www.elsevier.com/locate/ynimg

NeuroImage 33 (2006) 399–405

Anterior cingulate and prefrontal cortex activity in an FMRI study oftrial-to-trial adjustments on the Simon task

John G. Kerns⁎

Department of Psychological Sciences, 214 McAlester Hall, University of Missouri, Columbia, MO 65211, USA

Received 21 April 2006; revised 30 May 2006; accepted 4 June 2006Available online 28 July 2006

People alter their task performance on a trial-to-trial basis, forexample after an incongruent trial on tasks involving response conflict.Previous research has found that these adjustments are most robust inthe Simon task. One explanation for behavioral adjustments is theconflict-monitoring hypothesis, which posits that the dorsal anteriorcingulate cortex (ACC) responds to conflict and that this serves as asignal to recruit other brain regions such as the prefrontal cortex(PFC) to minimize conflict and improve performance. However,another independently supported explanation for behavioral adjust-ments on the Simon task is the feature integration view, which canaccount for behavioral adjustments as the result of stimulus repetitionsand alternations. Hence, by itself, evidence for behavioral adjustmentson the Simon task does not clearly provide evidence for the conflict-monitoring hypothesis. However, the conflict-monitoring hypothesisdoes predict that behavioral adjustments on the Simon task shouldinvolve ACC conflict activity and PFC post-conflict activity. In thecurrent study, consistent with the conflict-monitoring hypothesis,behavioral adjustments in performance on the Simon task werepredicted by ACC conflict-related activity. In addition, subsequentbehavioral adjustments were associated with PFC activity, withprevious trial ACC conflict-related activity predicting greater PFCactivity on subsequent trials. These results provide additional evidencethat behavioral adjustments on the Simon task are due in part to ACCconflict monitoring and the subsequent recruitment of PFC to minimizeconflict.© 2006 Elsevier Inc. All rights reserved.

Keywords: Anterior cingulate; Prefrontal cortex; Response conflict; Post-conflict adjustments; Cognitive control

Introduction

People adjust their ongoing performance on behavioral tasks ina number of ways, such as slowing down and being more accurateafter errors (Rabbit, 1966; Laming, 1979). In addition, on tasks

E-mail address: KernsJ@missouri.edu.Available online on ScienceDirect (www.sciencedirect.com).

⁎ Fax: +1 573 882 7710.

1053-8119/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.neuroimage.2006.06.012

involving response conflict (i.e., the simultaneous activation of atleast two different responses), after incongruent high conflict trialsparticipants also adjust their performance (Gratton et al., 1992). Forexample, a response conflict task such as the Stroop color namingtask can involve high conflict incongruent trials (i.e., the wordRED in green ink, correct response is green) and low conflictcongruent trials (i.e., the word GREEN in green ink). After a highconflict incongruent trial, responses are slower for a subsequentcongruent trial and faster for a subsequent incongruent trial.Specifically, on congruent trials responses are slower if precededby an incongruent trial (i.e., iC trials) than if preceded by acongruent trial (i.e., cC trials). Moreover, participants are faster ifan incongruent trial was preceded by an incongruent trial (i.e., iItrials) than if preceded by a congruent trial (i.e., cI trials). Hence,after conflict participants adjust their performance, respondingslower for a subsequent congruent trial and faster for a subsequentincongruent trial.

One explanation for these post-conflict adjustments is theconflict-monitoring hypothesis (Botvinick et al., 2001, 2004). Thishypothesis posits that at least two brain regions play distinct andcomplementary roles in behavioral adjustments. This hypothesisstates that the dorsal anterior cingulate cortex (ACC) is activatedby the occurrence of response conflict (Carter et al., 1998;Botvinick et al., 1999). The monitoring of response conflict by theACC then serves as a signal that results in the recruitment of otherbrain regions to minimize the amount of conflict on subsequentperformance (Botvinick et al., 2001; Cohen et al., 2000; Kerns etal., 2004). The prefrontal cortex (PFC) is one brain region thoughtto be engaged after the occurrence of conflict to subsequentlyminimize conflict (by providing a contextually or task-setappropriate biasing signal to poster regions; Botvinick et al.,2001; Miller and Cohen, 2001). Hence, from the conflict-monitoring view, ACC conflict monitoring serves as a signal thatresults in the recruitment of PFC to minimize subsequent conflictand improve performance.

