Neural correlates of stopping and self-reported impulsivity · Neural correlates of stopping and self-reported impulsivity Marieke M. Lansbergen a,*, Koen B.E. Bo¨cker a, Evelijne

www.elsevier.com/locate/clinph

Clinical Neurophysiology 118 (2007) 2089–2103

Neural correlates of stopping and self-reported impulsivity

Marieke M. Lansbergen a,*, Koen B.E. Bocker a, Evelijne M. Bekker b, J. Leon Kenemans a

a Departments of Experimental Psychology and Psychopharmacology, Heidelberglaan 2, 3584 CS Utrecht, Utrecht University, The Netherlandsb F.C. Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands

Accepted 12 June 2007

Abstract

Objective: To examine the relation between self-reported impulsivity, inhibitory control, and the neural correlates of stopping perfor-mance within the normal population.Methods: Healthy individuals scoring high and low on trait impulsivity performed an auditory stop-signal task. Stopping performanceand neural correlates of stopping (i.e. N1 and stop P3) were compared between the impulsive groups as well as between participants whowere slow and fast in stopping.Results: As expected, N1 and stop P3 were larger for successful relative to failed stops (i.e. N1 and stop P3 effects). Participants scoringhigh relative to low on impulsivity showed equal stopping performance, had larger stop P3, but similar N1 effects. Slow as compared tofast stoppers had reduced stop P3, but similar N1 effects.Conclusions: Participants scoring high relative to low on impulsivity may need more effortful inhibitory control to yield equal stoppingperformance. Slow relative to fast stoppers may have weaker inhibition processes and abnormal error processing. In contrast to ADHD,both high impulsives as well as slow stoppers had an intact N1 effect.Significance: Subjective impulsivity and slow stopping in healthy individuals cannot be generalized to ADHD.� 2007 Published by Elsevier Ireland Ltd. on behalf of International Federation of Clinical Neurophysiology.

Keywords: Stop-signal task; Impulsiveness; Inhibitory control; Attention deficit hyperactivity disorder; Frequency

1. Introduction

Impulsivity is associated with risky and spontaneousbehavior, acting without thinking, and recklessness, buthas also been specifically defined as the inability to inhibitinappropriate behavior or as an aversion to wait. Severalmeasures have been developed to assess impulsivity, includ-ing questionnaires that rely on individuals’ perceptions oftheir own behavior. High levels of impulsivity have oftenbeen seen as dysfunctional. The construct impulsivity isalso a defining characteristic of several psychiatric disor-ders, such as Attention Deficit Hyperactivity Disorder(ADHD; DSM-IV, American Psychiatric Association,1994). It is an interesting question as to what the relation

1388-2457/$32.00 � 2007 Published by Elsevier Ireland Ltd. on behalf of Inte

doi:10.1016/j.clinph.2007.06.011

* Corresponding author. Tel.: +31 (0)30 2534368; fax: +31 (0)302534511.

E-mail address: [email protected] (M.M. Lansbergen).

is between self-reported impulsivity in healthy individualsand impulsivity in psychopathology. The underlying defi-cits of impulsivity in ADHD patients may be an extremeform of normal subjective impulsivity. This may yield pos-sibilities to use healthy individuals, characterized byextreme self-reported impulsivity, as model samples forADHD.

Barkley (1997) argued that deficient inhibitory controlmay be the underlying deficit of impulsivity in ADHD.One important aspect of inhibitory control is the abilityto stop an on-going or prepotent response. The stop-signalparadigm has often been used to study this variety of inhib-itory control empirically (Logan, 1994). In the stop-signalparadigm, the primary task is typically a visual choice reac-tion time task. Participants have to respond as fast as pos-sible to a specific stimulus (go-stimulus). On some trials, astop-signal (auditory or visual) is presented and partici-pants must stop their response to the go-signal. Inhibitory

rnational Federation of Clinical Neurophysiology.

mailto:[email protected]

2090 M.M. Lansbergen et al. / Clinical Neurophysiology 118 (2007) 2089–2103

control in the stop task, reflected in the time needed to stopthe ongoing response (stop-signal reaction time; SSRT),can be estimated even though successful inhibitory controldoes not produce an overt response (Logan, 1994).

The stop-signal paradigm has often been used to inves-tigate inhibitory control in psychiatric patients with pre-sumed low impulse control (e.g. ADHD; Bekker et al.,2005c), but also in healthy individuals who are character-ized by high self-reported impulsivity. This may provideclues as to whether the impulsivity in ADHD is an extremeform of self-reported impulsivity in healthy individuals. Ifso, we would expect differences between high and lowimpulsive healthy individuals similar to those betweenADHD and controls. Robust differences have been foundbetween ADHD patients and controls, especially in termsof speed of stopping, with slower stopping reaction timesin ADHD, suggesting an inhibitory deficit (Lijffijt et al.,2005b; Oosterlaan et al., 1998). In high impulsive healthyindividuals, as assessed by self-report measures, inconsis-tent findings have been reported regarding deficient stop-ping performance (Avila and Parcet, 2001; Lijffijt et al.,2004; Logan et al., 1997; Marsh et al., 2002; Rodriguez-Fornells et al., 2002; Vigil-Colet and Codorniu-Raga,2004). One important factor here may concern the percent-age of stop trials in the stop-signal task. Most studies thatused a low probability of stop trials within their task (i.e.25% as compared to 40–50%) have found significant corre-lations between impulsivity and stopping performance.

How this exactly relates to differences in SSRT isunclear. The general effect of decreasing stop probabilityis shortened reaction times to go-signals (Logan, 1994).Ramautar and colleagues (Ramautar et al., 2004, 2006b)showed that participants develop a stronger tendency torespond to a go-stimulus when there is a low probabilityof stop-signals, resulting in an impulsive response style thatis characterized by faster reaction times to go-stimuli andmore failed stops. In this situation, more inhibitory effortis necessary to stop the stronger tendency to respond tothe go-signal. The stopping reaction time did not increasewith lower probability of stop-signals. Consistently, Dim-oska et al. (2006) demonstrated more failed stops, but nodifference for stop-signal reaction time between fast andslow groups (median split of reaction time to go-signal).So, in a stop-signal task with a low probability of stop-sig-nals, reaction times will be faster and the need for inhibi-tory control stronger, which may be especiallyproblematic in individuals characterized by high levels ofimpulsivity.

Another reason for the inconsistent findings betweenpoor stopping performance in ADHD and in healthyimpulsive individuals may be the use of self-report mea-sures to assess impulsivity in healthy participants. Self-report measures rely on participants’ own judgments whichcan be biased. For example, in adult ADHD it has beendemonstrated that parental reports have greater validitythan do self-reports (Barkley et al., 2002). Further, impul-sivity is a complex and multidimensional construct and

various self-report measures have been developed, eachmeasuring one or more components of impulsivity.Accordingly, impulsivity as assessed by self-report andbehavioral measures of inhibitory control may not capturesimilar processes (Reynolds et al., 2006). An alternativeway to classify healthy individuals as either high or lowimpulsive is by using stopping performance as a criterion.Poor stopping behavior is the most robust finding inADHD, especially in adult ADHD where longer stop-sig-nal reaction times were found without differences in goreaction times (Bekker et al., 2005b; Lijffijt et al., 2005b).Moreover, stopping behavior can be more objectivelyassessed than subjective impulsivity. Slow stoppers, whoare characterized by long stop-signal reaction times, mayshow similar deficits as individuals with ADHD, and sub-sequently may function as model samples for ADHD. Ithas already been demonstrated that groups divided by amedian split of stopping behavior can differ in behavioraleffects of medication (de Wit et al., 2000, 2002). Amphet-amine shortened SSRTs in slow, but not in fast, stoppers.

