Abstract

We investigated the functional properties of a previously described cingulo-opercular network (CON) putatively involved in cognitive control. Analyses of common fMRI task-evoked activity during perceptual and episodic memory search tasks that differently recruited the dorsal attention (DAN) and default mode network (DMN) established the generality of this network. Regions within the CON (anterior insula/frontal operculum and anterior cingulate/presupplementary cortex) displayed sustained signals during extended periods in which participants searched for behaviorally relevant information in a dynamically changing environment or from episodic memory in the absence of sensory stimulation. The CON was activated during all phases of both tasks, which involved trial initiation, target detection, decision, and response, indicating its consistent involvement in a broad range of cognitive processes. Functional connectivity analyses showed that the CON flexibly linked with the DAN or DMN regions during perceptual or memory search, respectively. Aside from the CON, only a limited number of regions, including the lateral pFC, showed evidence of domain-general sustained activity, although in some cases the common activations may have reflected the functional-anatomical variability of domain-specific regions rather than a true domain generality. These additional regions also showed task-dependent functional connectivity with the DMN and DAN, suggesting that this feature is not a specific marker of cognitive control. Finally, multivariate clustering analyses separated the CON from other frontoparietal regions previously associated with cognitive control, indicating a unique fingerprint. We conclude that the CON's functional properties and interactions with other brain regions support a broad role in cognition, consistent with its characterization as a task control network.

INTRODUCTION

Much research is concerned with identifying the neural mechanisms of cognitive control (Niendam et al., 2012; Dosenbach, Fair, Cohen, Schlaggar, & Petersen, 2008; Aron, Behrens, Smith, Frank, & Poldrack, 2007; Cole & Schneider, 2007). These mechanisms are thought to maintain task control signals (Miller & Cohen, 2001) that specify how stimuli are mapped into responses and that flexibly select and configure the elemental cognitive processes necessary to perform a task (Rogers & Monsell, 1995). A previous meta-analysis of task-evoked activity across multiple mixed block/event experiments identified a cingulo-opercular network (CON) that showed three properties consistent with a role in implementing a task set (Dosenbach et al., 2006): (i) significant start cue activity at the beginning of a task block (task set instantiation), (ii) positive sustained activity across the block (maintenance), and (iii) error-related feedback activity (adjustment). Further work expanded the CON to include the anterior pFC and the thalamus (Dosenbach et al., 2007) and showed that the network was partially segregated from a second, frontoparietal network that has also been associated with cognitive control (Dosenbach et al., 2008).

If the CON serves the broad role attributed by previous studies, then it should show sustained activity during the performance of different tasks that elicit very different spatial patterns of cortical activity. Here we provide a strong within-experiment test of this prediction by studying the response profile of the CON during two tasks that involved either attending to environment stimuli (i.e., perceptual search) or the retrieval of behaviorally relevant information from episodic memory (i.e., memory search). These tasks recruited, respectively, a dorsal attention network (DAN; Corbetta & Shulman, 2002; Kastner & Ungerleider, 2000) and a default mode network (DMN; Buckner, Andrews-Hanna, & Schacter, 2008; Raichle et al., 2001; Shulman et al., 1997) that operated in a push–pull dynamic competition (Sestieri, Shulman, & Corbetta, 2010), paralleling their negative correlation in the resting state (Fox et al., 2005). By hypothesis, the CON should show positive sustained activity during the performance of each task, regardless of the pattern of activation/deactivation in domain-specific networks.

Previous studies have proposed that a larger set of frontoparietal regions (Vincent, Kahn, Snyder, Raichle, & Buckner, 2008; Cole & Schneider, 2007) is activated during both the execution of externally and internally oriented tasks (Gao & Lin, 2012; Smallwood, Brown, Baird, & Schooler, 2012; Spreng, Stevens, Chamberlain, Gilmore, & Schacter, 2010). For example, Spreng and colleagues (Spreng, Sepulcre, Turner, Stevens, & Schacter, 2013; Spreng et al., 2010) found that an extensive set of regions in lateral and medial frontoparietal cortex, including the CON, showed common activity during two planning tasks that activated the DMN and DAN, respectively. The relatively large extent of common activity, which included large portions of the occipital cortex, may have partly reflected the fact that both the DMN- and DAN-oriented tasks involved cognitive processes related to planning and visual sensory input. Here, by testing DMN- and DAN-oriented tasks that involved very different cognitive processes, we determined whether the set of regions commonly activated during the two types of tasks was more restricted and yet still included the CON. In line with this goal, our experimental design allowed us to separate different phases within the perceptual and memory tasks. This design feature enabled us to (i) test for common activity during those parts of the two tasks that isolated very different processes and (ii) assess the generality of the CON across very different cognitive processes within each task. For example, we assessed the presence of sustained brain activity when participants searched for an object in a dynamically changing display, before target detection and response execution, versus when participants retrieved information from episodic memory over an extended period of time while fixating a blank screen, again before reaching a final decision and making a response. In addition, whereas the original meta-analysis of Dosenbach et al. (2006) examined whether brain activity was sustained over a block to show that task set signals were present also during the intertrial interval as well as during the trial, here we examined activity that was sustained within a prolonged phase of individual trials, that is, an extended period of task performance. Therefore, although the present tasks do not represent all the different elements of cognitive control, as in large meta-analytic approaches (Dosenbach et al., 2006), their features enabled a strong and novel test of the domain generality of activity within the CON.

A second important property of a “domain-general” network is the degree to which the network selectively interacts with domain-specific, task-relevant networks. Prior studies that focused on whether the CON dynamically linked with domain-specific networks did not examine both externally and internally oriented tasks in a single experimental design (Chiong et al., 2013; Higo, Mars, Boorman, Buch, & Rushworth, 2011; Zanto, Rubens, Thangavel, & Gazzaley, 2011; Sridharan, Levitin, & Menon, 2008). In contrast, those studies that have investigated the presence of modulations of functional/effective connectivity with the DMN and the DAN during the execution of externally and internally oriented tasks (Gao & Lin, 2012; Smallwood et al., 2012; Spreng et al., 2010) have not explicitly focused on the CON, making it difficult to appreciate its particular relationship with domain-specific networks. Here we specifically measured the functional connectivity of the CON with the DMN and DAN as those networks were activated and deactivated during memory and perceptual search.

Therefore, in this study, we first determined whether regions that exhibited sustained within-trial activity during both perceptual and memory search tasks included the CON (dorsal anterior cingulate/presupplementary cortex [dACC/pre-SMA] and anterior insula/frontal operculum [aINS/fO], as defined in Dosenbach et al., 2006). We then characterized the full profile of task-evoked activity in the CON during the different phases of each task. Next, we analyzed the pattern of task-evoked and resting functional connectivity to investigate whether the CON flexibly coupled with domain-specific regions of the DMN and DAN according to task demands (Spreng et al., 2010; Sridharan et al., 2008). Finally, we examined whether the profile of task-evoked activity and functional connectivity distinguished the CON from other cortical regions that showed any evidence for domain-general sustained activity.

METHODS

This article is based on a new analysis of a previously published experiment (Sestieri, Corbetta, Romani, & Shulman, 2011; Sestieri et al., 2010). Because stimuli, tasks, procedures, and linear modeling of task-evoked activity have been extensively described in these publications, here we present a briefer description.

Participants

Nineteen healthy right-handed participants (mean age = 26.8 years, range 23–32 years; three men) gave informed consent in accordance with guidelines set by the Human Studies Committee of Washington University in St. Louis, MO. Each participant performed a perceptual search and an episodic memory search task involving audiovisual material on different days, counterbalanced across participants. Participants were also scanned at rest for ∼30 min (rs-FC scans).

Procedure

The experimental paradigm is illustrated in Figure 1A, B.

Figure 1. 

(A) Trial structure in the perceptual search task. A sentence instructed participants to search for a specific target (object or character) that could appear at any time in the upcoming 12-sec video clip. Participants searched for the target while fixating a central cross and pressed a button as soon as the target was detected, depending on their confidence (high, low). Search duration was varied (early, middle, late) by manipulating the time at which the target was presented. After display offset, a variable ITI was interposed before the onset of the next sentence. (B) Trial structure in the episodic memory search task. Participants read a sentence describing a specific detail of a previously encoded episode from a TV show. They then retrieved information from episodic memory to judge the accuracy (i.e., true, false) of the sentence, which they indicated by pressing one of four buttons, depending on their confidence (high, low). Participants were given up to 15 sec to provide the judgment on each trial. An example of early, middle, and late search trials are provided. After participants' response, a variable ITI was interposed before the onset of the next sentence. (C) Mean accuracy for the memory and the perceptual (oddball target excluded) task. Vertical bars indicate SEM. (D) Mean RTs from target onset in the perceptual task, divided into early (1.678 sec), middle (1.691 sec), and late (1.615 sec) interval. Vertical bars indicate SEM. (E) Trial distribution (mean across participants) for each time bin (1 sec) of the allowed RT (15 sec). The graph shows the distribution of all trials (solid black), correct trials (solid gray), and incorrect (broken gray) trials. The asterisks indicate the time bins in which performance was significantly different from chance (one-sample t test against the chance level of 0.5). Different shades of gray on the graph illustrates the subdivision into early (E = 0–4 sec), middle (M = 4–8 sec) and late (L = 8–12 sec) trials for time course analysis. Trials in the 12- to 15-sec interval were discarded because of poor performance.

Figure 1. 

(A) Trial structure in the perceptual search task. A sentence instructed participants to search for a specific target (object or character) that could appear at any time in the upcoming 12-sec video clip. Participants searched for the target while fixating a central cross and pressed a button as soon as the target was detected, depending on their confidence (high, low). Search duration was varied (early, middle, late) by manipulating the time at which the target was presented. After display offset, a variable ITI was interposed before the onset of the next sentence. (B) Trial structure in the episodic memory search task. Participants read a sentence describing a specific detail of a previously encoded episode from a TV show. They then retrieved information from episodic memory to judge the accuracy (i.e., true, false) of the sentence, which they indicated by pressing one of four buttons, depending on their confidence (high, low). Participants were given up to 15 sec to provide the judgment on each trial. An example of early, middle, and late search trials are provided. After participants' response, a variable ITI was interposed before the onset of the next sentence. (C) Mean accuracy for the memory and the perceptual (oddball target excluded) task. Vertical bars indicate SEM. (D) Mean RTs from target onset in the perceptual task, divided into early (1.678 sec), middle (1.691 sec), and late (1.615 sec) interval. Vertical bars indicate SEM. (E) Trial distribution (mean across participants) for each time bin (1 sec) of the allowed RT (15 sec). The graph shows the distribution of all trials (solid black), correct trials (solid gray), and incorrect (broken gray) trials. The asterisks indicate the time bins in which performance was significantly different from chance (one-sample t test against the chance level of 0.5). Different shades of gray on the graph illustrates the subdivision into early (E = 0–4 sec), middle (M = 4–8 sec) and late (L = 8–12 sec) trials for time course analysis. Trials in the 12- to 15-sec interval were discarded because of poor performance.

