Functional connectivity patterns reflect individual differences in conflict adaptation

Neuropsychologia 70 (2015) 177–184

Contents lists available at ScienceDirect

Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia

Functional connectivity patterns reflect individual differences in conflict adaptation Xiangpeng Wang a, Ting Wang a,b, Zhencai Chen c, Glenn Hitchman a, Yijun Liu a, Antao Chen a,n a

Key Laboratory of Cognition and Personality of Ministry of Education, Faculty of Psychology, Southwest University, Chongqing 400715, China Laboratory of Cognition and Mental Health, Chongqing University of Arts and Sciences, Chongqing 402160, China c State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing 100875, China b

art ic l e i nf o

a b s t r a c t

Article history: Received 30 May 2014 Received in revised form 18 February 2015 Accepted 21 February 2015 Available online 23 February 2015

Individuals differ in the ability to utilize previous conflict information to optimize current conflict resolution, which is termed the conflict adaptation effect. Previous studies have linked individual differences in conflict adaptation to distinct brain regions. However, the network-based neural mechanisms subserving the individual differences of the conflict adaptation effect have not been studied. The present study employed a psychophysiological interaction (PPI) analysis with a color-naming Stroop task to examine this issue. The main results were as follows: (1) the anterior cingulate cortex (ACC)-seeded PPI revealed the involvement of the salience network (SN) in conflict adaptation, while the posterior parietal cortex (PPC)-seeded PPI revealed the engagement of the central executive network (CEN). (2) Participants with high conflict adaptation effect showed higher intra-CEN connectivity and lower intra-SN connectivity; while those with low conflict adaptation effect showed higher intra-SN connectivity and lower intra-CEN connectivity. (3) The PPC-centered intra-CEN connectivity positively predicted the conflict adaptation effect; while the ACC-centered intra-SN connectivity had a negative correlation with this effect. In conclusion, our data demonstrated that conflict adaptation is likely supported by the CEN and the SN, providing a new perspective on studying individual differences in conflict adaptation on the basis of large-scale networks. & 2015 Elsevier Ltd. All rights reserved.

Keywords: Conflict adaptation effect Individual differences Central executive network (CEN) Salience network (SN) Proactive control Psychophysiological interaction (PPI)

1. Introduction Individuals differ in their ability to adjust ongoing behaviors to adapt to changes in the environment. In the laboratory, the dynamic adjustment of behavior is often captured by the ‘conflict adaptation effect’ in tasks with congruent and incongruent conditions. For instance, in the Stroop task (Stroop, 1935), participants are required to respond to the ink color of a printed color name while ignore the word's meaning that may either be congruent with the ink color (congruent trial, C; e.g., RED printed in red) or incongruent with the ink color (incongruent trial, I; e.g., RED printed in blue). Typically, the interference effect (I–C) is reduced after incongruent trials as compared to after congruent trials, which reflects the conflict-induced adjustment (Egner and Hirsch, 2005b; Gratton et al., 1992). Although it has long been assumed that conflict adaptation is n

Corresponding author. Fax: þ 86 23 68253629. E-mail address: [email protected] (A. Chen).

http://dx.doi.org/10.1016/j.neuropsychologia.2015.02.031 0028-3932/& 2015 Elsevier Ltd. All rights reserved.

mediated by different brain areas (Botvinick et al., 2001; Egner et al., 2007), recent studies support the view that this effect might result from the dynamic interactions of distributed brain areas. In particular, the conflict monitoring model speculates that the anterior cingulate cortex (ACC) detects the occurrence of conflict and then conveys this information to the dorsolateral prefrontal cortex (DLPFC) to implement an optimal conflict resolution (Kerns et al., 2004; Matsumoto and Tanaka, 2004). However, direct functional integration between the ACC and DLPFC has not been examined yet. Using a variant of the Stroop task, Egner and Hirsch (2005a) observed the functional coupling between the DLPFC and the fusiform face area, demonstrating that cognitive control mechanisms resolve the conflict through cortical amplification of task-relevant information. Furthermore, one of our previous studies suggests that conflict adaptation is implemented by modulating the effective connectivity between parietal and rightfrontal regions (Tang et al., 2013). This converging evidence implies that conflict adaptation is achieved through the networkbased neural interactions. Nevertheless, only a few studies to date have attempted to

