Abstract

High density electrical mapping was used to index event-related brain activity in subjects performing parametric variations of the ‘AX’-type continuous performance task (AX-CPT) that differentially challenge control, and informative control conditions. In the AX-CPT, subjects must use context, created by a cue stimulus, to guide response to a target. Diseases such as schizophrenia and attention deficit hyperactivity disorder (ADHD) are associated with impaired AX-CPT performance. Event-related potentials (ERP) were analyzed as a function of both global and local stimulus context. The topographical analysis revealed that well-defined ERP are elicited under conditions where subjects must override a prepotent response. Activation patterns related to overriding a prepotent response (Go to No-Go) differed markedly from those associated with overriding a prepotent non-response (No-Go to Go). Dipole source mapping suggested that withholding a prepotent response is reflected primarily in anterior cingulate/dorsolateral prefrontal cortex activity during the 350–450 ms latency range following presentation of the No-Go. In contrast, preparing to respond is reflected in parietal (including area BA 40) activity during the same latency range, followed by a prolonged frontal negativity (contingent negative variation; CNV). Similar patterns of activation were observed whether the changes in preparation were triggered by cue or target stimuli, though target-elicited potentials peaked earlier.

Introduction

Although there is great interest in understanding the voluntary control of behavior, the brain mechanisms underlying this control are still poorly understood. To address this issue, many studies have used paradigms that create a global prepotency of response that must occasionally be overridden by voluntary control. One commonly used task is the Continuous Performance Task (CPT) (Rosvold et al., 1956), where the subject must make a motor response, usually a button press, to a rare target. This task, and its variations, have been used extensively to index voluntary control, and also sustained attention, vigilance and short-term memory in healthy subjects. They also have been used to assess brain dysfunction in disorders such as schizophrenia and attention deficit hyperactivity disorder (ADHD) (Rosvold et al., 1956; Cohen et al., 1999; Elvevag et al., 2000; Javitt et al., 2000; Umbricht et al., 2000; Riccio and Reynolds, 2001).

In the classical CPT task, subjects are presented with a stream of letters, one at a time, and are required to press a button when an infrequent target letter, usually an ‘X’, is presented [for a review, see Riccio and Reynolds (Riccio and Reynolds, 2001)]. A variation of this task, the so-called ‘AX-CPT’, uses a two-step version of this task, with different letters as cue and target, most commonly an ‘A’ for cue and an ‘X’ for target (Rosvold et al., 1956; Riccio and Reynolds, 2001). In this task, each trial consists of a sequence of two letters, that are presented one at a time, and the subject’s task is to respond (Go case), whenever the letter ‘A’ (correct cue), is followed by the letter ‘X’ (correct target). They are instructed not to respond (No-Go cases) when A is followed by any letter other than X, or when the cue is any letter other than A. Thus, targets represent ambiguous stimuli that must be disambiguated based upon the context provided by the cue stimulus. Accurate performance depends on both the accurate representation of context information and the maintenance of this representation across the delay period between the cue and target. Changing the proportion of the different types of trials, allows the context to be manipulated, providing a powerful method for varying the demands placed on ‘control’ systems of the brain (Cohen et al., 1996; Servan-Schreiber et al., 1996). This task has been recently used in studies of contextual memory, and the results have been applied to create models of underlying neural networks (Servan-Schreiber et al., 1996).

Neuroanatomical substrates of AX-CPT performance have been studied using functional imaging (fMRI) (Carter et al., 1998; Barch et al., 2001a). These studies have consistently demonstrated involvement of the anterior cingulate cortex (ACC) and the dorsolateral prefrontal cortex (DLPFC), and it was suggested that these areas may play important roles in monitoring conflict and asserting executive control, as had also suggested by experiments using other behavioral control tasks (Pardo et al., 1990). A severe limitation of the fMRI approach, however, is that it provides very limited temporal information and so the relative timing of activations across the network of regions involved in cognitive control cannot be well characterized.

Here we use methods of high-density electrical mapping from a relatively dense whole-scalp electrode array (64 channels) and inverse source localization techniques (Scherg and Picton, 1991; Murray et al., 2001) to assess the relative contributions from intracranial generators involved in cognitive control mechanisms and the relative timing of such contributions. High-density mapping and source analysis allow us to more thoroughly exploit the exquisite temporal resolution of the event-related potential (ERP) method. This ability is particularly strong when good a priori hypotheses regarding the location of probable intracranial sources can be assessed from the functional imaging literature, as in the case of AX-CPT. Knowledge of the characteristics and sources of the ERPs has important implications for understanding the neurophysiological substrate of disorders that cause impairment in the performance of AX-CPT.

Materials and Methods

Subjects

Eleven healthy, paid volunteers (seven males), participated in this study (aged 18–37 years, mean = 25.7 years). Written informed consent was obtained and the Institutional Review Board of the Nathan Kline Institute approved all procedures. All subjects had normal or corrected-to-normal vision, and all but one were right-handed, as assessed with the Edinburgh Handedness Inventory (Oldfield, 1971).

Procedure

Subjects performed a series of tasks, in order, and were allowed breaks whenever requested. They sat in a comfortable chair inside a darkened, electrically shielded and acoustically attenuated room, and attended stimuli that were presented on a computer screen located 137 cm in front of their eyes. The stimuli were easily recognizable white letters (Helvetica font), comprising ~2° of visual angle, presented on a black background. Subjects responded, in a task-specific way, by pressing a button on a response pad with the right index finger. Before the recording of brain activity on a given task, a practice block of trials was presented to familiarize the subject with the task. Subjects were instructed to respond as quickly and as accurately as possible.

AX-CPT Task

In the AX-CPT task, the subject was instructed to press the button every time there was a letter ‘X’ (target) following a letter ‘A’ (cue), but not if the cue was any other letter, referred to collectively as ‘B’, or if the target was any other letter than X, referred to collectively as ‘Y’. Consequently, there were four types of trials in this task: AX, the correct sequence; AY, correct cue, but incorrect target; BX, incorrect cue, correct target; and BY, where neither cue or target were correct.

Parametric manipulations of the task were performed to create conditions that differentially challenged control mechanisms, allowing the evaluation of the specific ERP components. The proportions of each type of trial were varied to create three versions of the original task. In each case, one type of trial is presented most of the time (70%), and the other three types of trials are equiprobable (10% each). The probabilities in each version of the task determined what was referred to as ‘Global Context’ of the task, i.e. they changed the expectation of the response to be either Go or No-Go (Table 1). The expectation created for each type of response (Go, No-Go) after the presentation of the cue is referred to here as the ‘Local Context’ of the trial.

In the basic task, hereafter termed AX-70, most of the target(X)s are correctly cued (AX sequence). Therefore, this task introduces a strong bias towards responding to the target. One challenge in performing this task is to voluntarily inhibit the prepotent motor response to target(X) during BX trials. Another challenge arises in AY trials when the subject must inhibit the prepotent motor response based upon immediate, non-contextual information (i.e. presentation of the incorrect target(Y)). The present study permits evaluation of the degree to which these two No-Go situations depend upon similar underlying neural processes.

In the first variant, the BX-70 task, the most common sequence is BX, determining that most trials will not require a response (Global Context = No-Go). Thus, the challenge in performing this task is to maintain a representation of the infrequent cue(A) in order to respond correctly to the target when the sequence is AX (Go). This task changes the prepotency of the target(X), because in most cases target(X) occurs but does not require a motor response.

In the second variant, the AY-70 task, there are mostly AY sequences (Global Context = No-Go). The appearance of cue A, in this case, indicates a higher probability of a No-Go trial (AY sequence), even though the cue is the same as in the correct sequence (AX sequence). Because only a low proportion of targets are X, there is no strong prepotency to prepare a motor response. Thus, the challenge in performing this task is to prepare a last-minute, unexpected response to the target when the sequence is AX. This task also serves as a control for the novelty effects of the letters used as No-Go cues, because it allows the differentiation of the brain activity following cue(B) that is due to a change in expectation of response, as in AX-70, from activity that is due to the proportions of the different types of cues [cue(A), 80%, cue(B), 20%, in both AX-70 and AY-70].

The probabilities of each type of trial in each task are tabulated in Table 1. The probability of each type of cue and the conditional probability of a Go trial given the correct cue (sequence AX) are also shown.

In all types of trials, the duration of the cue and the target were 250 ms, the interval between cue and target (interstimuli interval; ISI) was 1.1 s, and the inter-trial interval was 1.25 s. For each variation, six blocks of ~93 pairs of letters were presented, totaling 560 trials.

Control Tasks

Continuous Performance Task (CPT)

The original CPT task: a stream of letters is presented and subjects are instructed to press the button when the letter ‘X’ (target) is detected. Ten percent of the letters were targets, and thus the Global Context of the task was No-Go. The complete task consisted of six blocks of ~186 letters, totaling 1120 presentations, with 112 being targets. The stimuli were on for 250 ms and the ISI was 1.15 s.

