Emotion 2014, Vol. 14, No. 2, 321–330

© 2013 American Psychological Association 1528-3542/14/$12.00 DOI: 10.1037/a0035208

Conflict Adaptation in Emotional Task Underlies the Amplification of Target Natalia Chechko, Thilo Kellermann, Frank Schneider, and Ute Habel

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

RWTH Aachen University and JARA—Translational Brain Medicine, Aachen, Germany A primary function of cognitive control is to adjust the cognitive system according to situational demands. The so-called “conflict adaptation effect” elicited in laboratory experiments is supposed to reflect the above function. Neuroimaging studies suggest that adaptation of nonemotional conflict is mediated by the dorsolateral prefrontal cortex through a top-down enhancement of task-relevant (target), relative to task-irrelevant (distractor), stimulus representation in the sensory cortices. The adaptation of emotional conflict, on the other hand, is suggested to be related to the rostral anterior cingulate inhibiting the processing of emotional distractors through a top-down modulation of amygdala responsivity. In the present study, we manipulated, on a trial-by-trial basis, the levels of semantic interference conflict triggered by the incompatibility between emotional faces (targets) and emotional words (distractors) in a modified version of the emotional Stroop task. Similar to previous observations involving nonemotional interference effects, the behavioral adaptation of emotional conflict was found to be paralleled by a stronger recruitment of the fusiform face area. Additional areas related to the conflict adaptation effect were the bilateral insula, the bilateral frontal operculum (fO), the right amygdala, the left precentral and postcentral gyri, and the parietal cortex. These findings suggest that augmentation of cortical responses to task-relevant information in emotional conflict may be related to conflict adaptation processes in a way that has been observed in nonemotional conflict, challenging the view that brain circuitries underlying the conflict adaptation effect depend only on the nature of conflict. Keywords: emotional interference conflict, conflict adaptation, fusiform face area, Stroop, functional magnetic resonance imaging

Neuroimaging studies of brain circuitries underlying the conflict adaptation effect have suggested that these circuitries may vary depending on the nature of conflict (Egner & Hirsch, 2005; Egner, 2008; Etkin, Egner, Peraza, Kandel, & Hirsch, 2006). In a nonemotional analogue of the Stroop task, with participants requiring to respond to the faces of famous personalities (actors or politicians) while suppressing incongruent words (e.g., an actor’s face accompanied by a politician’s name or vice versa), behavioral conflict adaptation was paralleled by a stronger involvement of the fusiform face area (FFA). The functional coping mechanisms between the dorsolateral prefrontal cortex (DLPFC) and FFA suggested that the role of DLPFC in conflict adaptation effect is to augment the processing of task-relevant stimuli, inducing a stronger FFA involvement (Egner & Hirsch, 2005). Thus, nonemotional conflict resolution (or conflict adaptation) is thought to be mediated by top-down enhancement of task-relevant (target) relative to task-irrelevant (distractor) stimulus representations in sensory cortices. These findings appeared to support the notion of an attentional target-feature amplification, rather than the inhibition of taskirrelevant information (distractor), being the primary mechanism for interference conflict adaptation through cognitive control (Egner & Hirsch, 2005). However, other research has suggested that those mechanisms are specifically linked to the adaptation of nonemotional interference conflict, and that the adaptation of emotional conflict involves an alternative route. Thus, the emotional conflict adaptation experiment, in which subjects were asked to categorize faces according to their emotional expressions (happy vs. fearful), while trying to ignore emotionally congruent

In interference tasks such as the Stroop task (Stroop, 1992), behavioral improvement during incongruent trials with preceding incongruent stimuli, compared with incongruent trials following congruent stimuli, is known as the conflict adaptation effect, which has been interpreted as reflecting conflict-driven adjustments in cognitive control (Carter & van Veen, 2007). This effect has been exploited to dissociate brain regions involved in the generation of conflict from those involved in the resolution of or control over conflict. This has been done by contrasting neural activity in incongruent trials preceded by a congruent trial (high-conflict, low-control trials) with that in incongruent trials preceded by an incongruent trial (low-conflict, high-control trials; Botvinick, Nystrom, Fissell, Carter, & Cohen, 1999; Carter et al., 2000).

This article was published Online First December 30, 2013. Natalia Chechko, Thilo Kellermann, Frank Schneider, and Ute Habel, Department of Psychiatry, Psychotherapy and Psychosomatics, RWTH Aachen University, Aachen, Germany, and JARA—Translational Brain Medicine, Aachen, Germany. Conceived and designed the experiments: NC, UH. Prepared ethical approval and recruited patients: NC, UH. Performed the experiments: NC. Analyzed the data: NC, TK. Wrote the article: NC. Reviewed the manuscript: UH. Supervised the study: UH, FS. No funding from any external source was obtained for this study. We thank all participants. Correspondence concerning this article should be addressed to Natalia Chechko, Pauwelstrasse 30, 52074 Aachen. E-mail: nchechko@ ukaachen.de 321

