| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Center for the Neural Basis of Cognition, Mellon Institute, Pittsburgh, Pennsylvania; and 2Department of Neuroscience, University of Pittsburgh, Pittsburgh, Pennsylvania
Submitted 25 March 2004; accepted in final form 29 July 2004
 |
ABSTRACT |
|---|
|
|
|
INTRODUCTION |
|---|
|
). Tasks in which habitual or automatic responses are pitted against intended or voluntary responses include the Stroop task (Stroop 1935), the Eriksen Flanker task (Eriksen and Eriksen 1974), the AX-CPT task (Botvinick et al. 1999), and the antisaccade task (Everling and Fischer 1998; Hallett 1978; Hallett and Adams 1980). In none of these tasks is it possible to suppress completely the tendency to make an automatic or habitual but task-inappropriate response. When automatic and habitual tendencies oppose correct performance, longer reaction times and higher error rates are the result.
It has recently been proposed that the brain actively compensates for conflict by a series of steps originating with the detection of conflict by anterior cingulate cortex (ACC) and culminating in heightened activation of cortical networks responsible for maintaining task set (Botvinick et al. 1999, 2001; Braver et al. 2001; Carter et al. 1998, 2000; Cohen et al. 2000; van Veen et al. 2001). This proposal is compatible with a large number of human functional MRI (fMRI) studies showing that conflict elicits a BOLD response in the ACC (Badgaiyan and Posner 1998; D'Esposito et al. 1995; Doricchi 1997; Fan et al. 2003; Frith et al. 1991; Jenkins et al. 1994; Jueptner et al. 1997; Kerns et al. 2004; Merriam et al. 2001; O'Driscoll 1995; Pardo 1990; Paus et al. 1993, 1998; Petersen et al. 1988; Petit et al. 1998; Raichle et al. 1994; Sweeney et al. 1996). The finding of conflict-specific BOLD activation in the ACC is, however, merely suggestive and does not constitute conclusive evidence for involvement of the area in monitoring conflict. This is for two reasons. First, BOLD signals may be more tightly yoked to synaptic events than to local spiking activity (Logothetis and Wandell 2004). Second, conflict-related activity observed at the population level could arise from at least two different sources at the single-neuron level. On one hand, the ACC might contain conflict-monitoring neurons: neurons selectively active in the presence of conflict. On the other hand, it might contain neurons active in conjunction with planning particular actions. The net activity of this population might increase under conflict because subpopulations representing opposed actions are simultaneously active or, in other words, because conflict is present and not because conflict is being monitored.
The point that population activity could increase under conflict without an area's containing neurons that monitor conflict merits close consideration. Conflict, by definition, involves the co-activation of neuronal populations representing incompatible behavioral responses (Botvinick et al. 2001). Simultaneous activation of neurons representing opposed responses (under conditions of high conflict) might well sum to a higher level of population activity than exclusive activation of neurons representing a single response (under conditions of low conflict). To refer to the resulting increase as a "conflict monitoring" signal would be meaningless because the co-activation of neurons representing incompatible responses constitutes conflict and is thus the thing that must be monitored. Those who have speculated about conflict monitoring at the neuronal level accordingly have envisioned neurons carrying pure conflict-monitoring signals: neurons that do not represent the behavioral responses pitted against each other and yet do show an increase of firing rate when conflict is present.
To characterize conflict-related activity requires recording from single neurons while monkeys perform tasks in which, unpredictably, on any given trial, the level of conflict may be high or low. To date, this approach has been taken in studies of two adjacent areas in the medial frontal lobe: the supplementary eye field (SEF) and the ACC. In the SEF, three studies have revealed signs of conflict-related activity. 1) When monkeys perform a countermanding task in which presentation of a peripheral saccadic target is followed, on some trials, by a foveal cue instructing that the saccade be canceled, some SEF neurons fire at the instant of successful cancellation (Stuphorn et al. 2000). This pattern of neuronal activity has been taken as evidence for neuronal conflict monitoring, on the assumption that it arises from conflict between the command to execute a saccade and the command to maintain fixation, but it could depend on several other uncontrolled factors that distinguish trials in which a saccade is canceled from those in which it is not. 2) When monkeys perform a color-conditional saccade task, in which the direction of the required saccade is instructed by the color of a peripheral cue, SEF neuronal activity is higher overall on trials requiring a saccade away from the cue's location than on trials in which the location of the cue and the location of the target coincide (Olson and Gettner 2002). This effect does not, however, arise from the activity of neurons with a circumscribed "conflict-monitoring" function but rather takes the form of a subtle modulation of firing rate among neurons with direction-selective task-related activity. 3) When monkeys prepare to make an antisaccade, SEF neurons are more active than when they prepare to make a prosaccade (Amador et al. 20004). This effect does not involve conflict monitoring but rather preparation to follow a rule (the antisaccade rule) that incidentally involves the experience of conflict. In the ACC, a single study, based on the countermanding paradigm, has failed to reveal any sign of conflict-related activity (Ito et al. 2003).
The interpretations of studies carried out to date by means of single-neuron recording in the SEF and ACC is problematic for two reasons. 1) They did not incorporate direct behavioral measures of conflict, based on errors and reaction time, such as are commonly used in human studies, although indirect measures in the countermanding task did indicate that conflict was present during successful countermanding. In the absence of these measures, it is conceivable that the monkeys, highly overtrained as they were, had acquired a capacity for early suppression of competing response tendencies and thus were not subject to conflict. To circumvent this problem, we carried out recording in the context of tasks that allowed direct behavioral measures of conflict. 2) Each study employed a single behavioral paradigm. Thus conflict-related activity might have depended on circumstances incidentally correlated with conflict in the particular task. To circumvent this problem, we carried out recording under conditions in which conflict was induced by two different experimental manipulations. The monkeys performed a color-conditional eye movement task in which conflict was produced either through spatial incompatibility between the cue's location and the direction of the required response or through reversal of the instruction.
|
|
METHODS |
|---|
|
Two adult rhesus monkeys were used (Macaca mulatta; laboratory designations Nb and Al, male and female, respectively). Experimental procedures were approved by the Carnegie Mellon University Animal Care and Use Committee and were in compliance with the guidelines set forth in the United States Public Health Service Guide for the Care and Use of Laboratory Animals.
Preparatory surgery
At the outset of the training period, each monkey underwent sterile surgery under general anesthesia induced with ketamine (20 mg/kg) and diazepam (1 mg/kg) and maintained with isofluorane inhalation. The top of the skull was exposed, bone screws were inserted around the perimeter of the exposed area, a continuous cap of rapidly hardening acrylic was laid down so as to cover the skull and embed the heads of the screws, a head-restraint bar was embedded in the cap, and scleral search coils were implanted on the eyes, with the leads directed subcutaneously to plugs on the acrylic cap (Remmel 1984; Robinson 1963). Following training, a 2-cm-diam disk of acrylic and skull, centered on the midline of the brain at anterior 29 (monkey N) or 35 mm (monkey A) in Horsley-Clarke coordinates, was removed, and a cylindrical recording chamber (Crist Instruments) was cemented into the hole with its base just above the exposed dural membrane.
Single-neuron recording
At the beginning of each day's session, a varnish-coated tungsten microelectrode with an initial impedance of several megohms at 1 kHz (Frederick Haer and Co., Bowdoinham, ME) was advanced vertically through the dura into the immediately underlying cortex. The electrode could be placed reproducibly at points forming a square grid with 1-mm spacing (Crist et al. 1988). The action potentials of a single neuron were isolated from the multineuronal trace by means of an on-line spike-sorting system using a template-matching algorithm (Signal Processing Systems, Prospect, Australia). The spike-sorting system, on detection of an action potential, generated a pulse that was stored with 1-ms resolution.
