Abstract

Previous studies have demonstrated that stable and adaptive attention processes are mediated by partly overlapping, but distinct, brain areas. Dorsal medial PFC and anterior insula may form a “core network” for attention control, which is believed to operate on both temporal scales. However, both the existence of such a network as well as the unique functional topography for adaptive and stable attention processes is still highly debated. In this study, 87 healthy participants performed a clinical not-X continuous performance test optimized for use in a mixed block and event-related fMRI design. We observed overlapping activations related to stable and adaptive attention processes in dorsal medial PFC and anterior insula/adjacent cortex as well as in the right inferior parietal lobe and middle temporal gyrus. We also identified areas of activations uniquely related to stable and adaptive attention processes in widespread cortical, cerebellar, and subcortical areas. Interestingly, the functional topography within the PFC indicated a rostro-caudal distribution of adaptive, relative to stable, attention processes. There was also evidence for a time-on-task effect for activations related to stable, but not adaptive, attention processes. Our results provide further evidence for a “core network” for attention control that is accompanied by unique areas of activation involved in domain-specific processes operating on different temporal scales. In addition, our results give new insights into the functional topography of stable and adaptive attention processes and their temporal dynamics in the context of an extensively used clinical attention test.

INTRODUCTION

A considerable body of evidence suggests that attention is supported by widespread brain areas located in cortical (Ogg et al., 2008; Fan, McCandliss, Fossella, Flombaum, & Posner, 2005), subcortical (Balleine, Delgado, & Hikosaka, 2007; Heyder, Suchan, & Daum, 2004), and cerebellar (Ghajar & Ivry, 2009; Dosenbach et al., 2007) regions, which in turn are organized into neural networks supporting different attentional processes (Dosenbach et al., 2007; Posner & Rothbart, 2007; Raz & Buhle, 2006; Corbetta & Shulman, 2002). Some properties of attention are considered to be stable ongoing processes related to functions such as sustained attention (Ogg et al., 2008) and task-set maintenance (Altmann & Gray, 2002). Other aspects of attention are thought to be related to rapid adaptive processes such as conflict processing (Desmet, Fias, Hartstra, & Brass, 2011), error processing (Mathiak et al., 2011; Nee, Kastner, & Brown, 2011), and successful response inhibition (Boehler, Appelbaum, Krebs, Hopf, & Woldorff, 2010). Stable and adaptive attention processes are subserved by partly overlapping, but distinct, brain networks (Wilk, Ezekiel, & Morton, 2012; Dosenbach, Fair, Cohen, Schlaggar, & Petersen, 2008; Dosenbach et al., 2006, 2007; Seeley et al., 2007). Overlapping regions have been suggested to include the dorsal medial PFC as well as the insula and adjacent prefrontal regions, which may form a “core network” for task control comprising both adaptive and stable processes (Dosenbach et al., 2006). These regions are among the most commonly reported brain areas in imaging studies across a wide variety of tasks (Nelson et al., 2010), further underlining their potential role as core task regions. However, whether insula and dorsal medial PFC play a substantial role for stable task maintenance is highly debated. Other studies imply that these structures are primarily involved in a salience network that works on a more rapid time scale (Wilk et al., 2012; Menon & Uddin, 2010; Seeley et al., 2007). Also, despite several informative studies, the functional topography of the unique adaptive task control and stable task-set maintenance activations has yet to be elucidated.

The PFC has drawn particular interest from researchers investigating the functional topography of attentional control. Accumulated evidence supports a rostro-caudal distribution of control functions within PFC (Kim, Johnson, Cilles, & Gold, 2011; Badre & D'Esposito, 2007, 2009; Venkatraman, Rosati, Taren, & Huettel, 2009; Badre, 2008; Koechlin & Summerfield, 2007; Koechlin, Ody, & Kouneiher, 2003). However, controversy exists regarding which factors govern this distribution (Badre, 2008). One distinction between two acknowledged frameworks originating from the rostro-caudal perspective is to what degree they emphasize the temporal dimension of attentional control. The cognitive demand framework states that control functions should be located more rostral within PFC as control demands are increased, that is, by increasing task complexity or abstractness (Badre & D'Esposito, 2007, 2009; Badre, 2008). Accordingly, processes with lower cognitive demands, such as those more closely related to motor responses, should be located in more caudal regions (Kim et al., 2011; Venkatraman et al., 2009). In contrast, the episodic and contextual control framework (Koechlin & Summerfield, 2007; Koechlin et al., 2003) suggests that processing relying on immediate stimuli-related cues (contextual control) should be located caudally relative to processes based on cues more distant in time that have to be maintained throughout the task (episodic control). Although these two different frameworks both originate from the rostro-caudal perspective and overlap to a certain degree on a theoretical level, they provide different predictions for specific task paradigms. Determining whether the distribution of stable task-set maintenance and adaptive task control within PFC is best accommodated by the control demand or the episodic and contextual control framework may yield valuable information for delineating the functional organization of attentional control.

Another temporal aspect of attention is the effect of time-on-task (TOT). Typical TOT effects are fatigue and decrements in behavioral performance (Langner, Steinborn, Chatterjee, Sturm, & Willmes, 2010; Lim et al., 2010; Conners, Epstein, Angold, & Klaric, 2003). Self-reported levels of fatigue have been found to correlate with increased activity in typical task-positive brain regions and decreased activity in “default mode network” (DMN) or task-negative regions (Kelly, Uddin, Biswal, Castellanos, & Milham, 2008; Cook, O'Connor, Lange, & Steffener, 2007; Weissman, Roberts, Visscher, & Woldorff, 2006; Fox et al., 2005). However, other studies report both decreased and increased activations of both task-positive and task-negative regions using different types of tasks and with or without accompanying behavioral or state changes (Lim et al., 2010; Tana, Montin, Cerutti, & Bianchi, 2010; Cook et al., 2007; Butti et al., 2006). In context of these results, it is worth mentioning that limitations of most previous studies of TOT effects are relatively few participants, the use of predefined ROIs, and liberal statistical thresholding. Also, there are currently no published studies that have investigated the effects of TOT for both stable task-set maintenance and adaptive task control in the same study.

An excellent task for investigating the temporal dynamics of stable task-set maintenance and adaptive task control is the continuous performance test (CPT; Riccio, Reynolds, Lowe, & Moore, 2002; Riccio & Reynolds, 2001). The main feature of CPTs is that participants are asked to detect low-frequency target stimuli within a consecutive presentation of nontargets in a task that typically lasts for >10 min. The CPT is among the top five most frequently used test paradigms for assessment of attention in the U.S. and Canada (Rabin, Barr, & Burton, 2005). Among the most extensively used CPTs for assessment of dysfunctions of attention is the Conners' CPT (Conners et al., 2003). Conners' CPT falls within the category of CPTs, where participants are asked to respond to nontargets (letters from A–Z) and withhold their response to targets (the letter X), also commonly described as a not-X CPT (Riccio et al., 2002). Moreover, a previous proof of concept study using a block fMRI design has demonstrated that brain regions typically involved in attention function are reliably activated during this task (Ogg et al., 2008).

For this study, an in-house version of a not-X CPT, optimized for use in a mixed block and event-related fMRI design, was used to determine common and distinct brain areas related to stable task-set maintenance and adaptive task control. On the basis of previous research, the existence of a core network for attentional control consisting of anterior insular and dorso-medial prefrontal brain regions was investigated. In addition, domain-specific not-X CPT activations within cortical, cerebellar, and subcortical areas uniquely related to stable and adaptive processes were mapped. Extending previous research, this study aimed to answer two novel research questions, namely (1) how areas related to stable versus adaptive task processing are organized within PFC and (2) how stable task-set maintenance and adaptive task control are affected by TOT. In detail, for research question 1, two different frameworks giving opposite predictions for our particular not-X CPT were tested. The episodic and contextual control framework (Koechlin & Summerfield, 2007; Koechlin et al., 2003) made the prediction that stable task-set maintenance (episodic control) would be located rostral, relative to adaptive task control (contextual control). Contrary to this, the cognitive demand framework (Badre & D'Esposito, 2007, 2009; Badre, 2008) predicted that stable task-set maintenance (low complexity: task set dominated by 90% simple motor responses) would be located caudal, relative to adaptive task control (higher complexity: decision making/response inhibition). Finally, for research question 2, the prediction was that adaptive task control and stable task-set maintenance would be affected differently as an effect of TOT, something that would provide further evidence for dissociation between the proposed networks.

METHODS

Participants

One hundred three healthy participants were recruited through a cohort database from a clinical study as well as advertisements at a wide variety of workplaces in Trondheim, Norway. Volunteers were financially reimbursed with NOK 1000 for their participation. Inclusion criteria were absence of diagnosed neurological or psychiatric condition, ongoing substance abuse, previous head injury, ability to cooperate during fMRI testing, and fluency in the Norwegian language. Number of years of completed formal education was assessed using a self-report form, which was quality assured through a short interview performed by the experimenters. Of the 103 participants first enrolled, 16 were excluded from further analysis: two because of diagnosed psychiatric or neurological condition uncovered in the initial screening interview at site, one because of excessive fMRI artifacts, eight because of technical problems (missing data or aborted scans), and five because of excessive movement artifacts (defined as mean relative displacement > 0.5 mm in any direction), leaving 87 participants (34 women) ranging in age from 14.5 to 64.5 years, with a median age of 31.4 years in the final sample. The participants had between 9 and 18 years of completed formal education with a median of 12 years. All participants (and their parents if the participant was under 18 years old) gave written informed consent. The study protocol was approved by the National Ethics Committee in Norway and adhered to the Helsinki Declaration.