XX-CPT

The subject is instructed to press the button only when an occasional cue, usually X, is followed by another X (target), and not by other letters (Fitzpatrick et al., 1988). Thus the cue signals whether to prepare a movement whereas the target signals whether to initiate a movement or inhibit an already prepared response. Six blocks of ~93 pairs of letters were presented, totaling 560 trials. Cues were presented in 20% of the trials (112 cues). Targets were presented in 10% of the trials (56 targets), also creating a No-Go Global Context. The stimuli were on for 250 ms and the ISI was 1.12 s.

Identical Pairs

Subjects press the button when a letter is a repeat (target) of the previously presented letter (Rutschmann et al., 1977). Targets were present in 10% of the trials, in six blocks of ~93 cue–target pairs, totaling 560 trial pairs/56 targets. The stimuli were on for 250 ms and the ISI was 1.15 s.

ERP Recording

High-density electroencephalography (EEG) was acquired continuously by Neuroscan Synamps from 64 scalp electrodes referenced to an electrode on the nose (band-pass filtered from 0.05 to 100 Hz, digitized at 500 Hz, impedances < 5 kΩ). Digital tags were obtained for both cues and targets. For each stimulus, ERP epochs were constructed offline. Epochs began 100 ms before and ended 750 ms after the onset of the visual stimulus. An additional analysis consisted of epochs beginning 100 ms before and ending 1500 ms after the onset of the cue, to encompass the entire delay period. Trials with blinks and large eye movements were rejected offline on the basis of horizontal (HEOG) and vertical (VEOG) electro-oculograms. Epochs in which the ERP amplitude exceeded ±80 μV at any electrode site were eliminated and the remaining epochs were averaged for each subject for each experimental condition/trial type, and also grouped into grand mean averages across all subjects.

ERP Analysis

The topographical analysis evaluated the scalp distribution of ERP components across relevant sites. We were particularly interested in the effects of the different tasks on the cognitive potentials elicited by the cue, present in the delay period, and in the potentials following targets in Go versus No-Go cases.

We based our analysis of Cue effects on previous ERP studies which employed the AX-CPT task (similar to AX-70 in this paper) (Javitt et al., 2000; Tekok-Kilic et al., 2001). For cue(A) we inspected componentry of the ERPs over centro-parietal and parietal cortices and for cue(B) the componentry of ERPs over fronto-central and centro-parietal cortices. The analysis was performed for the mean amplitude of the ERP during the period from 350 to 450 ms. This period was chosen because it encompassed the mean peak latencies of the P3 that followed presentation of the cue in previous studies (Javitt et al., 2000; Tekok-Kilic et al., 2001). These mean values for each type of cue were subjected to a 2 × 3 × 3 repeated measures analysis of variance (ANOVA), which has no assumptions of independence of variables. The factors were Hemisphere (right versus left), Task (AX-70, AY-70 and BX-70) and Electrode [for cue(A): C1/C2; CP1/CP2; P1/P2; for cue(B): FC1/FC2; C1/C2; CP1/CP2 (these are pairs, treated as one level in the third factor)].

The analysis of the late Delay effects was based on previous ERP studies of periods between a cue and a target stimulus, which, depending on the instructions of the task, can generate a contingent negative variation (CNV) (Walter et al., 1964; Simson et al., 1977) and Bereitschaftspotentials (BPs) (Deecke et al., 1969). Because it is well established that the amplitude of these components increase (in the negative direction) throughout the delay, the statistical analysis was performed on the mean amplitude of the period from 1.050 to 1.090 ms, immediately preceding the appearance of the target on the screen (1.1 ms). The componentry of the ERP was inspected over the frontal cortex. In this case we tested the significance of the results by repeated measures ANOVA, using a 2 × 6 × 3 design, with factors being Hemisphere, Task Condition [AX-70, cue(A) and cue(B), AY-70, cue(A) and cue(B) and BX-70, cue(A) and cue(B)], and Electrode (C1/C2, FC1/FC2, FC3/FC4).

For the analysis of the Target effects, the period and topographical location for analysis were again chosen based on a previous study (Tekok-Kilic et al., 2001). For target(X), mean amplitudes were obtained for the period between 275 and 375 ms and for target(Y) the period was 325 to 425 ms. Repeated measures ANOVAs were performed separately for each target, as there was an expectation of different topography, based on previous findings. Thus for target(X) the amplitudes were measured over central-parietal scalp (electrodes C1/C2, CP1/CP2, P1/P2), and for target(Y) amplitudes were measured over fronto-central and centro-parietal scalp (electrodes FC1/FC2, C1/C2, CP1/CP2). The repeated measures ANOVA was performed with factors: hemisphere (2), cue (2), task (3) and electrode (3). Where significant effects were found by these measures, protected follow-up comparisons (t-tests) were employed to fully characterize the nature of a given effect. All significance levels are two-tailed with a preset level of significance of P < 0.05.

Two-dimensional topographical maps of the temporal sequences of scalp distribution of the potentials were created using a global interpolation function as implemented in the commercial program Neuroscan (Sterling, VA) to facilitate comparisons of the conditions. Additionally, three-dimensional topographical maps of potentials were derived using the CURRY multimodal neuroimaging analysis software package (Version 4.0, Philips Research, Hamburg, Germany), and represent interpolated potential distributions, derived from the 64-scalp measurements and based on the computation of a common average reference. These interpolated potential maps are displayed on the three-dimensional reconstruction of an average rendered scalp surface (derived from anatomical MRIs), using the boundary element method [BEM; e.g. Fuchs et al. (Fuchs et al., 1998)] as implemented in CURRY.

Exact electrode locations were assessed for each subject on the day of testing by 3D-digitization of the locations of the scalp electrodes with respect to fiduciary landmarks (i.e. the nasion and pre-auricular notches) using a magnetic digitization device (Polhemus Fastrak and 3DspaceDX software, Neuroscan, Inc.). Electrode placement was highly consistent across subjects due to the use of a custom-designed electrode cap that constrained inter-electrode spacing and placement. An averaged version of these electrode locations was projected onto the averaged rendered head for computation of the group topographic data.

Source Analysis

A dipole source analysis was performed to assess the relative contributions of ACC and DLPFC to the potentials generated in the different conditions created by the variations of the task The source analysis used electromagnetic source estimation (EMSE) as applied through CURRY software. This method assumes that there are a limited and distinct number of active brain regions during an epoch, each of which can be approximated by an equivalent dipole. Dipole generators are placed within a three-shell spherical volume conductor model and overlaid on and adjusted to one of our subjects’ segmented structural MRI. The forward solution to this dipole configuration is tested against the observed experimental data (Scherg and Picton, 1991).

To avoid solutions that landed the dipoles in mathematically correct, yet physiologically unrealistic locations, we used a ‘seeded’ dipole strategy, where dipoles are placed in previously characterized (through functional imaging studies) neural loci. In this case, the source solution is constrained to solving the ‘forward’ problem, and only the orientation parameters are allowed to vary. The positions of the dipoles were fixed to the Talairach (Talairach and Tournoux, 1988) coordinates of ACC and DLPFC obtained in previous fMRI studies (Petit et al., 1998; Barch et al., 2001a,b; Braver et al., 2001) (Fig. 4A). The coordinates used to position the dipoles for the ACC were obtained from a meta-analysis of ACC activations obtained with a variety of tasks that required a manual response (3, 19, 35 mm) (Barch et al., 2001a). Similar coordinates for ACC have been obtained during performance of an AX-CPT task (7, 17, 33 mm) (Barch et al., 2001b), though in this study they referred to ‘ACC/supplemental motor area’ at those coordinates, and in a study of working memory for faces (−1, 16, 33 mm) (Petit et al., 1998). Since these coordinates were obtained in fMRI studies, and the ACC is a medial wall structure, the resolution of this method, at least as applied to date, did not allow the discrimination of the centers of activities for both hemispheres. Here, in order to represent more accurately the locations in both hemispheres, we positioned a pair of dipoles in locations symmetric with relation to the midline. We also tested unilateral positioning of the dipole in each hemisphere.

The coordinates used for the DLPFC dipoles were obtained from the fMRI study of Barch and collaborators (Barch et al. 2001b) (left DLPFC: −34, 25, 26). For these dipoles also, both unilateral and symmetric position coordinates were used to represent activations on the brain (Fig. 4A), even though only left activation had been reported in the study above. This was done so the solution could inform if there was bilateral or unilateral activation in our results.

The orientations of the frontal dipoles for all tests were fixed to those obtained in the No-Go target condition, because this condition produced the largest potentials on the scalp, and these potentials were very well explained by the dipoles in ACC and DLPFC [right ACC: (0.81, 0.23, 0.54), left ACC: (−0.86, −0.51, −0.07), right DLPFC: (0.13, 0.92, −0.38), left DLPFC: (0.46, 0.88, −0.07)]. To obtain those orientations, the freely rotating dipoles were initially fit to a single time point at the peak activation (340 ms) and the orientation was adjusted to minimize the residual variance between the forward solution and the observed data. The dipoles were fixed and, for each task, the epoch was widened to encompass both the peaks and surrounding period, and the solution was re-calculated to show the variation of the explained variance and of source strength throughout the relevant period.