Experimental control and data collection
All aspects of the behavioral experiment, including presentation of stimuli, monitoring of eye movements, monitoring of neuronal activity, and delivery of reward, were under the control of a 486- or pentium-based computer running Cortex software provided by R. Desimone, Laboratory of Neuropsychology, National Institute of Mental Health. Eye position was monitored by means of a scleral search coil system (Riverbend Instruments, Birmingham, AL) and the x and y coordinates of eye position were stored with 10-ms resolution, during the period beginning after the presentation of the fixation point to 1,000 ms after the end of the reward period. Stimuli generated by an active matrix LCD projector were rear-projected on a frontoparallel screen 25 cm from the monkey's eyes. Reward in the form of 
0.1 ml of water or juice was delivered through a spigot under control of a solenoid valve on successful completion of each trial.
Spatial incompatibility task
The aim of the spatial incompatibility task was to pit the instructed response (select the left or right target) against the habitual tendency to look at the cue (which might appear either at the left or right). Essential features of the task are summarized in Fig. 1, A and B. Each trial began with presentation of a central fixation spot followed by the appearance of two potential targets at the neuron's preferred and antipreferred locations. A color cue then appeared in superimposition on one of the targets. A red (or green) cue instructed a response to the left (or right). On "compatible" trials, the cue occupied the location to which it instructed the monkey to respond; on "incompatible" trials, it occupied the opposite location. Simultaneously with appearance of the cue, the central fixation spot was extinguished. This instructed the monkey to make a saccade directly to the instructed target. On the eye's reaching the target, a feedback stimulus (of the same color as the cue) appeared in superimposition on the target. The monkey was required to maintain fixation at the target location for a variable interval in the range 300–450 ms before reward was delivered.
|
The sequence of conditions within a block was random subject to the constraint that one trial conforming to each condition had to be completed successfully before any condition could be repeated. Due to this nonrandom constraint, the nature of an impending trial could be predicted from the nature of the preceding trial at a level above chance. In particular, there was a relatively high probability (0.625) that the next trial would require a response in the opposite direction or would involve the opposite compatibility status. Even if monkeys developed an alternation set appropriate to the statistics of the sequence, this would not, however, have affected the analysis of neuronal effects due to conflict because the tendency for direction to alternate was the same under both compatible and incompatible conditions.
Reversal task
In this task as well, the monkey followed a color conditional rule, making a leftward (or rightward) saccade in response to a red (or green) cue. Conflict was induced by reversing the cue's color on some trials. The final color of the cue carried the instruction concerning which target to select. We reasoned that the monkey would begin to prepare a saccade on the basis of the cue's initial color and that the traces of this act would carry over into the ensuing phase of the trial, creating conflict. Essential features of the task are summarized in Fig. 2, A and B. Each trial began with presentation of a central fixation spot followed by the appearance of two potential targets at the neuron's preferred and antipreferred locations. A color cue then appeared at fixation. On "compatible" trials, the cue's color was consistent over an interval of 200 ms; following this, it was extinguished. On "incompatible" trials, the cue was of one color for 100 ms and then switched to the other color for 100 ms; following this, it was extinguished. The monkey was required to make a saccade directly to the target associated with the color of the cue on compatible trials and to the target associated with the cue's final color on incompatible trials. On the eye's reaching the target, a feedback stimulus (of the color associated with the target) appeared in superimposition on the target. The monkey was required to maintain fixation at the target location for a variable interval in the range 300–450 ms. Trials conforming to four conditions (red-red, green-green, red-green, and green-red) were presented in pseudorandom sequence subject to the constraint that one trial of each type had to be completed successfully before any condition could be repeated. Because of this constraint, the sequence of trial conditions fell short of being fully random, as discussed in connection with the spatial incompatibility task. Data collection continued until 16 trials had been competed successfully under each condition.
|
The fixation spot was a white, 0.38° square presented at the center of the screen. Targets were 0.38° white squares presented 10° from central fixation. The cue and feedback stimuli were 0.96° white, red, or green disks. The background of the display had a luminance of 6.9 cd/m2 and CIE x and y chromaticity coefficients of 0.27 and 0.31, respectively. White stimuli had a luminance of 193 cd/m2 and CIE x and y chromaticity coefficients of 0.31 and 0.34, respectively. Red stimuli had a luminance of 79 cd/m2 and CIE x and y chromaticity coefficients of 0.56 and 0.38, respectively. Green stimuli had a luminance of 181 cd/m2 and CIE x and y chromaticity coefficients of 0.28 and 0.61, respectively.
Memory-guided saccade task
Following identification of a neuron exhibiting apparent task-related activity in one of the above tasks, we characterized its best direction. We did this by recording during performance of a memory guided saccade task in which there were six target locations located at 10° eccentricity and distributed around the clock at 60° intervals. Two of the locations were directly to the left and right of fixation; the others were at +60° and –60° in the left and right hemifields. We defined the preferred location as the one associated with strongest firing and the antipreferred location as the one diametrically opposite. In subsequent data collection during performance of tasks involving conflict, targets were placed only at these two locations. Thus one target was always in the left hemifield and one always in the right hemifield. Regardless of absolute location, monkeys were required to select the left target in response to a red cue and the right in response to a green cue.
Behavioral data analysis
To assess the impact of the compatibility-incompatibility manipulation on behavior, we computed two measures for each recording session. 1) Error rate was defined as the number of trials on which a saccade was executed to the wrong target, expressed as a percentage of all trials on which a saccade was executed to either target. 2) Mean reaction time was defined as the average delay, on successful trials, from onset of the instructional cue to the moment when the eye left the central fixation window. For each recording session, we computed the error rate and the mean reaction time independently for compatible and incompatible trials. To assess whether values obtained under compatible and incompatible conditions were significantly different across sessions, we used a paired t-test.
Neuronal data analysis
EPOCHS OF INTEREST. Statistical analysis was based on neuronal firing rates during specific epochs defined relative to cue onset and saccade initiation. Postcue firing rates were measured during three epochs aligned on cue onset, C1 (50–150 ms after cue onset), C2 (150–250 ms after cue onset), and C3 (250–350 ms after cue onset). Perisaccadic firing rates were measured during three epochs aligned on saccade initiation, S1 (100–0 ms before saccade onset), S2 (0–100 ms after saccade onset), and S3 (100–200 ms after saccade onset). Epochs C2 and C3 corresponded roughly to epochs S1 and S2 because the behavioral reaction time was on average around 250 ms. The correspondence was not exact, however, due to reaction-time jitter. The decision to carry out analyses based on epochs aligned both to cue onset and to saccade initiation was dictated by the consideration that significant firing rate effects might be time-locked to either event.
SELECTION OF NEURONS. Neurons were accepted for study if they exhibited task-related activity by the criterion that the firing rate during at least one of the six analysis epochs (C1–C3 and S1–S3) should differ significantly (t-test, P < 0.05) from the firing rate during a baseline period 300–200 ms prior to cue onset. A baseline period well before cue onset was chosen because neurons in the SEF often show an anticipatory buildup of activity immediately before a predictable event. The use of a lenient winnowing criterion (P < 0.05 for 1 of 6 measures) was motivated by the desire to separate the wheat of potentially task-related units from the chaff of probably non–task-related units while not excluding units with weak but potentially interesting forms of task-related activity. Because this test was used simply for winnowing and not as the basis for a claim concerning the percentage of units with task-related activity, the use of a lenient criterion is justified. Each selected neuron's preferred direction (left or right) was defined as the direction associated with a higher average firing rate on compatible trials during the period between cue onset and saccade onset.
