for peer reviewpeople.socsci.tau.ac.il/mu/galsheppes/files/2016/08/neural-processing... · for peer...

For Peer Review

Neural Processing of Emotional-intensity Predicts Emotion

Regulation Choice

Journal: Social Cognitive and Affective Neuroscience

Manuscript ID SCAN-15-619.R3

Manuscript Type: Original Manuscript

Date Submitted by the Author: n/a

Complete List of Authors: Shafir, Roni; Tel-Aviv University, School of Psychological Sciences Thiruchselvam, Ravi ; Hamilton College, Psychology Suri, Gaurav ; Stanford University, Psychology Gross, James; Stanford University, Psychology Sheppes, Gal; Tel Aviv University, School of Psychological Sciences

Keywords: Emotion Regulation Choice, Distraction, Reappraisal, Emotional-intensity,

Late Positive Potential

http://mc.manuscriptcentral.com/scan

Manuscripts submitted to Social Cognitive and Affective Neuroscience

For Peer Review

1

Running Head: NEURAL PREDICTION OF REGULATORY CHOICES

Neural Processing of Emotional-intensity Predicts Emotion Regulation Choice

Roni Shafir1, Ravi Thiruchselvam

2*, Gaurav Suri

3*, James J. Gross

3

&

Gal Sheppes1

1. Tel Aviv University, Tel Aviv, Israel

2. Hamilton College, Clinton, NY, United States

3. Stanford University, Stanford, CA, United States

Author Note

Roni Shafir and Gal Sheppes, The School of Psychological Sciences, Tel Aviv

University, and Sagol School of Neuroscience. Ravi Thiruchselvam, Department of

Psychology, Hamilton College. Gaurav Suri, and James J. Gross, Department of

Psychology, Stanford University.

*R. Thiruchselvam and G. Suri contributed equally to this work.

G. Sheppes is supported by Israel Science Foundation Grant No.1130/16.

We wish to thank Pavel Goldstein and Yoav Kessler for statistical consultation, and Shreya Lakhan Pal and Dustin Goerlitz for research assistance.

Corresponding Address:

Roni Shafir, or Gal Sheppes, Ph.D.

The School of Psychological Sciences

Tel Aviv University,

Tel Aviv 69978, Israel

Email: [email protected] , [email protected]

Keywords: Emotion Regulation Choice, Distraction, Reappraisal, Emotional-intensity, Late Positive Potential.

Word count: 4556

Abstract: 184

Figures: 3

Page 1 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

2

Abstract

Emotional-intensity is a core characteristic of affective events that strongly

determines how individuals choose to regulate their emotions. Our conceptual

framework suggests that in high emotional-intensity situations, individuals prefer to

disengage attention using distraction, which can more effectively block highly potent

emotional information, as compared to engagement reappraisal, which is preferred in

low emotional-intensity. However, existing supporting evidence remains indirect

because prior intensity categorization of emotional stimuli was based on subjective

measures that are potentially biased and only represent the endpoint of emotional-

intensity processing. Accordingly, the present study provides the first direct evidence

for the role of online emotional-intensity processing in predicting behavioral

regulatory-choices. Utilizing the high temporal resolution of Event Related Potentials

we evaluated online neural processing of stimuli's emotional-intensity (Late Positive

Potential, LPP) prior to regulatory-choices between distraction and reappraisal.

Results showed that enhanced neural processing of intensity (enhanced LPP

amplitudes) uniquely predicted (above subjective measures of intensity) increased

tendency to subsequently choose distraction over reappraisal. Additionally,

regulatory-choices led to adaptive consequences demonstrated in finding that actual

implementation of distraction relative to reappraisal-choice resulted in stronger

attenuation of LPPs and self-reported arousal.

Page 2 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

3

Introduction

Individuals have an impressive ability to actively modify their emotions using

various regulatory strategies. In recent years it has become clear that while a certain

strategy may prove adaptive in a particular emotional situation, it may lead to

maladaptive outcomes in another. Therefore, the ability to adequately choose between

various regulatory strategies in accordance with differing situational demands is

considered to be crucial for healthy adaptation (Aldao & Tull, 2015; Bonanno &

Burton, 2013; Gross, 2015; Sheppes & Levin, 2013; Sheppes et al., 2015; Webb et al.,

2012 for conceptual reviews).

Despite its conceptual importance, empirical examination of regulatory-

choices has only recently become a focus of interest in the field of emotion regulation

(Aldao & Nolen-Hoeksema, 2013; Bonanno et al., 2004; Sheppes et al., 2014a;

Sheppes et al., 2011). In several studies, a special focus was devoted to examine how

the evaluation of central characteristics of emotional situations leads individuals to

favor one regulatory strategy over the other. For example, the evaluation of how

aversive a dental procedure is likely to be can influence the regulatory-choice between

thinking about unrelated errands during the dental procedure versus contemplating the

long-term benefits of healthy teeth.

Our recently developed framework (Sheppes & Levin, 2013) demonstrates

how the evaluation of one central characteristic of emotional situations, namely their

intensity level, influences the choice between central regulatory strategies (Dixon-

Gordon et al., 2015; Reisenzein, 1994 for a review). Emotional-intensity determines

the degree of activation in response systems that form emotional responses (Bradley

et al., 2001). The regulatory strategies in our conceptual framework differ in one

central dimension, namely whether strategies generally involve disengaging via

Page 3 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

4

diverting cognition away from emotional situations, or engaging via sustained

attention to, or work on, emotional responses (Parkinson & Totterdell, 1999).

Specifically, one central disengagement regulatory strategy is distraction, that

involves disengaging from emotional contents by producing unrelated neutral

thoughts at an early attentional deployment stage (Van Dillen & Koole, 2007). One

central engagement regulatory strategy is reappraisal, that involves attentional

engagement with emotional information prior to re-interpreting its initial meaning at a

late semantic meaning stage (Gross, 1998; Sheppes & Gross, 2011).

According to our conceptual framework (Sheppes & Levin, 2013), varying

intensity levels of emotional events lead individuals to differentially choose between

disengagement distraction and engagement reappraisal. Particularly, disengagement

distraction is more likely to be chosen over reappraisal in high emotional-intensity

situations, because distraction can more effectively block potent emotional

information early before it gathers force (Shafir et al., 2015; Sheppes et al 2014b;

Sheppes & Gross, 2011; Sheppes & Meiran, 2007). By contrast, engagement

reappraisal is more likely to be chosen over distraction in low emotional-intensity

situations, because reappraisal effectively modulates mild emotional reactions, while

also altering how emotional events are perceived (Blechert et al., 2012; Denny et al.,

2015; Macnamara et al., 2011; Thiruchselvam et al., 2011; Wilson & Gilbert, 2008).

Strong empirical evidence for these regulatory-choices was demonstrated in

many behavioral studies (e.g., Hay et al., 2015; Levy-Gigi et al., 2015; Scheibe et al.,

2015; Sheppes et al., 2014; Sheppes et al., 2011). In a typical study, emotional

pictures that vary in their subjective emotional-intensity level (high versus low) are

presented in different trials. Each picture is followed by participants' choice of

Page 4 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

5

whether they prefer to regulate their emotions via disengagement distraction or

engagement reappraisal.

