Differential modulation of the N2 and P3 event-related potentials by response conflict and inhibition

Highlights

• Response inhibition and response conflict yield dissociable effects on the N2 and P3.

• The N2 is modulated by flanker-induced response conflict but not inhibition.

• P3 amplitude displays a frontal shift in topography for response inhibition only.

• The specificity of these ERPs may improve understanding of action control.

• The specificity of these ERPs may improve understanding of disorders such as ADHD.

Abstract

Background: Developing reliable and specific neural markers of cognitive processes is essential to improve understanding of healthy and atypical brain function. Despite extensive research there remains uncertainty as to whether two electrophysiological markers of cognitive control, the N2 and P3, are better conceptualised as markers of response inhibition or response conflict. The present study aimed to directly compare the effects of response inhibition and response conflict on the N2 and P3 event-related potentials, within-subjects. Method: A novel hybrid go/no-go flanker task was performed by 19 healthy adults aged 18–25 years while EEG data were collected. The response congruence of a central target stimulus and 4 flanking stimuli was manipulated between trials to vary the degree of response conflict. Response inhibition was required on a proportion of trials. N2 amplitude was measured at two frontal electrode sites; P3 amplitude was measured at 4 midline electrode sites. Results: N2 amplitude was greater on incongruent than congruent trials but was not enhanced by response inhibition when the stimulus array was congruent. P3 amplitude was greater on trials requiring response inhibition; this effect was more pronounced at frontal electrodes. P3 amplitude was also enhanced on incongruent compared with congruent trials. Discussion: The findings support a role for N2 amplitude as a marker of response conflict and for the frontal shift of the P3 as a marker of response inhibition. This paradigm could be applied to clinical groups to help clarify the precise nature of impaired action control in disorders such as attention deficit/hyperactivity disorders (ADHD).

Introduction

The term cognitive control refers to a suite of cognitive processes that enable us to navigate the world in a goal-directed manner. There is consensus that the pre-frontal cortex and its anatomical connections with other brain regions, particularly the basal ganglia and parietal and motor cortices, underpin effective cognitive control (Arnsten and Rubia, 2012, Ridderinkhof et al., 2004). However, the degree to which the sub-processes of cognitive control can be fractionated remains unclear. More precise characterisation of the sub-processes of cognitive control and the degree to which they rely on distinct or overlapping neural circuits will refine understanding of human brain function. To achieve this it is crucial to identify neural markers specific to one or a set of cognitive processes and to understand the factors that influence the morphology and presentation of such markers.

Developing reliable and specific neural markers of cognitive processes also has implications for improving our understanding of atypical brain function. Impaired cognitive control has been reported in a range of neurodevelopmental disorders, particularly attention deficit/hyperactivity disorder (ADHD) (Groom et al., 2008, Groom et al., 2010, Pliszka et al., 2000, Yong-Liang et al., 2000) and schizophrenia (Groom et al., 2008, Kiehl et al., 2000, Roche et al., 2004). However the precise features of cognitive control that are impaired have not been well characterised, partly because of a lack of consistency in how these processes are measured and described. Improving the reliability of markers used to measure specific aspects of cognitive control will increase knowledge of the brain systems involved and how these systems are impaired in specific disorders.

One area which may benefit from more precise definition of neural markers is the study of action selection and control. In electrophysiological studies the N2 event-related potential (ERP), a fronto-central stimulus-locked component with a latency of between 200 and 350 ms, has traditionally been interpreted as an index of response inhibition (Falkenstein et al., 1999, Jodo and Kayama, 1992), defined in this context as the cancellation of a prepotent or prepared motor response. In the visual go/no-go task, in which a rapid unimanual response to a ‘go’ stimulus must be inhibited when an infrequent ‘no-go’ stimulus is presented, and in the stop signal task, in which the go stimulus is followed occasionally by a signal to cancel the prepared response, N2 amplitude is larger on no-go trials than on go trials and on successful compared to failed stop trials (Enriquez-Geppert et al., 2010, Kok et al., 2004, Ramautar et al., 2006, Schmajuk et al., 2006).

