1

NEURAL CORRELATES OF STRESS RECOVERY

IN A POSITIVE EMOTIONAL CONTEXT

BY

XI YANG

A Thesis Submitted to the Graduate Faculty of

WAKE FOREST UNIVERSITY GRADUATE SCHOOL OF ARTS AND SCIENCES

in Partial Fulfillment of the Requirements

for the Degree of

MASTER OF ARTS

Psychology

May 2017

Winston-Salem, North Carolina

Approved By:

Christian E. Waugh, PhD, Advisor

Terry D. Blumenthal, PhD, Chair

Eric R. Stone, PhD

Fadel Zeidan, PhD

ii

ACKNOWLEDGMENTS

I would first like to thank my advisor Dr. Christian Waugh. The door to Dr.

Waugh’s office was always open whenever I ran into a trouble spot or had a question

about my research or writing. He consistently allowed this paper to be my own work, but

steered me in the right direction whenever he thought I needed it. He spent dozens of

hours editing the script that we used for fMRI data analysis. Somehow the bubbly state he

was always in when debugging made me less fearful of debugging programs in future.

I would also like to thank the experts who were involved in the project: Dr. Youngkyoo

Jung, Dr. Benjamin Wagner, Dr. Michael Tobia, Dr. Kateri McRae, Dr. Tor Wager, and

Ms. Debra Fuller. I also greatly appreciate all the help and effort from my labmates and

participants.

I would also like to acknowledge Dr. Terry Blumenthal, Dr. Eric Stone and Dr.

Fadel Zeidan as readers of this thesis, and I am gratefully indebted to them for their

patience, kindness and very valuable comments on this thesis.

Finally, I must express my very profound gratitude to my parents, to my friends

and every other little thing for providing me with sustaining positive emotions, unfailing

support, and continuous encouragement throughout my years of study and through the

process of researching and writing this thesis. This accomplishment would not have been

possible without them. Thank you.

iii

TABLE OF CONTENTS

List of Illustrations and Tables ........................................................................................ v

List of Abbreviations ..................................................................................................... viii

Abstract .............................................................................................................................. x

Introduction ....................................................................................................................... 1 Top-Down Route: Positive Emotions Enhance Cognitive Flexibility and Openness ............. 3 Bottom-Up Route: Positive Emotions Bias and Provide Positive Information ....................... 7 The Present Study ................................................................................................................. 10

Method .......................................................................................................................................... 14

Participants ............................................................................................................................ 14 Behavioral pilot. .............................................................................................................................. 14 fMRI. ............................................................................................................................................... 14

Procedure............................................................................................................................... 16 Pre-scanner preparation. .................................................................................................................. 16 fMRI data acquisition. ..................................................................................................................... 16 fMRI data preprocessing. ................................................................................................................ 16 Scanner tasks. .................................................................................................................................. 17 Stress manipulation task. ................................................................................................................. 17 Experimental manipulation task. ..................................................................................................... 18 Experiment memory task. ............................................................................................................... 20

Materials ................................................................................................................................ 20 Mood ratings. .................................................................................................................................. 20 Post-task thoughts content and personality questionnaires. ............................................................ 21

Modified Decentering Questionnaire. ........................................................................................ 21 Thought Content Questionnaire. ................................................................................................ 22 Emotion Regulation Questionnaire. ........................................................................................... 23 Ego-Resiliency Scale. ................................................................................................................. 23 Extraversion and Neuroticism subscales from Neuroticism-Extraversion-Openness Personality Inventory. ................................................................................................................................... 23 Ruminative Response Scale. ...................................................................................................... 24

Data Analyses ............................................................................................................................... 24

fMRI Data ............................................................................................................................. 24 Emotion mask. ................................................................................................................................ 24 Change-point analysis. .................................................................................................................... 25 Inverse logit hemodynamic response modeling event-related analysis. .......................................... 27

Results ........................................................................................................................................... 29

Behavioral Pilot Data ............................................................................................................ 29 Stress manipulation check. .............................................................................................................. 29

Pleasant mood ratings. ................................................................................................................ 29 Unpleasant mood ratings. ........................................................................................................... 29

Experimental manipulation check. .................................................................................................. 29 NE recovery. ................................................................................................................................... 30

fMRI Study Behavioral Data................................................................................................. 31 Stress manipulation check. .............................................................................................................. 31

iv

Pleasant mood ratings. ................................................................................................................ 32 Unpleasant mood ratings. ........................................................................................................... 32

Experimental manipulation check. .................................................................................................. 33 PE recovery. ............................................................................................................................... 33 Thought content.......................................................................................................................... 34

NE recovery. ................................................................................................................................... 34 Multiple regression analyses. .......................................................................................................... 35 Mediation analyses with mood ratings and thought content. .......................................................... 38

PE recovery mediation: PosVT and NE recovery. ..................................................................... 38 PE recovery mediation: decentering and NE recovery. .............................................................. 39

Exploratory analyses in personality variables. ................................................................................ 40 fMRI Data – Change-Point Analyses .................................................................................... 41

Main effect of time - Overall intercept. ........................................................................................... 41 Stress reactivity. .............................................................................................................................. 44

NE: Low Stress Reactivity vs. High Stress Reactivity. .............................................................. 44 PE: Low Stress Reactivity vs. High Stress Reactivity. ............................................................... 45

Experimental manipulation. ............................................................................................................ 48 Positive Video vs. Neutral Video. .............................................................................................. 48

Mood recovery as covariates. .......................................................................................................... 49 Low PE Recovery vs. High PE Recovery. ................................................................................. 49 Low NE Recovery vs. High NE Recovery. ................................................................................ 52

Thought content as covariates. ........................................................................................................ 53 Low PosVT vs. High PosVT. ..................................................................................................... 53 Low Decentering vs. High Decentering. .................................................................................... 55

Personality variable as covariates. .................................................................................................. 55 Low Rumination vs. High Rumination. ..................................................................................... 55

Mediation Analyses......................................................................................................................... 57 fMRI Data - Event-Related Analysis of Re-Exposure to Anagrams ..................................... 60

Discussion ..................................................................................................................................... 63

References ..................................................................................................................................... 69

Appendix A ...................................................................................................................... 86

Appendix B ...................................................................................................................... 89

Appendix C ...................................................................................................................... 91

Appendix D ...................................................................................................................... 93

Appendix E ...................................................................................................................... 95

Appendix F ...................................................................................................................... 97

Curriculum Vitae ............................................................................................................ 99

v

LIST OF ILLUSTRATIONS AND TABLES

Figure 1. The Schematic Flow Chart For The Scanning Procedures .......................... 17

Figure 2. A Hard Level Of Difficulty Anagram From The First Word List And The Correct Answer Is 14523 “Opera” ................................................ 19 Figure 3. Three-Minute Reflection And Recovery Period With Positive “Waves” Video (Left) And Neutral “Sticks” Video (Right). ...................... 19 Figure 4. Time Points When Mood Ratings Were Collected In The Scanner. ........... 21 Figure 5. Time Points Information For Change-Points Categorization. ..................... 26 Figure 6. Pleasant Mood Ratings At Different Time Points (N = 89). ....................... 30 Figure 7. Unpleasant Mood Ratings At Different Time Points (N = 89). ................... 31 Figure 8. Correlation Between PE Recovery And NE Recovery (N = 89). ................ 32 Figure 9. Pleasant Mood Ratings At Different Time Points (N = 43). ....................... 33 Figure 10. Unpleasant Mood Ratings At Different Time Points (N = 43). ................... 35 Figure 11. Correlation Between PE Recovery And NE Recovery (N = 43). ................ 36 Figure 12. PE Recovery Mediation Between Positive Video Thoughts And NE Recovery Model............................................................................. 39 Figure 13. PE Recovery Mediation Between Decentering Thoughts And NE Recovery Model. ................................................................................... 40 Figure 14. NE Stress Reactivity Radiologist View Of The Medial Plane Of The Brain. ............................................................................................... 46 Figure 15. PE Stress Reactivity Radiologist View Of The Medial Plane Of The Brain.. .............................................................................................. 48 Figure 16. Radiologist View Of The Medial Plane Of The Brain. Blue Means Positive Correlation Between Pe Recovery And Regional Activation. ....... 51 Figure 17. vmPFC, dmPFC, pACC, dACC And Pre-Sma Activation Time Course At .05 Ratio Criterion. ................................................................................. 51

vi

Figure 18. Positive Video Thoughts Radiologist View Of The Medial Plane Of The Brain. ............................................................................................... 54 Figure 19. vmPFC Activation Time Course At .05 Ratio Criterion.............................. 55 Figure 20. Decentering Radiologist View Of The Medial Plane Of The Brain. ........... 56 Figure 21. Ruminative Response Scores Radiologist View Of The Medial Plane Of The Brain. ..................................................................................... 58 Figure 22. PE Recovery Mediation Between vmPFC Activation And NE Recovery Model. .......................................................................................................... 58 Table 1. Modified Experience And Thought Content Questionnaires Subscales Average By Video Valence. ........................................................................ 34 Table 2. Correlation Matrix Among PE Recovery, NE Recovery And Modified Experience And Thought Content Questionnaires Subscales. ..................... 37 Table 3. Summary Of Multiple Regression Analysis For PE Recovery. ................... 37 Table 4. Summary Of Multiple Regression Analysis For NE Recovery. .................. 38 Table 5. Correlation Matrix Among Personality Measures, PE Recovery And NE Recovery. ............................................................................................... 42 Table 6. Change-Point Analysis Of The Bold Responses To The Anagram Stressor And Reflection With Video – Average (Intercept) Across All Participants. .................................................................................................. 43 Table 7. Change-Point Analysis Of The Bold Responses To The Anagram – Regression Based On Post Anagram NE Change. ....................................... 45 Table 8. Change-Point Analysis Of The Bold Responses To The Anagram – Regression Based On Post Anagram PE Change. ....................................... 47 Table 9. Change-Point Analysis Of The Bold Responses To The Video – Regression Based On Video Condition. ...................................................... 49 Table 10. Change-Point Analysis Of The Bold Responses – Regression Based On PE Recovery ........................................................................................... 50 Table 11. Change-Point Analysis Of The Bold Responses – Regression Based On NE Recovery. ......................................................................................... 53