Although these prior behavioral studies provide significant evidence regarding

the role of emotional-intensity level in predicting which of the two regulatory

strategies is more likely to be chosen, this evidence is indirect. Particularly, in all of

these studies the intensity level categorization of pictorial emotional stimuli was

based on subjective ratings of intensity (IAPS normative ratings: Lang et al., 2008).

While important, subjective ratings suffer from several shortcomings. First, subjective

ratings are prone to multiple reporting biases. In our case, self-reports of emotional-

intensity may derive from tacit knowledge regarding how one should feel, rather than

how one actually feels in response to emotional pictures (e.g., Nisbett & Wilson,

1997; Rosenthal & Rosnow, 2009). Most importantly, self-report measures at best

only represent the endpoint or outcome of intensity information processing, rather

than actual underlying online processing of intensity information.

Therefore, the present study was set to provide the first direct evidence of

individuals' online processing of stimuli's emotional-intensity in emotion regulation

choice. Furthermore, examining online processing of emotional-intensity would prove

particularly useful if it could improve the prediction of subsequent regulatory-choices,

beyond subjective ratings.

In recent years, it has become apparent that neural measures can be utilized to

elucidate underlying processing of central attributes of stimuli, that can predict

various choice behaviors (see Fellows, 2004; Kable & Glimcher, 2009; Levy &

Glimcher, 2012; Rangel et al., 2008 for conceptual reviews). For example, recent

studies in neuro-economics applied underlying neural processing of products' value to

Page 5 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

6

predict products' chances to be subsequently chosen (e.g. Ravaja et al., 2013; Telpaz

et al., 2015. See Padoa-Schioppa, 2011; Fehr & Rangel, 2011 for conceptual reviews).

Drawing from this line of research, in the present study we examined the role

of online neural processing of emotional-intensity in predicting behavioral choices

between disengagement distraction and engagement reappraisal. To that end, we

employed Event Related Potentials (ERPs), which are temporally sensitive enough to

detect rapidly evolving emotional-intensity processing (Luck, 2014). Specifically, we

focused on the Late Positive Potential (LPP), perhaps the most well-known ERP

component for measuring online processing of emotional information that varies in

intensity (Hajcak et al., 2010 for a review). The LPP is a centro-parietal positive-

going wave that becomes evident approximately 300 ms following stimulus onset,

showing enhanced amplitudes as the processing of emotional-intensity increases

(Hajcak et al., 2009).

Consistent with our conceptual framework and previous behavioral findings,

we expected that subjective intensity level of emotional pictures (high versus low)

would strongly predict regulatory-choices. Specifically, low-intensity stimuli would

lead to reappraisal preference whereas high-intensity stimuli would lead to distraction

preference. Extending prior behavioral findings and consistent with our framework,

we expected that the neural intensity processing of emotional pictures (henceforth

pre-choice-LPP) would improve the prediction of subsequent regulatory-choices.

Particularly, an increase in pre-choice-LPP amplitudes would be uniquely associated

with higher chances of subsequently choosing distraction over reappraisal.

A secondary goal of the present study was to explore the consequences of

actually executing or implementing regulatory-choices. To that end, we examined

neural intensity processing during actual implementation of regulatory-choices

Page 6 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

7

(henceforth implementation-LPP), as well as self-reported arousal and unpleasantness

ratings immediately following regulatory implementation. Since the LPP reflects the

degree of emotional-intensity processing, its attenuation during regulatory

implementation denotes regulatory success (Dennis & Hajcak, 2009; Hajcak &

Nieuwenhuis, 2006; Moser et al., 2006). Congruent with previous studies showing

stronger LPP attenuation during distraction, as compared to reappraisal

implementation (Paul et al., 2013; Schönfelder et al., 2014; Thiruchselvam et al.,

2011), we hypothesized that distraction relative to reappraisal- choices'

implementation would result in enhanced attenuation of implementation-LPPs.

Correspondingly, Previous behavioral findings showed enhanced attenuation

of self-reported emotional-intensity following distraction, as compared to reappraisal

implementation (Sheppes et al., 2014b; Sheppes & Meiran, 2007). Accordingly we

predicted that distraction relative to reappraisal-choices' implementation would result

in enhanced attenuation of self-reported arousal and unpleasantness.

Method

Participants. Twenty five students participated in the experiment1. One

participant who did not complete the experiment was excluded from analyses.

Therefore, the final sample consisted of 24 participants (mean age 20.17 years, 14

men).

Stimuli. 144 negative pictures were chosen from previously validated pictorial

datasets2 (IAPS; Lang et al, 2008). High-intensity pictures (n=72, Marousal = 6.44,

Mvalence = 2.01) differed significantly from Low-intensity pictures (n=72, Marousal =

5.65, Mvalence = 2.85) in both valence and arousal subjective normative ratings (both

Fs > 42.2, ps < .001, ηp2s < .37, c.f. Sheppes et al., 2011, 2014a). Emotional pictures

Page 7 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

8

included various negative contents such as sadness, threat and fear, and were matched

across the two intensity categories when possible.

Procedure. Participants performed an emotion regulation choice task (Levy-

Gigi et al., 2015; Shafir et al., 2015; Sheppes et al., 2011; Sheppes et al., 2014a) while

continuous EEG was recorded. First, participants learned how to implement

distraction and reappraisal (order of learning counterbalanced across participants).

During this phase, the experimenter made sure that participants understood the

regulation strategies properly by asking them to talk out loud how they implement

each strategy. If needed, participants were corrected by the experimenter. Participants

then underwent an eight trial practice phase (four trials for each intensity category), in

which they practiced choosing between distraction and reappraisal.

Distraction instructions involved forming emotionally neutral thoughts that are

unrelated to the emotional picture (i.e., thinking about familiar streets or about

performing daily activities), thus disengaging attention from emotional contents.

Reappraisal instructions involved engaging with the emotional picture but

reinterpreting its emotional meaning (i.e., thinking that the situation would improve or

considering less negative elements of the emotional situation). We did not allow

participants to form reality challenge reappraisals (e.g., thinking of the situation as

unreal), because these reappraisals function as a form of disengagement (see Sheppes

et al., 2014a, Study 6 for details).

The task consisted of 144 trials (divided to 6 equally-long blocks, separated by

short breaks). Pictures of low and high emotional-intensity were presented randomly,

with the restriction that no more than 2 trials of the same emotional-intensity category

repeat in sequence.

Page 8 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

9

Each trial (see Figure 1) began with a 1500 ms fixation cross, followed by a

1000 ms initial preview of the emotional picture (henceforth "pre-choice window"),

followed by a 1500 ms fixation cross. Then a 3000 ms "Choice" screen was presented,

during which participants were asked to consider whether they prefer to regulate their

emotions via distraction or reappraisal. Participants were instructed to choose the

regulatory option which they assume would be more efficient in reducing their

negative emotions in reaction to each emotional picture. Another screen was then

presented and participants were asked to indicate their chosen strategy by pressing a

left or right button (assignment of button to strategy was counterbalanced across

trials). Following a 1500 ms fixation cross, the same picture was presented again for

5000 ms during which participants implemented their chosen strategy (henceforth

"implementation window"). The offset of each picture was followed by two 1-9 rating

scales in which participants reported their experienced valence (1= "pleasant", 9=

"unpleasant") and arousal (1= "calm", 9= "excited") using the Self-assessment

Manikin (SAM, Lang et al., 2008).

