2022; 20(3): 415-426  
Alterations of Power Spectral Density in Salience Network during Thought-action Fusion Induction Paradigm in Obsessive-compulsive Disorder
Sang Won Lee1,2,*, Eunji Kim3,*, Tae Yang Jang2,4, Heajung Choi3, Seungho Kim3, Huijin Song5, Moon Jung Hwang6, Yongmin Chang7,8, Seung Jae Lee2,4
1Department of Psychiatry, Kyungpook National University Chilgok Hospital, 2Department of Psychiatry, School of Medicine, Kyungpook National University, 3Department of Medical and Biological Engineering, Kyungpook National University, 4Department of Psychiatry, Kyungpook National University Hospital, Daegu, 5Biomedical Research Institute, Seoul National University Hospital, 6GE Health Korea, Seoul, 7Department of Molecular Medicine, School of Medicine, Kyungpook National University, 8Department of Radiology, Kyungpook National University Hospital, Daegu, Korea
Correspondence to: Seung Jae Lee
Department ofPsychiatry, School of Medicine, Kyungpook National University, 680 Gukchaebosang-ro, Jung-gu, Daegu 41944, Korea
E-mail: jayleemd@knu.ac.kr
ORCID: https://orcid.org/0000-0003-3648-9824
Yongmin Chang
Department ofMolecular Medicine, School of Medicine, Kyungpook National University, 680 Gukchaebosang-ro, Jung-gu, Daegu 41944, Korea
E-mail: ychang@knu.ac.kr
ORCID: https://orcid.org/0000-0002-0585-8714

*These authors contributed equally to this study.
Received: December 7, 2020; Revised: March 30, 2021; Accepted: April 29, 2021; Published online: August 31, 2022.
© The Korean College of Neuropsychopharmacology. All rights reserved.

Objective: Recent studies highlighted the triple-network model which illustrated the interactions among three large-scale networks including salience network (SN). The functional magnetic resonance imaging used in this study was designed to investigate the characteristics of three large-scale networks associated with the thought-action fusion (TAF) in patients with obsessive-compulsive disorder (OCD) using power spectral density (PSD) analysis.
Methods: This study included 32 OCD patients and 38 age-matched healthy controls (HC). The TAF task was modified from the experiment of Rassin. PSD from time courses in large-scale networks of each subject was measured to compare between the groups for both TAF and resting state.
Results: In SN, OCD reported lower power in the low-frequency domain of SN compared to HC using the two-sample t test during the TAF task (t = −2.395, p = 0.019) but not in the resting state. The PSD in the low-frequency domain of the SN had a significant negative correlation with state score in the guilty inventory (r = −0.361, p = 0.042) in OCD patients.
Conclusion: This study suggests that OCD patients showed reduced SN power which can be prominent in a certain situation, such as TAF. In addition, the PSD alterations in SN cause difficulty in processing ambiguous emotional cues in social situations, and the difficulty can be connected with a negative feeling (e.g., guilt).
Keywords: Obsessive-compulsive disorder; Power spectral density; Salience network; Triple-network model

Recent studies have emphasized the significance of the triple-network model which illustrated the interactions within and between default mode network (DMN), central executive network (CEN), and salience network (SN) in the pathophysiology of psychiatric disorders [1]. The DMN is deactivated during the cognitive task [2] and activated in self-referential processing [3]. Higher-order executive functions that need focused attention, working memory, and decision making were processed using functions of CEN [4]. The SN has been proposed to have an important role in switching brain activity between the DMN and CEN [5].

Obsessive-compulsive disorder (OCD) is characterized by intrusive thoughts termed obsessions and repetitive ritualistic behaviors so called compulsions and has been associated with diverse functional brain abnormalities with emphasis on the involvement of cortico-striato-thalamo- cortical (CSTC) circuits in OCD pathology. Further, recent neuroimaging studies have demonstrated the altered functional connectivity within and between these three brain networks may underlie the pathophysiology of OCD. In a study using resting state functional magnetic resonance imaging (fMRI), OCD patients had alteration in SN-DMN and SN-CEN connectivity, and switching difficulties between CEN and DMN [6]. Also, drug-naïve OCD patients had reduced intrafunctional connectivity among regions within the SN and decreased internetwork connectivity between the SN-DMN and CEN [7]. OCD patients had difficulties in decreasing the DMN activation in the non-resting condition [8]. We have paid attention to the fact that many components of CSTC loops overlap regions of SN, and these shared regions, such as anterior cingulate cortex (ACC) and striatum, indeed showed aberrant activity and functional connectivity in OCD patients [9].

While correlations in blood oxygen level dependent (BOLD) signal between brain regions, like aforementioned studies, have been the dominant class of methods for analyzing fMRI fluctuations, correlation-based techniques, focusing on a narrow frequency band, do not fully characterize changes in dynamics that occur across a broader range of frequencies. The power spectral density (PSD) is an alternative approach that quantifies the amount of specific time–frequency range, the temporal dimension of neural processing, in different brain regions. This functional approach of brain functioning enables the investigation of the temporal interplay between different intrinsic brain networks [10]. PSD approach using resting-state fMRI has been applied to find the characteristics of psychiatric illnesses, such as schizophrenia [11]. However, the PSD approach using task-based fMRI is seldom applied to psychiatric illnesses while PSD analysis using task-based fMRI has been applied to cognitive studies. Changes of PSD in DMN were measured during the n-back test to verify the age-related alteration in DMN [12]. Cognitive loads induced by the n-back task can modulate low-frequency fluctuation within the DMN [13]. Accordingly, certain tasks can alter the activity of large intrinsic networks, and these changes might be sensitively detected using the PSD approach. PSD is thought to reflect regional neuronal activity, and long-range neural synchronization affects PSD strengths although the physiological origin and significance of PSD were still not clear [14].