Perceptual Search Task

In the perceptual task, visually presented sentences instructed participants to search for a visual target that could be presented at any time during an upcoming 12-sec video clip, while maintaining central fixation. Search time was manipulated by varying the time of onset of the target to separate the neural signals associated with searching for the target (search), audiovisual constant stimulation (display), and target detection/motor response (detection; Shulman et al., 2003). On each trial, a sentence was presented on the screen for 4 sec instructing participants to search for a specific target in the next 12-sec video clip. The clips had a superimposed central fixation cross and were followed by a variable intertrial interval (ITI; ∼4.1/6.2/8.3 sec). Three kinds of clips were presented: clips with a target (“target”), clips with an oddball target (“oddball”), and clips with no target (“nontarget”). Trials were grouped according to target onset and the corresponding search time: early (0–4 sec after sentence offset), middle (4–8 sec), and late (8–12 sec) trials. When either the target or the oddball target was detected, participants had to press as quickly as possible one of the two “Yes” keys with their left hand, rating their confidence (high, low). If a target was not detected, they had to press one of the two “No” buttons with their right hand, depending on confidence, at the end of the clip. Responses were categorized in hits, misses, false alarms (FA) and correct rejections (CR) according to signal detection theory (Green & Swets, 1966). Participants knew that targets could appear only once during the clip, and they were instructed to passively watch the end of the clip after a target had been detected. “Catch” trials, in which the sentence was immediately followed by the ITI, were used to separate BOLD activity associated with sentence reading from subsequent phases of the trial (Ollinger, Corbetta, & Shulman, 2001; Ollinger, Shulman, & Corbetta, 2001; Shulman et al., 1999). Ten runs, each containing 25 trials, were administered.

Memory Search Task

In the memory task, participants judged the accuracy of sentences describing specific details about one of two movies that had been encoded while maintaining central fixation. In this task, the source of search duration variability was the participant's decision time, reflecting the time needed to retrieve the critical information and answer the question. This cued recollection task likely involves elaborative recall processes of imagery and scene reconstruction, in addition to a general sense of familiarity (Mendelsohn, Furman, & Dudai, 2010). Therefore, the term memory search is used here as a proxy for a broad range of processes that involve the retrieval of episodic information guided by the behavioral goal, the organization and evaluation of the retrieved information, and the accumulation of evidence that pointed to a particular response. At encoding, participants watched two episodes from an English language television sitcom (Curb Your Enthusiasm, by Larry David; HBO Network), separated by an hour break. The fMRI retrieval session was performed ∼24 hr later. On each trial, a sentence describing memory for details and events across the two episodes was presented for 4 sec, followed by a black display with a white central fixation cross. Participants were instructed to read the sentence, wait until it disappeared, and then take the time they needed, up to 15 sec, to retrieve the specific information and provide a yes/no judgment about the accuracy of the sentence (named detection for consistency with the perceptual task), with confidence rating (high, low). Judgments were made using four buttons with the same category-key mapping used for the perceptual task. Trials were grouped into early (0–4 sec after sentence offset), middle (4–8 sec), and late (8–12 sec), representing different search times (trials between 12 and 15 sec were discarded because of low accuracy). Following participant's response, the fixation cross turned red, indicating the onset of a variable ITI (4.1, 6.2, 8.3 sec). During catch trials, sentences were immediately followed by the ITI. Participants were asked not to retrieve information following catch trial sentences. Five runs of 20 trials pertaining to the episode encoded first were presented, followed by five runs pertaining to the episode encoded second.

Imaging Methods and Preprocessing of BOLD Images

Images were acquired with a Siemens (Erlangen, Germany) Allegra 3T scanner. Structural images were obtained during the first scanning session using a sagittal MP-RAGE T1-weighted sequence (repetition time [TR] = 1810 msec, echo time [TE] = 3.93 msec, flip angle = 12°, time for inversion = 1200 msec, voxel size = 1 × 1 × 1.25 mm) and a T2-weighted spin-echo sequence (TR = 3800 msec, TE = 90 msec, flip angle = 90°). BOLD contrast functional images were acquired using a gradient-echo echo-planar sequence (TR = 2064 msec, TE = 25 msec, flip angle = 90°, 32 contiguous 4 mm axial slices, 4 × 4 mm in-plane resolution).

Analysis of Task-evoked Activity

Linear Modeling

Data were analyzed using two kinds of general linear models (GLMs; Sestieri et al., 2010, 2011). The aim of the first model (process-GLM) was to separately estimate the BOLD signal for the different task processes that temporally overlapped in the course of a trial: sentence reading (sentence), sensory stimulation (display), perceptual search (search), target detection, and response (detection) in the perceptual task; sentence reading (sentence), memory search (search), and response (detection) in the memory task. A multiple parameter regression model was created that specified the effects of the several task processes on the observed BOLD response. The model assumed that the BOLD response on each trial was the sum of the hemodynamic responses that were generated by the above processes and was used to identify voxels activated or deactivated by each process. The assumed response shape for each process was generated by convolving a function representing the duration of the process (rectangle functions for sustained processes, delta functions for transient processes) with a standard hemodynamic response function (Boynton, Engel, Glover, & Heeger, 1996). The sentence-reading phase was separated using a catch-trial technique, in which, on a random 20% of the trials, the trial ended following the sentence-reading phase. For the perceptual task, separate parameters modeled the search and the detection components of the task according to participant's accuracy (Hit, Miss, FA, CR) and confidence (High, Low). Similarly, for the memory task, separate parameters modeled the search and the detection component of the task according to participant's accuracy (Correct, Incorrect) and confidence (High, Low).

To examine the overall time course of BOLD activity for different types of trials (e.g., early, middle, and late correct response trials), which reflects the sum of the BOLD signal for the processes operative on those trials, a second type of GLM was created that made no assumption about the shape of the hemodynamic response (frame-by-frame GLM). This model provided an unbiased estimate of the time course for each trial type (Ollinger, Corbetta, et al., 2001; Ollinger, Shulman, et al., 2001), generating separate delta function regressors for each MR frame up to ∼30 sec after trial onset. Time courses for 13 types of trials, all starting at sentence onset, were estimated for the perceptual task: sentence reading catch trials, three Hit-target, three Hit-oddball, and three Miss types of trials, depending on the interval of target presentation (early, middle, late), and CR, FA, and trials in which participants did not press any key. Time courses for eight types of trials, all starting at sentence onset, were estimated for the memory task: sentence reading catch trials, three correct response trials (early, middle, late), three incorrect response trials (early, middle, late), and trials in which participants did not press a key. One participant was removed from the time course analysis for the lack of early correct trials in the memory experiment.

Statistical Analyses

Although the two types of GLMs included regressors for correct and incorrect trials, all the statistical analyses have been conducted on correct trials only. Using the process-GLM, we created group voxelwise statistical maps corresponding to each process in which participant was treated as a random effect. We used a standard method to correct for multiple comparisons based on a region size/z-score criterion combination, determined by Monte Carlo simulations. We used the same combination (17 face contiguous voxels/z = 3) used in our previous publications (Sestieri et al., 2010, 2011), which corresponds to a probability of falsely rejecting the null hypothesis of p < .05. To focus on optimal task performance, voxelwise maps of perceptual and memory search-related activity were obtained using high-confidence correct trials. Regional one-sample t tests were performed to test whether regions activated by sustained parameters also responded to other parameters in either task. Using the process-GLM, we assessed the presence of significant activity for two additional parameters (sentence reading, detection) of the perceptual and the memory tasks. This approach allowed us to establish the full pattern of task-evoked activity for each ROI.

Definition of Domain-general ROIs

ROIs showing sustained activity for both tasks were formed using a procedure to maximize consistency across participants. Whereas the memory task had a single sustained process (search), the sustained processes of the perceptual task included the display and the search parameters, which were averaged to create a voxelwise map (z = 3, corrected). The display parameter was included because sustained signals in higher-level regions might be maintained after the participant's response while the meaningful audio-visual display continued. Therefore, including both the search and display parameters increased our ability to map sustained signals during the perception task.

Next, we formed a binary AND map between the memory and the perceptual maps to identify voxels positively activated by sustained process of both tasks at the group level. The same procedure was repeated for each participant to generate individual binary AND maps. These images were then summed to obtain a frequency map in which the value of each voxel represented the number of participants showing sustained activity in both tasks. The frequency map and the group AND map were multiplied and the resulting image was used to form ROIs (6 mm radius, peaks separated by at least 12 mm). ROIs with less than five voxels were excluded. This method guaranteed that ROIs were centered on those voxels where domain-general sustained activity was most consistent across participants and was present at the group level.

Definition of Domain-specific ROIs

We also defined two sets of domain-specific ROIs showing sustained activity during the memory and the perceptual task, respectively, to examine whether the pattern of connectivity between networks was modulated by task execution. The memory set was extracted from the voxelwise map corresponding to the memory search parameter, masked to exclude eventual voxels showing domain-general activity. Four ROIs (6 mm radius, peaks separated by at least 12 mm) were selected, based on their proximity to posterior nodes of the DMN (bilateral angular gyrus, bilateral posterior cingulate cortex/precuneus; Sestieri et al., 2011). The perceptual set of ROIs was extracted from the voxelwise map corresponding to the average voxelwise map of the search and display parameters, masked to exclude potential domain-general activity. Four ROIs were selected topographically corresponding to key frontoparietal regions of the DAN (bilateral FEF, bilateral posterior intraparietal sulcus; He et al., 2007; Corbetta & Shulman, 2002).

Functional Connectivity Analysis

Resting State and Task-induced Functional Connectivity

We conducted six runs (∼5 min each) in which the BOLD signal was measured while participants maintained fixation on a central cross in an otherwise blank display. After standard preprocessing of BOLD images, data were passed through an additional series of specific processing steps for rs-FC (Vincent et al., 2006; Fox et al., 2005). First, the runs were concatenated. For each voxel, temporal filtering retained frequencies < 0.1 Hz and data were spatially smoothed using a 6-mm FWHM Gaussian blur. Several sources of spurious or regionally nonspecific variance were removed by linear regression including six parameters obtained by rigid body head motion correction, the signal averaged over the whole brain, the signal averaged over the lateral ventricles, and the signal averaged over a region centered in the deep cerebral white matter. The analysis was also conducted without whole brain signal regression to test the stability of the results across different methods.

Functional connectivity was also assessed during the execution of the perceptual and the memory search tasks. The mean task-evoked response was removed by linear regression, adding a further set of regressors corresponding to the design matrix of the frame-by-frame GLM to the list of regressors for rs-FC preprocessing. We minimized the contribution of evoked responses associated with task structure, motor responses, and perceptual stimulation by using a GLM that made no assumptions about the shape of the HRF (frame-by-frame GLM). This procedure resulted in two task-induced functional connectivity datasets (Norman-Haignere, McCarthy, Chun, & Turk-Browne, 2011) for investigating potential changes in connectivity as a function of the task set. Compared with previously developed approaches for the analysis of task-related modulation of FC (i.e., based on regional betas, Chadick & Gazzaley, 2011, or partial least squares analyses, Spreng et al., 2010), the present approach was relatively conservative and aimed at emphasizing the functional coupling associated with the maintenance of the task set, rather than the pattern of regional coactivation induced by transient evoked activity.