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

322

CHECHKO, KELLERMANN, SCHNEIDER, AND HABEL

or incongruent affective words (“HAPPY,” “FEAR”) written across the faces (Etkin et al., 2006), produced trial-to-trial adaptation effects akin to what had been seen in nonemotional experiments (Egner & Hirsch, 2005). However, unlike nonemotional conflict, the emotional interference conflict was linked to an initial increase in amygdala activation as an involuntary response to incompatible emotional presentations. The behavioral conflict adaptation, on the other hand, was paralleled by a context-responsive restraining of amygdala reaction by the rostral part of the anterior cingulate. Furthermore, unlike in nonemotional conflict, the adaptation in emotional conflict was not linked to sensory cortex modulation (Etkin et al., 2006). In our previous research (Chechko et al., 2012), we compared the processing of emotional conflict with that of nonemotional conflict on the level of interference effect (comparison between the incongruent and congruent trials). In both versions of the interference task, we used the same emotional faces as targets with two different assignments— emotion recognition and age judging. In the emotional word-face interference task, a stable behavioral interference effect arose from the semantic incompatibility between task-relevant (recognition of emotional face) and taskirrelevant (reading of emotional word) information as described by Etkin et al., 2006. In the nonemotional word-face interference task, the conflict effect ensued from the semantic incompatibility between task-relevant (judgment of the person’s age) and taskirrelevant information (reading of a word whose meaning does not fit the target’s age category). In both tasks, we observed a common network including the dorsal anterior cingulate, the supplemental motor area, the bilateral insula and the inferior prefrontal cortex, indicating that these brain structures are markers of experienced conflict. In the emotional word-face interference task, the behavioral interference conflict was greater (compared with the nonemotional task) and was associated with a stronger recruitment of the posterodorsal medial frontal cortex, the left inferior frontal cortex (IFC)/anterior insula region and the right FFA. In the emotional task itself, incongruent trials (compared with congruent trials) were associated with stronger involvement of the dACC, the bilateral anterior insula, the bilateral lateral prefrontal cortex (LPFC), the bilateral occipitotemporal visual cortex, and the right FFA. Thus, in response to emotional interference conflict, we observed the involvement of regions (FFA, dACC, bilateral anterior insula, LPFC) that had been suggested to be specifically related to conflict adaptation in nonemotional tasks (Egner & Hirsch, 2005). Given that our previous research was based only on the comparison between incongruent and congruent trials, it was

not possible to determine whether those regions are related to conflict generation or conflict resolution. To address this issue in the present study, we sought to manipulate the level of semantic interference conflict on a trial-by-trial basis and study the conflict adaptation effect, using the emotional paradigm applied before. Contrasting the neural activity during incongruent trials preceded by a congruent trial (high- conflict, low-control trials) with that during incongruent trials preceded by an incongruent trial (lowconflict, high-control trials), we aimed to study the regions involved in conflict generation as opposed to the ones involved in conflict resolution. Given the results of previous research (Egner & Hirsch, 2005), we hypothesized a stronger involvement of the FFA and the lateral prefrontal cortex during conflict resolution (lowconflict, high-control trials) compared with conflict generation (high-conflict, low-control trials). As we used tasks that were based on the semantic interference between two emotional stimuli, we expected the amygdala to be involved in conflict, although without a clear prediction as to whether this involvement would be stronger in response to conflict resolution or conflict generation. If our results were to vindicate our assumption, they would challenge the notion that brain circuitries underlying conflict adaptation effect depend only on the nature of conflict, and suggest that other factors (e.g., processing load) should also be taken into account.

Method Participants The current study was based on a previously recorded data set, some results of which having already been published (Chechko et al., 2012). Twenty-four healthy, right-handed participants (12 women, 12 men), all native speakers of German, took part in the study. Demographic variables are summarized in Table 1 and Table 2. The participants were students from Aachen University, recruited by means of an advertisement. Participants with neurological, psychiatric, or other medical illnesses were excluded. Previous head injuries with loss of consciousness and substance abuse or dependence were the additional exclusion criteria. All participants were screened by an experienced psychiatrist (NC) from the university hospital Aachen. The screening included a short version of the structured clinical interview for Axis I disorder (SCID-I, German version; Wittchen, 1997) and an interview during which the participants’ medical history was recorded. During the latter, participants with Axis II disorders were identified and excluded

Table 1 Group Characteristics and Demographic Data Group characteristics and demographic data

Mean values and SD

Subject number Gender Age (years) Mean education (years) Processing speed (Trail Making Test A: TMT-A, s) Cognitive flexibility (Trail Making Test B: TMT-B, s) Verbal intelligence (Wortschatztest; Vocabulatory test: WST, IQ) Facial recognition (Benton Facial Recognition Test: BFRT–hits, %)

24 12 women,12 men 28.7 ⫾ 2.9 16.6 ⫾ 0.7 19.7 ⫾ 4.1 36.5 ⫾ 8.0 100.5 ⫾ 8.0 47.2 ⫾ 3.2

CONFLICT ADAPTATION IN EMOTIONAL TASK

323

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 2 Descriptive Statistics of Behavioral Data Trial

RT (ms)

SD

Accuracy (%)

SD

Total number of trials

Number of trials excluded from the analysis

Incongruent Congruent Congruent-incongruent Incongruent-incongruent

998 881 1020 976

123 153 164 147

93,9% 95,6% 92,9% 94,9%

2,2 1,6 2,6 2,2

1440 1440 720 720

176 140 104 72

from the examination process. The diagnosis was made based on clinical observation, information (obtained during the interview) related to the subjects’ life events, typical behavior and relationships, their inner experience, and capacity for self-reflection. A detailed description of the study protocol was provided and all participants gave written informed consent. The study protocol was in concordance with the Declaration of Helsinki and approved by the Institutional Review Board of the Medical Faculty, RWTH Aachen University. All subjects received standardized instruction and a 5-min training session outside the scanner. Two data sets from the previous analysis were replaced by new data of healthy controls. The high number of errors (⬎20 errors) committed by the subjects in incongruent conditions, especially at the beginning of the paradigm, and the fact that posterror trials were excluded from the analysis, led to technical difficulties pertaining to the use of those log files for first-level analysis.