Electrophysiological Recordings, Data Reduction, and Analysis. Scalp

EEG was recorded using the Neuroscan recording system (Neuroscan, Inc., Herndon,

VA, USA) from 19 electrode sites (Fz, FCz, Cz, CPz, POz, Pz, Oz, FP1/2, F3/4, C3/4,

CP1/2, P3/4, O1/2), as well as two additional electrodes placed on the left and right

mastoids. During recording, Pz electrode served as the online reference. The vertical

electrooculogram (EOG) was recorded from two electrodes placed approximately 1

cm beneath and above the left eye. EEG data was sampled at 1000 Hz.

Offline signal processing was carried out using EEGLAB and the ERPLAB

Toolbox (Delorme & Makeig, 2004; Lopez-Calderon & Luck, 2014). Data from all

electrodes were re-referenced to the average activity of the left and right mastoids.

Page 9 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

10

Continuous EEG data was band-pass filtered (cutoffs: 0.05 – 20Hz; 12 dB/oct rolloff).

Eye-movement artifacts were removed using an Independent Component Analysis

(ICA) approach (Delorme & Makeig, 2004; Mennes et al., 2010).

For the pre-choice-LPP analysis, the EEG was epoched into 1200 ms

segments, starting 200 ms (baseline) before the onset of the pre-choice window and

lasting 1000 ms (end of the pre-choice window). Similarly, for the implementation-

LPP analysis, the EEG was epoched into 5200 ms segments, starting 200 ms

(baseline) before the onset of the implementation window and lasting 5000 ms (end of

the implementation window). All trials exceeding 80 µV within 200 ms were rejected.

The mean rejection rate was 0.75%, SE = 0.01 for the initial preview analysis and

0.76%, SE = 0.01 for the implementation window analysis, and did not vary between

conditions [both Fs < 1.69, n.s.].

The pre-choice-LPP was defined as the mean amplitude between 300 (when it

becomes evident, see Hajcak et al., 2010) and 1000 ms (end of the preview window).

The implementation-LPP analysis was defined as the mean amplitude between 300

and 5000 ms (end of the implementation window). The LPP was measured as the

average activity over centro-parietal electrode sites, where it is maximal (CPz, CP1

and CP2, e.g. Thiruchselvam et al., 2011).

Results

Does Neural Processing of Emotional-intensity Improve Regulatory-Choice

Prediction?

Based on our conceptual framework and prior findings, we expected that for

the high relative to the low subjective emotional-intensity category participants would

be more likely to choose distraction over reappraisal. Importantly, we hypothesized

that beyond the predictive value of subjective emotional-intensity categorization for

Page 10 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

11

regulatory-choices, increased pre-choice-LPPs would be associated with higher odds

of subsequently choosing distraction over reappraisal.

To test these two hypotheses, we applied a Logistic Mixed Effects Model

(using PROC GLIMMIX procedure in SAS 9.3). Binary regulatory-choices of all

participants in all trials, either to distract (receiving the value 1) or to reappraise

(receiving the value 0), were entered as a criterion variable, with single-trial

subjective emotional-intensity category (high or low, based on IAPS norms) and

single-trial (continuous) pre-choice-LPPs as predictors. In order to minimize type I

error (see recommendations by Barr, Levy, Scheepers, & Tily, 2013), random slopes

were defined for the two predictors (see also Supplementary Table S1 showing higher

fit for this model relative to alternative models). Since LPP amplitudes differ between

low and high emotional intensities, for each subject separately every single-trial pre-

choice-LPP value of a picture of a certain intensity category (either high or low) was

centered relative to the average pre-choice-LPP amplitude of the same intensity

category3. This procedure enabled us to compare between pre-choice-LPPs of

different intensity categories.

Confirming our hypothesis, we found a main effect of subjective emotional-

intensity category [b = 1.9, F(1,23) = 113.43, p < .001], with 6.68 increase of the odds

for choosing distraction following high versus low-intensity stimuli (OR = 6.68, p <

.001, 95% CI: [4.63, 9.71]). Importantly, confirming our second hypothesis we also

found a main effect of pre-choice-LPP amplitudes [b = 0.02, F(1,23) = 17.78, p <

.001], such that 1 microvolt (µV) increase in pre-choice-LPP amplitudes was uniquely

related to 2% increase of the odds for subsequently choosing distraction over

reappraisal (OR = 1.02, p < .001, 95% CI: [1.009, 1.028])4 (see Figure 2). Given the

relatively large µV range of single-trial pre-choice-LPPs, there is a considerable

Page 11 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

12

accumulative increase of the odds for choosing distraction over reappraisal as pre-

choice-LPP amplitudes increase. Note that we did not expect nor did we find an

interaction between the two predictors [F = 0.04, 95% CI: [-0.02, 0.02]].

In addition to conventional null hypothesis significance testing, which suffers

from several flaws (Cumming, 2013), we estimated the observed effects using

likelihood intervals5 (LIs). LIs consist of a set of possible values for the parameter of

interest that are consistent with the observed data. For example, the LI of the pre-

choice-LPP predictor consists of Beta values (i.e., values which define the

relationship between pre-choice-LPPs and regulatory-choices) supported by the data.

Values within the 1/8 LI (corresponding to a 95.9% confidence interval) are

consistent with the data, as they are no less than 1/8 as likely as the maximum-

likelihood estimate (i.e., the observed Beta) (Van Der Tweel, 2005; Dienes, 2008).

Since the log of the regression slope in logistic regressions is approximately normally

distributed, we used a LI calculation for a mean of normal distribution with unknown

variance (Dienes, 2008). Observed LIs for the pre-choice-LPP: 1/8 LI [0.009, 0.028],

for emotional-intensity category: 1/8 LI [1.51, 2.28], for the non-significant

interaction between the two predictors: 1/8 LI [-0.02, 0.02]).

Additional analyses were performed in order to mitigate concerns related to

the neural pre-choice-LPP measure. First, since pre-choice-LPPs vary considerably

across trials and subjects, we confirmed in complementary analyses that the observed

pre-choice-LPP findings are not driven by outliers at the trial (see Supplementary

Figure S6) or subject (see Supplementary Figure S7) level. Second, we wished to

mitigate the concern that our continuous pre-choice-LPP measure has predictive value

simply because it is more sensitive than the dichotomous subjective emotional-

intensity category measure. Specifically, we ran two Logistic Mixed Effects Models

Page 12 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

13

where we replaced the dichotomous subjective emotional-intensity measure with

continuous arousal (first analysis) and valence6 (second analysis) emotional-intensity

measures (i.e., the two continuous subjective measures that define the low and high

emotional-intensity categories). Congruent with findings obtained with the

dichotomous subjective emotional-intensity measure, we found a main effect of

continuous arousal [b = 0.76, 1/8 LI [0.64, 0.87], F(1,23) = 210.59, p < .001, OR =

2.13, p < .001, 95% CI: [1.91, 2.37]] and valence [b = 1.36, 1/8 LI [1.11, 1.62],

F(1,23) = 138.04, p < .001, OR = 3.91, p < .001, 95% CI: [3.07, 4.97]]. Importantly,

in these models the pre-choice-LPP remained a meaningful predictor [b = 0.01, 1/8 LI

[0.004, 0.02], F(1,23) = 10.89, p = .003, OR = 1.01, p < .001, 95% CI: [1.005, 1.02]]

in the model including continuous arousal; b = 0.01, 1/8 LI [0.005, 0.02], F(1,23) =

10.72, p = .003, OR = 1.01, p < .001, 95% CI: [1.005, 1.02] in the model including

continuous valence]. Note that in an additional analysis that included pre-choice-LPPs

as well as both continuous arousal and valence as regulatory-choice predictors, pre-

choice-LPP remained a meaningful predictor [b = 0.01, 1/8 LI [0.004, 0.02], F(1,23)

= 9.9, p = .005, OR = 1.01, p < .005, 95% CI: [1.004, 1.022]].