Regional FC

The connectivity between each of the three regions of the CON and each domain specific ROI (DMN, DAN) at rest and during the two task conditions was assessed in each participant. Significant changes of connectivity across tasks and networks were assessed by means of two-way repeated-measures ANOVA for each CON region, in which the dependent variable was the average Fisher z-transformed correlation value between the CON ROI and the four domain-specific regions of each network. Post hoc analyses were performed using Duncan tests. The same procedure was used to compute measures of connectivity between domain-specific ROIs and each of the other domain-general ROIs identified in this study. As a control analysis to test the extent to which modulations of FC were driven by the pattern of task coactivation, a within-network analysis was computed, averaging the correlation values of ROIs from the same network (CON, DMN, DAN). Significant changes of connectivity across tasks and networks were assessed by a two-way repeated-measures ANOVA with Task and Network as factors and post hoc analyses.

Voxelwise rs-FC Maps

Voxelwise connectivity maps were generated in each participant using domain-general ROIs as seeds. The BOLD time series from the rs-FC session was averaged over all voxels in the ROI, the voxelwise Pearson correlation coefficients between the seed time course and all other voxels were computed, and the Fisher z-transform was applied. For the group statistical analysis, a one-sample t test with participant as a random effect was computed on the Fisher z-transformed values, and the resulting group maps were Monte Carlo corrected over the brain for multiple comparisons (voxel size, 3 × 3 × 3 mm; cluster size, 17 voxels; z = 3, corresponding to p < .05).

Hierarchical Clustering

A multivariate hierarchical clustering procedure was used to determine whether the 22 domain-general regions identified in this study could be divided into different groups according to their functional properties (see Andrews-Hanna, Reidler, Sepulcre, Poulin, & Buckner, 2010; Ploran et al., 2007, for a similar approach). This analysis was conducted on the following measures, averaged across participants: (1) the pattern of task-evoked activation for each task parameter using the process-GLM; (2) the time courses of task-evoked activity for each trial type using the frame-by-frame GLM; (3) the voxelwise pattern of rs-FC using a seed-based approach; (4) the regional pattern of connectivity with domain specific ROIs at rest and during task execution. Matrices were created for each measure: a 22 (ROI) × 7 (BOLD % signal change for 7 parameters) for (i), a 22 (ROI) × 108 (BOLD % signal change for 18 time points by 3 conditions by 2 tasks) for (ii), a 22 (ROI) × 65,523 (voxels in the brain mask) for (iii), a 22 (ROI) × 24 (4 domain-specific ROIs × 2 networks × 3 tasks). Using the correlation coefficient between pairs of regions, a weighted dissimilarity matrix was calculated combining the matrices of the individual measures and assigning equal value (1/4) to each measure to balance their weight. An agglomerative hierarchical cluster tree was created from the distances in the matrix. The threshold for the pruning of the cluster tree was calculated maximizing cluster size and two measures of clustering validity (searching for the local maximum value of the Dunn index and silhouette coefficient and for the local minimum value of the Davies–Bouldin index).

The intersubject consistency of the clustering approach was tested by measuring the reproducibility across participants of the dissimilarity matrix used for the hierarchical clustering. A dissimilarity matrix was created for a subgroups of individuals (n = 9) on the basis of single-subject parameters. One participant did not contribute to the analysis because of a lack of time courses relative to early correct trials in the memory experiment (see above). The procedure was repeated for different subgroup permutations (n = 3000) to obtain a distribution of Pearson r correlation values between any possible pair of dissimilarity matrices. In addition, the clustering analysis was performed using only a subset (n = 3) of the four measures used in the original analysis to test the consistency of the results across measures.

RESULTS

A brief summary of the behavioral performance is presented in Figure 1CE. A detailed assessment of behavioral performance has been reported in previous publications (Sestieri et al., 2010, 2011).

Domain-general Sustained Activity in the CON during Memory and Perceptual Search

To test whether the CON exhibited domain-general activity, we first identified regions showing sustained activity in both tasks (Figure 2), using a procedure that searched for the presence of an overlap at the group and at the individual level (see Methods). Overall, a large segregation between voxels showing sustained activity for the memory and the perceptual task was observed, with regions showing domain-specific activity located adjacent to one another in large portions of cortex (Figure 2A). This result extends previous findings of a topographical segregation in the parietal lobe (Sestieri et al., 2010) also to the organization of frontal and temporal lobes. Domain-general activity was sparse and was sometimes located at the boundary between extended, unambiguous regions of domain-specific activity. For example, the region in right IPL was bordered superiorly by a large extent of memory task activity and inferiorly by a large extent of perception task activity. One might argue that it would be efficient for domain-general regions to be interposed between domain-specific regions, as noted by a reviewer. But this topography also raises the possibility that the common activation in this and other areas with a similar topography resulted from variability in the functional-anatomy of domain-specific regions, spatial smoothing, and a lack of spatial resolution. Therefore, although these regions may in fact be domain-general, some caution is warranted.

Figure 2. 

(A) Voxelwise map of sustained activity during the memory task (red, multiple-comparison corrected group z map of the memory search process) and the perceptual task (green, multiple-comparison corrected group z map of the average of search and display processes). Voxels in yellow represent regions of overlap at the group level. The map is superimposed on the lateral and medial view of the bilateral inflated representation of the PALS Atlas (Caret 5.5 software; Van Essen, 2005, p. 56). Black and white circles have also been superimposed to indicate the nodes location of the CON and frontoparietal network (FPCN), respectively, based on the coordinates reported by Dosenbach and colleagues (2006, 2007, 2008). The node corresponding to the dorsal ACC/pre-SMA has been represented in both hemispheres because of its proximity to the midline (z = −1). Note the good spatial consistency between the three key regions of the CON (bilateral aINS/fO, dACC/pre-SMA) and regions showing sustained domain activity in this study. Domain-general activity was not observed in proximity of the additional CON regions (anterior pFC and thalamus) described by Dosenbach and colleagues (2007, 2008) and regions of the FPCN. (B) Map representing the number of individual participants showing overlap of sustained activity across tasks for each domain-general voxel identified in A. Individual maps were created with the same procedure of the group analysis.

Figure 2. 

(A) Voxelwise map of sustained activity during the memory task (red, multiple-comparison corrected group z map of the memory search process) and the perceptual task (green, multiple-comparison corrected group z map of the average of search and display processes). Voxels in yellow represent regions of overlap at the group level. The map is superimposed on the lateral and medial view of the bilateral inflated representation of the PALS Atlas (Caret 5.5 software; Van Essen, 2005, p. 56). Black and white circles have also been superimposed to indicate the nodes location of the CON and frontoparietal network (FPCN), respectively, based on the coordinates reported by Dosenbach and colleagues (2006, 2007, 2008). The node corresponding to the dorsal ACC/pre-SMA has been represented in both hemispheres because of its proximity to the midline (z = −1). Note the good spatial consistency between the three key regions of the CON (bilateral aINS/fO, dACC/pre-SMA) and regions showing sustained domain activity in this study. Domain-general activity was not observed in proximity of the additional CON regions (anterior pFC and thalamus) described by Dosenbach and colleagues (2007, 2008) and regions of the FPCN. (B) Map representing the number of individual participants showing overlap of sustained activity across tasks for each domain-general voxel identified in A. Individual maps were created with the same procedure of the group analysis.

A notable exception to this pattern, however, was observed in three regions that are anatomically consistent with the three main components of the CON: left aINS/fO (x: −29, y: +17, z: +2), right dACC/pre-SMA (x: +3, y: +6, z: +50), and right aINS/fO (x: +31, y: +15, z: +4). The peak of consistency (Figure 2B) of these regions was similar to the coordinates reported in previous studies (left aINS/fO = x: −35, y: +14, z: +5; dACC/pre-SMA = x: −1, y: +10, z: +46; right aINS/fO = x: +36, y: +16, z: +4; indicated by black circles in the figure; Dosenbach et al., 2006, 2008). Importantly, domain-general activity in these regions was not sandwiched between two larger swaths of domain-specific activity. For example, the majority of voxels activated in the left and right aINS showed domain-general rather than domain-specific activity. In the anterior cingulate, the region showing common activity was bordered anteriorly by a region showing memory task activity but was not bordered by any region showing only perception task activity. Consistent with the sustained activation of the CON during the search phase of the visual search task, similar regions had shown the same pattern of within-trial, sustained activity in a previous experiment on visual search (Shulman et al., 2003; cf. Figure 2) that involved a comparable analysis procedure. Domain-general activity was not found in regions (anterior pFC, thalamus) that showed sustained activity only for a subset of the tasks in the meta-analyses performed by Dosenbach et al. (2006) and that have been included in more recent formulations of the CON (Dosenbach et al., 2007, 2008).

Other regions showing domain-general activity included the right retrosplenial cortex (RSC), left dorsal precuneus, and left pFC. The complete list of domain-general clusters, sorted by consistency across participants, is provided in Table 1. The domain-general activity found in multiple left prefrontal locations is consistent with previous reports of task-independent, cognitive control signals in these regions (reviewed in Sakai & Passingham, 2003; Duncan & Owen, 2000), whereas domain-general activity in the dorsal precuneus has been observed in the context of task-switching paradigms (Greenberg, Esterman, Wilson, Serences, & Yantis, 2010; Chiu & Yantis, 2009). However, partly because this article is focused on the CON and partly because of the caution related to the “sandwich” topography noted above, we do not make strong claims regarding the domain generality of regions outside the CON. At the same time, however, because these regions exhausted the possible set of domain-general regions that showed sustained activations during the search phase of the perception and memory paradigms (i.e., sustained domain-general activity was not present outside these regions), they provided a useful control for evaluating the degree to which the domain-general properties of the CON were unique.

Table 1. 