Design of fMRI Task

The fMRI attention task was an in-house developed variant of the not-X CPT, inspired by the Conners' CPT (Conners et al., 2003), and modified to allow for both block and event-related fMRI analysis, giving the opportunity to investigate both adaptive task control (event-related) and stable task-set maintenance (block-related). Targets were randomly chosen from all letters other than X (A–Z), whereas the nontarget was the letter X. The task was presented in two consecutive runs, each lasting ∼15 min. Each run consisted of 240 stimuli with 24 (10%) of them being nontargets. Each stimulus was presented on the screen for 250 msec. The stimuli within run 1 were split into 16 blocks of different types (containing zero, one, two, or three nontarget/s). Block types were presented pseudorandomly with the constraint that no more than two consecutive blocks of the same type could be presented. Inside each block, which was composed of 15 stimuli, targets and nontargets were randomly scrambled, not allowing two consecutive nontargets to occur or nontargets to be the first stimulus of any block. The ISIs were randomly scrambled within each block (with five ISIs of 1 sec, five ISIs of 2 sec, and five ISIs of 4 sec) to ensure sampling at different time points of the hemodynamic response curve for all trial types, allowing for event-related fMRI analysis (Petersen & Dubis, 2012). ISIs preceding nontargets (X) were evenly distributed (eight of 1 sec, eight of 2 sec, and eight of 4 sec). Interblock intervals (IBIs) were randomly scrambled within the run (with six IBIs of 14 sec, five IBIs of 16 sec, and five IBIs of 18 sec). Run 2 was an inverse presentation of run 1, meaning that the same blocks with the same stimuli, stimuli order, ISIs, and IBIs as in run 1 were presented in the opposite order, thereby counterbalancing possible order effects after the two runs were collapsed. The total number of stimuli of both runs combined was 480 with 48 (10%) nontargets elements. Implementation of the CPT was done in MATLAB 2008 (The MathWorks, Inc., Natick, MA), using the functions randperm() and rand() for the generation of pseudorandom permutations and values.

CPT Paradigm Procedure

Participants were instructed to press a response button as fast as possible whenever a letter appeared on the screen, except for the letter “X.” They were also instructed to strive to make as few errors as possible. All participants completed a training session using a standard desktop computer, and the experimenter made sure that the participant performed the test as intended before entering the scanner room. Instructions were also repeated before each fMRI run during scanning. Stimulus presentation and timing of stimuli were achieved using E-prime 1.2 (Psychology Software Tools, Pittsburgh, PA). The paradigm was presented using MRI-compatible video-goggles (VisualSystem, Nordic NeuroLab, Bergen, Norway) for 41 participants. Because of technical problems with the goggles, an MRI-compatible monitor (Siemens AG, Erlangen, Germany) and a head-coil mounted mirror were used for the remaining 46 participants. The video goggles and monitor were tested for differences in timing of stimuli presentation using photodiodes and an oscilloscope. The test revealed that the monitor had a ∼60-msec stimulus onset delay relative to the video goggles. This was adjusted for during postprocessing of response and fMRI data. Responses were measured using fiberoptic response grips (ResponseGrip, Nordic NeuroLab, Bergen, Norway) and stored in log files by utilizing a custom-made Python-based log script that interacted with the E-Prime software.

MRI Scanning

Scanning was performed on a 3-T Siemens Trio scanner with a 12-channel head matrix coil (Siemens AG, Erlangen, Germany). Foam pads were used to minimize head motion. In both fMRI runs, 380 T2*-weighted BOLD sensitive images were acquired using an EPI pulse sequence (repetition time [TR] = 2400 msec, echo time [TE] = 35 msec, field of view [FOV] = 244 mm, slice thickness = 3.0 mm, slice number = 40, matrix = 80 × 80, giving an in-plane resolution of 3 × 3 mm) with slices positioned transversal along the A–P axis. For anatomical reference, a T1-weighted 3-D volume was acquired with a magnetization prepared rapid gradient echo sequence (TR = 2300 msec, TE = 2.88 msec, FOV = 256 mm, slice thickness = 1.20 mm, matrix 256 × 256, giving an in-plane resolution of 1.0 × 1.0 mm). In addition, two spin echo volumes were sampled to be used for distortion correction (Holland, Kuperman, & Dale, 2010): one acquired in the A–P direction, and the other, in P–A. Apart from different phase encoding directions, the spin echo protocols were identical (TR = 2010 msec, TE = 35 msec, FOV = 244 mm, slice thickness = 3 mm, matrix 80 × 80, giving an in-plane resolution of 3 × 3 mm).

Analysis of Behavioral Data

Calculation of CPT Performance Measures

The following measures were computed based on the behavioral raw data: omission errors, the number of targets a participant failed to respond to; commission errors, the number of nontargets a participant mistakenly responded to; hit RT, the mean RT of correct responses; and hit RT standard error of the mean (SEM), the standard error of the mean of hit RT. In addition, measures derived from signal detection theory (Green & Swets, 1966) were computed, namely detectability (d′) and response style (β). Detectability (d′) represents the relation between the signal (targets) and noise (nontargets) distribution and is considered to be a good measure of the discriminative power of an individual. The rationale behind this measure is that a participant with a good ability to distinguish and detect target and nontarget stimuli will have large differences between the signal and noise distributions, whereas an individual with decreased ability to distinguish and detect target and nontarget stimuli will have small differences. Detectability (d′) was computed by applying the equation d′ = zHzFA where H represented hits and FA represented false alarms, and the functions zH and zFA represented the inverse of the cumulative normal distribution of the hit rate and the false alarm rate. Hits were defined as correct responses to targets, whereas false alarms were defined as erroneous responses to nontargets (commission errors). Hit rate was found by dividing the actual number of hits in the CPT task by the number of possible hits (432), and false alarm rate was calculated by dividing number of false alarms by the number of possible false alarms (48), respectively. Response style (β) is a measure representing response tendency, which is considered related to how risky an individual is when responding to stimuli on a CPT. High β value reflects a cautious response style with emphasis on avoiding commission errors, whereas a low β value reflects a risky response style with more emphasis on avoiding omission errors. Response style (β) was computed in the following manner: C = −0.5(zHzFA), ln β = d′ × C, β = exp(ln β). For participants who had hit rates or false alarm rates of 1, d′ and β were calculated using a maximum value for hit rate to (N − 1)/N, which is a commonly used approximation in signal detection theory. Furthermore, if participants had a false alarm rate of 0, β was calculated using a minimum false alarm rate of 1/N.

Statistical Analyses

All behavioral data were analyzed with IBM SPSS 19.0. Means, standard deviations (SD), and confidence intervals (CI) for the overall means of not-X CPT measures were calculated. To investigate TOT effects, the CPT was divided into four equally long time epochs after collapsing run 1 and run 2. Each of the six CPT measures was included in separate repeated-measures ANOVAs with time epoch (1, 2, 3, and 4) as a fixed factor. Mauchley's test was used to investigate the assumption of sphericity of the data, and a Greenhouse–Geisser correction was utilized if this assumption was violated. Subsequent polynomial trend analyses were performed to explore the nature of significant effects from the repeated-measures ANOVAs. For all tests, the acceptance level for significant results was set to p < .05, and a Bonferroni correction giving a critical p > .0083 (Bonferroni corrected, α/n = 0.05/6 = 0.0083) was applied to control for multiple statistical tests where appropriate (univariate ANOVAs). Eta squared (η2) was calculated as a measure of effect size.

Analysis of MRI Data

Preprocessing

All fMRI data were processed utilizing the FMRIB's Software Library (FSL) toolbox version 4.1 (FMRIB Centre, Oxford, United Kingdom) and a custom algorithm for correction of susceptibility artifacts (Holland et al., 2010). After nonbrain removal using the Brain Extraction Tool (Smith, 2002), MCFLIRT (Jenkinson, Bannister, Brady, & Smith, 2002) was applied for affine motion correction. Correction of inhomogeneous static magnetic field-induced distortion was done by calculating displacement fields using spin-echo EPI sequences with opposite phase encoding, which was then applied for correction of the gradient EPI sequences (Holland et al., 2010). A Gaussian kernel of 6-mm FWHM was applied for spatial smoothing. Grand-mean intensity normalization of the entire 4-D data set was performed, and high-pass temporal filtering was set to 50 sec for the block-related analysis and 25 sec for the event-related analysis. Linear registration (7 df) using FLIRT (Jenkinson et al., 2002; Jenkinson & Smith, 2001) was performed to register each individual's fMRI data to their own high-resolution structural image. Nonlinear registration (12 df) from high-resolution structural to MNI152 1-mm standard template using an 8-mm warp resolution was done by utilizing FNIRT (Anderson, Jenkinson, & Smith, 2007a, 2007b). BOLD activity related to task blocks and individual trials was modeled using a general linear model (GLM). The hemodynamic response function was convolved using a Gamma variate, which is a normalization of the probability density function of the Gamma function (phase = 0 sec, SD = 3 sec, mean lag = 6 sec).

Main Contrasts

For each run, the following main contrasts were performed for all participants: (1) stable task-set maintenance (task block > fixation block) and (2) task-negative activation (fixation block > task block) for block analysis and (3) conflict processing (nontargets > targets), (4) hits (targets > nontargets), (5) error processing (commission errors > successful inhibition), and (6) successful inhibition (successful inhibition > commission errors) for event-related analysis (adaptive task control). Both runs were then combined in a higher level analysis for each individual using a fixed effects model. Finally, a mixed effects model was applied to create group average statistic images, which were corrected for multiple comparisons using GRF-theory-based maximum height thresholding with p < 10−13 (voxel corrected) for contrasts 1–4. Because of very robust activations, this stringent threshold was necessary to generate meaningful statistical parametric maps for interpretation. For the error processing and successful inhibition contrast, a more standard GRF-theory-based maximum height threshold of p < .01 (voxel corrected) was applied because of the lower number of trials included in this contrast as compared with the other two. Omission errors were not included in any separate contrasts because of low statistical power because of a large number of participants with very few or no such errors (floor effect). However, for exploratory purposes, omission errors were investigated in a subsequent regression model (see below). A conjunction analysis was performed to investigate brain areas that were activated above the statistical threshold (p < 10−13) in both adaptive task control (conflict processing) and stable task maintenance and therefore would represent a “core network.” Error processing was not included in this formalized conjunction analysis because of the considerable difference in statistical threshold for this contrast compared with the other two. However, all contrasts were also evaluated more qualitatively with regard to overlap in standard Montreal Neurological Institute (MNI) space.