Interpretation of the contributions of these dipoles to the scalp potentials must be made with appropriate caution. Because the dipoles are localized deeply inside the head (especially the ACC dipoles), there is an increased risk that they would be included in the solution of potentials that are in fact originating elsewhere. Thus the minimum value acceptable for explained variance by the solution was arbitrarily set at 80%. It is also important to note that modeling the activity in each area with a single equivalent current dipole represents an oversimplification of the activity in the areas. Thus these dipoles should be considered as representative of a ‘center of gravity’ and not necessarily discrete neural locations (Murray et al., 2002). Finally, because of our seeded dipole strategy, it is possible, even likely, that areas that are in fact contributing activity to the scalp potentials are not represented in the current dipole solution.

If the modeled sources were adequate, then addition of other sources (test dipoles) would not be expected to further reduce the residual variance, above that attributable to noise. The converse is also true, i.e. removal of necessary dipoles should result in a substantive increase in residual variance. To test the hypothesis that we needed all four dipoles (two symmetrical in ACC, two symmetrical in DLPFC) to explain the activity, we tested, in each case, the residual variance for a set of dipoles located only in ACC, both bilaterally and individually in each hemisphere (to check for a lateralization of the activity), only in DLPFC, also bilaterally and individually in each hemisphere, and compared the residual variance obtained in each case with that obtained when all dipoles were introduced.

In the cases where these dipoles were clearly insufficient to explain the activity recorded, an additional test was performed, with an extra pair of dipoles added bilaterally in the parietal cortex (Brodmann area 40, coordinates 40, −49, 46), which has also been shown to be activated by AX-CPT performance (Barch et al., 2001b). Finally, we tested the effects of adding additional dipoles at the other locations highlighted in the fMRI study, but those did not help explain the variance in any case.

Group-averaged evoked potential data were used in all cases in order to maintain the highest possible signal-to-noise ratio as well as to generalize our results across individuals. Parts of these results have been presented previously as abstracts (Dias et al., 2001).

Results

Behavioral Observations

Table 2 presents the reaction times (RT) and percentages of hits and false alarms (FA) averaged for all subjects. For all but the Identical Pairs task, correct responses to the Go condition were above 90%, and in all conditions the number of false alarms was below 2%.

Subjects were slowest and percent hits was poorest in AY-70. This is almost certainly related to the low conditional probability for Go trials, 12.5%, which required a last moment change of strategy. On the other hand, in the AX-70 task, which has a conditional probability of 87.5% for Go trials, subjects responded fastest. Not surprisingly, the AX-70 task, which had the highest rate of response, also produced the highest number of false alarms.

The behavioral results in the three variations of the AX-CPT task were tested using one-way ANOVAs, with task as a factor (three levels). There was a highly significant difference of RT between the tasks [F(2,10) = 19.8, P < 0.01]. Protected t-tests indicated highly significant differences in RT between AY-70 and the other tasks (AX-70 × AY-70, P < 0.01; BX-70 × AY-70, P < 0.01), but no difference between AX-70 and BX-70 (P = 0.41).

There was no significant difference in percentage of hits (P = 0.06), but there was a highly significant difference between tasks for percentage of false alarms [F(2,10) = 19.55, P < 0.01]. Protected t-tests indicated highly significant differences in false alarms between AX-70 and the other tasks (AX-70 × AY-70, P < 0.01; AX-70 × BX-70, P < 0.01), but no difference between AY-70 and BX-70 (P = 0.22).

ERPs

The potentials evoked by presentation of each type of cue and of each type of target are analyzed separately below.

ERP Responses to CUE Stimuli

Cue(A).

In both AX-70 and AY-70 tasks, cue(A) occurs in 80% of trials. However, in the AX-70 the Global Context is Go (conditional probability of target following the cue is 87.5%), whereas in AY-70 it is No-Go (conditional probability is 12.5%), so that in AX-70 cue(A) indicates prepare to respond, and in AY-70 cue(A) indicates that motor preparation is not necessary, and may in fact be deleterious for performance. In the BX-70 task, which also has a No-Go Global Context, cue(A) only occurs in 20% of the trials. In this task, the conditional probability of an X following the A was 50%.

Figure 1A illustrates the grand mean of the potentials evoked by the presentation of cue(A) in all three AX-CPT tasks, as recorded at scalp site CPZ (a central parietal locus). The ERPs were clearly modulated by the Global Context of each block of trials. The most prominent differences between the three conditions occurred in the 300–600 ms epoch.

A very robust positive potential following cue(A) in the 300–600 ms range was prominent in task BX-70, where cue(A) was less common. Analysis of the mean amplitude of the ERP during the 350–450 ms epoch (as explained in Methods), for the AX-CPT tasks, showed a highly significant effect of Task [F(2,10) = 13.65, P < 0.01] (Fig. 2). Pairwise comparisons showed there were significant differences in the amplitude of this component between the tasks (AX-70 versus AY-70, P < 0.001; AX-70 versus BX-70, P < 0.0001; AY-70 versus BX-70, P < 0.0001). The mean of the peak of this potential in task BX-70, in individuals, was at 422 ± 52.7 ms (range 351–512 ms) (Figs 1A, 2A and 3 ), and was centered over the central-parietal scalp (electrode CPZ). There was no effect of Hemisphere (P = 0.9).

Figure 3 shows temporal sequences of top views of the spatial distribution of the potentials on the scalp from 200 to 700 ms after presentation of the cue, at 100 ms intervals, for each task. The effect of Global Context is clearly illustrated, as the potential maps vary widely in the 300–600 ms range to the same visual stimulation.

Source analysis for cue(A): the posterior distribution of the scalp potentials suggested that most of the activity recorded on the scalp in the 350–450 ms range was originating in parietal areas. In order to isolate the component, the potentials recorded when cue(A) was presented in the AY-70 task, where there was no clear peak in this temporal range (see Fig. 1A), were subtracted from the equivalent potential in the BX-70 task, where the peak was most evident. The resulting potential difference was then subjected to a dipole source localization analysis (Fig. 4B).

The activity was well explained by dipoles placed in the parietal cortex (Brodmann area 40), with bilateral placement better explaining the variance than unilateral. Dipoles placed in these positions explained 90.75% of the variance at 450 ms.

As would be expected from the topographic analysis, dipoles fixed in the ACC and DLPFC accounted poorly for the potentials recorded. The explained variance at the time point when the amplitude of the difference potential was maximal (450 ms) was 74.6% with only these dipoles, and thus below our self-imposed threshold for acceptability. The source strength for each dipole showed little modulation with the task, with the DLPFC dipoles showing less modulation than the ACC dipoles. When tested independently, both sets of dipoles had weak effects (explained variances: only ACC: 75%; only DLPFC: 60.0%). These observations suggest that these areas are probably not contributing much, if any, activity to the scalp potentials in this case.

The activity was best explained by a combination of the bilateral parietal dipoles with the bilateral DLPFC dipoles. This solution brought the explained variance up a little, to 93.75%, which suggests that there are still other areas, not analyzed in this study, that are contributing to the scalp topography.

Control tasks: in the XX-CPT control task, which has the same Go/No-Go probabilities as BX-70, the ERP following the presentation of the first X (Go cue) was similar to that of BX-70-cue(A). However, in this task the potentials were of higher magnitude, both positive and negative (Figs 1C and 3). The potential had a maximal amplitude of 13.94 μV (at 438 ms), which is more than 50% greater than maximal amplitude measured for task BX-70, 8.87 μV (at 384 ms).

In the Identical Pairs task, in which the cue did not have a Go/No-Go instruction, and response type could only be ascertained after the presentation of the target stimulus, there was no clear peak in this time range (Fig. 1C).

Cue(B).

Cue(B), which always indicated a No-Go trial, was presented on only 20% of the trials of both AX-70 and AY-70 tasks. However, in AY-70, the Global Context was No-Go, whereas in AX, the Global Context was Go. Thus, only in AX-70, but not in AY-70, there was a change in the planned response based on the cue. This difference was clearly reflected in the different potentials evoked by these tasks. In BX-70, Cue(B) occurred in 80% of the trials. Because this task had a No-Go Global Context, there was no change in planned response.

The presentation of cue(B) in task AX-70 evoked a frontal negative component, with a negative peak ~270 ms, which was not evident for any of the other tasks. This component was followed by a frontal positivity, with a peak ~450 ms, again only present for the AX-70 task.

Analysis of the mean amplitude of the ERP during the 350–450 ms epoch showed a main effect of Task [F(2,10) = 6.74, P < 0.05]. Protected pairwise comparisons showed significant differences in amplitude between task AX-70 and the other tasks, but no difference between AY-70 and BX-70 (AX-70 versus AY-70, P < 0.0001; AX-70 versus BX-70, P < 0.0001; AY-70 versus BX-70, P = 0.79) (Figs 1B, 2 and 3). There was no effect of Hemisphere (P = 0.18), though there was a significant Task by Hemisphere interaction [F(2,10) = 7.27, P < 0.05). This differential potential in task AX-70 could not have been evoked solely by the novelty of cue(B), as it was not observed in AY-70, which had the same percentage of cue(B). We propose therefore that this potential indexes the decision to withhold a response in a prepotent Go context (Cue No-Go).