Domain-general ROIs

No.
Side
Region
x
y
z
Voxels
RSC +18 −56 +21 24 
aINS/fO −29 +17 +02 26 
dACC/pre-SMA +03 +06 +50 15 
aINS/fO +31 +15 +04 23 
SFS −26 −03 +57 27 
RSC −17 −59 +17 11 
dPreCu −01 −59 +48 11 
AG −41 −68 +18 14 
RSC +13 −50 +08 17 
10 paraHC +23 −39 −10 11 
11 paraHC −24 −41 −11 22 
12 AG +45 −64 +21 12 
13 dPreCu −11 −69 +49 15 
14 aSTS +51 −17 −09 12 
15 vIPS −35 −76 +27 13 
16 MFG −39 +16 +24 14 
17 MFG −47 +08 +34 19 
18 MFG −40 +02 +57 11 
19 pSTS −50 −19 −07 
20 MFG +34 −02 +60 17 
21 aSTS −53 −47 +06 
22 PreCeS −34 −06 +42 
No.
Side
Region
x
y
z
Voxels
RSC +18 −56 +21 24 
aINS/fO −29 +17 +02 26 
dACC/pre-SMA +03 +06 +50 15 
aINS/fO +31 +15 +04 23 
SFS −26 −03 +57 27 
RSC −17 −59 +17 11 
dPreCu −01 −59 +48 11 
AG −41 −68 +18 14 
RSC +13 −50 +08 17 
10 paraHC +23 −39 −10 11 
11 paraHC −24 −41 −11 22 
12 AG +45 −64 +21 12 
13 dPreCu −11 −69 +49 15 
14 aSTS +51 −17 −09 12 
15 vIPS −35 −76 +27 13 
16 MFG −39 +16 +24 14 
17 MFG −47 +08 +34 19 
18 MFG −40 +02 +57 11 
19 pSTS −50 −19 −07 
20 MFG +34 −02 +60 17 
21 aSTS −53 −47 +06 
22 PreCeS −34 −06 +42 

Cortical regions showing domain-general sustained activity, listed according to the consistency across individual participants. The table indicates hemisphere, name, Talairach coordinates, and voxel size. AG = angular gyrus; aSTS = anterior STS; dPreCu, dorsal precuneus; paraHC = parahippocampal gyrus; MFG = middle frontal gyrus; PreCeS = precentral sulcus; pSTS = posterior STS; RSC = retrosplenial cortex; vIPS = ventral intraparietal sulcus.

Transient and Sustained Activity in the CON within the Course of a Trial

We examined the full profile of task-evoked activity in the three regions of the CON during the memory and the perceptual search tasks (Figure 3). BOLD activity relative to each transient and sustained task process was obtained for each ROI with a model that assumed a shape of the hemodynamic response function (Figure 3A). Because the regions were defined on the basis of their sustained activity in both tasks, we show the response for each of the sustained processes for display purposes only. A significant BOLD response was observed during (i) the sentence reading phase that preceded the search phase in both the memory [left aINS/fO: t(18) = 3.79, p < .001; right dACC/pre-SMA: t(18) = 4.28, p < .001; right aINS/fO: t(18) = 4.56, p < .001] and the perceptual tasks [left aINS/fO: t(18) = 4.16, p < .001; right dACC/pre-SMA: t(18) = 3.43, p < .005; right aINS/fO: t(18) = 2.70, p < .05] and (ii) the detection/response phase that followed the search phase in both the memory [left aINS/fO: t(18) = 7.88, p < .001; right dACC/pre-SMA: t(18) = 4.17, p < .001; right aINS/fO: t(18) = 6.90, p < .001] and the perceptual tasks [left aINS/fO: t(18) = 15.98, p < .001; right dACC/pre-SMA: t(18) = 13.36, p < .001; right aINS/fO: t(18) = 16.19, p < .001]. Therefore, a significant BOLD response was observed during all the functionally distinct phases of the two tasks: An initial activation during sentence reading was followed by sustained activity over the entire duration of the search phase, which in turn was followed by transient activity evoked by detection/motor response. The robust response to the detection phase, especially during the memory task, was a highly distinctive property that was observed in only a few regions showing domain-general sustained activity, as shown in Figure 4 (last two columns in each graph) and Table 2 (sixth column). These additional regions were located in left lateral pFC (5, 16, 17, 22) and dorsal precuneus (7, 13). However, only the former group also exhibited significant activity for the sentence reading phase of both tasks, thus showing responses for all aspects of both task.

Figure 3. 

(A) The complete pattern of task-evoked activity for the three regions of the CON. The BOLD percent response change is relative to each transient and sustained process of the memory (red) and the perceptual (green) tasks obtained with a model that assumed a shape of the hemodynamic response function (process-GLM). Processes were grouped into sustained processes that led to ROI definition (left bars on gray background), processes related to the early reading of the sentence indicating the target of the search (center bars), and those related to the final transient detection/response phase (right bars). Two sustained processes were estimated for the perceptual task, corresponding to stimulus (display) and search duration. Asterisks indicate a significant response compared with the baseline. Error bars represent SEM. The statistical significance of the sustained parameters was not calculated, as domain-general regions were defined on the basis of this criterion. (B) Time courses of BOLD activity for trials of different duration (early, middle, and late correct response trials) in each task, starting at sentence onset. Trials of increasing duration are represented by lines of increasing size. The time courses reflect the sum of the BOLD signal for the processes operative on each trials and were obtained with a model that made no assumption about the shape of the hemodynamic response (frame-by-frame GLM). The approximate trial duration for the three intervals (top) and the temporal scale (bottom) are indicated in the left graph.

Figure 3. 

(A) The complete pattern of task-evoked activity for the three regions of the CON. The BOLD percent response change is relative to each transient and sustained process of the memory (red) and the perceptual (green) tasks obtained with a model that assumed a shape of the hemodynamic response function (process-GLM). Processes were grouped into sustained processes that led to ROI definition (left bars on gray background), processes related to the early reading of the sentence indicating the target of the search (center bars), and those related to the final transient detection/response phase (right bars). Two sustained processes were estimated for the perceptual task, corresponding to stimulus (display) and search duration. Asterisks indicate a significant response compared with the baseline. Error bars represent SEM. The statistical significance of the sustained parameters was not calculated, as domain-general regions were defined on the basis of this criterion. (B) Time courses of BOLD activity for trials of different duration (early, middle, and late correct response trials) in each task, starting at sentence onset. Trials of increasing duration are represented by lines of increasing size. The time courses reflect the sum of the BOLD signal for the processes operative on each trials and were obtained with a model that made no assumption about the shape of the hemodynamic response (frame-by-frame GLM). The approximate trial duration for the three intervals (top) and the temporal scale (bottom) are indicated in the left graph.

Figure 4. 

The pattern of task-evoked activity for each of the other regions showing domain-general sustained activity in both tasks (n = 19). The BOLD percent response change is relative to each transient and sustained process of the memory (red) and the perceptual (green) tasks obtained with the process-GLM. Processes were grouped into sustained processes, sentence-reading phase, and final transient detection/response phase. Asterisks indicate a significant activation/deactivation compared with the baseline for the additional processes. Error bars represent SEM.

Figure 4. 

The pattern of task-evoked activity for each of the other regions showing domain-general sustained activity in both tasks (n = 19). The BOLD percent response change is relative to each transient and sustained process of the memory (red) and the perceptual (green) tasks obtained with the process-GLM. Processes were grouped into sustained processes, sentence-reading phase, and final transient detection/response phase. Asterisks indicate a significant activation/deactivation compared with the baseline for the additional processes. Error bars represent SEM.

Table 2. 

Response Profile of Domain-general ROIs

No.
Side
Region
Memory Task
Perceptual Task
Sent.
Det.
Sent.
Det.
RSC 6.40** 1.20 6.30** 7.57** 
aINS/fO 3.79** 7.88** 4.16** 15.98** 
dACC/pre-SMA 4.28** 4.17** 3.43** 13.36** 
aINS/fO 4.56** 6.90** 2.69* 16.19** 
SFS 4.01** 3.96** 3.58** 6.20** 
RSC 4.60** 0.64 3.01* 4.38** 
dPreCu 0.44 2.91* −3.16** 5.16** 
AG 4.39** −0.38 3.39** 4.07** 
RSC 6.48** −0.17 7.15** 4.89** 
10 paraHC 2.83* −0.18 1.45 6.21** 
11 paraHC 5.13** 0.74 3.82** 8.35** 
12 AG 2.85* 1.20 −1.77 4.82** 
13 dPreCu 1.83 4.72** 1.07 6.88** 
14 aSTS 7.88** 1.99 0.95 1.70 
15 vIPS 6.54** −0.14 3.61** 4.67** 
16 MFG 4.89** 2.89* 4.53** 5.78** 
17 MFG 4.52** 5.37** 3.41** 8.54** 
18 MFG 5.24** −1.08 3.58** 0.17 
19 pSTS 9.56** 1.10 4.51** 2.86* 
20 MFG 3.65** 1.97 1.52 2.59* 
21 aSTS 7.16** 4.01** 7.11** 1.51 
22 PreCeS 3.20** 2.82* 3.83** 3.94** 
No.
Side
Region
Memory Task
Perceptual Task
Sent.
Det.
Sent.
Det.
RSC 6.40** 1.20 6.30** 7.57** 
aINS/fO 3.79** 7.88** 4.16** 15.98** 
dACC/pre-SMA 4.28** 4.17** 3.43** 13.36** 
aINS/fO 4.56** 6.90** 2.69* 16.19** 
SFS 4.01** 3.96** 3.58** 6.20** 
RSC 4.60** 0.64 3.01* 4.38** 
dPreCu 0.44 2.91* −3.16** 5.16** 
AG 4.39** −0.38 3.39** 4.07** 
RSC 6.48** −0.17 7.15** 4.89** 
10 paraHC 2.83* −0.18 1.45 6.21** 
11 paraHC 5.13** 0.74 3.82** 8.35** 
12 AG 2.85* 1.20 −1.77 4.82** 
13 dPreCu 1.83 4.72** 1.07 6.88** 
14 aSTS 7.88** 1.99 0.95 1.70 
15 vIPS 6.54** −0.14 3.61** 4.67** 
16 MFG 4.89** 2.89* 4.53** 5.78** 
17 MFG 4.52** 5.37** 3.41** 8.54** 
18 MFG 5.24** −1.08 3.58** 0.17 
19 pSTS 9.56** 1.10 4.51** 2.86* 
20 MFG 3.65** 1.97 1.52 2.59* 
21 aSTS 7.16** 4.01** 7.11** 1.51 
22 PreCeS 3.20** 2.82* 3.83** 3.94** 

BOLD response for the two additional processes of the memory and perceptual tasks obtained with the process-GLM in each of the regions showing domain-general sustained activity listed in Table 1. The table indicates the result of the one sample t test against the baseline for each process and the corresponding significance level (*p < .05; **p < .005). Data for the sustained processes of both tasks are not reported because they were used for region definition.

The sustained profile of activity observed in the CON was evident from the time course of BOLD activity, obtained with the frame-by-frame GLM, illustrated in Figure 3B. The peak of BOLD activity clearly shifted forward following the progressive increase of RTs, and this shift was evident both during the memory (red) and the perceptual task (green). The reason the display parameter in the perceptual task was significantly positive is that the BOLD signal did not drop immediately to baseline following detection but returned more gradually to baseline. Compared with the memory task, the CON exhibited a larger detection-related response during the perceptual task, which may be related to different task demands (see Discussion).