The current fMRI study used a thought–action fusion (TAF) induction task as an active task, which was adapted from an experimental paradigm [15] and validated for use in fMRI study [16]. In our task [16], participants were instructed that if they thought of the word apple, another person in a separate room would receive an electric shock (which did not really happen). The paradigm was designed to increase likelihood TAF responses, resulting in discomfort or feelings of guilt which, in turn, lead to attempts to reduce the thought of apple.

TAF, one of the major dysfunctional obsessive-compulsive (OC) beliefs in the cognitive theory for OCD, believes that thinking about something increases its likelihood of occurrence or are morally equivalent to actions [17,18]. TAF is significantly linked to OCD symptoms, such as obsession and guilt [19]. Increased beta frequency in the precuneus of individuals with high OC traits was shown in a previous electroencephalogram study [20]. An fMRI study using the TAF task also reported activation in the insula, dorsomedial prefrontal cortex, and precuneus, which are important components of SN and DMN [21]. Therefore, large-scale networks, such as SN and DMN, can be mainly involved in processing TAF-related stimuli.

The purpose of this fMRI study was to investigate the characteristics of three large-scale networks associated with the TAF cognitive model in patients with OCD using PSD analysis. It is believed that this is the first study to find the differences of three large-scale networks in OCD using the TAF task modified from the experiment of Rassin [15]. During the task, participants should monitor the thought of apple as a salient stimulus, which was supposed to potentially harm to another person. Therefore, based on the overlap between CSTC loop and SN and previous reports of impaired SN in OCD, it is hypothesized that OCD patients have PSD alterations in SN during the TAF task and these alterations relate to dysfunctional belief of TAF or associated psychological phenomenon.



This study included 32 OCD patients and 38 age- matched healthy controls (HC). All participants were male, 18−46 years old. The handedness of each participant was checked using the Edinburgh Handedness Inventory. All participants were provided written informed consent following the procedures approved by the Institutional Review Board of Kyungpook National University Hospital (2018-04-029). A previous study using part of the data from healthy participants has been published [16].

Experimental Design

All subjects were scanned during both resting state and TAF task. An 8-minutes resting-state fMRI was implemented to avoid being affected by prior tasks before getting the TAF task data. The subjects were asked to relax and keep still with eyes closed but to avoid falling asleep during the resting state. The subjects played the TAF task with visual stimuli using SuperLab (version 4.5; Cendus Corp., San Pedro, LA, USA) and binocularly presented through an MRI-compatible visual stimulator mounted on the head coil (NordicNeuroLab, Bergen, Norway).

The subjects were instructed that if they think of apple before entering the MR scanner, the brain activity could be detected and verified by monitoring the MR scanner. Additionally, the subjects received the instructions that an electrical shock would be administered to the person outside the MR shielding room whenever the thought of apple comes to mind. Subsequently, they tested a real but low-intensity electrical pulse as a test using a portable low-frequency massager. Finally, they met the other participant (one of the investigators) who has two sham electrode patches attached to her hands.

The TAF task paradigm consisted of three phases: the thought of apple, evaluation, and fixation. The thought of apple had four different conditions. These conditions were combinations of the electric shock (Esk+/Esk−) and suppression of the thought of apple (Sup+/Sup−). The subjects could recognize the existence of the electric shock through a red lightning bolt sign on the screen. In the following evaluation phase, the subjects responded to how much they thought about apple using the response grip according to four frequency ranges of 1 (0−25%), 2 (25−50%), 3 (50−75%), and 4 (75−100%). In total, the TAF task paradigm lasted 8 minutes (thought of apple, 10 seconds; evaluation, 4 seconds; and fixation, 10 seconds; 24 seconds for each trial × 4 conditions × 5 repetitions). This TAF task was applied in a previous fMRI study [16].

Psychological Measures

Psychological symptoms were assessed using the Obsessive-Compulsive Inventory-Revised (OCI-R) [22,23], Thought-Action Fusion Scale (TAFS) [18,24] and Guilt Inventory [25]. The OCI-R is an 18-item self-reported questionnaire that measures six symptom domains: washing, checking, obsessing, neutralizing, ordering, and hoarding, with each subscale ranging from 0 to 12. Korean version of the OCI-R was validated [23]. The TAFS [18] is a 19-item (12 for moral TAF and seven for likelihood TAF) self-reported measure that evaluates the tendency of TAF-like cognition. Each of the items is rated from “disagree strongly” (coded as 0) to “agree strongly” (coded as 4). Korean version of TAFS was used in our study [24]. The guilt inventory (GI) is a self-reported questionnaire to evaluate three domains of guilt (state-guilt, trait-guilt, and moral standards), consisting of 45-item with 5 point Likert scale [25]. We only used state-guilt score in this study.

Data Acquisition and Preprocessing

The whole-brain function was acquired by using a Discovery MR750w 3.0 T (GE Healthcare, Milwaukee, WI, USA) with a 24-channel head coil. The functional data were obtained with the parameters of repetition time (TR, 2,000 ms), echo time (TE, 30 ms), field of view (FOV, 23 cm), matrix (64 × 64), and slice thickness:gap (4:0 mm) using the echo-planar imaging sequence. Both functional data consisted of 240 volumes. The T1-weighted fast spoiled gradient brain volume imaging were acquired as a structural reference image with high resolution (TR = 8.5 ms, TE = 3.2 ms, FOV = 25.6 cm, flip angle (FA) = 12°, and voxel dimension = 1 × 1 × 1 during 4 minutes 14 seconds).