Source analysis for cue(B): a dipole source analysis was performed on the difference potential resulting from the subtraction of the potential elicited by cue(B) in task AY-70, where the Global Context was No-Go, from that elicited in task AX-70, where the Global Context was Go, and thus required a change in action plan (see Fig. 1B). Bilateral dipoles fixed in both ACC and DLPFC, explained 86.0% of the variance at 444 ms, thus supporting the suggestion that these regions are likely to represent sources of activation (Fig. 4C). There was a clear positive modulation of source strength around the time of the peak activity in the difference potential, indicating that the contribution of these sources to the scalp potential peaked at the same time as the potentials (Fig. 4C). The explained variances dropped when either pair of dipoles was removed from the analysis (only ACC, explained variance was 37.7%; only DLPFC, explained variance was 6.3%), suggesting that activity in both areas has to be considered for a solution, and that the dipoles in ACC were stronger contributors to the scalp potential than those in DLPFC. Removing even one of the ACC dipoles caused a strong drop in the explained variance, with the left ACC appearing to have stronger effects (removing left ACC, the explained variance dropped to 7.2%; removing right ACC, the explained variance dropped to 13.8%). Removing the dipole on the right DLPFC had the least effect on the explained variances, causing a drop in the explained variance to 79.2%, and removing the left DLPFC dropped the explained variance to 75.3%.

Control tasks: the ERP evoked by presentation of No-Go cues in the XX-CPT control task was similar to that evoked in AX-CPT tasks that did not call for a change in strategy (AY-70 and BX-70) (Figs 1D and 3). However, the later components of the Identical Pairs task were of higher magnitude than those in the XX-CPT task in the 350–450 ms range, possibly reflecting the fact that each new cue required additional processing, as it was a possible cue.

Interestingly, in task AY-70, there was no significant difference in the potentials evoked by cue (A) or cue(B) until ~600 ms (Fig. 3), indicating that there was no clear influence of novelty of the cue per se, suggesting that the differences seen in the other tasks are more related to the changes in expectations.

Late Negativity

When the cue was A, an increasing negative potential began to develop over the fronto-central cortex at ~600 ms. This potential was not apparent when the cue was B (Fig. 3, plots at 600 ms and 700 ms, Fig. 5). An analysis of this component, as described in Methods, showed a main effect of Condition [F(5,10) = 17.81, P < 0.01], and no effect of Hemisphere [F(1,10) = 1.02, P = 0.34] or Electrode [F(2,10) = 17.81, P = 0.06]. Protected t-tests showed a highly significant effect of Cue (A versus B, P < 0.01), and highly significant differences for pairwise comparisons of the tasks, for cue(A) (AX-70 versus AY-70, P < 0.0001; AX-70 versus BX-70, P < 0.01; AY-70 versus BX-70, P < 0.0001).

This negativity was similar to the previously described contingent negative variation (CNV) (Walter et al., 1964; Rosahl and Knight, 1995) and Bereitschaftspotentials (BPs) (Deecke et al., 1969), and was present whenever the cue indicated a high likelihood of a Go trial. The scalp topography of this potential was centered over fronto-central scalp (FCZ) in all tasks, and was distributed symmetrically around the midline. Figure 5 illustrates the temporal evolution of the potential and the spatial distribution at the time of target presentation, 1100 ms, for the different tasks. Among the variations of the AX-CPT tasks, the potential had largest amplitude in task BX-70, where it reached a peak negative amplitude value of −10.93 μV (Fig. 5). The peak amplitude of the potential for task AX-70, which had a higher likelihood of a Go, was larger (in the negative direction) than that for task AY-70.

Source Analysis for the Late Negativity.

Though previous studies indicate that the CNV and the BPs have distributed sources in several regions of the brain (Hamano et al., 1997), we wanted to verify to what extent activity in the ACC and in DLPFC might be contributing to the scalp potential. So a source analysis was performed on this negativity with dipoles fixed in the ACC and DLPFC (Fig. 4D). The analysis was performed on the difference potential between the potential recorded after presentation of cue(A) in task BX-70, which was followed by a very large late negativity, and the presentation of cue(B), which did evoke late negativity in the same task. The dipoles explained 86.7% of the variance at time 1100 ms, suggesting that these areas contribute to the scalp potential. This is further confirmed by the continuously increasing (in the negative direction) source strength, specially for the ACC dipoles, which follows the increasing negativity of the observed potentials. The explained variance was very much reduced if either the DLPFC or ACC dipoles were removed from the analysis, indicating that activity in both areas contributed to the solution. Placing only a pair of bilateral dipoles on the DLPFC caused the explained variance to plummet to 9.3%, while with only the pair of ACC dipoles, the explained variance dropped to 46.3%. This latter observation indicates that despite the stable level of activity (source strength) calculated for DLPFC, its contribution seems to be important for the final solution, as removal of these dipoles seriously impacts the solution. Removing dipoles in only one hemisphere in either region also diminished the explained variance, with removal of the left DLPFC having the least effects, causing the explained variance to drop to 82.2%, as opposed to 76.8% with removal of the right DLPFC, and big drops in explained variance, to 21.3% with removal of the left ACC and a drop to 10.5% in explained variance with removal of the right ACC.

Control Tasks.

The XX-CPT control task generated the largest amplitude late negativity, as can be seen in Figure 5. The late negativity reached an amplitude of −13.13 μV at 1100 ms, compared to −10.93 μV in BX-70. Dipole source analyses of the late negativity in this task had solutions similar to those obtained for task BX-70-cue(A). In the Identical Pairs task there was no clear late negativity following presentation of cue(B), though the potentials were slightly different from zero (not shown).

ERP Response to TARGET Stimuli

After the cue was presented there was an interval, after which a target was presented. In all trials with cue(B), which determined a No-Go trial, the ERPs evoked by the appearance of the target were independent of the type of target, that is, the responses to target(X) or target(Y), in all AX-CPT tasks, were indistinguishable (Fig. 6A, BX and BY).

In the cases with cue(A), there were two possible outcomes: for target(X), it was a Go trial, requiring a button press, and for target(Y) the response should be withheld, that is, it was a No-Go trial.

Go Trials.

A Go trial occurred in 70% of the trials in the AX-70 task, with a conditional probability of target(X) occurring 87.5% of the time for cue(A). Consequently, the Local Context of response in this task, after presentation of cue(A) is Go. The reaction time was fastest in this task (Table 2). In task AY-70, the subject had only a 10% chance of a Go, and only a 12.5% conditional probability of a Go given cue(A) (Local Context is No-Go). Thus the subject was usually expecting a No-Go, and had to change strategies and prepare a motor response when the rare target(X) appeared. Not surprisingly, all subjects were slowest and had the lowest percentage of hits in this task. In task BX-70, there was also only a 10% chance of a Go, but a 50% chance of a Go following cue(A) (Local Context was same for Go or No-Go), with corresponding decrease in reaction time.

The largest effects of task after target presentations were between 200 ms and 600 ms. A comparison of these potentials, measured as the mean amplitude between 275 ms and 375 ms (on scalp site CPZ) showed a main effect of Task [F(2,10) = 41.9, P < 0.0001] and of Cue [F(1,10) = 171.8, P < 0.0001], but no effect of Hemisphere. There was also a significant interaction between Cue and Task [F(2,10) = 21.0, P < 0.001].

In task AX-70, there was a peak in the late potential, centered over centro-parietal scalp (electrode CPZ), at 308 ms (14.55 μV) (Fig. 6). A centro-parietal peak was also observed in task BX-70, only in this case the potential peaked slightly later, at 329 ms, and had a higher amplitude (22.50 μV at CPZ) (Fig. 6).

In task AY-70 there was a negative peak ~265 ms (2.83 μV) followed by a positivity centered on electrode CPZ (peak at 399 ms, 17.39 μV). The longer latency of this peak was accompanied by longer reaction times in this task (Fig. 6A).

Source analysis for Go trials: Since the surface potentials clearly indicated a posterior distribution of the potentials, a dipole source analysis was performed with dipoles fixed in the parietal location. For this analysis, a difference potential was obtained by subtracting the potentials evoked in task BX-70 by the target(X) following cue(B) (Global Context was No-Go) from the potentials evoked for target(X) when following cue(A) (Global Context was Go) (Fig. 4E). The solution with only the bilateral parietal dipoles explained 91.4% of the variance at 404 ms.

To verify the extent to which ACC and/or DLPFC might be involved in the solution, another analysis was performed, including ACC and DLPFC dipoles. The solution with all six dipoles (two ACC, two DLPFC and two Parietal) increased the explained variance very slightly, to 94.80%, but the source strength of the ACC dipoles showed a clear modulation, suggesting that they may be contributing some activity to the scalp potential.

Control tasks: in the control tasks, similar potentials were observed in the Go trials (Fig. 7, Go). In the XX-CPT task, that followed the probabilities of task BX-70 for a Go trial, the temporal profile of the potential following the target was indistinguishable from that evoked by the presentation of the Go target in BX-70 for all recorded sites.

