP paradigms that involve attending to a target as well as discriminating between a target and a non-target. The P300 is most commonly referred to in the context of attention, working memory, and problem-solving.47 Although there remains debate surrounding the precise cognitive function that the P300 is most closely associated with, there seems to be a general consensus on its description as a neurophysiological index of information processing and updating in working memory.48 The latency and amplitude of the P300 have both been connected to behavioural success on executive tasks, including attention and memory, in healthy adult and patient populations, but this association has not been as widely researched in children.

The P300 is typically further subdivided into the P3a component and the P3b component. The P3a or ‘novelty P300’ activates in passive oddball tasks as a reaction to novel stimuli that do not call for an active response from the participant, whereas the P3b is engaged in active oddball tasks that involve intentional conscious discrimination as the participant responds to the novel stimulus, often by button press.49 The P3a is observed when a task requires orienting or novelty detection and has a frontocentral topography. The P3a is likely to originate in the frontal cortex and the hippocampus.50 The P3b is typically observed during active tasks that engage attention and working memory and shows a more parietal topography with sources in the temporal and parietal lobes, and cingulate cortex.32 Polich51 has proposed that the P300 is a result of the P3a, which responds to early attention-related processes and further drives the P3b, produced when enhanced attention drives the stimulus signal to temporal and parietal regions. In contrast, the No-go P3 is thought to reflect inhibitory control as it is observed in response to distractor items40, 52 and shows a different topography to P3a and P3b with maximum peaks in centro-parietal channels.

Contralateral delay activity: visuospatial working memory

The contralateral delay activity (CDA) is a lateralized ERP over the parietal cortex that reflects the amount of target and distractor stimuli that are encoded or maintained from one hemi-field during the memory display. The CDA increases in amplitude with the number of target and/or distractor items maintained in working memory and is correlated with working memory capacity.61 Working memory is the ability to temporarily mentally store and manipulate information. Classically, working memory has been divided into ‘slave systems’, which are separate for visuospatial and phonological information, and a supervisory system called the central executive. The capacity, or number of units of information that can be kept in working memory, is important in the development of academic skills and for general learning.62

Other attention and working-memory-related activities

ERPs have also been used to investigate other preparatory and inhibitory processes during cued attention and working memory tasks. In tasks where children are required to look towards a cued location and to ignore a distractor location, a series of ERP responses are observed. Early directing attention negativity, possibly reflecting early parietal activation within the frontoparietal network, precedes frontal activity reflected by the anterior directing-attention negativity. After these responses, a late widespread contralateral positivity is observed, which is thought to represent the oculomotor programming of the planned eye movement as well as the orienting of attention.68 Studies show that these early attentional responses are related to working memory abilities. Shimi et al.69 report that age-related differences in attention orienting processes before and after encoding stimuli in visual working memory (VWM) can explain differences in VWM performance between the developing brain and the adult brain. Differences on an individual level in the attention orienting processes before encoding can be biased so that relevant items are more efficiently encoded into VWM in children with high VWM capacity.69 A further study showed that children with large cueing benefits in VWM capacity elicited adult-like responses after selection of the stimuli, whereas children with low capacity did not elicit a contralateral negativity.38

Error-related negativity: response monitoring

The ability to monitor responses and adjust behavioural output according to set goals is another important executive function domain. Responses on tasks used to index monitoring (Go/No-go paradigm, Eriksen Flanker task, and the Simon task) are marked by specific ERP components following error. The ERN is a negative deflection between 80 milliseconds and 150 milliseconds with maximal amplitudes over frontocentral channels70 thought to be produced in the anterior cingulate cortex.71 The ERN response does not depend on the conscious awareness of the participant that an error was made.72 There is also a related response in correct trials with a similar time course and topography but with lower amplitudes, called the correct-related negativity. The ERN appears before a positive deflection (error-related positivity, Pe) with a maximum response over centro-parietal channels with a peak between 200 milliseconds and 500 milliseconds.73 In contrast to ERN, the Pe depends on conscious error awareness and is not present in all error-trials.74 On a behavioural level, increases in response accuracy, reaction time, and a reduction in response variability have been found using the Go/No-go paradigm75 and Eriksen Flanker task.76 These improvements have been found throughout childhood and adolescence77, 78 until adult-level performance is reached.75

