Electroencephalography of response inhibition tasks: Functional networks and cognitive contributions

Abstract

Response inhibition paradigms, as for example stop signal and go/no-go tasks, are often used to study cognitive control processes. Because of the apparent demand to stop a motor reaction, the electrophysiological responses evoked by stop and no-go trials have sometimes likewise been interpreted as indicators of inhibitory processes. Recent research, however, suggests a richer conceptual background. Evidence denotes an association of a frontal-midline N200/theta oscillations with premotor cognitive processes such as conflict monitoring or response program updating, and an anterior P300/delta oscillations with response-related, evaluative processing stages, probably the evaluation of motor inhibition. However, the data are still insufficient to unambiguously relate these electroencephalographic measures to specific inhibitory functions. Beta band activity only recently has become a focus of attention in this task context because of its association with the motor system and regions involved in inhibitory control. Its functional role in response inhibition tasks needs further exploration though. Hence, as things stand, any deduction of differences regarding actual inhibitory capabilities or loads between subject groups or conditions based on electroencephalographic measures has to be treated with caution.

Introduction

Human cognition and behavior vary dramatically from one moment to the other. These variations of course are not exclusively driven by the outcome of a random process, but rather reflect our ability, and the necessity, to adapt to an ever-changing environment and to adjust or maintain our goals accordingly. A broad class of processes, altogether often referred to as cognitive control, render such flexibility possible by supporting goal and stimulus-response representations, issuing task processing strategies or attentional allocation, or by managing interferences in information processing and inhibition of inappropriate response tendencies. Inhibitory mechanisms, especially, have gained much interest due to their putative relevance for a variety of mental disorders such as attention deficit/hyperactivity disorder, schizophrenia or psychopathy. Response inhibition paradigms are often used to study inhibitory control in both healthy subjects and patient groups. Here, not only behavioral performance measures but also electroencephalographic variables are often considered immediate indicators of inhibitory processes.

This review summarizes studies utilizing go/no-go and stop signal tasks and electroencephalography (EEG) to explore the cognitive processes underlying response inhibition. First, the electrophysiological phenomena associated with these tasks will thoroughly be assessed, accompanied by a description of anatomical networks contributing to the generation of these EEG signals. Then, we will examine EEG effects from experiments that investigated motor inhibition and will assess their behavioral relevance. A number of theories already exist that try to link event-related potentials (ERP) and EEG oscillations to cognition and behavior; thus, a section of this review is devoted to these frameworks. Integrating the reviewed information, we arrive at the conclusion that commonly postulated associations of EEG measures and inhibitory mechanisms are only insufficiently supported. Relevant implications for cognitive and clinical neuroscience are shortly discussed and some recommendations for future ventures are provided.

Most response inhibition paradigms impose the need to adjust response strategies in a multi-tasking situation. Subjects are required to respond with high pace to one sort of stimulus while they are instructed to withhold responses to a second type of stimulus. The most common paradigms employed to study response inhibition without doubt are go/no-go and stop signal tasks, which are depicted in Fig. 1. Although similar in nature, these tasks differ in one important aspect: while in the go/no-go task a stimulus on a given trial is unambiguously indicating to either respond or not, stop-trials in the stop signal task first elicit a response by presenting a go-stimulus that is subsequently followed by a stop-signal indicating the necessity to withhold the already initiated response.

Typical behavioral dependent variables derived from both tasks are the average reaction time to go-trials as well as the number of false alarms, i.e., the number of responses made in context of no-go or stop signal presentation. The stop signal task furthermore allows the calculation of the stop signal reaction time (SSRT) that aims at the quantification of the latency of the inhibition mechanism (e.g., Band et al., 2003a, Logan et al., 1984). SSRT estimation rests on the assumption that processes to produce and to stop a response, triggered by the presentation of a go and a stop signal, respectively, independently race for execution. Hence, when the stop process finishes before the go process does, the response is successfully inhibited. With respect to electrophysiological responses, both no-go and stop conditions evoke two event-related potentials: frontal-midline N200 and P300, often together referred to as the N2/P3-complex (e.g., Simson et al., 1977, De Jong et al., 1990). When subjected to time-frequency decompositions, augmented power in theta (4–8 Hz) and delta (0–4 Hz) frequencies is reliably discernible. More recent studies also found activity in the beta band (12–30 Hz) at fronto-central electrodes to be of relevance (e.g., Alegre et al., 2008, Krämer et al., 2011). These electrophysiological responses evoked in response inhibition paradigms have always been of special interest, because they were expected to more directly relate to inhibitory processes.