Do Regulatory-Choices Have Adaptive Consequences?

Attenuation of neural processing during implementation

In this analysis we tested whether distraction relative to reappraisal-choices'

implementation resulted in enhanced attenuation of implementation-LPPs. To equate

initial pre-choice-LPP differences between distraction-chosen and reappraisal-chosen

trials, for each condition the mean pre-choice-LPP amplitude served as a pre-

regulation baseline (see Figure 3a) from which we subtracted the mean

implementation-LPP amplitude (see Figure 3b)7. Thus, the outcome variable was

computed as (mean pre-choice-LPP) – (mean implementation-LPP), with higher

Page 13 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

14

scores indicating stronger attenuation of implementation-LPPs (i.e., stronger

regulatory success). We then employed a 2 x 2 analysis of variance (ANOVA) with

Emotional-Intensity Category (low, high) and Regulatory-Choice (distraction,

reappraisal) as repeated measures factors8. Confirming our hypothesis, we found a

Regulatory-Choice main effect [F(1,22) = 16.96, p < .001, ηp2 = .44, 1/8 LI [1.29,

4.18]], with higher attenuation of implementation-LPP amplitudes during distraction

(M = 3.97, SE = 1.11), as compared to reappraisal (M = 1.23, SE = 1.25)

implementation. Note that we did not expect nor did we find an Emotional-Intensity

Category X Regulatory-Choice interaction [F = 0.003, 95% CI: [-5.3, 5.05], 1/8 LI [-

5.51, 5.33]]. Additionally, this analysis revealed an Emotional-Intensity Category

main effect [F(1,22) = 6.66, p = .017, ηp2 = .23, 1/8 LI [0.33, 3.95]], with higher

attenuation of implementation-LPP amplitudes in high (M = 3.67, SE = 1.25) relative

to low (M = 1.53, SE = 1.16) intensity category, suggesting that regulatory

implementation was in general more efficient when facing high, as compared to low-

intensity stimuli.

Attenuation of self-reported post regulatory implementation ratings

In these analyses we tested whether distraction relative to reappraisal-choices'

implementation resulted in enhanced attenuation of self-reported arousal and

unpleasantness. Similar to the implementation-LPP analyses, the outcome variable

was computed as (mean IAPS normative pre-regulation arousal/valence6 ratings) -

(mean Subject's self-reported post-regulation arousal/valence ratings)7, with higher

scores indicating stronger reduction in arousal/unpleasantness (i.e., stronger

regulatory success). We then employed two 2 x 2 analyses of variance (ANOVAs) for

arousal and valence ratings separately, with Emotional-Intensity Category (low, high)

and Regulatory-Choice (distraction, reappraisal) as repeated measures factors8.

Page 14 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

15

Self-reported arousal analysis

Confirming our hypothesis that distraction relative to reappraisal-choice

implementation would result in reduced self-reported arousal levels, we found a

Regulatory-Choice main effect [F(1,22) = 28.72, p < .001, ηp2 = .57, 1/8 LI [0.6,

1.43]], with higher attenuation of self-reported arousal post distraction (M = 0.72, SE

= 0.25), as compared to reappraisal (M = -0.29, SE = 0.26) implementation. We did

not expect nor did we find an Emotional-Intensity Category X Regulatory-Choice

interaction [F = 0.31, 95% CI: [-1.2, 1.1], 1/8 LI [-1.22, 1.18]]. Additionally, this

analysis revealed an Emotional-Intensity Category main effect [F(1,22) = 104.98, p <

.001, ηp2 = .83, 1/8 LI [1.1, 1.7]], with higher reduction in self-reported arousal post

regulation of high (M = 0.92, SE = 0.23) relative to low (M = -0.48, SE = 0.26)

intensity stimuli, suggesting that regulatory implementation was generally more

efficient in reducing arousal levels of high, as compared to low-intensity stimuli.

Self-reported valence analysis

Somewhat unexpectedly the direction of the Regulatory-Choice main effect

[F(1,22) = 18.09, p < .001, ηp2 = .45, 1/8 LI [0.32, 1]] indicated that reappraisal

implementation was associated with higher reduction in self-reported unpleasantness

(M = 2.32, SE = 0.2), as compared to distraction (M = 1.67, SE = 0.18)

implementation. We did not expect nor did we find an Emotional-Intensity Category

X Regulatory-Choice interaction [F = 0.66, 95% CI: [-0.9, 0.8], 1/8 LI [-0.9 0.82]].

Additionally, this analysis revealed an Emotional-Intensity Category main effect

[F(1,22) = 5.46, p = .029, ηp2 = .2, 1/8 LI [0.02, 0.54]], with higher reduction in self-

reported unpleasantness post low (M = 2.13, SE = 0.17) relative to high (M = 1.86, SE

= 0.2) intensity, suggesting that regulatory implementation was generally more

Page 15 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

16

efficient in reducing self-reported unpleasantness of low relative to high emotional-

intensity stimuli.

Discussion

Although emotional-intensity is considered a central characteristic of

emotional events that strongly influences subsequent regulatory-choices, little is

known about the online processing of emotional-intensity that leads to choice. The

present study demonstrates for the first time that direct neural processing of stimuli's

emotional-intensity uniquely predicts behavioral regulatory-choices between

disengagement distraction and engagement reappraisal.

Consistent with our conceptual framework and previous behavioral findings

(Hay et al., 2015; Levy-Gigi et al., 2015; Scheibe et al., 2015; Sheppes et al., 2014a;

Sheppes et al., 2011), we found that subjective intensity level of emotional pictures

(high versus low) strongly predicted regulatory-choices. Specifically, in high relative

to low subjective emotional-intensity, disengagement distraction, which can more

effectively block highly potent emotional information early before it gathers force

(Sheppes & Gross, 2011), was more likely to be chosen, as compared to engagement

reappraisal. Extending these prior behavioral findings and consistent with our

framework, we showed that the neural intensity processing of emotional pictures

improved the prediction of regulatory-choices. Specifically, increased pre-choice-LPP

amplitudes were uniquely associated with higher odds for subsequent distraction

choice.

Additionally, in this study we tested the consequences of implementing

regulatory-choices. We predicted and found that distraction relative to reappraisal-

choice implementation resulted in stronger attenuation of implementation-LPPs as

well as self-reported arousal (but not unpleasantness).