MRI Data Acquisition Functional imaging was performed on a 3T Trio MR scanner (Siemens Medical Systems, Erlangen, Germany) using echoplanar imaging sensitive to BOLD (Blood-oxigen-level dependent) contrast (voxel size ⫽ 3.0 ⫻ 3.0 ⫻ 3.0 mm3, 64 ⫻ 64 matrix, FoV (field of view) ⫽ 192 mm2, 34 slices, gap 0. 75 mm, TR (repetition time) 2 s, TE (echo-time) 28 ms, ␣ ⫽ 77°). The scans were

acquired in interleaved mode. A 4-min magnetization-prepared rapid acquisition gradient echo image (MP-RAGE) T1-weighted sequence was used to acquire structural images (TR ⫽ 1,900 ms, TE ⫽ 2.52 ms, TI (inversion time) ⫽ 900 ms, matrix ⫽ 256 ⫻ 256, 176 slices, FoV ⫽ 250 ⫻ 250 mm2, ␣ ⫽ 9°, voxel size ⫽ 1 ⫻ 1 ⫻ 1 mm3) used for the coregistration step.

fMRI Paradigm In the event-related emotional interference task, single trials were combinations of emotional faces with words (distractors) printed across them in bold red letters to produce emotionally congruent or incongruent stimuli (Figure 1). The stimulus set consisted of 20 male and female persons, each photographed expressing sadness, fear, or happiness. The face images were taken from the set used in the Facial Emotions for Brain Activation (FEBA) test (Gur et al., 2002) and put in standardized positions of the eyes and mouth and normalized brightness. One hundred twenty single trials were presented in one run. For the analysis, they were classified as either congruent (C) or incongruent (I), and were further encoded according to the congruency (or incongruency) of the previous trial, resulting in four different order types (Figure 1). In this text, a lower case first letter denotes the previous trial (c for congruent and i for incongruent) and an upper case second letter denotes the current trial (C for

Figure 1. Emotional conflict paradigms. (A) Basic stimulus material consisting of congruent and incongruent face expression/word pairs from the FEBA face collection for emotional paradigm. Single trials are combinations of emotional faces (with sad, happy, or fearful facial expressions) and emotional words (distractors) “TRAUER,” ¨ CK” (German for “sadness,” “fear,” and “happiness”). Shown are examples of trials in “ANGST,” or “GLU which a congruent stimulus is followed by an incongruent stimulus, resulting in a high-conflict (low-control) trial, and an incongruent stimulus followed by an incongruent stimulus, resulting in a low-conflict (high-control) trial. (B) Trials were presented pseudorandomly so as to produce an equal number of each possible trial sequence with respect to high-conflict/low-control and low-conflict/high-control trials and were analyzed by trial type (congruent-congruent, congruent-incongruent, incongruent-congruent and incongruent-incongruent).

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

324

CHECHKO, KELLERMANN, SCHNEIDER, AND HABEL

congruent and I for incongruent) resulting in the labels cC, cI, iC, and iI for the respective order types. The first trial of the task was congruent and was excluded from the analysis. Error trials and posterror trials were also excluded from the analysis of reaction time (RT) and imaging data. The reason for exclusion of posterror trials was that in previous studies a slowing down had been observed during trials following an error trial (Egner & Hirsch, 2005). The number of congruent and incongruent trials, the number of order types, the number of trials representing each emotional category, and the number of face-word combinations were counterbalanced across the task. Stimulus occurrence did not include direct repetitions of the same emotional category of the face or word (Etkin et al., 2006; Mayr, Awh, & Laurey, 2003; Ullsperger, Bylsma, & Botvinick, 2005). Analogous to previous research involving the study of conflict adaptation effect by means of functional imaging (e.g., Egner & Hirsch, 2005; Egner, 2008; Etkin et al., 2006), trials were displayed for 1,000 ms with randomized interstimulus interval (4.00 ⫾ 0.38 s, range ⫽ 3–5 s) using Presentation software (Neurobehavioral Systems, San Francisco, CA). We used very short intertrial intervals (ITIs) because otherwise the sequential effects would not be detectable (as the influence of preceding trials diminishes within a few seconds). Participants watched the pictures via video goggles (VisuaStim XGA, Resonance Technology Inc., Los Angeles, CA) and were asked to judge the emotion of the faces while trying to ignore the words and answer as quickly and precisely as possible by pressing one of the three answer buttons with the right index, middle, or ring finger for sad, fearful, or happy faces, respectively.

Analysis of Behavioral Data RTs collected during the functional MRI (fMRI) experiment were analyzed by using analysis of variance (ANOVA) with repeated measures. To assess the behavioral trial-by-trial conflict adjustment, incongruent trials were divided into high-conflict/lowcontrol (cI) and low-conflict/high-control (iI) trials, which were further separated based on the emotional categories of the target stimuli. Thus, congruent-incongruent (cI) trials reflected conflict under low-control conditions, with incongruent-incongruent (iI) trials reflecting conflict under high- control conditions. Congruent trials were also split based on the character of previous trial, resulting in two types of congruent stimuli (cC or iC trials). To assess the effect of previous trial on the current trial, and its relation to the emotion type of the target, we performed a 3-way analysis of variance (ANOVA) of the RT with following factors: previous trial type (two levels: congruent or incongruent), current trial type (two levels: congruent of incongruent) and emotion type of target (three levels: sad, happy of fearful faces). To assess conflict adaptation effect and its relation to the targets’ emotional categories in greater detail, we also performed a 2-way target’s Emotion Type ⫻ Conflict Type ANOVA on RT data between the factor conflict type (2 levels: high-conflict/low-control and lowconflict/high-control trials) and the factor target’s emotion type (3 levels). For accuracy calculations, all types of errors (wrong answers and omissions) were considered. Posterror trials were excluded from the accuracy calculations. Analogous to the analysis of RTs, we also performed a 3-way Previous Trial Type ⫻ Current Trial

Type ⫻ Emotion Type of Target and 2 way Emotion Type of Target ⫻ Conflict Type analyses of variance (ANOVA) on accuracy data.