TOT Effects

To investigate TOT effects, two separate repeated-measures GLMs were applied. In the first GLM, we investigated the TOT effect with the stable task maintenance contrast as a dependent variable and time epoch (1, 2, 3, and 4) as a fixed factor. In the second GLM, the same procedure was performed for investigating the TOT effect for adaptive task control by using the conflict processing contrast as the dependent variable. Significant main effects were followed up by individual contrasts comparing the first time epoch (epoch 1) to each of the subsequent ones (2, 3, and 4; progressively more distant). Statistical images were thresholded using GRF-theory-based maximum height thresholding with p < .05 (voxel corrected).

Correlations with CPT Measures

Positive and negative neural correlates to the measures of attention (except commission errors, which were included in a separate contrast) computed from the behavioral data were investigated by including hit RT, hit RT SEM, omission errors, detectability (d′), and response style (β) in separate regression models for the two the main contrasts, stable task maintenance and conflict processing (adaptive task control). The final statistical images from the correlation analyses were thresholded using GRF-theory-based maximum height thresholding with p < .05 (voxel corrected).

Cluster Algorithm and Presentation of Imaging Data

For all contrasts, a cluster algorithm was applied to obtain the main peak Z values and size of clusters (number of voxels) in 1 × 1 × 1 mm MNI space. Only clusters consisting of >40 voxels were included. The Harvard Oxford cortical and subcortical structural brain atlases as incorporated in the FSL software and visual inspection were used to denote relevant anatomical structures.

RESULTS

Behavioral Results

Overall means, SD, and 95% CI for the different not-X CPT measures are presented in Table 1.

Table 1. 

Overall Not-X CPT Performance Data


N
Mean
SD
95% CI of the Mean
Hit RT (msec) 87 419.79 54.48 [408.18, 431.41] 
Hit RT SEM 87 6.31 2.44 [5.79, 6.83] 
Omission errors 87 7.69 13.49 [4.81, 10.57] 
Commission errors 87 15.57 8.12 [13.84, 17.30] 
Detectability (d′) 87 2.92 0.79 [2.75, 3.09] 
Response style (β) 87 0.12 0.17 [0.09, 0.16] 

N
Mean
SD
95% CI of the Mean
Hit RT (msec) 87 419.79 54.48 [408.18, 431.41] 
Hit RT SEM 87 6.31 2.44 [5.79, 6.83] 
Omission errors 87 7.69 13.49 [4.81, 10.57] 
Commission errors 87 15.57 8.12 [13.84, 17.30] 
Detectability (d′) 87 2.92 0.79 [2.75, 3.09] 
Response style (β) 87 0.12 0.17 [0.09, 0.16] 

TOT Effects

Mauchley's test indicated that the assumption of sphericity was violated for the main effect of hit RT, hit RT SEM, omission errors, and response style (β). For these variables, the degrees of freedom were corrected using the Greenhouse–Geisser estimates of sphericity (ɛ = 0.657 for the main effect of hit RT, ɛ = 0.684 for the main effect of hit RT SEM, ɛ = 0.621 for the main effect of omission errors, and ɛ = 0.675 for the main effect of response style). All effects are reported as significant at p < .05 (Bonferroni corrected, α/n = 0.05/6 = 0.0083). Results from separate one-way repeated-measures ANOVAs showed a main effect of TOT for hit RT, F(1.971, 85.029) = 12.3, p = .000, η2 = 0.125, omission errors, F(1.863, 85.137) = 9.309, p = .000, η2 = 0.098, detectability, F(3, 84) = 6.104, p = .001, η2 = 0.066, and response style, F(2.024, 84.976) = 6.344, p = .002, η2 = 0.069. Statistically significant results are presented in Figure 1. There were no significant main effects of TOT for hit RT SEM, F(2.053, 84.947) = 1.954, p = .144, η2 = 0.022, or commission errors, F(3, 84) = 3.187, p = .024. A polynomial trend analysis indicated that there was a significant linear increase in hit RT with TOT, F(1, 86) = 16.004, p = .000, η2 = 0.157. Furthermore, there was a significant cubic trend, F(1, 86) = 13.133, p = .0000, η2 = 0.132, indicating that hit RT first increased from block 1 to 2, then slightly decreased from block 2 to 3, and then increased from block 3 to 4 (Figure 1). There was also a significant linear increase in omission errors with TOT, F(1, 86) = 13.271, p = .0000, η2 = 0.134, and response style, F(1, 86) = 9.614, p = .003, η2 = 0.101, as well as a linear decrease in detectability score, F(1, 86) = 10.605, p = .002, η2 = 0.110. The detectability score also demonstrated a significant cubic trend, indicating that detectability was relatively stable from block 1 to 2, decreased substantially from block 2 to 3, and then slightly increased from block 3 to 4 (Figure 1).

Figure 1. 

TOT effects for not-X CPT performance. The graphs show the development of CPT performance (group means ± standard error) with TOT. Only CPT measures with a statistically significant (p < .05, Bonferroni corrected) effect of TOT are included. Results from a polynomial trend analysis (p < .05) are indicated by = linear trend, = quadratic trend, = cubic trend.

Figure 1. 

TOT effects for not-X CPT performance. The graphs show the development of CPT performance (group means ± standard error) with TOT. Only CPT measures with a statistically significant (p < .05, Bonferroni corrected) effect of TOT are included. Results from a polynomial trend analysis (p < .05) are indicated by = linear trend, = quadratic trend, = cubic trend.

Imaging Results

Activations for Stable Task-set Maintenance and Adaptive Task Control

Brain activation related to performance of the not-X CPT task was found in frontal, parietal, subcortical, and cerebellar regions of the brain. The conjunction analysis revealed overlapping areas of activation between stable task-set maintenance and adaptive task control in insular and adjacent cortices bilaterally, paracingulate gyrus, right inferior parietal cortex, and right middle temporal gyrus (Figure 2, Table 4). Nonoverlapping activity unique for stable task-set maintenance was found in the dorso-caudal parts of the medial frontal cortex (MFC), left precentral gyrus, and dorsal striatum (bilaterally in the caudate and left putamen), in addition to the right thalamus and cerebellum (Table 2, Figure 2). There were also unique activations for adaptive task control. Conflict processing showed unique activity in more anterior parts of the MFC (relative to stable task-set maintenance), left precentral gyrus, posterior cingulate, right thalamus, and left caudate (Table 3, Figure 2). Error processing revealed unique activations within the MFC, right posterior insula, right precentral gyrus, left postcentral gyrus, and bilaterally within the temporal lobes and parietal regions (including the precuneus) as well as the lingual gyrus (Tables 4 and 5, Figure 3). Successful inhibition had unique activations in the frontal poles as well as in areas bilaterally in the occipital lobe (Table 5). Within the adaptive task control network, there was an overlap in activation between conflict processing and error processing in the anterior cingulate gyrus and between conflict processing and successful inhibition in the right inferior parietal lobe (IPL). Finally, there was task-negative activation (fixation block > task block) in parietal (including the precuneus and posterior cingulate cortex [PCC]) and occipital regions (Table 2, Figure 3) as well as activation related to hits (targets > nontargets) in the left precentral gyrus (Table 3).

Figure 2. 

Stable task-set maintenance, adaptive task control, and core network for cognitive control. The figure shows SPMs for the contrasts in stable task-set maintenance (task block > fixation block, p < 10−13, voxel corrected) and adaptive task control (conflict processing, nontargets > Targets, p < 10−13, voxel corrected) and a conjunction analysis indicating areas that are activated above the statistical threshold for both adaptive task control and stable task-set maintenance (p < 10−13, voxel corrected). Targets = all letters except the letter X. Nontargets = the letter X. Results are presented on a 1 × 1 × 1 mm MNI template.

Figure 2. 

Stable task-set maintenance, adaptive task control, and core network for cognitive control. The figure shows SPMs for the contrasts in stable task-set maintenance (task block > fixation block, p < 10−13, voxel corrected) and adaptive task control (conflict processing, nontargets > Targets, p < 10−13, voxel corrected) and a conjunction analysis indicating areas that are activated above the statistical threshold for both adaptive task control and stable task-set maintenance (p < 10−13, voxel corrected). Targets = all letters except the letter X. Nontargets = the letter X. Results are presented on a 1 × 1 × 1 mm MNI template.

Table 2. 

Peak Activations across Whole Brain for Stable Task Control (Task Block vs. Fixation Block)

Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Stable Task-set Maintenance 
Frontal orbital cortex 6,212 10.8 30 26 −3 
Insular cortex 5,437 10.5 −28 24 −3 
Paracingulate gyrus 4,450 10.7 10 50 
Precentral gyrus 1,086 9.56 −44 −2 35 
Supramarginal gyrus, posterior division 533 9.01 53 −45 42 
Cerebellum 494 9.33 −75 −20 
Precentral gyrus 209 8.99 45 38 
Middle temporal gyrus 195 9.04 56 −28 −11 
Putamen 180 8.97 −29 −8 
Caudate 171 9.21 14 
Thalamus 64 8.76 −7 −5 
Caudate 42 8.68 16 23 
 
Task-negative 
Occipital pole 3,249 10.8 31 −97 −11 
Occipital pole 2,201 10.2 −25 −98 −13 
Precuneus 1,380 9.98 −15 −65 20 
Precuneus 1,360 9.85 16 −61 20 
Precuneus 936 9.39 −69 61 
Cingulate gyrus, posterior division 124 8.94 −49 
Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Stable Task-set Maintenance 
Frontal orbital cortex 6,212 10.8 30 26 −3 
Insular cortex 5,437 10.5 −28 24 −3 
Paracingulate gyrus 4,450 10.7 10 50 
Precentral gyrus 1,086 9.56 −44 −2 35 
Supramarginal gyrus, posterior division 533 9.01 53 −45 42 
Cerebellum 494 9.33 −75 −20 
Precentral gyrus 209 8.99 45 38 
Middle temporal gyrus 195 9.04 56 −28 −11 
Putamen 180 8.97 −29 −8 
Caudate 171 9.21 14 
Thalamus 64 8.76 −7 −5 
Caudate 42 8.68 16 23 
 
Task-negative 
Occipital pole 3,249 10.8 31 −97 −11 
Occipital pole 2,201 10.2 −25 −98 −13 
Precuneus 1,380 9.98 −15 −65 20 
Precuneus 1,360 9.85 16 −61 20 
Precuneus 936 9.39 −69 61 
Cingulate gyrus, posterior division 124 8.94 −49 

Results were achieved using a mixed effects model corrected for multiple comparisons using GRF-theory-based maximum height thresholding with p< 10−13 (voxel corrected). Only the most significant (main) peak within each cluster is reported in the present table. n = 87 (34 women). Naming of anatomical regions associated with main peaks was based on the Harvard Oxford cortical and subcortical structural atlases as implemented in the FSL software. R = right, L = left. Note that some clusters are particularly large and therefore span over several brain regions (see Figures 2 and 3 as well as the Results and Discussion sections in the main text for more information).