The estimation of the SSRT is an indirect measure ofstopping performance and depends on the assumption ofindependent go and stop processes, which is not alwayseasy to verify. Event related potentials (ERPs) have beenused to obtain more detailed insights into the nature ofresponse inhibition mechanisms. In general, previous liter-ature using the stop-signal task has focused on ERPs elic-ited by the stop-signal, and especially on the differencesbetween stop-ERPs to successful and to failed stops. Themost consistent finding thus far consists of a larger, cen-trally distributed positivity (150–250 ms after stop-signalonset) for successful, relative to failed, stops. This ‘stopP3’ has been interpreted as reflecting an inhibitory mecha-nism per se, but also as being related to error or conflictprocessing in case of failed stops (Bekker et al., 2005a;De Jong et al., 1990; Dimoska et al., 2006; Ramautaret al., 2004, 2006b; Schmajuk et al., 2006). A further recentfinding, observed for auditory stop-signals among visualgo-signals, is a larger negativity (‘N1’) for successful versusfailed stops as early as 100 ms after stop-signal onset,thought to mainly reflect the enhanced activation of audi-tory cortex associated with successful stops (Bekker et al.,2005a). The N1 may reflect the trial-to-trial varying impactthe stop-signal has in the auditory cortex. An enhanced N1amplitude to successful relative to failed stops suggests thatthe probability of successful stopping at least partlydepends on the amount of attention that is shifted to theauditory stop-signal (Bekker et al., 2005a,c; Kenemanset al., 2005). In addition to the stimulus-locked ERPs,recent studies have also examined response-locked ERPcomponents, especially the error-related negativity (ERNor Ne) and the error positivity (Pe) (Dimoska et al.,2006; Dimoska and Johnstone, 2007; Liotti et al., 2005;Ramautar et al., 2004, 2006a; Stahl and Gibbons, 2007;van Boxtel et al., 2001). The response-locked ERN hasbeen described as a negative deflection around 50–100 msafter an incorrect response and may reflect error detection

M.M. Lansbergen et al. / Clinical Neurophysiology 118 (2007) 2089–2103 2091

or error monitoring (Falkenstein et al., 1991; Gehring andKnight, 2000; Ullsperger and von Cramon, 2006). It is fol-lowed by the Pe, thought to reflect the awareness of conflictor the activation of compensatory processes after conflict(Overbeek et al., 2005). It has been speculated that theERN and Pe may overlap the stop P3 elicited by failedstops (Dimoska et al., 2006; Ramautar et al., 2004).

Previous studies have mostly found significantly reducedstop related ERPs in children and adults with ADHD rel-ative to controls. This especially holds for the stop P3 (Bek-ker et al., 2005c; Liotti et al., 2005; Overtoom et al., 2002).Regarding the N1, Bekker et al. (2005a) showed an N1effect (i.e. greater negativity to successful than failed stops)in healthy controls, but not in ADHD adults, indicatingdecreased or even absent inhibition from the auditory onthe motor cortex. In healthy individuals, there is a linkbetween the impact of the stop-signal in the sensory cortexand the probability of stopping the ongoing response,which is absent in ADHD. With respect to the ERN andPe, a reduced ERN (Liotti et al., 2005) and a reduced rel-atively late positivity (500–700 ms) in the ERP elicited byfailed stops, which may be related to the Pe (Overtoomet al., 2002), have been found in ADHD relative to controlchildren.

Especially, the visibility of the N1 effect has been shownto depend on the manner in which overlap between theERP to the stop-signal and the ERP to the preceding (oftena few hundreds of ms) go-stimulus is dealt with. Bekkeret al. (2005c) employed what is probably the most stringentcorrection procedure (ADJAR level 2, Woldorff, 1993),and showed that the stop N1 effect was only visible afterthis correction, not without it. In the present study,ADJAR level 2 was applied to correct for the overlap dis-tortion. The first aim of the present study was to explorestopping performance and the neural correlates of stoppingbehavior in healthy individuals that vary on self-reportimpulsivity. It was expected that individuals scoring high,relative to individuals scoring low on impulsivity, havelonger SSRTs and smaller or even no N1 and stop P3effects (i.e. greater ERP amplitudes for successful vs. failedstops).

During preparation of the present paper, Dimoska andJohnstone (2007) published a study that addressed the rela-tion between stopping performance and the neural corre-lates of stopping in healthy individuals, characterized bylow and high self-reported impulsivity scores. Whereasthe groups did not differ with respect to stopping perfor-mance (i.e. SSRT), high relative to low impulsive partici-pants showed a larger N1 and stop P3 to successful stopsas well as a larger ERN to failed stops. The enhanced N1for high impulsives was interpreted as reflecting moreattention to the stop-signal (Dimoska and Johnstone,2007). Given previous findings of an enhanced N1 to visualand auditory stimuli with increased stimulus intensity inhealthy individuals scoring high on self-reported impulsiv-ity and sensation seeking (Barrat et al., 1987; Carrillo-de-la-Pena and Barrat, 1993), Dimoska and Johnstone

(2007) argued that the enhanced N1 for high impulsivesmay reflect sensation seeking behavior. However, it shouldbe noted that the association between ERP augmentingand sensation seeking behavior is far from clear (Carrillo-de-la-Pena, 1992, 2001; Wang et al., 1999). Dimoska andJohnstone (2007) interpreted the enhanced P3 and ERNfor high impulsive individuals as greater inhibitory activa-tion in order to yield equal task performance, followed byenhanced error-related processing. With regard to theERN, contradictory findings have also been reported: Alarger ERN to failed stops as well as to successful stopsin healthy individuals scoring low versus high on self-reported impulsivity (Stahl and Gibbons, 2007).

The present study focused on the N1 and P3 componentsto failed and successful stops. ADJAR level 2 was applied tocorrect for overlap distortion. As already mentioned, self-report measures are biased by participants’ own perceptionsand a more objective method may function better as a modelfor ADHD. Therefore, the second aim of the present studywas to explore whether the neural mechanisms associatedwith poor stopping performance in healthy individuals arequalitatively similar to those observed in ADHD (i.e.reduced N1 and P3 stop effects). To this end, within thesample of subjects, two groups of healthy individuals, cre-ated by applying a median split of SSRT, were comparedwith respect to stop-related ERPs. It was expected that slowrelative to fast stoppers would show smaller or even no N1and stop P3 effects. In the present study, two versions of thestop task were used, one with a high (80%), and one with alow (20%), stop-signal probability. It was expected thatwhen stop-signals are presented less frequently, participantswill develop an impulsive response style, resulting in fastergo reaction times. In addition, in the stop task with a lowprobability of stop-signals more inhibitory effort is neededto suppress the stronger tendency to respond, andconsequently the stop-related N1 and P3 effects may begreater. Assuming that high impulsive individuals and slowstoppers already have more problems in inhibiting theirresponse tendency, a stop task where stop-signals are pre-sented less frequently may be a more sensitive test to detectpotential differences between high and low impulsives as wellas between slow and fast stoppers.

2. Methods

2.1. Participants

Participants were selected from a group of 435 psychol-ogy students (357 females and 78 males) that filled out theDutch version of the I7 questionnaire (Eysenck andEysenck, 1978; Lijffijt et al., 2005a). Seventeen students witha low impulsivity score (0 or 1) and 20 students with a highimpulsivity score (10–19) were tested. Three participants didnot complete the experiment and five participants wereexcluded due to technical problems. Finally, 14 studentswith low (3 males and 11 females) and 15 students with highimpulsivity scores (3 males and 12 females) participated in


the present study. Mean age was 20.71 (SD = 1.69) for thelow and 21.53 (SD = 1.96) for the high impulsive group.

Slow and fast stoppers were created by a median split ofthe mean stop-signal reaction time across the high- andlow-frequency stop tasks (SSRT; estimated as describedby Logan, 1994). Eight out of 14 participants with low I7

scores and six out of 15 participants with high I7 scoreswere categorized as slow stoppers. Mean age was 21.13(SD = 1.51) for fast stoppers (i.e. short SSRTs) and 21.14(SD = 2.21) for slow stoppers (i.e. long SSRTs).

Participants had neither a current neurological or psy-chological disorder nor a history of one and were not onpsychoactive medication. They reported to be right-handedand vision was normal or corrected-to-normal. They vol-unteered to participate in the study for course credit orcould earn monetary compensation. All participants signedinformed consent. The study was approved by the localEthical Committee of the Faculty of Social Sciences.

