Anodal tDCS over the right dorsolateral prefrontal cortex modulates cognitive processing of emotional information as a function of trait rumination in healthy volunteers

Accepted Manuscript Title: Anodal tDCS over the right dorsolateral prefrontal cortex modulates cognitive processing of emotional information as a function of trait rumination in healthy volunteers Author: Marie-Anne Vanderhasselt Alvaro Sanchez Haeike Josephy Chris Baeken Andre R. Brunoni Rudi De Raedt PII: DOI: Reference:

S0301-0511(16)30369-6 http://dx.doi.org/doi:10.1016/j.biopsycho.2016.12.006 BIOPSY 7305

To appear in: Received date: Revised date: Accepted date:

31-3-2016 5-12-2016 10-12-2016

Please cite this article as: Vanderhasselt, Marie-Anne, Sanchez, Alvaro, Josephy, Haeike, Baeken, Chris, Brunoni, Andre R., De Raedt, Rudi, Anodal tDCS over the right dorsolateral prefrontal cortex modulates cognitive processing of emotional information as a function of trait rumination in healthy volunteers.Biological Psychology http://dx.doi.org/10.1016/j.biopsycho.2016.12.006 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Running head: THE INFLUENCE OF TDCS ON COGNITIVE CONTROL

Anodal tDCS over the right dorsolateral prefrontal cortex modulates cognitive processing of emotional information as a function of trait rumination in healthy volunteers

Marie-Anne Vanderhasselt1,2,3*, Alvaro Sanchez 3, Haeike Josephy 4, Chris Baeken1,5, Andre R Brunoni 6, & Rudi De Raedt3

1

2

3

Department of Psychiatry and Medical Psychology, Ghent University, Belgium

Faculty of Medicine and Pharmacy, Free University Brussels, Brussels, Belgium

Department of Experimental-Clinical and Health Psychology, Ghent University, Belgium 4

5

6

Department of Data Analysis, Ghent University, Belgium

Department of Psychiatry, Free University Brussels, Belgium

Service of Interdisciplinary Neuromodulation (SIN), Laboratory of Neurosciences (LIM-27), Department and Institute of Psychiatry, University of São Paulo, São Paulo, Brazil

* Corresponding Author: Marie-AnneVanderhasselt, Ph. D. Department of Psychiatry and Medical Psychology, De Pintelaan 185-1K12F, B-9000 GENT Email: [email protected]

Highlights   

Individual differences in trait rumination were associated with the tDCS induced effects on cognitive costs Effects on cognitive costs were observed for both positive and negative stimuli the right DLPFC is causally involved in the modulation of cognitive control in healthy ruminators

Abstract Healthy individuals reporting higher (as compared to lower) levels of trait rumination recruit more neural activity in dorso-cortical regions (mostly in the right hemisphere) when inhibiting negative information, possibly to compensate their difficulty to disengage from it. In the present study, we investigated whether these latter neural correlates are causally implicated in cognitive control in these individuals. We administered the Cued Emotional Control Task, a measure of cognitive control indexed by cognitive costs for inhibiting versus providing a habitual response for emotional information, in thirty-five healthy volunteers reporting a broad range of trait rumination levels. Participants completed the task after receiving both real and sham-placebo (counterbalanced order) anodal transcranial Direct Current Stimulation (tDCS) over the right dorsolateral prefrontal cortex (DLPFC). Results reveal that the tDCS induced effects on cognitive costs for emotional information were associated with individual differences in trait rumination: the higher the trait rumination level, the less cognitive costs following real neuromodulation of the right DLPFC. Interestingly, these effects were observed for both positive and negative stimuli, and not only negative information as hypothesized. Overall, the data suggest that the right DLPFC is causally involved in the alteration of cognitive control in healthy individuals who tend to ruminate, possibly by helping them to disengage from emotional material.

Keywords: Rumination - Cognitive Control – right DLPFC - tDCS

Anodal tDCS over the right dorsolateral prefrontal cortex modulates cognitive processing of emotional information as a function of trait rumination in healthy volunteers

Rumination is defined as a maladaptive form of recurrent thoughts focused on the causes, symptoms, and implications of one‟s negative mood (Nolen-Hoeksema, 1991), and is considered a core cognitive risk factor for the onset, duration, severity and relapse of major depression (e.g., Nolen-Hoeksema, 2000). Even though rumination has mostly been investigated in depressed patients, the basic/fundamental cognitive mechanisms underlying rumination tendencies (i.e., without interference from current or past depression) remain to be determined. Within this vein, research in healthy individuals has shown that a higher tendency to ruminate (i.e., elevated trait rumination) is associated with deficits in cognitive control, such as impairments in the inhibition of previously relevant mental sets (using non-emotional stimuli, Whitmer & Banich, 2007), sustained processing of negative information (Duque, Sanchez, & Vazquez, 2014), and difficulties inhibiting a reflexive tendency towards emotional but not neutral information (De Lissnyder, Koster, Derakshan, and De Raedt, 2010). In their influencing theoretical model, De Raedt and Koster (2010) propose that an abnormal recruitment of cognitive control result in deficient emotional processes that characterize habitual ruminators (De Raedt & Koster, 2010). This suggestion was confirmed in a recent training study in healthy volunteers (Cohen, Mor, & Henik, 2014), in which an experimental manipulation was used to enhance executive control immediately prior to the presentation of negative pictures. This manipulation resulted in a reduction of rumination level following a resting period, indicating that cognitive control is crucially involved in rumination mechanisms.