Table 3. 

Peak Activations across Whole Brain for Adaptive Task Control (Nontargets vs. Targets)

Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Conflict Processing 
Cingulate gyrus, anterior division 14,547 11.8 34 25 
Frontal orbital cortex 11,272 12.7 37 22 −7 
Frontal orbital cortex 6,643 12.8 −35 19 −12 
Angular gyrus 5,411 10.1 60 −47 21 
Cingulate gyrus, posterior division 738 9.68 −21 24 
Supramarginal gyrus, posterior division 653 9.98 −55 −46 30 
Middle temporal gyrus, posterior division 588 9.78 57 −27 −13 
Caudate 369 9.03 12 
MFG 283 8.92 48 44 
Right Thalamus 82 8.87 −25 −4 
Caudate 72 8.81 −11 
 
Hits 
Precentral gyrus 340 9.08 −40 −27 62 
Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Conflict Processing 
Cingulate gyrus, anterior division 14,547 11.8 34 25 
Frontal orbital cortex 11,272 12.7 37 22 −7 
Frontal orbital cortex 6,643 12.8 −35 19 −12 
Angular gyrus 5,411 10.1 60 −47 21 
Cingulate gyrus, posterior division 738 9.68 −21 24 
Supramarginal gyrus, posterior division 653 9.98 −55 −46 30 
Middle temporal gyrus, posterior division 588 9.78 57 −27 −13 
Caudate 369 9.03 12 
MFG 283 8.92 48 44 
Right Thalamus 82 8.87 −25 −4 
Caudate 72 8.81 −11 
 
Hits 
Precentral gyrus 340 9.08 −40 −27 62 

Results were achieved using a mixed effects model corrected for multiple comparisons using GRF-theory-based maximum height thresholding with p < 10−13 (voxel corrected). Only the most significant (main) peak within each cluster is reported in the present table. n = 87 (34 women). Naming of anatomical regions associated with main peaks was based on the Harvard Oxford cortical and subcortical structural atlases as implemented in the FSL software. Targets = all letters except the letter X; nontargets = the letter X. Note that some clusters are particularly large and therefore span over several brain regions (see Figures 2 and 3 as well as the Results and Discussion sections in the main text for more information).

Table 4. 

Peak Activations across Whole Brain for Conjunction Analysis (Adaptive Task Control and Stable Task-set Maintenance), Representing the “Core Network” of Cognitive Control

Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Conjunction Analysis 
Frontal orbital cortex/insula 4,838 10.8 32 26 −3 
Insula/frontal orbital cortex 3,029 10.4 −29 23 −4 
Paracingulate gyrus 451 9.23 20 41 
Supramarginal gyrus, posterior division 353 8.95 54 −45 43 
Middle temporal gyrus, posterior division 191 9.04 56 −28 −11 
Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Conjunction Analysis 
Frontal orbital cortex/insula 4,838 10.8 32 26 −3 
Insula/frontal orbital cortex 3,029 10.4 −29 23 −4 
Paracingulate gyrus 451 9.23 20 41 
Supramarginal gyrus, posterior division 353 8.95 54 −45 43 
Middle temporal gyrus, posterior division 191 9.04 56 −28 −11 

Brain areas activated above the statistical threshold (p < 10−13) in both conflict processing (adaptive task control) and stable task-set maintenance were investigated in a conjunction analysis. The most significant (main) peak within each cluster is reported in the present table. n = 87 (34 women). Naming of anatomical regions associated with main peaks was based on the Harvard Oxford cortical and subcortical structural atlases as implemented in the FSL software.

Table 5. 

Peak Activations across Whole Brain for Adaptive Task Control (Commission Errors vs. Correct Inhibition)

Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Error Processing 
Lingual gyrus 19,334 7.11 −8 −62 
Lingual gyrus 4,981 6.16 17 −55 −1 
Central opercular cortex 4,443 7.02 59 −15 11 
Precentral gyrus 2,703 6.66 55 −5 36 
Postcentral gyrus 1,699 6.38 −49 −15 38 
Precuneus 1,583 6.33 −1 −53 63 
Cingulate gyrus, anterior division 1,356 5.89 −3 22 28 
Superior frontal gyrus 451 5.75 −7 11 61 
Temporal pole 310 5.94 28 −28 
Cingulate cortex, anterior division 224 5.38 −1 −10 38 
Parahippocampal gyrus, posterior division 177 5.29 −19 −26 −13 
Superior temporal gyrus, anterior division 137 5.12 66 −1 
Temporal pole 91 5.15 −29 15 −36 
 
Successful Inhibition 
Occipital pole 2,108 5.78 −28 −95 11 
Frontal pole 757 6.04 18 48 −19 
Supramarginal gyrus, posterior division 724 5.39 42 −44 46 
Frontal pole 588 5.69 −38 54 −8 
Frontal pole 559 5.29 47 36 22 
Frontal pole 389 5.51 37 58 −4 
Lateral occipital cortex, superior division 213 5.41 32 −69 36 
Lateral occipital cortex, inferior division 180 5.37 −43 −75 −3 
Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Error Processing 
Lingual gyrus 19,334 7.11 −8 −62 
Lingual gyrus 4,981 6.16 17 −55 −1 
Central opercular cortex 4,443 7.02 59 −15 11 
Precentral gyrus 2,703 6.66 55 −5 36 
Postcentral gyrus 1,699 6.38 −49 −15 38 
Precuneus 1,583 6.33 −1 −53 63 
Cingulate gyrus, anterior division 1,356 5.89 −3 22 28 
Superior frontal gyrus 451 5.75 −7 11 61 
Temporal pole 310 5.94 28 −28 
Cingulate cortex, anterior division 224 5.38 −1 −10 38 
Parahippocampal gyrus, posterior division 177 5.29 −19 −26 −13 
Superior temporal gyrus, anterior division 137 5.12 66 −1 
Temporal pole 91 5.15 −29 15 −36 
 
Successful Inhibition 
Occipital pole 2,108 5.78 −28 −95 11 
Frontal pole 757 6.04 18 48 −19 
Supramarginal gyrus, posterior division 724 5.39 42 −44 46 
Frontal pole 588 5.69 −38 54 −8 
Frontal pole 559 5.29 47 36 22 
Frontal pole 389 5.51 37 58 −4 
Lateral occipital cortex, superior division 213 5.41 32 −69 36 
Lateral occipital cortex, inferior division 180 5.37 −43 −75 −3 

Results were achieved using a mixed effects model corrected for multiple comparisons using GRF-theory-based maximum height thresholding with p < .01 (voxel corrected). Only the most significant (main) peak within each cluster is reported in the present table. n = 87 (34 women). Naming of anatomical regions associated with main peaks was based on the Harvard Oxford cortical and subcortical structural atlases as implemented in the FSL software. Commissions = failure to withhold button press when the letter X was presented; successful inhibition = correctly withholding the button press when the letter X was presented. Note that some clusters are particularly large and therefore span over several brain regions (see Figures 3 and 4 as well as the Results and Discussion sections in the main text for more information).

Figure 3. 

Stable task-set maintenance, task-negative activity, error related activity, and TOT effects. The figure shows SPMs for stable task-set maintenance (p < 10−13, voxel corrected), task-negative activity (p < 10−13, voxel corrected), error processing (p < .01), and TOT increase and decrease (p < .05). Commissions = failure to withhold button press when the letter X was presented. Correct = correctly withholding the button press when the letter X was presented. Results are presented on a 1 × 1 × 1 mm MNI template.

Figure 3. 

Stable task-set maintenance, task-negative activity, error related activity, and TOT effects. The figure shows SPMs for stable task-set maintenance (p < 10−13, voxel corrected), task-negative activity (p < 10−13, voxel corrected), error processing (p < .01), and TOT increase and decrease (p < .05). Commissions = failure to withhold button press when the letter X was presented. Correct = correctly withholding the button press when the letter X was presented. Results are presented on a 1 × 1 × 1 mm MNI template.

TOT Effects

Repeated-measures GLMs revealed a significant main effect of TOT for stable task-set maintenance, but no such effect for adaptive task control (conflict processing). For stable task-set maintenance, follow-up analyses showed increased activation in the left frontal pole and bilaterally in the occipital poles in time epochs 3 and 4 relative to epoch 1 (Table 6, Figure 3). The activations in the occipital poles were clearly overlapping with task-negative areas described in the previous section (Figure 3). For epoch 4 relative to epoch 1, there was also increased activation in other frontal, parietal, temporal, and occipital areas that, to some degree, overlapped or were adjacent to the task-negative network (Table 6, Figure 3). We also found decreased activity as an effect of TOT from time epoch 1 to epoch 4. Areas of decreased activations were located to frontal, parietal, and temporal brain regions in addition to the cerebellum and were, to a large degree, overlapping with the task-positive regions described in the previous section (Table 6, Figure 3).

Table 6. 