However, in addition to the conflict-monitoring hypothesis,there are other explanations for post-conflict behavioral adjust-ments, such as the feature integration view. From this perspective,behavioral adjustments are the result of exact stimulus repetitionsand alternations (Hommel et al., 2004). For example, consider the

400 J.G. Kerns / NeuroImage 33 (2006) 399–405

Simon task (Simon and Small, 1969; Lu and Proctor, 1995) onwhich two stimulus dimensions can vary from trial to trial:stimulus location (e.g., right or left) and stimulus identity (e.g., redor green object, necessitating either a right or left response,respectively). On the Simon task, only cC and iI trials involveexact stimulus repetitions (e.g., a green object on the left on twoconsecutive trials). Given that the post-conflict adjustment effectinvolves cC trials being faster than iC trials and iI trials being fasterthan cI trials, then faster responses on exact stimulus repetitions foronly cC and iI trials can account for some evidence of post-conflictbehavioral adjustments (Mayr et al., 2003). But in addition, on allother cC and iI trials not involving an exact stimulus repetition,they involve an exact stimulus alternation (e.g., green on leftfollowed by red on right). It has been argued that people processperceptual stimuli as integrated event files (integration of bothstimulus and response features, i.e., as object on left necessitating aright-hand response; Hommel et al., 2001; Stoet and Hommel,1999). If both stimulus dimensions change, this completealternation can be used to indicate a faster response. However,when only one stimulus dimension changes (i.e., red but still onleft), RT has been found to increase because of the partial match tothe previous trial. Hence, from this feature integration view,behavioral adjustments on the Simon task can be accounted for bycC and iI trials always involving either exact stimulus repetitions oralternations (thereby decreasing RT) whereas iC and cI trialsalways involve a partial match with the previous trial’s stimuli(thereby increasing RT). Importantly, evidence for the featureintegration view has been found in situations without responseconflict (Hommel et al., 2004). This suggests that evidence ofbehavioral adjustments on some response conflict tasks may notnecessarily indicate conflict monitoring.

Some research that does support a role for conflict monitoringin behavioral adjustments comes from brain imaging. The conflict-monitoring hypothesis predicts that ACC conflict-related activityshould predict adjustments in performance. Moreover, it predictsthat during post-conflict adjustment trials that PFC should beactive. Consistent with this, in a study using the Stroop task, it wasfound that ACC conflict activity predicted greater post-conflictadjustments (Kerns et al., 2004). At the same time, greatersubsequent post-conflict adjustments were associated with greaterPFC activity. Hence, on the Stroop task, it appears that conflictmonitoring may play a role in behavioral adjustments.

However, it is unclear whether these results for the Stroop taskwould also be found on other response conflict tasks. In fact, thereis some evidence from previous research that the size of trial-to-trial adjustments might vary across response conflict tasks. Forexample, some studies have not reported significant trial-to-trialadjustments in the Eriksen flanker task after the removal ofstimulus repetitions (Mayr et al., 2003; Nieuwenhuis et al., inpress). On the Stroop task, conflict adjustments have been reportedwithout stimulus repetitions in at least three studies (Egner andHirsch, 2005a; Kerns et al., 2004, 2005), although in two of thesestudies the overall effect was of only moderate to large size andwas not always consistent across both current trial types (e.g., onlyiI differed from cI or only iC differed from cC; Kerns et al., 2004,2005). In contrast, the response conflict task, which seems toconsistently produce the largest conflict adjustment effect sizes, isthe Simon task (Sturmer et al., 2002; and even without exactstimulus repetitions, Sohn and Carter, 2006). On the surface, thisseems counterintuitive because the Simon task involves a smallerbehavioral RT interference effect (i.e., incongruent RT minus

congruent RT) than either the Eriksen or the Stroop tasks (e.g., in arecent behavioral study with n>300, interference effects for theSimon, Eriksen, and Stroop tasks were 26.9, 69.2, and 183.7 ms,respectively; Kerns, 2006).

In part, one explanation for the variation in the size of post-conflict adjustment effects across tasks is the extent, consistentwith the feature integration view (Hommel et al., 2004), to whichiC and cI trials involve a partial match to the previous trial’sstimuli. On the Simon task, with two possible relevant andirrelevant stimuli, all iC and cI trials involve a partial match, whichcan account for a large adjustment effect. In contrast, the Strooptask usually involves more than two possible relevant andirrelevant stimuli, with some iC and cI trials involving exactstimulus alternations (Kerns et al., 2004), which can account forthe reduced size of the adjustment effect. In support of this, the oneimaging Stroop study that found the largest post-conflict adjust-ment effect involved only two possible relevant and irrelevantstimuli and therefore involved all partial matches on iC and cI trials(Egner and Hirsch, 2005a).