To date, research is more and more identifying neural correlates of individual differences in rumination and cognitive control, as they can clarify vulnerability mechanisms for rumination and depression (Joormann, Yoon, Zetkste, 2007). Several studies have shown that, while successfully inhibiting habitual responses towards negative information, healthy individuals who report higher tendencies to ruminate display more neural activation in the right dorsolateral prefrontal cortex (rDLPFC) (Vanderhasselt, Kuhn, & De Raedt, 2011). Also in healthy never-depressed individuals, it has been reported that individual differences in trait rumination are positively correlated to the rDLPFC volume (Wang, Wei, Yang, Hao, & Qiu, 2015), and that lower resting-state Regional Homogeneity (ReHo) within the rDLPFC is related to more self-reported unwanted thoughts during rest (Kuhn, Vanderhasselt, De Raedt, & Galinat, 2014). However, given the correlational nature of neuroimaging studies, it remains an open question whether these neural correlates are causally involved in how ruminators process emotional information. This causal question can be investigated using neuromodulation techniques. Transcranial Direct Current Stimulation (tDCS) is a non-invasive technique that is able to induce changes in cortical excitability by applying a weak electrical current - via subthrehold changes in the membrane potential (for a review, see Nitsche et al., 2008). Anodal tDCS leads to an increase in cortical excitability (Nitsche and Paulus, 2000), and prior research has consistently shown that this stimulation protocol can be used to enhance cognitive control functioning (for a review, see Brunoni & Vanderhasselt, 2014). tDCS of the right prefrontal network has been shown to successfully influence inhibitory processes (Jacobson, Javitt, & Lavidor, 2011; Penolazzi, Stramaccia, Braga, Mondini, & Galfano, 2014; Stramaccia, Penolazzi, Sartori, Braga, Mondini, & Galfano, 2015; Silas, & Brandt, 2016; Campanella et al., 2016). A cognitive control paradigm that (1) displays neural substrates associated with individual differences in trait rumination (Vanderhasselt et al., 2013a), and (2)

has been used in tDCS research (Vanderhasselt et al., 2013b), is the Cued Emotional Control Task (CECT, Vanderhasselt et al., 2012; see also Figure 1). In this task, cognitive control is conceptualized as the difference between responding to the actual emotion as compared to the opposite emotion of an emotional stimulus (i.e., cognitive costs of inhibiting versus providing a habitual emotional response). Using this task, Vanderhasselt and co-workers (2013a) demonstrated that, whereas healthy participants at high and low trait rumination levels did not differ in their task performance, high (as compared to low) rumination tendencies were related to more activation in the posterior part of the dorsal anterior cingulate cortex (pdACC) during the inhibition of a dominant response to negative information. In addition, Vanderhasselt et al. (2013b) showed that anodal tDCS over the left DLPFC specifically decreased cognitive costs for positive relative to negative information. These emotion-specific effects on cognitive control are in line with the proposal of a side-lateralized activity of the prefrontal cortex in emotional processing (e.g., Davidson & Fox, 1982; Canli et al., 1998). According to this view, the cognitive processing of positive stimuli would be lateralized towards the left prefrontal cortex, and the cognitive processing of negative stimuli would be lateralized towards the right prefrontal cortex. In line with this theory, and as stated in our prior work (Vanderhasselt et al. (2013b), anodal tDCS over the right DLPFC would specifically modulate cognitive control for negative information, the latter being specifically important for rumination.1 Indeed, recent reviews show that stimulation effects are influenced by individual differences (for a review, see Krause & Kadosh, 2014), such as the use of habitual ruminative thinking style (i.e., trait rumination). In the present study, to investigate the causal question, a group of healthy volunteers reporting a broad range of trait rumination levels received both sham (placebo) and active anodal tDCS over the rDLPFC. Based on the abovementioned literature, we expected that 1

Because we expected effects of tDCS over the left DLPFC to have effects on positive emotional material, and this valenced material is not associated specifically to rumination tendencies, we had no hypotheses regarding the association with rumination in Vanderhasselt et al. (2013b).

anodal tDCS over the rDLPFC would facilitate cognitive control (e.g., less cognitive costs) for negative material, specifically in healthy individuals with higher tendencies to ruminate. This implies that the positive effects of tDCS over the rDLPFC on cognitive costs for negative information would be moderated by individual differences in trait rumination.

Method Participants Internet postings were used to recruit 35 (11M/24F) healthy volunteers with a mean age of 23.40 years (SD = 4.43) to participate in this study. All participants were (1) right handed, (2) had normal or corrected to normal (using glasses or contact lenses) vision, (3) females were not pregnant at the time of stimulation, and (4) had no metal in or around their scalp. Moreover, participants reported to have no current or past neurological disorder, and a current or past psychiatric disorder was excluded using the MINI screen (Sheehan et al., 1998; Overbeek, Schruers, & Griez, 1999). Three participants were excluded from the original sample (n=38) because of a current psychiatric disorder (n=2) or an irremovable nose piercing (n=1). Procedure The study protocol is in line with the World Medical Association's Declaration of Helsinki and was approved by the institutional ethics committee of the University Hospital Ghent (2014/0433). At the start of the study, written informed consent was obtained from all participants. A single blind, placebo controlled, crossover study design was used for which each participant received real and sham (placebo) stimulation of rDLPFC during two separated days. With an interval of at least one week between sessions (depending on the participant‟s availability, the maximum was 3 weeks) the order of both stimulation sessions (real tDCS and sham stimulation) was counterbalanced between participants. In each session,