STATISTICAL ANALYSIS OF THE DEPENDENCE OF FIRING RATE ON TASK FACTORS.
In all phases of analysis, we focused exclusively on data from correct trials. The pattern of neuronal activity on the rare trials when monkeys selected the wrong target is the subject of a separate study (Nakamura et al. 2002). We carried out independent ANOVAs on data from the six epochs C1–C3 and S1–S3. In each ANOVA, firing rate was the dependent variable, while response direction (right or left) and conflict level (compatible or incompatible) were factors. In assessing main and interaction effects, we employed a criterion for statistical significance of P < 0.05.
Dissociating the neuronal correlates of reaction time and conflict
To determine whether neuronal activity continued to depend on conflict when the effects of behavioral reaction time were factored out, we performed a multivariate regression analysis, fitting three models
 | (1) |
 | (2) |
 | (3) |
 |
0.05. Localization of recording sites
Recording sites were localized relative to major morphological landmarks by use of MR images. Slices of 2 mm thickness, spanning the entire brain, were collected while the anesthetized monkey was supported by an MR-compatible stereotaxic device inside a Brükker 4.7 T magnet. Fiducial markers containing a contrast agent were placed at known locations inside the chamber during the scan. These provided registration for projecting the recording sites onto the images and thus for judging their location relative to landmarks including the interhemispheric cleft and the genu of the corpus callosum. Recording sites were localized relative to functional areas by mapping out the entire recording zone with low-current intracortical microstimulation (200-ms trains of 1.65-ms biphasic pulses at 300 Hz and <100 µA). The SEF was identifiable by the property that microstimulation elicited saccadic eye movements. Skeletomotor areas behind the SEF and ventral to it posteriorly were identifiable by the property that microstimulation elicited arm and trunk movements. A broad zone of cortex anterior and medial to the SEF (categorized here as anterior cingulate) was distinguished by the property that microstimulation did not elicit motor responses.
|
|
RESULTS |
|---|
|
SPATIAL INCOMPATIBILITY TASK. This task was designed to induce conflict by pitting the automatic tendency to look at a suddenly appearing visual cue against the voluntary behavior of looking in the direction associated with its color. Saccades were executed according to a color-conditional rule (red and green instructing leftward and rightward responses, respectively). A cue of either color could be presented at either target location. Thus the location of the cue and the direction of the response were compatible on one-half of trials and incompatible on the other half (Fig. 1B). Both monkeys achieved stable performance on this task, exhibiting consistent effects of incompatibility. The percentage of errors was higher, and the reaction time was longer, under incompatible compared with compatible conditions (Fig. 1, C and D; Table 1). Both the error rate and the reaction time differences were highly significant in each monkey and in the combined data (P < 0.0001, paired t-test on session means across all recording sessions). We conclude that the mismatch between cue location and response direction in incompatible conditions interfered with correct performance.
|
REVERSAL TASK. In the preceding task, neuronal activity specific to incompatible trials might have depended on response conflict (simultaneous preparation of leftward and rightward saccades) or on visuomotor mismatch (presentation of a visual stimulus in one direction followed by preparation of a saccade in the opposite direction). To allow distinguishing between these possibilities, we employed a second task in which visuomotor mismatch was held constant while response conflict varied. In the reversal task, the monkey selected a response according to the color of a foveal cue (leftward for red and rightward for green). Conflict was induced by changing the color of the cue 100 ms after its onset on one-half of the trials, with the monkey required to select the direction of the saccade on the basis of the final color (Fig. 2, A and B). Both monkeys exhibited consistent effects of incompatibility in this task (Fig. 2, C and D; Table 2). The behavioral reaction time was significantly longer and the error rate was significantly higher under incompatible compared with compatible conditions. These effects achieved a high level of significance in data from each monkey, as well as in the combined data (P < 0.0001, paired t-test on session means across all recording sessions).
|
The SEF, identified as a region in which low-threshold microstimulation reliably elicited saccades, was centered at a frontal level corresponding approximately to the genu of the corpus callosum (32 mm anterior to the interaural plane in monkey N and 33 mm anterior in monkey A). Data were collected from any neuron that seemed to exhibit task-related activity during any phase of performance of the task under study (either the spatial incompatibility or the reversal task). The data were included in the SEF data base if the recording site was in the region defined above and the mean firing rate during at least one of six trial epochs defined in METHODS (C1–C3, aligned to cue, and S1–S3, aligned to saccade) differed significantly (t-test, P < 0.05) from the baseline rate measured during a period 300–200 ms prior to cue onset. These criteria would be met by a neuron exhibiting pure movement-related activity, pure conflict-related activity, or any mix of the two. They were satisfied by 344 neurons (232 in monkey N and 112 in monkey A) studied in the context of the spatial compatibility task and by 253 neurons (142 in monkey N and 111 in monkey A) studied in the context of the reversal task. Out of these neurons, 221 (127 in monkey N and 94 in monkey A) were studied in the context of both tasks. The SEF is indicated by yellow highlighting and neurons in the SEF database are indicated by red symbols set against the yellow backdrop in Fig. 3. The indicated locations are approximate. The coordinates of each site were computed on the basis of the assumption that the electrode advanced precisely vertically into cortex. On passing through the dural membrane, however, the electrode was subject to some degree of deflection. This accounts for the fact that some recording sites are shown as in white matter although all were in gray matter as indicated by the nature of the background activity.
|
INTRODUCTION.
The central goal of this study was to determine whether SEF neurons exhibited different firing patterns when the cue and response were spatially incompatible versus compatible. To resolve this issue, we analyzed neuronal data from trials on which the monkeys made a correct response. Data from the rare trials on which they chose the wrong target are the subject of a separate analysis concerned with error-related activity (Nakamura et al. 2002). On casual inspection of histograms, we observed that effects of incompatibility were common but that they varied over the course of the interval between cue and response. Representative data from one neuron are shown in Fig. 4. Even prior to presentation of the directional cue, this neuron's activity ramped up raggedly, giving rise in some histograms to the appearance (Fig. 4, A and C) that there was a visual response at an impossibly short latency (SEF neurons frequently exhibit a nonspecific increase in firing rate in advance of the occurrence of an expected cue). However, our comments focus on the period beginning 75 ms after cue onset when activity began to be modulated in response to the cue. On trials requiring a response in the preferred direction (defined on the basis of data from compatible trials), this neuron began (before the saccade) by firing more strongly under the compatible condition (Fig. 4, A and C) but ended (after the saccade) by firing more strongly under the incompatible condition (Fig. 4, B and D). On trials requiring a response in the antipreferred direction, firing under the incompatible condition (Fig. 4, G and H) became stronger than firing under the compatible condition (Fig. 4, E and F) late in the cue-response period. All of the noted effects were significant. These observations make clear that any analysis of the impact of compatibility on neuronal activity must take into account time during the trial. Accordingly, in analyzing data from the entire recorded population, we adopted time-specific measures in the form of population histograms and statistical analyses confined to brief (100 ms) epochs locked to cue onset or saccade initiation.
|
|
250 ms, C2 and C3 corresponded roughly to S1 and S2. The results of this analysis are presented in full in Table 3 and are summarized in the next two paragraphs.
|
2 test to compare the observed counts to the counts predicted by chance (estimated conservatively at 5%). We found that the counts of such neurons were significantly above chance in all epochs (black bars representing counts greater than expected by chance at a criterion of P < 0.05 are marked by asterisks in Fig. 6A). In contrast, the count of cases in which direction selectivity was enhanced under the incompatible condition did not attain significance in any epoch (Fig. 6A, white bars). We conclude that the reduction of directional signal that was evident in the population at large also showed up as a statistically significant interaction effect in a substantial fraction of neurons. We note further that this pattern was most common in epochs (C2 and S1) during which the reduction of the population directional signal was maximal (Fig. 5, C and D).
|
2 test, P < 0.05) during all measurement epochs except C1. The frequency of significant effects peaked in late epochs (C3 and S2) during which the enhancement of net activity in the population was maximal (Fig. 5, C and D). Surprisingly, we also found that neurons firing significantly less under incompatible than under compatible conditions were present in numbers significantly greater than expected by chance during two epochs (C2 and S1); however, these counts barely exceeded the significance threshold. We conclude that the enhancement of net activity evident in the population at large was also present as a statistically significant effect in a substantial fraction of neurons.