Similarly, the potential evoked by the CPT target closely mimics the potential observed after presentation of the Go target in the AY-70 task. These potentials are indistinguishable across the scalp. In both tasks there is a target-driven change in the expected motor response.

The appearance of a repeat letter in the Identical Pairs task evoked a longer latency positive potential. Interestingly, this task produced the longest RT in the study.

No-Go Trials.

A No-Go target presented after a Go cue required the subject to interrupt an ongoing process directed at preparing for a button press. This happened for trials AY, or, in the XX-CPT control task, when X was followed by another letter than X.

In the AX-70 task, No-Go targets were presented in 20% of the trials, with a conditional probability of 12.5% for cue(A). In task AY-70, the appearance of target(Y) matched both the Global and Local Context, which was No-Go [90% of the trials were No-Go, 87.5% after cue(A)]. In BX-70, the chances of a No-Go were again 90%, however, the conditional probability of target(Y) after cue(A) was 50% (Table 1).

The largest differences in the potentials recorded after presentation of the No-Go target occurred in the 200 ms to 600 ms time period. A comparison of the potentials, measured as mean amplitude between 325 ms and 425 ms, in the different AX-CPT task showed an effect of Task [F(2,10) = 23.12, P < 0.001] and of Cue [F(2,10) = 210.46, P < 0.0001], but not of Hemisphere or Electrode. A highly significant Cue by Task interaction was also present [F(2,10) = 75.05, P < 0.0001].

In task AX-70 there was a negative deflection, peaking at 235 ms (−0.91 μV at FCZ), located anteriorly on the scalp (Fig. 6). This negativity, though less prominent was clearly identifiable also in BX-70 (at FCZ, peak: 217 ms, 6.73 μV). There was no evidence of such a negative deflection in the AY-70 task.

A positive peak followed this negative deflection. This positivity presented more anterior than those observed in the Go trials, and was centered on CZ. In AY-70 it peaked earlier than in the other tasks, at 295 ms (8.98 μV). In BX-70 the peak was at 340 ms (25.47 μV), and in task AX-70 the peak was at 362 ms (20.94 μV).

Source analysis for no-go trials: a dipole source analysis was performed on a difference potential obtained by subtracting the potentials evoked by the target(Y) in task AX-70 following cue(B) (Global Context was No-Go) from the potentials evoked for target(Y) when following cue(A) (Global Context was Go) (Fig. 4F).

The dipoles located bilaterally in ACC and DLPFC explained 93.7% of the variance, indicating that these were reasonable locations for the sources of the activities recorded on the scalp. There was a clear peak in source strength that followed the peak in the potential. The explained variances of the solution were much lower if either the ACC (28.5%) or the DLPFC (73.0%) dipoles were removed from the analyses (Fig. 4F).

Interestingly, the scalp distribution of the difference potentials and the solution of the dipoles in this case (target No-Go) were similar to those obtained when cue(B) was presented (cue No-Go), suggesting similar mechanisms may be involved in both cases (compare Fig. 4C and F).

Control tasks: in the XX-CPT control task No-Go trials, that is, when target(Y) followed cue(X), the ERPs were indistinguishable from those in BX-70. There was a negativity (peak: 222 ms, 8.05 μV at FCZ), followed by a positivity (peak: 344 ms, 25.70 μV at FCZ). This is in agreement with the fact that all probabilities in that case were the same as in BX-70. These peaks were not recognizable in the potentials evoked to No-Go targets in the CPT task and in the Identical Pairs task, where most stimuli were No-Go (Fig. 7).

Discussion

This study directly compared the spatial and temporal distribution of brain activations evoked by variations of the widely used CPT task that differentially challenged contextual memory and voluntary control of behavior. Although some of these tasks have been extensively studied with fMRI, no prior study has investigated ERP correlates of the local modulation of global prepotencies in CPT.

In particular, the study tested the effects of general changes in strategy, as a model of behavioral control mechanisms, which could be a change from No-Go to Go situation or a switch from Go to No-Go situation. The change could occur either as a result of the cue stimulus or the target stimulus, depending on the task (Global Context) or the specific stimulus sequence (Local Context).

The results seem to show: (i) That strategic changes are not always followed by the same pattern of brain activity. (ii) That changing into an action plan (No-Go to Go) prominently involves the parietal area, whereas (iii) changing into an inhibition plan (Go to No-Go) involves the ACC and PFC more. (iv) This happens whether the changes are triggered by the cue or the target stimuli.

ERPs to Cue Stimuli

When the cue determined a No-Go response in a prepotent Go situation [cue(B) in AX-70], the main activation was located more anteriorly in the scalp, and dipole localization suggested that both the ACC and DLPFC were activated. Note that the potential was clearly not a result of the ‘novelty’ of the invalid cue (where a variety of letters were presented), because in task AY-70 there were as many cases of cue(B) as in AX-70, and no differential potential was present in this time range. If there had been a ‘novelty effect’, the potentials after cue(B) should have been the same in both tasks.

In contrast, when the cue indicated a possible Go in a No-Go Global Context situation [cue(A) in BX-70], the focus of activation was more posterior. The distribution of scalp-recorded potentials in this time range was fit best by dipoles in the parietal lobe, with little or no significant contribution from ACC or DLPFC. Nevertheless, though our parietal dipoles were restricted to area BA 40, it is unlikely that this is the only posterior area activated in the task. The position of these dipoles was chosen to represent the parietal activity because it is activated during AX-CPT in fMRI studies (Barch et al., 2001b).

These parietal potentials had both the temporal and spatial characteristics of the previously described P300. The P300 is a centro-parietal potential generated in tasks requiring voluntary detection of infrequent and task-relevant stimuli, which are intrinsic to the present task (Knight and Scabini, 1998). However, the main point of this task was to suddenly change from the expectation of not making a response to preparing a response. Thus the potential probably reflects additional processing of information, as also reflected by the remaining unexplained variance of the dipole fit.

Previous studies have proposed that both ACC and DLPFC are activated by conflict, and/or by the initiation of goal-directed behavior (Pardo et al., 1990; Cohen et al., 1999; Carter et al., 2000; Barch et al., 2001a). Our data supports these propositions, but a complete generalization cannot be made. In the two cases mentioned above, the probabilities of the uncommon cue were the same, so the amount of conflict, theoretically, is the same. Nonetheless, the brain activations had very different distributions, where both ACC and DLPFC were more activated, at least in this time range, by a voluntary inhibition of behavior, than they were by the potential initiation of behavior [cue(A) in BX-70], indicating that there are several factors involved.

Interestingly, schizophrenic patients have a reduced positivity to the appearance of cue(B) in task AX-70, and also a marked deficit in inhibiting the prepotent Go response (Javitt et al., 2000). Our results suggest that the observed decrease in this positivity are probably due to impaired processing in one or both components of the ACC-DLPFC circuit.

Late Negativity

This potential had characteristics of the contingent negative variation (CNV), a slow potential that is thought to reflect cortical priming functions (Walter et al., 1964), with the later part of the potential likely also reflecting motor preparatory potentials (Bereitschaftspotentials, BPs) (Deecke et al., 1969). CNVs are typically elicited in two-step tasks, in the delay period between the cue and the test stimulus (Walter et al., 1964). Many areas, mainly in frontal and central cortex, but also in parietal cortex, have been implicated in generating the CNV, both in humans and in non-human primates (Rebert, 1972; Hamano et al., 1997). In this study we were specially interested in whether the ACC and DLPFC contributed to the CNV and the BPs.

When the cue was A, signaling the possibility of a response, the slow negativity developed, but in a graded way. The negativity in task AX-70 had a larger amplitude than in task AY-70, that created a lower expectancy of response, suggesting that the potential was, at least in part, reflecting motor preparation (Loveless and Sanford, 1974). However, in tasks BX-70 (and in control task XX-CPT, that had the same probabilities of task BX-70) the peak amplitude of the negativity was even higher, though the conditional probability of a motor response was only 50%, suggesting the relationship is not linear, and more than one factor are influencing the potential.

Another factor suggesting that the potential was reflecting more than just motor preparation was that the potential appeared symmetrical about the midline in every case tested, even though the manual response was always made with the right hand. Preparatory motor potentials would be expected to have a more left-hemisphere distribution (Saron et al., 2001).

Dipole localization suggested that these potentials are likely to contain a significant contribution from generators localized in ACC and DLPFC, supporting the hypotheses that DLPFC and ACC participate in maintaining representations of the cue, and/or of the possibility of impending action (Rosahl and Knight, 1995; Ruchkin et al., 1995). In addition, damage of DLPFC, in neurological patients, abolishes the late part of the CNV (Rosahl and Knight, 1995). Also, blood flow increases have been shown during working memory tasks both in the DLPFC (McCarthy et al., 1994, 1996) and in the medial surfaces of the brain, in areas overlapping the areas used for dipole positioning (Petit et al., 1998).

In addition to these areas, it is likely that some (or even most), of the activity is being generated in the midline by the supplementary motor areas (SMA and Pre-SMA). Activity in these areas would also appear close to the midline, and these areas have also been strongly implicated in motor preparation (Cui and Deecke, 1999; Yazawa et al., 2000; Cunnington et al., 2002). The proximity of these areas to ACC does not allow differentiation between potentials generated in both areas in a group analysis.