Feedback-related negativity: feedback monitoring

In addition to being able to detect errors in self-generated responses, children must also be able to respond to external feedback to reach optimal performance. Feedback monitoring is mostly elicited in tasks with either probabilistic or random outcome. In probabilistic learning tasks, participants learn to associate stimuli with certain risks for gains or losses. Other tasks look at the effects of positive or negative feedback presented randomly. Feedback typically elicits a negative deflection with a maximum over medio-frontal electrodes with a peak around 270 milliseconds after feedback onset.87 This component is described as the feedback-related negativity (FRN). The amplitude of this component is consistently larger in response to negative feedback than to positive feedback.87 The dorsal anterior cingulate cortex is the proposed source of the FRN.88 Genetic studies suggest that variation in the FRN may be linked to individual-level differences in the dopamine and serotonin systems, which have been previously associated with reward processing and decision-making.89

ERP markers of atypical executive function development

A large part of the available ERP literature that investigates differences in executive function development in childhood disorders focuses on children with ADHD.95 ADHD is characterized by deficits in attention, hyperactivity, and impulsivity.28 However, considerable heterogeneity in symptoms and a higher prevalence of co-occurring disorders pose significant challenges to the diagnosis, treatment and investigation of ADHD.96 Neurophysiological methods are one avenue to identify endophenotypes that could serve as biomarkers and help to distinguish between ADHD subtypes. Indeed, differences in N200 amplitude and latency have been described as a potential endophenotype for ADHD.97 The following section will review the literature on ERP investigations of ADHD spanning all previously discussed aspects of executive function.

ERP markers of attention and inhibition in adhd

Differences in ERP markers of attention and inhibition have been identified in children with ADHD. For instance, attenuated P300 amplitude and decreased latency in Go/No-go tasks in children with ADHD are thought to reflect early signs of atypical attention development.98, 99 Further, reduced P200 and N200 effects during Go/No-go and Stop Signal tasks have been attributed to poorer recruitment of neural resources.95, 100 One study reported a more anterior P300 for children with ADHD, which may indicate a greater requirement for frontal inhibitory processes.101 Differences are not consistently found across studies with either increased, reduced, or absent amplitude and latency effects in different studies.102 The mixed results may be due to the varying age groups used, differences in task design or analysis techniques, and the heterogeneity within ADHD groups.103 Nonetheless, ERP components can be used as sensitive markers of executive function as evidenced by a recent randomized control trial that looked at the impact of treatment in 112 school-age children with ADHD and found increases in the P300 amplitude after intervention alongside improvements in response inhibition.100

ERP markers of working memory in adhd

The CDA has been used to study working memory in adolescents and adults with ADHD. One study administered a change detection task both to adolescents aged 12 to 16 years, and to adults with and without ADHD and found that performance deteriorated more for the adolescents with (n=15) and without (n=19) ADHD than either adult group in the presence of distracters and when there was a higher working memory load.104 The CDA showed that adults were able to more efficiently remove distracting details from memory later in the retention interval, resulting in better working memory. ADHD diagnosis was related to smaller CDA amplitude in adolescents and adults with ADHD than in the comparison group when maintaining a low load, which could reflect an inability to maintain focused attention to cued stimuli when there are low task demands. Thus, overall, the ERP results discussed here suggest no differences in filtering efficiency and visuospatial working memory storage capacity in adolescents and adults with ADHD.

ERP markers of reward processing and response monitoring in adhd

ERP markers have been used to investigate potential differences in reward processing and response monitoring in ADHD. One study reported smaller ERN amplitudes in 68 children aged 8 to 15 years with ADHD as well as intermediate amplitudes in unaffected siblings than in a matched comparison group.105 Similarly, Pe amplitude was found to be reduced in 7- to 11-year-old (n=16) children with ADHD but not for adults with ADHD.106 Only children with ADHD who had additional learning difficulties showed reduced Pe amplitude in a larger-scale study.107

Van Meel et al.108 found no significant differences in feedback processing when investigating the FRN in 8- to 12-year-old (n=21) children with ADHD, but observed reduced amplitudes in later time windows. Similarly, another study found that while FRN amplitude decreased after the first reward in 8- to 12-year-old typically developing children, it increased in children with ADHD (n=14),109 which may indicate differences in motivation. In summary, studies indicate differences in ERP markers of reward processing and response monitoring in children with ADHD, but the specificity of this association will need to be further elucidated in future research.