178

X. Wang et al. / Neuropsychologia 70 (2015) 177–184

identify the neural mechanisms underlying individual differences in conflict adaptation. With the employment of resting-state functional magnetic resonance imaging (fMRI), Wang et al. (2014) observed a significant positive correlation between the conflict adaptation effect and the left dorsolateral prefrontal cortex (DLPFC) regional homogeneity (REHO) values, thus highlighting the crucial role of the DLPFC in conflict adaptation. On the other hand, Egner (2011) reported that individual differences in conflict adaptation might be mediated by the right ventrolateral prefrontal cortex (VLPFC). Still, no study has ever examined the networkbased neural mechanisms of individual differences in conflict adaptation. Therefore, the present study aimed to unravel the brain networks subserving individual differences in conflict adaption. In the present study, we employed an event-related fMRI design with the color-naming Stroop task. First, we examined regional activation during the processing of conflict adaptation in the whole subject group. Then we divided the subjects into two groups (high conflict adaptation effect (hereafter, high-score) group vs. low conflict adaptation effect (hereafter, low-score) group) based on the median value of behavioral scores and compared the between-group differences in functional connectivity patterns. The task-dependent functional connectivity in the present study was performed using the voxel-vise psychophysiological interaction (PPI) analysis (Egner and Hirsch, 2005b). PPI is a regression-based method for measuring functional connectivity that allows the detection of regionally specific responses in one brain area in response to inputs from another brain region (Friston et al., 1997). Since PPI provides a comprehensive characterization of task-dependent connectivity between brain regions, this approach offers novel insights into how separated regions dynamically change their connectivity in support of cognitive control. We expected to observe the activation of several regions in the prefrontal and parietal lobes, such as the ACC, the DLPFC, the VLPFC, and the posterior parietal cortex (PPC) in response to conflict adaptation, all of which have been demonstrated to play important roles in the conflict adaptation process (Kerns et al., 2004; Mansouri et al., 2009; Soutschek et al., 2013). Furthermore, existing studies suggest that these regions could also be assigned into different control systems: one is to the proactive top-down control system, including the DLPFC (Braver et al., 2007) and the PPC (Yoon et al., 2008), and the other is to the reactive bottom-up control system, involving the ACC (Braver et al., 2009) and the insula (Menon and Uddin, 2010). The tradeoffs between proactive control and reactive control would determine the final performance. Therefore, conflict adaptation performance would result from the relative involvement of proactive control and reactive control. Accordingly, we expected that the high-score group would show higher engagement of the proactive control network and lower involvement of the reactive control network, while the lowscore group would show higher engagement of the reactive control network and lower involvement of the proactive control network.

obtained from each subject prior to the experiment. 2.2. Stimuli Stimuli were standard Stroop color words, consisting of four Chinese characters “Hong” (red), “Huang” (yellow), “Lan” (blue) and “Lv” (green). Each character was presented in one of the four colors (i.e., red (255, 0, 0), yellow (255, 255, 0), green (0, 255, 0), blue (0, 0, 255); 16 stimuli altogether). The display background was always black. Each color corresponded to a response button, the mappings of which were counterbalanced across participants. For example, the participant was instructed to respond with the index finger of their left hand when the color of the word was green, the middle finger of their left hand when the color of the word was red, the index finger of their right hand when the color of the word was yellow, and the middle finger of their right hand when the color of the word was blue. 2.3. fMRI task Participants performed the color-naming task during one fMRI session. There were a total of 212 trials across four runs. Each trial lasted for 3000 ms, starting with a 1500 ms central fixation cross, followed by a 1500 ms target display, during which participants were required to respond to the colors of the stimuli as fast and accurately as possible. The trials were sequenced in a pseudorandom way in order to result in an equal proportion of each type of stimulus. In addition, there was no direct feature integration (stimulus–stimulus repetition or stimulus–response repetition) (Mayr et al., 2003). Before participants performed the main experiment, they took part in a 40-trial task for practice. In addition, we used an event-related design with a constant ITI. One may argue that a jittered ITI might be more optimal for an fMRI design, and there are studies using a jittered ITI in the field of conflict control (Egner and Hirsch, 2005a; Liston et al., 2006). The reasons that we did not choose a jittered ITI were as follows: the first one was that the magnitude of the conflict adaptation effect would be affected by increasing ITI (Peter, 2005); the second reason was that a jittered ITI is likely to interact with cognitive control processes in unknown ways (Kerns, 2006; Kerns et al., 2004). Thus, we chose the constant ITI in our design to ensure the purity of conflict adaptation effect. 2.4. fMRI image acquisition Images were acquired with a Siemens Trio 3.0T scanner in the Laboratory of Cognition and Personality in Southwest University, China. Functional data were acquired in an interlaced way along the AC–PC line with a T2-weighted EPI sequence of 24 axial slices (TR ¼1500 ms, TE ¼30 ms, flip angle ¼ 90°, acquisition matrix¼ 64  64) of 5 mm thickness with 1 mm inter-slice gap. Within a session, a total of 644 EPI images were acquired. At the end of the experiment, a T1-weighted spin echo data set (TR ¼1900 ms, TE ¼ 2.52 ms, flip angle ¼90°, acquisition matrix¼ 256  256) was acquired.