Dynamic Coupling between the CON and Domain-specific Networks Depending on Task Demands

We next asked whether the pattern of functional connectivity between the CON and domain-specific regions was modulated by task demands. We first identified two set of regions (Table 3) that showed domain-specific sustained activity and corresponded to key nodes of the DMN and the DAN. These two networks are known to generate a pattern of dynamic competition both in terms of functional connectivity (Kelly, Uddin, Biswal, Castellanos, & Milham, 2008; Fox et al., 2005) and task-evoked activity (Sestieri et al., 2010; Shulman et al., 2003). Then we assessed the strength of connectivity between the CON and these two sets of domain-specific regions at rest and during the memory and the perceptual task sets. The results are illustrated in Figure 5A (filled bars). At rest, the CON exhibited negative coupling with the DMN (p < .001 for all the regions of the CON; one sample t test) and positive coupling (p < .005 for all the regions of the CON; one sample t test) with the DAN. Notably, among the regions that showed any evidence of domain-general sustained activity, only those of the CON showed significant negative coupling at rest with DMN regions, as illustrated in Figure 6 (left gray bars) and Table 4 (fourth column). The sign of the negative correlations during the memory task could have reflected the use of whole brain signal regression (Murphy, Birn, Handwerker, Jones, & Bandettini, 2009, but see Fox, Zhang, Snyder, & Raichle, 2009). When the analysis was redone without whole brain signal regression, the correlation coefficients were close to zero (empty bars in Figure 5A). However, irrespective of the true “zero-point” of the correlation scale, the correlations during the perception task were considerably and significantly more positive than the correlations during the memory task. Therefore, at rest there was greater positive coupling between the CON and the DAN than between the CON and DMN.

Table 3. 

Selected Domain-specific ROIs

Side
Region
x
y
z
Voxels
Memory Set 
AG −43 −65 +25 31 
AG +39 −70 +40 31 
PCC/PreCu −04 −64 +24 32 
PCC/PreCu 11 −63 +25 29 
 
Perceptual Set 
Precentral sulcus (FEF) −22 −06 +51 24 
Precentral sulcus (FEF) +36 −07 +46 35 
pIPS −19 −72 +44 29 
pIPS +23 −74 +42 28 
Side
Region
x
y
z
Voxels
Memory Set 
AG −43 −65 +25 31 
AG +39 −70 +40 31 
PCC/PreCu −04 −64 +24 32 
PCC/PreCu 11 −63 +25 29 
 
Perceptual Set 
Precentral sulcus (FEF) −22 −06 +51 24 
Precentral sulcus (FEF) +36 −07 +46 35 
pIPS −19 −72 +44 29 
pIPS +23 −74 +42 28 

Cortical regions showing domain-specific sustained activity during the memory and the perceptual task selected as representative of the DMN and the DAN, respectively. PCC/PreCu = posterior cingulate/precuneus; pIPS = posterior intraparietal sulcus.

Figure 5. 

(A) The pattern of functional connectivity between the CON and domain-specific regions of the DMN and DAN that were selectively activated by sustained processes of the memory and perceptual task, respectively. The measure of functional connectivity was assessed at rest and during each task after the removal of the event-related activity by averaging the z-transformed correlation values between the region of the CON and the four representative ROIs of each domain-specific network. Filled and empty bars represent the z-transformed correlation values obtained with and without regression of the whole brain signal, respectively. Error bars represent SEM. Asterisks and circles represent significant differences across conditions. (B) The pattern of within-network connectivity in the three networks identified in this study (CON, DMN, DAN) as a function of task. No increase of within-network connectivity was observed compared with the resting state, supporting the hypothesis that task-induced modulations did not simply reflect the task-specific pattern of coactivation.

Figure 5. 

(A) The pattern of functional connectivity between the CON and domain-specific regions of the DMN and DAN that were selectively activated by sustained processes of the memory and perceptual task, respectively. The measure of functional connectivity was assessed at rest and during each task after the removal of the event-related activity by averaging the z-transformed correlation values between the region of the CON and the four representative ROIs of each domain-specific network. Filled and empty bars represent the z-transformed correlation values obtained with and without regression of the whole brain signal, respectively. Error bars represent SEM. Asterisks and circles represent significant differences across conditions. (B) The pattern of within-network connectivity in the three networks identified in this study (CON, DMN, DAN) as a function of task. No increase of within-network connectivity was observed compared with the resting state, supporting the hypothesis that task-induced modulations did not simply reflect the task-specific pattern of coactivation.

Figure 6. 

The pattern of functional connectivity between each of the other regions showing domain-general sustained activity and domain-specific regions of the DMN and DAN. Error bars represent SEM. Asterisks represent significant differences across conditions calculated with Duncan t tests.

Figure 6. 

The pattern of functional connectivity between each of the other regions showing domain-general sustained activity and domain-specific regions of the DMN and DAN. Error bars represent SEM. Asterisks represent significant differences across conditions calculated with Duncan t tests.

Table 4. 

Resting State and Task-evoked Connectivity between Domain-general ROIs and Domain-specific ROIs

No.
Side
Region
Resting State
Memory
Perceptual
3 × 2 ANOVA Interaction
DMN
DAN
DMN
DAN
DMN
DAN
RSC 0.49** 0.01 0.56 0.11 0.31 0.20 ** (**) 
aINS/fO −0.12** 0.11** 0.08 0.06 −0.05 0.16 ** (**) 
dACC/pre-SMA −0.24** 0.26** 0.05 0.27 −0.10 0.30 ** (**) 
aINS/fO −0.22** 0.18** 0.05 0.09 −0.09 0.18 ** (**) 
SFS −0.08 0.43** 0.09 0.43 −0.04 0.47 ** (*) 
RSC 0.44** −0.02 0.48 0.11 0.25 0.16 ** (**) 
dPreCu 0.33** 0.05 0.36 0.15 0.19 0.19 ** (**) 
AG 0.43** −0.04 0.53 0.04 0.31 0.08 ** (**) 
RSC 0.43** −0.07* 0.48 0.08 0.26 0.13 ** (**) 
10 paraHC 0.26** −0.01 0.27 0.09 0.13 0.15 ** (**) 
11 paraHC 0.33** −0.03 0.32 0.09 0.17 0.16 ** (**) 
12 AG 0.38** −0.07* 0.44 0.05 0.28 0.07 ** (**) 
13 dPreCu −0.04 0.43** 0.07 0.38 −0.03 0.44 * (ns
14 aSTS 0.27** −0.15** 0.35 −0.01 0.18 0.02 ** (**) 
15 vIPS 0.36** 0.11* 0.48 0.15 0.24 0.25 ** (**) 
16 MFG 0.03 0.11* 0.31 0.09 0.09 0.18 ** (**) 
17 MFG 0.00 0.14** 0.26 0.10 0.07 0.18 ** (**) 
18 MFG 0.17** −0.09* 0.26 −0.03 0.10 0.00 ** (*) 
19 pSTS 0.21** −0.11** 0.28 −0.02 0.17 0.00 ** (*) 
20 MFG 0.03 0.21** 0.16 0.16 0.04 0.19 ** (*) 
21 aSTS 0.05 −0.10** 0.13 −0.05 0.09 −0.01 ns (ns
22 PreCeS −0.05 0.20** 0.13 0.22 0.00 0.28 ** (**) 
No.
Side
Region
Resting State
Memory
Perceptual
3 × 2 ANOVA Interaction
DMN
DAN
DMN
DAN
DMN
DAN
RSC 0.49** 0.01 0.56 0.11 0.31 0.20 ** (**) 
aINS/fO −0.12** 0.11** 0.08 0.06 −0.05 0.16 ** (**) 
dACC/pre-SMA −0.24** 0.26** 0.05 0.27 −0.10 0.30 ** (**) 
aINS/fO −0.22** 0.18** 0.05 0.09 −0.09 0.18 ** (**) 
SFS −0.08 0.43** 0.09 0.43 −0.04 0.47 ** (*) 
RSC 0.44** −0.02 0.48 0.11 0.25 0.16 ** (**) 
dPreCu 0.33** 0.05 0.36 0.15 0.19 0.19 ** (**) 
AG 0.43** −0.04 0.53 0.04 0.31 0.08 ** (**) 
RSC 0.43** −0.07* 0.48 0.08 0.26 0.13 ** (**) 
10 paraHC 0.26** −0.01 0.27 0.09 0.13 0.15 ** (**) 
11 paraHC 0.33** −0.03 0.32 0.09 0.17 0.16 ** (**) 
12 AG 0.38** −0.07* 0.44 0.05 0.28 0.07 ** (**) 
13 dPreCu −0.04 0.43** 0.07 0.38 −0.03 0.44 * (ns
14 aSTS 0.27** −0.15** 0.35 −0.01 0.18 0.02 ** (**) 
15 vIPS 0.36** 0.11* 0.48 0.15 0.24 0.25 ** (**) 
16 MFG 0.03 0.11* 0.31 0.09 0.09 0.18 ** (**) 
17 MFG 0.00 0.14** 0.26 0.10 0.07 0.18 ** (**) 
18 MFG 0.17** −0.09* 0.26 −0.03 0.10 0.00 ** (*) 
19 pSTS 0.21** −0.11** 0.28 −0.02 0.17 0.00 ** (*) 
20 MFG 0.03 0.21** 0.16 0.16 0.04 0.19 ** (*) 
21 aSTS 0.05 −0.10** 0.13 −0.05 0.09 −0.01 ns (ns
22 PreCeS −0.05 0.20** 0.13 0.22 0.00 0.28 ** (**) 

Pattern of connectivity between each of the regions showing domain-general sustained activity and the four representative regions of the DMN and the DAN as a function of task demands (rest, memory task, perceptual task). Connectivity is reported in terms of z-transformed correlation values. For the resting state condition, the asterisks indicate whether a significance difference against null correlation was observed (*p < .05; **p < .005; one sample t test). The asterisks in the last column indicates whether the two-way ANOVA showed a significant interaction effect between task and network (*p < .05; **p < .005). The asterisks in parentheses indicates a significant interaction effect when the analysis was repeated without regression of the whole brain signal.

The pattern of internetwork connectivity was strongly modulated by task execution. Importantly, the CON showed higher connectivity with the currently active task-relevant network (DMN for memory, DAN for perceptual task), compared with the task-irrelevant one. The task dependent dynamic coupling was assessed in each ROI through two-way repeated-measures ANOVAs with Network (DMN, DAN) and Task (rest, memory, perceptual) as factors, which revealed a significant Network × Task interaction [left aINS/fO: F(2, 36) = 31.27, p < .001; right dACC/pre-SMA: F(2, 36) = 16.61, p < .001; right aINS/fO: F(2, 36) = 45.30, p < .001]. An even stronger result was obtained when internetwork connectivity during memory and perceptual tasks were directly compared in an ANOVA that excluded the resting state connectivity. Post hoc analyses revealed that the CON–DMN connectivity was different across the two search tasks in all three CON regions (p < .001), whereas the CON–DAN was greater in the perceptual compared with the memory task in left (p < .001) and right (p < .001) aINS/fO but not in the right dACC/pre-SMA. However, this dynamic coupling was observed in almost all the other regions that putatively showed domain-general sustained activity, as shown by Figure 6 (red and green bars) and Table 4 (last column). Therefore, flexible task-dependent functional connectivity with domain-specific regions was not a distinctive property of the CON.