vii

Table 12. Change-Point Analysis Of The Bold Responses – Regression Based On Positive Video Thoughts. ....................................................................... 54 Table 13. Change-Point Analysis Of The Bold Responses – Regression Based On Decentering. ........................................................................................... 56 Table 14. Change-Point Analysis Of The Bold Responses – Regression Based On Ruminative Response Scale. .................................................................. 57 Table 15. PE Recovery Mediation Of Relationship Between vmPFC Activation And NE Recovery Based On Positive Video Thoughts Contrast. ............... 59 Table 16. Average Ratios Of Correctly Recognized Anagrams Of Different Event Types. ................................................................................................ 61

viii

LIST OF ABBREVIATIONS

AIDS acquired immune deficiency syndrome

dACC dorsal anterior cingulate cortex

DES dorsal executive system

dlPFC dorsal lateral prefrontal cortex

dmPFC dorsal medial prefrontal cortex

EPI Echo Planar Imaging

FOV field of view

lPFC lateral prefrontal cortex

MNI Montreal Neurological Institute

mPFC medial prefrontal cortex

MPRAGE Magnetization-Prepared Rapid Gradient-Echo

NaC nucleus accumbens

OFC orbitofrontal cortex

pACC pregenual anterior cingulate cortex

PCASL Pseudo-Continuous Arterial Spin Labeling

PCC posterior cingulate cortex

PTSD posttraumatic stress disorder

SMA supplementary motor area

SPM Statistical Parametric Mapping

TE echo time

TR repetition time

ix

VAS ventral affective system

vmPFC ventral medial prefrontal cortex

vlPFC ventral lateral prefrontal cortex

x

ABSTRACT

Despite accruing evidence showing that positive emotions facilitate stress recovery, the

neural basis for this effect remains unclear. To identify the mechanism underlying the

beneficial effects from positive emotions on stress recovery, we compared stress recovery

for people reflecting on a stressor while in a positive emotional context with that for

people in a neutral context. We first conducted a behavioral pilot (n = 89), then recruited

healthy community participants (n = 50) to complete the study in the MRI scanner. In the

study, the participants did a stressful anagram task followed by a recovery period when

they reflected on the stressor and watched a positive or neutral emotion inducing video.

Participants reported in the moment pleasant and unpleasant mood ratings, and thoughts

and personality data in retrospect. In addition to examining the between group differences

in stress recovery based on the emotional context assignment, we also investigated how

changes in positive mood correlated with neural activation changes to the stressor. We

found evidence for positive emotions facilitating negative emotion recovery through

enhancing cognitive control (decentering thoughts) and providing positive contextual

information (positive video thoughts). Positive video thoughts also correlated with

stronger vmPFC activation during stress recovery.

Keywords: positive emotion, stress recovery, vmPFC, change-point analysis

11

INTRODUCTION

The nationwide annual survey in Stress in America reported that in 2015, people

experienced increased stress; money, work and family responsibilities are the top three

stressors, and greater percentages of adults indicated suffering extreme levels of stress

and more stress symptoms (APA, 2016). Poorly managed stress responses in the long run

greatly add to the cumulative biological burdens, the allostatic load, thus making the

organism more vulnerable to develop diseases or exacerbate the disease symptoms

(McEwen, 2000; Sapolsky, 2007; Seeman, McEwen, Rowe, & Singer, 2001). Via

negative emotions, psychological stress is tightly associated with the persisting symptoms

and progression of diseases such as cardiovascular diseases, acquired immune deficiency

syndrome (AIDS), asthma, and slowed wound healing (Cohen et al., 2007; Maple et al.,

2015; Trueba, Simon, Auchus, & Ritz, 2016). Psychologically, people develop or have

worsened symptoms of mental illnesses including anxiety disorders, depression (Cohen et

al., 2007; Miller, Chen, & Cole, 2009), posttraumatic stress disorder (PTSD), substance

abuse (Herman, 2012), and schizophrenia (Nugent, Chiappelli, Rowland, & Hong, 2015).

One protective factor for some of these stress-related diseases is positive

emotions. In a large Nova Scotian sample, researchers found increased positive affect

protective against ten-year incident coronary heart disease (Davidson, Mostofsky, &

Whang, 2010). With AIDS patients, higher positive emotions are significantly associated

with lower mortality while controlling for other risk factors (Moskowitz, 2003). On one

hand, stress triggers psychopathology including the dysregulation of normative positive

emotions (Gilbert, 2012) and low positive emotions are related to depression, anxiety,

2

eating disorders, and borderline personality disorder (Carl, Soskin, Kerns, & Barlow,

2013; Gilbert, 2012; Jacob et al., 2013); on the other hand, positive emotions

significantly mediate relations between positive mental health and psychopathology and

moderate relations between negative emotions and psychopathology; specifically,

positive emotions such as self-compassion are found to be an important component of

resilience and of adaptive emotion regulation (Trompetter, Kleine, & Bohlmeijer, 2016).

Positive emotions may be achieving this protective effect through facilitating

stress recovery. In fact, much evidence shows that positive emotions corroborate stress

recovery as if “undoing” negative effects from lingering negative emotions. For example,

researchers conducted a few physiological experiments where participants were first

stressed (watching a negative emotion inducing video or anxiously preparing for a judged

speech), then were exposed to various positive stimuli such as amusement or contentment

generating videos. Results showed that participants who were induced positive emotions

usually had faster cardiovascular recovery that was more than a replacement or

distraction effect from positive emotions (Fredrickson & Levenson, 1998; Fredrickson,

Mancuso, Branigan, & Tugade, 2000). In addition to the physiological “undo” effect

(Ong & Allaire, 2005), positive emotions have been tested to show similar cognitive

modulating effects (Falkenstern, Schiffrin, Nelson, Ford, & Keyser, 2009; Hughes &

Kendall, 2008). During stress recovery, an individual may remain insensitive to changes

in the environmental demands and still experience enduring negative emotions. If

positive emotions are induced, cues from the changed (safe) environment may become

more salient. This resembles putting a psychological break to negative emotions (Tugade,

3

Devlin, & Fredrickson, 2014), resulting in physiological (Tugade, Fredrickson, & Barrett,

2004) and cognitive “undo” effects.

Top-Down Route: Positive Emotions Enhance Cognitive Flexibility and Openness

Several routes exist for positive emotions to facilitate stress recovery. One way is

to influence the information processing style within an individual. Negative emotional

states generally shift a person’s default tendency (such as holistic processing) with a

detail-oriented analytic information-processing style, thus narrowing attention and focus

(Clore & Palmer, 2009; Huntsinger, Isbell, & Clore, 2014; Oatley et al., 2006). For

example, fear will initiate abrupt attention focusing that is narrowed on the threat or

induce a non-specific state of vigilance (Lang, Davis, & Öhman, 2000). Despite benefits

from negative emotions in eschewing stereotypes and promoting spontaneous reactions to

acute stressors, persistent negative emotions confine new information intake and

processing.

As emotions are self-perpetuating processes, components and products of

negative emotions reciprocally maintain or exacerbate negative states (Larsen & Prizmic,

2004). One such component is rumination, which is an ineffective problem-solving

strategy composed of inflexible and repetitive negative thoughts with inner-directed

attention (Gross, 2008; Larsen & Prizmic, 2004). So people in a negative emotional state

may have narrowed attention and think in a rigid and convoluted way (Garland, Farb,

Goldin, & Fredrickson, 2015). This rigidity and insensitivity to new environment

demands will impede stress recovery.

In contrast, based on the broaden-and-build theory (Fredrickson, 2001), positive

emotions have an immediate and short-term enhancing effect on the flexibility and

4

openness of cognitive processes (Fredrickson & Cohn, 2008). The flexibility is

manifested in the flexible deployment of attention under different situations: positive

emotions, more than simply broadening cognition, provide an easier access to broadened

cognitive style and better processing of salient information (Isen, 2008). These effects

have been tested and reviewed in attention (Fredrickson & Branigan, 2005), cognitive

control (Xue et al., 2013), decision-making (Estrada, Isen, & Young, 1997), social

perception (Waugh & Fredrickson, 2006), and visual perception (Johnson, Waugh, &

Fredrickson, 2010; Vanlessen, De Raedt, Koster, & Pourtois, 2016).

The cognitive facilitation effect from positive emotions may result in faster stress

recovery as evidence shows that higher cognitive capacity (such as working memory, a

capacity for processing multiple streams of information simultaneously) enhances

emotion regulation ability in both experimental and naturalistic settings (Schmeichel &

Demaree, 2010). With higher cognitive capacity, during stress recovery, the utilization of

cognitive reappraisal may be augmented. Cognitive reappraisal is to reframe and/or alter

the appraisal of components within a situation in a different way, influencing the

individual-environment relationship that eventually changes the emotions (Gross, 2008;

Larsen & Prizmic, 2004). It is considered to be an early intervention technique with high

ecological validity and no conspicuous cognitive cost (Gross, 2008). Positive emotions

may boost the quality and quantity of the reappraisal repertoire related to the stressful

event.