2.2. Stop-signal task

The stop-signal task involves two types of trials: go trialsand stop trials. Whereas go trials only contained go-sig-nals, stop trials contained go- and stop-signals. Go-signalswere square-wave, black-white vertical gratings of either ahigh (3.62 cycles per degree; cpd) or a low (0.46 cpd) funda-mental spatial frequency (size was 7.6� · 7.6�). The go-stimuli were presented in the center of the screen one byone on a gray background. Participants were required todiscriminate between the two gratings and press the correct(left or right) button. After the presentation of a fixationcross for 500 ms, a grating was presented for 750 ms. Thevariable time interval between the end of a grating andthe start of the fixation cross was 1000–1250 ms. Stop trialsconsisted of a go- and a stop-signal (a 1000 Hz tone,400 ms in duration, 83 dB intensity), generated by the com-puter and presented binaurally through earplugs. In stop-trials on which a grating was followed by a tone, theresponse to the grating had to be suppressed.

Two versions of the stop-signal task were administered.The percentage of stop-trials was 80% in the high-fre-quency stop task, and 20% in the low-frequency stop task.Each stop task contained 520 trials presented in 4 differentblocks of 130 trials. Blocks in the high-frequency stop taskconsisted of 104 stop- and 26 go-trials, and blocks in thelow-frequency stop task consisted of 26 stop- and 104 go-trials. Go–stop intervals were adjusted before each blockaccording to a tracking algorithm to yield a performanceof about 50% successful inhibitions (Pic) (De Jong et al.,1995; Logan et al., 1997).1 Before the go-stop interval forthe next block was estimated, the percentage of successful

1 In the original experiment, 34 participants performed the stop-signaltask. However, due to technical problems, the tracking algorithm did notalways yield an average percentage of 50% successful inhibition. In thepresent study, only participants with more than 15% or less than 85%percentage of successful inhibition were included.

stops was corrected for the amount of omissions on go-tri-als in the previous block (Tannock et al., 1989). To avoidstrategies induced by the predictability of the timing ofthe stop-signal and to reduce the confounding effect ofthe overlap between the ERPs elicited by go- and stop-sig-nals, the interval between the go-stimulus and the stop-sig-nal was jittered in a range of 250 ms surrounding thecalculated SOA (Pliszka et al., 2000) (i.e. 26 go–stop inter-vals ranging from �125 ms to 125 ms after the initialadjusted interval). Each of these intervals was used fourtimes in the blocks of the high-frequency stop task andonce in the blocks of the low-frequency stop task.

Gratings of a high and a low fundamental spatial fre-quency were equally divided across stop- and go-trials. Inhalf of the blocks, subjects had to react with the left fingerto the grating of the high fundamental spatial frequencyand with their right finger to the grating of the low funda-mental spatial frequency. For the remaining blocks, stimu-lus-response mapping was reversed. All trials were pseudo-randomized within blocks. Whereas in the high-frequencystop task never more than two succeeding go-trials werepresented without being followed by a stop-trial, in thelow-frequency stop task never more than two succeedingstop-trials were presented without being followed by ago-trial. The order of presentation of the two stop taskswas counterbalanced between subjects.

Before the participants performed the stop tasks, twopractice sessions were presented. In the first session, onlygo-trials (i.e. 30 gratings with narrow and 30 gratings withwide bars) were randomly presented and subjects had topress the correct button as quickly as possible. Second,before the participant performed each stop task, they prac-ticed one block of the stop task. In these practice blocks,the go-stop interval was 250 ms.

2.3. Electrophysiological recordings

EEG and EOG activities were recorded using an Elec-trocap with 58 tin electrodes referenced to the right mas-toid. The ground electrode was placed within the capbetween Fpz and Fz. Vertical electrooculogram (VEOG)was recorded from electrodes attached above and belowthe left eye and the horizontal electrooculogram (HEOG)from the outer canthi of both eyes. Electrode impedancewas kept below 5 kX. EEG and EOG were amplified witha Brain-Amp amplifier (Brain Products GmbH) with abandwidth of 0.04–100 Hz. The sampling rate was 500 Hz.

2.4. Procedure

In the laboratory, participants filled out various self-reportmeasures of impulsivity and performed three cognitive tasksin a dimly lit room: two stop-signal tasks that differedin the frequency of stop trials and an antisaccade task.2

2 Relations between the self-report scales have been reported elsewhere.


The tasks were presented while EEG was recorded. Theorder of the stop-signal and antisaccade task was counter-balanced across participants. This article presents anddiscusses the behavioral and psychophysiological datafrom the stop task.

2.5. Data analysis

2.5.1. Performance data

Mean reaction times (MRT) to go-trials (not followedby a stop-signal) and overall choice error rate (Per) wererecorded for both versions of the stop task (20% and80% stop trials). MRT was based on correct responsesbetween 150 ms and 1500 ms poststimulus. Overall choiceerror rate (in %) was calculated by dividing the numberof incorrect choice responses to go-trials by total numberof incorrect choice responses + number of correctresponses. The percentage of omissions (Pom) was calcu-lated by dividing the number of omissions by the numberof go-trials. Furthermore, the corrected percentage of inhi-bition (Pic) was calculated according to the proceduredescribed by Tannock et al. (1989). The average time inter-val between the grating and the stop-signal (SOA) was alsocalculated. The stop-signal reaction time (SSRT) was esti-mated as described by Logan (1994).

2.5.2. ERP analyses

EEG and EOG data were analyzed using Brain VisionAnalyzer software (1.05). EEG signals were rereferencedoff-line to the average of all electrodes and down-sampledto 250 Hz with a low-pass filter of 30 Hz. The continuousEEG data were segmented into epochs from �100 ms to1552 ms relative to the onset of both go- and stop-stimulus.Baseline for the ERPs was defined as the mean of the100 ms period prior to the go-stimulus. Trials with incor-rect responses or artifacts were rejected from further anal-yses (absolute amplitude criterion of 100 lV; low activitycriterion of 0.3 lV within a 50 ms time window) and ocularartifact correction was conducted according to the Grattonet al. (1983) algorithm. ADJAR level 2 (Woldorff, 1993)was used to remove the overlap between the go and stopERPs. Following the procedure of Bekker et al. (2005c),ADJAR-correction was performed from �100 to 700 ms,separately for successful and failed stops. Afterwards, a50 ms post-stimulus baseline correction was applied forstop-ERPs, to account for possible residual overlap distor-tion. For each subject and each stop task, average ERPswere computed from the corrected data, separately for suc-cessful and failed stops. ERP analyses focused on the N1and stop P3, and specifically on the contrast between suc-cessful and failed stops (here labeled as the N1 effect andstop P3 effect, respectively). Mean amplitude of individualaverage ERPs in the 80–120 ms time window was deter-mined for successful and failed stops, separately for thehigh- and low-frequency stop task at FCz and Cz to quan-tify the N1. Based on previous research (Bekker et al.,2005c) and after inspection of the grand-average wave-

forms and difference waves (Figs. 3, 4, and 7), we decidedto analyze the stop P3 by calculating the mean amplitudefor the 140–350 ms time window. The average number oftrials used for deriving ERPs ranged from 19 to 67 andfrom 27 to 290 for the low- and high-frequency stop task,respectively.

Average response-locked ERPs (r-ERPs) were calcu-lated, separately for failed stops and correct go-trials (onlyfor responses faster than 750 ms), to investigate whethererror-related processes contribute to the potential groupdifferences in the stop P3 effect. It should be noted thatADJAR level 2 could not be applied to response-lockedERPs. The ERN was defined as the largest negative peakin the 20–120 ms time window after response onset (peaklatency locked at FCz). The Pe was defined as the largestpositive peak in the 175–450 ms time window afterresponse onset (peak latency locked at Cz).

2.6. Statistical analyses

Fisher’s exact test was conducted to examine the associ-ation between I7 impulsivity and SSRT. Separate repeatedmeasures analyses of variance (ANOVAs) were conductedfor MRT, SSRT, Pic, Per, Pom, and SOA with probability(20% vs. 80% stop-trials) as within-subjects factor andimpulsivity (high vs. low impulsivity) as between-subjectsfactor. Similar ANOVAs were performed for SSRT (fastvs. slow stoppers) as between-subjects factor.

Regarding the ERP components, repeated-measuresANOVAs were conducted separately for the N1 and P3,including three within-factors, i.e. frequency (low vs.high-frequency of stop-signals), stop (failed vs. successfulstops), and lead (FCz vs. Cz) and one between-factor group(high vs. low impulsivity). Similar ANOVAs were con-ducted for the stop-related ERPs with SSRT as between-subjects factor (fast vs. slow stoppers).