participants first filled in the two trait questionnaires (see below), followed by the corresponding neuromodulation session for 20 minutes. After receiving the corresponding stimulation, participants performed the CECT. In addition, subjective mood ratings were recorded using Visual Analogue Scales (VAS), being administered at baseline and after task performance in each session. Material and apparatus CECT. The CECT was programmed in MATLAB 2012b software (The MathWorks Inc., Natick, MA, 2012). Eighteen faces (9F/9M) from the Karolinska Directed Emotional Faces dataset (Lundqvist, Flykt, & Öhman, 1997) were used as stimuli, comprising one frontal-view happy and one frontal-view sad expression each (i.e., 36 pictures). For more information regarding the stimuli that were used as well as an illustration of the CECT, we refer to Vanderhasselt et al. (2014). Each trial started with one of two word cues presented for 500 ms: “actual”, which instructed participants to press a key corresponding to the valence of the emotional expression in the upcoming target face (i.e., press “positive” for an upcoming happy face, press “negative” for an upcoming sad face); and “opposite”, which indicated that participants should make the response corresponding to the opposite valence of the emotional expression in the upcoming target face (i.e., press “negative” for an upcoming happy face, press “positive” for an upcoming sad face). Following the cue word, a black screen was presented for 1,500 ms. After this fixed inter-stimulus interval, either a happy or sad face was presented for 2,000 ms. Participants completed 20 practice trials using five faces not shown in the experimental blocks, followed by 4 blocks of 36 trials each. Each block contained nine trials of each cue/face combination (2 cues x 2 faces), resulting the total task in 36 trials per condition. Participants were instructed to respond as quickly and accurately as possible immediately after the face presentation, even though the face stayed on the screen after the response was made.

The assignment of labels (happy-sad) to the two buttons in the response box was counterbalanced across participants. Trait questionnaires. Because the interval between the two stimulation sessions could be more than two weeks for some participants, the trait questionnaires were administered at the start of each of the two experiment sessions (i.e., receiving real or sham stimulation). An average of the two scores was calculated, and used in our analyses. The Ruminative Response Scale (RRS, Nolen-Hoeksema & Morrow, 1991; Dutch translation by Raes & Hermans, 2007) was administered to measure trait rumination. The RRS consists of items that describe responses to a depressed mood, related to focussing on the self, on symptoms, and on the origin and consequences of the distress. This self-report questionnaire consists of 22 questions to which participants respond on a 4-point Likert scale how often they engage in these responses (i.e. 1 = almost never, 2 = sometimes, 3 = often, 4 = most of the times). Mean internal consistency of the RRS in this study was .91. The correlation between separate administrations of the test is high (.87), suggesting good test– retest reliability. The Beck Depression Inventory II (BDI-II; Beck, Steer, & Brown, 1996, Dutch translation by Van der Does, 2002) was administered to evaluate depressive symptoms. The BDI-II is a widely used self-report questionnaire consisting of 21 multiple choice format items (4 point scale), to assess the presence and severity of cognitive, motivational, affective, and somatic symptoms of depression. Past reports demonstrated established reliability and validity in clinical and non-clinical samples (Hautzinger, Bailer, Worall, & Keller, 1995). Mean internal consistency of the BDI-II in this study was .94. The correlation between separate administrations of the test is high (.87), suggesting good test–retest reliability. Mood Rating. In order to evaluate possible temporary changes in mood before (Tpre) versus after (Tpost) the CECT, mood ratings were administered using six visual analogue

scales (VAS) providing measures of fatigue, tension, anger, vigour, depression and cheerfulness (McCormack, Horne, Sheather, 1988). Participants were asked to describe how they felt „at that moment‟ by indicating on horizontal 100 cm lines whether they experienced the five above-mentioned mood states, from „ totally not‟ to „very much‟. tDCS application. Direct electrical current was applied by a saline-soaked pair of surface sponge electrodes (35 cm2) and delivered by a battery-driven stimulator (NeuroConn DC Stimulator). To stimulate the right DLPFC, the anode electrode was positioned centered over F4 according to the 10–20 international system for electroencephalogram electrode placement (vertical placement). The cathode was placed over the contra lateral supraorbital area (horizontal placement). A constant, direct current of 2 mA with 15 s of a ramped up and 15 s ramped down was applied for 20 min. For sham stimulation, the electrodes were positioned similar as when administering tDCS stimulation; however, the current was ramped down after 15 seconds. This procedure is a reliable sham condition (Nitsche et al., 2008). Data pre-processing In total, the CECT consisted of 144 trials, 36 trials per condition. We considered two response types within this data set, the reaction times (in ms) and the accuracy rates (in percentage). For the analysis of reaction times (RTs), only correct responses were considered, but overall, accuracy rates for all CECT trial types were high (83.49% - 88.25%). In accordance with our prior studies using this task (e.g., Vanderhasselt et al., 2014), median RTs for each of the four conditions were computed from the resulting valid trials with correct responses. Median RTs for each condition are described by the cue and then the facial emotion (e.g., “opposite/happy” refers to the RTs in trials with an “opposite” cue followed by a happy face, which would require pressing the button labelled with “sad”). For the accuracy (AC), we considered the percentage of correct trials within each condition (i.e., “actual/happy”, “actual/sad”, “opposite/happy”, “opposite/sad”).