The basic concepts behind go/no-go and stop signal tasks and the phenomenology of electrophysiological responses are highly similar and hence corresponding effects are interpreted virtually the same irrespective of the precise task context. There are only few studies that directly compared go/no-go and stop signal tasks and they do not generally argue against such a comparability. Using ERP methodology van Boxtel et al. (2001) showed similar ERP patterns in no-go- and stop trials, which suggests similar underlying mechanisms. Rubia et al. (2001) were the first to use functional magnetic resonance imaging (fMRI) to assess common and specific patterns of activations with go/no-go and stop signal variants. They found the majority of activation maxima in conditions associated with response inhibition to be congruent. A common network was spanned by bilateral middle and inferior frontal gyri, midcingulate and parietal cortices as well as the pre-supplementary motor area (preSMA). Minor differences were observed in middle and medial frontal as well as inferior parietal regions. Can we, however, actually expect to have zero differences in fMRI when comparing tasks with varying stimulus contexts, presentation times or interstimulus intervals, even though they indeed assess the same cognitive concept? Whether stop signal and go/no-go tasks are based on the same cognitive processes has also been addressed on a more conceptual level. Verbruggen and Logan (2008a), for example, argue that the inhibitory process in the go/no-go task is automatic due to consistent stimulus-response mappings, whereas successful stop signal inhibition may rely on a more controlled process. Still, their experimental data show that even in the stop signal task automatic inhibition does occur. Hence, let us for now put our doubts aside and accept the notion that both tasks basically rely on the same cognitive constructs. We will revisit this issue in the discussion after having reviewed experimental findings of both tasks.

Other task types used to study inhibitory processes often strongly resemble go/no-go and stop signal tasks in basic characteristics. Some variants of the continuous performance test (CPT), for example, do not only probe capabilities in sustained attention. Here, participants are subjected to a large number of different stimuli delivered at a high pace (e.g., every 500 ms). In some versions of this paradigm the subjects are asked to respond to every letter of the alphabet except for the letter X, with all letters having the same probability of presentation. This procedure strongly resembles a go/no-go task with varying go stimuli and a very low probability of no-go trials. Not surprisingly, no-go stimuli in this CPT also evoke an N2/P3-complex (e.g., Duncan et al., 2005). Another frequently utilized paradigm is the anti-saccade task during which target stimuli are presented to the visual periphery, thereby usually triggering an orienting response (pro-saccade). Previously presented cues, however, might instruct the subject to look in the opposite direction (anti-saccade). It is argued, though, that anti-saccade trials not only involve inhibition of the pro-saccade but also an active mechanism for the inversion of the response (e.g., Munoz and Everling, 2004). Recent evidence indeed suggests that tasks involving not only the necessity to stop but also to change a response pattern qualitatively differ from simple no-go or stop trials. Krämer et al. (2011) for example reported absence of a frontal-midline N200 in change as opposed to stop conditions. Hence, for the remainder of this review we will focus on go/no-go and stop signal paradigms as these by far outnumber other task types to study cognitive control in the context of response inhibition. In addition, they are considered rather pure measures of inhibitory processes.

Because of the apparent demand to suppress motor reactions, the electrophysiological responses evoked by stop and no-go trials have sometimes likewise been interpreted as indicators of inhibitory processes. Hence, these tasks and associated psychophysiological dependent variables are of clinical relevance, as many psychiatric symptoms are believed to originate from inhibitory deficits (e.g., Chamberlain and Sahakian, 2007). Current theories on the families of N200- and P300-like potentials, as well as brain oscillations, suggest a richer conceptual background though. In addition, the concept of inhibition itself has partially been challenged. Without question, inhibition exists in various forms in a neurophysiological context as, for example, seen with inhibitory neurons in simple muscle reflexes or in the form of inhibitory postsynaptic activity. Whether the concept of inhibition is of relevance in more cognitive domains, for example serving the suppression of activated cognitive contents or the control of interfering information, is still being disputed. It has been suggested that many observed experimental effects could well be explained in terms of facilitated processing of task-relevant rather than the suppression of task-irrelevant information (e.g., MacLeod et al., 2003, Egner and Hirsch, 2005, Aron, 2007).

However, behavioral inhibition – the control of overt behavior – is an exceptional case. Using transcranial magnetic stimulation (TMS), it has been shown that in both go/no-go and stop signal tasks intracortical inhibition in cortical motor regions indeed is increased in conditions calling for the suppression of a manual response (Chambers et al., 2006, Coxon et al., 2006, Waldvogel et al., 2000). For example, van den Wildenberg et al. (2009) used TMS to show increased excitability of inhibitory interneurons and decreased corticospinal excitability in stop relative to go trials. These inhibitory effects became evident as early as 130 ms after stop signal presentation.

Still, although TMS studies provide convincing evidence for the contribution of proper inhibitory mechanisms to response inhibition performance, these are not necessarily reflected in scalp EEG recordings. EEG relates to the summation of synchronous postsynaptic events of millions of neurons rather than to their spiking activity. Therefore, EEG is more informative with respect to these neurons' afferent input. Whether their efferent output has been increased, decreased, or not changed at all, cannot easily be inferred from EEG measurements.

Hence, there is no a-priori reason to assume that any of the observed electrophysiological responses with stop or no-go trials serves as a sensitive indicator of inhibitory processes. In addition, one has to differentiate to what degree potential EEG measures reflect the activity of inhibitory control regions (e.g., the inferior frontal cortex or the pre-supplementary motor area) or the outcome of high-level inhibitory control at specific target sites (e.g., the motor cortex).

Despite the conceptual problems raised, many studies tried to elucidate which cognitive mechanisms contribute to an adequate performance in go/no-go and stop signal tasks, and how they relate to the electrophysiological phenomena observed by means of scalp EEG recordings. The next section will describe relevant EEG effects originating from brain activity in contexts calling for withdrawal or suppression of overt behavior.