2. Materials and methods 2.5. Imaging data analysis 2.1. Subjects The experimental procedure was approved by the local ethics committee. Thirty-one university students (23 women and 8 men) from Southwest University, China, participated in the experiment for payment. The age of participants ranged from 18 to 24 years (20.77 7 1.37). All participants were right-handed, native Chinese speakers, and had normal or corrected-to-normal vision, without achromatopsia or color blindness. Signed informed consent was

2.5.1. fMRI data preprocessing All images were analyzed using SPM8 (the Wellcome Trust Centre for Neuroimaging, University College London, UK; http:// www.fil.ion.ucl.ac.uk/spm/software/spm8/). The first five volumes of each run were excluded from the analysis to allow for signal stability following onset transients. Functional images were corrected for differences in slice-timing, realigned, and co-registered with the structural images. Then these images were normalized to

X. Wang et al. / Neuropsychologia 70 (2015) 177–184

the MNI template brain (voxel size: 3  3  3 mm3) and smoothed with a Gaussian kernel of 6  6  6 mm3 full-width, half-maximal. 2.5.2. Statistical analyses The variance in the BOLD signal was decomposed in a general linear model. Separate regressors were coded according to the 4 combinations of previous trial type (C, I) and current trial type (C, I) as well as excluded trials (trials excluded in behavioral analyses). For each regressor, onsets were created and then convolved with the canonical hemodynamic response function (HRF). The duration of event in the model was set to 0. A default high-pass filter of 1/128 Hz was applied to the data and the model. Model parameters were estimated using Restricted Maximum Likelihood. Effects of interest were assessed in a random-effects analysis. Contrast estimates for each of the combinations were entered into a full factorial design using non-sphericity correction. Within this design, we used the previous congruency and the current congruency as factors, each consisting of two levels. We were therefore able to obtain the main effects of all factors and their interaction effects. 2.5.3. Psychophysiological interaction (PPI) analyses In the current study, PPI analysis employs one regressor representing the conflict adaptation effect (the psychological variable), one regressor representing the time course in a given volume of interest (VOI) (the physiological variable), and a third

179

regressor representing the cross-product of the previous two (the psychophysiological interaction term). Because conflict adaptation is calculated as RT(c(I–C)–i(I–C)) (Egner, 2011), we created a contrast of (  1 1 1  1) to represent the four conditions (CC, CI, IC, II) correspondingly. Based on the regional activation results and the literature, we defined the ACC, the VLPFC and the PPC as seed regions.VOIs' time courses were extracted from the ACC (x¼3, y ¼21, z ¼39), the VLPFC (x ¼51, y¼15, z ¼12) and the PPC (x¼  9, y¼  75, z ¼39) (6 mm radius sphere at the local peak). Then the psychophysiological interaction term was created with the time course and the psychological variable. PPI analyses were then carried out for each VOI in each subject and a followed group-level one-sample t-test was conducted to examine the significant functional connectivity. Considering individual differences (Fig. 1 C), we divided the 31 subjects into two groups: 15 subjects whose conflict adaptation effects scores were below the median ( o21.32 ms, low-score group), and 16 subjects whose effects scores were at or above the median (Z21.32 ms, high-score group). The difference in behavioral performances between these two groups was significant, t29 ¼7.34, po 0.001 (Fig. 1D). Note, however, that they did not differ in terms of the Stroop effect RT (I–C) (t29 ¼  0.334, p 40.1), nor the learning effects over time (t29 ¼ 0.79, p 40.1). In addition, there were no between-group differences in age (t29 ¼  1.01, p4 0.1) or gender (t29 ¼  0.88, p ¼0.389). Finally, a two-sample ttest was performed to calculate the between-group differences.

Fig. 1. Behavioral conflict adaptation effects in Stroop task. Panel A and B represent the average reaction time (RT) and accurate rate for each condition, respectively. Note: c ¼ congruent; i ¼incongruent. Panel C indicates the individual differences related to conflict adaptation effect. Conflict adaptation was calculated as (RT(CI–CC)–RT(II–IC)). Note: Each circle represents a subject’s conflict adaptation score, and the line represents the median value. Panel D displays the mean values of the conflict adaptation effects of the high-score (left) and the low-score group (right). The high-score group and the low-score group were divided by the median value of the conflict adaptation effect in the whole group.

180

X. Wang et al. / Neuropsychologia 70 (2015) 177–184

3. Results

Table 2 Brain regions susceptible to previous  current congruency interaction effect.