ERPs to TARGET Stimuli

In the cases where the cue had determined No-Go [cue(B)], there were no differential task-related potentials. After cue(A), however, the identity, and consequent behavioral implication of the targets presented, had a strong effect on the potentials. When a Go target was presented, there was a graded centro-parietal potential, again similar to a P300. This potential was stronger in BX-70 than in AX-70. The difference in amplitude may be reflect the fact that in BX-70 there was only a 50% chance of Go, and thus a stronger brain activation could be necessary to activate the response process. In AY-70 there was a temporally delayed potential with very similar spatial distribution, suggesting that the change in response planning caused the delay. While dipole analysis suggested that both ACC and DLPFC may contribute to the potentials, the major contributors appear to reside in the parietal cortex.

In the cases where there was a strong probability of a Go response (BX-70, 50%; AX-70, 87.5%), presentation of the No-Go target evoked a negative-positive sequence of potentials with a time range and distribution similar to that of the previously described No-Go-N2 and No-Go-P3 potentials (Kiefer et al., 1998; Falkenstein et al., 1999). The potentials appeared to be indexing the change in strategy required when the unexpected target was presented, indicating the need to withhold a motor command. The amplitudes of these potentials were related to the conditional probability of a Go trial, that is, the stronger the likelihood of a Go that had to be overridden in order to avoid an error, the larger the amplitudes of the potentials. On the other hand, in task AY-70, which had a No-Go Global Context, presentation of the No-Go target did not elicit a negative potential and elicited a positive potential with much smaller amplitude. It is not clear, however, how these potentials relate to the actual decision to withhold the response, because they occur too late to have a causal effect (Filipovic et al., 1999). Nonetheless they are clearly related to that decision, and are always present in No-Go tasks.

A dipole source localization indicated that the ACC and DLPFC participated in generating the positive potential. Previous work in both monkeys (Gemba et al., 1986; Gemba and Sasaki, 1989, 1990; Sasaki et al., 1993) and humans (Falkenstein et al., 1995; Kiefer et al., 1998; Falkenstein et al., 1999; Liddle et al., 2001) has argued for a frontal localization of behavioral control in No-Go tasks, including the ventrolateral prefrontal cortex, and in addition to other areas such as parietal regions (Iversen and Mishkin, 1970; Casey et al., 2001; Liddle et al., 2001; Durston et al., 2002). In humans, a previous study investigating sources of brain activations during response inhibition mapped similar locations for the No-Go potentials (Kiefer et al., 1998). In monkeys, Gemba and collaborators (Gemba et al. 1986; Gemba and Sasaki 1989) recorded intracranially and showed that there was a surface negative potential in the dorsal bank of the principal sulcus. They also showed that electrically microstimulating that area suppressed or delayed responses, further implicating the area in control of action (Sasaki et al., 1989). Interestingly, electrical microstimulation of the frontal eye fields, an area just behind the principal sulcus in monkeys, delayed the initiation of eye movements in oculomotor tasks (Burman and Bruce, 1997), indicating that this may be a common strategy of prefrontal cortex.

Our data support the hypothesis of frontal cortex involvement in behavioral inhibition, and suggest that the impaired behavioral inhibition observed in schizophrenics may be related to impaired processing in DLPFC and/or ACC.

The fact that the activity of the DLPFC and ACC appear related is not unexpected, considering the extensive anatomical connectivity between these areas (Pandya et al., 1981; Picard and Strick, 1996; Barbas, 2000). Studies of patients with brain lesions also support a strong interaction between these areas (Gehring and Knight, 2000). Comparison of these results with those obtained in pathological states should contribute to understanding the underlying neural dysfunction.

Notes

We thank Ms Beth Higgins for technical assistance. The work was supported by NIH grants MH49334 and MH63434 and the Burroughs Welcome Fund.

Address correspondence to Elisa C. Dias, Cognitive Neurophysiology Laboratory, Program in Cognitive Neurosciences and Schizophrenia, The Nathan Kline Institute for Psychiatric Research, 140 Old Orangeburg Rd, Orangeburg, NY 10962, USA. Email: dias@nki.rfmh.org.

Table 1

Task probabilities

Task Trial type Global prepotency P(A) (%) P(X) (%) P(X|A) (%) Local prepotency after cue(A) 
 Go No-Go      
 AX BX AY BY      
AX-70 70 10 10 10 Go 80 80 87.5 Go 
AY-70 10 10 70 10 No-Go 80 20 12.5 No-Go 
BX-70 10 70 10 10 No-Go 20 80 50 Go = No-Go 
Task Trial type Global prepotency P(A) (%) P(X) (%) P(X|A) (%) Local prepotency after cue(A) 
 Go No-Go      
 AX BX AY BY      
AX-70 70 10 10 10 Go 80 80 87.5 Go 
AY-70 10 10 70 10 No-Go 80 20 12.5 No-Go 
BX-70 10 70 10 10 No-Go 20 80 50 Go = No-Go 
Control tasks Go    Global prepotency  P(X) (%) P(X|X) (%) Local prepotency 
CPT 10    No-Go  10 n.a.  
XX-CPT 10    No-Go  20 50 Go = No-Go 
Control tasks Go    Global prepotency  P(X) (%) P(X|X) (%) Local prepotency 
CPT 10    No-Go  10 n.a.  
XX-CPT 10    No-Go  20 50 Go = No-Go 
 Go    Global prepotency   P(B|B) Local prepotency 
Identical pairs 10    No-Go   10% No-Go 
 Go    Global prepotency   P(B|B) Local prepotency 
Identical pairs 10    No-Go   10% No-Go 
Table 2

Behavioral results

Tasks n RT (SD) % hits (SD) % FA (SD) 
AX-70 11 343 (82) 97.5 (2.9) 1.6 (1.1) 
AY-70 11 518 (85) 91.3 (8.0) 0.1 (0.2) 
BX-70 11 360 (56) 95.2 (6.6) 0.1 (0.1) 
Identical pairs 10 596 (91) 82.8 (11.9) 0.1 (0.1) 
CPT 11 455 (71) 98.8 (1.5) 0.1 (0.0) 
XX-CPT 10 380 (84) 97.5 (2.9) 0.1 (0.1) 
Tasks n RT (SD) % hits (SD) % FA (SD) 
AX-70 11 343 (82) 97.5 (2.9) 1.6 (1.1) 
AY-70 11 518 (85) 91.3 (8.0) 0.1 (0.2) 
BX-70 11 360 (56) 95.2 (6.6) 0.1 (0.1) 
Identical pairs 10 596 (91) 82.8 (11.9) 0.1 (0.1) 
CPT 11 455 (71) 98.8 (1.5) 0.1 (0.0) 
XX-CPT 10 380 (84) 97.5 (2.9) 0.1 (0.1) 
Figure 1.

Global Context effects. Comparison of the potentials evoked by presentation of the cue (time = 0 ms) in the different tasks. Group-averaged ERP are shown from the electrodes where maximal response was recorded, as illustrated on the insets. (A) For cue(A), the data from scalp site CPZ is shown. The trace for AX-70 is in red, AY-70 is in blue and BX-70 in green. (B) For cue(B), the data from electrode FCZ is shown. Conventions are the same as in (A). (C) For Controls (Go) the data from electrode CPZ is shown. (D) For Controls (No-Go) the data from electrode FCZ is shown. The trace for the Identical Pairs task is in red and for the XX-CPT task is in green. The dashed lines mark the period used for the statistical analysis.

Figure 1.

Global Context effects. Comparison of the potentials evoked by presentation of the cue (time = 0 ms) in the different tasks. Group-averaged ERP are shown from the electrodes where maximal response was recorded, as illustrated on the insets. (A) For cue(A), the data from scalp site CPZ is shown. The trace for AX-70 is in red, AY-70 is in blue and BX-70 in green. (B) For cue(B), the data from electrode FCZ is shown. Conventions are the same as in (A). (C) For Controls (Go) the data from electrode CPZ is shown. (D) For Controls (No-Go) the data from electrode FCZ is shown. The trace for the Identical Pairs task is in red and for the XX-CPT task is in green. The dashed lines mark the period used for the statistical analysis.

Figure 2.

Histograms comparing mean amplitude of potentials recorded between 350 and 450 ms after presentation of cue A and cue B.

Figure 2.

Histograms comparing mean amplitude of potentials recorded between 350 and 450 ms after presentation of cue A and cue B.

Figure 3.

Topographical voltage distribution (top view, anterior is up, left is left) of the group-averaged potentials recorded after presentation of the cues, at six time points (200 to 700 ms, at 100 ms intervals). Each dot represents an electrode location in the 64-electrode array. Scale represents activities from −8 to 10 μV. Boxes include electrodes used for statistical analysis.

Figure 3.

Topographical voltage distribution (top view, anterior is up, left is left) of the group-averaged potentials recorded after presentation of the cues, at six time points (200 to 700 ms, at 100 ms intervals). Each dot represents an electrode location in the 64-electrode array. Scale represents activities from −8 to 10 μV. Boxes include electrodes used for statistical analysis.