An alternative school of thought is that the N2 is not modulated by response inhibition specifically but by conflict between competing responses (Donkers and van Boxtel, 2004, Randall and Smith, 2011, Smith et al., 2010, van Veen and Carter, 2002). According to this model, the N2 is modulated in inhibitory control tasks because the ratio of go to no-go or stop trials creates conflict between the prepotent response tendency and the infrequent requirement to inhibit the response (Braver, Barch, Gray, Molfese, & Snyder, 2001), not because of inhibition per se. In support of this, N2 amplitude is greater on go than no-go trials when the ratio of go:no-go trials is reversed (Donkers and van Boxtel, 2004, Enriquez-Geppert et al., 2010, Nieuwenhuis et al., 2003) and on incongruent than congruent flanker trials in the visual flanker task (Bartholow et al., 2005, Clayson and Larson, 2011, Kopp et al., 1996, Purmann et al., 2011). In the flanker task participants must respond to a central target stimulus while simultaneously suppressing an opposing response associated with the flanking stimuli. Thus, response inhibition is not required but N2 amplitude enhancement is observed, providing support for the hypothesis that the N2 is modulated by response conflict rather than inhibition specifically.

Few studies have directly compared no-go and incongruent flanker trials and it is therefore unclear whether they produce equivalent effects on the N2. Of the studies conducted to date, Brydges et al. (2012) report a more frontal topography for the N2 on no-go trials than on incongruent flanker trials and Heil, Osman, Wiegelmann, Rolke, and Hennighausen (2000) report N2 amplitude enhancement on both no-go and incongruent flanker trials. However, neither study successfully dissociated inhibition and conflict: the congruence of the array differed between go and no-go stimuli and in Brydges et al. (2012) the stimuli also differed in colour. Kopp, Mattler, Goertz, and Rist (1996) compared no-go trials in which a central no-go stimulus was flanked either by stimuli associated with a left/right hand response (specific primes) or by neutral stimuli (non-specific primes). The N2 was larger to no-go stimuli flanked by specific primes which the authors suggested reflected inhibition of the primed response; however there was no direct comparison between no-go and go flanker trials and it could be argued that the specific prime induced conflict between competing response options rather than inhibition of the primed response. It is therefore unclear from the research published so far whether the N2 responds differently to flanker-induced response conflict and response inhibition.

The lack of clarity over the role of the N2 as a marker of response inhibition and/or response conflict is reflected in ADHD research. Studies using go/no-go and stop signal tasks have often reported high rates of inhibitory errors coupled with reduced N2 amplitude (Barry et al., 2003, Brandeis et al., 1998, Groom et al., 2010, Liotti et al., 2005) interpreted as poor inhibitory control. However, as outlined above, response inhibition paradigms may reflect conflict between competing responses as well as (or instead of) the cancellation of a planned or prepotent response. Moreover, recent research suggests more widespread impairment in action regulation and monitoring in ADHD (Johnson et al., 2007, Kuntsi and Klein, 2012, Simmonds et al., 2007). Increasing understanding of the factors that influence the N2 could lead to more precise characterisation of the nature of impaired performance on action control tasks in ADHD and other disorders associated with impaired action control.

Another ERP which is frequently measured in studies of action control is the P3. This ERP appears between 300 and 500 ms post-stimulus and is observed in a range of cognitive tasks. In tasks requiring attention to a target stimulus, the P3 is maximal over posterior scalp electrodes and is thought to reflect updating of working memory (Polich, 2004). In response inhibition tasks it is larger in amplitude on no-go trials as well as successful stop trials (Enriquez-Geppert et al., 2010, Fallgatter and Strik, 1999, Smith et al., 2007) and has a more frontal topography (termed ‘no-go anteriorisation, NGA) (Fallgatter et al., 1997, Fallgatter and Strik, 1999), potentially reflecting activity in pre-SMA and inferior frontal cortex (Huster et al., 2011) or cingulate regions (Fallgatter et al., 2002, Strik et al., 1998). This has led to the suggestion that in the context of action regulation, the P3 is a marker of response inhibition with some authors proposing it as a more reliable marker of response inhibition than the N2 (Kropotov et al., 2011, Meel, 2005, Randall and Smith, 2011, Smith et al., 2007).