Neural anatomical connections provide some evidence of the cognition

facilitating effect from positive emotions on cognitive reappraisal. In general, various

emotion regulation strategies follow a broad neural model of prefrontal regions regulating

5

the subcortical affective regions (Ochsner, Silvers, & Buhle, 2012). The dorsal executive

system (DES) plays an important role in the top-down control in cognitive reappraisal.

Cognitive reappraisal involves the prefrontal cortex (dorsal lateral PFC (dlPFC) and

ventral lateral PFC (vlPFC), medial PFC (mPFC)) and cingulate system (dorsal anterior

cingulate cortex (dACC)) exerting control and modulating activity in posterior and

subcortical systems that generate emotional responses (Miller & Cohen, 2001; Morris,

Leclerc, & Kensinger, 2014; Ochsner et al., 2012). Within DES, dlPFC is important in

cognitive control and maintenance of goal-relevant information (Iordan & Dolcos, 2015;

Spielberg et al., 2012; Spielberg, Stewart, Levin, Miller, & Heller, 2008). It provides rich

input to dACC which is related to the regulation of adaptive behavior (autonomic

reactions) responsive to the environment (Pruessner et al., 2008), error detection and

decision making to minimize potential loss (especially between two aversive outcomes)

(Spielberg et al., 2008). dACC then provides critical output to the amygdala (Ray & Zald,

2012); the amygdala is important in initiating the learning of fear (Schiller, Levy, Niv,

LeDoux, & Phelps, 2008) and it processes emotional salient signals from the environment

and interacts with autonomic motor systems to express emotions (Ghashghaei & Barbas,

2002).

Within cognitive reappraisal, even though it is believed that PFC control systems

influence affect systems including the amygdala, there are sparse direct neural fiber

connections between dlPFC and the amygdala (Ray & Zald, 2012). In addition to the

dlPFC-dACC-amygdala pathway discussed above, other possible mediators that allow

functional communications to occur may exist (Ochsner et al., 2012). The vmPFC is an

emotion coding area that is highly relevant to positive emotions (Etkin, Egner, & Kalisch,

6

2011; Harris, McClure, van den Bos, Cohen, & Fiske, 2007; Winecoff et al., 2013).

Because of its anatomical location, it has connections with both lateral regions of PFC

and with limbic areas like the amygdala (Ghashghaei, Hilgetag, & Barbas, 2007). Thus

vmPFC may be part of the cognitive reappraisal pathway that connects dlPFC and

amygdala such that cognitive control of choice, such as selections of different

reappraisals, involves interactions among the systems for maintaining choice goals in the

dlPFC and the systems representing the value of the choices with respect to those goals in

the vmPFC (Ochsner et al., 2012).

In addition to the anatomical connections evidence, functional imaging and brain

volume studies also support the important role that the vmPFC plays in cognitive

reappraisal. PFC activation has been found positively correlated with the vmPFC and

negatively correlated to amygdala responses (Ochsner et al., 2012). Older people have

thinner lPFC but thicker vmPFC that is highly connected to the amygdala (Ochsner et al.,

2012), and for those who succeed in maintaining healthy stress hormone level, they have

more bilateral vmPFC activation when reappraising negative images (Urry et al., 2006).

The vmPFC may assist reappraisals for the senior population as well as for women as

evidence exists for a correlation between higher brain volume in vmPFC in women and

more use of cognitive reappraisal and a correlation between comparatively lower vmPFC

volume in men and more use of suppression (Welborn et al., 2009). Hence, the vmPFC,

an area related to positive emotions and emotion coding, may be part of the link in

positive emotions facilitating cognitive reappraisal and DES cognitive control.

Along with vmPFC, nucleus accumbens (NaC) is another region closely

associated with positive emotions (Spielberg et al., 2008). NaC may also engage in the

7

top-down route. Within positive emotions, motivational (anticipatory) states are activated

from the core of NaC with increased release of dopamine and these states feature greater

cognitive elaborations that are based on PFC (Turowski, Man, & Cunningham, 2014). So

is increased opioid activity which may enhance cognitive ability (Fredrickson, 2004) and

decrease the amygdala intensity (Koepp et al., 2009).

Bottom-Up Route: Positive Emotions Bias and Provide Positive Information

In addition to the top-down cognitive facilitation route, positive emotions may

influence information selection and provide information through a bottom-up fashion.

Negative emotions not only feature a detailed and analytic cognitive style, but also

impact the unconscious or conscious filtration of incoming information (Oatley et al.,

2006). Based on the affect infusion model and emotion congruence theory, emotion

influences attention, memory and judgment (Clore & Palmer, 2009; Oatley et al., 2006).

The more deviation from prototypes or cognitive processing a task needs, the more

congruence is shown with the current emotional state (Bartlett, 1932). People have

perceptual bias for emotion-congruent stimuli, which perpetuates the same emotional

state cycle (Bartlett, 1932). People also tend to remember things that are more congruent

with their present emotions and reconstruct the memory with schemas and emotional

attitudes towards the event (Bartlett, 1932). People with anxiety disorders have a pre-

attentive bias towards potentially threatening information (Mathews, 1990; Puliafico &

Kendall, 2006). Therefore, it is likely that when a person is in a negative emotional state,

because of congruence and affect infusion, the person is more likely to bias attention to

negative stimuli (Garland et al., 2015). Because of this negativity bias, a negative

8

emotional state will likely enable additional processing that facilitates congruent

activations of the components within negative emotions (Turowski et al., 2014).

In contrast to the information that is selected in a negative emotional state, the

presence of positive emotions may make emotion-congruent memories more accessible

and create incongruence in the negative emotional state, and thus interrupt the self-

perpetuating process of negative emotions. This process resembles a psychological break

from stress (Tugade et al., 2014) that can potentially short-circuit rumination (Folkman &

Moskowitz, 2000). Indeed, even a break or delayed reaction time that gives people more

time to think about alternatives can tilt a person’s negative valence bias towards

ambiguous objects in the positive direction (Neta & Tong, 2016).

Emotions not only bias information selection but also serve as information

themselves. Emotions provide evaluative information including the value and importance

of an object (Clore & Palmer, 2009). As information, feelings give people a quick, more

correct than chance, heuristic evaluation of the target (Clore & Palmer, 2009). In general,

negative emotions signal negative attributes and events: there is a problem in the

environment that needs to be addressed. Whereas positive emotions signal positive

attributes and events: there is an absence of problem or even better (Clore & Palmer,

2009; Turowski et al., 2014). For example, Duchenne laughter is considered, through

biological and cultural evolution, an approach to convey the signal of safety and satiation

to others (Gervais & Wilson, 2005). Safety and satisfaction information may also make

people less critical thus engage in less self-critical rumination.

The shared neural underpinnings between positive emotions and fear extinction

indicate safety signaling. This mainly happens within the ventral affective system (VAS).

9

It is a large aggregate involving different brain regions, such as the amygdala for basic

emotion processing and negative affect, vlPFC for emotion regulation, portions of medial

frontal cortex especially the vmPFC for emotion integration, as well as the visual cortex

and other ventral regions influenced by emotion modulation and task-specific

performance (Iordan & Dolcos, 2015; Lindquist, Wager, Kober, Bliss-Moreau, & Barrett,

2012). For instance, bottom-up fear generation activates attentional vigilance, memory

encoding systems and the amygdala (with dorsal amygdala activation correlating with

reported emotion intensity) (Ochsner et al., 2009). The vmPFC, an area responsible for

generation of affective meaning (Roy, Shohamy, & Wager, 2012), has been found active

in fear extinction. In rodent fear conditioning studies, initial acquisition of extinction

depends on the amygdala (Ochsner et al., 2012). The ability to retain and express

memory for extinction is related with the vmPFC, with evidence showing that the

magnitude of activation and thickness of vmPFC predict the extinction speed (Ochsner et

al., 2012). During reversal learning (fear flexibility as conditioned fearful stimuli can be

easily switched), a unique dissociation within vmPFC is found, in that a previously

learned danger stimulus engages more in vmPFC than a new safe stimulus (Schiller et al.,

2008). This suggests that the vmPFC may be selectively providing a safety signal to

facilitate the selective fear inhibition by preventing fear response generalization and

perseverance (Schiller et al., 2008). In addition to safety signal, the vmPFC may also

provide a reward signal due to negative reinforcement (Schiller et al., 2008).

NaC may play a parallel role in the bottom-up route. Evidence shows that via

dense reciprocal neural connections among NaC, the amygdala and the orbitofrontal

cortex (OFC), behavior guidance such as reward sensitivity is achieved through the

10

integration of information from the amygdala and NaC in OFC (Engen & Singer, 2015;

Turowski et al., 2014). In one study, depression was found to be positively correlated

with right OFC and right dlPFC coupling and negatively with left OFC and left dlPFC

coupling (Spielberg et al., 2014). Explanation for this result is that for the depressed

patients, the right OFC highlights the aversive aspect of the objects and is not well

regulated by right dlPFC; the left OFC usually draws on the positive aspects of the

objects, yet in depressed patients, the left dlPFC fails to up-regulate the left OFC

(Spielberg et al., 2014). Hence the NaC activation during positive emotions may provide

input into the OFC to bias positive aspects of events.

The Present Study

Although theoretical work and experiments demonstrate that positive emotions

after stressful events may facilitate stress recovery, the neural mechanisms linking

positive emotions to stress recovery are not adequately understood. In this current study,

we conducted behavioral and fMRI experiments to test the routes through which positive

emotions facilitate stress recovery. The participants were first given a challenging

anagram task, inducing negative emotions, and then instructed to reflect on the stressor

while watching either a positive or a neutral video. After this recovery period,

participants were exposed to the anagrams again to identify if they recognized each trial

of the anagram as old or new. This task was masked as a memory task to capture

participants’ emotional responses to reminders of the previous negative experience.

