Response priming in the Go/NoGo task: The N2 reflects neither inhibition nor conflict
Introduction
Several robust differences in the event-related potential (ERP) are observed when responses must be executed or inhibited in the Go/NoGo task. The N2 component is increased in the frontal region, and the P3 is increased in the frontocentral region, on NoGo compared to Go trials (e.g., Pfefferbaum et al., 1985, Kok, 1986, Jodo and Inoue, 1990, Jodo and Kayama, 1992, Roberts et al., 1994, Fallgatter et al., 1997, Fallgatter and Strik, 1999, Van’t Ent and Apkarian, 1999, Bruin and Wijers, 2002, Nieuwenhuis et al., 2003, Bekker et al., 2004, Smith et al., 2006). Debate is ongoing in the literature as to which component best reflects inhibitory processing, since results from other inhibitory tasks show both similarities and differences relative to results from the Go/NoGo task. For example, in the stop-signal task, the N2 is larger on trials where inhibition is unsuccessful, while the P3 is larger on successful trials (De Jong et al., 1990, Brandeis et al., 1998, van Boxtel et al., 2001, Dimoska et al., 2003, Kok et al., 2004, Ramautar et al., 2004, Bekker et al., 2005). Researchers using the stop-signal task generally agree that the N2 may represent an overlap with error-related processing (Kok et al., 2004), while the P3 may reflect evaluation of the outcome of inhibition (Dimoska et al., 2003), or the inhibitory process itself (Kok et al., 2004, Ramautar et al., 2004, Bekker et al., 2005). In the Eriksen flanker task, the N2 is interpreted as reflecting an inhibitory process, since the N2 shows robust effects of target-flanker incompatibility, while the P3 shows almost none (Gehring et al., 1992, Kopp et al., 1996a, Kopp et al., 1996b, Heil et al., 2000, van Veen and Carter, 2002, Bartholow et al., 2005). In the Posner task (Posner et al., 1978), where cues validly or invalidly predict the identity of a target (and thus the required response), N2 and P3 are both larger on invalidly than validly-cued trials (e.g., Gehring et al., 1992, Band et al., 2003). Importantly, an overt response is still executed on invalidly cued trials in the Posner task, and incongruent-flanker trials in the Eriksen task, suggesting that the N2 and P3 “inhibition” effects in other tasks may reflect conflict between competing responses, rather than total inhibition of any response. Indeed, many of the results cited above could be reinterpreted in light of the conflict hypothesis (e.g., Nieuwenhuis et al., 2003): the observed N2 and P3 effects could be due to a signal to change the planned response to a different one. Conflict between the Go and NoGo response can occur, as well as conflict between the Go and Stop responses, the responses demanded by the target and by flanker stimuli, and between validly and invalidly cued responses. The functional significance of the N2 and P3 ‘inhibitory’ effects is thus unclear.
A potential confound with many of these experimental paradigms is the comparison of trials where movement-related potentials do or do not occur (e.g., Go vs. NoGo, unsuccessful vs. successful stop). Pfefferbaum et al. (1985) showed that the N2 NoGo > Go effect was greater, but the P3 effect was unchanged, when participants counted Go stimuli, compared to when they pressed a button in response. A study by Bruin and Wijers (2002) is often cited as showing similar effects (a NoGo > Go N2 effect for counting comparable to that for pressing, and a more anterior P3 focus for NoGo than Go in both tasks), yet the usual effect of larger P3 amplitudes for NoGo than Go trials was not found in the counting task. The NoGo P3 in the count condition was never larger than the Go P3, even at frontocentral sites. Thus the contribution of movement-related potentials to the observed Go/NoGo effects remains a question (Salisbury et al., 2001, Salisbury et al., 2004).
The problem of movement-related potential overlap can be avoided, however, by comparing the outcomes when responses are differentially primed (Bruin et al., 2001). The use of cues or primes results in response preparation in both Posner-type tasks (e.g., Gehring et al., 1992, Band et al., 2003, Leuthold, 2003) and the Eriksen task, where analyses of the lateralised readiness potential (LRP) shows response processing according to the flankers (Gratton et al., 1988, Smid et al., 1990, Kopp et al., 1996a, Kopp et al., 1996b, Heil et al., 2000). Where this response preparation is inappropriate, inhibition is necessary, and presumably greater inhibition is required following greater response preparation.
Bruin et al. (2001) utilised differences in response preparation to examine response inhibition processes, by presenting participants with three different targets, requiring a left button press, a right button press, or a NoGo response. On each trial, the target was preceded by one of four different equiprobable cues which predicted either (a) a NoGo target on 100% of those trials (the Specific NoGo cue), (b) a Go Left or Go Right target on 25% of trials each, or a NoGo on 50% of trials (the Non-specific Go cue), (c) a Go Left or NoGo target on 50% of trials each (the Specific Go Left cue), or (d) a Go Right or NoGo target on 50% of trials each (the Specific Go Right cue). The authors reasoned that response preparation should be elicited according to the cue (and thus, according to what targets could be presented), and that the strength of inhibition required should increase over these varying levels of response preparation. Despite finding an N2 NoGo effect following Non-Specific and Specific cues, the critical test of ERPs to NoGo stimuli following each of the cues revealed no difference in amplitude in the N2 time range (200–280 ms post-target). However, the NoGo P3 did show an increase in amplitude according to the prediction of the cue. Analyses of the LRP timelocked to targets revealed a slightly earlier onset of significant lateralisation to Go targets following Specific than Non-specific cues, but no LRP activity was present to NoGo targets following Specific cues. In addition, no lateralisation according to the expected responding side was found in the cue-target interval. The authors interpreted their lack of an N2 difference to NoGo targets following different cues as evidence against an inhibitory interpretation of this component, and concluded instead that the P3 may reflect the inhibitory process.