Statistical plan The statistical analysis for the effects on mood was performed using SPSS software package (version 22), while the behavioural CECT data were analysed using R (version 3.2.3). In order to examine effects of tDCS stimulation in mood changes, a 2 x 2 repeated measures ANOVA was conducted, with Stimulation (sham, tDCS) and Time (pre, post) as within-subject factors for each VAS (i.e., fatigue, tension, anger, vigor, sadness and cheerful). For the CECT, the AC and RT data were both first analysed using mixed models, with Cue (Opposite=1, Actual=0), Emotion (Sad=1, Happy=0) and Stimulation (tDCS=1, sham=0) as binary-coded predictors, and Rumination (using the mean RRS score) as a continuous predictor. We opted for a mixed modelling approach instead of repeated measures ANOVA, as it offers more flexibility in deriving a parsimonious model (e.g., Gueorguieva and Krystal, 2004), as the one tested in this study. Given the overlap between rumination and depression, we also aimed to statistically control for individual differences in depressive symptoms, entering Depressive symptoms (using the mean BDI-II score) in the mixed models as a covariate. Therefore, we started from a mixed full model, including the four-way interaction between Cue x Emotion x Stimulation x Rumination (as well as the implied lower level interactions and main effects), while controlling for Depressive symptoms. Gradually, we simplified these models (one for AC and one for RT) through a backward selection procedure by removing the non-significant highest order interactions one-by-one (with a significance threshold of 0.05).

Results Self-reported questionnaire data Table 1 presents means and standard deviations in the whole sample for the RRS and the BDI-II at each of the two stimulation sessions, as well as for their averaged scores across the experiment. Both questionnaires were correlated with each other at each time point, ps <.001. For the exact correlations, we refer to Table 1. Effects on mood Table 2 presents the mean scores for each VAS scale at both pre and post stimulation assessment (for both active and sham stimulation conditions). Significant time effects were found for all VASs, all F’s > 6.07, all ps < .02, except for depression, F(1,28) = 1.07, p = .31, indicating increased fatigue, tension and anger and decreased vigor and cheerfulness from pre-stimulation to post-stimulation. However, no significant Stimulation by Time interaction emerged, F’s < 1.03, ps > .05. Therefore, tDCS (as compared to sham stimulation) over the right DLPFC did not differentially influence mood. Behavioural data – Accuracy Table 3 shows the descriptive statistics of the ACs in the four CECT trial types, in both stimulation conditions. For the backward selection procedure, we started from a full mixed model which included the four-way interaction Stimulation x Cue x Emotion x Rumination (as well as its implied lower level interactions and main effects), while also controlling for Depressive symptoms (i.e. BDI-II score). Eliminating all non-significant terms yielded a model with a significant two-way interaction: Cue x Emotion (p = .019). Of the main effects, only Cue (p= .033), Emotion (p=.038), and Stimulation (p= .006), were found to be significant, as the main effects of Rumination and Depressive symptoms all exhibited p-values larger than .12. The

significant two-order interaction Cue x Emotion revealed that for happy faces, opposite cues showed less accuracy than actual ones, while the inverse holds for sad faces (see Figure 1). Besides these moderated effects of Cue and Emotion, the accuracy is generally higher for sham compared to tDCS stimulation (when keeping all other predictors constant). Finally, for AC, trait rumination or depressive symptoms did not yield any significant interactions.

Behavioural data – Reaction Times Table 4 shows the descriptive statistics of the RTs in the four CECT trial types, in both stimulation conditions. For the backward selection procedure, we started from a full mixed model which included the four-way interaction Stimulation x Cue x Emotion x Rumination (as well as its implied lower level interactions and main effects), while also controlling for Depressive symptoms (i.e. BDI-II score). Eliminating all non-significant terms yielded a model with a significant three-way interaction: Stimulation x Cue x Rumination (p = .035). Besides this significant interaction, the remaining two-way interactions were all non-significant (all pvalues > .18). Of the main effects, only Cue was found to be significant (p < .001), as the main effects of Stimulation, Rumination and Depressive symptoms all exhibited p-values larger than .18. The significant highest-order interaction Stimulation x Cue x Rumination revealed that the differential effect of Stimulation by Cue varied across levels of rumination. To gain a better understanding of this interaction effect between Stimulation x Cue x Rumination, we calculated a median value of the Rumination score and categorised people as either being below or above this value. This way, a visual inspection could clarify these results of RTs at each cue condition by each stimulation condition for each rumination level (i.e., high trait ruminators, low trait ruminators). Those reaction time values in terms of Rumination, Cue and Stimulation conditions are depicted in Figure 2. It can be observed that

for low ruminators their RTs are qualified by the main effect of Cue, irrespective of the Stimulation condition (longer RTs following the opposite cue than following the actual cue), whereas for high ruminators this only holds following a sham Stimulation: in high ruminators, we observe a much smaller effect of Cue following the real Stimulation condition, compared to the sham (i.e., a smaller difference in reaction times between actual and opposite cues in the group after receiving real tDCS). To follow-up the above-described three-way interaction (i.e., effects in cognitive control for emotional information in general, so not valence-specific effects), we calculated a cognitive cost score of cognitive control for emotional information, by subtracting the median RTs following actual cues (over both emotions) from the median RTs following opposite cues (over both emotions). This cognitive cost score of cognitive control for emotional information was calculated for each individual separately. Based on the above-mentioned fitted model supporting an interaction between Stimulation and Rumination in cognitive costs for emotional information, we looked at the difference in this cognitive cost score between tDCS and sham Stimulation as a function of Rumination level. Therefore, a moderation analysis examining the interaction between Stimulation and Rumination on the differences in the cognitive cost score was performed. Although there was no effect of Stimulation on this score at average levels of Rumination (-10.95 [-51.50; 29.60], p = 0.60), Rumination proved to be a significant moderator of the effect of Stimulation on the differences in the cognitive cost score (-44.79 [-85.94; -3.65], p = .035). More specifically, compared to sham, real tDCS stimulation over the rDLPFC decreased the values of the cognitive cost score (i.e., reduced the difference in RTs between opposite and actual cues, thereby suggesting an increase in cognitive control) significantly more in high ruminators (when controlling for Depressive symptoms). These effects can be observed in Figure 3.