FIRING-RATE-BASED MEASURES IN INDIVIDUAL NEURONS.
The statistical measures described above were coarse in the sense that they categorized neurons by use of an arbitrary cut-off for significance. To assess the robustness of the observed effects, we carried out a second analysis based on firing rate without regard to significance. We focused in this analysis on epoch S1 (100 ms before saccade onset), during which the "reduction of directional signal" effect was strong, and epoch S2 (100 ms after saccade onset), during which the "enhancement of net activity" effect was strong. Results for presaccadic epoch S1 are plotted in Fig. 6, C and D. When the impending saccade was in the neuron's preferred direction, firing tended to be less on incompatible than on compatible trials (Fig. 6C). This effect was highly significant (2 test, P < 0.0001). When the impending saccade was in the neuron's antipreferred direction, firing tended to be greater on incompatible than on compatible trials (Fig. 6D). This effect was also highly significant (P < 0.0001). These effects, taken together, indicate that there was a decrease in the directional signal but little change in net activity during epoch S1, just as suggested by previous measures. Earlier in the trial, during epoch C1, 50–150 ms after cue-onset, the same form of analysis gave rise to closely similar results. Results for postsaccadic epoch S2 are plotted in Fig. 6, E and F. Regardless of whether the impending saccade was in the neuron's preferred direction (Fig. 6E) or in its antipreferred direction (Fig. 6F), firing tended to be greater on incompatible than on compatible trials. The increase was highly significant (2 test, P < 0.0001) for both directions. This pattern indicates that there was a net enhancement in firing rate but little change in the directional signal during epoch S2, just as suggested by previous measures.
FUNCTIONAL PROPERTIES OF "CONFLICT MONITORING" NEURONS. It seemed possible that neurons "monitoring conflict" (showing a significant enhancement of net activity on trials involving high conflict) might form a distinct population lacking functional attributes, such as direction selectivity, characteristic of other SEF neurons. To assess this possibility, we divided neurons into categories exhibiting or not exhibiting a significant enhancement of net activity on high-conflict trials during the period from 0 to 300 ms after saccade initiation, when the enhancement effect was strong at the level of the entire population (Fig. 5D, gray curve). Then we constructed population histograms representing firing rate as a function of time for the enhancing and nonenhancing populations (Fig. 7, A–D). It is clear from these histograms that the statistically based cut between the two categories was effective: the conflict-monitoring population showed a large and prolonged enhancement of firing rate on high-conflict trials (Fig. 7B, gray curve), whereas the other population showed almost no trace of such an effect (Fig. 7D, gray curve). Restricting consideration to trials in which the location of the cue and the direction of the response were compatible, we asked whether the pattern of task-related activity was different in the two populations. We found that both populations exhibited robust, direction selective activity prior to onset of the saccade. We conclude that "conflict monitoring" neurons do not constitute a sharply distinct class sensitive to conflict alone. There were, however, several apparent differences between the populations. 1) Phasic activity preceding the saccade was stronger in the enhancing population (the thick gray and thick black curves attain a higher level in Fig. 7A than in Fig. 7B). 2) Following execution of the saccade, the firing rate of the enhancing population dropped to a lower level (the thick gray and thick black curves are lower at trial's end in Fig. 7A than in Fig. 7B). 3) Signals reflecting the direction of the saccade were truncated earlier in the enhancing population (the thick gray and thick black curves converge at 0 ms in Fig. 7A compared with 200 ms in Fig. 7B). Thus neurons "monitoring conflict," although by no means constituting a "pure" conflict-monitoring population, were set apart from other neurons by several functional properties in addition to their conflict-sensitivity.
|
; Tremblay et al. 2002). Nevertheless, we assessed whether direction selectivity was genuinely present by comparing data collected from each neuron on compatible color-conditional trials to data collected on spatial trials. The spatial attributes of cues and responses were the same in spatial and compatible color-conditional trials (Fig. 1B), although the colors of the cues and the rule for response selection were different. On examining population histograms (Fig. 8, A and B), we found that neurons exhibited closely similar patterns of task-related activity on spatial and compatible color-conditional trials. However, in accordance with a previous report (Olson and Gettner 2002), the mean firing rate between cue-onset and saccade-initiation was moderately enhanced on compatible color-conditional trials as compared with spatial trials (mean = 15.5 vs. 15.0 spikes/s, P < 0.05, paired t-test). For each neuron, we computed a directional index for each condition: (R – L)/(R + L), where R and L were the firing rates between cue-onset and saccade-initiation on trials requiring rightward and leftward responses, respectively. There was a trend for directional indices measured under compatible conditions to be greater than those measured under spatial conditions (mean = 5.0 vs. 4.6 spikes/s, P < 0.05, paired t-test). The main finding, however, was that directional indices under spatial and color conditions (Fig. 8C) were strongly correlated (monkey N: r = 0.59, P < 0.0001; monkey A: r = 0.72, P < 0.0001). We conclude that SEF neurons were selective for the response direction associated with the cue's color rather than for the color itself.
|
INTRODUCTION. Conflict-related activity in the preceding task might have been dependent on response conflict (simultaneous activation of two incompatible motor programs) or on visual-motor incompatibility (simultaneous processing of a visual cue at one location and programming of a motor response in the opposite direction). To distinguish between these possibilities, we monitored neuronal activity in the context of a second task, the reversal task, in which visual-motor incompatibility was held constant across conditions but response conflict varied (Fig. 2, A and B). Conflict-related activity continued to be observable in the context of this second task, as shown by data from one neuron in Fig. 9. Time, in this and all subsequent figures concerned with the reversal task, is defined relative to the instant of the "switch," the time at which the display first conveyed information about the required response, either through color-reversal on incompatible trials or through failure to reverse on compatible trials. The impact of conflict on the firing of this neuron was especially evident on trials requiring a response in the antipreferred direction (defined on the basis of data from compatible trials). It fired more strongly under the incompatible (Fig. 9, G and H) than under the compatible (Fig. 9, E and F) condition because early onset of a cue instructing a response in the preferred direction stirred up activity that persisted despite reversal of the cue's color at "switch" time. More subtle features of its pattern of activity, such as the dip between early and late phases of strong activity on nonreversing trials requiring that the saccade is in the preferred direction (Fig. 9A), are of uncertain significance.
|
|
2 test, P < 0.05) during epochs between the cue and initiation of the saccade (C1, C2, and S1). Cases in which neuronal activity was significantly enhanced under conflict (Fig. 11B, black bars) were significantly more common than expected by chance (2 test, P < 0.05) during all but the latest postsaccadic analysis epoch (S2). Surprisingly, during certain epochs, there was also an above-chance rate of incidence of cases in which the net firing rate was attenuated under conflict (Fig. 11B, white bars for epochs C2, S1, and S2). Overall, however, cases of enhancement were more numerous, in harmony with the observation of enhancement in the population histograms.