Pleasant and unpleasant mood ratings were collected throughout the study. A battery of

questionnaires including the ones collecting participants’ thoughts during the recovery

period and a post interview were completed at the end.

11

We expected that this experiment design would allow us to test both the top-down

and bottom-up routes. The thoughts we collected from the participants may provide hints

for the two routes. The top-down route is about positive emotions facilitating cognition.

To test potentially enhanced cognitive ability, we had items for decentering and thoughts

about the anagram task. Decentering is about the ability to detach oneself from the

thoughts and feelings and understand them as temporary and not necessarily true (Fresco

et al., 2007). This meta-cognitive ability probably is effortful and involves activation in

many DES regions. More positive thoughts about the anagram task include reappraisals

that directly target the stressor. Generating these thoughts were presumed to be effortful

with much need for cognitive control. The bottom-up route is about positive emotions

biasing and providing positive information. In our study, the new positive information

was probably from contextual cues – the positive video. So positive video thought items

could measure contextual positive information influence.

Before testing these routes, we first checked the stress manipulation. Successful

stress manipulation would mean that post anagram pleasant mood ratings to be

significantly lower than the post baseline pleasant mood ratings and post anagram

unpleasant mood ratings to be significantly higher than the post baseline unpleasant mood

ratings. These results assisted us in excluding the data from the participants who showed

no stress responses (i.e., no changes in the predicted directions in either pleasant or

unpleasant mood ratings) and at the same time, expressed explicit suspicion about the

stress manipulation during the post interview.

In addition to the stress manipulation check, we examined the experimental

manipulation. Successful experimental manipulation would mean that the participants in

12

the positive video condition during the recovery period showed more positive emotion

recovery than participants in the neutral video condition. Positive emotion recovery is

operationalized as post video pleasant mood ratings controlled for post anagram pleasant

mood ratings and post baseline pleasant mood ratings. Following this definition, the more

the residual, the better a participant’s recovery in pleasant mood is. Based on previous

studies conducted in the lab, we predicted that participants in the positive video condition

would achieve better negative emotions recovery than participants in the neutral video

condition. Negative emotion recovery is operationalized as post video unpleasant mood

ratings controlled for post anagram unpleasant mood ratings and post baseline unpleasant

mood ratings. Following this definition, the more the residual, the worse a participant’s

recovery in unpleasant mood is.

We also examined group differences in thought content to see if presenting a

positive video influenced thoughts of those participants in a predicted direction (more

positive and fewer negative thoughts about the anagram task and the video; more

decentering and fewer rumination thoughts). We then followed up with correlation

analyses of positive emotion recovery, negative emotion recovery and the thought

content, though we were not able to make any causal conclusions based on correlation

results.

After the manipulation checks, we aimed to examine the following hypotheses to

test the mechanisms underlying beneficial effects from positive emotions on stress

recovery:

Hypothesis A: positive emotion (PE) recovery negatively correlates with negative

emotion (NE) recovery.

13

Hypothesis B: PE recovery is associated with more positive thoughts and/or fewer

negative thoughts (including decentering, rumination and thoughts about the anagram

task and video content).

Hypothesis C: PE recovery mediates the relationship between positive and/or

negative thoughts and NE recovery.

Hypothesis D: PE recovery and positive thoughts will be positively correlated

with vmPFC and/or NaC activation.

Hypothesis E: PE recovery will mediate the relationship between activation in

vmPFC and/or NaC areas and NE recovery.

Hypothesis F: when exposed to the negative stimuli again (during the memory

task for the previous anagrams), the participants who experienced better PE recovery will

have weaker negative affect related neural responses to the negative stimuli (compared to

novel stimuli) than the participants who had poor PE recovery.

14

METHOD

Participants

Behavioral pilot. Behavioral pilot consisted of two stages. First stage,

community participants (n = 20) were recruited through flyers posted around Wake

Forest University’s campus. To be eligible, those participants received phone screening

to see if they were native English speakers and had no history of psychiatric, neurological

disorders, severe learning disabilities or head trauma. Informed consent was obtained

from all participants, and all procedures were conducted in accordance with the Wake

Forest Institutional Review Board ethical standards. Participants were compensated $25

for their participation. In the results, we noticed much missing data and based on

participants’ feedback, we improved the experiment design. Due to the missing data and

differences in the experiment procedure, we did not include results from this sample in

the present paper.

At the second stage of behavioral pilot, Wake Forest University undergraduates (n

= 89) participated in the study through SONA system. These undergraduate participants

were notified that the requirements for signing up included age of 18 or above and

English as first language. Informed consent was obtained from all participants, and all

procedures were conducted in accordance with the Wake Forest Institutional Review

Board ethical standards. SONA subject pool participants received .5 experiment credit.

fMRI. Three hundred sixty-six community participants responded to the

advertisement for the study from the Be Involved clinical research service at Wake Forest

Baptist Medical Center and flyers posted around Wake Forest University’s campus. To

be eligible, participants had to be native English speakers, have no history of psychiatric

15

or neurological disorders, no severe learning disabilities or head trauma, or any other

physical limitations that prevented them from entering the magnetic resonance imaging

(MRI) machine. Right-handed community participants (n = 79) between the ages of 18

and 64 years were invited to participate the study at Wake Forest Baptist Medical Center.

In the end, 50 participants (18 to 57 years of age, M = 28.40 years, SD = 11.780, nFemale =

25, nWhite = 28) completed the study. Informed consent was obtained from all participants,

and all procedures were conducted in accordance with the Wake Forest Institutional

Review Board ethical standards. Participants were paid $25 for their participation. During

the post study interview (before debriefing), we used structured interview questions to

examine the response process validity of the anagram task. Specifically, we asked the

participants how they felt about the anagram task and how they felt when being compared

with other participants. If the participants mentioned either that some anagrams seemed

unsolvable or that the statistics from other participants appeared fake, we decided that the

participants had suspicion about the difficulty of anagram/stressor task and their poor

performance. If those participants also had no reduction in pleasant mood after the

stressor and no increase in unpleasant mood after the stressor, we viewed this as evidence

that those participants already had suspicion about the intention of the stressor task

during the stressor task. Based on these criteria (expressed suspicion during post study

interview and no predicted pleasant and unpleasant mood changes post stressor task), six

participants’ data were excluded from data analyses (one from positive video condition,

five from neutral condition). Another participant discontinued due to physical discomfort,

so we analyzed the data for 43 participants (18 to 57 years of age, M = 28.19 years, SD =

11.843, nFemale = 23, nWhite = 22).

16

Procedure

Pre-scanner preparation. Upon arrival, the participants were screened again for

MRI eligibility. In the waiting room, the participants were trained to use the mood rating

scales and practiced an easy version of the anagram task in the E-Prime program on a

laptop.

fMRI data acquisition. fMRI data were acquired using a 3T Siemens Magnetic

Skyra Resonance scanner with a high-resolution, 32-channel human head/neck coil. To

minimize head movement, padding was applied around the participant’s head and the

participants were instructed to keep still. High-resolution T1 anatomic images (5’30’’)

were obtained using a 3D volumetric MPRAGE sequence with isotropic voxel-resolution

of 0.9 mm. BOLD data were acquired with a single channel, whole-head imaging coil

from 36 slices with a single shot EPI (TR 2 seconds, TE 25 milliseconds, 3.5 mm

isotropic spatial resolution, FOV 22.4x22.4 cm2, 64x64 matrix size).

fMRI data preprocessing. We applied the Wake Forest University pipeline

(Maldjian, Baer, Kraft, Laurienti, & Burdette, 2009) to preprocess the BOLD data in

SPM12 (Wellcome Trust Centre for Neuroimaging). This automatic standard imaging

preprocessing process included data cleaning, realignment, coregistration, normalization

to MNI space, spatially smoothing the data with an 8x8x10 mm Full Width at Half

Maximum Gaussian smoothing kernel, temporally smoothing with a six second

exponential kernel, detrended, and high-pass filtering with a frequency criterion of one

cycle per minute. After motion correcting the files, we separated each whole NIFTI file

into single TR NIFTI files for analysis in SPM12.

17

Scanner tasks. In the scanner, participants underwent a long BOLD scan

(10’02’’, 299 volumes). During the baseline period, participants were instructed to relax

and look at the fixation screen at the same time. They then performed the anagram and

stress recovery tasks. Afterwards, they engaged in a memory task during another blood-

oxygen-level dependent (BOLD) scan (4’20’’, 128 volumes). In between the tasks, the

participants were mainly instructed to relax and look at the screen. The scans other than

the two BOLD scans of interest will not be addressed in this paper. See figure 1 for a

detailed schematic flow chart of the scanning procedures in scanner.

Figure 1. The schematic flow chart for the scanning procedures.

Stress manipulation task. While BOLD data was being collected, the

participants were instructed to relax for three minutes with their eyes open during the

baseline period. Then the participants did an adapted anagram task that was designed to

frustrate the participants. Two lists of 15 anagrams of five unique letters were created.

Each list consisted of five easy, five hard and five literally impossible anagrams that were

selected (Bradshaw, 1984; Proctor & Vu, 1999), edited and tested to show matched

4’34’’

64’’

10’02’’

5’30’’

E-Prime Start

Sag MPRAGE

PCASL

BOLD

BOLD

Stress

PCASL

fMAP

64’’

4’34’’

4’20’’

67’’

Memory

18

difficulty across lists based on the piloting data. Participants were randomly assigned to

either of the anagram lists. We created two lists of anagrams to ensure in the later

memory task, participants saw the anagrams they saw before and new anagrams.