Separate repeated measures ANOVAs were conductedfor ERN and Pe including lead (FCz and Cz for ERN;FCz, Cz, and Pz for Pe), frequency (high vs. low), and trial(failed stops vs. correct go-trial) as within-subjects factorsand group (high vs. low impulsivity) as between-subjectsfactor. Similar ANOVAs were performed with SSRT (orslow vs. fast stoppers) as between-subjects factor. Onlygroup main effects or interactions with group effects arereported.

3. Results

3.1. Behavioral data

Fisher’s exact test revealed no significant associationbetween I7 impulsivity and mean stop-signal reaction time(SSRT) across the high- and low-frequency stop task(p = 0.466). Hence, the complete group of 29 participantscould be used to create a slow and fast stop group byapplying a median split of SSRT.

0

200

400

600

high low slow fast

Impulsivity SSRTmean

MR

T (in

ms)

high freq Stop low freq Stop

Fig. 1. Mean reaction times (MRT) to go-signals by stop task version forparticipants scoring high and low on impulsivity (left panel) and for slowand fast stoppers (right panel). Error bars represent the standard error ofthe mean.


Performance data with regard to differences between thetwo impulsive groups have been described elsewhere (Lans-bergen et al., 2007) and will only be briefly reported here.Table 1 presents performance data for the high- and thelow-frequency stop task, for slow and fast stoppers. Figs. 1and 2 illustrate mean reaction time (MRT) and mean SSRTfor the high-and low-frequency stop task, separately for indi-viduals with high and low I7 impulsivity scores (left panels)and for slow and fast stoppers (right panels). In the low-fre-quency stop task, MRTs were faster (F(1,27) = 39.53,p < .001), standard deviations of the reaction time (SDRT)were smaller (F(1,27) = 31.03, p < .001), go-stop intervals(SOA) were shorter (F(1,27) = 34.91, p < .001), and percent-age of omissions (Pom) were smaller (F(1,27) = 16.23,p < .001), relative to the high-frequency stop task. Althoughthe tracking algorithm was used to yield a performance ofabout 50% successful inhibitions, the percentage of correctedinhibition (Pic) was smaller when stop-signals were pre-sented less frequently F(1, 27) = 18.70, p < .001). SSRTand choice-reaction time errors did not differ between thehigh- and low-frequency tasks.

No differences were found between participants with highand low self-reported impulsivity for the parameters.Whereas slow stoppers had longer SSRTs (F(1, 27) =27.86, p < .001), greater SDRTs (F(1, 27) = 9.92, p = .004),and made more choice-reaction time errors than fast stop-pers (F(1,27) = 10.85, p = .003), no group differences werefound for SOA, Pom, Pic, and choice reaction time. Further,no significant interaction between frequency and group wasfound.

3.2. Electrophysiological data

In the reported ERP results, the terms N1 and stop P3refer to the negativity around 100 ms and the positivityaround 200–300 ms after the presentation of the stop-sig-nal, respectively. The greater N1 elicited by successful rela-tive to failed stops will be labeled as the ‘N1 effect’. Theenhanced stop P3 to successful as compared to failed stopswill be labeled as the ‘stop P3 effect’.

Table 1Mean values and standard deviation of the performance data obtained fromstoppers

Fast stoppers

Low-frequency stop task High-frequency stop tas

Mean SD Mean SD

MRT (in ms) 388.3 64.0 514.8 106.0SSRT (in ms) 137.8 17.4 142.5 16.5SDRT 87.6 18.5 124.7 20.1SOA 236.8 59.4 311.1 80.2Pom 1.8 1.7 4.4 3.7Pic 48.5 6.4 62.2 12.6Per 2.7 1.9 2.7 2.6

Note: MRT = mean reaction time to go-stimulus; SSRT = stop-signal reactSOA = delay between go-stimulus and stop-stimulus Pom = percentage of omnumber of omissions on stop-trials; Per = percentage of choice errors.

3.2.1. High versus low I7 impulsive groupsFig. 3 presents grand average ERPs elicited by stop-sig-

nals at FCz, Cz, and Pz for successful and failed stops forhigh and low impulsive participants, separately for the low-and high-frequency stop task. Mean amplitudes and stan-dard deviations are summarized in Table 2.

3.2.1.1. N1. As illustrated by the ERPs in Fig. 3 and the dif-ference waves in Fig. 4 (left panel), the N1 elicited by suc-cessful stops was larger than the N1 elicited by failed stops(i.e. N1 effect; main effect of stop: F(1,27) = 12.61,p = 0.001). A marginally significant stop · lead · groupeffect (F(1,27) = 4.11, p = 0.053) revealed a stop · leadeffect for participants scoring high (F(1, 27) = 16.06,p < .001), but not for those scoring low on self-reportedimpulsivity. Post-hoc analyses for the high impulsive groupindicated an N1 effect at Cz, but not at FCz. Furthermore,an enhanced N1 was found for the low-frequency relativeto the high-frequency stop task, especially at Cz (frequency· lead: F(1, 27) = 32.43, p < .001; main effects of frequency:FCz: F(1, 27) = 12.27, p = .002; Cz: F(1,27) = 23.03,p < .001).

the high- and low-frequency stop-signal task, separately for fast and slow

Slow stoppers

k Low-frequency stop task High-frequency stop task

Mean SD Mean SD

413.6 69.6 527.2 148.4195.5 35.7 191.8 45.2119.8 23.2 149.4 46.8214.2 62.7 290.0 95.8

2.5 2.5 6.0 5.342.1 11.8 54.0 23.07.3 5.3 7.4 7.0

ion time; SDRT = standard deviation of reaction times to go-stimulus;issions; Pic = percentage of successful inhibitions, corrected for estimated

0

50

100

150

200

250

high low slow fast

Impulsivity SSRTmean

SSR

T (in

ms)

high freq Stop low freq Stop

Fig. 2. Stop-signal reaction times (SSRT) by stop task version forparticipants scoring high and low on impulsivity (left panel) and for slowand fast stoppers (right panel). Error bars represent the standard error ofthe mean.


3.2.1.2. Stop P3. Figs. 3 and 4 illustrate the enhanced stop-related P3 for successful relative to failed stops (i.e. stop P3effect). Statistical analyses revealed a trend for a stop ·group effect (F(1, 27) = 3.08, p = .090), indicating a largerstop P3 effect for high than for low impulsives(F(1, 27) = 39.39, p < .001; F(1, 27) = 13.12, p = .001,respectively; see Fig. 4, left panel). Group differences forsuccessful stops, F(1,27) = 3.64, p = .067, rather than forfailed stops, F < 1, might have contributed to the largerstop P3 effect for high as compared to low impulsive

FCz

-20

-15

-10

-50

5

10

15

20

25

μV

Cz

-100 0 100

Low frequency stop task

High I7, SuccessfulHigh I7, Failed

Pz

Fig. 3. Grand-average ERPs for participants scoring high and low on I7 impulsfor the high- (80% stop trials) and low- (20% stop trials) frequency stop task.

individuals. Further, a significant main effect of frequencyindicated a larger frontal P3 for the low- relative to thehigh-frequency stop task (F(1, 27) = 49.52, p < .001).

3.2.1.3. Error-related ERP components. Fig. 5 presents thegrand average response-locked ERPs at FCz, Cz, and Pzfor failed stops and correct go-trials for high and lowimpulsive participants, separately for the low- and high-fre-quency stop task. As illustrated in Fig. 5, the impulsivegroups did not differ significantly with respect to theERN or Pe amplitude (see also the difference waves inFig. 6, left panel).

3.2.2. Slow versus fast stoppers

Fig. 7 presents grand average ERPs elicited by stop-sig-nals at FCz, Cz, and Pz for successful and failed stops forfast and slow stoppers, separately for the low- and high-fre-quency stop task. Mean amplitudes and standard devia-tions are summarized in Table 3. Effects of frequency andlead were qualitatively the same as those in previous anal-yses that included impulsivity as between-subjects factor(i.e. greater amplitudes in the low- vs. high-frequency taskand greater amplitudes for successful vs. failed stops).