Peak Activations across Whole Brain for TOT Effects for Stable Task-set Maintenance

Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Time Epoch 1 > 4 
Cingulate gyrus, anterior division 2,400 5.19 11 41 
Precentral gyrus 1,794 5.53 −42 −4 60 
Insular cortex 1,514 5.29 30 21 
Cerebellum 1,421 5.47 −1 −74 −18 
Insular cortex 1,303 5.34 −42 11 −2 
Precentral gyrus 685 5.12 42 −4 44 
Frontal pole 260 4.68 32 42 23 
Parietal opercular cortex 249 4.63 −47 −34 22 
Juxtapositional Lobule cortex 208 5.10 −8 −6 60 
Superior frontal gyrus 193 4.77 14 59 
Superior temporal gyrus, posterior division 143 4.90 −66 −34 18 
Cerebellum 90 4.76 −35 −60 −29 
MFG 79 4.59 44 59 
Superior frontal gyrus 42 4.61 −30 −50 70 
 
Time Epoch 3 > 1 
Occipital pole 1,586 6.06 30 −97 −11 
Occipital pole 927 5.59 −26 −98 −12 
Frontal pole 59 4.67 −18 37 −18 
 
Time Epoch 4 > 1 
Precuneus 4,610 5.08 −64 26 
Frontal pole 3,437 6.32 −19 37 −18 
Occipital pole 1,671 6.15 29 −97 −11 
Occipital pole 1,354 6.18 −25 −98 −12 
Frontal pole 353 4.79 −41 40 −19 
Frontal pole 266 4.81 −5 69 16 
Frontal orbital cortex 166 4.57 19 34 −18 
Superior frontal gyrus 120 4.74 −14 34 39 
Precentral gyrus 88 4.62 14 −28 66 
Frontal pole 79 4.61 −13 48 37 
Precentral gyrus 58 4.59 −10 −30 69 
Subcallosal cortex 56 4.72 −7 −17 
Lateral occipital cortex 52 4.54 −38 −71 31 
Middle temporal cortex 49 4.48 −52 −4 −25 
Anatomical Region
R/L
Size (Number of Voxels)
Z
Coordinates for Peak Activation (MNI)
x
y
z
Time Epoch 1 > 4 
Cingulate gyrus, anterior division 2,400 5.19 11 41 
Precentral gyrus 1,794 5.53 −42 −4 60 
Insular cortex 1,514 5.29 30 21 
Cerebellum 1,421 5.47 −1 −74 −18 
Insular cortex 1,303 5.34 −42 11 −2 
Precentral gyrus 685 5.12 42 −4 44 
Frontal pole 260 4.68 32 42 23 
Parietal opercular cortex 249 4.63 −47 −34 22 
Juxtapositional Lobule cortex 208 5.10 −8 −6 60 
Superior frontal gyrus 193 4.77 14 59 
Superior temporal gyrus, posterior division 143 4.90 −66 −34 18 
Cerebellum 90 4.76 −35 −60 −29 
MFG 79 4.59 44 59 
Superior frontal gyrus 42 4.61 −30 −50 70 
 
Time Epoch 3 > 1 
Occipital pole 1,586 6.06 30 −97 −11 
Occipital pole 927 5.59 −26 −98 −12 
Frontal pole 59 4.67 −18 37 −18 
 
Time Epoch 4 > 1 
Precuneus 4,610 5.08 −64 26 
Frontal pole 3,437 6.32 −19 37 −18 
Occipital pole 1,671 6.15 29 −97 −11 
Occipital pole 1,354 6.18 −25 −98 −12 
Frontal pole 353 4.79 −41 40 −19 
Frontal pole 266 4.81 −5 69 16 
Frontal orbital cortex 166 4.57 19 34 −18 
Superior frontal gyrus 120 4.74 −14 34 39 
Precentral gyrus 88 4.62 14 −28 66 
Frontal pole 79 4.61 −13 48 37 
Precentral gyrus 58 4.59 −10 −30 69 
Subcallosal cortex 56 4.72 −7 −17 
Lateral occipital cortex 52 4.54 −38 −71 31 
Middle temporal cortex 49 4.48 −52 −4 −25 

After a repeated-measures GLM yielded a statistically significant main effect of TOT, subsequent paired t tests were performed which compared time epoch 1 with each of the other time epochs (2, 3, and 4) progressively further away in time. Only statistically significant results from these analyses are presented in this table. Results were achieved using a mixed effects model corrected for multiple comparisons using GRF-theory-based maximum height thresholding with p < .05 (voxel corrected). Only the most significant (main) peak within each cluster is reported in the present table. n = 87 (34 women). Naming of anatomical regions associated with main peaks was based on the Harvard Oxford cortical and subcortical structural atlases as implemented in the FSL software. Note that some clusters are particularly large and therefore span over several brain regions (see Figures 3 and 4 as well as the Results and Discussion sections in the main text for more information).

Correlation Analyses

There was a positive correlation between brain activation related to adaptive task control (conflict processing) and detectability (d′), with main peak in the juxtapositional (former SMA) cortex (main peak MNI coordinates: x = −4, y = −6, z = 55, size = 83 voxels, Z = 4.9). None of the other CPT measures correlated with brain activation in neither the adaptive task control (conflict processing) nor stable task maintenance contrasts.

DISCUSSION

The present fMRI study investigating the neural underpinnings of not-X CPT performance revealed three main findings: (1) overlapping as well as nonoverlapping brain activation in cortical, subcortical, and cerebellar regions related to stable task-set maintenance and adaptive task control (including conflict processing, error processing, and successful inhibition); (2) activations within the frontal cortex were by and large localized to more rostral regions during adaptive task control as compared with stable task-set maintenance; and (3) brain activity decreased in task-positive and increased in task-negative/DMN regions with TOT for stable task-set maintenance, whereas no TOT effects were found for adaptive task control.

A Core Network for Cognitive Control

During the not-X CPT task, a widespread cortical–subcortical cerebellar network was engaged, where overlapping areas of activation in stable set maintenance and adaptive task control were located to the insula and adjacent cortex, paracingulate cortex, right inferior partial lobe, and right middle temporal gyrus.

Our findings support the existence of a core system for task control, which includes the insula and the MFC (Dosenbach et al., 2006). By using a conservative threshold (p < 10−13, voxel corrected) compared with most studies, there was an increased risk for type II errors in this study, which could potentially lead to a bias toward large effects (Lieberman & Cunningham, 2009). However, the proposed core network has previously been found to be relatively robust (Dosenbach et al., 2006). Taking this into consideration, in addition to the large sample size (Thyreau et al., 2012), a conservative threshold was chosen in this study, as type I errors were considered to be less desirable than type II errors when trying to determine a true core network.

Contrary to previous research, we failed to find overlapping activity directly related to error processing in the anterior insula and MFC (Dosenbach et al., 2006). Neither did we find overlapping activity there for successful inhibition. This indicates that, although adaptive task control in general may belong to a core network together with stable task-set maintenance, certain adaptive properties are likely to be domain or task specific. A more conventional threshold was used when investigating error processing and successful inhibition (p < .01, voxel corrected), because of the fact that there were considerably less trials included in these contrasts (lower statistical power) as compared with the more-general conflict processing contrast. The lack of overlap, despite more liberal thresholds, supports the interpretation that they do not belong to the core network but rather are domain specific.

We also found overlapping activation for stable task-set maintenance and adaptive task control in the IPL and middle temporal gyrus. The IPL is recognized as part of a dorsal attention system that is connected to orienting (Fan et al., 2005; Corbetta & Shulman, 2002), but has also been related to not-X CPT performance (Tana et al., 2010; Ogg et al., 2008) and moment-to-moment adjustment during task performance (Wilk et al., 2012) as well as semantic and phonological processing and categorization of visual stimuli (Stoeckel, Gough, Watkins, & Devlin, 2009). This brain region was also activated for the individual types of control networks in the study performed by Dosenbach and colleagues; however, they failed to find a direct overlap (Dosenbach et al., 2006), indicating that these brain areas may represent more domain (task) specific processing rather than a generalized top–down cognitive control. Whereas Dosenbach and colleagues based their analyses on data from several tasks with different task demands and stimulus types (Dosenbach et al., 2006), this study is limited by the use of only one task. This may simply lead to a larger degree of overlap because both block- and event-related analyses are likely to be affected by the same domain-specific demands of the task. If this is the case, low demand on spatial orienting in our task makes it plausible that the overlapping IPL activity we found could be partly related to semantic and phonological processing and categorization of the letters (Stoeckel et al., 2009) rather than spatial orientation (Fan et al., 2005). The most pronounced and explicit demand on semantic processing in the not-X CPT was the instruction to distinguish between the semantic meaning given to the letter X (“stop responding”) versus that of all the other letters (“respond”). In addition to this, participants were engaged in continuously processing letters throughout the task. Such processing could have activated the IPL, as semantic and phonological processing of meaningful stimuli has previously shown to recruit this region regardless of explicit task demands (Binder, Desai, Graves, & Conant, 2009; Stoeckel et al., 2009). Overlapping activity in the right middle temporal gyrus is also likely associated with semantic processing (Visser, Jefferies, Embleton, & Lambon Ralph, 2012; Laufer, Negishi, Lacadie, Papademetris, & Constable, 2011; Binder et al., 2009). Increased activity in both the IPL and middle temporal gyrus during successful inhibition further supports the view that these regions are involved in successful performance.

Interestingly, conflict processing revealed additional activation in the left supramarginal gyrus, indicating that additional resources may be recruited specifically in relation to the cognitive conflict that arises when processing nontargets as opposed to targets (Ettinger et al., 2008). Conflict cannot easily be distinguished from differences related to responding or not responding (Kim, Chung, & Kim, 2012). However, it is unlikely that the additional activation in the left supramarginal gyrus in the current study is merely caused by differences related to responding and not responding. Typical activation related to responding was found only in motor areas when investigating activation directly related to hits (targets > nontargets). Also, if the activation had been directly related to responding versus not responding, it would be expected that this difference would be isolated in the successful inhibition or error processing contrasts. This was not the case, as the activation in the left supramarginal gyrus was unique for the conflict processing contrast (nontargets > targets). Finally, a recent study demonstrated that the left supramarginal gyrus has a causal role in relation to the left dorsal premotor cortex in rapid action reprogramming (Hartwigsen et al., 2012). Evidence from this study supports that the supramarginal gyrus is involved in controlling the release of action programs regardless of whether they involve responding or not responding (Hartwigsen et al., 2012).