Image data were preprocessed and statistically analyzed using the Statistical Parametric Mapping program (SPM12; Wellcome Centre for Human Neuroimaging, London, UK). The functional data were preprocessed with motion correction for the first volume, slice timing correction, co-registration with high-resolution T1 image and realigned mean echo-planar imaging for normalization to standard space, and spatial smoothing using an 8-mm Gaussian kernel.

Independent Components Analysis and Power Spectral Density

The spatial independent component for the TAF task and the resting state was separately analyzed with GroupICAT v4.0b (GIFT v3.0b; http://mialab.mrn.org/ software/gift) running in MATLAB 2018b (MathWorks, Natick, CA, USA). Spatial independent components analysis (ICA) was used to classify the neural networks with the InfoMax algorithm to separate the mixed signal. The preprocessed TAF data were decomposed into 39 components selected by minimum description length (MDL) criteria to estimate the optimal number of components as a dimension estimation using principal component analysis. The estimated optimal number of components for the resting state was 40 using the MDL criteria, but the resting state data was separated into 39 components similar to the TAF data to compare with the TAF data based on a previous study [26]. ICA was performed using the InfoMax algorithm for each subject and repeated 20 times in ICASSO to estimate the stability of the components. All the data were back-reconstructed into single-subject space scaled to Z-scores using the grounded intersubjective concept analysis method.

Spatial correlation sorting using the GIFT program was performed with established templates that had consistent meta-analysis activation to sort components which represent the triple-network model [27]. The sorted components had the greatest absolute spatial correlation value with each default mode, executive, and salience map.

A voxel-wise one-sample t test was used to generate each network map of groups for both functional data, and the results were at a threshold of p < 0.05 with family-wise error (FWE) correction. A two-sample t test was used to find group differences within large-scale networks at uncorrected p < 0.001 (cluster size > 85).

The Welch method in MATLAB Signal Processing Toolbox was used to calculate PSD from time courses in independent components of each subject to compare between the groups for both TAF and resting state. Each PSD was divided into five equally spaced frequency bins between 0 and 0.25 Hz with 0.05 Hz intervals. Several previous studies equally divided the frequency bins [28-30]. Thus, the current study equally divided five frequency bins for the inclusion of one trial of the TAF task which lasted 24 seconds. The extracted PSD was converted to fractional PSD as normalized with summation PSD in total frequency in each bin.

Additionally, general linear model (GLM) analyses were conducted to confirm the brain activation at each contrast. Methods and results of GLM analyses were presented in Supplementary Materials (Supplementary Fig. 1; available online).

Statistical Analysis

Statistical analyses were performed using the IBM SPSS Statistics for Windows version 25 (IBM Corp., Armonk, NY, USA). The independent two-sample t test at p < 0.05 measured the group differences in psychological measurements, response data during the TAF task, and frequency domains of PSD. Additionally, Pearson’s correlation analyses were conducted to assess the relationship between psychological measurements and frequency domains of PSD to find group comparison results in PSD analyses.


Physiological Assessment and Behavioral Data

The demographic and clinical variables for each group are presented in Table 1. The subjects in both groups did not have a significant age difference, but the ratio of handedness showed a significant difference between the groups (χ2 = 5.02, p = 0.042). OCD patients showed higher OC, state guilt, and TAFS scores on average compared to the HC. Eleven patients (35%) were drug-naïve or had been drug-free for three months, whereas 21 patients (65%) were taking antidepressants, mostly selective serotonin reuptake inhibitors (SSRIs).

Both OCD and healthy groups showed the successful thought suppression of apple in the suppression condition, regardless of the presence of electrical shock, during the performance of the TAF task. However, OCD showed a lower degree of thought of apple in Esk+/Sup− (t = −2.06, p = 0.043) and Esk−/Sup− (t = −1.96, p = 0.054), indicating that OCD patients thought about apple less than HC in the non-suppression condition (Table 2). No differences were evident in reaction time in each condition.

In addition, participants as a whole showed positive correlation between total scores of TAFS and degree of thought of apple (r = 0.28, p = 0.019) as well as reaction time (r = 0.29, p = 0.015) in Esk−/Sup+. Scores of state guilt also had a relationship with degree of thought of apple (r = 0.28, p = 0.019) in Esk+/Sup+.

Independent Components Analysis

The triple networks separately sorted 39 independent components by spatial correlation implemented in the GIFT toolbox with a template (Table 3). Moreover, sorted components of each network showed highly spatial similarity between groups when using dice similarity regardless of experimental condition.

In resting and TAF condition, both groups showed that DMN was included in the middle, posterior cingulate, middle occipital, middle temporal cortices, and angular gyrus. CEN was shown in the left inferior temporal, inferior parietal, precentral, bilateral inferior frontal, middle occipital cortices, right angular gyrus, and precuneus. Also, SN was shown in the putamen, hippocampus, parahippocampus, anterior insula, ACC, and middle prefrontal cortex (Figs. 1 and 2; FWE-corrected p < 0.05).

OCD showed reduced activity in the ACC within SN during the TAF task (uncorrected p < 0.001, cluster size ≥ 85; Fig. 3) in the two-sample t test. DMN and CEN did not have significant differences during the TAF task, and no significant difference existed within each network in the resting state.

Analyses of Power Spectrum Density

DMN and CEN had no difference in all of the frequency domains of PSD in a comparison between OCD and HC in the TAF and the resting state (Fig. 4). In SN, OCD reported lower power in bin1 of SN compared to HC using two sample t test during TAF task (t = −2.395, p = 0.019, Fig. 4C) but not in the resting state. The PSD in bin1 of SN had significant negative correlation with state subscale score in guilty inventory (r = −0.361, p = 0.042) in OCD patients (Fig. 5). However, no correlation between bin1 PSD and scores of guilty inventory was evident in the HC (r = −0.021, p = 0.903).