P3 amplitude is also enhanced by response conflict (Clayson and Larson, 2011, Smith et al., 2007). However, to our knowledge no studies have compared P3 amplitude and topography on no-go and incongruent flanker trials within the same task design and it remains unclear whether the amplitude enhancement and frontal shift in topography are specific to response inhibition or also occur in relation to flanker-induced response conflict. Moreover, the P3 is altered in clinical groups such as those with schizophrenia and ADHD (Bekker et al., 2005, Groom et al., 2008, Groom et al., 2010, Hughes et al., 2012, Wiersema and Roeyers, 2009) and the NGA described by Fallgatter et al. is also reduced in these groups (Fallgatter, 2001, Fallgatter and Muller, 2001, Fallgatter et al., 2004). Improving understanding of the role of the P3 as a marker of action regulation could therefore enhance knowledge of the neuro-cognitive factors underpinning these disorders.

To determine whether the N2 and P3 are modulated by response inhibition, response conflict or both, we administered a hybrid go/no-go flanker task to healthy young adults, with a view to applying the same measure in a future study of clinical groups. The task included trials with varying levels of response conflict and inhibitory control but with minimal differences in visual characteristics between trial types. The design facilitated a within-subjects comparison of the effects of response inhibition and flanker-induced response conflict on N2 amplitude and P3 amplitude and topography. On each trial a 5-element stimulus-array was presented (depicted in Table 1, column 3) and participants were instructed to respond to the central target stimulus. Stimuli were directional arrows signalling either a left- or right-hand response or no-go (up arrow). The congruence of the central target stimulus and flanking stimuli was manipulated between trials to produce trials with minimal conflict and no response inhibition (low conflict, no inhibition), trials requiring inhibition but within a congruent stimulus array (low conflict, inhibition), trials with conflict between competing left and right responses (high conflict, no inhibition) and trials with conflict between a unimanual right-hand response and response inhibition (high conflict, inhibition). The task included trials in which a no-go stimulus was flanked by stimuli associated with a unimanual response and vice versa to ensure that any stimulus (right, left, no-go) could be part of a congruent or an incongruent array. This provided greater separation of Congruence and Response Type and consistency while maintaining the visual properties of the different trial types.

With reference to Table 1, we reasoned that if the N2 is a marker of response inhibition, amplitude will be significantly greater on congruent no-go (NNN) than go (RRR) trials. The task was designed so that the frequency of congruent go and no-go trials was equal. This ensured minimal conflict between ‘go’ and ‘no-go’, providing a strong test of whether either ERP is modulated by response inhibition when response conflict is minimal. If the N2 is a marker of conflict between competing responses rather than response inhibition, amplitude will be greater on incongruent left and right (RLR, LRL) than congruent left and right (LLL, RRR) but will not differ between NNN and RRR.

With regard to the P3, we predicted that if this ERP is a marker of response inhibition, amplitude will be significantly greater and topography more fronto-central on trials requiring response inhibition (NNN, RNR) than on other trial types. If amplitude enhancement and NGA are specific to response inhibition, there will be no difference between high conflict (LRL, RLR) and low conflict (RRR, LLL) trials where the inhibition of a response is not required. Alternatively, if the P3 responds to trials with high response conflict as well as to those with inhibitory requirements, amplitude will be greater and topography more frontal for all trial types requiring response inhibition or the resolution of response conflict (NNN, RNR, LRL, RLR, NRN) relative to those free from inhibition or conflict (RRR, LLL).

Finally, if response inhibition and response conflict produce different effects on either the N2 and/or the P3, we reasoned that this would provide evidence for at least some separation of the neural circuits underlying these cognitive processes and would suggest that response inhibition and response conflict are not wholly equivalent.