3.1. Behavioral results

Side Cluster size MNI coordinate

The first trial of each block, error trials, the post-error trials, and trials with RTs more than 3 SDs from the mean were excluded from the RT analyses. A 2(previous congruency: C, I)  2(current congruency: C, I) repeated measures ANOVA was conducted on these data. RT data are depicted in Fig. 1 A. The main effects of previous congruency (F1, 30 ¼5.81, p o0.05) and current congruency (F1, 30 ¼188.50, p o0.001) were significant. Importantly, their interaction effect was also significant (F1, 30 ¼ 7.88, p o0.01), indicating a significant conflict adaptation effect. The accuracy results are presented in Fig. 1B. The main effect of current congruency (F1, 30 ¼10.85, p o0.01) was significant but the main effect of previous congruency was not. In addition, there emerged a significant interaction effect (F1, 30 ¼6.19, p ¼0.019). Furthermore, there were no learning effects in task performance over time (F1, 30 ¼0.71, p ¼0.19). 3.2. Regional brain activations during conflict adaptation Group analysis of the main effect of previous congruency revealed significant activation in the left inferior frontal gyrus (IFG), the right primary motor cortex (PMC), and the left medial frontal gyrus (MFG) (p o0.01, FWE corrected). Brain regions that showed the significant activation on the main effect of current trial type were the bilateral middle frontal gyrus, the left ventrolateral prefrontal cortex (VLPFC), the left superior frontal gyrus, the ACC, the left occipital lobe and the right posterior cerebellum lobe (p o0.01, FWE corrected). The brain regions which were responsible for the main effects of current and/or previous congruency are summarized in Table 1. Very importantly, as listed in Table 2, a number of clusters revealed significant previous  current congruency interaction effects (i.e., conflict adaptation effects): voxels in the left MFG, the bilateral PMC, the right ventrolateral prefrontal cortex (VLPFC), the ACC and the PPC, as well as some subcortical brain regions like the left insula, the bilateral thalamus and the cerebellum (p o0.05, AlphaSim corrected), see also Fig. 2 A. To investigate whether the two groups differ in conflictTable 1 Brain regions susceptible to main effects of current congruency and/or previous congruency. Side Cluster size MNI coordinate

Effect of previous congruency Inferior frontal gyrus L Precentral gyrus R Medial frontal gyrus L Sub-gyral L

Effect of current congruency Middle frontal gyrus L Middle frontal gyrus R Inferior frontal gyrus R Superior frontal L gyrus Extra-nuclear R Cingulate gyrus L Sub-gyral R Cerebellum posterR ior lobe Occipital lobe L

Peak intensity

x

y

54 29 47 37

 42 33 6  42

6  24 12 27

30 60 57 18

44.58 35.72 43.87 37.43

1256 27 29 109

 48 51 42 6

9 30 9 12

33 24 33 54

79.49 40.58 41.78 62.35

25 20 54 17

12 9 30 27

3 12 30  69

3 30 6  45

44.70 35.91 39.59 39.55

28

 45

 60

9

42.76

Note: The thresholding was po 0.01 (FWE corrected).

z

x Medial frontal gyrus Precentral gyrus Precentral gyrus Inferior frontal gyrus Insula Insula Thalamus Thalamus Cingulate gyrus Cingulate gyrus Cingulate gyrus Precuneus Precuneus Cerebellum posterior lobe Cerebellum anterior lobe

y

Peak intensity z

L L R R L L L R R L R L L L

22 170 43 119 15 21 24 29 24 23 15 19 18 16

3  36 36 51  39  33  18 24 12 15 3  18 9  39

12 51  24 57  24 57 15 12 18 3 3 9  27 9  27 6  39 27 21 30 21 39  57 39  75 39  69  30

14.24 34.73 23.81 25.83 15.28 22.60 14.44 16.09 16.59 18.47 15.24 15.29 14.83 14.63

R

28

18

 66  33

19.07

Note: The thresholding was p o 0.05 (AlphaSim corrected).