The results of the present study showed that OCD patients represented a reduced low-frequency PSD signal in the SN during the TAF task while the other large-scale networks of DMN and CEN did not show any significant group differences. Also, reduced power in low-frequency PSD in the SN was related to a state of guilt. In ICA analyses, OCD patients reported reduced activities in the ACC within the SN. This study reported that alterations in the SN during the TAF situation can be a potential neural feature of OCD patients.

The behavioral data in this study demonstrated that both OCD and healthy groups could successfully suppress the target thought in Sup+ compared to Sup–, regardless of the presence of electrical shock. These findings indicated the lack of paradoxical effect of thought suppression. In fact, two meta-analyses found that thought suppression reduces rather than increases the occurrence of target thoughts during thought suppression [31,32]. These studies argued that the ironic effect cannot be developed since people can successfully suppress thought over short periods of time, as was in our paradigm. However, although statistically insignificant, the target thought tended to increase with Esk+ during thought suppression while it seemed to decrease with Esk+ during thought non-suppression. These findings are consistent with a recent report that immediate enhancement effects were observed only in the presence of cognitive load [33]. Furthermore, overall patterns of OCD seemed to show more thinking of apple on suppression conditions and less thinking of apple on non-suppression conditions compared to healthy participants (Table 2). Consistent with our results, Rassin et al. (2008) [34] concluded that evidence for amplification of thought suppression failures in psychopathology is generally weak. OCD samples experienced even less amplification of intrusive thoughts while suppressing than did non-clinical groups [32].

Although general response patterns were similar between two groups in this study, between-group differences on the Likert scale were also found. The OCD group suppressed thought of apple more in Esk+/Sup− condition compared to the healthy control. One may assume that OCD patients may believe that thinking about apple increases the likelihood of occurrence of hurting the other person and thus still inhibit their thoughts more greatly than healthy participants even in the condition when thinking apple was freely allowed. Another explanation is that the condition contains two conflicting instructions—think of apple freely vs. thoughts of apple harm another person—that may affect the result. This highly conflicting situation can arouse overactive monitoring in OCD patients as mentioned in a previous study [35]. Based on these observations, while the paradoxical effect of thought suppression was small, TAF may exert its influence throughout our paradigm. These results may also validate this study.

OCD patients represented reduced low-frequency PSD in SN compared to healthy participants during the TAF task. Low-frequency PSD is known to be generated by both spontaneous activity and task-related responses. Similar to that of resting-state fMRI [36,37], spontaneous fluctuation of activity during task-fMRI can affect low-frequency PSD. In terms of task-related responses, the length of the task block in this study was 24 seconds which corresponds to 0.0417 Hz in the frequency domain. Responses of large-scale networks on TAF stimuli were included in low-frequency PSD (range, 0−0.05 Hz). A large-scale network such as the SN is important for the integration of internal/external stimuli and the modulation of the function of other large-scale networks [5]. However, the findings were rather inconsistent although several studies with resting-state fMRI reported aberrant function of SN in OCD patients. A recent meta-analysis revealed a reduced intrinsic functional connectivity of supramarginal gyrus (SMG) within the SN. However, the SMG is an overlapped region with other large-scale networks (e.g., CEN and DMN) rather than a specific component of SN [38]. Also, a study showed that OCD patients have increased functional connectivity in the ACC within the SN network [6]. Another recent study represented reduced intrinsic functional connectivity among SN subregions, such as ACC and insula [7], but not SMG. Furthermore, it should be noticed in the current study that no significant group differences were found in ICA and PSD analyses when using resting-state fMRI data. Only significant group differences were shown in the SN in PSD analysis when using the TAF data. Previous literature implicated that the SN activity can change depending on the difficulties or ambiguity of the task. The activities in the SN components (e.g., insula and ACC) increased with tasks with high ambiguity or difficulty [39,40]. The SN may be engaged with sensory integration to resolve stimuli ambiguity [39-41]. The results of the current suggest that OCD patients may fail to engage more SN function in ambiguous and emotional situations, such as TAF, following previous studies. Therefore, alterations in the SN of OCD patients can be prominent in the TAF compared with the resting condition. Moreover, the results of the present study also suggest that the PSD analysis can be a sensitive method to reveal the responses of large-scale networks on TAF. The PSD approach has an advantage in measuring the intrinsic functional dynamics of specific regions compared with the functional connectivity approach [10]. This study found that only PSD analyses but not ICA showed significant group differences in the SN.

The TAF response increase thought-related uncomfortable feelings, such as guilt [42]. In Rassin’s experiment that was modified in our study, the subjects felt guilty during the experiment, and the guilty feeling was negatively correlated with a socially desirable behavior [15]. Also, the previous fMRI study applying the same TAF task revealed that activities of visual association cortexes were related to a feeling of guilt [16]. The TAF task of the current study contains a social context that has a target person who could suffer from the subjects’ thoughts of apple, and the specific object might amplify the subjects’ feelings of guilt. The CSTC loop of the SN has been described to be involved in filtering a cortical input that leads to different responses [43]. The result of this study suggests that OCD patients showed an SN alteration which is related to inappropriate filtering of ambiguous situations, and this alteration can contribute to a maladaptive response, such as the feeling of guilt. However, the DMN and CEN did not show any group differences in this study. The result suggested that the TAF task of the current study was focused on the evaluation of salience of the stimuli rather than the cognitive or self-referential processing.