Analysis of fMRI Data Images were processed using Statistical Parametric Mapping (SPM) software (version SPM5, http://www.fil.ion.ucl.ac.uk/spm) on a Linux workstation. The first five images of each time series were excluded because of T1 stabilization effects. All remaining images were slice-time corrected and realigned to the first image. Images were normalized to a standard EPI template (interpolation to 2 ⫻ 2⫻2 mm3 resolution) and smoothed with an isotropic Gaussian kernel (8 mm full width at half maximum). None of the data sets revealed movement parameters exceeding one voxel size. Structural images were obtained prior to functional scanning by using a standard T1-weighted pulse sequence. Images for all subjects were coregistered and normalized to a common reference structural MRI template. Analogous to the statistical analysis of behavioral data, as we used an event-related design, we modeled the single-subject fMRI time series with 12 regressors of interest: order types (cC, iC, iI, Ci). Thus, all single events were modeled and classified according to the preceding trial (apart from the very first stimulus which had no preceding trial) and split by the targets’ emotional category (sad, fearful or happy). Delta functions of the trial types were convolved with the HRF in order to build regressors in the model. Because the ITI is 4 s and the duration was modeled as zero, the resulting regressors were independent of one another. An additional regressor was included for wrong trials and posterror trials, and another one modeled the average across the whole time series. Sessions were high-pass filtered with a cut-off period of 128 s and serial auto-correlations were accounted for with an AR(1) model. Contrast estimates of the 12 regressors of interest from each subject were entered in a three-way ANOVA with dependent observations (previous trial type [2 levels: congruent vs. incongruent] by current trial type [2 levels: congruent vs. incongruent] by target’s emotion type [3 levels]). Contrasts were specified for conflict generation or high-conflict/low-control (congruentincongruent trials ⬎ incongruent-incongruent trials) and conflict resolution or low-conflict/high-control (incongruent-incongruent trials ⬎ congruent-incongruent trials) sequences, whereas each of the two levels of conflict contrasts was further split based on the targets’ emotional category. Three directional t-contrasts comparing incongruent high- and low-conflict (low- and high-control) trials separated by the target’s emotional character were defined. Finally, to demonstrate the effect of interference conflict independently of the previous trial type, we subjected all four contrasts (iI ⬎ cC; iI ⬎ iC; cI ⬎ cC; cI ⬎ iC), comparing the congruent and incongruent items, to a conjunction analysis. Unless otherwise stated, the significance level for all main effects of the imaging data with regard to conflict adaptation effect was set to p ⬍ .05 family wise error (or FWE) corrected at the cluster level using a cluster-defining threshold of p ⬍ .001 at the voxel level. The significance level for each single contrast in the conjunction was set to an uncorrected threshold of p ⬍ .001. The probability of falsely declaring a voxel as significant across

CONFLICT ADAPTATION IN EMOTIONAL TASK

325

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 3 Regions Involved in Emotional Interference Conflict (Conjunction Analysis)

Anatomical region

Side

Cluster size

Superior parietal lobule

L

520

Supplementary motor area

R

290

Insula lobe

R

125

Inferior frontal gyrus Insula lobe Inferior frontal gyrus Precentral gyrus Middle occipital gyrus Inferior parietal lobule Middle occipital gyrus Fusiform gyrus

L

110

L L L R L R

91 67 43 30 28 18

MNI-coordinates of peak voxels z score

x

y

4.58 4.38 3.84 4.58 3.87 3.81 3.93 3.28 4.13 3.91 3.96 4.22 3.67 3.53 3.55 3.71

⫺26 ⫺32 ⫺20 4 0 ⫺8 42 36 ⫺46 ⫺44 ⫺52 ⫺54 ⫺42 36 ⫺28 34

⫺60 ⫺54 ⫺64 14 8 10 22 20 12 8 18 2 ⫺76 ⫺48 ⫺76 ⫺48

z 46 54 40 50 52 54 2 6 0 2 22 40 ⫺2 48 28 ⫺22

Note. MNI ⫽ Montreal Neurological Institute. Significance level for each of the four single contrasts in the conjunction was set to an uncorrected threshold of p ⬍ .001. The clusters of activation in the conjunction were displayed only when they contained ⬎10 contiguous voxels.

four tests in conjunction is 0.0014 (or 0.000000000001) uncorrected if the four tests were independent of one another. The clusters of activation in the conjunction were displayed only when they contained ⬎10 contiguous voxels.

The data are reported in Montreal Neurological Institute (MNI) space coordinates. Activations were anatomically localized using version 1.5 of the SPM anatomy toolbox (http://www .fz-juelich.de/inm/spm_anatomy_toolbox; Eickhoff et al., 2005).

Figure 2. Regions involved in emotional interference conflict (comparison between congruent and incongruent trials independent of the previous trial type (conjunction analysis). (A) Recruitment of the right fusiform face area (MNI: x ⫽ 34, y ⫽ ⫺48, z ⫽ ⫺22). (B) Response of the right (MNI: x ⫽ 42, y ⫽ 22, z ⫽ 2) and the left (MNI: x ⫽ ⫺44, y ⫽ 8, z ⫽ 2) insula. (C) Involvement of the supplementary motor area (MNI: x ⫽ 4, y ⫽ 14, z ⫽ 50). (D) Response of the left superior parietal lobule (MNI: x ⫽ ⫺26, y ⫽ ⫺60, z ⫽ 46)/ Significance level for each of the four single contrasts in the conjunction was set to an uncorrected threshold of p ⬍ .001.

CHECHKO, KELLERMANN, SCHNEIDER, AND HABEL

326

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 4 Effects of Previous Trial Type ⫻ Current Trial Type Interaction

Anatomical region

Side

Precuneus Precuneus Pallidum Superior frontal gyrus Caudate nucleus Superior medial gyrus Middle frontal gyrus Parahippocampal gyrus Fusiform gyrus

L R L L R R R R R

MNI-coordinates of peak voxels

Cluster size 664 236 114 108 76 59 61 35

z score

x

y

z

4.66 3.99 4.93 4.69 4.86 4.44 4.01 4.43 3.94

⫺8 16 ⫺10 ⫺22 20 12 24 26 42

⫺58 ⫺46 4 50 22 46 48 ⫺32 ⫺44

42 40 0 0 ⫺6 28 28 ⫺14 ⫺18

Note. MNI ⫽ Montreal Neurological Institute. Activity is shown at p ⬍ .001 uncorrected. The clusters of activation were displayed only when they contained ⬎30 contiguous voxels.