Given evidence for the effect of stimulus repetitions andalternations on behavioral adjustment, this leaves open thepossibility that behavioral adjustments on the Simon task maynot be due to conflict monitoring. If true, this would suggest thatthe conflict-monitoring hypothesis may not apply to all responseconflict tasks, which might suggest important limits on the theory.The current study examined whether conflict adjustments occurredon the Simon task and whether they were associated with ACC andPFC activity. According to the conflict-monitoring hypothesis,although repetitions and alternations should influence performance(Cho et al., 2002; Hommel et al., 2004), a previous trial’s ACCconflict activity should still predict the subsequent occurrence ofbehavioral adjustments (Jones et al., 2002). Moreover, trialsinvolving large behavioral adjustments (i.e., both speeding up on iItrials and slowing down on iC trials) should be associated withPFC activity, which should be associated with amount of ACCactivity on the previous trial.

Materials and methods

Participants

Twenty-six right-handed individuals (14 females) with anaverage age of 24.2 years (SD=4.5, range 18–36) participated inthis study and were paid $30 for their participation.

Behavioral task

In the scanner, participants performed the Simon task (Lu andProctor, 1995; Simon and Small, 1969). On each trial, participantssaw a circle in either red or green ink on the left or right side oftheir visual field. Participants needed to respond with their rightindex finger for green circles and with their left index finger for redcircles. Each trial began with a fixation cross presented for 250 ms.Then a green or red circle appeared for 1500 ms, followed by ablank screen for 750 ms (given an average reaction time of lessthan 700 ms, following Wuhr and Ansorge, 2005, the averageresponse to stimulus interval was a little more than 1800 ms).Participants completed 8 blocks of 40 trials each. Each trial waseither a congruent or an incongruent trial. On congruent trials, thelocation of the stimulus matched the side of response (e.g., a greencircle, necessitating a right-hand response, presented on the right

401J.G. Kerns / NeuroImage 33 (2006) 399–405

side). On incongruent trials, the location of the stimulus did notmatch the side of response (e.g., a green circle on the left side). Theproportion of congruent and incongruent trials was 50% each. Bothspeed and accuracy was stressed equally. Errors were infrequent(average of 3.4 error trials per participant) with only 2 participants(8%) having more than 9 errors. For the basic Simon effect, thedependent variable was the difference in reaction times betweenincongruent and congruent trials.

In analyzing post-conflict adjustments, there are four differenttrial types, depending upon the congruence or incongruence of thepreceding and the current trial: (1) iC trials (i.e., a congruent trialpreceded by an incongruent trial), (2) cC trials, (3) iI trials, and (4)cI trials. The proportion of each of these four types of trials was25%. The post-conflict adjustment effect is an interaction betweenprevious and current trial congruency, with iC trials being slowerthan cC trials and iI trials being faster than cI trials.

Functional imaging

Functional scans were acquired using a 1.5-T SiemensSymphony whole body scanner with a standard head coil. Pillowswere used to minimize head movement. Three functional T1-weighted scouts (in the axial, coronal, and sagittal planes) wereused to localize the AC/PC line. Structural images were obtainedprior to and in the same plane as functional images using a standardT1-weighted pulse sequence. Twenty-six contiguous 3.8-mm-thickaxial slices with 3.75×3.75 mm in-plane resolution were obtainedbeginning 11.4 mm below the AC-PC line. Functional images wereacquired using an echo-planar sequence (TR 2500 ms, TE 35 ms,flip angle 70°, FOV 24 cm). During the scanning session, therewere 4 BOLD runs, with each BOLD run consisting of two blocksof 40 trials of the Simon task, with 30 s of baseline fixation (i.e.,the presentation of a fixation cross) occurring in-between the twoblocks.

Data analysis

Incremental (scan to scan) and total movement were correctedusing FSL (Jenkinson et al., 2002). Participants were excludedfrom the study with mean estimated absolute movement from thereference scan greater than either 4 mm or degrees. Structuralimages were cross-registered to a reference brain by minimizingsignal intensity differences with 12-parameter AIR (Woods et al.,1998). Imaging data were linearly detrended (removing linear trendwithin each BOLD run) and then aligned to structural scans usingAIR (Woods et al., 1998), after which images were set to a standardmean intensity and spatially smoothed (8 mm FWHM). Finally,functional images were high-pass filtered (50 sec cut off) and pre-whitened using FSL (Woolrich et al., 2001).

Imaging data were analyzed with a random effects single-subject general linear model using FSL, AFNI (Cox, 1996), andNIS software (Fissell et al., 2003). To examine response conflictactivity on the Simon task, one analysis was performed with threecovariates: (1) incongruent correct trials; (2) error trials; and (3) anall trials covariate (either congruent or incongruent). Hence, in thisanalysis, significant conflict activity would be reflected in asignificant incongruent covariate (i.e., in the regression analysisactivity on incongruent trials even after statistically accounting forgeneral level of activity on all trials). Note that tolerance (adiagnostic measure of multicollinearity, with values ranging from 0to 1, with values less than 0.1 indicating problematic multi-

collinearity; Hays, 1988) was 0.70 for this response conflictanalysis (for the overlap between incongruent and all trialscovariates). Z-maps for individual subjects were generated thatreflected the extent to which each voxel’s activity conformed to ana priori canonical double gamma hemodynamic response function.Statistical threshold was p<0.005 and 8 contiguous voxels(Forman et al., 1995).