adaptation activation, a further 2(previous congruency)  2(current congruency)  2(group) ANOVA analysis was performed. At the uncorrected level of p o0.005, k ¼30, no cluster that showed significant between-group differences was detected. 3.3. Psychophysiological interaction analyses For all participants, the one-sample t-test showed that during conflict adaptation, the ACC was positively connected to the bilateral insula, the right anterior frontal cortex and the cerebellum, see also in Fig. 3 A; whereas, the PPC was positively connected to the left anterior frontal cortex, the left middle frontal gyrus and the left parietal lobe, for details, see Fig. 3B. Note that no negative functional connectivity of the ACC or the PPC was detected. However, the VLPFC only showed a significant positive connectivity with the middle occipital gyrus. All these results were corrected at p o0.05 by AlphaSim. As displayed in Fig. 4, the two-sample t-test revealed different connectivity patterns between these two groups. For the ACC-seed analysis, in the high-score group, the ACC showed weaker functional connectivity with the right insula and the superior occipital gyrus relative to the low-score group (po 0.05, AlphaSim corrected). At the uncorrected level of p o0.005, k¼ 10, the left insula also showed a greater connectivity with the ACC in the low-score group. No voxel in the high-score group was detected showing greater functional connectivity with the ACC (for details, see Fig. 4). For the PPC-seed analysis, the PPC in the high-score group showed greater functional connectivity with the superior frontal gyrus (SFG), the bilateral superior parietal lobule (SPL), the bilateral cerebellum posterior lobe and some voxels in the right temporal lobe (p o0.05, AlphaSim corrected). No voxel in the high-score group that showed weaker functional connectivity with the PPC was detected. This result is displayed in Fig. 5. In addition, the VLPFC-seed PPI analysis did not show any results concerning the between-group differences at the significant level of p o0.005, k¼ 30. To confirm our results based on the median split, we performed an additional analysis of PPI-based connectivity as a function of conflict adaptation scores. We extracted the beta values of the PPI connectivity between the seed regions and these that showed significant between group differences, and then correlated them with the conflict adaptation scores. As displayed in Fig. 5, the PPI connectivity between the ACC and the regions (the bilateral insula,

X. Wang et al. / Neuropsychologia 70 (2015) 177–184

181

Fig. 2. Regional brain activations during conflict adaptation. Panel A displays voxels activated during conflict adaptation process. All these results were corrected at po 0.05 (AlphaSim corrected). The color represents the F-value, ranges from 0 to 34. Panel B displays the average percent signal change for each condition in the anterior cingulate cortex (ACC) and Panel C displays the average percent signal change for each condition in the left posterior parietal cortex (PPC). Note: error bars represent the standard errors (SE).

Fig. 3. Functional connectivity patterns of the ACC and the PPC in conflict adaptation. Panel A displays the one-sample t-test results of the ACC-seed PPI analysis, and Panel B displays the results of the PPC-seed PPI analysis. The significant level is set at p o0.05, AlphaSim corrected. Color bar shows a scale of the t values. The color represents positive functional connectivity.

182

X. Wang et al. / Neuropsychologia 70 (2015) 177–184

Fig. 4. Individual differences in the ACC-seed PPI connectivity patterns during conflict adaptation. The colored regions refer to voxels that showed significant functional connectivity with the ACC (high-score group4low-score group). Color bar shows a scale of the t values. The line charts represent the correlations between the behavioral conflict adaptation scores and the PPI connectivity values. ACC-occipital represents the connectivity between the ACC and the occipital lobe, and ACC-insula_r represents the connectivity between the ACC and the right insula. The letter “r” refers to right. R2 represents the coefficient of determination and ranges from 0 to 1.

Fig. 5. Individual differences in the PPC-seed PPI connectivity patterns during conflict adaptation. The colored regions refer to voxels that showed significant functional connectivity with the PPC (high-score group 4low-score group). Color bar shows a scale of the t values. The line charts represent the correlations between the behavioral conflict adaptation scores and the PPI connectivity values. SFG refers to the superior frontal cortex, SPL refers to superior parietal lobule, and the cere refers to cerebellum. PPC-SFG represents the functional connectivity between the PPC and the SFG. R2 represents the coefficient of determination and ranges from 0 to 1.

X. Wang et al. / Neuropsychologia 70 (2015) 177–184

the occipital lobe) could negatively predict behavioral conflict adaptation, while the PPI connectivity between the PPC and the regions (SFG, the bilateral SPL, and the bilateral cerebellum posterior lobe) could positively predict behavioral conflict adaptation. All these results reached a significance level of p o0.05.