|
|
2 test, P < 0.0001) during epoch S1 when the saccade was in the neuron's antipreferred direction (Fig. 11D). This confirms the qualitative pattern seen in the population histograms, whereby, during epoch S1 of trials requiring a saccade in the antipreferred direction, firing was stronger under the incompatible than under the compatible condition (Fig. 10B, thin vs. thick black curve). FUNCTIONAL PROPERTIES OF "CONFLICT MONITORING" NEURONS. We divided neurons into categories exhibiting or not exhibiting a significant enhancement of net activity on high-conflict trials during the period from 300 to 0 ms before saccade initiation, when the enhancement effect was strong at the level of the entire population (Fig. 10D, gray curve). Then we constructed population histograms representing firing rate as a function of time for the two populations (Fig. 12, A–D). The population histograms shown here are aligned on saccade onset but the essential results were the same with alignment on cue onset. On trials in which the location of the cue and the direction of the response were compatible, we found that the enhancing population exhibited robust, direction selective activity around the period of the saccade. Thus "conflict monitoring" neurons did not constitute a sharply distinct class sensitive to conflict alone. They were, however, distinguished as a population by several functional traits evident in the context of compatible trials (Fig. 12, A and B, thick gray and black lines). 1) Phasic activity preceding the saccade was much stronger in the enhancing population (the thick gray and thick black curves attain a higher level in Fig. 12A than in Fig. 12B). 2) Following execution of the saccade, the firing rate of the enhancing population dropped to a lower level (the thick gray and thick black curves are lower at trial's end in Fig. 7A than in Fig. 7B). 3) Signals reflecting the direction of the saccade were truncated earlier in the enhancing population (the thick gray and thick black curves converge at 150 ms in Fig. 12A compared with 350 ms in Fig. 12B).
|
Although SEF neurons, considered as a population, exhibited enhanced activity under conditions of conflict in both tasks, the timing of the effect differed between tasks. In the spatial incompatibility task, conflict induced a late (postsaccadic) enhancement (Fig. 5D, gray curve), whereas, in the reversal task, conflict induced an early (presaccadic) enhancement (Fig. 10D, gray curve). These effects may have depended on the same mechanism but in a manner dependent on the temporal pattern of crucial events in the task. In the spatial incompatibility task, the event eliciting conflict (presentation of a cue opposite the target) was simultaneous with the instruction in what direction to move (conveyed by the cue's color). In the reversal task, the event eliciting conflict (presentation of the 1st colored cue) occurred 100 ms before the instruction in what direction to move (conveyed by the 2nd colored cue). If the effects observed in the two tasks did indeed originate from the same mechanism, we would expect that neurons exhibiting conflict-induced enhancement in one task should do so as well in the other task. We assessed whether this was so by analyzing data from 221 SEF neurons studied in the context of both tasks. For each neuron in each task, we computed an index of enhancement: I – C, where I was the mean firing rate on trials when cue location and response direction were incompatible, and C was the mean firing rate on trials when the cue location and response direction were compatible. This measure was based on the period 0–300 ms after saccade initiation for the spatial incompatibility task and 0–300 ms before the saccade for the reversal task. Across the population of neurons, the values obtained in the two tasks were positively and significantly (P < 0.005) correlated. The correlation coefficient was comparatively low (R = 0.19). Nevertheless, the existence of a significant positive correlation indicates that the conflict-induced firing-rate enhancement tended to occur in the same neurons across tasks and suggests that one mechanism underlay the enhancement observed in both tasks.
SEF: relation of conflict-related activity to saccadic reaction time
Behavioral reaction times were significantly shorter on low-conflict than on high-conflict trials. This raises the question: was conflict-related neuronal activity directly correlated with the level of conflict or was it, alternatively, directly correlated with the behavioral reaction time and correlated with the level of conflict only through this dependence? To resolve this issue, for each neuron, we used a multiple least-squares regression approach to optimize the parameters of three models representing firing rate as a linear function of reaction time and compatibility status: 1) a reduced model incorporating reaction time only, 2) a reduced model incorporating compatibility status only and 3) a full model incorporating both. This analysis was based on the period in each task during which conflict-enhancement was maximal (300 ms after saccade onset in the spatial incompatibility task and 300 ms before saccade onset in the reversal task). It was carried out independently for trials requiring responses in each neuron's preferred and antipreferred directions because the relation between reaction time and neuronal activity might vary as a function of response direction. We compared each of the reduced models to the full model using a nested F-test (see METHODS). Then we calculated the percentage of neurons in each area that showed a significant improvement of fit when the variable of compatibility status was added to the model and, likewise, those that showed a significant improvement of fit when reaction time was added.
The number of neurons showing a significant improvement of fit when the variable of compatibility status was added to the model (those in which conflict-related activity could not be accounted for by reaction-time-related activity) exceeded the number expected by chance at a highly significant level (2 test, P < 0.0001). Of 344 neurons studied in the context of the spatial incompatibility task, 41 exhibited a significant dependence on compatibility status in preferred-direction trials (with 35/41 exhibiting a decrease of firing rate under high conflict) and 39 exhibited a significant dependence in antipreferred-direction trials (with 37/39 exhibiting an increase of firing rate under high conflict). Of 253 neurons studied in the context of the reversal task, 21 showed a significant dependence on compatibility status when the required response was in the preferred direction (with 14/21 exhibiting reduced activity under high conflict) and 30 showed a significant dependence when the required response was in the antipreferred direction (with 27/30 exhibiting increased activity under high conflict). The number of neurons exhibiting a significant dependence on reaction time was commensurate with the number exhibiting a significant dependence on compatibility status. The tendency to exhibit one trait was not, however, correlated with the tendency to exhibit the other. Nor, in any of the four analyses (2 tasks x 2 response directions), was there a significant difference between the number of neurons exhibiting enhanced versus reduced firing when the reaction time was short. We conclude that compatibility status exerted an impact on neuronal firing rate that was independent of the impact of reaction time.
SEF: trial sequence effects
We have described up to this point several features of behavior and neuronal physiology that were correlated with the presence or absence of conflict in a given trial. It might seem reasonable to suppose that these behavioral and physiological variables were dependent on the level of conflict in the trial during which they were measured. We must, however, allow for the possibility that they depended on the level of conflict during the preceding trial. This is possible because the sequencing of trials in our tasks was not absolutely random. There was a probability of 0.625 that an incompatible trial would be followed by a compatible trial and vice versa. Thus the current trial and one preceding it tended to have opposite levels of conflict.
To determine whether the monkeys' behavior was influenced by the nature of the preceding trial, we carried out an analysis of behavioral performance in which trials were subdivided more finely than in our initial analyses (Figs. 1C and 2C). Each trial was placed into one of four categories: a compatible trial following a compatible trial (C-C), a compatible trial following an incompatible trial (IC-C), an incompatible trial following a compatible trial (C-IC), or an incompatible trial following an incompatible trial (IC-IC). There were eight analyses as determined by crossing measure (percent correct or reaction time) with trial type (compatible or incompatible) and task (spatial incompatibility or reversal). Every analysis revealed a significant effect of the nature of the preceding trial (2-tailed t-test with Bonferroni correction, P < 0.01 for reaction time on incompatible trials in the reversal task and P < 0.0001 for all other comparisons). The pattern was absolutely consistent: performance was faster and more accurate when a trial of a given type (compatible or incompatible) followed another trial of the same type (Fig. 13). This cannot have been a result of the monkeys' making a strategic adjustment on the basis of the trial-sequence statistics. Incompatible trials were more common after compatible trials and vice versa, with the result that strategizing on the basis of the trial-sequence statistics would have led to better performance when there was a change of trial-type, not when trial-type remained the same. Nor is this the pattern that one would expect from the monkeys' developing a more focused task-set and a more conservative response criterion following conflict, as humans have been suggested to do (Botvinick et al. 2001). This would have led to slower and more accurate performance on all trials following an incompatible trial, regardless of type. Rather, it seems that performing under a given condition (compatible or incompatible) primed performing under the same condition on the next trial.