Anteriorization of Adaptive Relative to Stable Task Control in the Frontal Cortex

In addition to the “core” activation located within the MFC, stable task-set maintenance activated caudal parts of the medial superior frontal gyrus, whereas adaptive task control (conflict and error processing) activated rostral regions of ACC. The present results are therefore best accommodated by the cognitive demand framework (Badre & D'Esposito, 2007, 2009; Badre, 2008) rather than the episodic and contextual control framework (Koechlin & Summerfield, 2007; Koechlin et al., 2003). Accordingly, the results indicate that, with an event-related analysis, we can identify activation related to the more specific and perhaps demanding actions of conflict and error processing, whereas the activity present in the block analysis appears to be dominated by the ongoing task-set related to simple motor responses (Kim et al., 2011; Venkatraman et al., 2009).

Moreover, evidence has demonstrated that the more dorsal region of ACC as well as the pre-SMA are involved in action selection and conflict resolution (Forstmann, van den Wildenberg, & Ridderinkhof, 2008; Taylor, Nobre, & Rushworth, 2007), whereas the relatively more ventro-rostral part of ACC is mainly activated during error detection and anticipatory prediction of response selection (Nee et al., 2011). In accordance with this, our results revealed that main activations related to conflict and error processing was located in more ventro-rostral regions of ACC, relative to that of stable task-set maintenance.

One exception from the anteriorization of adaptive task control within the PFC was present in our data, namely the error-related activation in the very posterior ACC. The activated region is, however, bordering the PCC and may hence not be representative for the prefrontal region. Activation of the PCC has been related to both a task-negative network involved in lapses of attention (Weissman et al., 2006) and preparatory motor inhibition (Hu & Li, 2012). The rostral PCC may be specifically involved in processing errors, as the more general conflict processing contrast activated an area far more caudal in the same region. From this, it could be hypothesized that there is a rostro-caudal distribution within the PCC with regards to more general versus complex processing.

The pre-SMA and adjacent areas were activated by both stable task-set maintenance and error processing, with the latter located more rostral (and somewhat dorsal). Block-related fMRI analysis has previously demonstrated a larger number of activated voxels in dorsal, relative to ventral, areas during a not-X CPT performance (Ogg et al., 2008). Furthermore, stable task-set maintenance has been found to activate more dorsally within the MFC relative to adaptive task control (Wilk et al., 2012). Novel in our study is that we investigated error processing more directly. The pre-SMA area is hypothesized to be involved in switching between and/or reactivating task-sets when this is required, and error-related activity in the same area may signal re-engagement of decaying task-sets (Nee et al., 2011; Altmann & Gray, 2002). Memory for task-sets is thought to be a noisy system, where decay in task-sets leads to lower discriminative power (given by lower d′) for distinguishing one stimulus from the other (Altmann & Gray, 2002). In our task, we found a positive correlation between the conflict processing and detectability (d′) in the pre-SMA area, giving direct support for its role in separating signal from noise.

There was also evidence for a distribution gradient within the lateral prefrontal lobe, where adaptive task control was associated with more anterior activation than stable task-set maintenance. Stable task-set maintenance activation spreads from its main peak in the precentral gyrus into the middle frontal gyrus (MFG), which is known to be involved in maintaining task goals as well as manipulating items in working memory (Rypma, Prabhakaran, Desmond, Glover, & Gabrieli, 1999). Conflict processing was also associated with activation more anterior in the MFG relative to stable task-set maintenance. Moreover, in addition to enhanced activity in primary visual areas, successful inhibition revealed increased activation with main peaks bilaterally in the frontal poles. In the right hemisphere, this activation spreads into the MFG, anterior to both stable task-set maintenance and conflict processing, indicating that successful inhibition relies on endogenous control processes located in the most anterior regions in the PFC (Kim et al., 2011; Taren, Venkatraman, & Huettel, 2011).

Distinct Subcortical and Cerebellar Regions Recruited by Stable Task-set Maintenance and Adaptive Task Control

Activations related to stable task-set maintenance were found in the dorsal striatum (bilaterally in the caudate and left putamen), right thalamus, and cerebellum. The dorsal striatum has an important role in cognitive control and categorization of visual stimuli (Seger, 2008; Balleine et al., 2007; Heyder et al., 2004). Stable task-set maintenance activated anterior parts (the head) and also more posterior parts spreading into the body of the caudate, in addition to the left putamen. These areas are thought to be key structures in separate cortico-striatal loops, namely the executive (head) and visual (body) loop (Seger, 2008). Conflict processing also activated the head of the caudate, although separated from, and located dorso-caudal relative to the stable task-set maintenance activation.

Largely overlapping thalamic activations related to both alerting and executive control have been reported in a previous event-related fMRI study (Fan et al., 2005). In our study, using a more stringent threshold, we were able to find separate thalamic activation for stable task-set maintenance and adaptive task control. Whereas stable task-set maintenance showed activation in the ventral anterior nuclei region previously known to be involved in executive function (Little et al., 2010; Van der Werf, Witter, Uylings, & Jolles, 2000), conflict processing revealed activation in the pulvinar region (in vicinity of the peak activation related to executive control in the Fan et al. study), which plays a role in working-memory-guided visual selection (Rotshtein, Soto, Grecucci, Geng, & Humphreys, 2011). Previous research has suggested a role for the thalamus as a part of a cingulo-opercular network, which is supposed to mainly engender stable task-set maintenance (Dosenbach et al., 2008). Our results refine this view, suggesting that distinct subregions of the thalamus may be specifically involved in both stable task-set maintenance and adaptive task control.

The cerebellum has also been proposed to be involved in a cingulo-opercular network by providing error codes through interacting with the thalamus and a fronto-parietal network through connections via the dorsolateral PFC and IPL (Dosenbach et al., 2007). Another theoretical perspective sees the cerebellum as crucial for maintenance of anticipatory brain activity that is subsequently synchronized with the expected sensory stimuli, facilitating a more sustained “predictive brain state” (Ghajar & Ivry, 2009). In our study, we failed to directly relate activation in the cerebellum to error processing, as we only found activations related to stable task-set maintenance. These results support a role for the cerebellum as a comparator or internal template in attention and executive control (Ghajar & Ivry, 2009) and not an error detector as such.

Task-negative Network and Error Processing

Task-negative not-X CPT activations were located to posterior brain regions similar to those previously found in a CPT study (Ogg et al., 2008) and coinciding with the DMN (Fox et al., 2005). We also found error-related activity in several of the same regions (lateral and medial parietal regions including the precuneus), possibly indicating a failure to effectively deactivate the task-negative network when errors occurred (Weissman et al., 2006). However, not all error processing areas overlapped with the task-negative regions. Also, in addition to the previously described activations in PFC, PCC, and task-negative regions, error processing was uniquely supported by the right posterior insula, left postcentral gyrus, bilaterally in the temporal lobes, and central opercular cortices. This activity may be related to more specific error activity such as affective or cognitive reactions related to making errors (Mathiak et al., 2011) or reactive activation of domain-/task-specific neural networks.

TOT Effects

This study is the first to investigate TOT effects for both stable task-set maintenance and adaptive task control in the same fMRI experiment. Interestingly, we found a statistically significant TOT effect only for stable task-set maintenance and not for adaptive task control. This finding gives further support for dissociation between adaptive versus stable networks. Furthermore, for stable task-set maintenance, there was primarily a decrease of activation in task-positive and an increase in task-negative (our task) or DMN regions (Fox et al., 2005; Fransson, 2005) as a function of TOT.

A previous study found reduction of CBF in a fronto-parietal task-positive attention network after 20 min of performing a psychomotor vigilance task, which was related to decreased task performance (Lim et al., 2010). Others have found decreased BOLD response in task-positive networks that were unrelated to behavioral performance as the task progressed, leading the authors to attribute their findings mainly to habituation effects (Tana et al., 2010; Butti et al., 2006). The fact that TOT effects in our study were both positive and negative and overlapping with the original task-positive and task-negative networks makes it unlikely that the BOLD response changes are because of general habituation effects or global signal changes (Fox, Zhang, Snyder, & Raichle, 2009). The global fMRI signal may change over time, particularly in paradigms lasting for an extended period (e.g., because of scanner drift). Such effects were minimized in this study, both by application of conventional filtering of the data as well as by using a well-balanced task design and analysis approach (see Methods section). Although this balanced design theoretically reduced the sensitivity for detecting TOT effects (e.g., by collapsing two runs), it actually increased the specificity. Global signal effects are also unlikely for the present findings in particular, as there is evidence that the global signal is primarily not localized in the currently activated regions, resembling anticorrelated regions typically found in resting-state fMRI studies (Fox et al., 2009).

Both increases in attentional demands and self-reported level of fatigue have previously been associated with increased activations within typical task-positive areas and decreased activations in typical task-negative areas (Cook et al., 2007; McKiernan, Kaufman, Kucera-Thompson, & Binder, 2003), which is the opposite of the TOT effect in this study. Moreover, stimulus-independent thoughts have been associated with decreased activity in task-positive areas and increased activity in task-negative regions (McGuire, Paulesu, Frackowiak, & Frith, 1996). However, it has also been suggested that activation in DMN areas is involved in prospective planning (Buckner, Andrews-Hanna, & Schacter, 2008), which may suggest that increased activation in DMN areas may play a role in a proactive attention control system (Braver, 2012).

The general behavioral effect of TOT in our task was a linear decrease in performance on detectability and increase in RT and omission errors as well as increased β. This general worsening of performance with increased TOT is in accordance with previous research (Langner et al., 2010). Although there was a general decrease in performance, the RT and detectability curves could not be fully understood without also investigating nonlinear effects. The fact that these measures were more fluctuating with time could mean that they are supported by different underlying neural networks than those that elicit omission errors and higher β. However, we were not able to confirm this hypothesis in our fMRI analyses, which failed to reveal any statistically significant relationships between any of the behavioral CPT measures and BOLD activity for the stable task-set maintenance contrast.

Interestingly, there was a lack of significant TOT effects for commission errors. Along with increased RT, omission errors, and response style (β) scores, this may be an indication of a change to a more cautious response style with TOT, represented by an increased threshold for not-X responses. This interpretation may shed further light on the finding that TOT affected only neural activity related to stable task-set maintenance rather than dynamic control adjustments, as changes in the overall strategy (task-set) would be expected to mainly influence the former.