To examine whether activity in certain brain regions on conflicttrials predicted making post-conflict adjustments, following Kernset al. (2004), iC and iI trials were divided into fast and slow trialsusing a median split. Good adjustment for iC trials means goingslower, whereas good adjustment for iI trials means going faster.To examine whether activity on the preceding trial predictedactivity on iI trials, an analysis was performed with the followingcovariates: (1) correct trials immediately preceding correct fast iItrials; (2) correct trials immediately preceding slow iI trials; (3)error trials; (4) all other incongruent trials; and (5) an all trialscovariate. Then, activity for trials preceding fast iI trials wascompared to trials preceding slow iI trials. A separate analogousanalysis was completed for to examine whether activity on thepreceding trial predicted activity on iC trials (note that this analysiswas separated from the analysis for iI trials to prevent multi-collinearity from affecting the results). Finally, to examine whichbrain regions predicted post-conflict adjustments both for iI and ICtrials, a conjunction analysis was performed that examined regionsthat exhibited significant activity preceding both good iI and goodiC trials, using an a priori p<0.05 threshold within each analysis(Friston et al., 2005; Kerns et al., 2004).

To examine which brain regions were active on goodadjustment trials, following Kerns et al. (2004), two additionalanalyses were performed. In one analysis examining activity on iItrials, the following covariates were used: (1) correct fast iI trialsthat were preceded by a correct trial; (2) correct slow iI trialspreceded by a correct trial; (3) error trials; (4) all other incongruenttrials; and (5) an all trials covariate. Then, activity for fast iI trialswas compared to activity for slow iI trials. A similar separateanalysis was conducted to examine activity on iC trials. Then aconjunction analysis was performed to examine which brainregions were active during post-conflict adjustments for both iI andIC trials.

One final analysis examined whether a brain region thatpredicted making good post-conflict adjustments (such as theACC) was directly associated with activity in a brain region thatwas active during good post-conflict adjustment trials (such as thePFC). To do this, within each subject, conflict and post-conflictscans were correlated. Given a TR of 2.5 s, the BOLD responseshould be at its peak elevation in between 2 and 4 scans for bothconflict and post-conflict scans, respectively. In order to notcorrelate activity for temporally overlapping scans, the secondconflict scan was correlated with the second post-conflict scan(which was the third conflict scan), the third conflict scan wascorrelated with the third post-conflict scan, and the fourth conflictscan was correlated with the fourth post-conflict scan. Thesecorrelation values were averaged within each subject and thentreated as the dependent variable to test whether the correlationbetween conflict and post-conflict scans were significant. Inaddition, to rule out general activity across brain regions, partialcorrelations were computed two different ways. One set of partialcorrelations controlled for activity in a brain region also activatedby response conflict. The other set of partial correlations controlledfor baseline activity (i.e., at scan 1 of the conflict trial). These

Table 1Post-conflict behavioral adjustment effect on the Simon test

Trial type Mean SD

cC 650.6 102.5iC 699.3 97.5cI 712.3 100.7iI 672.8 100.7

Table 2Regions significantly activated during incongruent Simon trials

Region (Brodmann's area) Numberofvoxels

Peak ActivityTalairach coordinates

x y z

Anterior cingulate/supplementarymotor (32/6)

82 −7 3 49

Inferior parietal (BA 40) 346 −44 −52 48179 41 −38 53

Middle frontal gyrus (BA 9) 15 −40 30 3318 33 37 35

402 J.G. Kerns / NeuroImage 33 (2006) 399–405

correlation analyses provide information about whether ACCactivity was correlated with PFC activity. However, given thetemporal sluggishness of the BOLD response (Huettel et al., 2004),one cannot definitively conclude that this correlation reflects onlyconflict (ACC) and post-conflict (PFC) trial activity.

Results

Behavioral data

Behavioral data were first analyzed in a 2 (previous trialtype: congruent vs. incongruent)×2 (current trial type: con-gruent vs. incongruent) ANOVA. As expected, as can be seenin Table 1 and Fig. 1, the effect of current trial type (i.e., thebasic Simon effect) was significant, F(1,25)=18.80, p<0.001, asreaction times were slower for current incongruent trials (683.3,SD=97.9) than for current congruent trials (mean=700.1,SD=102.8). The effect of previous trial type was notsignificant, F(1,25)=2.03, p=0.17. However, there was asignificant previous by current trial type interaction (i.e., thepost-conflict adjustment effect), F(1,25)=102.53, p<0.001. Ascan be seen in Fig. 1, iI trials were significantly faster than cItrials, t(25)=8.96, p<0.001, whereas iC trials were significantlyslower than cC trials, t(25)=7.76, p<0.001. This was true evenafter removing exact stimulus repetitions, iI vs. cI, t(25)=5.21,p<0.001; iC vs. cC, t(25)=4.19, p<0.001.