The results were shown on the T1 average of a cohort of participants.

Behavioral Results Analysis of RT. In a 3-way Previous Trial Type ⫻ Current Trial Type ⫻ Emotion Type of Target analysis of variance, we observed interaction of previous by current trial type, F(1,23) ⫽ 32.8, p ⬍ .001, revealing the effect of preceding trial type on the RT of current trial. Significant main effect of the factor current trial type, F(1,23) ⫽ 86.5, p ⬍ .001, indicated that RTs for correct trials were longer during incongruent trials (998 ms ⫾ 123 vs. 881 ms ⫾ 153 [mean ⫾ SD] for incongruent and congruent trials, respectively). The significant main effect of the factor emotion type of target, F(2,46) ⫽ 19.6, p ⬍ .001, revealed that stimuli with happy facial expressions were associated with faster RTs compared with the stimuli with sad or fearful expressions (mean for happy faces ⫽ 871 ms ⫾ 144; mean for sad faces ⫽ 986 ms ⫾ 156; mean for fearful faces ⫽ 962 ms ⫾ 143 [mean ⫾ SD]). No other effects were significant. The 2-way Conflict Type ⫻ Emotion Type of Target ANOVA on RT data revealed a significant main effect of conflict, F(1,23) ⫽ 15.6, p ⫽ .001, because RTs for incongruent trials were faster in low-conflict/high-control (iI) trials than in high-conflict/lowcontrol (cI) trials, t23 ⫽ ⫺4.1, p ⬍ .001; 976 ms ⫾ 147 vs. 1,020

ms ⫾ 164 (mean ⫾ SD). As the significant main effect of the factor emotion type of target, F(2,46) ⫽ 6.8, p ⫽ .003, indicated, incongruent stimuli with happy faces as targets were processed faster (mean for happy faces ⫽ 943 ms ⫾ 168; mean for sad faces ⫽ 1,038 ms ⫾ 182; mean for fearful faces ⫽ 1,014 ms ⫾ 163 [mean ⫾ SD]). There was no interaction between the factors conflict type and emotion type of target, F(2,46) ⫽ 0.3, p ⫽ .696, suggesting that the conflict adaptation effect was not driven disproportionately by any emotional category of target. Analysis of accuracy. A 3-way Previous Trial Type ⫻ Current Trial Type ⫻ Emotion Type of Target analysis of accuracy (ANOVA) on RT data was significant, F(2,46) ⫽ 11.6, p ⫽ .001. We also observed the interaction of previous and current trial types, F(1,23) ⫽ 32.0, p ⬍ .001, revealing the effect of preceding trial on accuracy in current trial. Significant main effect of the factor current trial, F(1,23) ⫽ 30.5, p ⬍ .001, indicated that accuracy for current trials were higher during congruent trials (95.6% ⫾ 1.6 vs. 93.9% ⫾ 2.2 [mean ⫾ SD] for congruent and incongruent trials, respectively). The main effect of emotion type of target, F(2,46) ⫽ 15.6, p ⫽ .001, revealed that stimuli with happy facial expressions were associated with higher accuracy compared with the stimuli with sad or fearful expressions (mean for happy faces ⫽ 95.6% ⫾ 1.2 [mean ⫾ SD]; mean for sad faces ⫽ 94.3% ⫾ 2.1 [mean ⫾ SD]; mean for

Table 5 Effects of Previous Trail Type ⫻ Conflict Interaction

Side

Cluster size

Insula

R

146

Insula Temporal pole Rolandic operculum Superior parietal lobule Fusiform gyrus Precentral gyurs

L R R R R L

70 69

Anatomical region

57 57 38

MNI-coordinates of peak voxels z score

x

y

z

4.49 3.87 3.21 3.90 4.20 3.19 4.05 4.26 3.67

40 42 42 ⫺38 54 56 16 42 ⫺34

2 0 0 ⫺2 8 6 ⫺56 ⫺44 ⫺30

⫺8 ⫺4 0 6 ⫺4 4 72 ⫺18 64

Note. MNI ⫽ Montreal Neurological Institute. Activity is shown at p ⬍ .001 uncorrected. The clusters of activation were displayed only when they contained ⬎ 30 contiguous voxels.

CONFLICT ADAPTATION IN EMOTIONAL TASK

327

fearful faces ⫽ 94.3% ⫾ 2.1 [mean ⫾ SD]). No other effects were significant. The 2-way Conflict Type⫻ Emotion Type of Target ANOVA revealed a main effect of conflict type, F(1,23) ⫽ 20.5, p ⬍ .001, as accuracy for incongruent trials were higher in low-conflict/highcontrol (iI) trials than in high-conflict/low-control (cI) trials (94.9% ⫾ 2.2 vs. 92.9% ⫾ 2.6 [mean ⫾ SD]). There were no other significant effects.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