3.2.2.1. N1. With regard to the N1, a frequency · lead ·group effect (F(1.27) = 6.97, p = .014) was found. Post-hoc analyses per lead indicated a significantly larger N1

Cz

FCz

200 300 400 500ms

High frequency stop task

Low I7, SuccessfulLow I7, Failed

Pz

ivity, elicited with successful and failed stops at FCz, Cz, and Pz, separately

Table 2Means and standard deviations of the mean amplitudes of the N1 (80–120 ms), and stop P3 (120–160 ms and 140–350 ms) at Cz, elicited by successful andfailed stops, in the high- and low-frequency stop-signal task, separately for participants scoring high and low on impulsivity

High I7 Low I7

Low-frequency stop task High-frequency stop task Low-frequency stop task High-frequency stop task

Mean SD Mean SD Mean SD Mean SD

N1-SS (80–120 ms) �13.6 6.0 �8.6 1.9 �12.1 6.6 �8.3 3.4N1-FS (80–120 ms) �11.9 6.2 �7.7 3.3 �9.9 6.3 �7.1 3.2Stop P3-SS (120–160 ms) �6.7 5.2 0.1 3.4 �6.4 6.0 �1.2 2.4Stop P3-FS (120–160 ms) �9.9 7.8 �4.7 3.9 �7.6 6.6 �3.4 2.9Stop P3-SS (140–350 ms) 13.1 7.8 7.0 5.2 8.5 7.2 3.3 4.5Stop P3-FS (140–350 ms) 7.1 6.3 0.1 5.0 4.5 6.7 �0.4 3.0

Note: SS = successful stops; FS = failed stops.


in the low- than in the high-frequency stop task for fast,but not for slow stoppers at FCz and Cz (FCz: frequency· group: F(1,27) = 5.31, p = .029; Cz: frequency · group:F(1, 27) = 8.43, p = .007). Furthermore, a marginally sig-nificant stop · lead · group effect (F(1,27) = 4.20,p = .050) was decomposed using post-hoc analyses for eachgroup, revealing a stop · lead effect for fast(F(1,27) = 16.20, p < .001), but not for slow, stoppers; thisreflected an N1 effect for slow stoppers at both leads, butfor fast stoppers only at Cz and not at FCz.

3.2.2.2. Stop P3. Regarding the stop P3, both fast and slowstoppers showed a stop P3 effect, but it was larger for fast

-5

0

5

10

15

Cz

μV

FCz

High I7, low freqHigh I7, high freq

Low I7, low freqLow I7, high freq

m-100 0 100 2

Pz

Fig. 4. Difference waves (successful minus failed stops) for the high- (80% stopfor the impulsive (left panel) and SSRT groups (right panel).

than for slow stoppers (see Fig. 4 (right panel) and Fig. 7;stop · group: (F(1,27) = 8.70, p = .007; F(1, 27) = 57.11,p < .001, and F(1,27) = 10.24, p = .003 for fast and slowstoppers, respectively). Post-hoc analyses indicated thatthis enhanced stop P3 for fast relative to slow stopperswas significant for successful (F(1,27) = 26.24, p < .001)as well as for failed stops (F(1,27) = 8.12, p = .008.

3.2.2.3. Error-related ERP components. Fig. 8 presents thegrand average response-locked ERPs at FCz, Cz, and Pzfor failed stops and correct go-trials for slow and fast stop-pers, separately for the low- and high-frequency stop task.Statistical analyses for slow versus fast stoppers yielded no

FCz

Cz

Slow stop, low freqSlow stop, high freq

Fast stop, low freqFast stop, high freq

s00 300 400 500

Pz

trials) and low- (20% stop trials) frequency stop task at FCz, Cz, and Pz,

FCz

Cz

FCz

μV

Cz

Low frequency stop task High frequency stop task

High I7, FailedHigh I7, Go correct

Low I7, FailedLow I7, Go correct

Pz Pz

ms-200 -100 -0 100 200 300 400 500

15

-15

-10

-5

0

5

10

Fig. 5. Grand-average ERPs for participants scoring high and low on I7 impulsivity, elicited by failed stops and correct go-trials at FCz, Cz, and Pz,separately for the high- (80% stop trials) and low- (20% stop trials) frequency stop task.

FCz

Cz

FCz

μV

Cz

Pz Pz

ms-200 -100 -0 100 200 300 400 500

High I7, low freqHigh I7, high freq

Low I7, low freqLow I7, high freq

Slow stop, low freqSlow stop, high freq

Fast stop, low freqFast stop, high freq

20

-5

0

5

10

15

-10

Fig. 6. Difference waves (failed stops minus correct go-trials) for the high- (80% stop trials) and low- (20% stop trials) frequency stop task at FCz, Cz, andPz, for the impulsive (left panel) and SSRT groups (right panel).


FCz FCz

-20

-15

-10

-5

0

5

10

15

20

25

μV

Cz

-100ms


Pz Pz

Slow stop, SuccessfulSlow stop, Failed

Fast stop, SuccessfulFast stop, Failed

Cz

0 100 200 300 400 500

Fig. 7. Grand-average ERPs for slow and fast stoppers, elicited by successful and failed stops at FCz, Cz, and Pz, separately for the high- (80% stop trials)and low- (20% stop trials) frequency stop task.


group or interaction with group effects for ERN. Regard-ing the Pe amplitude, a significant frequency · trial · groupwas found (F(1, 27) = 10.36, p = .003). Post-hoc analysesrevealed a significant trial · group effect for the low(F(1,27) = 8.54, p = .007) but not the high frequency stoptask version, indicating that the enhanced Pe amplitudefor failed stops versus correct go-trials was greater for fastthan for slow stoppers. (F(1,27) = 81.24, p < .001 andF(1, 27) = 21.56, p < .001 for fast and slow stoppers,respectively).

3.2.3. Post-hoc analyses in 120–160 ms time window

Differences in stop P3 effects were accompanied bySSRT differences for slow versus fast stoppers, but not

Table 3Means and standard deviations of the mean amplitudes of the N1 (80–120 ms),failed stops, in the high- and low-frequency stop-signal task, separately for fa

Fast stoppers

Low-frequency stop task High-frequency s

Mean SD Mean

N1-SS (80–120 ms) �15.2 5.8 �9.1N1-FS (80–120 ms) �13.6 5.3 �7.5Stop P3-SS (120–160 ms) �6.2 4.6 0.8Stop P3-FS (120–160 ms) �11.2 6.5 �3.1Stop P3-SS (140–350 ms) 15.5 6.3 8.6Stop P3-FS (140–350 ms) 8.5 6 1.3

Note: SS = successful stops; FS = failed stops.

for high versus low impulsives. This combined pattern ofresults makes a unitary interpretation of the stop P3 effectas a manifestation of an inhibitory control process unat-tractive. Based on visual inspection (Figs. 3, 4, and 7), weformulated the hypothesis that the early phase of the stopP3 effect was different for slow versus fast, but not for highversus low impulsives. Such a result would be consistentwith the notion that this early phase reflects inhibition,whereas the later phase would reflect more strategic pro-cesses like error monitoring. For the 120–160 ms time win-dow (the early phase of the stop P3), we found nosignificant group · stop interaction for high versus lowimpulsives. For slow versus fast stoppers, a significant fre-quency · stop · group effect (F(1, 27) = 7.57, p < .010)

and stop P3 (120–160 ms and 140–350 ms) at Cz, elicited by successful andst and slow stoppers

Slow stoppers

top task Low-frequency stop task High-frequency stop task

SD Mean SD Mean SD

1.9 �10.4 6.0 �7.7 3.32.9 �8.2 6.0 �7.3 3.63.3 �6.9 6.5 �1.9 1.92.8 �6.2 7.2 �5.2 3.84.5 5.9 5.9 1.5 2.74.6 3 8.5 �1.7 2.8

FCz

Cz

FCz

μV

Cz


Slow stop, FailedSlow stop, Go correct

Fast stop, FailedFast stop, Go correct

Pz Pz

ms-200 -100 -0 100 200 300 400 500

15

-15

-10

-5

0

5

10

Fig. 8. Grand-average ERPs for slow and fast stoppers, elicited by failed stops and correct go-trials at FCz, Cz, and Pz, separately for the high- (80% stoptrials) and low- (20% stop trials) frequency stop task.


reflected a significant stop · group effect for the low, butnot the high-frequency task version (F(1,27) = 11.22,p = .002). As illustrated in Fig. 4 (right panel), the stopP3 effect in the low-frequency task was significant onlyfor fast and not for slow stoppers (F(1, 27) = 16.20,p < .001; F < 1, respectively).