|
|
We categorized as ACC an extensive region ventral to the SEF including cortex in both banks of the cingulate sulcus and on adjacent gyral cortex (Paus 2001). The distinguishing trait of this region was that intracortical microstimulation at currents <100 µA did not elicit observable motor responses. In agreement with previous studies (Ito et al. 2003; Niki and Watanabe 1976), we observed no signs of regional differentiation of function within this zone. Accordingly, we considered it as a whole for purposes of data analysis. A neuron in this region was included in the ACC data base if the mean firing rate during at least one of six trial epochs (C1–C3 and S1–S3) differed significantly (t-test, P < 0.05) from the baseline rate measured during a period 300–200 ms prior to cue onset. These criteria were satisfied by 172 neurons (65 in monkey N and 107 in monkey A) studied in the context of the spatial compatibility task and by 144 neurons (68 in monkey N and 76 in monkey A) studied in the context of the reversal task. Neurons in this category are indicated by red symbols overlying a white background in Fig. 3.
ACC: neuronal activity in the spatial incompatibility task
The impact of conflict on neuronal activity in ACC during performance of the spatial incompatibility task appeared primarily as a reduction in the strength of the directional signal during the period before saccade execution. This is evident on inspection of population histograms representing the activity of all 172 neurons (Fig. 15). It is also apparent from consideration of the counts of neurons exhibiting a significant interaction of conflict and response direction: neurons in which directional signals were reversed or decreased under conditions of conflict consistently (but not significantly) outnumbered those in which directional signals were increased (Fig. 16A; Table 5). Finally, a reduction of the directional signal is revealed by analysis of firing rates during epoch S1 immediately preceding the saccade. During this epoch, ACC neurons tended to fire less strongly under conflict when the impending saccade was in their preferred direction (Fig. 16C; Table 5) and more strongly under conflict when the impending saccade was in their antipreferred direction (Fig. 16D; Table 5). In contrast to the impact of conflict on directional signals, there was no demonstrable impact on net firing rate. The absence of a tendency toward enhanced firing on trials involving conflict is evident in the population histograms (Fig. 15) and in the counts of neurons exhibiting significant main effects of the level of conflict (Fig. 16B; Table 5).
|
|
|
The impact of conflict on neuronal activity in the ACC during performance of the reversal task primarily took the form of a reduction in the strength of presaccadic directional signals. The reduction in the strength of the directional signal is evident on inspection of population histograms representing the activity of all 144 neurons (Fig. 17), on consideration of the counts of neurons exhibiting a significant interaction of conflict and response direction (Fig. 18A; Table 6), and on comparison of incompatible-trial and compatible-trial firing rates during epoch S1 (Fig. 18D). In contrast to the impact of conflict on directional signals, there was little or no impact on net firing rate. The net activity of the population was not obviously enhanced on incompatible trials (Fig. 17, C and D); however, during postsaccadic epochs S2 and S3, neurons exhibiting significantly increased firing under incompatible conditions were observed at a rate barely significantly in excess of the one expected by chance (Fig. 18B; Table 6).
|
|
|
|
|
DISCUSSION |
|---|
|
The tasks used in this study clearly induced conflict, as evidenced by behavioral and electrophysiological measures. In both the spatial incompatibility task (Fig. 1) and the reversal task (Fig. 2), conflict was reflected in reduced percent correct scores and lengthened reaction times. Moreover, in both tasks, population activity in the SEF gave direct evidence for the presence of conflict as defined according to the criterion that there be simultaneous activation of incompatible response tendencies (Botvinick et al. 2001). On the assumption that for any given neuron there is an antineuron with symmetric functional properties, the gray and black curves in Figs. 5, A and B, and 10, A and B can be regarded as representing data collected from two populations of neurons encoding saccades in opposite directions and thus encoding incompatible responses. It is evident from inspection of these histograms that neurons encoding responses opposite to the required response showed markedly enhanced activity under conditions presumed to induce conflict (the thin black curve is greatly elevated relative to the thick black curve), whereas neurons encoding the required response showed a less marked reduction in firing rate or even an enhancement (the thin gray line is moderately below or is even above the thick gray line). In consequence, there was an increase in the level of co-activation of the two neuronal populations, which would have been evident in standard neuronal measures of conflict such as the Hopfield energy measure (Botvinick et al. 2001).
Apparent absence of neuronal conflict monitoring in the SEF
Although population activity in the SEF was enhanced under conditions of conflict, there was no evidence that the SEF contained neurons with conflict monitoring as their essential function. Neurons exhibiting a conflict-associated firing-rate enhancement also exhibited standard forms of task-related activity, including direction-selective phasic activation, even when no conflict was present (Figs. 7A and 12A, thick gray and black curves). Thus they did not carry pure conflict-monitoring signals. Moreover, enhanced firing was arguably no more than a direct consequence of the co-activation of neuronal populations encoding opposed responses, a condition present, by definition, in conflict (Botvinick et al. 2001). In a push-pull system such as the neural circuit governing saccade generation, simultaneous inputs commanding opposite responses combine subadditively because populations of neurons associated with the two responses inhibit each other. Nevertheless, the two inputs together may elicit a mean level of activity higher than that elicited by either input alone. This is just what we observed in the SEF.
It might be argued that the enhancement of population activity, whatever its origin, could be used by other parts of the brain as an indicator for the presence of conflict, and thus could constitute a conflict-monitoring signal. This line of reasoning is specious because it ignores an important distinction. Computational models of conflict make a distinction between events at the output layer, where units controlling incompatible responses are co-active during conflict, and events at the monitoring layer, where units sensitive to co-activation in the output layer are selectively active during conflict (Botvinick et al. 2001). Events in the output layer do reflect conflict. Otherwise, it would be impossible for conflict-monitoring units sensitive to events in the output layer to detect conflict. However, events in the output layer embody conflict rather than monitor it.
Our failure to find conflict-monitoring neurons in the SEF stands in apparent contrast to an earlier finding that 13% of SEF neurons studied in the context of the countermanding task fired exclusively on trials involving conflict (Stuphorn et al. 2000). In the example shown in the earlier report (Stuphorn et al. 2000, Fig. 2), strong neuronal activity occurred exclusively on trials involving conflict. It did not take the form of a subtle enhancement of classic direction-selective task-related activity as noted in our study. However, it is clear from the supplementary on-line materials that many neurons exhibiting conflict-enhancement in the former study did also exhibit other forms of task-related activity. Thus the degree of discrepancy between the two sets of results may not be great. We also note that other factors co-varied with conflict in the former study and that firing could conceivably have been related to these concomitant factors. The identification of conflict-monitoring neurons was based on comparing activity during trials in which a saccade was successfully countermanded (a condition argued to involve conflict) to activity on trials in which cancellation of the saccade either was not commanded or was not achieved (conditions argued not to involve conflict). The condition in which a saccade is successfully countermanded differs in several respects from the conditions to which it was compared: it is the only condition in which there is successful cancellation of a saccade, it is the only condition in which a foveal stimulus, the countermanding cue, is stably present on the fovea, and it is the only condition in which attention is divided between the fovea and the periphery. In light of our failure to find neurons with pure conflict-related activity in the SEF, we suggest that the neurons described in the earlier report must have been sensitive to some aspect of countermanding trials other than the occurrence of conflict.