Page 16 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

17

The present results enrich our original conceptual framework (Sheppes &

Levin, 2013) by directly elucidating the role of emotional-intensity in determining

regulatory-choices. In our original conceptual framework the intensity level predictor

of regulatory-choices was based on subjective intensity categorization of emotional

stimuli. While important, subjective intensity only represents the end point of

intensity information processing. Therefore, the present study provides the first

evidence for the role of direct neural intensity processing in predicting regulatory-

choices. Moreover, finding similar directionality in the neural and subjective

measures strengthens the conceptual relationship between enhanced intensity (either

neural or subjective) and distraction over reappraisal preference.

Relative to the strong effect of subjective emotional-intensity category on

regulatory-choices, the neural pre-choice-LPP measure was modest in its contribution.

However, we wish to make several empirical and conceptual arguments that together

increase our confidence in the validity of our neural predictor. Empirically, we were

able to show that the unique predictive value of pre-choice-LPPs could not be

attributed to the influence of trial or subject level outliers, or to the continuous nature

of the pre-choice-LPP measure. In addition, despite the relatively small odds ratio

value, the relatively large µV range of single-trial pre-choice-LPPs allows a

considerable accumulative increase of the odds for choosing distraction over

reappraisal as pre-choice-LPP amplitudes increase. Furthermore, converging support

for the validity of our neural measure was obtained with analyses that overcome

limitations associated with null hypothesis significance testing. Conceptually, the pre-

choice-LPP represents a direct online neural measure of emotional-intensity

processing. As such, the pre-choice-LPP may hint at distal causal relationships that

Page 17 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

18

are crucial for detecting the actual mechanism by which intensity processing

influences regulatory-choices (Lance & Vandenberg, 2009).

The present study extends the important contribution of neural measures in

understanding various classic choice behaviors, to a relatively new choice domain that

concerns behavioral choices in emotion regulation. While it shares a common logic

with other choice domains, the study of emotion regulation choice is also unique.

Neural processing of information in other choice domains usually predicts external

behavioral preferences, such as whether or not to buy a product, which career path to

pursue etc. By contrast, neural measures of information processing in emotion

regulation choice may serve as predictors of internal preferences for one way to

cognitively regulate emotions over another.

Our findings concerning the consequences of regulatory-choices showed the

expected enhanced attenuation of implementation-LPP amplitudes and self-reported

arousal ratings for distraction relative to reappraisal-chosen implementation. These

converging results reflect the notion that the LPP is highly sensitive to the level of

arousal elicited by emotional stimuli (Hajcak et al., 2010; Weinberg & Hajcak, 2010).

Additionally, both the LPP and the arousal measures conceptually reflect the degree

of activation of the defensive system which corresponds, according to some theories,

to emotional-intensity (Bradley et al., 2001). However, the self-reported

unpleasantness results were not consistent with the implementation-LPP and the

arousal results. Specifically, distraction, as compared to reappraisal-choice, resulted in

decreased attenuation of reported unpleasantness post-implementation. This pattern of

results may reflect the notion that while distraction involved producing neutral but not

positive thoughts, reappraisal may have involved reinterpreting negative emotional

Page 18 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

19

stimuli as more positive (e.g., "this man would eventually recover from his wounds".

See also McRae et al., 2010; Parkinson & Totterdell, 1999).

Several limitations of the present study should be mentioned. First, we did not

measure subjective ratings of stimuli's emotional-intensity on a trial-by-trial basis,

before participants made their regulatory-choices. It could be argued that using this

measure as a predictor of regulatory-choices could have further improved the

prediction of regulatory-choices. While possible, individual's self-reported emotional-

intensity ratings do not represent underlying online processing of intensity. Moreover,

having participants explicitly rate their perceived intensity in each trial may increase

the saliency of this measure and bias subsequent regulatory-choices.

Second, since we observed greater pre-choice-LPPs for distraction relative to

reappraisal choice, it could be argued that the greater subsequent reduction in

implementation-LPPs during distraction relative to reappraisal was due to regression

to the mean. More broadly, an exclusive regression to the mean argument would

predict that the higher the pre-choice-LPPs are, the greater reduction in

implementation-LPPs would be observed. However, the actual observed means

suggest that at least in some instances this is not the case. For example, while pre-

choice-LPPs for high-intensity reappraisal-choice (M = 6.02) were higher than those

of low-intensity distraction-choice (M = 5.28), less subsequent reduction was

observed in implementation-LPPs of high-intensity reappraisal-choice (M = 2.29)

relative to low-intensity distraction-choice (M = 2.88). In general, when participants

freely choose between regulatory strategies it is impossible to control for initial

intensity differences in distraction versus reappraisal chosen stimuli. Therefore, initial

intensity differences could influence post-choice implementation findings. However,

similar implementation results were obtained in a study that did not involve

Page 19 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

20

regulatory-choice, and thus controlled for initial intensity differences by randomly

assigning pictures to regulatory conditions (Shafir et al., 2015).

Page 20 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

21

Footnotes

1. Sample size was determined with reference to the vast majority of ERP studies

(including recent papers, e.g., Paul et al., 2013; Schönfelder, Kanske, Heissler,

& Wessa, 2014) in the field of emotion regulation. We concentrated on all of

the measures that were central to our theoretical predictions, all of which we

report. An additional measure of Visual Working Memory capacity (Cowan,

2001) was also collected for exploratory purposes.

2. Picture codes were as follows: Low-intensity: 207, 208, 209, 214, 215, 217,

219, 224, 225, 226, 227, 228, 246, 1090, 1201, 1205, 1303, 1930, 2095, 2100,

2110, 2120, 2130, 2141, 2205, 2375.1, 2661, 2683, 2688, 2691, 2692, 2694,

2700, 2710, 2800, 2810, 2900.1, 3160, 3181, 3220, 3230, 3300, 3350, 3500,

6211, 6260, 6300, 6312, 6313, 6315, 6550, 6560, 6570.1, 6821, 6830, 6831,

6834, 6840, 7359, 7380, 8231, 9006, 9041, 9220, 9402, 9421, 9435, 9480,

9594, 9800, 9810, 9920; High-intensity: 102, 105, 107, 112, 121, 123, 126,

127, 231, 232, 234, 235, 236, 237, 238, 239, 240, 242, 243, 244, 245, 247,

248, 250, 1111, 2352.2, 2730, 2981, 3000, 3005, 3010, 3015, 3030, 3051,

3053, 3060, 3062, 3063, 3064, 3068, 3069, 3071, 3080, 3100, 3101, 3102,

3110, 3120, 3130, 3140, 3150, 3168, 3170, 3261, 3266, 3301, 3400, 3550,

6212, 6555, 9040, 9250, 9252, 9253, 9265, 9300, 9301, 9400, 9405, 9420,

9433, 9570.

3. Results remain essentially unchanged when re-conducting this analysis

without centering pre-choice-LPP amplitudes.

4. In the main analyses reported we adopted a conservative approach that does

not involve excluding participants. However, it is important to note that eight

participants showed a very strong tendency to choose a certain regulatory

Page 21 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

22

option, which resulted in having less than 8 trials in some conditions – the

minimal recommended number of trials required to produce a stable LPP (see

Moran, Jendrusina, & Moser, 2013). However, re-conducting all analyses

reported in the main text while excluding these participants left all results

essentially unchanged.

5. In addition to likelihood intervals, we also performed Bayesian analyses for

the main results (see Supplementary Materials for complete details). Note that

the estimated Bayes Factors supported our key findings.