4. Discussion

The present study examined the relation between self-report measures of impulsivity and stopping performanceas well as the underlying neural mechanisms. Stopping per-formance was assessed using two stop-signal tasks that dif-fered in the probability of stop-signals. Consistent withprevious findings (Ramautar et al., 2004), participants inthe present study developed an impulsive response stylein the stop-signal task with a low probability of stop-sig-nals, with faster reaction times to go-trials and a higherpercentage of failed stops. So, in a stop task where stop-sig-nals are presented less frequently, speed of processing thego-signal is traded for the probability of successful inhibi-tion. Stop-signal reaction time (SSRT) was not affectedby probability of stop-signals, suggesting that SSRT maynot be influenced by strategic processes.

No differences were found in behavioral performancebetween individuals scoring low and scoring high on theI7 impulsiveness subscale, not even in the low-frequency

stop task where an impulsive response style might haveaffected stopping performance more in participants scoringhigh on impulsivity. Together with the inconsistent findingsof previous behavioral studies (e.g. Lijffijt et al., 2004;Logan et al., 1997), it seems that there is no clear relationbetween impulsivity, as assessed by self-report measures,and stopping performance in the stop-signal task. Consis-tently, previous studies did not provide evidence for anassociation between self-reported impulsivity in healthyindividuals and other related forms of inhibitory control,such as stopping a response in a continuous performancetask (Fallgatter and Herrmann, 2001; Horn et al., 2003)or interference control as quantified in the Stroop test(Avila and Parcet, 2001; Cheung et al., 2004; Kirkebyand Robinson, 2005; Visser et al., 1996).

With respect to event-related potentials, as expected,successful stops were associated with greater negativityaround 80–120 ms and subsequently greater positivitybeginning around 120 ms after the onset of the stop-signal,relative to failed stops (i.e. N1 and stop P3 effect, respec-tively). As discussed in the Introduction, the N1 effectmay reflect a functional link between auditory-cortex acti-vation and motor inhibition, and the P3 effect may bethought to reflect an inhibitory mechanism and/or errorprocessing. Given the absence of behavioral differencesbetween participants scoring high and low on self-reportedimpulsivity, the finding of similar N1 effects is not


surprising. In contrast, Dimoska and Johnstone (2007)reported a larger N1 elicited by successful stops forindividuals scoring high relative to low on the I7 impulsive-ness subscale. They interpreted this as more attention tothe stop-signal due to more sensation seeking in high versuslow impulsives. Inconsistency between the findings of thepresent study and Dimoska and Johnstone (2007) may beexplained by the difference in intensity of the stop tone(60 dB vs. 83 dB in Dimoska and in our study, respectively).A positive relation between N1 amplitude and intensity of astimulus (i.e. augmenting dimension of sensory perfor-mance) in individuals characterized by high self-reportedimpulsivity or more sensation seeking behavior has beenexplained as reflecting searching for an optimal level ofstimulation for cognitive processes (Carrillo-de-la-Penaand Barrat, 1993). In this perspective, a stop tone of 83 dBsound pressure level may already be an optimal level of stim-ulation in contrast to a tone of 60 dB. Note that the ADHDstudy of Bekker et al. (2005c) used a stop tone of 80 dB.

Whereas the enhanced stop P3 for successful relative tofailed stops was present in both groups, it was attenuatedin participants scoring low, relative to high on self-reportedimpulsivity. Post-hoc analyses for each stop-trial indicatedthat probably the enhanced stop P3 elicited by successfulin high relative to low impulsives, but not the stop P3 forfailed stops, contributed to this effect, confirming the resultsof Dimoska and Johnstone (2007). In individuals withADHD, it has consistently been found that the stop P3 effectis substantially reduced compared to controls (see Introduc-tion). The reduced stop P3 effect in ADHD has always beenreported in combination with deficient stopping perfor-mance (i.e. longer SSRTs). In the present study, stoppingperformance did not differ between individuals scoring lowand high on self-reported impulsivity. To the extent thatthe stop P3 effect reflects the activation of an inhibition sys-tem, these results suggest that participants scoring high onself-reported impulsivity needed more inhibitory effort forequal stopping performance as compared to those scoringlow on self-reported impulsivity (see also Dimoska andJohnstone, 2007). However, as discussed in the Introduction,the stop P3 effect may reflect both inhibition as well as error-related processing (Dimoska et al., 2006; Ramautar et al.,2004). As shown in Figs. 3 and 7, the ERN and Pe may over-lap the stop P3 elicited by failed stops.

It can be speculated that the earlier phase of the stop P3effect (before 200 ms-latency) reflects inhibition, and thelater phase error-related processing. Accordingly, similarstopping performance in high and low impulsives wasaccompanied by differences in the later error-related phase,but not in the earlier inhibition phase. However, the impul-sive groups did not differ with respect to the ERN and Peamplitudes. For the time being, it is difficult to decidebetween these two alternatives, and we conclude that inhi-bition, rather than error processing, is different betweenindividuals with high and low self-reported impulsivity(although in a manner opposite to what was expected).In the stop-signal task, participants scoring high on

self-reported impulsivity may need more inhibitory effortfor equal stopping performance.

The second aim of the present study concerned the neu-ral basis of individual differences in stopping performance.There was no association between the efficiency of stopping(as indexed by stopping reaction time) and I7 impulsivity,justifying the division of the same sample into slow and faststoppers. Slow stoppers also displayed more variable reac-tion times to go-signals and larger percentages of choicereaction time errors at the behavioral level. These resultssuggest lapses in attention (Castellanos and Tannock,2002). However, larger percentage of choice reaction timeserrors in slow stoppers may also indicate a pattern ofimpulsive responding (Bekker et al., 2005b). In terms ofthe speed-accuracy trade-off, a larger percentage of choicereaction time errors without an increase in go RT mayreflect favoring speed over accuracy, indicating a fast andimpulsive response style. The dissociation between SSRTand go RT (no difference between slow and fast stoppers)is consistent with a recent report by Dimoska et al.(2006), who found no differences between fast and slowgroups (median split of reaction time to go-signal) forstop-signal reaction time.

Contrary to our expectation, no differences were foundbetween slow and fast stoppers with respect to theenhanced N1 for successful relative to failed stops. Consis-tent with our prediction, the increased stop P3 for success-ful relative to failed stops was attenuated in slow stoppersfor the earlier (120–160 ms) as well as for the later phase(140–350 ms). These results suggest that in healthy individ-uals, impaired stopping performance is associated withweaker activation of the inhibition system (earlier phase)as well as less error processing (later phase), but the linkbetween the amount of attention paid to the stop-signaland the probability of successful stopping, as reflected inthe N1 effect, is not impaired.

Further, the contrast between fast and slow stopperswith regard to the stop P3 effect concerned a decreased stopP3 to successful as well as to failed ones. Assuming that thestop P3 elicited by successful stops reflects inhibitory con-trol and the stop P3 elicited by failed stops reflects error-related processes (Dimoska et al., 2006; Ramautar et al.,2004), it may be concluded that differences between fastand slow stoppers refer to both inhibition and error-relatedprocessing. Comparing the error-related ERPs between thegroups, no significant association between stopping perfor-mance and ERN, but a smaller Pe was found for slow stop-pers relative to fast stoppers. In conclusion, presentfindings suggest that slow stoppers show deficient inhibi-tory control (as reflected in the stop P3 to successful stops),‘normal’ error monitoring (ERN), but are less aware ofconflict or engage less attentional control after conflict(Pe; stop P3 to failed stops). This pattern of ERP resultsonly partly resembles the abnormal brain correlates ofstopping performance in ADHD.

Note that the slow stoppers had similar stop P3 effectsas the fast stoppers in the early time window (i.e.


120–160 ms) in the high-, but not in the low-frequency task(see Fig. 4; right panel). However, as illustrated in Fig. 4(right panel), after approximately 180 ms the stop P3 effectwas smaller in slow as compared to fast stoppers in bothfrequency versions.