The analysis conducted without regression of the whole brain signal, while resulting in a general increase of connectivity regardless of network and task, provided evidence for a very similar pattern of task-induced modulations (the presence of a significant interaction effect is indicated in last column of Table 4, within parentheses). We also conducted a control within-network analysis to address whether task-induced modulations of FC always reflected the pattern of task-specific coactivation. The results are illustrated in Figure 5B. The two-way ANOVA with Network (CON, DMN, DAN) and Task (rest, memory, perceptual) as factors showed a significant interaction effect, F(4, 72) = 18.47, p < .0001. Post hoc analyses revealed no significant increases of FC from rest to task within any of the three networks, but the trend was in the direction expected based on the degree of activation. Therefore, the results are not conclusive concerning whether regions that are coactivated during task execution (e.g., DAN during the perceptual task) also tend to show an increase of task-dependent FC.

Distinctiveness of the CON among Other Regions Showing Domain-general Sustained Activity

To further investigate the functional distinctiveness of the CON, we conducted a multivariate hierarchical clustering procedure on the whole set of regions that putatively showed domain-general sustained activity. This analysis took several measures of task-evoked activity and functional connectivity into account, assigning equal weight to each measure. The results of the clustering procedure are illustrated in Figure 7. Importantly, the three regions of the CON (purple) were not only grouped in the same cluster but were also the only members of this cluster. The closest cluster (green) and thus the one exhibiting the most similar properties to the CON included a region within the superior frontal sulcus (SFS) and a region of the dorsal precuneus. Three other prefrontal regions were grouped in a more distant cluster (yellow). Many regions of the ventral parietal, retrosplenial, and parahippocampal gyrus formed a distinct cluster (red) showing different functional properties. This analysis confirmed the common clustering of regions within the CON and their segregation from other regions showing putative domain-general sustained activity.

Figure 7. 

(A) Anatomical location of the cortical regions showing domain-general sustained activity. Regions with the same color were grouped in the same cluster according to the multivariate hierarchical clustering analysis conducted on measures of task-evoked activity and functional connectivity. (B) Graph representing the correlation distance between each region illustrated in A. The threshold for the pruning of the cluster tree was calculated maximizing cluster size and measures of clustering validity (see Methods).

Figure 7. 

(A) Anatomical location of the cortical regions showing domain-general sustained activity. Regions with the same color were grouped in the same cluster according to the multivariate hierarchical clustering analysis conducted on measures of task-evoked activity and functional connectivity. (B) Graph representing the correlation distance between each region illustrated in A. The threshold for the pruning of the cluster tree was calculated maximizing cluster size and measures of clustering validity (see Methods).

Control analyses further tested the stability and reproducibility of the hierarchical clustering results across subjects and measures. The mean value of the correlation between dissimilarity matrices obtained creating different subgroup permutations was .94 ± .02, indicating a strong interindividual consistency. The clustering analysis was also repeated using a subset of the four functional measures. The regions of the CON were always in the same cluster, independent of the subset of parameters taken into consideration. However, for three of four permutations, this cluster also included the left SFS and left dorsal precuneus region, confirming their functional similarity to the CON. In general, the analysis confirmed that regions of the CON formed a cluster and that the regions showing the most similar characteristics to the CON were located in left SFS and left dorsal precuneus.

DISCUSSION

The results demonstrated the generality of the CON, both across two tasks that involved very different cognitive processes and across the different cognitive processes that occurred within each task. The CON was among a small number of regions that showed sustained activity during extended periods in which people searched for behaviorally relevant information in the environment or from long-term episodic memory. This common activity was observed despite the fact that each task recruited a specific domain-specific network, the DAN and DMN, that shows a mutual competitive relationship (Sestieri et al., 2010; Fox et al., 2005). Consistent with a role in cognitive control, within each task, the CON was significantly activated by all of the sustained and transient processes that were operative within a trial, from the initial sentence reading phase to the final transient response phase. The CON was dynamically coupled with task-specific networks, but this property was shared by almost all regions that showed any evidence of sustained, domain-general activity. A multivariate clustering procedure, which combined measures of task-evoked activity and functional connectivity, demonstrated the functional distinctiveness of the CON from lateral frontoparietal regions that are also thought to be involved in task control.

Domain-general Signals in the CON

In this study, very limited portions of cortex showed domain-general sustained activity, compared with the large extent of regions that exhibited sustained activity within a single domain. Domain-specific activity for memory and perceptual tasks showed a striking topographical relationship characterized by spatial contiguity, which was previously described within the parietal cortex (Sestieri et al., 2010), but here was extended to frontal and temporal cortex. Sustained activity in the CON was observed during periods in which participants searched through dynamically presented scene episodes for the presence of a particular object or in which they retrieved information from episodic memory while fixating a blank screen. The common, sustained activation of the CON under these very different circumstances confirmed its general involvement across highly disparate cognitive processes that were distinguished by the presence or absence of sensory stimulation and perceptual processing and whether attention was directed toward external/on-line or internal/off-line representations. Moreover, this common sustained activity was observed during tasks that produced domain-specific activity in one or the other of two brain networks, the DAN or the DMN, that show a competitive relationship.

In addition, the experimental design allowed us to separately assess the involvement of the CON across a variety of processes that occurred within each task, from the reading of the cue sentence through to the motor response or end of sensory stimulation. The CON was significantly activated across all of these processes, again attesting to its general involvement across different cognitive processes. Although the CON was activated during all task phases, however, the magnitude of activation also depended on specific task demands. For example, the transient activity related to the detection/response phase was considerably more pronounced in the perceptual search than episodic memory task. This result may reflect the fact that the memory task was self-paced, whereas in the perception task, the timing of the response was dictated by the abrupt target onset.

Several previous studies have suggested that the CON is part of a salience network that facilitates the detection of behaviorally important or salient environmental stimuli (Seeley et al., 2007) and triggers a cascade of cognitive control signals (Menon & Uddin, 2010). The current results indicate that this description may be incomplete. The sustained involvement of the CON over the extended course of a trial adds support to the idea that the network is involved in maintaining a task set (Dosenbach et al., 2006), coordinating or sequencing task processes, or maintaining sustained effort (Sterzer & Kleinschmidt, 2010), perhaps coupled with transient processes related to error and salience detection.

Finally, we did not find evidence for the presence of sustained domain-general activity in regions that appeared in later definitions of the CON based on rsFC analyses (Dosenbach et al., 2007, 2008), namely the anterior pFC (BA 10) and the anterior thalamus. Whereas Dosenbach and colleagues also did not observe sustained activity in the thalamus, they did observe sustained activity in BA 10 for a subset of the tasks included in their meta-analysis (Dosenbach et al., 2006). It is possible that BA 10 only responds to tasks requiring more complex forms of cognitive control, such as relational integration, planning, or multitasking, with respect to our tasks (Gerlach, Spreng, Gilmore, & Schacter, 2011; Badre & D'Esposito, 2007). However, the present tasks were not chosen to cover all the possible elements of cognitive control but were selected to show minimal functional overlap and strongly activate either the DAN or DMN. The lack of anterior prefrontal sustained activity may also reflect the direct contrast of the BOLD task-related response against the baseline. It has been shown that fixation can be accompanied by complex forms of internally directed thought processes (i.e., mind-wandering), which have been associated with the activity of medial anterior prefrontal regions (Christoff, Gordon, Smallwood, Smith, & Schooler, 2009). It is thus possible that common activity during tasks and fixation might have obscured some patterns of domain-general brain activity.

Dynamic Coupling between Domain-general and Domain-specific Regions

Previous studies have proposed that frontoparietal regions regulate the competition between the DMN and the DAN. Several studies have analyzed functional and effective connectivity across different tasks (Gao & Lin, 2012), including externally or internally oriented tasks (Smallwood et al., 2012; Spreng et al., 2010) that activated the DAN or DMN, respectively. However, these studies did not focus on the CON but examined changes in the pattern of connectivity within a larger network of frontoparietal regions, making it difficult to examine the specific relationship between the CON and domain-specific regions. Conversely, previous analyses of task-related changes in the functional and effective connectivity of specific components of the CON have only been conducted separately using either externally oriented (Higo et al., 2011; Zanto et al., 2011; Sridharan et al., 2008) or internally oriented tasks (Chiong et al., 2013). This study demonstrates the presence of robust, task-specific changes of connectivity between regions of the CON and other higher-order regions belonging to the DMN and the DAN, indicating flexible interactions across networks that depend on task demands (see also Chadick & Gazzaley, 2011). Such internetwork functional interactions may be mediated by anatomical connections between key nodes of the CON and relevant frontoparietal networks. Although the pattern of anatomical connectivity of the dACC/pre-SMA and aINS/fO in humans has not been comprehensively characterized (see Menon & Uddin, 2010; Beckmann, Johansen-Berg, & Rushworth, 2009), a recent study has provided evidence that the structural integrity of the cingulo-opercular connection is necessary for the efficient regulation of activity in the DMN, both in terms of functional connectivity and behavioral performance (Bonnelle et al., 2012).

However, we found a similar pattern of flexible task-dependent functional connectivity in almost all regions that exhibited any evidence of domain-general sustained activity. Although the functions of regions outside the CON were not the main focus of the present report, it seems unlikely that they all involved cognitive control. For example, both tasks may have activated regions involved in representing scenes, either during on-line perception or during episodic memory retrieval (Chun & Johnson, 2011), see below. Following this logic, the modulation of task-related functional connectivity of domain-general regions with domain-specific networks is not a unique marker of cognitive control.

Caution also should be exercised in interpreting task-dependent increases in functional connectivity as enhanced functional communication between regions (Spreng et al., 2010). First, the pattern of task-dependent functional connectivity matched the pattern of mean activity across regions and tasks. Therefore, although this study adopted a conservative approach aimed at minimizing the contribution of the mean event-related activity on the time series, it is still possible that residual trial-to-trial variations in the magnitude of the task signal biased measures of functional connectivity. In addition, the neurophysiological basis of BOLD functional connectivity remains controversial at best, both at rest and during task performance. Current models emphasize that resting state BOLD connectivity reflects either fluctuations of slow cortical potentials (He, Snyder, Zempel, Smyth, & Raichle, 2008) or slow band-limited fluctuations of signals in alpha and beta bands between distant regions of cortex (Hipp, Hawellek, Corbetta, Siegel, & Engel, 2012; Brookes et al., 2011; de Pasquale et al., 2010). In contrast, cortical activations as measured electrophysiologically are characterized by decrements of alpha/beta power/coherence and increases in power/coherence at higher frequencies (e.g., gamma; Miller, 2010; Siegel, Donner, Oostenveld, Fries, & Engel, 2008; Fries, 2005). Independent measures of fMRI functional connectivity coupled with electrophysiological measures (Daitch et al., 2010) may provide useful constraints on interpretation.