Figure 4.

(A) Illustration of the dipole positions used, projected onto the cortical surface in top, front and side views. Tailarach coordinates for the positions were: ACC (3, 19, 35 mm/–3, 19, 35 mm), DLPFC (34, 25, 26 mm/–34, 25, 26 mm), BA 40 (40, −49, 46 mm/–40, −49, 46 mm) (BF). Left: Top view of the scalp distribution of the potentials at the moment they reach their peak amplitude, for each case. Dipoles are projected onto the surface for comparison. The relative sizes of the dipoles are scaled to the source strength for each dipole in each case. Values under scalp representations indicate the scale values for each case. Right: The top graph in each composite illustrates the temporal evolution of the strength of source in μA mm × ms for all dipoles tested. The bottom graph illustrates the variance explained by the dipoles in % × ms. (B) Cue Go. Difference potential: cue(A) in BX-70 minus cue(A) in AY-70. Parietal dipoles were also included. (C) Cue No-Go: difference potential: cue(B) in AX-70 task minus cue(B) in AY-70 task. (D) Late negativity: difference potential: cue(A) in BX-70 task minus cue(B) in BX-70 task. The potentials could be mostly explained by dipoles localized in ACC and DLPFC, but their orientations were opposite to those of the No-Go situations, i.e. negative values. (E) Target Go. Difference potential: target(X) in BX-70 task, cue(A) minus target(X) in BX-70 task, cue(B). Parietal dipoles also included. Conventions as in B. (F) Target No-Go: difference potential: target(Y) in AX-70 task, cue(A) minus target(Y) in AX-70 task, cue(B). Notice that in both cases where there is a No-Go situation (C and F) the potential are represented more anteriorly than when there is a change in task expectation, but is a Go situation (B and E).

Figure 4.

(A) Illustration of the dipole positions used, projected onto the cortical surface in top, front and side views. Tailarach coordinates for the positions were: ACC (3, 19, 35 mm/–3, 19, 35 mm), DLPFC (34, 25, 26 mm/–34, 25, 26 mm), BA 40 (40, −49, 46 mm/–40, −49, 46 mm) (BF). Left: Top view of the scalp distribution of the potentials at the moment they reach their peak amplitude, for each case. Dipoles are projected onto the surface for comparison. The relative sizes of the dipoles are scaled to the source strength for each dipole in each case. Values under scalp representations indicate the scale values for each case. Right: The top graph in each composite illustrates the temporal evolution of the strength of source in μA mm × ms for all dipoles tested. The bottom graph illustrates the variance explained by the dipoles in % × ms. (B) Cue Go. Difference potential: cue(A) in BX-70 minus cue(A) in AY-70. Parietal dipoles were also included. (C) Cue No-Go: difference potential: cue(B) in AX-70 task minus cue(B) in AY-70 task. (D) Late negativity: difference potential: cue(A) in BX-70 task minus cue(B) in BX-70 task. The potentials could be mostly explained by dipoles localized in ACC and DLPFC, but their orientations were opposite to those of the No-Go situations, i.e. negative values. (E) Target Go. Difference potential: target(X) in BX-70 task, cue(A) minus target(X) in BX-70 task, cue(B). Parietal dipoles also included. Conventions as in B. (F) Target No-Go: difference potential: target(Y) in AX-70 task, cue(A) minus target(Y) in AX-70 task, cue(B). Notice that in both cases where there is a No-Go situation (C and F) the potential are represented more anteriorly than when there is a change in task expectation, but is a Go situation (B and E).

Figure 5.

Temporal evolution of the ERP throughout the delay period (Cue On = 0 ms), illustrating the late negativity. The scalp distribution of potentials at 1100 ms after presentation of the cue (dashed line), i.e. just before the presentation of the target, is illustrated for the cases where the cue was Go. Tasks represented: AX-70, AY-70, BX-70 and XX-CPT.

Figure 5.

Temporal evolution of the ERP throughout the delay period (Cue On = 0 ms), illustrating the late negativity. The scalp distribution of potentials at 1100 ms after presentation of the cue (dashed line), i.e. just before the presentation of the target, is illustrated for the cases where the cue was Go. Tasks represented: AX-70, AY-70, BX-70 and XX-CPT.

Figure 6.

(A) Comparison of grand mean potentials recorded after presentation of the target in all three AX-CPT tasks for each type of trial. All four types of trials were presented in each task, in pseudo-random order and following predetermined probabilities. Arrowheads under potentials in AX trials mark the mean response reaction time for each task. (B) Scalp distribution of the potentials at six time points (200 to 700 ms, at 100 ms intervals). Scale represents activities from −12 to 20 μV. Other conventions as in Figure 3.

(A) Comparison of grand mean potentials recorded after presentation of the target in all three AX-CPT tasks for each type of trial. All four types of trials were presented in each task, in pseudo-random order and following predetermined probabilities. Arrowheads under potentials in AX trials mark the mean response reaction time for each task. (B) Scalp distribution of the potentials at six time points (200 to 700 ms, at 100 ms intervals). Scale represents activities from −12 to 20 μV. Other conventions as in Figure 3.

Figure 7.

Top: Grand mean potentials recorded after presentation of the target in control tasks CPT, XX-CPT and Identical Pairs. Bottom: Scalp distribution of these potentials. All conventions as in Figure 6.

Top: Grand mean potentials recorded after presentation of the target in control tasks CPT, XX-CPT and Identical Pairs. Bottom: Scalp distribution of these potentials. All conventions as in Figure 6.

References

Barbas H (
2000
) Connections underlying the synthesis of cognition, memory, and emotion in primate prefrontal cortices.
Brain Res Bull
 
52
:
319
–330.
Barch DM, Braver TS, Akbudak E, Conturo T, Ollinger J, Snyder A (
2001
) Anterior cingulate cortex and response conflict: effects of response modality and processing domain.
Cereb Cortex
 
11
:
837
–848.
Barch DM, Carter CS, Braver TS, Sabb FW, MacDonald A, 3rd, Noll DC, Cohen JD (
2001
) Selective deficits in prefrontal cortex function in medication-naive patients with schizophrenia.
Arch Gen Psychiatry
 
58
:
280
–288.
Braver TS, Barch DM, Gray JR, Molfese DL, Snyder A (
2001
) Anterior cingulate cortex and response conflict: effects of frequency, inhibition and errors.
Cereb Cortex
 
11
:
825
–836.
Burman DD, Bruce CJ (
1997
) Suppression of task-related saccades by electrical stimulation in the primate’s frontal eye field.
J Neurophysiol
 
77
:
2252
–2267.
Carter CS, Braver TS, Barch DM, Botvinick MM, Noll D, Cohen JD (
1998
) Anterior cingulate cortex, error detection, and the online monitoring of performance.
Science
 
280
:
747
–749.
Carter CS, Macdonald AM, Botvinick M, Ross LL, Stenger VA, Noll D, Cohen JD (
2000
) Parsing executive processes: strategic vs. evaluative functions of the anterior cingulate cortex.
Proc Natl Acad Sci USA
 
97
:
1944
–1948.
Casey BJ, Forman SD, Franzen P, Berkowitz A, Braver TS, Nystrom LE, Thomas KM, Noll DC (
2001
) Sensitivity of prefrontal cortex to changes in target probability: a functional MRI study.
Hum Brain Mapp
 
13
:
26
–33.
Cohen JD, Braver TS, O’Reilly RC (
1996
) A computational approach to prefrontal cortex, cognitive control and schizophrenia: recent developments and current challenges.
Philos Trans R Soc Lond B Biol Sci
 
351
:
1515
–1527.
Cohen JD, Barch DM, Carter C, Servan-Schreiber D (
1999
) Context-processing deficits in schizophrenia: converging evidence from three theoretically motivated cognitive tasks.
J Abnorm Psychol
 
108
:
120
–133.
Cui RQ, Deecke L (
1999
) High resolution DC-EEG analysis of the Bereitschaftspotential and post movement onset potentials accompanying uni- or bilateral voluntary finger movements.
Brain Topogr
 
11
:
233
–249.
Cunnington R, Windischberger C, Deecke L, Moser E (
2002
) The preparation and execution of self-initiated and externally-triggered movement: a study of event-related fMRI.
Neuroimage
 
15
:
373
–385.
Deecke L, Scheid P, Kornhuber HH (
1969
) Distribution of readiness potential, pre-motion positivity, and motor potential of the human cerebral cortex preceding voluntary finger movements.
Exp Brain Res
 
7
:
158
–168.
Dias EC, Foxe JJ, Higgins BA, Schroeder CE, Javitt DC (
2001
) Physiological recordings of contextual memory.
Schizophr Res
 
49
:
201
.
Durston S, Thomas KM, Worden MS, Yang Y, Casey BJ (
2002
) The effect of preceding context on inhibition: an event-related fMRI study.
Neuroimage
 