Although the conclusions of Bruin et al. (2001) are compelling, there are some problems with their methodology and interpretations. For instance, whether the NoGo N2 amplitude is affected by the cue may depend on whether response preparation according to the cue is actually elicited: Bruin et al. did not find LRP activity signalling this response preparation in the cue-target interval. The lack of response preparation according to the cue may in turn be caused by the low global response probability (37.5% overall) and/or by the percentage of validly cued trials (although participants could trust the cue information on 100% of NoGo cue trials, Non-specific and Specific cues predicted the correct target on only half the trials). If the cue predicts the target correctly only at chance levels, and if NoGo targets are presented on a large proportion of trials (62.5%), then high levels of cue-related preparation for a Go response are unlikely.
The current study modifies the task of Bruin et al. (2001) in several ways, with the aim of gaining further understanding of the functional correlates of the N2 and P3. The global response probability was raised to 66%, in order to ensure that response inhibition was difficult whenever a NoGo stimulus was presented. With respect to cue validity, the overall percentage of validly cued trials was similar to that of Bruin et al., but for individual cue types, the percentage was slightly higher (60% compared to 50%). In order to further examine the response conflict interpretations of N2 and P3, invalid cues were introduced, such that specific cues could be followed by not only the expected valid target (e.g., a Specific Left cue followed by a Go Left target) or a NoGo stimulus, but also by an invalidly cued target (Go Right), so that conflict between the planned and required response could be induced.
By increasing both cue validity and response probability, participants would be expected to increase their use of the information given by the cue to prepare fast responses. This would be revealed by differences in amplitude and topography of the late CNV to Specific, Non-specific and NoGo cues. It was also expected that the usual N2 and P3 NoGo effects would be seen in comparisons of Go and NoGo targets after Non-specific and Specific cues, but the primary tests of interest were comparisons of NoGo targets after different cues, and of Invalidly vs. Validly cued Go targets following Specific cues. If a component reflects motoric inhibition, then its amplitude should increase as a function of prior levels of preparation; if it reflects response conflict, then it should be larger for Invalid than Valid targets.
Section snippets
Subjects
Participants were 26 adults (11 male) with a mean age of 22.6 years (SD 7.2 years) who participated to fulfil an undergraduate course requirement. All participants were right-handed and had not consumed caffeine in the 2 h prior to testing, alcohol or illicit drugs in the previous 24 h, or illicit drugs more than once a month for the past six months. No participants reported any neurological disorders or problems with vision or hearing. The research protocol was approved by the joint University
Behavioural performance
Responses to targets were faster following Valid than Non-Specific cues (345 vs. 360 ms, F = 9.9, p < .01), but there was no difference in RT following Invalid and Non-Specific cues (360 vs. 365 ms, F = 1.7, p > .05). In addition, there was an effect of response side which approached significance (F = 4.1, p = .055), with faster responses for the right than left hand (352 vs. 362 ms).
Regarding inhibitory accuracy following the four cues, participants made more errors following Go than NoGo cues (0.83 vs.
Discussion
This study was designed to examine response competition processes via the use of informative cues. Cue stimuli were used to induce varying levels of response preparation, with the aim of examining variations in response inhibition and response conflict when the planned response was inappropriate. The major comparisons of interest were the late CNV following the cue, indexing preparation of the expected response, increased inhibitory activity to NoGo targets according to the preparation for a Go
Acknowledgements
The authors thank Rodney Davies for his enthusiastic and expert assistance in writing the stimulus presentation programs.
References (65)
- et al.
Effect of task and stimulus probability on evoked potentials
Biol Psychol
(1981) - et al.
Electrophysiological correlates of attention, inhibition, sensitivity and bias in a continuous performance task
Clin Neurophysiol
(2004) - et al.
The pure electrophysiology of stopping
Int J Psychophysiol
(2005) - et al.
Age-related differences in auditory event-related potentials during a cued attention task
Clin Neurophysiol
(2004) - et al.
Neuroelectric mapping reveals precursor of stop failures in children with attention deficits
Behav Brain Res
(1998) - et al.
Inhibition, response mode, and stimulus probability: a comparative event-related potential study
Clin Neurophysiol
(2002) - et al.
Response priming in a go/nogo task: do we have to explain the go/nogo N2 effect in terms of response activation instead of inhibition?
Clin Neurophysiol
(2001) - et al.
Electroencephalographic nogo potentials in a no-movement context: the case of motor imagery in humans
Neurosci Lett
(2004) - et al.
The N2 in go/no-go tasks reflects conflict monitoring not response inhibition
Brain Cogn
(2004) - et al.
ERP components in Go/Nogo tasks and their relation to inhibition
Acta Psychol
(1999)