Functional Distinctiveness of the CON

Several studies have proposed that regions of the CON are part of a larger frontoparietal control network (FPCN) that also includes regions in lateral prefrontal (e.g., MFG, IFG) and inferior parietal cortex (Vincent et al., 2008; see also Falkenberg, Specht, & Westerhausen, 2011; Cole & Schneider, 2007, for alternative versions of the network topography). The FPCN is thought to be anatomically interposed between the DMN and the anticorrelated DAN (Fox et al., 2005) and thus well positioned to facilitate functional integration between these two networks (Spreng et al., 2010, 2013; Vincent et al., 2008). However, the present study showed that, although many lateral frontoparietal regions showed memory-specific sustained activity, they were poorly recruited by the perceptual task. The region in the SFS that showed domain-general activity was clearly more dorsal than the lateral prefrontal regions of the FPCN (Figure 2). The ventral lateral prefrontal regions that showed evidence of domain-general activity are likely located in the FPCN, although the correspondence was not exact (see Figure 2). Overall, however, a fair number of regions in the FPCN were likely not involved in regulating the competition between the DMN and DAN or coordinating the two networks. Importantly, the lack of activation in these FPCN regions during the perceptual search task was not because that task was easy or automatic. The perception task was quite demanding, producing long RTs and relatively low accuracy. Moreover, the perception task resulted in very sustained, domain-specific signals in IPS, FEF, and other regions putatively involved in top–down control of sensorimotor cortex (Corbetta & Shulman, 2002) as well as sustained activity in the CON. It is the case that the domain-general activity observed here reflected the particular perception and memory tasks that were studied, which did not cover the entire spectrum of processes associated with cognitive control. Nonetheless, for these two tasks, we observed a differential response between the CON and many frontoparietal regions, and we are not aware of studies showing a converse dissociation in which frontoparietal regions of the FPCN are commonly activated across very different tasks in the absence of common activation in the CON.

The CON was also distinguished from most frontoparietal regions that showed evidence in the current study for domain-general activity, as the multivariate hierarchical clustering analysis indicated that these regions and the CON formed separate clusters. The two domain-general regions that were most similar to the CON, according to the clustering analysis, were located in the dorsal precuneus and SFS. The dorsal precuneus has been associated with the domain-independent function of shifting between task sets (Greenberg et al., 2010; Chiu & Yantis, 2009). Consistent with this characterization, the dorsal precuneus showed no activation or even a significant deactivation (see Table 2 and Figure 4) during the sentence-reading phase of both tasks. Therefore, the dorsal precuneus did not respond during the transition from a resting state to a task state, but only during the course of a trial in which transitions likely occurred between different active task processes (e.g., the transition from visual search to target detection; see Shulman et al., 2009; Yantis et al., 2002). In contrast, the CON was significantly activated during the sentence reading phase of both tasks, consistent with a role in instantiating a task set (Dosenbach et al., 2006) during a transition from rest to a task state. These considerations suggest that, although the dorsal precuneus showed some functional similarity with the CON, as indicated by the multivariate clustering procedure, it nevertheless has some distinguishing characteristics. In contrast, the current results did not clearly distinguish the CON from the SFS region.

Finally, the cluster of domain-general regions showing a functional profile that differed the most from the CON contained many regions that have been associated with scene perception, including bilateral retrosplenial and left posterior parahippocampal cortex (Vann, Aggleton, & Maguire, 2009; Walther, Caddigan, Fei-Fei, & Beck, 2009; Epstein, 2008). The common activation in these regions may have reflected the recruitment of the same “representational” regions, in terms of on-line scene perception and off-line scene reconstruction during the perceptual and memory tasks, respectively. A representational function related to scene processing is consistent with the task-evoked fingerprint of these regions, which were strongly activated by the display component of the perception task but showed significantly less activation during the detection/response components of both tasks relative to the CON.

Overall, the present results support the idea that the CON is a network with distinctive properties (Dosenbach et al., 2006), functionally separate from other frontal and parietal regions (Dosenbach et al., 2008). Additional support for this conclusion has been provided in a recent study on the dynamics of large-scale brain functional networks during recollection (Fornito, Harrison, Zalesky, & Simons, 2012), which reported a functional distinction between the CON and frontoparietal systems, as well as between left- and right-lateralized components of the FPCN. In particular, each of these networks displayed diverse modes of context-dependent interaction with the DMN and a different relationship with behavioral performance. Taken together, these findings indicate the functional specificity of the CON with respect to other frontal and parietal regions putatively involved in task control.

Acknowledgments

We are grateful to Abraham Z. Snyder, Annalisa Tosoni, and Chris Lewis for helpful discussions, Anna L. Pope for technical help, and Mark McAvoy for guidance with statistical analyses. This work was supported by National Institutes of Mental Health grant R01 1R01MH096482.

Reprint requests should be sent to Carlo Sestieri, Department of Neuroscience and Imaging, G. d'Annunzio University, ITAB Institute for Advanced Biomedical Technologies, Via dei Vestini, 31, 66013 Chieti, Italy, or via e-mail: c.sestieri@unich.it.