In the anagram task, participants were given four seconds to view an anagram

then five seconds to solve the anagram by using the keypad to reorder the scrambled

letters on the screen. They then received an immediate performance feedback about

accuracy and a fake/inflated percentage statistic about other participants’ performance

that was meant to produce social comparison pressure. Specifically, we used inline codes

in E-Prime to generate pseudo-random statistics such that for the easy anagrams, the

participants were guaranteed to receive a percentage between 70 and 80, hard from 60 to

70, and impossible from 55 to 65 to show how many other participants were able to solve

that particular anagram. See figure 2 for a schematic flow chart of the anagram task. At

the end of the three-minute anagram task, the participants received a performance

summary of how many anagrams they got correct out of the fifteen and a comment that

most of the other participants (73%) did better than them. This anagram task has been

shown to reliably induce negative emotions and make participants ruminate over the

stressor and feel negative about themselves based on pilot data. In this sample, the

average anagram accuracy is 3.26 out of 15 anagrams (ten possible and five impossible)

with a SD of 2.64.

Experimental manipulation task. After the anagram task, the participants rated

their mood and entered a three-minute post stress recovery period. They were instructed

to “think back to the anagram task” and to “engage in any thoughts and feelings that

come naturally” while viewing a video. Participants in the positive emotion condition

19

Figure 2. A hard level of difficulty anagram from the first word list and the correct answer is 14523 “opera.” viewed a mute “Waves” video about ocean waves which was designed to induce

contentment, a low arousal positive emotion that broaden cognition (Fredrickson &

Branigan, 2005). Participants in the control condition viewed a mute “Sticks” video about

brightly colored moving sticks that was akin to an old computer screensaver, which was

supposed to be neutral. Both videos were selected from previous studies to induce

expected emotions and similar level of interest (Fredrickson & Levenson, 1998). See

figure 3 for video screenshots.

Figure 3. Three-minute reflection and recovery period with positive “Waves” video (left) and neutral “Sticks” video (right).

Think about how you felt during the anagram task. Think about how you felt during the anagram task.

View Only

O R A P E

1 2 3 4 5

Type Answer

O R A P E

1 2 3 4 5

Incorrect!

65% participants got this

anagram correct!

So far you got 0 anagram

correct.

4 sec

5 sec

1 sec

20

Experiment memory task. This task was masked as a surprise memory task but

was intended to expose the participants to the anagrams that they failed to solve for a

second time to examine their responses towards the previous stressor. After the recovery

task and two additional scanning periods (6’45’’ elapsed), the participants were instructed

to view and respond to the 29 anagrams (15 from the first list and 14 from the second list)

that flashed on the screen one at a time with intermittent fixation cross for jittered

durations of time (between four and 12 seconds). Only 14 instead of 15 anagrams from

the second list were presented due to a programming glitch. Each anagram was

programmed to stay on the screen for four seconds. The participants pressed one or two

on the keypad to indicate if they recognized the anagram as old or new. Old anagrams

were the ones they tried to solve during the stress task and new anagrams were from the

other wordlist that the other group of participants were randomly chosen to solve during

their anagram task.

Materials

Mood ratings. The participants self-reported their moods throughout the study.

They were first trained to report their pleasant mood (“how pleasant do you feel right

now”) and unpleasant mood (“how unpleasant do you feel right now”) on separate rating

scales. The pleasant mood scale was always presented first and then the unpleasant mood

scale. The scale was from 1 (not at all) to 5 (extremely) with .5 increments. This scale

was designed to increase the sensitivity of the mood ratings through better utilizing the 5-

key pressing keyboard in the scanner. The participants were instructed to press the same

number twice for a whole digit mood rating (i.e., 11 equals 1) and press one number and

the adjacent larger number to reflect a .5 increment (i.e., 12 equals 1.5). The participants

21

were given 6.5 seconds to respond to each mood rating scale and there was a one-second

fixation slide in-between the two scales. To decrease the missing data rate, we also added

a one-second mood rating reminder slide before the onset of the mood rating to cue the

participant of the coming scales. For the mood ratings right before and after the anagram

task during the BOLD scan, the duration of the scales was lengthened to 11 seconds each

to ensure data collection. The experimenters were also monitoring the whole mood rating

process and asked the participants to provide the ratings if the participants failed to

respond to the screen in time. See figure 4 for the time points that mood ratings were

collected.

Figure 4. Time points when mood ratings were collected in the scanner.

Post-task thoughts content and personality questionnaires. At the end of the

experiment, outside of the scanner, participants filled out a questionnaire battery to report

their thoughts about the anagram task and the video during the recovery period and

provide personality data, answered a few questions during post interview, were then

debriefed and received compensation.

Modified Decentering Questionnaire. The modified decentering questionnaire is

based on the Experiences Questionnaire (Fresco et al., 2007). It has two factors of

decentering and rumination. Decentering is about the ability to detach oneself from the

thoughts and feelings and understand them as temporary and not necessarily true (Fresco

22

et al., 2007). We edited the wordings of the items to make it less generalized and more

relevant to the anagram stressor. The 17 items are rated on a scale from 1 (not at all) to 5

(very much). The construct of the questionnaire has been validated. In this sample (N =

43), decentering subscale had acceptable reliability (11 items; α = .704) and rumination

subscale showed acceptable internal consistency (6 items; α = .662).

Thought Content Questionnaire. The thought content items were developed

based on thought data collected from a previous study of similar experiment design in the

lab. It is used to record participants’ thought process during the reflection period. The 14

items were intentionally grouped into 3x2 categories: Positive, negative or irrelevant

thoughts and feelings about the anagram task or the video content. Items are rated on a

scale from 1 (not at all) to 5 (very much). The items have face validity. In our sample, all

subscales demonstrated acceptable internal consistency except positive thoughts about

the video: positive thoughts about the anagram task (7 items; α = .738), about the video

(3 items; α = .530), negative thoughts about the anagram task (8 items; α = 885) and

about the video (3 items; α = .744). After closer examination of the positive thoughts

about the video subscale, we realized one item “I think the video is exciting” was

probably measuring a slightly different factor than the other items asking about the

“calming” and “pleasant” thoughts about the video. However, since “exciting” feature is

a positive description of the video, which is coherent with the goal of measuring having

pleasant responses towards the video and considering the generally lower reliability for a

3-item scale, we decided to not drop this item only for the sake of increasing Cronbach’s

α. The three items together describe the whole positive experience and any item may be

significant to the theoretic expectations for the underlying construct. For this video

23

thoughts scale, the mean of inter-item correlations is .499 (.505, .394 and .598), which is

acceptable (Briggs & Cheek, 1986).

Emotion Regulation Questionnaire. The Emotion Regulation Questionnaire is

used to assess individual differences in the habitual use of emotion regulation strategies:

cognitive reappraisal and expressive suppression (Gross & John, 2003). Habitual

emotional regulation tendencies will contribute to individual differences in spontaneous

stress recovery. The ten items are rated on a scale from 1 (strongly disagree) to 7

(strongly agree). It has good face validity and in our final fMRI sample, reappraisal

factor showed good internal consistency (5 items; α = .856) and suppression factor had

acceptable reliability (5 items; α = .707).

Ego-Resiliency Scale. The Ego-Resiliency Scale (ER89) is designed to measure

the characteristic level of ego-control (active adjustment) in response to the stressful

environment demand (Block & Kremen, 1996). In our study, this personality may be

highly relevant to participants using PE to recover from the stressor. The scale consists of

14 items and are rated on a scale from 1 (does not apply at all) to 4 (applies very

strongly). It has been validated and used to show good reliability (Fredrickson, Tugade,

Waugh, & Larkin, 2003; Ong, Bergeman, Bisconti, & Wallace, 2006; Tugade &

Fredrickson, 2004). In this sample, it demonstrated good reliability (14 items; α = .865).

Extraversion and Neuroticism subscales from Neuroticism-Extraversion-

Openness Personality Inventory. The Extraversion and Neuroticism subscales from

Neuroticism-Extraversion-Openness Personality Inventory (NEO) is used to capture the

extravert and neurotic traits (McCrae & John, 1992). In our study, extraversion trait may

be related to PE recovery and neuroticism trait may play an important role in rumination

24

which may hinder PE recovery and NE recovery. Together the subscales have 24 items

and the items are rated from 1 (strongly disagree) to 5 (strongly agree). The traits and

items were validated (McCrae & John, 1992). In our sample, extraversion subscale

satisfactory internal consistency (12 items; α = .772) and neuroticism subscale had good

internal consistency (12 items; α = .836).

Ruminative Response Scale. The Ruminative Response Scale (RRS) is a

validated measure for rumination with three factors: reflection, brooding and depression

(Treynor, Gonzalez, & Nolen-Hoeksema, 2003). Rumination is expected to be in our

study a critical component in prolonging NE and hindering PE recovery and NE

recovery. The 22-item scale is rated from 1 (almost never) to 4 (almost always). All three

factors displayed satisfactory to good reliability, reflection (5 items; α = .775), brooding

(5 items; α = .837) and depression (12 items; α = .871). Because the three factors together

capture the ruminative thought pattern, we used the mean of the sum scores of three

factors for later analysis (22 items; α = .920).

See all questionnaires in the supplemental material section at the end.

Data Analyses

fMRI Data

Emotion mask. Because the goal of this study was to investigate the emotional

changes at different time points and/or contexts and the reasons behind them, we adopted

a specific emotion mask to isolate our analyses to the brain regions that were most likely

to be relevant to the emotional processes. The emotion mask was created from

Neurosynth.org based on forward inferencing the results from an automated meta-

analysis of 790 studies with the keyword “emotion” (Yarkoni, 2011; Yarkoni, Poldrack,

25

Nichols, Van Essen, & Wager, 2011). This forward referenced emotion map contained z

values of each voxel that was consistently activated in emotion studies (Waugh et al.,

under review).