6. The normative valence scale was reversed so that higher scores indicated

higher unpleasantness, as in the valence scale used in our study.

7. The main results remain essentially unchanged when re-conducting the

implementation-LPP as well as the self-reported ratings analyses without

performing the aforementioned subtractions.

8. One subject that had no distraction-chosen trials in low-intensity was not

included in the ANOVA analyses.

Page 22 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

23

References

Aldao, A., & Nolen-Hoeksema, S. (2013). One versus many: capturing the use of multiple

emotion regulation strategies in response to an emotion-eliciting stimulus. Cognition

& Emotion, 27(4), 753–60.

Aldao, A., & Tull, M. T. (2015). Putting emotion regulation in context. Current Opinion in

Psychology, 3, 100–107.

Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013). Random effects structure for

confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language,

68(3), 255–78.

Blechert, J., Sheppes, G., Di Tella, C., Williams, H., & Gross, J. J. (2012). See what you

think: reappraisal modulates behavioral and neural responses to social stimuli.

Psychological Science, 23(4), 346–53.

Bonanno, G. A., & Burton, C. L. (2013). Regulatory flexibility: an individual differences

perspective on coping and emotion regulation. Perspectives on Psychological Science,

8(6), 591–612.

Bonanno, G. A., Papa, A., O’Neill, K., Westphal, M., & Coifman, K. (2004). The

importance of being flexible the ability to both enhance and suppress emotional

expression predicts long-term adjustment. Psychological Science, 15(7), 482–7.

Bradley, M. M., Codispoti, M., Cuthbert, B. N., & Lang, P. J. (2001). Emotion and

motivation I: defensive and appetitive reactions in picture processing. Emotion, 1(3),

276–98.

Cowan, N. (2001). The magical number 4 in short-term memory: a reconsideration of

mental storage capacity. Behavioral and Brain Sciences, 24(1), 87–114.

Cumming, G. (2013). The New Statistics: Why and how. Psychological Science, 25(1), 7–

29.

Delorme, A., & Makeig, S. (2004). EEGLAB: an open source toolbox for analysis of

single-trial EEG dynamics including independent component analysis. Journal of

Neuroscience Methods, 134(1), 9–21.

Dennis, T. A., & Hajcak, G. (2009). The late positive potential: a neurophysiological

marker for emotion regulation in children. Journal of Child Psychology and

Psychiatry, 50(11), 1373–83.

Denny, B. T., Inhoff, M. C., Zerubavel, N., Davachi, L., & Ochsner, K. N. (2015). Getting

over it: long-lasting effects of emotion regulation on amygdala response.

Psychological Science. 26(9), 1377-88.

Dienes, Z. (2008). Understanding psychology as a science: An introduction to scientific

and statistical inference. London, England: Palgrave Macmillan.

Page 23 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

24

Dixon-Gordon, K. L., Aldao, A., & De Los Reyes, A. (2015). Emotion regulation in

context: Examining the spontaneous use of strategies across emotional intensity and

type of emotion. Personality and Individual Differences, 86, 271–76.

Fellows, L. K. (2004). The cognitive neuroscience of human decision making: a review

and conceptual framework. Behavioral and Cognitive Neuroscience Reviews, 3(3),

159–72.

Gross, J. J. (1998). Antecedent-and response-focused emotion regulation: divergent

consequences for experience, expression, and physiology. Journal of personality and

social psychology, 74(1), 224-37.

Hajcak, G., Dunning, J. P., & Foti, D. (2009). Motivated and controlled attention to

emotion: time-course of the late positive potential. Clinical Neurophysiology, 120(3),

505–10.

Hajcak, G., MacNamara, A., & Olvet, D. M. (2010). Event-related potentials, emotion, and

emotion regulation: an integrative review. Developmental Neuropsychology, 35(2),

129–55.

Hajcak, G., & Nieuwenhuis, S. (2006). Reappraisal modulates the electrocortical response

to unpleasant pictures. Cognitive, Affective & Behavioral Neuroscience, 6(4), 291–7.

Kable, J. W., & Glimcher, P. W. (2009). The neurobiology of decision: consensus and

controversy. Neuron, 63(6), 733–45.

Lance, C. E., & Vandenberg, R. J. (2009). Statistical and methodological myths and urban

legends: Doctrine, verity and fable in the organizational and social sciences. Taylor

& Francis.

Lang, P., Bradley, M., & Cuthbert, B. (2008). International affective picture system

(IAPS): Affective ratings of pictures and instruction manual. Technical Report, A–8,

Gainesville,FL: University of Florida.

Levy, D. J., & Glimcher, P. W. (2012). The root of all value: a neural common currency

for choice. Current Opinion in Neurobiology, 22(6), 1027–38.

Levy-Gigi, E., Bonanno, G. a., Shapiro, a. R., Richter-Levin, G., Keri, S., & Sheppes, G.

(2015). Emotion regulatory flexibility sheds light on the elusive relationship between

repeated traumatic exposure and posttraumatic stress disorder symptoms. Clinical

Psychological Science, 1-12.

Lopez-Calderon, J., & Luck, S. (2014). ERPLAB: an open-source toolbox for the analysis

of event-related potentials. Frontiers in Human Neuroscience, 8, 213.

Luck, S. J. (2014). An Introduction to the Event-Related Potential Technique. 2nd Edition.

Cambridge, MA: MIT Press.

Page 24 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

25

Macnamara, A., Ochsner, K. N., & Hajcak, G. (2011). Previously reappraised: the lasting

effect of description type on picture-elicited electrocortical activity. Social Cognitive

and Affective Neuroscience, 6(3), 348–58.

McRae, K., Hughes, B., Chopra, S., Gabrieli, J. D. E., Gross, J. J., & Ochsner, K. N.

(2010). The neural bases of distraction and reappraisal. Journal of Cognitive

Neuroscience, 22(2), 248–62.

Mennes, M., Wouters, H., Vanrumste, B., Lagae, L., & Stiers, P. (2010). Validation of ICA

as a tool to remove eye movement artifacts from EEG/ERP. Psychophysiology, 47(6),

1142–50.

Moran, T. P., Jendrusina, A. a, & Moser, J. S. (2013). The psychometric properties of the

late positive potential during emotion processing and regulation. Brain Research,

1516, 66–75.

Moser, J. S., Hajcak, G., Bukay, E., & Simons, R. F. (2006). Intentional modulation of

emotional responding to unpleasant pictures: an ERP study. Psychophysiology, 43(3),

292–6.

Nisbett, R. E., & Wilson, T. D. (1997). Telling more than we can know: verbal reports on

mental processes. Psychological Review, 84(3), 231–59.

Parkinson, B., & Totterdell, P. (1999). Classifying Affect-regulation Strategies. Cognition

& Emotion, 13(3), 277–303.

Paul, S., Simon, D., Kniesche, R., Kathmann, N., & Endrass, T. (2013). Timing effects of

antecedent- and response-focused emotion regulation strategies. Biological

Psychology, 94(1), 136–42.

Rangel, A., Camerer, C., & Montague, P. R. (2008). A framework for studying the

neurobiology of value-based decision making. Nature Reviews Neuroscience, 9(7),

545-56

Ravaja, N., Somervuori, O., & Salminen, M. (2013). Predicting purchase decision: The

role of hemispheric asymmetry over the frontal cortex. Journal of Neuroscience,

Psychology, and Economics, 6(1), 1–13.