Discussion In this within-subjects, placebo controlled experiment, the effects of anodal tDCS of the rDLPFC on cognitive cost (as an indication of cognitive control) for emotional material were tested, taking the habitual tendency to ruminate into account. We expected a moderating role of trait rumination (controlling for the influence of depressive symptom levels) in the effects of tDCS in cognitive control for negative material, suggesting a causal role of dorsocortical correlates in cognitive control in high ruminators. First, no changes in mood states were observed after a single session of tDCS over the right DLPFC, which is completely in line with our prior studies (e.g., Vanderhasselt et al., 2013b) and a recent review on the effects on mood after single neurostimulation studies in healthy volunteers (Remue, Baeken, De Raedt, 2016). Only multiple-session tDCS protocols seem to influence mood and emotional reactivity, and such effects are specific for depressed patients (e.g., Vanderhasselt et al., 2016). Therefore, changes in mood did not interfere with effects of tDCS over the rDLPFC on cognitive control responses in the present study. It should be noted that mood was assessed after the experimental task, and not immediately after stimulation. In future studies mood could be assessed immediately after the stimulation (and thus before the experimental task) for a complete picture of fluctuations in mood. Anodal tDCS over the rDLPFC causally influenced cognitive costs as a function of the individuals‟ trait rumination level. This was the case for response times, but not for accuracy rates, possibly because accuracy was very high in general. The more individuals reported to ruminate in daily life, the less cognitive costs for emotional material (both positive and negative) following real neuromodulation of the rDLPFC. It should be noted that reaction times to both actual and opposite conditions did not differ significantly between the two stimulation sessions (also when controlling for individual differences in rumination and depression). On the other hand, the delta score of response times (and not the absolute scores)

to actual and opposite cues after real stimulation were related to individual differences in trait rumination. This means that high ruminators demonstrate less cognitive cost (i.e. the relative difference between actual and opposite trials), suggesting more cognitive control, for emotional information after real stimulation. It is an interesting finding that trait rumination scores moderated the influence of tDCS over the rDLPFC in cognitive costs for emotional material in general, rather than specifically for negative material. This was unexpected, as prior research has shown that ruminators have particular difficulties in inhibiting or disengaging from negative material (e.g., Duque et al., 2014). Moreover, neuroimaging data also show that healthy individuals who report a habitual tendency to ruminate display more neural activation in the rDLPFC while successfully inhibiting a habitual response towards negative information (Vanderhasselt et al., 2011). However, in a prior study we observed that tDCS over the rDLPFC, as compared to sham stimulation, led to impairments in attentional disengagement from both positive and negative faces, and not from a specific emotional stimulus (Sanchez, Vanderhasselt, Baeken, & De Raedt, 2016). In another line of research using non-emotional stimuli, trait rumination level in healthy volunteers has been found to be associated with deficits in inhibiting previously relevant mental sets (Whitmer & Banich, 2007) or impaired inhibition (De Lissnyder, Derakshan, De Raedt, & Koster, 2011), the latter for which the impairments were stronger for high versus low ruminators. Moreover, according to Martin & Tesser (1996), rumination is a cognitive style characterized by a general attentional inflexibility, associated with ineffective inhibition, leading to perseverative thoughts. Based on the current data, tDCS of the rDLPFC might therefore causally modulate cognitive control processes for different types of emotional materials in healthy individuals reporting a habitual tendency to ruminate in daily life, even though these ruminators have most difficulties to inhibit or disengage from negative material.

In our study, we used a sample of well-functioning/non psychiatric volunteers and statistically controlled for individual differences in depressive symptoms. The National Institute of Mental Health (NIMH) encouraged a dimensional approach to investigate the neural correlates of psychopathology, in order to better understand the development from healthy to high risk, and eventually to clinical depression. The current results suggest that healthy individuals who report higher tendencies to ruminate in daily life, and are thus more vulnerable for future depression (e.g., Nolen-Hoeksema, 2000), benefit from an additional boost/neuromodulation of the rDLPFC. Possibly, this indicates an inefficient recruitment of the DLPFC in habitual ruminators when having to inhibit or disengage from emotional material. Similarly, depressed patients have been found to show higher activity in cognitive control related neural regions (such as the DLPFC), even though their cognitive performance was similar as healthy volunteers (Harvey et al., 2005; Wagner et al., 2006; Langenecker et al., 2007). Also Wang and co-workers (2015) recently reported a negative correlation between ReHo in the DLPFC and individual differences in rumination, explained by these authors as an “inhibition compensation dysfunction” in healthy ruminators. This inefficient activation of the DLPFC might represent an early risk biomarker of cognitive control dysfunctions, resulting in sustained negative mood cycles, and hence the development of depression (De Raedt and Koster, 2010). At this moment, longitudinal studies are mandatory to confirm this presumed biomarker within a dimensional approach to investigate and understand the neural correlates of psychopathology. Possibly, as a prevention strategy, healthy individuals reporting higher tendencies to ruminate in daily life (vulnerable for future depression) may especially benefit from cognitive control training that activates the prefrontal network, and which is found to specifically reduce ruminative thoughts in depressed patients (e.g., Siegle et al., 2007, 2014; Vanderhasselt et al., 2015). In line with this reasoning, the current findings demonstrate that cognitive control enhancements for emotional material following dLPFC