Change-point analysis. We adopted change point analyses to identify brain

regions responsive to tasks – both during stressor and/or during recovery. At first level,

we retrieved from the NIFTI files each participant’s BOLD signal time series at every

voxel to calculated and create exponentially weighted moving-average (EWMA) z

matrix. This was to estimate activation for each time point after baseline using an

exponentially weighted moving average with baseline subtracted from the activation

during task periods of interest (Waugh, Hamilton, Chen, Joormann, & Gotlib, 2012). At

second level, we put in covariates to create t time series for each voxel for each

participant to apply hierarchical exponentially weighted moving-average (HEWMA)

model. The existing script treated the continous regressors as dicotomized variables,

meaning that the continous variables were split into halves based on the median. This will

be discussed as a limitation at the end of the paper. Then we applied the family-wise error

corrected alpha to determine the threshold for a t value to be considered significant.

Among those time points with significant t values, we defined a change point as the onset

of the time points when the voxels had no shorter than three consecutive time points with

above threshold t values. This way change points in voxels were less likely to be sporadic

time points of random activation. Based on these voxels which contained change points,

we generated k-mean clusters that were no smaller than ten spatially and temporally

connected voxels (see Lindquist et al., 2007 for more details of the script and change-

point analysis).

26

Within a cluster there were multiple voxels with different time series and change

points. Simply averaging the time series of all the voxels cannot provide us with

information of change points of a cluster. So we devised a way to represent the most

consistent activation pattern of a cluster. Specifically, for a given cluster, we calculated at

each time point, the ratio of its voxels that had above threshold t value out of its all

voxels. The ratio ranged from 0 to 1. To better categorize the cluster activation results,

we set three ratio criteria .05 (meaning more than 5% of the voxels within the cluster had

significant activation), .25 (more than a quarter of the voxels) and .50 (more than a half

of the voxels). These three ratio criteria from low to high represents increasingly stringent

criterion for us to consider a particular time point activation as a representation of the

whole cluster’s activation. When interpreting the cluster activation, we mainly focused on

the time points which at least met the .05 ratio criteria. For descriptive purposes in

reporting the results, we categorized the change-point onsets as occurring during the

stressor/anagram period (103–147 time points first half, 148-192 time points second half)

and recovery period (205-254 time points first half, 255-299 time points second half)

including recovery instruction and reflection and video task. see Figure 5.

Figure 5. Time points information for change-points categorization.

One important advantage of applying change-point analysis is that compared to

the standard General Linear Model that is used to analyze long blocks, change-point

analysis increases the sensitivity of the contrasts and collects more information about the

27

onset, duration and intensity of the activation among different regions of interest (M. A.

Lindquist et al., 2007). In our study, this analysis enabled us to detect when and for how

long changes in activation occur during the stressor and recovery periods relative to the

baseline period.

To test Hypothesis D that is PE recovery and positive thoughts will be positively

correlated with vmPFC and/or NaC activation, we conducted change-point analyses to do

contrasts between the brain activation during baseline and periods after baseline with

covariates of PE recovery and significant positive thoughts variables that were selected

from behavioral data analyses.

Inverse logit hemodynamic response modeling event-related analysis. We

conducted event-related analysis to test Hypothesis F, when exposed to negative stimuli

again, participants who experienced better PE recovery would have weaker negative

affect related regional responses to negative stimuli (compared to novel stimuli) than

participants who had poor PE recovery. We adopted the inverse logit (IL) hemodynamic

response modeling. For each voxel, we used the time-varying hemodynamic response

functions (HRF) to model the BOLD responses to different types of events. IL HRF

modeling enabled us to separately estimate the height, time to peak, and the width of the

BOLD responses (Waugh, Hamilton, & Gotlib, 2010). After this, we used different

contrasts for the height, time to peak and width data to detect differences in the BOLD

responses towards different types of events (Waugh et al., 2010; Waugh, Lemus, &

Gotlib, 2014). Then we used robust regression at the group level to analyze the random

effects on the linear contrasts (Waugh et al., 2010; Waugh, Lemus, & Gotlib, 2014).

Based on the Monte Carlo simulations (2,000 iterations), we used an AFNI script

28

(3dClustSim; using 3dFWHMx, 1-sided thresholding; Cox, 1996; Cox, Chen, Glen,

Reynolds, & Taylor, in press) to decide the cluster size (k = 5) for different contrasts

holding a per-voxel threshold at .001 and a cluster-level corrected p-value at .05.

29

RESULTS

Behavioral Pilot Data

Stress manipulation check. For pleasant and unpleasant mood changes after the

anagram task respectively, we conducted independent t-tests to check stress

manipulation. Dependent variables (DVs) were mood ratings and independent variables

(IVs) were time (within-subject: post baseline, post anagram).

Pleasant mood ratings. Pleasant mood ratings significantly decreased from

baseline (M = 2.865, SD = 1.179) to the anagram task (M = 2.000, SD = 1.215), t(88) =

7.355, p < .001. Stress manipulation in pleasant mood was successful as participants’

pleasant mood significantly decreased from baseline to anagram task.

Unpleasant mood ratings. Unpleasant mood ratings significantly increased from

baseline (M = 1.697, SD = .858) to the anagram task (M = 2.910, SD = 1.212), t(88) = -

10.196, p < .001. Stress manipulation in unpleasant mood was successful as participants’

unpleasant mood significantly increased from baseline to anagram task.

Experimental manipulation check. To check positive emotion induction

success, we conducted an ANCOVA to check experimental manipulation. DV was post

video pleasant mood ratings; IV was video condition (between-subject: positive video

and neutral video); covariates were post baseline and post anagram pleasant mood

ratings.

Participants in positive video condition (M = 2.772, SD = 1.255) achieved PE

recovery significantly better than those in neutral video condition (M = 2.156, SD =

1.107), F(1, 85) = 9.271, p = .003. So the positive video compared to the neutral video

successfully facilitated PE recovery in participants (see Figure 6).

30

NE recovery. To test Hypothesis A (PE recovery correlates with NE recovery),

we conducted an ANCOVA to compare between group difference in NE recovery. DV

was post video unpleasant mood ratings; IV was video condition (between-subject:

positive video and neutral video); covariates were post baseline and post anagram

unpleasant mood ratings.

Figure 6. Pleasant mood ratings at different time points (N = 89).

Participants in positive video condition (M = 2.023, SD = 1.110) did not recover

in NE differently from those in neutral video condition (M = 2.444, SD = 1.179), F(1, 85)

= 1.968, p = .164. Despite better PE recovery, those participants did not achieve better

NE recovery than participants in the neutral video condition. Unpleasant mood ratings at

three time points are presented in Figure 7.

1

1.5

2

2.5

3

3.5

4

4.5

5

PostBaseline PostAnagram PostVideo

Ple

asa

nt

Mo

od

Ra

tin

gs

Time Points

PosVid

NeuVid

31

Figure 7. Unpleasant mood ratings at different time points (N = 89).

We tested the correlations between PE recovery (standardized residuals: post

video pleasant mood ratings controlling for post anagram and post baseline pleasant

mood ratings) and NE recovery (standardized residuals: post video unpleasant mood

ratings controlling for post anagram and post baseline unpleasant mood ratings), r =

-.361, p = .001. Despite no between group difference in PE recovery and NE recovery,

we found evidence for Hypothesis A, PE recovery correlated with NE recovery.

Correlation scatterplot is presented in Figure 8. PE recovery positively correlated with

NE recovery.

fMRI Study Behavioral Data

Stress manipulation check. One improvement upon behavioral piloting in fMRI

study is that we also collected pre-baseline mood ratings from the participants. So we

included ratings at this time point in stress manipulation check. We performed two one-

way repeated measure ANCOVAs for pleasant and unpleasant mood ratings (DV). The

1

1.5

2

2.5

3

3.5

4

4.5

5

PostBaseline PostAnagram PostVideo

Un

ple

asa

nt

Mo

od

Ra

tin

gs

Time Points

PosVid

NeuVid

32

IV was time (within-subject: post baseline and post anagram) and the pre-baseline ratings

were a covariate.

Figure 8. Correlation Between PE Recovery and NE Recovery (N = 89).

Pleasant mood ratings. We performed a repeated measure ANCOVA to examine

the pleasant mood reduction in the participants. The post anagram pleasant mood ratings

(M = 2.244, SD = 1. 093) were not significantly lower than pre-anagram ratings (M =

2.791, SD = 1.264), F(1, 41) = 2.961, p = .093. Surprisingly, participants did not show

strong responses to the stressor task.

Unpleasant mood ratings. We performed an RMANCOVA to examine the

unpleasant mood reduction in participants. The post anagram pleasant mood ratings (M =

2.686, SD = 1.160) were significantly higher than pre-anagram ratings (M = 1.721, SD

= .734), F(1, 41) = 7.190, p = .011. Participants experienced NE increase after the

anagram task.

Hence the anagram task successfully produced negative emotional changes in the

participants, more so in unpleasant mood ratings than in pleasant ratings.

-4

-3

-2

-1

0

1

2

3

4

-3 -2 -1 0 1 2 3

NE

Re

cove

ry

PE Recovery

33

Experimental manipulation check.

PE recovery. To check positive emotion induction success, we conducted a two-

way ANCOVA to check experimental manipulation. DV was post video pleasant mood

ratings; IV was video condition (between-subject: positive video and neutral video);

covariates were post baseline and post anagram pleasant mood ratings.

Participants in positive video condition (M = 2.750, SD = 1.285) did not achieve

PE recovery better than those in neutral video condition (M = 2.342, SD = .851), F(1, 39)

= .007, p = .933. So in the fMRI study sample, the positive video compared to the neutral

video did not facilitate PE recovery in participants. This may be due to no significant PE

reduction after the anagram task in the first place. See Figure 9.