Apparent absence of neuronal conflict monitoring in the ACC
We have found no evidence that neurons in the ACC monitor conflict. In the ACC, unlike the SEF, there was not even a tendency for population activity to be heightened under conditions of conflict. This finding is compatible with an earlier report to the effect that ACC neurons in the monkey are not selectively active during the countermanding of saccades, an operation assumed to involve conflict (Ito et al. 2003). However, it stands in contrast to a large literature, based on functional imaging in humans, indicating that BOLD activation in the ACC is heightened under conditions of conflict (Badgaiyan and Posner 1998; Botvinick et al. 1999, 2001; Braver et al. 2001; Carter et al. 1998, 2000; Cohen et al. 2000; D'Esposito et al. 1995; Doricchi et al. 1997; Fan et al. 2003; Frith et al. 1991; Jenkins et al. 1994; Jueptner et al. 1997; Kerns et al. 2004; Merriam et al. 2001; O'Driscoll 1995; Pardo 1990; Paus et al. 1993, 1998; Petersen et al. 1988; Petit et al. 1998; Raichle et al. 1994; Sweeney et al. 1996; van Veen et al. 2001). There are three lines of reasoning by which this discrepancy might be explained. 1) It is possible that there is a species difference such that neurons in the human ACC monitor conflict whereas those in the monkey ACC do not. This seems improbable because, in general, anatomically homologous areas seem to serve similar functions in the two species (Paus 2001). It cannot, however, be altogether ruled out. The human ACC possesses a cell type not found in the monkey ACC (Allman et al. 2002). It might, therefore, serve a function not served by the monkey ACC. 2) It is possible that our recording sites lay outside the region of the ACC where neurons exhibit conflict-related activity. This seems improbable because we sampled widely in the ACC. Moreover, we deliberately recorded in the subregion that is connected to the SEF (Huerta and Kaas 1990) and in which, accordingly, it would be most reasonable to expect to find neurons sensitive to conflict in an oculomotor task. Some neurons in this region exhibited error-related activity (Nakamura et al. 2002), a trait typical of the ACC (Ito et al. 2003; Niki and Watanabe 1976), which has been suggested to be related to conflict monitoring (Botvinick et al. 2001). However, even these neurons did not exhibit conflict-monitoring activity. 3) It is possible that, through selectively sampling neurons with relatively large cell bodies, such as are most likely to be isolated in extracellular recording, we failed to detect a subpopulation possessing small cell bodies and engaged in monitoring of conflict. This possibility cannot be absolutely ruled out. However, in light of the dense interconnectivity among neurons in a cortical column, it seems improbable that a functional trait present in one subpopulation would be absent in another subpopulation occupying the same column. 4) It is possible that the conflict-specific BOLD signals detected in human fMRI studies are related to neural events other than spiking activity, for example, to presynaptic potentials (Logothetis and Wandell 2004). It might be the case that ACC neurons, even in humans, do not exhibit enhanced spiking activity under conditions of conflict. For the ACC to serve a conflict-monitoring function (alerting the rest of the cortex to the presence of conflict) would require enhanced spiking activity because spikes are the currency of commerce between the ACC and other cortical areas. Thus the fourth line of reasoning leads to the conclusion that the ACC does not, after all, monitor conflict.
|
|
GRANTS |
|---|
|
|
|
ACKNOWLEDGMENTS |
|---|
|
Present address of K. Nakamura: Laboratory of Sensorimotor Research, National Eye Institute, National Institutes of Health, Bldg. 49, Rm. 2A50, Bethesda, MD 20892.
|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: C. R. Olson, Center for the Neural Basis of Cognition, Mellon Inst., Rm. 115, 4400 Fifth Ave., Pittsburgh, PA 15213-2683 (E-mail: colson{at}cnbc.cmu.edu)
|
|
REFERENCES |
|---|
|
Amador N, Schlag-Rey M, and Schlag J. Primate antisaccade. II. Supplementary eye field neuronal activity predicts correct performance. J Neurophysiol 91: 1672–1689, 2004.
Badgaiyan RD and Posner MI. Mapping the cingulate cortex in response selection and monitoring. Neuroimage 7: 255–260, 1998.[CrossRef][Web of Science][Medline]
Botvinick MM, Braver TS, Barch DM, Carter CS, and Cohen JD. Conflict monitoring and cognitive control. Psycho Rev 108: 624–652, 2001.[CrossRef]
Botvinick MM, Nystrom LE, Fissell K, Carter CS, and Cohen JD. Conflict monitoring versus selection-for-action in anterior cingulate cortex. Nature 402: 179–181, 1999.[CrossRef][Medline]
Braver TS, Barch DM, Gray JR, Molfese DL, and Snyder A. Anterior cingulate cortex and response conflict: effects of frequency, inhibition and errors. Cereb Cortex 11: 825–836, 2001.
Carter CS, Braver TS, Barch DM, Botvinick MM, Noll D, and Cohen JD. Anterior cingulate cortex, error detection, and the online monitoring of performance. Science 280: 747–749, 1998.
Carter CS, Macdonald AM, Botvinick M, Ross LL, Stenger VA, Noll D, and Cohen JD. Parsing executive processes: strategic vs. evaluative functions of the anterior cingulate cortex. Proc Natl Acad Sci USA 97: 1944–1948, 2000.
Cohen JD, Botvinick M, and Carter CS. Anterior cingulate and prefrontal cortex: who's in control? Nat Neurosci 3: 421–423, 2000.[CrossRef][Web of Science][Medline]
Crist CF, Yamasaki DS, Komatsu H, and Wurtz RH. A grid system and a microsyringe for single cell recording. J Neurosci Methods 26: 117–122, 1988.[CrossRef][Web of Science][Medline]
D'Esposito M, Detre JA, Alsop DC, Shin RK, Atlas S, and Grossman M. The neural basis of the central executive system of working memory. Nature 378: 279–281, 1995.[CrossRef][Medline]
Doricchi F, Perani D, Incoccia C, Grassi F, Cappa SF, Bettinardi V, Galati G, Pizzamiglio L, and Fazio F. Neural control of fast-regular saccades and antisaccades: an investigation using positron emission tomography. Exp Brain Res 116: 50–62, 1997.[CrossRef][Web of Science][Medline]
Eriksen BA and Eriksen CW. Effects of noise letters upon the identification of a target letter in a nonsearch task. Percept Psychophys 16: 143–149, 1974.[Web of Science]
Everling S and Fischer B. The antisaccade: a review of basic research and clinical studies. Neuropsychologia 36: 885–899, 1998.[CrossRef][Web of Science][Medline]
Fan J, Flombaum JI, McCandliss BD, Thomas KM, and Posner MI. Cognitive and brain consequences of conflict. Neuroimage 18: 42–57, 2003.[CrossRef][Web of Science][Medline]
Frith CD, Friston K, Liddle PF, and Frackowiak RS. Willed action and the prefrontal cortex in man: a study with PET. Proc R Soc Lond B Biol Sci 244: 241–246, 1991.[Medline]
Hallett PE. Primary and secondary saccades to goals defined by instructions. Vision Res 18: 1279–1296, 1978.[CrossRef][Web of Science][Medline]
Hallett PE and Adams BD. The predictability of saccadic latency in a novel voluntary oculomotor task. Vision Res 20: 329–339, 1980.[CrossRef][Web of Science][Medline]
Huerta MF and Kaas JH. Supplementary eye field as defined by intracortical microstimulation: connections in macaques. Comp Neurol 293: 299–330, 1990.[CrossRef][Web of Science][Medline]
Ito S, Stuphorn V, Brown JW, and Schall JD. Performance monitoring by the anterior cingulate cortex during saccade countermanding. Science 302: 120–122, 2003.