16
:
449
–453.
Elvevag B, Weinberger DR, Suter JC, Goldberg TE (
2000
) Continuous performance test and schizophrenia: a test of stimulus–response compatibility, working memory, response readiness, or none of the above?
Am J Psychiatry
 
157
:
772
–780.
Falkenstein M, Koshlykova NA, Kiroj VN, Hoormann J, Hohnsbein J (
1995
) Late ERP components in visual and auditory Go/Nogo tasks.
Electroencephalogr Clin Neurophysiol
 
96
:
36
–43.
Falkenstein M, Hoormann J, Hohnsbein J (
1999
) ERP components in Go/Nogo tasks and their relation to inhibition.
Acta Psychol (Amst)
 
101
:
267
–291.
Filipovic SR, Jahanshahi M, Rothwell JC (
1999
) Cortical potentials related to decision-making: comparison of two types of go/no-go decision.
Neuroreport
 
10
:
3583
–3587.
Fitzpatrick P, Klorman R, Brumaghim JT, Keefover RW (
1988
) Effects of methylphenidate on stimulus evaluation and response processes: evidence from performance and event-related potentials.
Psychophysiology
 
25
:
292
–304.
Fuchs M, Drenckhahn R, Wischmann HA, Wagner M (
1998
) An improved boundary element method for realistic volume-conductor modeling.
IEEE Trans Biomed Eng
 
45
:
980
–997.
Gehring WJ, Knight RT (
2000
) Prefrontal–cingulate interactions in action monitoring.
Nat Neurosci
 
3
:
516
–520.
Gemba H, Sasaki K (
1989
) Potential related to no-go reaction of go/no-go hand movement task with color discrimination in human.
Neurosci Lett
 
101
:
263
–268.
Gemba H, Sasaki K (
1990
) Potential related to no-go reaction in go/no-go hand movement with discrimination between tone stimuli of different frequencies in the monkey.
Brain Res
 
537
:
340
–344.
Gemba H, Sasaki K, Brooks VB (
1986
) ‘Error’ potentials in limbic cortex (anterior cingulate area 24) of monkeys during motor learning.
Neurosci Lett
 
70
:
223
–227.
Hamano T, Luders HO, Ikeda A, Collura TF, Comair YG, Shibasaki H (
1997
) The cortical generators of the contingent negative variation in humans: a study with subdural electrodes.
Electroencephalogr Clin Neurophysiol
 
104
:
257
–268.
Iversen SD, Mishkin M (
1970
) Perseverative interference in monkeys following selective lesions of the inferior prefrontal convexity.
Exp Brain Res
 
11
:
376
–386.
Javitt DC, Shelley AM, Silipo G, Lieberman JA (
2000
) Deficits in auditory and visual context-dependent processing in schizophrenia: defining the pattern.
Arch Gen Psychiatry
 
57
:
1131
–1137.
Kiefer M, Marzinzik F, Weisbrod M, Scherg M, Spitzer M (
1998
) The time course of brain activations during response inhibition: evidence from event-related potentials in a go/no go task.
Neuroreport
 
9
:
765
–770.
Knight RT, Scabini D (
1998
) Anatomic bases of event-related potentials and their relationship to novelty detection in humans.
J Clin Neurophysiol
 
15
:
3
–13.
Liddle PF, Kiehl KA, Smith AM (
2001
) Event-related fMRI study of response inhibition.
Hum Brain Mapp
 
12
:
100
–109.
Loveless NE, Sanford AJ (
1974
) Slow potential correlates of preparatory set.
Biol Psychol
 
1
:
303
–314.
McCarthy G, Blamire AM, Puce A, Nobre AC, Bloch G, Hyder F, Goldman-Rakic P, Shulman RG (
1994
) Functional magnetic resonance imaging of human prefrontal cortex activation during a spatial working memory task.
Proc Natl Acad Sci USA
 
91
:
8690
–8694.
McCarthy G, Puce A, Constable RT, Krystal JH, Gore JC, Goldman-Rakic P (
1996
) Activation of human prefrontal cortex during spatial and nonspatial working memory tasks measured by functional MRI.
Cereb Cortex
 
6
:
600
–611.
Murray MM, Foxe JJ, Higgins BA, Javitt DC, Schroeder CE (
2001
) Visuo-spatial neural response interactions in early cortical processing during a simple reaction time task: a high-density electrical mapping study.
Neuropsychologia
 
39
:
828
–844.
Murray MM, Wylie GR, Higgins BA, Javitt DC, Schroeder CS, Foxe JJ (
2002
) The spatiotemporal dynamics of illusory contour processing: combined high-density electrical mapping, source analysis, and functional magnetic resonance imaging.
J Neurosci
 
22
:
5055
–5073.
Oldfield RC (
1971
) The assessment and analysis of handedness: the Edinburgh inventory.
Neuropsychologia
 
9
:
97
–113.
Pandya DN, Van Hoesen GW, Mesulam MM (
1981
) Efferent connections of the cingulate gyrus in the rhesus monkey.
Exp Brain Res
 
42
:
319
–330.
Pardo JV, Pardo PJ, Janer KW, Raichle ME (
1990
) The anterior cingulate cortex mediates processing selection in the Stroop attentional conflict paradigm.
Proc Natl Acad Sci USA
 
87
:
256
–259.
Petit L, Courtney SM, Ungerleider LG, Haxby JV (
1998
) Sustained activity in the medial wall during working memory delays.
J Neurosci
 
18
:
9429
–9437.
Picard N, Strick PL (
1996
) Motor areas of the medial wall: a review of their location and functional activation.
Cereb Cortex
 
6
:
342
–353.
Rebert CS (
1972
) Cortical and subcortical slow potentials in the monkey’s brain during a preparatory interval.
Electroencephalogr Clin Neurophysiol
 
33
:
389
–402.
Riccio CA, Reynolds CR (
2001
) Continuous performance tests are sensitive to ADHD in adults but lack specificity. A review and critique for differential diagnosis.
Ann N Y Acad Sci
 
931
:
113
–139.
Rosahl SK, Knight RT (
1995
) Role of prefrontal cortex in generation of the contingent negative variation.
Cereb Cortex
 
5
:
123
–134.
Rosvold HE, Mirsky AF, Sarason I, Bransome ED, Beck LH (
1956
) A continuous performance test of brain damage.
J Consult Psychol
 
20
:
343
–350.
Ruchkin DS, Canoune HL, Johnson R, Jr, Ritter W (
1995
) Working memory and preparation elicit different patterns of slow wave event-related brain potentials.
Psychophysiology
 
32
:
399
–410.
Rutschmann J, Cornblatt B, Erlenmeyer-Kimling L (
1977
) Sustained attention in children at risk for schizophrenia. Report on a continuous performance test.
Arch Gen Psychiatry
 
34
:
571
–575.
Saron CD, Schroeder CE, Foxe JJ, Vaughan HG, Jr. (
2001
) Visual activation of frontal cortex: segregation from occipital activity.
Brain Res Cogn Brain Res
 
12
:
75
–88.
Sasaki K, Gemba H, Tsujimoto T (
1989
) Suppression of visually initiated hand movement by stimulation of the prefrontal cortex in the monkey.
Brain Res
 
495
:
100
–107.
Sasaki K, Gemba H, Nambu A, Matsuzaki R (
1993
) No-go activity in the frontal association cortex of human subjects.
Neurosci Res
 
18
:
249
–252.
Scherg M, Picton TW (
1991
) Separation and identification of event-related potential components by brain electric source analysis.
Electroencephalogr Clin Neurophysiol Suppl
 
42
:
24
–37.
Servan-Schreiber D, Cohen JD, Steingard S (
1996
) Schizophrenic deficits in the processing of context. A test of a theoretical model.
Arch Gen Psychiatry
 
53
:
1105
–1112.
Simson R, Vaughan HG, Jr, Ritter W (
1977
) The scalp topography of potentials in auditory and visual Go/NoGo tasks.
Electroencephalogr Clin Neurophysiol
 
43
:
864
–875.
Talairach J, Tournoux P (
1988
) Co-planar stereotaxic atlas of the human brain. Stuttgart: Thieme.
Tekok-Kilic A, Shucard JL, Shucard DW (
2001
) Stimulus modality and Go/NoGo effects on P3 during parallel visual and auditory continuous performance tasks.
Psychophysiology
 
38
:
578
–589.
Umbricht D, Schmid L, Koller R, Vollenweider FX, Hell D, Javitt DC (
2000
) Ketamine-induced deficits in auditory and visual context-dependent processing in healthy volunteers: implications for models of cognitive deficits in schizophrenia.
Arch Gen Psychiatry
 
57
:
1139
–1147.
Walter W, Cooper R, Aldridge V, McCallum W, Winter A (
1964
) Contingent negative variation: an electrical sign of sensorimotor association and expectancy in the human brain.
Nature
 
203
:
380
–384.
Yazawa S, Ikeda A, Kunieda T, Ohara S, Mima T, Nagamine T, Taki W, Kimura J, Hori T, Shibasaki H (
2000
) Human presupplementary motor area is active before voluntary movement: subdural recording of Bereitschaftspotential from medial frontal cortex.
Exp Brain Res
 
131
:
165
–177.