REFERENCES

REFERENCES
Andrews-Hanna
,
J. R.
,
Reidler
,
J. S.
,
Sepulcre
,
J.
,
Poulin
,
R.
, &
Buckner
,
R. L.
(
2010
).
Functional-anatomic fractionation of the brain's default network.
Neuron
,
65
,
550
562
.
Aron
,
A. R.
,
Behrens
,
T. E.
,
Smith
,
S.
,
Frank
,
M. J.
, &
Poldrack
,
R. A.
(
2007
).
Triangulating a cognitive control network using diffusion-weighted magnetic resonance imaging (MRI) and functional MRI.
Journal of Neuroscience
,
27
,
3743
3752
.
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
.
Beckmann
,
M.
,
Johansen-Berg
,
H.
, &
Rushworth
,
M. F.
(
2009
).
Connectivity-based parcellation of human cingulate cortex and its relation to functional specialization.
Journal of Neuroscience
,
29
,
1175
1190
.
Bonnelle
,
V.
,
Ham
,
T. E.
,
Leech
,
R.
,
Kinnunen
,
K. M.
,
Mehta
,
M. A.
,
Greenwood
,
R. J.
,
et al
(
2012
).
Salience network integrity predicts default mode network function after traumatic brain injury.
Proceedings of the National Academy of Sciences, U.S.A.
,
109
,
4690
4695
.
Boynton
,
G. M.
,
Engel
,
S. A.
,
Glover
,
G. H.
, &
Heeger
,
D. J.
(
1996
).
Linear systems analysis of functional magnetic resonance imaging in human V1.
Journal of Neuroscience
,
16
,
4207
4221
.
Brookes
,
M. J.
,
Woolrich
,
M.
,
Luckhoo
,
H.
,
Price
,
D.
,
Hale
,
J. R.
,
Stephenson
,
M. C.
,
et al
(
2011
).
Investigating the electrophysiological basis of resting state networks using magnetoencephalography.
Proceedings of the National Academy of Sciences, U.S.A.
,
108
,
16783
16788
.
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
.
Chadick
,
J. Z.
, &
Gazzaley
,
A.
(
2011
).
Differential coupling of visual cortex with default or frontal-parietal network based on goals.
Nature Neuroscience
,
14
,
830
832
.
Chiong
,
W.
,
Wilson
,
S. M.
,
D'Esposito
,
M.
,
Kayser
,
A. S.
,
Grossman
,
S. N.
,
Poorzand
,
P.
,
et al
(
2013
).
The salience network causally influences default mode network activity during moral reasoning.
Brain
,
136
,
1929
1941
.
Chiu
,
Y. C.
, &
Yantis
,
S.
(
2009
).
A domain-independent source of cognitive control for task sets: Shifting spatial attention and switching categorization rules.
Journal of Neuroscience
,
29
,
3930
3938
.
Christoff
,
K.
,
Gordon
,
A. M.
,
Smallwood
,
J.
,
Smith
,
R.
, &
Schooler
,
J. W.
(
2009
).
Experience sampling during fMRI reveals default network and executive system contributions to mind wandering.
Proceedings of the National Academy of Sciences, U.S.A.
,
106
,
8719
8724
.
Chun
,
M. M.
, &
Johnson
,
M. K.
(
2011
).
Memory: Enduring traces of perceptual and reflective attention.
Neuron
,
72
,
520
535
.
Cole
,
M. W.
, &
Schneider
,
W.
(
2007
).
The cognitive control network: Integrated cortical regions with dissociable functions.
Neuroimage
,
37
,
343
360
.
Corbetta
,
M.
, &
Shulman
,
G. L.
(
2002
).
Control of goal-directed and stimulus-driven attention in the brain.
Nature Reviews Neuroscience
,
3
,
201
215
.
Daitch
,
A. L.
,
Snyder
,
A. Z.
,
Astafiev
,
S.
,
Bundy
,
D.
,
Freudenburg
,
Z.
,
Gaona
,
B.
,
et al
(
2010
).
Temporal dynamics of stimulus-driven attention shifts as studied through the combined use of ECoG and fMRI.
Paper presented at the Society for Neuroscience
.
de Pasquale
,
F.
,
Della Penna
,
S.
,
Snyder
,
A. Z.
,
Lewis
,
C.
,
Mantini
,
D.
,
Marzetti
,
L.
,
et al
(
2010
).
Temporal dynamics of spontaneous MEG activity in brain networks.
Proceedings of the National Academy of Sciences, U.S.A.
,
107
,
6040
6045
.
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
.
Duncan
,
J.
, &
Owen
,
A. M.
(
2000
).
Common regions of the human frontal lobe recruited by diverse cognitive demands.
Trends in Neurosciences
,
23
,
475
483
.
Epstein
,
R. A.
(
2008
).
Parahippocampal and retrosplenial contributions to human spatial navigation.
Trends in Cognitive Sciences
,
12
,
388
396
.
Falkenberg
,
L. E.
,
Specht
,
K.
, &
Westerhausen
,
R.
(
2011
).
Attention and cognitive control networks assessed in a dichotic listening fMRI study.
Brain and Cognition
,
76
,
276
285
.
Fornito
,
A.
,
Harrison
,
B. J.
,
Zalesky
,
A.
, &
Simons
,
J. S.
(
2012
).
Competitive and cooperative dynamics of large-scale brain functional networks supporting recollection.
Proceedings of the National Academy of Sciences, U.S.A.
,
109
,
12788
12793
.
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
.
Fries
,
P.
(
2005
).
A mechanism for cognitive dynamics: Neuronal communication through neuronal coherence.
Trends in Cognitive Sciences
,
9
,
474
480
.
Gao
,
W.
, &
Lin
,
W.
(
2012
).
Frontal parietal control network regulates the anti-correlated default and dorsal attention networks.
Human Brain Mapping
,
33
,
192
202
.
Gerlach
,
K. D.
,
Spreng
,
R. N.
,
Gilmore
,
A. W.
, &
Schacter
,
D. L.
(
2011
).
Solving future problems: Default network and executive activity associated with goal-directed mental simulations.
Neuroimage
,
55
,
1816
1824
.
Green
,
D. M.
, &
Swets
,
J. A.
(
1966
).
Signal detection theory and psychophysics.
New York
:
Wiley
.
Greenberg
,
A. S.
,
Esterman
,
M.
,
Wilson
,
D.
,
Serences
,
J. T.
, &
Yantis
,
S.
(
2010
).
Control of spatial and feature-based attention in frontoparietal cortex.
Journal of Neuroscience
,
30
,
14330
14339
.
He
,
B. J.
,
Snyder
,
A. Z.
,
Vincent
,
J. L.
,
Epstein
,
A.
,
Shulman
,
G. L.
, &
Corbetta
,
M.
(
2007
).
Breakdown of functional connectivity in frontoparietal networks underlies behavioral deficits in spatial neglect.
Neuron
,
53
,
905
918
.
He
,
B. J.
,
Snyder
,
A. Z.
,
Zempel
,
J. M.
,
Smyth
,
M. D.
, &
Raichle
,
M. E.
(
2008
).
Electrophysiological correlates of the brain's intrinsic large-scale functional architecture.
Proceedings of the National Academy of Sciences, U.S.A.
,
105
,
16039
16044
.
Higo
,
T.
,
Mars
,
R. B.
,
Boorman
,
E. D.
,
Buch
,
E. R.
, &
Rushworth
,
M. F.
(
2011
).
Distributed and causal influence of frontal operculum in task control.
Proceedings of the National Academy of Sciences, U.S.A.
,
108
,
4230
4235
.
Hipp
,
J. F.
,
Hawellek
,
D. J.
,
Corbetta
,
M.
,
Siegel
,
M.
, &
Engel
,
A. K.
(
2012
).
Large-scale cortical correlation structure of spontaneous oscillatory activity.
Nature Neuroscience
,
15
,
1884
1890
.
Kastner
,
S.
, &
Ungerleider
,
L. G.
(
2000
).
Mechanisms of visual attention in the human cortex.
Annual Review of Neuroscience
,
23
,
315
341
.
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
.
Mendelsohn
,
A.
,
Furman
,
O.
, &
Dudai
,
Y.
(
2010
).
Signatures of memory: Brain coactivations during retrieval distinguish correct from incorrect recollection.
Frontiers in Behavioral Neuroscience
,
4
,
18
.
Menon
,
V.
, &
Uddin
,
L. Q.
(
2010
).
Saliency, switching, attention and control: A network model of insula function.
Brain Structure & Function
,
214
,
655
667
.
Miller
,
E. K.
, &
Cohen
,
J. D.
(
2001
).
An integrative theory of prefrontal cortex function.
Annual Review of Neuroscience
,
24
,
167
202
.
Miller
,
K. J.
(
2010
).
Broadband spectral change: Evidence for a macroscale correlate of population firing rate?
Journal of Neuroscience
,
30
,
6477
6479
.
Murphy
,
K.
,
Birn
,
R. M.
,
Handwerker
,
D. A.
,
Jones
,
T. B.
, &
Bandettini
,
P. A.
(
2009
).
The impact of global signal regression on resting state correlations: Are anti-correlated networks introduced?
Neuroimage
,
44
,
893
905
.
Niendam
,
T. A.
,
Laird
,
A. R.
,
Ray
,
K. L.
,
Dean
,
Y. M.
,
Glahn
,
D. C.
, &
Carter
,
C. S.
(
2012
).
Meta-analytic evidence for a superordinate cognitive control network subserving diverse executive functions.
Cognitive, Affective, and Behavioral Neuroscience
,
12
,
241
268
.
Norman-Haignere
,
S. V.
,
McCarthy
,
G.
,
Chun
,
M. M.
, &
Turk-Browne
,
N. B.
(
2011
).
Category-selective background connectivity in ventral visual cortex.
Cerebral Cortex
,
22
,
391
402
.
Ollinger
,
J. M.
,
Corbetta
,
M.
, &
Shulman
,
G. L.
(
2001
).
Separating processes within a trial in event-related functional MRI: II. Analysis.
Neuroimage
,
13
,
218
229
.
Ollinger
,
J. M.
,
Shulman
,
G. L.
, &
Corbetta
,
M.
(
2001
).
Separating processes within a trial in event-related functional MRI: I. The method.
Neuroimage
,
13
,
210
217
.
Ploran
,
E. J.
,
Nelson
,
S. M.
,
Velanova
,
K.
,
Donaldson
,
D. I.
,
Petersen
,
S. E.
, &
Wheeler
,
M. E.
(
2007
).
Evidence accumulation and the moment of recognition: Dissociating perceptual recognition processes using fMRI.
Journal of Neuroscience
,
27
,
11912
11924
.
Raichle
,
M. E.
,
MacLeod
,
A. M.
,
Snyder
,
A. Z.
,
Powers
,
W. J.
,
Gusnard
,
D. A.
, &
Shulman
,
G. L.
(
2001
).
A default mode of brain function.
Proceedings of the National Academy of Sciences, U.S.A.
,
98
,
676
682
.
Rogers
,
R. D.
, &
Monsell
,
S.
(
1995
).
Costs of a predictable switch between simple cognitive tasks.
Journal of Experimental Psychology: General
,
124
,
207
231
.
Sakai
,
K.
, &
Passingham
,
R. E.
(
2003
).
Prefrontal interactions reflect future task operations.
Nature Neuroscience
,
6
,
75
81
.
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
.
Sestieri
,
C.
,
Corbetta
,
M.
,
Romani
,
G. L.
, &
Shulman
,
G. L.
(
2011
).
Episodic memory retrieval, parietal cortex, and the default mode network: Functional and topographic analyses.
Journal of Neuroscience
,
31
,
4407
4420
.
Sestieri
,
C.
,
Shulman
,
G. L.
, &
Corbetta
,
M.
(
2010
).
Attention to memory and the environment: Functional specialization and dynamic competition in human posterior parietal cortex.
Journal of Neuroscience
,
30
,
8445
8456
.
Shulman
,
G. L.
,
Astafiev
,
S. V.
,
Franke
,
D.
,
Pope
,
D. L.
,
Snyder
,
A. Z.
,
McAvoy
,
M. P.
,
et al
(
2009
).
Interaction of stimulus-driven reorienting and expectation in ventral and dorsal frontoparietal and basal ganglia-cortical networks.
Journal of Neuroscience
,
29
,
4392
4407
.
Shulman
,
G. L.
,
Fiez
,
J. A.
,
Corbetta
,
M.
,
Buckner
,
R. L.
,
Miezin
,
F. M.
,
Raichle
,
M. E.
,
et al
(
1997
).
Common blood flow changes across visual tasks: II. Decreases in cerebral cortex.
Journal of Cognitive Neuroscience
,
9
,
648
663
.
Shulman
,
G. L.
,
McAvoy
,
M. P.
,
Cowan
,
M. C.
,
Astafiev
,
S. V.
,
Tansy
,
A. P.
,
d'Avossa
,
G.
,
et al
(
2003
).
Quantitative analysis of attention and detection signals during visual search.
Journal of Neurophysiology
,
90
,
3384
3397
.
Shulman
,
G. L.
,
Ollinger
,
J. M.
,
Akbudak
,
E.
,
Conturo
,
T. E.
,
Snyder
,
A. Z.
,
Petersen
,
S. E.
,
et al
(
1999
).
Areas involved in encoding and applying directional expectations to moving objects.
Journal of Neuroscience
,
19
,
9480
9496
.
Siegel
,
M.
,
Donner
,
T. H.
,
Oostenveld
,
R.
,
Fries
,
P.
, &
Engel
,
A. K.
(
2008
).
Neuronal synchronization along the dorsal visual pathway reflects the focus of spatial attention.
Neuron
,
60
,
709
719
.
Smallwood
,
J.
,
Brown
,
K.
,
Baird
,
B.
, &
Schooler
,
J. W.
(
2012
).
Cooperation between the default mode network and the frontal-parietal network in the production of an internal train of thought.
Brain Research
,
1428
,
60
70
.
Spreng
,
R. N.
,
Sepulcre
,
J.
,
Turner
,
G. R.
,
Stevens
,
W. D.
, &
Schacter
,
D. L.
(
2013
).
Intrinsic architecture underlying the relations among the default, dorsal attention, and frontoparietal control networks of the human brain.
Journal of Cognitive Neuroscience
,
25
,
74
86
.
Spreng
,
R. N.
,
Stevens
,
W. D.
,
Chamberlain
,
J. P.
,
Gilmore
,
A. W.
, &
Schacter
,
D. L.
(
2010
).
Default network activity, coupled with the frontoparietal control network, supports goal-directed cognition.
Neuroimage
,
53
,
303
317
.
Sridharan
,
D.
,
Levitin
,
D. J.
, &
Menon
,
V.
(
2008
).
A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks.
Proceedings of the National Academy of Sciences, U.S.A.
,
105
,
12569
12574
.
Sterzer
,
P.
, &
Kleinschmidt
,
A.
(
2010
).
Anterior insula activations in perceptual paradigms: Often observed but barely understood.
Brain Structure & Function
,
214
,
611
622
.
Van Essen
,
D. C.
(
2005
).
A Population–Average, Landmark-, and Surface-based (PALS) atlas of human cerebral cortex.
Neuroimage
,
28
,
635
662
.
Vann
,
S. D.
,
Aggleton
,
J. P.
, &
Maguire
,
E. A.
(
2009
).
What does the retrosplenial cortex do?
Nature Reviews Neuroscience
,
10
,
792
802
.
Vincent
,
J. L.
,
Kahn
,
I.
,
Snyder
,
A. Z.
,
Raichle
,
M. E.
, &
Buckner
,
R. L.
(
2008
).
Evidence for a frontoparietal control system revealed by intrinsic functional connectivity.
Journal of Neurophysiology
,
100
,
3328
3342
.
Vincent
,
J. L.
,
Snyder
,
A. Z.
,
Fox
,
M. D.
,
Shannon
,
B. J.
,
Andrews
,
J. R.
,
Raichle
,
M. E.
,
et al
(
2006
).
Coherent spontaneous activity identifies a hippocampal-parietal memory network.
Journal of Neurophysiology
,
96
,
3517
3531
.
Walther
,
D. B.
,
Caddigan
,
E.
,
Fei-Fei
,
L.
, &
Beck
,
D. M.
(
2009
).
Natural scene categories revealed in distributed patterns of activity in the human brain.
Journal of Neuroscience
,
29
,
10573
10581
.
Yantis
,
S.
,
Schwarzbach
,
J.
,
Serences
,
J. T.
,
Carlson
,
R. L.
,
Steinmetz
,
M. A.
,
Pekar
,
J. J.
,
et al
(
2002
).
Transient neural activity in human parietal cortex during spatial attention shifts.
Nature Neuroscience
,
5
,
995
1002
.
Zanto
,
T. P.
,
Rubens
,
M. T.
,
Thangavel
,
A.
, &
Gazzaley
,
A.
(
2011
).
Causal role of the prefrontal cortex in top–down modulation of visual processing and working memory.
Nature Neuroscience
,
14
,
656
661
.