Figure 9. Pleasant mood ratings at different time points (N = 43).

1

1.5

2

2.5

3

3.5

4

4.5

5

PostBaseline PostAnagram PostVideo

Ple

asa

nt

Mo

od

Ra

tin

gs

Time Points

PosVid

NeuVid

34

Thought content. We checked the between group difference in the thought

content collected from the Modified Decentering and Thought Content Questionnaire

which was designed to capture participants thinking pattern and content. We thus

performed a MANOVA (IV: video valence (positive and neutral video); DVs:

Decentering, Rumination, Positive Anagram Thoughts, Positive Video Thoughts,

Negative Anagram Thoughts and Negative Video Thoughts).

MANOVA results showed that there was no difference in the thought pattern and

content between video conditions, F(6, 36) = 1.801, p = .127. So the experimental

positive emotion induction was too weak to significantly impact participants’ thoughts.

For subscale results see Table 1.

Table 1. Modified Experience and Thought Content Questionnaires Subscales Average by Video

Valence

Video Decentering Rumination PosAT PosVT NegAT NegVT

Positive 3.230 3.068 2.696 3.264 3.297 1.681 (SD) (.687) (.824) (.738) (.755) (.924) (.807) Neutral 3.067 3.054 2.383 2.754 3.368 2.491 (SD) (.441) (.740) (.756) (.942) (1.050) (.849)

Note. Decentering and rumination are two factors of the modified Experience Questionnaire. PosAT is positive anagram thoughts; PosVT is positive video thoughts; NegAT is negative anagram thoughts; NegVT is negative video thoughts.

NE recovery. Despite failed experimental manipulation in PE recovery, we still

conducted an ANCOVA to compare between group difference in NE recovery. DV was

post video unpleasant mood ratings; IV was video condition (between-subject: positive

35

video and neutral video); covariates were post baseline and post anagram unpleasant

mood ratings.

Participants in positive video condition (M = 2.000, SD = 1.294) did not recover

in NE differently from those in neutral video condition (M = 2.026, SD = .935), F(1, 39)

= .148, p = .702. Unpleasant mood ratings at three time points are presented in Figure 10.

Lastly, we conducted a correlation test between PE recovery and NE recovery in

this sample, r = -.651, p < .001 (see figure 11). Despite failed experimental manipulation

in PE, we still found evidence for Hypothesis A that PE recovery correlated with NE

recovery. See scatterplot Figure 11. PE recovery positively correlated with NE recovery.

Figure 10. Unpleasant mood ratings at different time points (N = 43).

Multiple regression analyses. To test the Hypothesis B that PE recovery

correlates with more positive thoughts and/or fewer negative thoughts, we examined

correlations among PE recovery and six thought variables from the questionnaires. NE

recovery is the ultimate goal for stress recovery, so we also looked at correlations among

1

1.5

2

2.5

3

3.5

4

4.5

5

PostBaseline PostAnagram PostVideo

Un

ple

asa

nt

Mo

od

Ra

tin

gs

Time Points

PosVid

NeuVid

36

thoughts variables and NE recovery. PE recovery was significantly correlated with

decentering and positive video thoughts, and marginally associated with positive

Figure 11. Correlation Between PE Recovery and NE Recovery (N = 43).

anagram thoughts. NE recovery was negatively associated with decentering, positive

video thoughts and negative anagram thoughts. NE recovery was marginally associated

with positive anagram thoughts (see Table 2).

Consistent with Hypothesis B, PE recovery was associated with more decentering

thoughts, positive video thoughts and marginally more positive anagram thoughts.

Among those variables, we conducted multiple regression to measure the explanatory

power of them in predicting PE recovery and NE recovery. Decentering uniquely

predicted PE recovery above and beyond positive video and anagram thoughts, see Table

3. Positive video thoughts uniquely predicted NE recovery above and beyond positive

video and anagram thoughts, and negative anagram thoughts, see Table 4.

The results from previous analyses together provided evidence to Hypothesis A

and B: despite failed experimental manipulation in PE recovery in positive video

-4

-3

-2

-1

0

1

2

3

4

-3 -2 -1 0 1 2 3

NE

Re

cove

ry

PE Recovery

37

Table 2. Correlation Matrix Among PE Recovery, NE Recovery and Modified Experience and Thought Content Questionnaire Subscales. Video Decentering Rumination PosAT PosVT NegAT NegVT

PE Recovery r .524 -.131 .258 .318 -.160 -.228 p < .001 .402 .095 .038 .304 .142 NE Recovery r -.440 .323 -.293 -.441 .331 .103 p .003 .035 .057 .003 .030 .512

Note. PE Recovery is the residual of post video pleasant mood ratings controlled for post anagram and post baseline pleasant mood ratings; NE Recovery is the residual of post video unpleasant mood ratings controlled for post anagram and post baseline unpleasant mood ratings. Decentering and rumination are two factors of the modified Experience Questionnaire. PosAT is positive anagram thoughts; PosVT is positive video thoughts; NegAT is negative anagram thoughts; NegVT is negative video thoughts. Table 3. Summary of Multiple Regression Analysis for PE Recovery. Variables β t p

Decentering .509 2.897 .006 PosVT .111 .736 .466 PosAT -.058 -.355 .724 R2 F p

Whole model .400 27.375 < .001

Note. PE Recovery is the residual of post video pleasant mood ratings controlled for post anagram and post baseline pleasant mood ratings; Decentering is one factor of the modified Experience Questionnaire; PosAT is positive anagram thoughts; PosVT is positive video thoughts. condition group and the null finding in between group difference in NE recovery, PE

recovery was correlated with NE recovery and PE recovery was positively correlated

with decentering, positive thoughts about the anagram and video.

38

Table 4. Summary of Multiple Regression Analysis for NE Recovery. Variables β t p

Decentering -.212 -1.170 .249 PosVT -.340 -2.238 .031 PosAT .230 1.384 .174 NegAT .041 .228 .821 R2 F p

Whole model .308 4.224 .006

Note. NE Recovery is the residual of post video unpleasant mood ratings controlled for post anagram and post baseline unpleasant mood ratings; Decentering is one factor of the modified Experience Questionnaire; PosAT is positive anagram thoughts; PosVT is positive video thoughts; NegAT is negative anagram thoughts.

Mediation analyses with mood ratings and thought content. Positive video

thoughts and decentering stood out respectively as significant predictors of NE recovery

and PE recovery. On a temporal basis, participants first reflected on the anagram task

experience while watching the video, then rated their pleasant and unpleasant moods

afterwards. This prompted us to test the mediating role of PE recovery between NE

recovery and positive video thoughts and decentering. According to Hypothesis C, PE

recovery will mediate the relationship between NE recovery and positive and/or negative

thoughts.

PE recovery mediation: PosVT and NE recovery. We first tested the mediating

role of PE recovery on positive video thoughts and NE recovery. Positive thoughts about

the video significantly correlated with NE recovery, r = -.441, p = .003. PE recovery

39

significantly correlated with both positive thoughts about the video, r = .318, p = .038,

and NE recovery, r = -.651, p <.001. The mediation analysis (bootstrap 10,000 times)

results showed that in addition to the significant direct effect, t = -2.180, p = .035, there

was a significant indirect effect with a coefficient of -.2024, indicating that positive video

thoughts had direct relations with NE recovery and positive video thoughts also had

indirect relations with NE recovery through PE recovery. See Figure 12. Based on the

large significant β (-.568) of PE recovery in the model of predicting NE recovery along

with positive video thoughts, we considered the possible existence of an alternative

model that was NE recovery mediated the relations between positive video thoughts and

PE recovery. We then reversed the positions of PE recovery and NE recovery and found

no significant changes to the direct and indirect relations. This suggested that positive

video thoughts had indirect relation with NE recovery via PE recovery as well as had

indirect relation with PE recovery through NE recovery.

Figure 12. PE recovery mediation between positive video thoughts and NE recovery model. Standardized coefficients. c is standardized coefficient of total effect of positive video thoughts on NE recovery. c’ is direct effect coefficient. * means p value < .05. **

PE recovery mediation: decentering and NE recovery. We then tested the

mediating role of PE recovery between decentering thought pattern and NE recovery.

Decentering variable achieved significant correlations with negative emotions recovery, r

-.568*** PE Recovery

NE Recovery

.318*

c = -.441**

c’ = -.261* Positive Video

Thoughts

40

= -.440, p = .003, and positive emotions recovery, r = .524, p < .001. The mediation

analysis (bootstrap 10,000 times) results showed that there was a significant indirect

effect with a coefficient of -.5025, indicating that more decentering thoughts were related

to less NE recovery and this relationship was fully mediated through PE recovery. See

Figure 13.

The mediation analyses results demonstrated that PE recovery may be the channel

through which positive video thoughts and decentering thoughts facilitated successful

stress recovery. Positive video thoughts may also influence NE recovery directly or

through additional channels. These correlation results supported Hypothesis C that PE

recovery mediated the relationship between positive thoughts and NE recovery.

Figure 13. PE recovery mediation between decentering thoughts and NE recovery model. Standardized coefficients. c is standardized coefficient of total effect of decentering on NE recovery. c’ is direct effect coefficient. ** p value < .01. *** p value < .001.

Exploratory analyses in personality variables. We also examined the

correlations among personality measures, PE recovery and NE recovery and the thought

content. Ruminative response scale scores had significant negative correlation with PE

recovery. We also checked if the personality variables were related to baseline mood

ratings. Resilience measure positively correlated with post baseline pleasant mood

ratings, r = .434, p = .004. Reappraisal measure also positively correlated with post

-.580*** PE Recovery

NE Recovery

.524***

c = -.440**

c’ = -.137 Decentering

41

baseline pleasant mood ratings, r = .395, p = .009. Personality variables particularly the

resilience/reappraisal played a strong role in participants’ baseline pleasant moods.