Furthermore, the results in this study for group differences in brain activities within the SN showed reduced activation in the ACC in OCD patients. The ACC was important in making an appropriate prediction on the current stimuli within the SN [44]. Also, the ACC involves in high-demand cognitive processes, such as conflict monitoring and attentional control [45,46]. Conflicting situations induced by the TAF task in this study can arouse ACC activation. Another important role of ACC is inhibitory control, such as thought suppression [47]. Successful thought retrieval and suppression can be related to the function of ACC [48]. Deficits of inhibitory control have been an important model for OCD symptoms [49], and reduced activity in ACC during response inhibition was reported in OCD patients [50]. Although alterations of ACC seems rather inconsistency by types of stimuli [51], functional and structural alterations of the ACC in OCD patients have been frequently reported in previous studies [52,53]. The results of the present study, which reported reduced activity in ACC, were in line with previous studies. In sum, there were three possible effects that can contribute to reduced activities in SN, especially the ACC, during TAF situation: 1) deficits in monitoring the importance of the current stimuli, 2) deficits in predicting the consequences of the stimuli, and 3) deficits in inhibitory control, such as thought suppression.

Several limitations should be noticed. First, the statistical level of ICA analyses of group differences was relatively low (uncorrected p < 0.001). Therefore, the interpretation of the results was minimized. However, the ACC cluster size was the largest in the uncorrected analyses of the present study, and the results were in line with previous literature that reported alterations in the ACC in OCD patients. Second, task-based PSD analyses need more replicative results because only small amounts of previous studies were conducted using task-based PSD. Additionally, in our study, the statistical power in PSD comparison between OCD and HC was relatively low (uncorrected p < 0.05). It does not show significant results with correction about multiple comparison problems. Third, medication effects can affect our fMRI results. About two-thirds of patients take their medications, mostly SSRIs. SSRI medication has significant effects on the strengths of brain functional connectivity [54,55]. Fourth, limitations on sample characteristics exist. The handedness showed group differences, and only young and male subjects were recruited in this study. Further pieces of evidence would be needed to generalize the results.

In conclusion, the results of the present study suggest that OCD patients showed reduced SN power which can be prominent in certain situations, such as the TAF. In addition, the PSD alterations in the SN enables difficulty in processing ambiguous emotional cues in social situations, and the difficulty can be connected with a negative feeling (e.g., guilt).


This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (NRF-2018R1A2B6007374).

Conflicts of Interest

No potential conflict of interest relevant to this article was reported.

Author Contributions

Conceptualization: Sang Won Lee, Yongmin Chang, and Seung Jae Lee. Data acquisition: Sang Won Lee, Tae Yang Jang, Eunji Kim, Heajung Choi, Seungho Kim, Huijin Song, Moon Jung Hwang, and Seung Jae Lee. Formal analysis: Eunji Kim, Heajung Choi, Seungho Kim, Huijin Song, Moon Jung Hwang, and Yongmin Chang. Writing-original draft: Sang Won Lee, Eunji Kim, Yongmin Chang, and Seung Jae Lee. Writing-review & editing: Sang Won Lee, Eunji Kim, Yongmin Chang, and Seung Jae Lee.

Fig. 1. Spatial maps of the default mode network (A, D), central executive network (B, E), and salience network (C, F) during the thought-action fusion (TAF) task (one sample t test, FWE corrected p < 0.05).
Fig. 2. Spatial maps of the default mode network (A, D), central executive network (B, E), and salience network (C, F) during the resting state (one sample t test, FWE corrected p < 0.05).
Fig. 3. Group differences in brain activity within salience network (two sample t test, uncorrected p < 0.001, cluster size > 85). Healthy controls showed higher activation in the ACC than OCD patients.
Fig. 4. Results of group differences during the TAF task (A−C) and resting state (D−F) with power spectrum density (PSD) analyses (two sample t test, *p < 0.05). Error bars indicated standard deviation of the mean PSD. The frequency bands are: 0.00−0.05 Hz (bin 1), 0.05−0.1 Hz (bin 2), 0.1−0.15 Hz (bin 3), 0.15−0.2 Hz (bin 4), and 0.2−0.25 Hz (bin 5).
Fig. 5. Relationship between guilty feeling and power of low frequency (0.00−0.05 Hz of bin 1) power spectrum density (PSD) in OCD patients. OCD, obsessive-compulsive disorder; a.u., arbitrary unit.

Demographic, clinical and psychological data

Variable OCD (n = 32) HC (n = 38) t2 p value
Demographic data
Age (yr) 25.3 ± 7.2 23.0 ± 2.1 1.8 0.085
Level of education (yr) 14.3 ± 1.7 14.7 ± 1.2 −1.2 0.231
Handedness (right/left/both) 26 (81.2)/6 (18.8) 37 (97.4)/1 (2.6) 5.0 0.042
Clinical data
Age of onset 19.0 ± 5.4 - - -
Duration of illness 6.4 ± 5.6 - - -
Medication vs. no current medication or drug-naïve 21 (65.6)/11 (35.4) - - -
Types of medication
Antidepressants (escitalopram/others) 19 (90.5)/9 (42.9) - - -
Anxiolytics (benzodiazepines/others) 14 (66.7)/11 (52.4) - - -
Antipsychotics 8 (38.1) - - -
Depressive disorder (except MDD) 3 - - -
Panic disorder or agoraphobia 3 - - -
Tic disorder 1 - - -
Psychological symptoms
Dimensional OC scale
Contamination 7.7 ± 4.7 3.2 ± 1.8 5.3 < 0.001
Responsibility for harm 8.6 ± 5.0 2.8 ± 2.5 6.3 < 0.001
Unacceptable thoughts 9.6 ± 4.5 3.0 ± 2.7 7.6 < 0.001
Symmetry 6.1 ± 6.0 1.8 ± 2.1 4.2 < 0.001
Obsessive compulsive inventory-revised, total 34.7 ± 13.6 12.3 ± 5.9 8.7 < 0.001
Checking 6.1 ± 3.3 1.8 ± 1.4 6.9 < 0.001
Hoarding 4.3 ± 3.1 2.5 ± 1.8 2.8 0.008
Neutralizing 4.8 ± 3.6 1.4 ± 1.7 5.0 < 0.001
Obsessing 8.6 ± 2.7 2.3 ± 1.9 11.2 < 0.001
Ordering 5.3 ± 2.9 3.1 ± 2.0 3.7 0.001
Washing 5.6 ± 3.3 1.2 ± 1.2 7.1 < 0.001
Guilty inventory, state 33.7 ± 6.6 25.7 ± 5.9 5.4 < 0.001
Thought-action fusion scale 30.8 ± 13.5 18.4 ± 12.5 4.0 < 0.001
Beck depression inventory 19.5 ± 11.6 4.9 ± 4.4 6.8 < 0.001