Imaging data

First, response conflict activity on incongruent trials wasexamined. As can be seen in Table 2, as expected and consistentwith previous research (Nee et al., 2004), there was significantconflict-related activity in the caudal ACC extending superiorlyinto supplementary motor areas. Among other regions activated byconflict were the left PFC and bilateral parietal areas.

Fig. 1. Simon task behavioral performance, with significant effects of bothcurrent conflict and post-conflict adjustments (i.e., previous trial by currenttrial interaction).

Next, a conjunction analysis was conducted to examine whichbrain regions on conflict trials predicted making good behavioraladjustments on both subsequent iI and iC trials. As can be seen inFig. 2, only one brain region, the caudal ACC (Brodmann’s area,BA, 32; Talairach coordinates −3, 8, 41), significantly predictedmaking better adjustments on the following trial. Interestingly, theACC region in the current study is very similar to the ACCregion found in a previous study to predict post-conflictadjustments on the Stroop task (Kerns et al., 2004; Talairachcoordinates 1, 10, 40).

Next, a conjunction analysis was conducted to examine whichbrain regions were active during good adjustment trials for both iIand iC trials. As can be seen in Fig. 3, two adjacent brain regions inthe left PFC (BA 9/46, coordinates −44, 28, 28; and BA 6/9,coordinates −52, 5, 36) were significantly active on goodadjustment trials.

Finally, it was examined whether previous trial ACC conflictactivity was directly associated with post-conflict adjustmentdorsolateral (BA 9/46) PFC activity. Previous trial ACC conflictactivity was significantly associated with current trial PFC activity,p<0.01. This was true even after partialling out activity for acontrol brain region, the left inferior parietal region which was alsoactivated by conflict, or after partialling out baseline PFC activity(i.e., amount of activity during the first scan of the conflict trial).

Discussion

The results of this study are consistent with the conflict-monitoring hypothesis. In this study, ACC conflict-related activitypredicted subsequent adjustments in performance. In addition,

Fig. 2. Region of anterior cingulate cortex significantly predictingsubsequent post-conflict behavioral adjustments (BA, 32; coordinates −3,8, 41).

Fig. 3. Left prefrontal regions significantly activated during post-conflictadjustment trials (BA 9/46, coordinates −52, 5, 36).

403J.G. Kerns / NeuroImage 33 (2006) 399–405

high adjustment trials were associated with increased PFCactivity, with ACC activity on the previous trial being associatedwith amount of PFC activity on the subsequent trial. Overall,these results are consistent with the hypothesis that ACC conflictmonitoring serves as a signal that results in the recruitment ofgreater PFC activity and that this contributed to behavioraladjustments.

The current results using a Simon task are consistent with thoseof a previous study using the Stroop task (Kerns et al., 2004).Moreover, recently a third study has also replicated and extendedthis pattern of results using a task-switching paradigm (involvingcolor and movement direction; Liston et al., 2006). Hence, itappears that conflict monitoring might contribute to behavioraladjustments in performance across a number of response conflicttasks.

One issue in future research testing predictions of the conflict-monitoring hypothesis is to account for other possible sources ofvariance that can contribute trial-to-trial adjustments. For example,as discussed, post-conflict adjustments after removing exactrepetitions on the Eriksen flanker task have not always beenreported (Mayr et al., 2003). However, it has been argued thatnegative priming might reduce adjustment effects on this task(Ullsperger et al., 2005). Consistent with this, a flanker task thatremoved negative priming effects did result in significantbehavioral adjustments (Ullsperger et al., 2005). In the currentstudy, exact stimulus repetitions and alternations also probablycontributed to post-conflict adjustments (Hommel et al., 2004). Infact, the crossover interaction, with iI trials being significantlyfaster than iC trials in the current study might be accounted for bythe joint effects of both conflict monitoring and repetitions/alternations. Importantly, recent behavioral research by Wuhr(2005) has found evidence for significant post-conflict behavioraladjustments using a modified Simon task even after removing theeffects of repetitions and alternations. Future imaging researchusing the Simon task to investigate behavioral adjustments thatremoved the effect of repetitions and alternations might produceeven clearer neural evidence of ACC conflict activity predictingbehavioral adjustments and PFC post-conflict activity beingassociated with behavioral adjustments.