Reisenzein, R. (1994). Pleasure-arousal theory and the intensity of emotions. Journal of

Personality and Social Psychology, 67(3), 525–39.

Rosenthal, R., & Rosnow, R. L. (2009). Artifacts in behavioral research: Robert Rosenthal

and Ralph L. Rosnow’s classic books: a re-issue of Artifact in behavioral research,

Experimenter effects in behavioral research and The volunteer subject. New York:

Oxford University Press.

SAS Institute Inc, SAS/STAT® 9.2 user’s guide. The GLIMMIX procedure. SAS Institute

Inc., Cary, NC, 2008.

Page 25 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

26

Schönfelder, S., Kanske, P., Heissler, J., & Wessa, M. (2014). Time course of emotion-

related responding during distraction and reappraisal. Social Cognitive and Affective

Neuroscience, 9(9), 1310–19.

Shafir, R., Schwartz, N., Blechert, J., & Sheppes, G. (2015). Emotional intensity influences

pre-implementation and implementation of distraction and reappraisal. Social

Cognitive and Affective Neuroscience, 10, 1329-37.

Sheppes, G., Brady, W. J., & Samson, A. C. (2014b). In (visual) search for a new

distraction: the efficiency of a novel attentional deployment versus semantic meaning

regulation strategies. Frontiers in Psychology, 5, 346.

Sheppes, G., & Gross, J. J. (2011). Is timing everything? Temporal considerations in

emotion regulation. Personality and Social Psychology Review, 15(4), 319–31.

Sheppes, G., & Levin, Z. (2013). Emotion regulation choice: selecting between cognitive

regulation strategies to control emotion. Frontiers in Human Neuroscience, 7, 179.

Sheppes, G., & Meiran, N. (2007). Better late than never? On the dynamics of online

regulation of sadness using distraction and cognitive reappraisal. Personality & Social

Psychology Bulletin, 33(11), 1518–32.

Sheppes, G., Scheibe, S., Suri, G., & Gross, J. J. (2011). Emotion-regulation choice.

Psychological Science, 22(11), 1391–6.

Sheppes, G., Scheibe, S., Suri, G., Radu, P., Blechert, J., & Gross, J. J. (2014a). Emotion

regulation choice: a conceptual framework and supporting evidence. Journal of

Experimental Psychology: General, 143(1), 163–81.

Telpaz, A., Webb, R., & Levy, D. J. (2015). Using EEG to predict consumers’ future

choices. Journal of Marketing Research.

Thiruchselvam, R., Blechert, J., Sheppes, G., Rydstrom, A., & Gross, J. J. (2011). The

temporal dynamics of emotion regulation: an EEG study of distraction and

reappraisal. Biological Psychology, 87(1), 84–92.

Van der Tweel, I. (2005). Repeated looks at accumulating data: To correct or not to

correct?. European journal of epidemiology, 20(3), 205-11

Van Dillen, L. F., & Koole, S. L. (2007). Clearing the mind: a working memory model of

distraction from negative mood. Emotion, 7(4), 715–23.

Webb, T., & Gallo, I. S. (2012). Effective regulation of affect: An action control

perspective on emotion regulation. European Review of Social Psychology, 23(1),

143–86.

Weinberg, A., & Hajcak, G. (2010). Beyond good and evil: the time-course of neural

activity elicited by specific picture content. Emotion, 10(6), 767–82.

Page 26 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

27

Wilson, T. D., & Gilbert, D. T. (2008). Explaining away: A model of affective adaptation.

Perspectives on Psychological Science, 3(5), 370–86.

Page 27 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

1

Figures

Fig. 1. Trial Structure.

Trial structure of the emotion regulation choice task. SAM stands for Self-Assessment

Manikin.

Page 28 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

2

Fig. 2. Prediction of regulatory-choices based on subjective emotional-intensity

categories and pre-choice-LPPs.

Probability of choosing to regulate via distraction (y-axis) as a function of microvolt

(µV) increase in pre-choice-LPP amplitudes (x-axis), in low (continuous line) and

high (dotted line) subjective emotional-intensity categories. Density of the pre-choice-

LPPs and data from individual subjects are included in Supplementary materials

(Figures S4, S5, respectively). Note that for this Figure we used raw (non-centered)

pre-choice-LPP amplitudes.

Page 29 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

3

Fig. 3. Neural consequences of regulatory-choices.

Pre-choice-LPP amplitudes (A) and implementation-LPP amplitudes (that do not

involve subtraction from mean pre-choice-LPP amplitudes) (B) for distraction and

reappraisal-choice in low and high subjective emotional-intensity categories.

Waveforms are averaged across CPz, CP1 and CP2 electrodes. The x-axis runs from

the beginning of the baseline (-200 ms pre picture onset) to the end of the picture's

presentation – 1000 ms for the pre-choice window (A) and 5000 ms for the

implementation window (B).

Page 30 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

1

Supplementary Materials

Neural Processing of Emotional-intensity Predicts Emotion Regulation Choice

Comparison between alternative models.

Model

number

Intercept Emotional-

intensity

predictor

Pre-choice-

LPP

predictor

Interaction

between

predictors

BIC AIC

1 fixed+random fixed+random fixed+random no 3962.17 3956.28

2 fixed+random fixed+random fixed+random yes

(fixed+randm) 3964.75 3956.68

3 fixed only fixed+random fixed+random no 3978.35 3973.64

4 fixed+random fixed only fixed+random no 3993.37 3988.65

5 fixed+random fixed+random fixed only no 3965.17 3956.28

Table. S1. Estimation of alternative Logistic Mixed Effects Models to predict

regulatory-choices, using Bayesian Information Criterion (BIC) and Akaike

Information Criterion (AIC) values.

Lower BIC and AIC values are indicative of better model fit. This Table shows that

the characteristics of the model we report in the manuscript (model 1) show the best

fit relative to other models.

Page 31 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

2

Density of pre-choice-LPPs.

Fig. S4. Density of the pre-choice-LPP amplitudes in both low (A) and high (B)

emotional-intensity categories. In both panels the x-axis represents the range of pre-

choice-LPP amplitudes in microvolt (µV). The bottom histograms represent the

frequency of reappraisal-chosen trails, and the upper inverted histograms represent the

frequency of distraction-chosen trials (see also the right y-axis). For example, in panel

B that represents high-intensity pictures, there were approximately 226 reappraisal-

chosen trials and approximately 338 distraction-chosen trials in the 0 to 10 pre-

choice-LPP amplitude range. The red line represents the probability of choosing

distraction across pre-choice-LPP amplitudes (see also the left y-axis). For example,

in panel B that represents high-intensity, for the 20 to 40 pre-choice-LPP amplitude

range, the red line ranges between approximately 0.82 – 0.97 probability of

distraction-choice. The box plots next to each histogram represent the interquartile

range (between 25th

and 75th

percentile) of each distribution.

As can be seen, in both emotional intensities the majority of pre-choice-LPPs of

distraction-chosen trials were higher than those of reappraisal-chosen trials.

Page 32 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

3

Data from individual subjects.