The global frequency of stop-signals had an affect onprocessing the stop-signal in general. The N1 and stopP3 for both successful and failed stops were larger inthe low than in the high-frequency stop task. The N1is known to be sensitive to the global temporal probabil-ity, rareness of the evoking stimulus (Davis et al., 1966),or the intensity of the stimulus. Rare auditory stimuliand relatively intense auditory stimuli may evoke largeN1s, which may reflect a larger impact of these stimuliin auditory cortex, or stated differently, more attentionto these stimuli; the N1 can also be modulated by volun-tary attention (Woldorff et al., 1993). Furthermore, theN1 is larger when participants are engaged in a taskrather than relaxing (Naatanen, 1992). The enhancedN1 in the low- versus high-frequency task was muchmore pronounced and significant only for fast as com-pared to slow stoppers, suggesting an increase in sensorysensitivity to the auditory stop-signal in the low as com-pared to the high-frequency task, for fast, but not forslow, stoppers. However, it is not completely straightfor-ward, as no interaction effects of group and frequencywere found on task performance. A similar line of rea-soning may apply to the larger stop P3 to low-frequencystop-signals observed specifically in fast stoppers. Thisfrequency effect on the stop P3 may be viewed as asequel of the N1 interaction, at the level of an inhibitorymechanism. However, the enhanced P3 to rare stop-sig-nals may also reflect aspects of both the classical, fron-tal-centrally distributed P3a, related to orienting(Courchesne et al., 1975), as well as the classical, cen-tral-parietally distributed P3b, which responds especiallyto oddball stimuli (Donchin, 1981).

In summary, prototypical stopping deficits as repeat-edly reported for ADHD could not be confirmed for ques-tionnaire-based high impulsivity in healthy participants.Even stronger, at the level of neurophysiological correlates,healthy highly impulsives exhibit a pattern opposite to thatusually observed for ADHD. It may be speculated thatindividuals with relatively high self-reported impulsivityneed more inhibitory effort to attain normal stopping lev-els, rather than having a stopping deficit. Our findings sug-gest that the underlying mechanisms of low impulse controlin pathological samples (as in ADHD) are qualitatively dif-ferent from that in healthy, highly impulsive individuals.Therefore, self-reported impulsivity within the normal pop-ulation may not function as a model for impulsivity in psy-chopathology. So, conclusions about, for example, thegenetics, the psychopharmacology, and the neurophysio-logical basis of inhibition that are based on comparingeither high versus low subjective impulsives or slow versusfast stoppers cannot be generalized to ADHD, or viceversa.

With respect to stopping performance, stopping perfor-mance in ADHD may be an extreme form, but also quali-tatively different from healthy individuals. Whereas theneural mechanism underlying the implementation of inhib-itory control and/or error processing in ADHD patientsmay partly be the same as that in healthy volunteers, char-acterized by poor stopping performance, individuals withADHD have an additional deficit, that is, the lack of a linkbetween the amount of attention paid to the stop-signaland the outcome of the inhibition process. It is speculatedthat this discrepancy may be due to the additional attentiondeficits in ADHD patients, which may not be present in theslow stoppers in the present study.

Concluding, although we did not directly comparehealthy individuals with ADHD patients, the findingsof the present study in combination with a previousADHD study (Bekker et al., 2005c) suggest at least threegroups that can be differentiated according to stoppingbehavior in the stop-signal task and the neural correlatesof stopping. The first group involves healthy participantsscoring high relative to low on subjective self-reportimpulsivity. They had no deficiency in stopping behavior(similar speed of stopping), showed a normal relationbetween the amount of attention switched to the stop-sig-nal and subsequent stopping performance (similar N1effect), showed normal error processing (similar ERNand Pe), but may be less efficient in implementing theinhibition process (increased stop P3 effect). A secondgroup of healthy individuals was characterized by slowrelative to fast stop reaction times. They had a normallink between the amount of attention switched to thestop-signal and the probability of successful stopping(similar N1 effect), but may have weaker inhibition pro-cesses (reduced stop P3 effect) and may be less awareof conflict and/or engage less attentional control afterconflict (reduced Pe). Third, Bekker et al. (2005a)reported on adults with ADHD (ADHD combined sub-type), who were slow in stopping, showed a compromisedlink between attentional switching to the stop-signal andsubsequent stopping (no N1 effect) as well as weakerinhibition processes (reduced stop P3 effect). Furtherresearch including an ADHD group, a healthy highlyimpulsive group, and a group with relatively slow stop-pers is warranted to determine the exact differencesbetween these groups with respect to stopping perfor-mance and the neural correlates of stopping.

We conclude that impulsivity in ADHD is qualita-tively different from impulsive behavior within the nor-mal population, as assessed by self-report measures.Healthy, but relatively deficient, stoppers only partlyshow the same abnormal neurophysiological correlatesof stopping that characterize ADHD. Consequently, theneural mechanisms underlying impulsivity in ADHDshould not be investigated in healthy highly impulsivevolunteers or in healthy volunteers, characterized bypoor stopping, and then extrapolated to an ADHDpopulation.


Acknowledgements

This study was supported by a grant from the DutchOrganization for Scientific Research (NWO: 425-20-302).We thank Durk Talsma for the opportunity to use hisimplementation of ADJAR.

References

American Psychiatric Association. Diagnostic and statistical manual ofmental disorders. 4th ed. Washington; 1994.

Avila C, Parcet MA. Personality and inhibitory deficits in the stop-signaltask: the mediating role of Gray’s anxiety and impulsivity. Pers IndDiffer 2001;31(6):975–86.

Barkley RA. ADHD and the nature of self-control. New York: Guil-ford; 1997.

Barkley RA, Fischer M, Smallish L, Fletcher K. The persistence ofattention-deficit/hyperactivity disorder into young adulthood as afunction of reporting source and definition of disorder. J AbnormPsychol 2002;111(2):279–89.

Barrat ES, Pritchard WS, Faulk DM, Brandt ME. The relationship betweenimpulsiveness subtraits, trait anxiety, and visual N100 augmentintg/reducing: a topographic analysis. Pers Ind Differ 1987;8:43–51.

Bekker EM, Kenemans JL, Hoeksma MR, Talsma D, Verbaten MN. The pureelectrophysiology of stopping. Int J Psychophysiol 2005a;55(2):191–8.

Bekker EM, Overtoom CC, Kenemans JL, Kooij JJ, De Noord I,Buitelaar JK, et al. Stopping and changing in adults with ADHD.Psychol Med 2005b;35(6):807–16.

Bekker EM, Overtoom CC, Kooij JJ, Buitelaar JK, Verbaten MN,Kenemans JL. Disentangling deficits in adults with attention-deficit/hyperactivity disorder. Arch Gen Psychiatry2005c;62(10):1129–36.

Carrillo-de-la-Pena MT. ERP augmenting/reducing and sensation seek-ing: a critical review. Int J Psychophysiol 1992;12(3):211–20.

Carrillo-de-la-Pena MT. One-year test-retest reliability of auditory evokedpotentials (AEPs) to tones of increasing intensity. Psychophysiology2001;38(3):417–24.

Carrillo-de-la-Pena MT, Barrat ES. Impulsivity and ERP augmenting/reducing. Pers Ind Differ 1993;15(1):25–32.

Castellanos FX, Tannock R. Neuroscience of attention-deficit/hyperac-tivity disorder: the search for endophenotypes. Nat Rev Neurosci2002;3(8):617–28.

Cheung AM, Mitsis EM, Halperin JM. The relationship of behavioralinhibition to executive functions in young adults. J Clin ExpNeuropsychol 2004;26(3):393–404.

Courchesne E, Hillyard SA, Galambos R. Stimulus novelty, taskrelevance, and the visual evoked potential in man. Electroenceph ClinNeurophysiol 1975;39(2):131–43.

Davis H, Mast T, Yoshie N, Zerlin S. The slow response of the humancortex to auditory stimuli: recovery process. Electroenceph ClinNeurophysiol 1966;21(2):105–13.

De Jong R, Coles MG, Logan GD, Gratton G. In search of the point ofno return: the control of response processes. J Exp Psychol Human1990;16(1):164–82.

De Jong R, Coles MGH, Logan GD. Strategies and mechanisms innonselective and selective inhibitory motor control. J Exp PsycholHuman 1995;21(3):498–511.

de Wit H, Crean J, Richards JB. Effects of d-amphetamine and ethanol ona measure of behavioral inhibition in humans. Behav Neurosci2000;114(4):830–7.

de Wit H, Enggasser JL, Richards JB. Acute administration ofd-amphetamine decreases impulsivity in healthy volunteers. Neuro-psychopharmacology 2002;27(5):813–25.