stimulation was more manifest for participants reporting higher levels of trait rumination, confirming the idea that high ruminators benefitted more from the neuromodulation session to enhance cognitive control for emotional information. Notably, the present study revealed no direct (without taking into account rumination) main effects of a single session of anodal tDCS over the rDLPFC on response times for emotional information. These results contrast with the effects found for tDCS over the lDLPFC (Vanderhasselt et al., 2013b, using the same stimulation parameters). In this latter study, tDCS over lDLPFC resulted in augmented cognitive control for positive relative to negative information. This direct and emotion specific influence in cognitive control thus seems to appear specifically for stimulation of the lDLPFC, whereas this was not the case for tDCS over the rDLPFC. Further research is needed to investigate and explain these hemispheric specific results, and elucidate how they relate to individual differences in rumination. Moreover, further research is warranted to examine the effects of rDLPFC on cognitive control for non-emotional information. This is critical to rule out generalized effects in order to support the ultimate interpretation of a specific role of rDLPFC in the modulation of cognitive control for emotional information processing per se. Finally, because the effects are not ascribed specifically to the opposite cue condition, but to the relative difference between actual and opposite cues, further research is warranted to test the specific influence of cognitive control processes In conclusion, the results of the current study indicate that anodal tDCS over the rDLPFC influences cognitive control for emotional information as a function of individual differences in trait rumination. The data suggest that the DLPFC is causally involved in the modulation of cognitive control processes in healthy individuals who tend to ruminate, possibly to help them disengage from emotional material.

Acknowledgments This research was supported by a grant of the Research Foundation Flanders (FWO) awarded to Alvaro Sanchez and a grant BOF16/GOA/017 for a Concerted Research Action of Ghent University awarded to Rudi De Raedt and Chris Baeken. This work was also supported by the Ghent University Multidisciplinary Research Partnership “The integrative neuroscience of behavioral control”.

References Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck depression inventory (2nd ed.). San Antonio, TX: The Psychological Corporation. Brunoni, A.R., Vanderhasselt, M.A. (2014). Working memory improvement with noninvasive brain stimulation of the dorsolateral prefrontal cortex: a systematic review and metaanalysis. Brain & Cognition, 86, 1-9. DOI: 10.1016/j.bandc.2014.01.008. Campanella, S., Schroder, E., Monnart, A., Vanderhasselt, M. A., Duprat, R., Rabijns, M., ... & Baeken, C. (in press) Transcranial Direct Current Stimulation over the Right Frontal Inferior Cortex decreases neural activity needed to achieve inhibition: A double-blind ERP study in a male population. Clinical EEG and Neuroscience. Cohen, N., Mor, N.,& Henik, A. (2014). Training to link executive control and emotion – A training procedure to reduce rumination. Clinical Psychological Science, 3(1), 15-25. De Lissnyder, E., Koster, E. H. W., Derakshan, N., & De Raedt, R. (2010). The assocation between depressive symptoms and executive control impairments in response to emotional and non-emotional information. Cognition & Emotion, 24, 264−280. De Lissnyder, E., Derakhshan, N., De Raedt, R., & Koster, E. H.W. (2011). Depressive symptoms and cognitive control in a mixed antisaccade task: Specific effects of depressive rumination. Cognition and Emotion, 25, 886-897. De Raedt, R., & Koster, E. H. W. (2010). Understanding vulnerability for depression from a cognitive neuroscience perspective: a reappraisal of attentional factors and a new conceptual model. Cognitive Affective and Behavioral Neuroscience, 10, 50−70. Duque, A., Sanchez, A., & Vazquez, C. (2014). Gaze-fixation and pupil dilation in the processing of emotional faces: The role of rumination. Cognition & Emotion, 28(8), 1347–66. doi:10.1080/02699931.2014.881327 Gueorguieva, R., Krystal, J. H. (2004). Move over ANOVA: progress in analyzing repeatedmeasures data and its reflection in papers published in the Archives of General Psychiatry Archives of General Psychiatry, 61(3), 310-7. Harvey, P. O., Fossati, P., Pochon, J. B., Levy, R., Lebastard, G., Lehéricy, S., et al. (2005). Cognitive control and brain resources in major depression: An fMRI study using the n-back task. Neuroimage, 26(3), 860–869. Hautzinger, M., Bailer, M., Worall, H., & Keller, F. (1995). Beck-Depressions-Inventary (BDI). Test handbuch. Bern: Hans Huber. Jacobson, L., Javitt, D. C., & Lavidor, M. (2011). Activation of inhibition: diminishing impulsive behavior by direct current stimulation over the inferior frontal gyrus. Journal of Cognitive Neuroscience, 23, 3380-3387.
 