fMRI Results Interference effect in fMRI data. Comparison of the congruent and incongruent conditions (conjunction analysis) revealed that in response to incongruent stimuli the participants demonstrated stronger involvement of the bilateral LPFC, the bilateral insula, posterior medial frontal cortex (pMFC), and the extrastriate (including FFA) visual cortex. The same applied to the bilateral parietal cortex (Table 3; Figure 2). Increased cortical responses to task-relevant information as reflected in the involvement of FFA. Analogous to behavioral data, we observed a significant Previous Trial Type ⫻ Current Trial Type interaction in the imaging data. At whole brain analysis of p ⬍ .001 uncorrected (the clusters of activation were reported only when they contained ⬎30 contiguous voxels) the effect of preceding trial on the current trial was also seen in the right fusiform gyrus, among other regions (Table 4). There was no significant Previous Trial Type ⫻ Current Trial Type ⫻ Emotion Type of Target interaction. Finally, we also saw a significant interaction between the Previous Trial Type ⫻ Conflict Type at p ⬍ .001 uncorrected (the clusters of activation were considered only when they contained ⬎30 contiguous voxels) in the bilateral right temporal pole/ frontal operculum region, the left precentral gyrus, the right fusiform gyrus and the right superior parietal lobule (SPL; Table 5; Figure 3). Regions associated with conflict adaptation (resolution) in emotional task. In the directional incongruent-incongruent trials ⬎ congruent-incongruent trials t-contrast, we observed involvement of the bilateral insula and the bilateral fO, the left precentral and postcentral gyri, the right fusiform gyrus, and the right superior parietal lobule (SPL; Table 6; Figure 4). At a more liberal threshold of p ⬍ .005 uncorrected (by a priori investigation of activity in amygdala as a region of interest), we saw that the cluster including the right insula/fO region (Z ⫽ 4.63; MNI x ⫽ 40 y ⫽ 2 z ⫽ ⫺8; 1,139 voxels) extended to the right amygdala (Figure 5). Conflict generation in emotional task. In the opposite contrast (incongruent-incongruent trials ⬍ congruent-incongruent trials), we did not detect any significant effects.

Discussion We applied emotional variations of the Stroop task using emotional face stimuli as targets and emotional word stimuli as distractors. The distractors generated a significant amount of interference conflict, which was lower in incongruent trials with preceding incongruence (low-conflict, high-control trials) than in the incongruent trials that followed congruence (high-conflict, low-control trials). The incongruent trials with preceding incongruence were associated with significantly higher levels of accuracy. Thus, we observed conflict adaptation effects by manipulating the levels of conflict on a trial-by-trial basis, reflecting the

Figure 3. Effects of Previous Trial Type ⫻ Conflict Type interaction in the right fusiform face area. (A) Response of the right fusiform face area (MNI: x ⫽ 42, y ⫽ ⫺44, z ⫽ ⫺18). (B) In response to conflict trials preceded by incongruent trials (low-conflict/high-control trials) compared with conflict trials preceded by congruent trials (high-conflict/low-control trails), subjects showed stronger activation in the right fusiform face area. Activity is shown at p ⬍ .001 uncorrected (the clusters of activation were displayed only when they contained ⬎30 contiguous voxels).

engagement of strategic processes and the reduction of conflict (Carter & van Veen, 2007): Interference conflict (as measured by the difference in RT between incongruent and congruent trials) led to the involvement of the ventral cortical regions including the fusiform (with the FFA) and the inferior occipital gyri. Furthermore, we noticed that within incongruent trials, the low-conflict (highcontrol) trials were also associated with a stronger involvement of the right fusiform gyrus. Other regions associated with conflict adaptation were the bilateral fO/insula, the bilateral parietal cortex and the right amygdala. Since we used faces as targets, involvement of the right fusiform gyrus (a region in the visual cortex specialized for face processing; Kanwisher, McDermott, & Chun, 1997) suggests augmentation of task-relevant stimulus information as a source of conflict reduction. The fO is suggested to regulate the involvement of FFA when task performance is guided by faces (Higo, Mars, Boorman, Buch, & Rushworth, 2011). As part of the ventral attentional network, the insula and the fO are involved in higher-order cognitive

CHECHKO, KELLERMANN, SCHNEIDER, AND HABEL

328

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 6 Regions Associated With Emotional Conflict Adaptation Effect

Anatomical region

Side

Insula Temporal pole Rolandic operculum Precentral gyrus Postcentral gyrus Inferior parietal lobule Insula Rolandic operculum Fusiform gyrus Superior parietal lobule

R R R L L L L L R R

Cluster size 394 161 116 78 77

MNI-coordinates of peak voxels z score

x

y

z

4.64 4.35 3.10 3.90 3.69 3.64 4.06 3.33 4.41 4.21

40 54 56 ⫺36 ⫺36 ⫺34 ⫺38 ⫺48 42 16

2 8 6 ⫺28 ⫺40 ⫺42 ⫺2 ⫺2 ⫺44 ⫺56

⫺8 ⫺4 4 66 54 53 6 8 ⫺18 72

Note. MNI ⫽ Montreal Neurological Institute. Activity is shown at p ⬍ .05 (cluster-level family-wise error-corrected; cluster-forming threshold at voxel-level p ⬍ .001).

processes such as changes in task sets (Corbetta, Patel, & Shulman, 2008) and stimulus-driven attentional reorientation (Cieslik, Zilles, Kurth, & Eickhoff, 2010). Amygdala activation in response to emotional faces has been repeatedly shown to correlate with activation in the visual cortex, including the FFA

(Morris, Öhman, & Dolan, 1999; Pessoa, McKenna, Gutierrez, & Ungerleider, 2002). While semantic interference conflict (in our study) involved emotional stimuli, its adaptation (not tested directly but assumed, given the involvement of FFA, insula, and fO) appeared to employ

Figure 4. Regions associated with emotional conflict’s adaptation effect. (A) Right fusiform face area (MNI: x ⫽ 42, y ⫽ ⫺44, z ⫽ ⫺18). (B) Response of the right (MNI: x ⫽ 40, y ⫽ 2, z ⫽ ⫺8) and the left (MNI: x ⫽ ⫺38, y ⫽ ⫺2, z ⫽ 6) insula. Activity is shown at p ⬍ .05 cluster-level family wise error-corrected; cluster-forming threshold at voxel-level p ⬍ .001.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