Values are presented as mean ± standard deviation or number (%).

OCD, obsessive-compulsive disorder; HC, healthy controls; MDD, major depressive disorder.

Behavioral data on thought-action fusion experiment during magnetic resonance scan

Variable OCD (n = 32) HC (n = 38) t p value
Degree of thought of applea
Sup+/Esk– 1.5 ± 0.5 1.3 ± 0.4 1.5 0.146
Sup+/Esk+ 1.6 ± 0.7 1.4 ± 0.4 1.5 0.133
t = −1.4, p = 0.223 t = −0.9, p = 0.391
Sup–/Esk– 2.9 ± 0.7 3.3 ± 0.6 −2.0 0.054
Sup–/Esk+ 2.7 ± 0.7 3.1 ± 0.8 −2.1 0.043
t = 2.1, p = 0.042 t = 1.6, p = 0.121
Reaction time (msec)
Sup+/Esk– 1,958.2 ± 590.5 1,852.5 ± 551.5 0.8 0.442
Sup+/Esk+ 1,800.2 ± 565.7 1,727.9 ± 538.2 0.6 0.586
t = 2.3, p = 0.026 t = 1.7, p = 0.092
Sup–/Esk– 1,917.1 ± 650.8 1,881.8 ± 457.2 0.3 0.798
Sup–/Esk+ 1,931.7 ± 536.4 1,810.7 ± 454.7 1.0 0.311
t = −0.2, p = 0.865 t = 1.2, p = 0.255    

Values are presented as mean ± standard deviation.

OCD, obsessive-compulsive disorder; HC, healthy controls; Esk, electric shock condition; Sup, thought suppression condition.

Paired t test for within-group comparison, independent t test for between-group comparison.

aLikert degree was measured by frequency range with 1 (0−25%), 2 (25−50%), 3 (50−75%), and 4 (75−100%).

Spatial correlation value with each network templates and dice similarity between OCD and HC

 Description TAF Resting
Spatial correlation
(a) DMN 0.526 0.580
(b) CEN 0.399 0.438
(c) SN 0.317 0.316
Dice similarity between groups
(a) DMN 0.767 0.791
(b) CEN 0.740 0.783
(c) SN 0.803 0.770

TAF, thought-action fusion; OCD, obsessive-compulsive disorder; HC, healthy controls; DMN, default mode network; CEN, central executive network; SN, salience network.