4. Discussion Employing a color-naming Stroop task, the present study aimed to explore the individual differences in conflict adaptation and to identify the underlying neural mechanisms. At the behavioral level, we obtained a significant conflict adaptation effect after controlling repetition priming effects. Two groups, which differed in behavioral performance, were identified by a median split based on conflict-adaptation scores. In the fMRI data, we observed the activations of the ACC, the VLPFC, and the PPC in conflict adaptation processing. The PPI results indicated that during conflict adaptation, the ACC and the PPC showed different connectivity patterns. Meanwhile, individual difference analyses revealed that compared to the low-score group, the high-score group had the PPC showing relatively greater connectivity with the left superior parietal lobule (SPL), the dorsal prefrontal gyrus, and the cerebellum. In contrast, the ACC in the low-score group showed greater connectivity with the insula and the superior occipital gyrus relative to the high-score group. However, the VLPFC-seed PPI analysis did not show any significant effect. Moreover, the correlation analysis between the connectivity patterns and the behavioral scores also supported the results based on the median split. As expected, we observed activation of the ACC and the PPC in conflict adaptation. In the present study, the ACC could only be evoked by conflict-related conditions and its activation was higher in CI trials than in II trials. As suggested by Egner and Raz (2007), CI trials represent high conflict due to a low level of cognitive control, while II trials represent low conflict due to a high level of cognitive control. Thus, the present results are in accordance with the previous studies highlighting the function of the ACC in conflict monitoring (Botvinick et al., 1999; Clayson and Larson, 2011). Interestingly, the activation of the PPC was also correlated with conflict adaptation. Although the engagement of the PPC in conflict adaptation has been reported by several studies (Egner et al., 2007; Mansouri et al., 2009), the most direct evidence may be the observation of reduced conflict adaptation when stimulating the PPC using transcranial magnetic stimulation (TMS) (Soutschek et al., 2013). Moreover, the PPC might contribute to conflict adaptation by influencing the processing of response selection or identifying task-relevant information (Bunge et al., 2002; Soutschek et al., 2013). Note that, the present study did not detect activation of the DLPFC. However, this does not mean that the DLPFC is not engaged in this process, for that the PPI results showed that functional connectivity between the PPC and the DLPFC significantly supported conflict adaptation. Despite the fact that the ACC and PPC could be co-activated during multiple cognitive tasks (Dosenbach et al., 2006; Hwang et al., 2010), their PPI results indicated distinct connectivity patterns. While the ACC was connected to the bilateral insula and the anterior prefrontal cortex, the PPC was connected to the SPL and the DLPFC. These dissociable results were consistent with the segregation of two cognitive control systems: the fronto-parietal central executive network (CEN) and the cingulo-opercular salience network (SN) (Bressler and Menon, 2010; Seeley et al., 2007). The CEN, consisting of the DLPFC and PPC, engages in high-level cognitive functions and is thought to be involved in adaptive cognitive control (Dosenbach et al., 2008; Power and Petersen, 2013). The SN, anchored by the right insula and the ACC, supports

183

the maintenance and implementation of task set across trials (Dosenbach et al., 2006) and/or is responsible for detecting and orienting to salient external stimuli and internal events (Menon and Uddin, 2010; Seeley et al., 2007). Thereby, the present study demonstrates that conflict adaptation is achieved through the cooperation of two separated networks, supporting the dual-network model (Dosenbach et al., 2008). Results from individual difference analyses provided further support that the two networks carried out dissociable functions in conflict adaptation. The high-score group showed relatively greater intra-CEN connectivity and lower intra-SN connectivity, while the low-score group showed lower intra-CEN connectivity and greater intra-SN connectivity. In addition, the PPC-centered intra-CEN connectivity was positively correlated with the conflict adaptation effect, while the ACC-centered intra-SN connectivity had a negative correlation with this effect. Speculatively, participants who recruited the CEN more and the SN less during the Stroop task would score higher in conflict adaptation; whilst, those who recruited the SN more and the CEN less would score lower in conflict adaptation. The higher involvement of the CEN in conflict adaptation is consistent with our previous findings that the intrinsic DLPFC activity and the amplitude of the parietal sustained potential could be linked to individual differences in conflict adaptation (Tang et al., 2013; Wang et al., 2014). The CEN seems to actively maintain task-relevant information about one or a small number of trials and to implement control parameter adjustments rapidly (Dosenbach et al., 2008, 2007). While the SN maps aberrant stimuli and initiates control signals to facilitate access to cognitive resources needed for goal-directed action (Menon, 2011). Therefore, with the recruitment of the CEN, the high-score group may adopt a top-down control manner (Corbetta and Shulman, 2002; Elton and Gao, 2014). In contrast, the lowscore group may implement a SN-centered bottom-up control (Braver et al., 2009; Menon and Uddin, 2010). In terms of the dual control model (Braver, 2012; Braver et al., 2007), top-down control is similar to proactive control, and bottom-up control is like reactive control. Interestingly, existing evidence indicates that proactive control is associated with the LPFC and the parietal cortex (Boulinguez et al., 2009; Braver et al., 2007), and reactive control is related to the ACC and probably the insula (Braver et al., 2009; Menon and Uddin, 2010). Therefore, the high-score group may implement proactive control during the Stroop task, which allows them to fully utilize previous conflict signals to improve current conflict resolution, manifesting a high conflict adaptation effect in their performance. Meanwhile, the low-score group may engage reactive control, wherein previous conflict signals hardly influence current conflict resolution, which may account for the low conflict adaptation in this group. In the present study, the contrast of “  CCþ CI þIC–II” was used to perform the PPI analyses. One may argue that this contrast might not reflect conflict adaptation, but lead to results representing changes in connectivity for trial-type alternations (CI and IC) vs. repetitions (II and CC). However, the conflict adaptation is defined as an effect that conflict resolution is more efficient when congruency levels/trial types repeat than alternate. Consequently, this contrast used in the PPI analyses was appropriate to reflect conflict adaptation. However, one limitation of the present approach was that the voxel-vise PPI analysis can only explore the instantaneous intra-network connectivity patterns, and it could not provide a direct examination about the relationships between networks. Therefore, although the present results suggested that the CEN and the SN played different roles in conflict adaptation, we could not rule out the possibility that these two networks may interact in some ways. In fact, previous studies have suggested the ACC could function to detect the occurrence of conflict and then recruit the LPFC (and/or the PPC) to minimize conflict (Kerns