CONFLICT ADAPTATION IN EMOTIONAL TASK

Figure 5. Involvement of the right amygdala in response to emotional conflict’s adaptation effect. Activity is shown at p ⬍ .005 uncorrected (the clusters of activation were displayed only when they contained ⬎30 contiguous voxels).

the top-down mechanism of selective attention, as had been previously shown in a nonemotional word-face task (Egner & Hirsch, 2005). A study conducted by Egner et al. showed that the prefrontal cortex and the insula augment the processing of task-relevant stimuli by top-down induction of FFA involvement, thus playing a crucial role in conflict adaptation. Based on this observation, they maintained that performance adjustments in response to conflict are mediated by amplified neural processing of task-relevant (target) stimulus features and not by inhibited processing of taskirrelevant (distractor) stimulus features. However, the described effect was suggested to be limited only to the processing of nonemotional conflict. Using emotional interference stimuli akin to those used in the study by Etkin et al., our work appears to contradict the previous finding. The study by Etkin et al., 2006 indicated that while emotional conflict, as an involuntary response to emotional interference, takes place in the amygdala, the rACC is involved in conflict adaptation by inhibiting the processing of emotional distractors through top-down modulation of amygdala responsivity. We, on the other hand, suggest that a pathway including the fusiform gyrus and the fronto-parietal regions is involved in the detailed analysis of information required for emotional face perception and, finally, for emotional conflict adaptation. While Etkin et al. detected the involvement of the amygdala in conflict generation processes (high-conflict/low- con-

329

trol trials), suggesting an involuntary response of the amygdala to emotional interference, we observed the amygdala’s involvement in relation to conflict resolution (low-conflict/high- control trials). Based on that, we suggest that heightened activation in the amygdala and the FFA during conflict resolution is likely to ensue from increased focus on the emotional face dimension, which is consistent with the view that attended (compared with unattended) faces evoke stronger involvement of those areas (Pessoa et al., 2002). The differences in the results yielded by these two studies might be linked to the disparity in the employed tasks. We used three different emotional categories as opposed to two (happy and fearful faces) in the study conducted by Etkin and his group. In tasks applying only two types of incongruent stimuli, a stronger contribution of associative processes such as episodic memory to conflict adaptation is suggested (Egner, Etkin, Gale, & Hirsch, 2008). Furthermore, correct emotional recognition of two emotional categories seems to be based on whether the target’s emotion was positive or negative, without requiring a more explicit categorization. The lack of differences in RT in responses to happy and fearful faces in the study by Etkin et al. is in line with this assumption. While negative emotions such as fear may be detected faster (Yang, Zald, & Blake, 2007), detailed cognitive analysis of such emotions (involving recognition/categorization) requires longer processing (Barrett, 2006). Thus, while the study by Etkin et al. proposed unintended processing of emotional distractors leading to emotional conflict, our findings suggest that the effect of a distractor, even if emotional, is likely to be suppressed when attentional resources are bound by the processing of task-relevant (target) information. While the findings of Etkin et al. propose that emotion processing is automatic and independent of attentional resources (e.g., Dolan, 2002), our observations corroborate the notion that emotion processing depends on manipulations affecting the availability of processing resources associated with the demands of the main task (competing view; e.g., Pessoa et al., 2002). Similarly, it has been observed that increased cognitive demand can lead to modulation of the “emotional brain regions” (including the amygdala), as with increased cognitive demand, neural responses to emotional pictures are considerably reduced in the amygdala and the orbitofrontal cortex, whereas stronger activation is triggered in an extensive frontoparietal network (Kellermann et al., 2012). Moreover, an increasing body of evidence suggests that the notion that emotion processing is automatic does not contradict the so-called “competing view,” and vice versa. A recent study investigating systematic manipulations of both task difficulty and emotional challenge in a “lower-level” perceptual task has provided evidence that processing of emotional distraction is both automatic and modulated by attention (Shafer et al., 2012), thus supporting both the “traditional” and “competing” views. Collective evidence indicates a dynamic and context-specific function of the emotional and cognitive brain regions, which are capable of producing both attentional and affective responses based on task demands (Pessoa, 2008; Shafer et al., 2012). Previous studies have suggested that brain circuitries underlying the conflict adaptation effect vary strongly depending on the nature of conflict (emotional vs. nonemotional; Egner & Hirsch, 2005; Egner, 2008; Etkin et al., 2006). Our findings, however, challenge this view, as in emotional conflict adaptation we observed stronger

CHECHKO, KELLERMANN, SCHNEIDER, AND HABEL

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

330

involvement of the regions (FFA, LPFC, insula) previously thought to be related only to nonemotional conflict adaptation (Egner & Hirsch, 2005; Egner et al., 2008). Because we used a greater variety of emotional stimuli compared with previous work related to conflict adaptation in emotional task (Etkin et al., 2006), our results successfully corroborate the notion that emotional conflict processing can be both automatic and modulated by attention (as opposed to being only automatic, as had been previously suggested; Etkin et al., 2006), and that the modulation by attention depends on the processing load (e.g., high vs. low; Shafer et al., 2012). Because conflict resolution in our emotional task was also linked to the involvement of the amygdala, our results further suggest that the cognitive and affective origins of modulation associated with conflict adaptation cannot always be separated. Finally, our observations support the notion that conflict adaptation processes are domain-specific and that they may be mediated by multiple independent conflict-control loops (Egner et al., 2008).