  1. Menon V. Large-scale brain networks and psychopathology: a unifying triple network model. Trends Cogn Sci 2011;15:483-506.
    Pubmed CrossRef
  2. Greicius MD, Krasnow B, Reiss AL, Menon V. Functional connectivity in the resting brain: a network analysis of the default mode hypothesis. Proc Natl Acad Sci U S A 2003;100:253-258.
    Pubmed KoreaMed CrossRef
  3. Whitfield-Gabrieli S, Ford JM. Default mode network activity and connectivity in psychopathology. Annu Rev Clin Psychol 2012;8:49-76.
    Pubmed CrossRef
  4. Sridharan D, Levitin DJ, Menon V. A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks. Proc Natl Acad Sci U S A 2008;105:12569-12574.
    Pubmed KoreaMed CrossRef
  5. Menon V, Uddin LQ. Saliency, switching, attention and control: a network model of insula function. Brain Struct Funct 2010;214:655-667.
    Pubmed KoreaMed CrossRef
  6. Fan J, Zhong M, Gan J, Liu W, Niu C, Liao H, et al. Altered connectivity within and between the default mode, central executive, and salience networks in obsessive-compulsive disorder. J Affect Disord 2017;223:106-114.
    Pubmed CrossRef
  7. Chen YH, Li SF, Lv D, Zhu GD, Wang YH, Meng X, et al. Decreased intrinsic functional connectivity of the salience network in drug-naïve patients with obsessive-compulsive disorder. Front Neurosci 2018;12:889.
    Pubmed KoreaMed CrossRef
  8. Gonçalves ÓF, Soares JM, Carvalho S, Leite J, Ganho-Ávila A, Fernandes-Gonçalves A, et al. Patterns of default mode network deactivation in obsessive compulsive disorder. Sci Rep 2017;7:44468.
    Pubmed KoreaMed CrossRef
  9. Peters SK, Dunlop K, Downar J. Cortico-striatal-thalamic loop circuits of the salience network: a central pathway in psychiatric disease and treatment. Front Syst Neurosci 2016;10:104.
    Pubmed KoreaMed CrossRef
  10. Duff EP, Johnston LA, Xiong J, Fox PT, Mareels I, Egan GF. The power of spectral density analysis for mapping endogenous BOLD signal fluctuations. Hum Brain Mapp 2008;29:778-790.
    Pubmed KoreaMed CrossRef
  11. Hoptman MJ, Zuo XN, Butler PD, Javitt DC, D'Angelo D, Mauro CJ, et al. Amplitude of low-frequency oscillations in schizophrenia: a resting state fMRI study. Schizophr Res 2010;117:13-20.
    Pubmed KoreaMed CrossRef
  12. Sambataro F, Murty VP, Callicott JH, Tan HY, Das S, Weinberger DR, et al. Age-related alterations in default mode network: impact on working memory performance. Neurobiol Aging 2010;31:839-852.
    Pubmed KoreaMed CrossRef
  13. Tommasin S, Mascali D, Gili T, Assan IE, Moraschi M, Fratini M, et al. Task-related modulations of BOLD low-frequency fluctuations within the default mode network. Front Phys 2017;5:31.
    Pubmed KoreaMed CrossRef
  14. Di X, Kim EH, Huang CC, Tsai SJ, Lin CP, Biswal BB. The influence of the amplitude of low-frequency fluctuations on resting-state functional connectivity. Front Hum Neurosci 2013;7:118.
    Pubmed KoreaMed CrossRef
  15. Rassin E, Merckelbach H, Muris P, Spaan V. Thought-action fusion as a causal factor in the development of intrusions. Behav Res Ther 1999;37:231-237.
    Pubmed CrossRef
  16. Lee SW, Kim E, Chung Y, Cha H, Song H, Chang Y, et al. Believing is seeing: an fMRI study of thought-action fusion in healthy male adults. Brain Imaging Behav 2021;15:300-310.
    Pubmed CrossRef
  17. Rachman S. Obsessions, responsibility and guilt. Behav Res Ther 1993;31:149-154.
    Pubmed CrossRef
  18. Shafran R, Thordarson DS, Rachman S. Thought-action fusion in obsessive compulsive disorder. J Anxiety Disord 1996;10:379-391.
  19. Rachman S, Thordarson DS, Shafran R, Woody SR. Perceived responsibility: structure and significance. Behav Res Ther 1995;33:779-784.
    Pubmed CrossRef
  20. Jones R, Bhattacharya J. A role for the precuneus in thought- action fusion: evidence from participants with significant obsessive-compulsive symptoms. Neuroimage Clin 2013;4:112-121.
    Pubmed KoreaMed CrossRef
  21. Lee SW, Cha H, Chung Y, Kim E, Song H, Chang Y, et al. The neural correlates of thought-action fusion in healthy adults: a functional magnetic resonance imaging study. Depress Anxiety 2019;36:732-743.
    Pubmed CrossRef
  22. Foa EB, Huppert JD, Leiberg S, Langner R, Kichic R, Hajcak G, et al. The Obsessive-Compulsive Inventory: development and validation of a short version. Psychol Assess 2002;14:485-496.
    Pubmed CrossRef
  23. Woo CW, Kwon SM, Lim YJ, Shin MS. The Obsessive- Compulsive Inventory-Revised (OCI-R): psychometric properties of the Korean version and the order, gender, and cultural effects. J Behav Ther Exp Psychiatry 2010;41:220-227.
    Pubmed CrossRef
  24. Lee SJ, Lee S. Korean version of the Thought-Action Fusion Scale: the evidence for its psychometric properties. Psychiatry Investig 2021;18:348-356.
    Pubmed KoreaMed CrossRef
  25. Jones WH, Schratter AK, Kugler K. The Guilt Inventory. Psychol Rep 2000;87(3 Pt 2):1039-1042.
    Pubmed CrossRef
  26. Zhang S, Li CS. A neural measure of behavioral engagement: task-residual low-frequency blood oxygenation level-dependent activity in the precuneus. Neuroimage 2010;49:1911-1918.
    Pubmed KoreaMed CrossRef
  27. Yarkoni T, Poldrack RA, Nichols TE, Van Essen DC, Wager TD. Large-scale automated synthesis of human functional neuroimaging data. Nat Methods 2011;8:665-670.
    Pubmed KoreaMed CrossRef
  28. Malinen S, Vartiainen N, Hlushchuk Y, Koskinen M, Ramkumar P, Forss N, et al. Aberrant temporal and spatial brain activity during rest in patients with chronic pain. Proc Natl Acad Sci U S A 2010;107:6493-6497.
    Pubmed KoreaMed CrossRef