Another issue for future research on conflict monitoring andbehavioral adjustments is how general the effect of the recruitment

of cognitive control is across different types of tasks. For example,in the current study significant post-conflict PFC activity occurredon the left side of the brain, whereas in a previous study PFCactivity was on the right (Kerns et al., 2004; in contrast, the area ofthe ACC that predicted post-conflict adjustments in the currentstudy is very similar to the area reported in a previous study;current study Talairach coordinates: −3, 8, 41; in Kerns et al.,2004: 1, 10, 40). As argued previously, different regions of thePFC might be recruited depending upon the particular nature of thetask (Kerns et al., 2004). In a previous study, the Stroop task,which involves attending to color, activated right PFC. In theSimon task, it is possible that participants are more likely to use averbal strategy (e.g., “right for green, left for red,” which were theexact instructions given to participants), which might be likely toactivate left rather than right PFC. Importantly, several recentbehavioral studies have found evidence that post-conflict adjust-ments do not transfer across different tasks (Cho et al., 2006; Egneret al., 2006; Wendt et al., in press). This suggests that the post-conflict recruitment of PFC activity may be somewhat specific andnot general. This also seems consistent with the view that the roleof the PFC is the maintenance of specific, task-relevant information(Miller and Cohen, 2001), information that would vary substan-tially across tasks.

Another issue for future research on trial-to-trial adjustments isto examine whether results could be influenced by BOLDnonlinearity. Wager and colleagues have outlined a scenario inwhich observed brain activity could differ on a trial-to-trial basisdue to nonlinearity (Wager et al., 2005). If one trial type has greateractivity on a previous trial, there might be relatively reducedactivity on the subsequent trial of that same trial type due tononlinearity (i.e., BOLD saturation). This suggests that controllingfor the effects of nonlinearity might be critical in some contexts toaccurately assess neural activity across two trials (Wager et al.,2005). However, for the current study (and for Kerns et al., 2004),it is not immediately clear how nonlinearity could account for theresults. The results of the current study differ in two ways from thescenario outlined by Wager and colleagues. If nonlinearityproduced the trial-to-trial changes in observed BOLD response,then increased activity on one trial should be followed bydecreased activity on the next trial. However, in the current study,increased activity on one trial (conflict trial) was followed byincreased (not decreased) activity on the next trial (post-conflicttrial), which is the exact opposite of what would be expected by aneffect of nonlinearity. In addition, another difference from thescenario outlined by Wager and colleagues is that in the currentstudy, significant activity was not found across trials in the samebrain region. For example, activity in PFC differed significantlybetween trials only on post-conflict trials, but not on the previoustrial. Nonlinearity cannot account for these results because if thereis no difference in previous trial activity then there is no reason toexpect a difference in current trial activity. Nevertheless, given thatnonlinearity could produce BOLD changes across trials, futureimaging research on trial-to-trial effects should take into accountthe effects that nonlinearity could have on their results at short ITIs(Wager and Nichols, 2003; Wager et al., 2005).

One other issue for future research on the nature of post-conflictadjustments is to more precisely delineate the time course of theengagement of the PFC and of control processes (Durston et al.,2003). To answer this question, ideally it would be possible toconduct research using jittered ITIs and/or at least occasional longITIs. However, one can imagine that this type of imaging research

404 J.G. Kerns / NeuroImage 33 (2006) 399–405

may not be feasible in studying post-conflict adjustments. Onereason is that there might be only a short temporal window inwhich greater cognitive control is engaged. In fact, on the Simontask it has been found that behavioral adjustments decrease withincreasing ITI, although significant adjustments have still beenreported with a 6-s ITI (Wuhr and Ansorge, 2005). A secondreason is that it has been argued that the immediate engagement ofcognitive control may be less likely to occur if there can be a longdelay period before control is needed (Weissman et al., 2005).However, two studies using a jittered ITI have now reportedsignificant post-conflict behavioral adjustments and significantpost-conflict PFC activation, one using a jittered ITI of between 3and 5 s (Egner and Hirsch, 2005b), the other using a jittered ITI ofbetween 0.5 and 12.5 s (Liston et al., 2006). Hence, it does appearthat the recruitment of the PFC lasts for a number of seconds andeven with variability in the onset of the next trial. One additionalway future research could examine the time course of theengagement of greater cognitive control is to use ERPs (West,2004).

Acknowledgments

Thanks to Christy Watson, Joni Schupp, and the rest of the staffat the Radiology Department at the University of Missouri MedicalSchool and Hospital. Thanks also to Raymond Y. Cho for helpfuldiscussions of these research ideas.

References

Botvinick, M.M., Nystrom, L.E., Fissell, K., Carter, C.S., Cohen, J.D., 1999.Conflict monitoring versus selection-for-action in anterior cingulatecortex. Nature 402, 179–181.