Jenkins IH, Brooks DJ, Nixon PD, Frackowiak RS, and Passingham RE. Motor sequence learning: a study with positron emission tomography. J Neurosci 14: 3775–3790, 1994.[Abstract]
Jueptner M, Stephan KM, Frith CD, Brooks DJ, Frackowiak RS, and Passingham RE. Anatomy of motor learning. I. Frontal cortex and attention to action. J Neurophysiol 77: 1313–1324, 1997.
Kerns JG, Cohen JD, MacDonald AW III, Cho RY, Stenger VA, and Carter CS. Anterior cingulate conflict monitoring and adjustments in control. Science 303: 1023–1026, 2004.
Logothetis NK and Wandell BA. Interpreting the BOLD signal. Annu Rev Physiol 66: 735–769, 2004.[CrossRef][Web of Science][Medline]
Merriam EP, Colby CL, Thulborn KR, Luna B, Olson CR, and Sweeney JA. Stimulus-response incompatibility activates cortex proximate to three eye fields. Neuroimage 13: 794–800, 2001.[Web of Science][Medline]
Nakamura K, Roesch MR, and Olson CR. Error Signal in Monkey Anterior Cingulate Cortex During Performance of Oculomotor Task With Conflict. Program No. 464.7. Washington, DC: Society for Neuroscience, 2002.
Niki H and Watanabe M. Cingulate unit activity and delayed response. Brain Res 110: 381–386, 1976.[CrossRef][Web of Science][Medline]
O'Driscoll GA, Alpert NM, Matthysse SW, Levy DL, Rauch SL, and Holzman PS. Functional neuroanatomy of antisaccade eye movements investigated with positron emission tomography. Proc Natl Acad Sci USA 92: 925–929, 1995.
Olson CR and Gettner SN. Macaque SEF neurons encode object-centered directions of eye movements regardless of the visual attributes of instructional cues. J Neurophysiol 81: 2340–2346, 1999.
Olson CR and Gettner SN. Neuronal activity related to rule and conflict in macaque supplementary eye field. Physiol Behav 77: 663–670, 2002.[CrossRef][Medline]
Pardo JV, Pardo PJ, Janer KW, and Raichle ME. The anterior cingulate cortex mediates processing selection in the Stroop attentional conflict paradigm. Proc Natl Acad Sci USA 87: 256–259, 1990.
Paus T. Primate anterior cingulate cortex: where motor control, drive and cognition interface. Nat Rev Neurosci 2: 417–424, 2001.[CrossRef][Web of Science][Medline]
Paus T, Koski L, Caramanos Z, and Westbury C. Regional differences in the effects of task difficulty and motor output on blood flow response in the human anterior cingulate cortex: a review of 107 PET activation studies. Neuroreport 9: R37–R47, 1998.[Web of Science][Medline]
Paus T, Petrides M, Evans AC, and Meyer E. Role of the human anterior cingulate cortex in the control of oculomotor, manual, and speech responses: a positron emission tomography study. J Neurophysiol 70: 453–469, 1993.
Petersen SE, van Mier H, Fiez JA, and Raichle ME. The effects of practice on the functional anatomy of task performance. Proc Natl Acad Sci USA 95: 853–860, 1998.
Petit L, Courtney SM, Ungerleider LG, and Haxby JV. Sustained activity in the medial wall during working memory delays. J Neurosci 18: 9429–9437, 1998.
Raichle ME, Fiez JA, Videen TO, MacLeod AM, Pardo JV, Fox PT, and Petersen SE. Practice-related changes in human brain functional anatomy during nonmotor learning. Cereb Cortex 4: 8–26, 1994.
Remmel RS. An inexpensive eye movement monitor using the scleral search coil technique. IEEE Trans Biomed Eng 31: 388–390, 1984.[Web of Science][Medline]
Robinson DA. A method of measuring eye movement using a scleral search coil in a magnetic field. IEEE Trans Biomed Eng 10: 137–145, 1963.[Medline]
Stroop JR. Studies of interference in serial verbal reactions. J. Exp Psychol 18: 643–662, 1935.[CrossRef][Web of Science]
Stuphorn V, Taylor TL, and Schall JD. Performance monitoring by the supplementary eye field. Nature 408: 857–860, 2000.[CrossRef][Medline]
Sweeney JA, Mintun MA, Kwee S, Wiseman MB, Brown DL, Rosenberg DR, and Carl JR. Positron emission tomography study of voluntary saccadic eye movements and spatial working memory. J Neurophysiol 75: 454–468, 1996.
Tremblay L, Gettner SN, and Olson CR. Neurons with object-centered spatial selectivity in macaque SEF: do they represent locations or rules? J Neurophysiol 87: 333–350, 2002.
van Veen V, Cohen JD, Botvinick MM, Stenger VA, and Carter CS. Anterior cingulate cortex, conflict monitoring, and levels of processing. Neuroimage 14: 1302–1308, 2001.[CrossRef][Web of Science][Medline]
This article has been cited by other articles:
|
|
 |
|
  B. B. Averbeck, A. Battaglia-Mayer, C. Guglielmo, and R. Caminiti Statistical Analysis of Parieto-Frontal Cognitive-Motor Networks J Neurophysiol, September 1, 2009; 102(3): 1911 - 1920. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. B. Holroyd, O. E. Krigolson, R. Baker, S. Lee, and J. Gibson When is an error not a prediction error? An electrophysiological investigation Cogn Affect Behav Neurosci, March 1, 2009; 9(1): 59 - 70. [Abstract] [PDF]
|
|
|
|
 |
|
 D. M. Barch, T. S. Braver, C. S. Carter, R. A. Poldrack, and T. W. Robbins CNTRICS Final Task Selection: Executive Control Schizophr Bull, January 1, 2009; 35(1): 115 - 135. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Isoda and O. Hikosaka A Neural Correlate of Motivational Conflict in the Superior Colliculus of the Macaque J Neurophysiol, September 1, 2008; 100(3): 1332 - 1342. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. E. Emeric, J. W. Brown, M. Leslie, P. Pouget, V. Stuphorn, and J. D. Schall Performance Monitoring Local Field Potentials in the Medial Frontal Cortex of Primates: Anterior Cingulate Cortex J Neurophysiol, February 1, 2008; 99(2): 759 - 772. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. A. Mansouri, M. J. Buckley, and K. Tanaka Mnemonic Function of the Dorsolateral Prefrontal Cortex in Conflict-Induced Behavioral Adjustment Science, November 9, 2007; 318(5852): 987 - 990. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Michelet, B. Bioulac, D. Guehl, L. Escola, and P. Burbaud Impact of Commitment on Performance Evaluation in the Rostral Cingulate Motor Area J. Neurosci., July 11, 2007; 27(28): 7482 - 7489. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Amiez, J.P. Joseph, and E. Procyk Reward Encoding in the Monkey Anterior Cingulate Cortex Cereb Cortex, July 1, 2006; 16(7): 1040 - 1055. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. D. Davis, K. S. Taylor, W. D. Hutchison, J. O. Dostrovsky, M. P. McAndrews, E. O. Richter, and A. M. Lozano Human Anterior Cingulate Cortex Neurons Encode Cognitive and Emotional Demands J. Neurosci., September 14, 2005; 25(37): 8402 - 8406. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Hoshi, H. Sawamura, and J. Tanji Neurons in the Rostral Cingulate Motor Area Monitor Multiple Phases of Visuomotor Behavior With Modest Parametric Selectivity J Neurophysiol, July 1, 2005; 94(1): 640 - 656. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
TOP
ABSTRACT
INTRODUCTION