  29. Otti A, Guendel H, Wohlschläger A, Zimmer C, Noll-Hussong M. Frequency shifts in the anterior default mode network and the salience network in chronic pain disorder. BMC Psychiatry 2013;13:84.
    Pubmed KoreaMed CrossRef
  30. Garrity AG, Pearlson GD, McKiernan K, Lloyd D, Kiehl KA, Calhoun VD. Aberrant "default mode" functional connectivity in schizophrenia. Am J Psychiatry 2007;164:450-457.
    Pubmed CrossRef
  31. Abramowitz JS, Tolin DF, Street GP. Paradoxical effects of thought suppression: a meta-analysis of controlled studies. Clin Psychol Rev 2001;21:683-703.
    Pubmed CrossRef
  32. Magee JC, Harden KP, Teachman BA. Psychopathology and thought suppression: a quantitative review. Clin Psychol Rev 2012;32:189-201.
    Pubmed KoreaMed CrossRef
  33. Wang DA, Hagger MS, Chatzisarantis NLD. Ironic effects of thought suppression: a meta-analysis. Perspect Psychol Sci 2020;15:778-793.
    Pubmed CrossRef
  34. Rassin E, Merckelbach H, Muris P. Paradoxical and less paradoxical effects of thought suppression: a critical review. Clin Psychol Rev 2000;20:973-995.
    Pubmed CrossRef
  35. Ursu S, Stenger VA, Shear MK, Jones MR, Carter CS. Overactive action monitoring in obsessive-compulsive disorder: evidence from functional magnetic resonance imaging. Psychol Sci 2003;14:347-353.
    Pubmed CrossRef
  36. Calhoun VD, Kiehl KA, Pearlson GD. Modulation of temporally coherent brain networks estimated using ICA at rest and during cognitive tasks. Hum Brain Mapp 2008;29:828-838.
    Pubmed KoreaMed CrossRef
  37. Fox MD, Raichle ME. Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat Rev Neurosci 2007;8:700-711.
    Pubmed CrossRef
  38. Gürsel DA, Avram M, Sorg C, Brandl F, Koch K. Frontoparietal areas link impairments of large-scale intrinsic brain networks with aberrant fronto-striatal interactions in OCD: a meta-analysis of resting-state functional connectivity. Neurosci Biobehav Rev 2018;87:151-160.
    Pubmed CrossRef
  39. Lamichhane B, Adhikari BM, Dhamala M. The activity in the anterior insulae is modulated by perceptual decision-making difficulty. Neuroscience 2016;327:79-94.
    Pubmed CrossRef
  40. Lamichhane B, Adhikari BM, Dhamala M. Salience network activity in perceptual decisions. Brain Connect 2016;6:558-571.
    Pubmed CrossRef
  41. Wiech K, Lin CS, Brodersen KH, Bingel U, Ploner M, Tracey I. Anterior insula integrates information about salience into perceptual decisions about pain. J Neurosci 2010;30:16324-16331.
    Pubmed KoreaMed CrossRef
  42. Ladouceur R, Rhéaume J, Freeston MH, Aublet F, Jean K, Lachance S, et al. Experimental manipulations of responsibility: an analogue test for models of obsessive-compulsive disorder. Behav Res Ther 1995;33:937-946.
    Pubmed CrossRef
  43. Choi EY, Yeo BT, Buckner RL. The organization of the human striatum estimated by intrinsic functional connectivity. J Neurophysiol 2012;108:2242-2263.
    Pubmed KoreaMed CrossRef
  44. Kleckner IR, Zhang J, Touroutoglou A, Chanes L, Xia C, Simmons WK, et al. Evidence for a large-scale brain system supporting allostasis and interoception in humans. Nat Hum Behav 2017;1:0069.
    Pubmed KoreaMed CrossRef
  45. Crottaz-Herbette S, Menon V. Where and when the anterior cingulate cortex modulates attentional response: combined fMRI and ERP evidence. J Cogn Neurosci 2006;18:766-780.
    Pubmed CrossRef
  46. Eichele T, Debener S, Calhoun VD, Specht K, Engel AK, Hugdahl K, et al. Prediction of human errors by maladaptive changes in event-related brain networks. Proc Natl Acad Sci U S A 2008;105:6173-6178.
    Pubmed KoreaMed CrossRef
  47. Anderson MC, Bunce JG, Barbas H. Prefrontal-hippocampal pathways underlying inhibitory control over memory. Neurobiol Learn Mem 2016;134(Pt A):145-161.
    Pubmed KoreaMed CrossRef
  48. Kuhl BA, Kahn I, Dudukovic NM, Wagner AD. Overcoming suppression in order to remember: contributions from anterior cingulate and ventrolateral prefrontal cortex. Cogn Affect Behav Neurosci 2008;8:211-221.
    Pubmed KoreaMed CrossRef
  49. Chamberlain SR, Blackwell AD, Fineberg NA, Robbins TW, Sahakian BJ. The neuropsychology of obsessive compulsive disorder: the importance of failures in cognitive and behavioural inhibition as candidate endophenotypic markers. Neurosci Biobehav Rev 2005;29:399-419.
    Pubmed CrossRef
  50. Page LA, Rubia K, Deeley Q, Daly E, Toal F, Mataix-Cols D, et al. A functional magnetic resonance imaging study of inhibitory control in obsessive-compulsive disorder. Psychiatry Res 2009;174:202-209.
    Pubmed CrossRef
  51. van Velzen LS, Vriend C, de Wit SJ, van den Heuvel OA. Response inhibition and interference control in obsessive- compulsive spectrum disorders. Front Hum Neurosci 2014;8:419.
    Pubmed KoreaMed CrossRef
  52. Yücel M, Wood SJ, Fornito A, Riffkin J, Velakoulis D, Pantelis C. Anterior cingulate dysfunction: implications for psychiatric disorders? J Psychiatry Neurosci 2003;28:350-354.
    Pubmed KoreaMed
  53. Rotge JY, Guehl D, Dilharreguy B, Tignol J, Bioulac B, Allard M, et al. Meta-analysis of brain volume changes in obsessive-compulsive disorder. Biol Psychiatry 2009;65:75-83.
    Pubmed CrossRef
  54. Arnone D, Wise T, Walker C, Cowen PJ, Howes O, Selvaraj S. The effects of serotonin modulation on medial prefrontal connectivity strength and stability: a pharmacological fMRI study with citalopram. Prog Neuropsychopharmacol Biol Psychiatry 2018;84(Pt A):152-159.
    Pubmed KoreaMed CrossRef
  55. Shin DJ, Jung WH, He Y, Wang J, Shim G, Byun MS, et al. The effects of pharmacological treatment on functional brain connectome in obsessive-compulsive disorder. Biol Psychiatry 2014;75:606-614.
    Pubmed CrossRef

This Article