Botvinick, M.M., Braver, T.S., Barch, D.M., Carter, C.S., Cohen, J.D., 2001.Conflict monitoring and cognitive control. Psychol. Rev. 108, 624–652.

Botvinick, M.M., Cohen, J.D., Carter, C.S., 2004. Conflict monitoring andanterior cingulate cortex: an update. Trends Cogn. Sci. 8, 539–546.

Carter, C.S., Braver, T.S., Barch, D.M., Botvinick, M.M., Noll, D., Cohen,J.D., 1998. Anterior cingulate cortex, error detection, and the onlinemonitoring of performance. Science 280, 747–749.

Cho, R.Y., Nystrom, L.E., Brown, E.T., Jones, A.D., Braver, T.S., Holmes,P.J., Cohen, J.D., 2002. Mechanisms underlying dependencies ofperformance on stimulus history in a two-alternate forced-choice task.Cogn. Affect. Behav. Neurosci. 2, 283–299.

Cho, R.Y., Forster, S.E., Orr, J.M., Carter, C.S., 2006. Generalizability ofsignaling for control processes, part II: across task-level representations.Poster presented at the annual meeting of the Cognitive NeuroscienceSociety, San Francisco, CA.

Cohen, J.D., Botvinick, M.M., Carter, C.S., 2000. Anterior cingulate andprefrontal cortex: who’s in control? Nat. Neurosci. 3, 421–423.

Cox, R.W., 1996. AFNI: Software for analysis and visualization offunctional magnetic resonance neuroimages. Comput. Biomed. Res.29, 173–1672.

Durston, S., Davidson, M.C., Thomas, K.M., Worden, M.S., Tottenham, N.,Martinez, A., Watts, R., Ulug, A.M., Casey, B.J., 2003. Parametricmanipulation of conflict and response competition using rapid mixed-trial event-related fMRI. NeuroImage 20, 2135–2141.

Egner, T., Hirsch, J., 2005a. The neural correlates and functional integrationof cognitive control in a Stroop task. NeuroImage 24, 539–547.

Egner, T., Hirsch, J., 2005b. Cognitive control mechanisms resolve conflictthrough cortical amplification of task-relevant information. Nat.Neurosci. 8, 1784–1790.

Egner, T., Delano, M., Hirsch, J., 2006. On the specificity of cognitivecontrol: conflict resolution does not generalize across different sources

of conflict. Poster Presented at the Annual Meeting of the CognitiveNeuroscience Society, San Francisco, CA.

Fissell, K., Tseytlin, E., Cunningham, D., Karthickeyan, I., Carter, C.S.,Schneider, W., Cohen, J.D., 2003. Fiswidgets: a graphical computingenvironment for neuroimaging analysis. Neuroinformatics 1, 111–125.

Forman, S.D., Cohen, J.D., Fitzgerald, M., Eddy, W.F., Mintun, M.A., Noll,D.C., 1995. Improved assessment of significant activation in functionalmagnetic imaging (fMRI): use of a cluster-size threshold. Magn. Reson.Med. 33, 636–647.

Friston, K.J., Penny, W.D., Glaser, D.E., 2005. Conjunction revisited.NeuroImage 23, 661–667.

Gratton, G., Coles, M.G., Donchin, E., 1992. Optimizing the use ofinformation: strategic control of activation of responses. J. Exp. Psychol.Gen. 121, 480–506.

Hays, W.L., 1988. Statistics. Holt, Rinehart, and Winston, Chicago, IL.Hommel, B., Musseler, J., Aschersleben, G., Prinz, W., 2001. The theory of

event coding (TEC): a framework for perception and action planning.Behav. Brain Sci. 24, 849–937.

Hommel, B., Proctor, R.W., Vu, K.P.L., 2004. A feature-integration accountof sequential effects in the Simon task. Psychol. Res. 68, 1–17.

Huettel, S.A., Song, A.W., McCarthy, G., 2004. Functional MagneticResonance Imaging. Sinauer, Sunderland, MA.

Jenkinson, M., Bannister, P., Brady, M., Smith, S., 2002. Improvedoptimization for the robust and accurate linear registration and motioncorrection of brain images. NeuroImage 17, 825–841.

Jones, A.D., Cho, R.Y., Nystrom, L.E., Cohen, J.D., Braver, T.S., 2002. Acomputational model of anterior cingulate function in speeded responsetasks: effects of frequency, sequence, and conflict. Cogn. Affect. Behav.Neurosci. 2, 300–317.

Kerns, J.G., 2006. Disorganized schizotypy, emotion traits, and cognitivecontrol task performance. Manuscript in preparation.