184

X. Wang et al. / Neuropsychologia 70 (2015) 177–184

2006; Kerns et al., 2004). We expect that, in the future, alternative approaches (e.g., multi-regional PPI analysis, independent component analysis, or dynamic causal model analysis) will be used and offer further data to address the relationships between these two networks. To summarize, the present study revealed that the CEN and the SN are key networks in conflict adaptation processing, and their dissociable functions could interpret individual differences in this phenomenon. In particular, the SN initiates reactive bottom-up control to detect the occurrence of conflict, while the CEN exerts proactive top-down control to optimize conflict resolution. Therefore, our data revealed the network-based mechanisms underpinning conflict adaptation and the individual differences in those mechanisms.

Acknowledgments We thank Ming Li (University of Nebraska-Lincoln) for his patient help for proofreading our manuscript. This work was supported by grants from the National Natural Science Foundation of China (31170980, 81271477), the Foundation for the Author of National Excellent Doctoral Dissertation of the People's Republic of China, China (201107), the Program for New Century Excellent Talents in University, China (NCET-11-0698), and the Fundamental Research Funds for the Central Universities (SWU1009001, SWU1309351).

References Botvinick, M., Nystrom, L.E., Fissell, K., Carter, C.S., Cohen, J.D., 1999. Conflict monitoring versus selection-for-action in anterior cingulate cortex. Nature 402, 179–181. Botvinick, M.M., Braver, T.S., Barch, D.M., Carter, C.S., Cohen, J.D., 2001. Conflict monitoring and cognitive control. Psychol. Rev. 108, 624–652. Boulinguez, P., Ballanger, B., Granjon, L., Benraiss, A., 2009. The paradoxical effect of warning on reaction time: demonstrating proactive response inhibition with event-related potentials. Clin. Neurophysiol. 120, 730–737. Braver, T.S., 2012. The variable nature of cognitive control: a dual mechanisms framework. Trends Cogn. Sci. 16, 106–113. Braver, T.S., Gray, J.R., Burgess, G.C., 2007. Explaining the many varieties of working memory variation: Dual mechanisms of cognitive control. Var. Work. Mem., 76–106. Braver, T.S., Paxton, J.L., Locke, H.S., Barch, D.M., 2009. Flexible neural mechanisms of cognitive control within human prefrontal cortex. Proc. Natl. Acad. Sci. USA 106, 7351–7356. Bressler, S.L., Menon, V., 2010. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn. Sci. 14, 277–290. Bunge, S.A., Hazeltine, E., Scanlon, M.D., Rosen, A.C., Gabrieli, J., 2002. Dissociable contributions of prefrontal and parietal cortices to response selection. NeuroImage 17, 1562–1571. Clayson, P.E., Larson, M.J., 2011. Conflict adaptation and sequential trial effects: support for the conflict monitoring theory. Neuropsychologia 49, 1953–1961. Corbetta, M., Shulman, G.L., 2002. Control of goal-directed and stimulus-driven attention in the brain. Nat. Rev. Neurosci. 3, 201–215. Dosenbach, N.U., Fair, D.A., Cohen, A.L., Schlaggar, B.L., Petersen, S.E., 2008. A dualnetworks architecture of top-down control. Trends Cogn. Sci. 12, 99–105. Dosenbach, N.U., Fair, D.A., Miezin, F.M., Cohen, A.L., Wenger, K.K., Dosenbach, R.A.,