Dimoska A, Johnstone SJ, Barry RJ. The auditory-evoked N2 and P3components in the stop-signal task: indices of inhibition, response-conflict, or error-detection? Brain Cogn 2006;62(2):98–112.

Dimoska A, Johnstone SJ. Neural mechanisms underlying trait impulsivityin non-clinical adults: stop-signal performance and event-related poten-tials. Prog Neuropsychopharmacol Biol Psychiatry 2007;31(2):443–54.

Donchin E. Surprise! surprise? Psychophysiology 1981;18:493–513.Eysenck SB, Eysenck HJ. Impulsiveness and venturesomeness: their

position in a dimensional system of personality description. PsycholReports 1978;43(3 Pt 2):1247–55.

Falkenstein M, Hohnsbein J, Hoormann J, Blanke L. Effects ofcrossmodal divided attention on late ERP components: II. Errorprocessing in choice reaction tasks. Electroenceph Clin Neurophysiol1991;78(6):447–55.

Fallgatter AJ, Herrmann MJ. Electrophysiological assessment of impulsivebehavior in healthy subjects. Neuropsychologia 2001;39(3):328–33.

Gehring WJ, Knight RT. Prefrontal-cingulate interactions in actionmonitoring. Nat Neurosci 2000;3(5):516–20.

Gratton G, Coles MG, Donchin E. A new method for off-line removal ofocular artifact. Electroenceph Clin Neurophysiol 1983;55(4):468–84.

Horn NR, Dolan M, Elliott R, Deakin JF, Woodruff PW. Responseinhibition and impulsivity: an fMRI study. Neuropsychologia2003;41(14):1959–66.

Kenemans JL, Bekker EM, Lijffijt M, Overtoom CC, Jonkman LM,Verbaten MN. Attention deficit and impulsivity: selecting, shifting,and stopping. Int J Psychophysiol 2005;58(1):59–70.

Kirkeby BS, Robinson MD. Impulsive behavior and stimulus-responsevariability in choice reaction time. J Res Pers 2005;39(2):263–77.

Lansbergen MM, Schutter DJLG, Kenemans JL. Subjective impulsivityand baseline EEG in relation to stopping performance. Brain Res2007;1148:161.

Lijffijt M, Bekker EM, Quik EH, Bakker J, Kenemans JL, Verbaten MN.Differences between low and high trait impulsivity are not associatedwith differences in inhibitory motor control. J Atten Disord2004;8(1):25–32.

Lijffijt M, Caci H, Kenemans JL. Validation of the Dutch translation ofthe I7 questionnaire. Pers Ind Differ 2005a;38(5):1123–33.

Lijffijt M, Kenemans JL, Verbaten MN, van Engeland H. A meta-analyticreview of stopping performance in attention-deficit/hyperactivitydisorder: deficient inhibitory motor control? J Abnorm Psychol2005b;114(2):216–22.

Liotti M, Pliszka SR, Perez R, Kothmann D, Woldorff MG. Abnormalbrain activity related to performance monitoring and error detection inchildren with ADHD. Cortex 2005;41(3):377–88.

Logan GD. On the ability to inhibit thought and action: a user’s guide tothe stop-signal paradigm. In: Dagenbach D, Carr TH, editors.Inhibitory processes in attention, memory, and language. San Diego,CA: Academic Press; 1994. p. 189–239.

Logan GD, Schachar RJ, Tannock R. Impulsivity and inhibitory control.Psychol Sci 1997;8:60–4.

Marsh DM, Dougherty DM, Mathias CW, Moeller FG, Hicks LR.Comparisons of women with high and low trait impulsivity usingbehavioral models of response-disinhibition and reward-choice. PersInd Differ 2002;33(8):1291–310.

Naatanen R. Attention and brain function. Hillsdale, NJ: Erlbaum; 1992.Oosterlaan J, Logan GD, Sergeant JA. Response inhibition in AD/HD,

CD, comorbid AD/HD + CD, anxious, and control children: a meta-analysis of studies with the stop task. J Child Psychol Psychiatry1998;39(3):411–25.

Overbeek TJM, Nieuwenhuis S, Ridderinkhof KR. Dissociable compo-nents of error processing: On the functional signifance of the Pe vis-a-vis the ERN/Ne. J Psychophysiol 2005;19(4):319–29.

Overtoom CC, Kenemans JL, Verbaten MN, Kemner C, van der MolenMW, van Engeland H, et al. Inhibition in children with attention-deficit/hyperactivity disorder: a psychophysiological study of the stoptask. Biol Psychiatry 2002;51(8):668–76.

Pliszka SR, Liotti M, Woldorff MG. Inhibitory control in childrenwith attention-deficit/hyperactivity disorder: event-related potentials


identify the processing component and timing of an impaired right-frontal response-inhibition mechanism. Biol Psychiatry 2000;48(3):238–46.

Ramautar JR, Kok A, Ridderinkhof KR. Effects of stop-signal probabil-ity in the stop-signal paradigm: the N2/P3 complex further validated.Brain Cogn 2004;56(2):234–52.

Ramautar JR, Kok A, Ridderinkhof KR. Effects of stop-signal modalityon the N2/P3 complex elicited in the stop-signal paradigm. BiolPsychol 2006a;72(1):96–109.

Ramautar JR, Slagter HA, Kok A, Ridderinkhof KR. Probability effectsin the stop-signal paradigm: the insula and the significance of failedinhibition. Brain Res 2006b;1105(1):143–54.

Reynolds B, Ortengren A, Richards JB, de Wit H. Dimensions ofimpulsive behavior: personality and behavioral measures. Pers IndDiffer 2006;40(2):305–15.

Rodriguez-Fornells A, Lorenzo-Seva U, Andres-Pueyo A. Are high-impulsive and high risk-taking people more motor disinhibited in thepresence of incentive? Pers Ind Differ 2002;32(4):661–83.

Schmajuk M, Liotti M, Busse L, Woldorff MG. Electrophysiologicalactivity underlying inhibitory control processes in normal adults.Neuropsychologia 2006;44(3):384–95.

Stahl J, Gibbons H. Dynamics of response-conflict monitoring andindividual differences in response control and behavioral control: anelectrophysiological investigation using a stop-signal task. Clin Neu-rophysiol 2007;118(3):581–96.

Tannock R, Schachar RJ, Carr RP, Chajczyk D, Logan GD. Effects ofmethylphenidate on inhibitory control in hyperactive children. JAbnorm Child Psychol 1989;17(5):473–91.

Ullsperger M, von Cramon DY. How does error correction differ fromerror signaling? An event-related potential study. Brain Res2006;1105(1):102–9.

van Boxtel GJ, van der Molen MW, Jennings JR, Brunia CH. Apsychophysiological analysis of inhibitory motor control in the stop-signal paradigm. Biol Psychol 2001;58(3):229–62.

Vigil-Colet A, Codorniu-Raga MJ. Aggression and inhibition deficits, therole of functional and dysfunctional impulsivity. Pers Ind Differ2004;37(7):1431–40.

Visser M, Das-Smaal E, Kwakman H. Impulsivity and negative priming:evidence for diminished cognitive inhibition in impulsive children. Br JPsychol 1996;87(Pt 1):131–40.

Wang W, Mei XF, Du L, Lu SW, Fu XM, Wang YH. Personalitycorrelates of auditory augmenting response to clicks repeated around2 Hz. J Neural Transm 1999;106(5-6):559–68.

Woldorff MG. Distortion of ERP averages due to overlap fromtemporally adjacent ERPs: analysis and correction. Psychophysiology1993;30(1):98–119.

Woldorff MG, Gallen CC, Hampson SA, Hillyard SA, Pantev C, Sobel D,et al. Modulation of early sensory processing in human auditorycortex during auditory selective attention. Proc Natl Acad Sci USA1993;90(18):8722–6.

Documents

Neural correlates of stopping and self-reported impulsivity · Neural correlates of stopping and self-reported impulsivity Marieke M. Lansbergen a,*, Koen B.E. Bo¨cker a, Evelijne