References Barrett, L. F. (2006). Solving the emotion paradox: Categorization and the Experience of emotion. Personality and Social Psychology Review, 10, 20 – 46. doi:10.1207/s15327957pspr1001_2 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. Carter, C. S., Macdonald, A. M., Botvinick, M., Ross, L. L., Stenger, V. A., & Noll, D. (2000). Parsing executive processes: Strategic vs. evaluative functions of the anterior cingulate cortex. Proceedings of the National Academy of Sciences, USA, 97, 1944 –1948. doi:10.1073/pnas.97.4.1944 Carter, C. S., & van Veen, V. (2007). Anterior cingulate cortex and conflict detection: An update of theory and data. Cognitive, Affective & Behavioral Neuroscience, 7, 367–379. doi:10.3758/CABN.7.4.367 Chechko, N., Kellermann, T., Zvyagintsev, M., Augustin, M., Schneider, F., & Habel, U. (2012). Brain circuitries involved in semantic interference by demands of emotional and non-emotional distractors. PLoS ONE, 7, e38155. doi:10.1371/journal.pone.0038155 Cieslik, E. C., Zilles, K., Kurth, F., & Eickhoff, S. B. (2010). Dissociating bottom-up and top-down processes in a manual stimulus–response compatibility task. Journal of Neurophysiology, 104, 1472–1483. doi: 10.1152/jn.00261.2010 Corbetta, M., Patel, G., & Shulman, G. L. (2008). The reorienting system of the human brain: From environment to theory of mind. Neuron, 58, 306 –324. doi:10.1016/j.neuron.2008.04.017 Dolan, R. J. (2002). Emotion, cognition, and behavior. Science, 298, 1191–1194. doi:10.1126/science.1076358 Egner, T. (2008). Multiple conflict-driven control mechanisms in the human brain. Trends in Cognitive Sciences, 12, 374 –380. doi:10.1016/ j.tics.2008.07.001 Egner, T., Etkin, A., Gale, S., & Hirsch, J. (2008). Dissociable neural systems resolve conflict from emotional versus nonemotional distracters. Cerebral Cortex, 18, 1475–1484. doi:10.1093/cercor/bhm179 Egner, T., & Hirsch, J. (2005). Cognitive control mechanisms resolve conflict through cortical amplification of task-relevant information. Nature Neuroscience, 8, 1784 –1790.

Eickhoff, S. B., Stephan, K. E., Mohlberg, H., Grefkes, C., Fink, G. R., & Amunts, K. (2005). A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. NeuroImage, 25, 1325–1335. doi:10.1016/j.neuroimage.2004.12.034 Etkin, A., Egner, T., Peraza, D., Kandel, E., & Hirsch, J. (2006). Resolving emotional conflict: A role for the rostral anterior cingulate cortex in modulating activity in the amygdala. Neuron, 51, 871– 882. doi:10.1016/ j.neuron.2006.07.029 Gur, R. C., Sara, R., Hagendoorn, M., Marom, O., Hughett, P., & Macy, L. (2002). A method for obtaining 3-dimensional facial expressions and its standardization for use in neurocognitive studies. Journal of Neuroscience Methods, 115, 137–143. doi:10.1016/S0165-0270(02)00006 –7 Higo, T., Mars, R. B., Boorman, E. D., Buch, E. R., & Rushworth, M. F. S. (2011). Distributed and causal influence of frontal operculum in task control. Proceedings of the National Academy of Sciences, USA, 108, 4230 – 4235. doi:10.1073/pnas.1013361108 Kanwisher, N., McDermott, J., & Chun, M. M. (1997). The fusiform face area: A module in human extrastriate cortex specialized for face perception. The Journal of Neuroscience, 17, 4302– 4311. Kellermann, T. S., Sternkopf, M. A., Schneider, F., Habel, U., Turetsky, B. I., & Zilles, K. (2012). Modulating the processing of emotional stimuli by cognitive demand. Social Cognitive and Affective Neuroscience, 7, 263–273. doi:10.1093/scan/nsq104 Mayr, U., Awh, E., & Laurey, P. (2003). Conflict adaptation effects in the absence of executive control. Natural Neuroscience, 6, 450 – 452. Morris, J. S., Öhman, A., & Dolan, R. J. (1999). A subcortical pathway to the right amygdala mediating “unseen” fear. Proceedings of the National Academy of Sciences, USA, 96, 1680 –1685. doi:10.1073/pnas.96.4.1680 Pessoa, L. (2008). On the relationship between emotion and cognition. Nature Reviews Neuroscience, 9, 148 –158. doi:10.1038/nrn2317 Pessoa, L., McKenna, M., Gutierrez, E., & Ungerleider, L. G. (2002). Neural processing of emotional faces requires attention. Proceedings of the National Academy of Sciences, USA, 99, 11458 –11463. doi:10.1073/ pnas.172403899 Shafer, A. T., Matveychuk, D., Penney, T., O’Hare, A. J., Stokes, J., & Dolcos, F. (2012). Processing of emotional distraction is both automatic and modulated by attention: Evidence from an event-related fMRI investigation. Journal of Cognitive Neuroscience, 24, 1233–1252. doi: 10.1162/jocn_a_00206 Stroop, J. R. (1992). Studies of interference in serial verbal reactions (reprinted from Journal Experimental-Psychology (Vol. 18, pp. 643– 662, 1935). Journal of Experimental Psychology-General, 121, 15–23. doi:10.1037/0096-3445.121.1.15 Ullsperger, M., Bylsma, L., & Botvinick, M. (2005). The conflict adaptation effect: It’s not just priming. Cognitive, Affective & Behavioral Neuroscience, 5, 467– 472. doi:10.3758/CABN.5.4.467 Wittchen, H. (1997). SKID-I. Strukturiertes Klinisches Interview for DSM– IV. Achse I: Psychische Störungen [SCID-II. Structured Clinical Interview for DSM-IV Axis I Mental Disorders]. Göttingen, Germany: Hogrefe. Yang, E., Zald, D. H., & Blake, R. (2007). Fearful expressions gain preferential access to awareness during continuous flash suppression. Emotion, 7, 882– 886. doi:10.1037/1528 –3542.7.4.882

Received July 13, 2012 Revision received September 9, 2013 Accepted October 3, 2013 䡲