Fox, M.D., Snyder, A.Z., Vincent, J.L., Raichle, M.E., 2007. Distinct brain networks for adaptive and stable task control in humans. Proc. Natl. Acad. Sci. USA 104, 11073–11078. Dosenbach, N.U., Visscher, K.M., Palmer, E.D., Miezin, F.M., Wenger, K.K., Kang, H.C., Burgund, E.D., Grimes, A.L., Schlaggar, B.L., Petersen, S.E., 2006. A core system for the implementation of task sets. Neuron 50, 799–812. Egner, T., 2011. Right ventrolateral prefrontal cortex mediates individual differences in conflict-driven cognitive control. J. Cogn. Neurosci. 23, 3903–3913. Egner, T., Delano, M., Hirsch, J., 2007. Separate conflict-specific cognitive control mechanisms in the human brain. NeuroImage 35, 940–948. Egner, T., Hirsch, J., 2005a. Cognitive control mechanisms resolve conflict through cortical amplification of task-relevant information. Nat. Neurosci. 8, 1784–1790. Egner, T., Hirsch, J., 2005b. The neural correlates and functional integration of cognitive control in a Stroop task. NeuroImage 24, 539–547. Egner, T., Raz, A., 2007. Cognitive control processes and hypnosis. Hypn. Conscious States: Cogn. Neurosci. Perspect., 29–50. Elton, A., Gao, W., 2014. Divergent task-dependent functional connectivity of executive control and salience networks. Cortex 51, 56–66. Friston, K., Buechel, C., Fink, G., Morris, J., Rolls, E., Dolan, R., 1997. Psychophysiological and modulatory interactions in neuroimaging. NeuroImage 6, 218–229. Gratton, G., Coles, M.G., Donchin, E., 1992. Optimizing the use of information: strategic control of activation of responses. J. Exp. Psychol. Gen. 121, 480. Hwang, K., Velanova, K., Luna, B., 2010. Strengthening of top-down frontal cognitive control networks underlying the development of inhibitory control: a functional magnetic resonance imaging effective connectivity study. J. Neurosci. 30, 15535–15545. Kerns, J.G., 2006. Anterior cingulate and prefrontal cortex activity in an FMRI study of trial-to-trial adjustments on the Simon task. NeuroImage 33, 399–405. Kerns, J.G., Cohen, J.D., MacDonald, A.W., Cho, R.Y., Stenger, V.A., Carter, C.S., 2004. Anterior cingulate conflict monitoring and adjustments in control. Science 303, 1023–1026. Liston, C., Matalon, S., Hare, T.A., Davidson, M.C., Casey, B.J., 2006. Anterior cingulate and posterior parietal cortices are sensitive to dissociable forms of conflict in a task-switching paradigm. Neuron 50, 643–653. Mansouri, F.A., Tanaka, K., Buckley, M.J., 2009. Conflict-induced behavioural adjustment: a clue to the executive functions of the prefrontal cortex. Nat. Rev. Neurosci. 10, 141–152. Matsumoto, K., Tanaka, K., 2004. Neuroscience. Conflict and cognitive control. Science 303, 969–970. Mayr, U., Awh, E., Laurey, P., 2003. Conflict adaptation effects in the absence of executive control. Nat. Neurosci. 6, 450–452. Menon, V., Uddin, L.Q., 2010. Saliency, switching, attention and control: a network model of insula function. Brain Struct. Funct. 214, 655–667. Menon, V., 2011. Large-scale brain networks and psychopathology: a unifying triple network model. Trends Cogn. Sci. 15, 483–506. Peter, W., 2005. Evidence for gating of direct response activation in the Simon task. Psychon. Bull. Rev. 12, 282–288. Power, J.D., Petersen, S.E., 2013. Control-related systems in the human brain. Curr. Opin. Neurobiol. 23, 223–228. Seeley, W.W., Menon, V., Schatzberg, A.F., Keller, J., Glover, G.H., Kenna, H., Reiss, A. L., Greicius, M.D., 2007. Dissociable intrinsic connectivity networks for salience processing and executive control. J. Neurosci. 27, 2349–2356. Soutschek, A., Taylor, P.C., Müller, H.J., Schubert, T., 2013. Dissociable networks control conflict during perception and response selection: a transcranial magnetic stimulation study. J. Neurosci. 33, 5647–5654. Stroop, J.R., 1935. Studies of interference in serial verbal reactions. J. Exp. Psychol. 18, 643–662. Tang, D., Hu, L., Li, H., Zhang, Q., Chen, A., 2013. The neural dynamics of conflict adaptation within a look-to-do transition. PLoS One 8, e57912. Wang, T., Chen, Z., Zhao, G., Hitchman, G., Liu, C., Zhao, X., Liu, Y., Chen, A., 2014. Linking inter-individual differences in the conflict adaptation effect to spontaneous brain activity. NeuroImage 90, 146–152. Yoon, J., Minzenberg, M., Ursu, S., Walters, R., Wendelken, C., Ragland, J., Carter, C., 2008. Association of dorsolateral prefrontal cortex dysfunction with disrupted coordinated brain activity in schizophrenia: relationship with impaired cognition, behavioral disorganization, and global function. Am. J. Psychiatry 165, 1006–1014.