Kerns, J.G., Cohen, J.D., MacDonald III, A.W., Cho, R.Y., Stenger, V.A.,Carter, C.S., 2004. Anterior cingulate conflict monitoring and adjust-ments in control. Science 303, 1023–1026.

Kerns, J.G., Cohen, J.D., MacDonald III, A.W., Johnson, M.K., Stenger,V.A., Aizenstein, H., Carter, C.S., 2005. Decreased conflict and error-related activity in the anterior cingulate cortex in schizophrenia. Am. J.Psychiatry 162, 1833–1839.

Laming, D.R.J., 1979. Autocorrelation of choice-reaction time. ActaPscyhol. 43, 381–412.

Liston, C., Matalon, S., Hare, T.A., Davidson, M.C., Casey, B.J., 2006.Anterior cingulate and posterior parietal cortices are sensitive todissociable forms of conflict in a task-switching paradigm. Neuron 50,643–653.

Lu, C.H., Proctor, R.W., 1995. The influence of irrelevant locationinformation on performance: a review of the Simon and spatial Stroopeffects. Psychon. Bull. Rev. 2, 174–207.

Mayr, U., Awh, E., Laurey, P., 2003. Conflict adaptation effects in theabsence of executive control. Nat. Neurosci. 6, 450–452.

Miller, E.K., Cohen, J.D., 2001. An integrative theory of prefrontal cortexfunction. Ann. Rev. Neurosci. 24, 167–202.

Nee, D.E., Jonides, J., Wager, T.D., 2004. A meta-analysis of interferenceresolution tasks. Poster Presented at the Annual Meeting of theCognitive Neuroscience Society, San Francisco, CA.

Nieuwenhuis, S., Stins, J.F., Posthuma, D., Polderman, T.J.C., Boomsma,D.I., de Geus, E.J., in press. Accounting for sequential trial effects in theflanker task: conflict adaptation or associative priming? Mem. Cognit.

Rabbit, P.M., 1966. Errors and error correction in choice-response tasks.J. Exp. Psychol. 71, 264–272.

Simon, J.R., Small Jr., A.M., 1969. Processing auditory information:interference from an irrelevant cue. J. Appl. Psychol. 53, 433–435.

Sohn, M.H., Carter, C.S., 2006. Conflict adaptation effect is independent ofstimulus repetition: evidence for the conflict monitoring model.Manuscript under review.

Stoet, G., Hommel, B., 1999. Action planning and the temporal binding ofresponse codes. J. Exp. Psychol. Hum. Percept. Perform. 23,1625–1640.

405J.G. Kerns / NeuroImage 33 (2006) 399–405

Sturmer, B., Leuthold, H., Soetens, E., Schroter, H., Sommer, W., 2002.Control over location-based response activation in the Simon task:behavioral and electrophysiological evidence. J. Exp. Psychol. Hum.Percept. Perform. 28, 1345–1363.

Ullsperger, M., Bylsma, L.M., Botvinick, M.M., 2005. The conflictadaptation effect: it’s not just priming. Cogn. Affect. Behav. Neurosci.5, 467–472.

Wager, T.D., Nichols, T.E., 2003. Optimization of experimental design infMRI: a general framework using a genetic algorithm. NeuroImage 18,293–309.

Wager, T.D., Vazquez, A., Hernandez, L., Noll, D.C., 2005. Accounting fornonlinear BOLD effects in fMRI: parameter estimates and a model forprediction in rapid event-related studies. NeuroImage 25, 206–218.

Weissman, D.H., Gopalakrishnan, A., Hazlett, C.J., Woldorff, M.G., 2005.Dorsal anterior cingulate cortex resolves conflict from distracting stimuliby boosting attention toward relevant events. Cereb. Cortex 15, 229–237.

Wendt, M., Kluwe, R.H., Peters, A., in press. Sequential modulations ofinterference evoked by processing task-irrelevant stimulus features.J. Exp. Psychol. Hum. Percept. Perfom.

West, R., 2004. The effects of aging on controlled attention and conflictprocessing in the Stroop task. J. Cogn. Neurosci. 16, 103–113.

Woods, R.P., Grafton, S.T., Holmes, C.J., Cherry, S.R., Mazziotta, J.C.,1998. Automated image registration: I. general methods andintrasubject, intramodality validation. J. Comput. Assist. Tomogr. 22,139–152.

Woolrich, M.W., Ripley, B.D., Brady, M., Smith, S.M., 2001. Temporalautocorrelation in univariate linear modeling of FMRI data. NeuroImage14, 1370–1386.

Wuhr, P., 2005. Evidence for gating of direct response activation in theSimon task. Psychon. Bull. Rev. 12, 282–288.

Wuhr, P., Ansorge, U., 2005. Exploring trial-by-trial modulations of theSimon effect. Q. J. Exp. Psychol., A 58, 705–731.