 Joormann, J., Yoon, K. L., & Zetsche, U. (2007). Cognitive inhibition in depression. Applied and Preventive Psychology, 12(3), 128–139. http://doi.org/10.1016/j.appsy.2007.09.002

Kühn, S., Vanderhasselt, M.-A., De Raedt, R., and Gallinat, J. (2014). The neural basis of unwanted thoughts during resting state. Social Cognitive and Affective Neuroscienece. 9(9), 1320-1324.doi: 10.1093/scan/nst117 Langenecker, S. A., Kennedy, S. E., Guidotti, L. M., Briceno, E. M., Own, L. S., Hooven, T., et al. (2007). Frontal and limbic activation during inhibitory control predicts treatment response in major depressive disorder. Biological Psychiatry, 62(11), 1272–1280. Lundqvist, D., Flykt, A., & Öhman, A. (1997). The Karolinska Directed Emotional Faces (KDEF). Department of Neurosciences, Karolinska Hospital, Stockholm. Martin, L., & Tesser, A. (1996). Some ruminative thoughts. In R. S. Wyer (Ed.), Advances in Social Cognition, Vol. 9, Hillsdale, NJ: Lawrence Erlbaum Associates. McCormack, H.M., Horne, D.J.L. and Sheather, S. (1988). Clinical application of visual analogue scales: a critical review. Psychological Medicine, 18, 1007–1019. Nitsche, M. A., Cohen, L., Wassermann, E. M., Priori, A., Lang, N., Antal, A., ... PascualLeone, A. (2008). Transcranial direct current stimulation: State of the art 2008. Brain Stimulation, 1(3), 206–23. http://doi.org/10.1016/j.brs.2008.06.004 Nitsche, M.A., Paulus, W. (2000). Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. Journal of Physiology, 527, 633–9. doi:10.1111/j.1469-7793.2000.t01-1-00633.x. Nolen-Hoeksema, S., & Morrow, J. (1991). A prospective study of depression and posttraumatic stress symptoms after a natural disaster: the 1989 Loma Prieta earthquake. Journal of Personality and Social Psychology, 61, 115-121. Nolen-Hoeksema, S. (1991). Responses to depression and their effects on the duration of depressive episodes. Journal of Abnormal Psychology, 100, 569–582, http://dx.doi.org/ 10.1037/0021-843X.100.4.569. Nolen-Hoeksema, S. (2000). The role of rumination in depressive disorders and mixed anxiety/depressive symptoms. Journal of Abnormal Psychology, 109(3), 504–511, http://dx.doi.org/10.1037/0021-843x.109.3.504. Overbeek, T., Schruers, K., & Griez, E. (1999). Mini International Neuropsychiatric Interview. Maastricht: University of Maastricht. Penolazzi, B., Stramaccia, D. F., Braga, M., Mondini, S., & Galfano, G. (2014). Human memory retrieval and inhibitory control in the brain: Beyond correlational evidence. Journal of Neuroscience, 34, 6606-6610.
 
 Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Janavs, J., Weiller, E., Keskiner, A., Schinka, J., Knapp, E., Sheehan, M. F., & Dunbar, G. C. (1998). The validity of the Mini International Neuropsychiatric Interview (MINI) according to the SCID-P and its reliability. European Psychiatry, 12, 232-241. Raes, F., & Hermans, D. (2007). The revised version of the Dutch Ruminative Response Scale. Unpublished instrument.

Remue, J., Baeken, C., & De Raedt, R. (2016). Does a single neurostimulation session really affect mood in healthy individuals?: A systematic review. Neuropsychologia, 85, 184-198. Sanchez, A., Vanderhasselt, M.A., Baeken, C., & De Raedt, R. (in press). Effects of tDCS over the right DLPC on attentional disengagement from positive and negative faces: An eyetracking study. Cognitive, Affective and Behavioral Neuroscience. Silas, J., & Brandt, K. R. (2016). Frontal transcranial direct current stimulation (tDCS) abolishes list-method directed forgetting. Neuroscience Letters, 616, 166-169. Stramaccia, D. F., Penolazzi, B., Sartori, G., Braga, M., Mondini, S., & Galfano, G. (2015). Assessing the effects of tDCS over a delayed response inhibition task by targeting the right Inferior Frontal Gyrus and right Dorsolateral Prefrontal Cortex. Experimental Brain Research, 233, 2283-2290. 
 
 Siegle, G.J., Ghinassi, F., & Thase, M.E. (2007). Neurobehavioral therapies in the 21st century: Summary of an emerging field and an extended example of cognitive control training for depression. Cognitive Therapy and Research, 31(2), 235–262, http://dx.doi.org/10.1007/s10608-006-9118-6. Siegle, G.J., Price, R.B., Jones, N.P., Ghinassi, F., Painter, T., & Thase, M.A. (2014). You gotta work at it: Pupillary indices of task focus are prognostic for response to a neurocognitive intervention for rumination in depression. Clinical Psychological Science, 2(4), 455–471, http://dx.doi.org/10.1177/2167702614536160. Van der Does, A. J. W. (2002). De Nederlandse versie van de Beck depression inventory Tweede Editie. The Dutch version of the Beck depression inventory - Second edition.Lisse: Swets & Zeitlinger B.V Vanderhasselt, M.A., Kuhn, S., & De Raedt, R. (2011). Healthy brooders employ more attentional resources when disengaging from the negative: an event-related fMRI study. Cognitive, Affective, & Behavioral Neuroscience, 11(2), 207–216, http://dx.doi. org/10.3758/s13415-011-0022-5. Vanderhasselt, M.A., De Raedt, R., Dillon, D.G., Dutra, S.J., Brooks, N., Pizzagalli, D.A. (2012). Decreased cognitive control in response to negative information in remitted depression: An Event Related Potential study. Journal of Psychiatry and Neuroscience, 37, 250-258. Doi: 10.1503/jpn.110089. Vanderhasselt, M.A., Baeken, C., Van Schuerbeek, P., Luypaert, R., De Maey, J., De Raedt, R. (2013a). How brooding minds inhibit negative material: an event-related fMRI study. Brain and Cognition, 81, 352-359. DOI: 10.1016/j.bandc.2013.01.007. Vanderhasselt, M.A., De Raedt, R., Brunoni, A.R., Campanhã, C., Baeken, C., Jonathan Remue, & J., & Boggio, P.S. (2013b). tDCS over the left prefrontal cortex enhances cognitive control for positive affective stimuli. PlosOne, 8(5), e62219. DOI: 10.1371/journal.pone.0062219. Vanderhasselt, M.A., De Raedt, R., De Paepe, A., Aarts, K., Otte, G., Van Dorpe, J., Pourtois, G. (2014). Abnormal proactive and reactive cognitive control during conflict processing in