Summary and Conclusion

This study demonstrates novel aspects of the neural underpinnings of stable task-set maintenance and adaptive task control in 87 healthy participants, using a mixed block and event-related fMRI design. The results support the existence of an overlapping core network for cognitive control. In addition, we were able to map distinct brain regions underlying cognitive control during not-X CPT performance, which operates and reacts on different temporal scales. The results also indicate a rostro-caudal distribution in the frontal cortex where stable task-set maintenance is located more posteriorly in regions considered to be related to more general functions, whereas adaptive task control is located more anteriorly where more demanding and perhaps domain-specific operations are considered to be performed. Only the stable task-set maintenance network, and not the adaptive task control network, exhibited a TOT effect. The TOT effects in the stable task-set maintenance network were related to a decrease in task-positive activation and a parallel increase in task-negative/DMN activation.

A particular strength of the current study is the high number of participants and statistical power, which allowed for whole-brain analyses without abandoning strict thresholds for statistical significance and correction for multiple comparisons. This study contributes new knowledge by combining one of the most commonly administered cognitive tests with more recent, innovative neurocognitive theoretical perspectives and methods. This knowledge may give rise to valuable new questions within basic and clinical attention research as well as new perspectives for interpretation of clinical CPT results.

Reprint requests should be sent to Alexander Olsen, MI Lab and Department of Circulation and Medical Imaging, Norwegian University of Science and Technology, Postboks 8905, 7491 Trondheim, Norway, or via e-mail: alexander.olsen@ntnu.no.