Conclusion

ERP paradigms provide us with a direct means of analysing the brain basis of typically and atypically developing executive skills in children and adolescents. They also offer valuable insights that cannot be gleaned from behavioural research alone. ERPs can inform cognitive interpretations by indexing constituent processes that contribute to behavioural performance on a particular task. For example, a study of 8- to 10-year-olds with a history of concussion on a Go/No-go task found that children who had experienced recent concussion (n=15) made more commission errors behaviourally than those who did not (n=15).110 These behavioural differences were accompanied by longer N200 latencies and more diminished P300 amplitudes on a neural level. Similarly, an ERP study of adolescents with unilateral and bilateral frontal stroke (n=11) due to sickle cell disease on a fast-response task found that these patients showed evidence of a diminished ERN response compared with patients with sickle cell disease only (n=11) and comparison siblings (n=11), despite no differences on a behavioural level. However, the N200 and P300 were not impacted by the lesions, which suggests that although these executive processes were still relatively intact, performance monitoring was not.111 These studies demonstrate how ERPs can help in the assessment of acquired brain injury and other disorders by contributing to the development of executive profiles that highlight specific strengths and weaknesses, bringing us closer to an ‘executive fingerprint’.112 As described by Ozonoff and Jensen112 in their report almost two decades ago, a better understanding of an executive dysfunction can lead to a more successful diagnosis and intervention.

Developmental studies show changes in all of the discussed ERP components with increasing age from preschool to adolescence. These changes are likely to reflect the structural and functional maturation of the neural substrates underlying executive skills and help inform theories of executive development.7, 8 The prolonged developmental changes in the frontal lobe and its related systems mean that the timing of brain injury onset can have differential effects on the executive system, depending on its developmental stage, with earlier insult often resulting in wider-reaching dysfunction across executive domains.113 It can be more difficult to assess the impact of frontal brain injury early in development on later emerging executive skills. Promising new research suggests that neurophysiological indices of executive functions can be identified before they are behaviourally assessable and may even be predictive of future executive performance.114 For example, Brydges et al.114 recently showed that the N200 difference waveform and the P3b amplitude in a group of 7- to 9-year-olds were predictive of a unitary executive factor, showing observable indices of executive functioning before the specific associated behaviours could be distinguished from one another using psychometric assessment. We are yet to fully understand the interpretation of the individual neural correlates that underlie specific executive functions and to grasp how these relate to one another in the context of the developing brain. The potential contribution of factors such as sex, environment, disease, and hormones require further investigation to better understand the significance of sometimes subtle differences in ERP responses.

ERP methods are being used more frequently to assess the efficacy of interventions designed to improve cognition and behaviour.111, 115, 116 For instance, one intervention study found specific changes in the N200 response, which implied that emotional regulation training successfully worked by increasing inhibition rather than decreasing emotional arousal.117 However, ERP techniques also have their own unique design and interpretation issues. For instance, EEG data are often ‘noisier’ in younger populations because of differences in compliance. This problem is aggravated by arbitrary age groupings, variation in task implementation, and small sample sizes in the available literature. There is currently no general consensus on best practices in paediatric ERP research that would aid interpretation and cross-study comparison.118

There are some specific limitations to the current ERP literature on executive function development that should be considered in the development of future studies. For one, the association between specific components and behaviourally defined executive function constructs is often unclear. Irreconcilable conflicts between neurophysiological findings and cognitive theory may necessitate the development of new models. Second, certain domains of executive function such as switching are well established on a behavioural level, but few studies have investigated them with ERP methods so there are insufficient studies for appropriate review.119 Third, while there is a substantial body of ERP research investigating executive functions in ADHD, other disorders with well-known executive deficits are less well studied. Based on these limitations, we suggest that future studies aim to investigate the relation between the development of ERP components and behavioural executive performance longitudinally in developmental populations, as our current knowledge is limited by a lack of longitudinal focus. Greater sample sizes are also required to account for the substantial ERP changes in this period to enhance power and to better establish ERP correlates of developing executive functions. In this way, we can better understand the influence of some factors investigated in studies discussed in this review such as temperament, puberty, and sex, as well as the influence of age and behavioural ability. We also suggest that more focus should be placed on using the ERP method to focus on less well-established components, such as the neural response associated with switching. Finally, we suggest that researchers use the ERP as a methodology to better understand executive development and dysfunction in less well-studied patient populations such as children with developmental disorders like Tourette syndrome, obsessive–compulsive disorder, and children with acquired brain injury. By taking ERP research further in these directions, we will be better equipped to interpret the significance of individual differences in ERP components and be better able to utilize this method for more informed diagnosis and treatment.

In summary, all of the components reviewed show developmental changes through adolescence and have been linked to specific regions of the brain networks underlying executive skills (Fig. 1). Future research may take advantage of using these components as markers of functional development or dysfunction of these brain regions and as an index of developmental differentiation of the executive system.