major depression. Journal of Abnormal Psychology, 123(1), 68-80. DOI: 10.1037/a0035816. Vanderhasselt, M.A., De Raedt, R., Namur, V., Lotufo, P.A., Bensenor, I.M., Boggio, P.S., & Brunoni, A.R. (2015). Transcranial electric stimulation and neurocognitive training in clinically depressed patients: A pilot study of the effects on rumination. Progress in NeuroPsychopharmacology & Biological Psychiatry, 3, 57, 93-99. doi: 10.1016/j.pnpbp.2014.09.015. Vanderhasselt, M.A, De Raedt, R., Namur, V., Valiengo, L.C.L, Lotufo, P.A., Bensenor, I.M., Baeken, C., Boggio, PS., & Brunoni, A.R. (2016). Emotional reactivity to valence-loaded stimuli are related to treatment response of neurocognitive therapy. Journal of Affective Disorders. 190, 443–449. Doi: 10.1016/j.jad.2015.10.022. Wagner, G., Sinsel, E., Sobanski, T., Kohler, S., Marinou, V., & Mentzel, H. J. (2006). Cortical inefficiency in patients with unipolar depression: An event-related MRI study with the Stroop task. Biological Psychiatry, 59, 958–965. Wang, K., Wei, D., Yang, J., Xie, P., Hao, X., & Qiu, J. (2015). Individual differences in rumination in healthy and depressive samples: association with brain structure, functional connectivity and depression. Psychological Medicine, 45 (14), 2999-3008 Whitmer, A., & Banich, M. (2007). Inhibition versus switching deficits in different forms of rumination. Psychological Science, 18, 546-553.

Table 1: Rumination and depression scores on both experiment days, as well as the mean of both scores.

tDCS Session Sham Session Averaged AcrossSessions Score Pearson r correlation coefficient between repeated administrations ** p < .001

RRS Mean (SD) 40.57 (11.53) 42.14 (11.50) 41.36 (11.13)

BDI Mean (SD) 5.34 (8.16) 4.91 (7.42) 5.23 (7.16)

Pearson r correlation coefficient between questionnaires over stimulation sessions .57** .46** .52**

.87

.87

/

Table 2: VAS mood scores tDCS Session Mean (SD) Fatigue Tension Anger Vigour Depression Cheerful

pre 2,99 (1,90) 1,24 (1,55) 0,57 (0,71) 5,58 (2,11) 0,96 (1,28) 6,28 (2,05)

post 5,83 (2,54) 2,35 (2,69) 1,32 (2,12) 4,08 (2,32) 1,00 (1,43) 5,13 (2,10)

Sham Session Mean (SD) pre 2,88 (2,05) 2,00 (2,16) 0,87 (2,21) 5,72 (2,11) 1,04 (1,49) 6,20 (2,27)

post 5,36 (2,04) 2,82 (2,64) 2,23 (2,51) 4,67 (2,27) 1,18 (1,53) 5,01 (2,05)

Table 3: AC scores in the CECT

tDCS Session Median (SD) Actual Happy Actual Sad Opposite Happy Opposite Sad

86,67 (7,03) 85,08 (8,50) 83,49 (8,95) 86,14 (7,37)

Sham Session Median (SD) 88,25 (6,69) 85,93 (7,66) 87,41 (5,97) 87,09 (6,31)

Table 4: RT scores in the CECT

Actual Happy Actual Sad Opposite Happy Opposite Sad Opposite min Actual

tDCS Session Median (SD) 784,79 (225,10) 783,59 (209,51) 854,99 (244,07) 816,21 (215,48) 52,68 (80,02)

Sham Session Median (SD) 768,59 (222,39) 829,44 (246,05) 782,37 (199,57) 835,05 (222,62) 57,85 (58,73)

Figure 1: Accuracy (in percentage, with SE bars), for each combination of Cue (actual vs. opposite cues) and Emotion (sad vs. happy faces).

Figure 2: Median reaction times (in ms, with SE bars), for each combination of Cue and Stimulation, for low (blue) and high values (red) of Rumination.

Figure 3: Compound scores for cognitive control (Opposite Cues – Actual Cues), for tDCS and sham stimulations (red and blue, respectively), as a function of Rumination (with a 95% confidence band).

Figure 3 is a reduced form of Figure 2 because we subtracted the ‘Actual cue’ bars (the first and third bars of Figure 1) from the ‘Opposite cue’ bars (the second and last bars of Figure 2). Here, we do not see an effect of Stimulation for average levels of Rumination (averaging the blue and red bars provides the same compound score for a sham and tDCS Stimulation). We do, however, see that Rumination significantly moderates the effect of Stimulation on the compound scores: for high ruminators, the compound score significantly decreases when tDCS is administered, as compared to sham stimulation.