REFERENCES

REFERENCES
Altmann
,
E. M.
, &
Gray
,
W. D.
(
2002
).
Forgetting to remember: The functional relationship of decay and interference.
Psychological Science
,
13
,
27
33
.
Anderson
,
J. L. R.
,
Jenkinson
,
M.
, &
Smith
,
S.
(
2007a
).
Non-linear optimisation.
Oxford
:
FMRIB Centre
.
Anderson
,
J. L. R.
,
Jenkinson
,
M.
, &
Smith
,
S.
(
2007b
).
Non-linear registration aka spatial normalisation.
Oxford
:
FMRIB Centre
.
Badre
,
D.
(
2008
).
Cognitive control, hierarchy, and the rostro-caudal organization of the frontal lobes.
Trends in Cognitive Sciences
,
12
,
193
200
.
Badre
,
D.
, &
D'Esposito
,
M.
(
2007
).
Functional magnetic resonance imaging evidence for a hierarchical organization of the prefrontal cortex.
Journal of Cognitive Neuroscience
,
19
,
2082
2099
.
Badre
,
D.
, &
D'Esposito
,
M.
(
2009
).
Is the rostro-caudal axis of the frontal lobe hierarchical?
Nature Reviews Neuroscience
,
10
,
659
669
.
Balleine
,
B. W.
,
Delgado
,
M. R.
, &
Hikosaka
,
O.
(
2007
).
The role of the dorsal striatum in reward and decision-making.
Journal of Neuroscience
,
27
,
8161
8165
.
Binder
,
J. R.
,
Desai
,
R. H.
,
Graves
,
W. W.
, &
Conant
,
L. L.
(
2009
).
Where is the semantic system? A critical review and meta-analysis of 120 functional neuroimaging studies.
Cerebral Cortex
,
19
,
2767
2796
.
Boehler
,
C. N.
,
Appelbaum
,
L. G.
,
Krebs
,
R. M.
,
Hopf
,
J. M.
, &
Woldorff
,
M. G.
(
2010
).
Pinning down response inhibition in the brain-conjunction analyses of the stop-signal task.
Neuroimage
,
52
,
1621
1632
.
Braver
,
T. S.
(
2012
).
The variable nature of cognitive control: A dual mechanisms framework.
Trends in Cognitive Sciences
,
16
,
106
113
.
Buckner
,
R. L.
,
Andrews-Hanna
,
J. R.
, &
Schacter
,
D. L.
(
2008
).
The brain's default network: Anatomy, function, and relevance to disease.
Annals of the New York Academy of Sciences
,
1124
,
1
38
.
Butti
,
M.
,
Pastori
,
A.
,
Merzagora
,
A.
,
Bianchi
,
A.
,
Bardoni
,
A.
,
Branca
,
V.
,
et al
(
2006
).
Combining near infrared spectroscopy and functional MRI during continuous performance test in healthy subjects.
Conference Proceedings: IEEE Engineering in Medicine and Biology Society
,
1
,
1944
1947
.
Conners
,
C. K.
,
Epstein
,
J. N.
,
Angold
,
A.
, &
Klaric
,
J.
(
2003
).
Continuous performance test performance in a normative epidemiological sample.
Journal of Abnormal Child Psychology
,
31
,
555
562
.
Cook
,
D. B.
,
O'Connor
,
P. J.
,
Lange
,
G.
, &
Steffener
,
J.
(
2007
).
Functional neuroimaging correlates of mental fatigue induced by cognition among chronic fatigue syndrome patients and controls.
Neuroimage
,
36
,
108
122
.
Corbetta
,
M.
, &
Shulman
,
G. L.
(
2002
).
Control of goal-directed and stimulus-driven attention in the brain.
Nature Reviews Neuroscience
,
3
,
201
215
.
Desmet
,
C.
,
Fias
,
W.
,
Hartstra
,
E.
, &
Brass
,
M.
(
2011
).
Errors and conflict at the task level and the response level.
Journal of Neuroscience
,
31
,
1366
1374
.
Dosenbach
,
N. U.
,
Fair
,
D. A.
,
Cohen
,
A. L.
,
Schlaggar
,
B. L.
, &
Petersen
,
S. E.
(
2008
).
A dual-networks architecture of top–down control.
Trends in Cognitive Sciences
,
12
,
99
105
.
Dosenbach
,
N. U.
,
Fair
,
D. A.
,
Miezin
,
F. M.
,
Cohen
,
A. L.
,
Wenger
,
K. K.
,
Dosenbach
,
R. A.
,
et al
(
2007
).
Distinct brain networks for adaptive and stable task control in humans.
Proceedings of the National Academy of Sciences, U.S.A.
,
104
,
11073
11078
.
Dosenbach
,
N. U.
,
Visscher
,
K. M.
,
Palmer
,
E. D.
,
Miezin
,
F. M.
,
Wenger
,
K. K.
,
Kang
,
H. C.
,
et al
(
2006
).
A core system for the implementation of task sets.
Neuron
,
50
,
799
812
.
Ettinger
,
U.
,
Ffytche
,
D. H.
,
Kumari
,
V.
,
Kathmann
,
N.
,
Reuter
,
B.
,
Zelaya
,
F.
,
et al
(
2008
).
Decomposing the neural correlates of antisaccade eye movements using event-related fMRI.
Cerebral Cortex
,
18
,
1148
1159
.
Fan
,
J.
,
McCandliss
,
B. D.
,
Fossella
,
J.
,
Flombaum
,
J. I.
, &
Posner
,
M. I.
(
2005
).
The activation of attentional networks.
Neuroimage
,
26
,
471
479
.
Forstmann
,
B. U.
,
van den Wildenberg
,
W. P.
, &
Ridderinkhof
,
K. R.
(
2008
).
Neural mechanisms, temporal dynamics, and individual differences in interference control.
Journal of Cognitive Neuroscience
,
20
,
1854
1865
.
Fox
,
M. D.
,
Snyder
,
A. Z.
,
Vincent
,
J. L.
,
Corbetta
,
M.
,
Van Essen
,
D. C.
, &
Raichle
,
M. E.
(
2005
).
The human brain is intrinsically organized into dynamic, anticorrelated functional networks.
Proceedings of the National Academy of Sciences, U.S.A.
,
102
,
9673
9678
.
Fox
,
M. D.
,
Zhang
,
D.
,
Snyder
,
A. Z.
, &
Raichle
,
M. E.
(
2009
).
The global signal and observed anticorrelated resting state brain networks.
Journal of Neurophysiology
,
101
,
3270
3283
.
Fransson
,
P.
(
2005
).
Spontaneous low-frequency BOLD signal fluctuations: An fMRI investigation of the resting-state default mode of brain function hypothesis.
Human Brain Mapping
,
26
,
15
29
.
Ghajar
,
J.
, &
Ivry
,
R. B.
(
2009
).
The predictive brain state: Asynchrony in disorders of attention?
Neuroscientist
,
15
,
232
242
.
Green
,
D. M.
, &
Swets
,
J. A.
(
1966
).
Signal detection theory and psychophysics.
New York
:
John Wiley & Sons Ltd.
Hartwigsen
,
G.
,
Bestmann
,
S.
,
Ward
,
N. S.
,
Woerbel
,
S.
,
Mastroeni
,
C.
,
Granert
,
O.
,
et al
(
2012
).
Left dorsal premotor cortex and supramarginal gyrus complement each other during rapid action reprogramming.
Journal of Neuroscience
,
32
,
16162
16171
.
Heyder
,
K.
,
Suchan
,
B.
, &
Daum
,
I.
(
2004
).
Cortico-subcortical contributions to executive control.
Acta Psychologica (Amsterdam)
,
115
,
271
289
.
Holland
,
D.
,
Kuperman
,
J. M.
, &
Dale
,
A. M.
(
2010
).
Efficient correction of inhomogeneous static magnetic field-induced distortion in echo planar imaging.
Neuroimage
,
50
,
175
183
.
Hu
,
S.
, &
Li
,
C. S.
(
2012
).
Neural processes of preparatory control for stop signal inhibition.
Human Brain Mapping
,
33
,
2785
2796
.
Jenkinson
,
M.
,
Bannister
,
P.
,
Brady
,
M.
, &
Smith
,
S.
(
2002
).
Improved optimization for the robust and accurate linear registration and motion correction of brain images.
Neuroimage
,
17
,
825
841
.
Jenkinson
,
M.
, &
Smith
,
S.
(
2001
).
A global optimisation method for robust affine registration of brain images.
Medical Image Analysis
,
5
,
143
156
.
Kelly
,
A. M.
,
Uddin
,
L. Q.
,
Biswal
,
B. B.
,
Castellanos
,
F. X.
, &
Milham
,
M. P.
(
2008
).
Competition between functional brain networks mediates behavioral variability.
Neuroimage
,
39
,
527
537
.
Kim
,
C.
,
Chung
,
C.
, &
Kim
,
J.
(
2012
).
Conflict adjustment through domain-specific multiple cognitive control mechanisms.
Brain Research
,
1444
,
55
64
.
Kim
,
C.
,
Johnson
,
N. F.
,
Cilles
,
S. E.
, &
Gold
,
B. T.
(
2011
).
Common and distinct mechanisms of cognitive flexibility in prefrontal cortex.
Journal of Neuroscience
,
31
,
4771
4779
.
Koechlin
,
E.
,
Ody
,
C.
, &
Kouneiher
,
F.
(
2003
).
The architecture of cognitive control in the human prefrontal cortex.
Science
,
302
,
1181
1185
.
Koechlin
,
E.
, &
Summerfield
,
C.
(
2007
).
An information theoretical approach to prefrontal executive function.
Trends in Cognitive Sciences
,
11
,
229
235
.
Langner
,
R.
,
Steinborn
,
M. B.
,
Chatterjee
,
A.
,
Sturm
,
W.
, &
Willmes
,
K.
(
2010
).
Mental fatigue and temporal preparation in simple reaction-time performance.
Acta Psychologica (Amsterdam)
,
133
,
64
72
.
Laufer
,
I.
,
Negishi
,
M.
,
Lacadie
,
C. M.
,
Papademetris
,
X.
, &
Constable
,
R. T.
(
2011
).
Dissociation between the activity of the right middle frontal gyrus and the middle temporal gyrus in processing semantic priming.
PLoS One
,
6
,
e22368
.
Lieberman
,
M. D.
, &
Cunningham
,
W. A.
(
2009
).
Type I and type II error concerns in fMRI research: Re-balancing the scale.
Social Cognitive & Affective Neuroscience
,
4
,
423
428
.
Lim
,
J.
,
Wu
,
W. C.
,
Wang
,
J.
,
Detre
,
J. A.
,
Dinges
,
D. F.
, &
Rao
,
H.
(
2010
).
Imaging brain fatigue from sustained mental workload: An ASL perfusion study of the time-on-task effect.
Neuroimage
,
49
,
3426
3435
.
Little
,
D. M.
,
Kraus
,
M. F.
,
Joseph
,
J.
,
Geary
,
E. K.
,
Susmaras
,
T.
,
Zhou
,
X. J.
,
et al
(
2010
).
Thalamic integrity underlies executive dysfunction in traumatic brain injury.
Neurology
,
74
,
558
564
.
Mathiak
,
K. A.
,
Klasen
,
M.
,
Weber
,
R.
,
Ackermann
,
H.
,
Shergill
,
S. S.
, &
Mathiak
,
K.
(
2011
).
Reward system and temporal pole contributions to affective evaluation during a first person shooter video game.
BMC Neuroscience
,
12
,
66
.
McGuire
,
P. K.
,
Paulesu
,
E.
,
Frackowiak
,
R. S.
, &
Frith
,
C. D.
(
1996
).
Brain activity during stimulus independent thought.
NeuroReport
,
7
,
2095
2099
.
McKiernan
,
K. A.
,
Kaufman
,
J. N.
,
Kucera-Thompson
,
J.
, &
Binder
,
J. R.
(
2003
).
A parametric manipulation of factors affecting task-induced deactivation in functional neuroimaging.
Journal of Cognitive Neuroscience
,
15
,
394
408
.
Menon
,
V.
, &
Uddin
,
L. Q.
(
2010
).
Saliency, switching, attention and control: A network model of insula function.
Brain Structure & Function
,
214
,
655
667
.
Nee
,
D. E.
,
Kastner
,
S.
, &
Brown
,
J. W.
(
2011
).
Functional heterogeneity of conflict, error, task-switching, and unexpectedness effects within medial prefrontal cortex.
Neuroimage
,
54
,
528
540
.
Nelson
,
S. M.
,
Dosenbach
,
N. U.
,
Cohen
,
A. L.
,
Wheeler
,
M. E.
,
Schlaggar
,
B. L.
, &
Petersen
,
S. E.
(
2010
).
Role of the anterior insula in task-level control and focal attention.
Brain Structure & Function
,
214
,
669
680
.
Ogg
,
R. J.
,
Zou
,
P.
,
Allen
,
D. N.
,
Hutchins
,
S. B.
,
Dutkiewicz
,
R. M.
, &
Mulhern
,
R. K.
(
2008
).
Neural correlates of a clinical continuous performance test.
Magnetic Resonance Imaging
,
26
,
504
512
.
Petersen
,
S. E.
, &
Dubis
,
J. W.
(
2012
).
The mixed block/event-related design.
Neuroimage
,
62
,
1177
1184
.
Posner
,
M. I.
, &
Rothbart
,
M. K.
(
2007
).
Research on attention networks as a model for the integration of psychological science.
Annual Review of Psychology
,
58
,
1
23
.
Rabin
,
L. A.
,
Barr
,
W. B.
, &
Burton
,
L. A.
(
2005
).
Assessment practices of clinical neuropsychologists in the United States and Canada: A survey of INS, NAN, and APA division 40 members.
Archives of Clinical Neuropsychology
,
20
,
33
65
.
Raz
,
A.
, &
Buhle
,
J.
(
2006
).
Typologies of attentional networks.
Nature Reviews Neuroscience
,
7
,
367
379
.
Riccio
,
C. A.
, &
Reynolds
,
C. R.
(
2001
).
Continuous performance tests are sensitive to ADHD in adults but lack specificity. A review and critique for differential diagnosis.
Annals of the New York Academy of Sciences
,
931
,
113
139
.
Riccio
,
C. A.
,
Reynolds
,
C. R.
,
Lowe
,
P.
, &
Moore
,
J. J.
(
2002
).
The continuous performance test: A window on the neural substrates for attention?
Archives of Clinical Neuropsychology
,
17
,
235
272
.
Rotshtein
,
P.
,
Soto
,
D.
,
Grecucci
,
A.
,
Geng
,
J. J.
, &
Humphreys
,
G. W.
(
2011
).
The role of the pulvinar in resolving competition between memory and visual selection: A functional connectivity study.
Neuropsychologia
,
49
,
1544
1552
.
Rypma
,
B.
,
Prabhakaran
,
V.
,
Desmond
,
J. E.
,
Glover
,
G. H.
, &
Gabrieli
,
J. D.
(
1999
).
Load-dependent roles of frontal brain regions in the maintenance of working memory.
Neuroimage
,
9
,
216
226
.
Seeley
,
W. W.
,
Menon
,
V.
,
Schatzberg
,
A. F.
,
Keller
,
J.
,
Glover
,
G. H.
,
Kenna
,
H.
,
et al
(
2007
).
Dissociable intrinsic connectivity networks for salience processing and executive control.
Journal of Neuroscience
,
27
,
2349
2356
.
Seger
,
C. A.
(
2008
).
How do the basal ganglia contribute to categorization? Their roles in generalization, response selection, and learning via feedback.
Neuroscience & Biobehavioral Reviews
,
32
,
265
278
.
Smith
,
S. M.
(
2002
).
Fast robust automated brain extraction.
Human Brain Mapping
,
17
,
143
155
.
Stoeckel
,
C.
,
Gough
,
P. M.
,
Watkins
,
K. E.
, &
Devlin
,
J. T.
(
2009
).
Supramarginal gyrus involvement in visual word recognition.
Cortex
,
45
,
1091
1096
.
Tana
,
M. G.
,
Montin
,
E.
,
Cerutti
,
S.
, &
Bianchi
,
A. M.
(
2010
).
Exploring cortical attentional system by using fMRI during a Continuous Performance Test.
Computational Intelligence and Neuroscience
,
329213
.
Taren
,
A. A.
,
Venkatraman
,
V.
, &
Huettel
,
S. A.
(
2011
).
A parallel functional topography between medial and lateral prefrontal cortex: Evidence and implications for cognitive control.
Journal of Neuroscience
,
31
,
5026
5031
.
Taylor
,
P. C.
,
Nobre
,
A. C.
, &
Rushworth
,
M. F.
(
2007
).
Subsecond changes in top down control exerted by human medial frontal cortex during conflict and action selection: A combined transcranial magnetic stimulation electroencephalography study.
Journal of Neuroscience
,
27
,
11343
11353
.
Thyreau
,
B.
,
Schwartz
,
Y.
,
Thirion
,
B.
,
Frouin
,
V.
,
Loth
,
E.
,
Vollstädt-Klein
,
S.
,
et al
(
2012
).
Very large fMRI study using the IMAGEN database: Sensitivity–specificity and population effect modeling in relation to the underlying anatomy.
Neuroimage
,
61
,
295
303
.
Van der Werf
,
Y. D.
,
Witter
,
M. P.
,
Uylings
,
H. B.
, &
Jolles
,
J.
(
2000
).
Neuropsychology of infarctions in the thalamus: A review.
Neuropsychologia
,
38
,
613
627
.
Venkatraman
,
V.
,
Rosati
,
A. G.
,
Taren
,
A. A.
, &
Huettel
,
S. A.
(
2009
).
Resolving response, decision, and strategic control: Evidence for a functional topography in dorsomedial prefrontal cortex.
Journal of Neuroscience
,
29
,
13158
13164
.
Visser
,
M.
,
Jefferies
,
E.
,
Embleton
,
K. V.
, &
Lambon Ralph
,
M. A.
(
2012
).
Both the middle temporal gyrus and the ventral anterior temporal area are crucial for multimodal semantic processing: Distortion-corrected fMRI evidence for a double gradient of information convergence in the temporal lobes.
Journal of Cognitive Neuroscience
,
24
,
1766
1778
.
Weissman
,
D. H.
,
Roberts
,
K. C.
,
Visscher
,
K. M.
, &
Woldorff
,
M. G.
(
2006
).
The neural bases of momentary lapses in attention.
Nature Neuroscience
,
9
,
971
978
.
Wilk
,
H. A.
,
Ezekiel
,
F.
, &
Morton
,
J. B.
(
2012
).
Brain regions associated with moment-to-moment adjustments in control and stable task-set maintenance.
Neuroimage
,
59
,
1960
1967
.