Fig. S5. Individual actual proportions (circles) and predicted probabilities (triangles)

of choosing distraction (y-axis) in both low (A) and high (B) emotional-intensity

categories. The x-axis represents pre-choice-LPP amplitudes in microvolt (µV). For

each individual, the value on the x-axis represents their mean pre-choice-LPP

amplitudes in each emotional-intensity category (i.e., in graphs A and B, these values

represent mean pre-choice-LPP amplitudes for each individual in low and high

emotional-intensity, respectively). Each individual is represented by a different color.

Note that the distance between actual proportions and predicted probabilities provides

visual representation of model fit.

Page 33 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

4

Bayesian analyses.

Does Neural Processing of Emotional-intensity Improve Regulatory-Choice

Prediction?

Bayes factors (BFs) were calculated to compare our model, which contains both

subjective emotional-intensity category and pre-choice-LPP amplitude as regulatory-

choice predictors, with two alternative models containing each of these predictors

alone. BFs were calculated based on BIC approximation, where differences in BICs

are converted into an approximate BF (e.g., Raftery, 1995; Wagenmakers, 2007 for

details). BFs were interpreted based on Jeffreys (1961) interpretation guidelines.

Supporting our pre-choice-LPP effect, BFboth predictors/emotional-intensity category only = 4.48

(interpreted as "substantial support") suggested that the data were 4.48 times more

likely to occur under a model including both predictors (emotional-intensity category

and pre-choice-LPP), rather than a model including only emotional-intensity category.

Supporting our emotional-intensity category effect, BFboth predictors/ pre-choice-LPP

amplitudes only = 335120256 (interpreted as "decisive support") suggested that the data

were 335120256 times more likely to occur under a model including both predictors

(emotional-intensity category and pre-choice-LPP), rather than a model including

only the pre-choice-LPP.

Do Regulatory-Choices Have Adaptive Consequences?

Bayesian repeated measures analyses of variance (ANOVAs) ware

conducted using JASP (version 0.7.5.5. See JASP, 2014).

Page 34 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

5

Modulation of neural processing during implementation

For the implementation-LPP analysis, an estimated BF suggested that the

data were 6.24 times more likely to occur under the Emotional-Intensity main effect

than the null model (interpreted as reflecting "substantial support" for H1).

Additionally, the Regulatory-Choice main effect model was preferred over the null

model (BF = 68.77, "very strong support"). Finally, BF suggested that the data were

3.68 times more likely to occur under a model including two main effects (Emotional-

Intensity and Regulatory-Choice), rather than a model including two main effects and

an interaction ("substantial support").

Modulation of self-reported post regulatory implementation arousal

For the self-reported arousal analysis, an estimated BF suggested that the


than the null model ("strong support"). Additionally, the Regulatory-Choice main

effect model was preferred over the null model (BF = 929.79, "decisive support").

Finally, BF suggested that the data were 1.6 times more likely to occur under a model

including two main effects (Emotional-Intensity and Regulatory-Choice) and an

interaction, rather than a model including two main effects ("anecdotal support").

Although this result is not consistent with our prediction, we do not discuss it further

because it is anecdotal.

Modulation of self-reported post regulatory implementation valence

For the self-reported valence analysis, an estimated BF suggested that the


than the null model ("anecdotal support"). Additionally, the Regulatory-Choice main

effect model was preferred over the null model (BF = 15235.89, "decisive support").

Page 35 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

6

Finally, BF suggested that the data were 2.6 times more likely to occur under a model

including two main effects (Emotional-Intensity and Regulatory-Choice), rather than

a model including two main effects and an interaction ("anecdotal support").

Page 36 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

7

Single-trial & single-subject level outliers.

Cook's distance function of the R package ‘influence.ME’ was used to identify

outliers at the single-trial level as well as the single-subject level.

Fig. S6. Single trial-level Cook’s distance. The x-axis represents single-trial Cook's

distance values. In our case, the cut-off value of Cook’s distance (4/n) equals

4/3425=0.001167. The y-axis represents each of the 3425 single-trials in the data,

arranged by their Cook's distance values (smallest to largest). As can be seen, the

results seem to be most strongly influenced by 10 observations (represented by the red

triangles) that were identified as outliers.

To evaluate trial outlier influence we re-ran the Logistic Mixed Effects Model (using

PROC GLIMMIX procedure in SAS 9.3) excluding these 10 trials, and results

remained essentially unchanged. Specifically, we found a main effect of subjective

emotional-intensity category [b = 1.89, 1/8 LI [1.5, 2.28], F(1,23) = 112.2, p < .001,

Page 37 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

8

OR = 6.65, p < .001, 95% CI: [4.59, 9.62]] as well as pre-choice-LPP amplitudes [b =

0.02, 1/8 LI [0.01, 0.03], F(1,23) = 17.5, p < .001, OR = 1.02, p < .001, 95% CI:

[1.009, 1.028]]. Additionally, the sigtest function of the R package ‘influence.ME’

was used to estimate the influence of each single-trial on the observed results. The

analysis confirmed that none of the single trials influenced the significance of the

effects we originally observed (i.e., the subjective emotional-intensity category and

the pre-choice-LPP effects).

Fig. S7. Single-subject level Cook’s distance. The x-axis represents single-subject

Cook's distance values. In our case, the cut-off value of Cook’s distance (4/n) equals

4/24=0.16667. The y-axis represents each of the 24 single subjects, arranged by their

Cook's distance values (smallest to largest). As can be seen, the results seem to be

most strongly influenced by one subject (number 14, represented by the red triangle)

who was identified as an outlier.

Page 38 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

9

Similar to the previous trial-level analyses, to evaluate subject outlier influence we

first re-ran the Logistic Mixed Effects Model (using PROC GLIMMIX procedure in

SAS 9.3) excluding subject 14, and results remained essentially unchanged.

Specifically, we found a main effect of subjective emotional-intensity category [b =

1.9, 1/8 LI [1.49, 2.28], F(1,22) = 108.44, p < .001, OR = 6.59, p < .001, 95% CI:

[4.53, 9.6] as well as pre-choice-LPP amplitudes [b = 0.02, 1/8 LI [0.01, 0.03],

F(1,22) = 16.26, p < .001, OR = 1.02, p < .001, 95% CI: [1.009, 1.028]]. Additionally,

the sigtest function of the R package ‘influence.ME’ was used to estimate the

influence of each single subject on the observed results. This function tests whether

excluding the influence of each single subject changes the statistical significance of

any of the predictors in the model. The analysis confirmed that none of the single

subjects influenced the significance of the effects we originally observed (i.e., the

subjective emotional-intensity category and the pre-choice-LPP effects).

Page 39 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review

10

References

JASP. (2014). Retrieved from https://jasp-stats.org/

Jeffreys, H. (1961). Theory of probability. New York: Oxford University Press.

Raftery, A. E. (1995). Bayesian Model Selection in Social Research. Sociological

Methodology, 25, 111–64.

SAS Institute Inc, SAS/STAT® 9.2 user’s guide. The GLIMMIX procedure. SAS

Institute Inc., Cary, NC, 2008.

Wagenmakers, E. J. (2007). A practical solution to the pervasive problems of p

values. Psychonomic Bulletin & Review, 14(5), 779–804.

Page 40 of 40



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

for peer reviewpeople.socsci.tau.ac.il/mu/galsheppes/files/2016/08/neural-processing... · for peer...

Documents