All significant correlations between personality measures and the thought content

were in expected directions. Decentering had strong relations with most variables and the

relations were compatible with the decentering construct. More positive anagram

thoughts were related to both stronger resilience/reappraisal personality and weaker

rumination/neuroticism tendencies, whereas more negative anagram thoughts were only

associated with the rumination/neuroticism tendencies. Interestingly, compared to

thoughts about the anagram tasks, positive video thoughts were only related to the

resilience/reappraisal personality. Rumination was only related to ruminative response

measure and negative thoughts about the video had no significant relations with any

personality measure (see Table 5).

Considering the correlation results, ruminative response scores which were

negatively correlated with PE recovery will also be used as a covariate in the following

fMRI analyses.

fMRI Data – Change-Point Analyses

Main effect of time - Overall intercept. The results of the main effect of time

showed that compared to baseline brain activation, various brain regions (mainly medial

and right hemisphere) were more activated mostly during the stressor and some during

the recovery period. See Table 6.

At the beginning of the stressor and the recovery task, more activation in the right

temporoparietal junction (a region that usually engaged in other-related processing

(Atique, Erb, Gharabaghi, Grodd, & Anders, 2011) was probably related with participants

42

Table 5. Correlation Matrix Among Personality Measures, PE Recovery and NE Recovery

Ruminative Response Resilience Neuroticism Extraversion Reappraisal Suppression

PE Recovery -.307* .132 -.271 .121 .046 .140

NE REcovery .259 -.144 .296 -.069 -.029 -.197

Decentering -.530** .434** -.537** .548** .467** -.132

Rumination .325* -.048 .270 .029 -.026 -.109

PosAT -.420** .360* -.330* .283 .351* -.080

PosVT -.216 .361* -.172 .081 .327* .137

NegAT .517** -.106 .491** -.283 -.046 .091

NegVT .100 -.187 .136 -.076 -.25 .018 Note. PE Recovery is the residual of post video pleasant mood ratings controlled for post anagram and post baseline pleasant mood ratings; NE Recovery is the residual of post video unpleasant mood ratings controlled for post anagram and post baseline unpleasant mood ratings; Decentering is one factor of the modified Experience Questionnaire; PosAT is positive anagram thoughts; PosVT is positive video thoughts; NegAT is negative anagram thoughts. * means p value <. 05. ** means p value <.01. trying to understand the task instructions for the anagram task, mood ratings and

reflection task. The thalamus has been shown to relate with vigilance and attention and

involve control of neocortex and effort in working memory (Engström, Karlsson, &

Landtblom, 2014; Hirata & Castro-Alamancos, 2010). More activation in the thalamus

may suggest that participants were more engaged in the stressor, exerting effort in solving

anagrams with the use of working memory. Right BA 6 area and right precentral gyrus

traditionally involve in motor (tactile) and phonetic behaviors (Pishnamazi, Nojaba,

Ganjgahi, Amousoltani, & Oghabian, 2016). In one interesting study, the researchers

found that right precentral gyrus and BA 6 activation was responding to emotions by

eliciting a preparatory defensive response (Kolesar, Kornelsen, & Smith, 2017). In our

study, the right BA 6 and precentral gyrus activation may result from participants

43

Table 6. Change-Point Analysis of the BOLD Responses to the Anagram Stressor and Reflection with Video – Average (Intercept) Across All Participants.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

Activation > Baseline

Temporoparietal junction; BA39; Temporal Mid

R

-52

-56

20

39

2496

7.74

S1; R1

S1; R1

S1; R1

Thalamus; Thalamus; Thalamus

0 -16 8 25 1600 6.42 S1 S1 S1

BA6; Precentral

R -44 4 32 44 2816 6.75 S S S1

Precuneus; BA31; Precuneus

R -4 -60 28 14 896 5.86 S1

Parietal; BA39; Angular, Parietal Sup

R -28 -60 48 15 960 6.35 S S S1

Pre-motor area; BA6; Supp Motor

0 12 52 21 1344 6.44 S S1 S1

Activation < Baseline V5; fusiform, BA37; Temporal Inf

L 48 -72 0 10 640 -5.45

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half. touching the keypad, speaking to themselves while doing the anagram task and/or

responding to the negative emotions induction. The right parietal cortex and angular

gyrus stronger activation happened solely during mood ratings and anagram task. This

corresponded to the finding that these regions were involved in intentional numeric

processing (Kadosh, Bien, & Sack, 2012; Yang et al., 2017). Pre-motor area (preSMA) is

important in domain general executive/cognitive control as well as semantic control

(Hallam, Whitney, Hymers, Gouws, & Jefferies, 2016; Lu, Liang, Yang, & Li, 2010).

44

Activation during the anagram task was likely due to the task demand in executive and

semantic control.

Stress reactivity. NE: Low Stress Reactivity vs. High Stress Reactivity. Using

post anagram unpleasant ratings controlled for pre- and post baseline unpleasant ratings

as a covariate, we examined the regions associated with increased NE (part of stress

reactivity). Overall, participants who were more responsive to the NE induction from the

anagram task had stronger activation in the medial part of the brain (see Table 7).

Those participants surprisingly had stronger vmPFC activation at the beginning of

the anagram task (see Figure 14). One explanation is the stronger engagement from

vmPFC indicated those participants’ conscious process of increasing interoceptive and

subjective awareness of their stress level (Sinha, Lacadie, Constable, & Seo, 2016).

Because those participants indeed felt more stressful based on unpleasant mood ratings

they reported, more vmPFC activation may show their awareness of their increased stress

level.

In healthy people, stress will induce decrease in right vlPFC activation as its

activation probably involves negative thoughts which may interfere with the cognitive

task (Koric et al., 2012). In our study, participants who had NE increase due to the

anagram task also had increased right vlPFC activation during the first half of the

anagram task. This suggests that these participants probably had more negative

rumination with the task at the beginning and indeed felt stressful during the anagram

task.

Those participants also had stronger dACC activation at the beginning of the

anagram task. In a large scale analysis of dACC function study, the authors concluded

45

that the best description of dACC psychological function was related to pain processing

(Lieberman & Eisenberger, 2015). We interpret that for those participants, stronger

dACC activation showed their more intense response to negative feedback (Diekhof &

Ratnayake, 2016).

Table 7. Change-Point Analysis of the BOLD Responses to the Anagram – Regression Based on Post Anagram NE Change.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

High NE Increase > Low NE Increase

vmPFC; BA32; Frontal Med Orb

0

44

-4

30

1920

6.76

S1

S1

S1

vlPFC; BA45, BA44; Frontal Inf Tri

R -48 24 8 43 2752 6.85 S1 S1 S1

dACC; BA32, BA8; Cingulum Mid, Frontal Sup Medial

L 4 24 40 64 4096 6.63 S1 S1 S1

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half.

PE: Low stress reactivity vs. high stress reactivity. PE increase was defined as

post anagram pleasant mood ratings controlled for pre- and post baseline pleasant ratings.

Consistent with NE stress reactivity results, the participants who had more PE decrease

(low PE increase) had more dACC activation near the end of the anagram task (see Table

8).

46

Figure 14. Radiologist view of the medial plane of the brain. Blue means positive correlation between NE stress reactivity and regional activation.

Interestingly, while participants who had post anagram NE increase also had

stronger right vlPFC activation in the first half of the stressor, participants who had post

anagram PE decrease actually had weaker right vlPFC activation near the end of the

stressor. One explanation is this is the normal temporal dynamic neural activation for

vlPFC during stress. In an elegantly designed study, researchers found that if participants

felt stressful during the task, opposite to neutral task, their left vlPFC first had stronger

activation at the beginning of the stressor then the activation gradually tapered off (Sinha

et al., 2016). In our study, right vlPFC may be following the same stress reactive pattern.

So participants who had post anagram PE reduction were expected to have weaker right

vlPFC activation near the end of the stressor.

More PE decrease was also associated with stronger pre-SMA and dACC

activation near the end of the stressor. This may be related to those participants (who

experienced more PE reduction) processing and being responsive to the frustrating

performance summary (see Figure 15).

47

Table 8. Change-Point Analysis of the BOLD Responses to the Anagram – Regression Based on Post-anagram PE Change.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

High PE Increase > Low PE Increase

vlPFC; BA44; Frontal Inf Tri

R

-48

16

24

15

960

5.88

S2

S2

S2

High PE Increase < Low PE Increase

vmPFC; BA10, BA11; Frontal Med Orb

L

4

52

-4

10

640

-5.85

BA6; Precentral

R -44 4 36 22 1408 -6.39

Pre-SMA, dACC; BA32, BA8; Cingulum Mid, Frontal Sup Medial

L 4 24 48 40 2560 -6.09 S2

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half.

The change-point analyses results showed the neural activation in the medial part

of the brain and right vlPFC were consistent with participants’ self-reported stress

reactivity.

48

Figure 15. Radiologist view of the medial plane of the brain. Red means negative correlation between PE stress reactivity and regional activation.

Experimental manipulation. Positive Video vs. Neutral Video. At the start of

the video, participants in the neutral video condition had stronger left pre-SMA

activation. Left pre-SMA has been shown to specifically involve in making implicit

relation synthesis (Lu, Liang, Yang, & Li, 2010). We suspect those participants were

trying to understand and find a pattern in the neutral video based on post interview as

well as thoughts collected from a previous study in the lab (see Table 9).

Also at the beginning of the positive video, the other group of participants had

more thalamus activation which may suggest they were more attentive to the video

content. At the same time, those participants had more dACC and pre-SMA activation.

We were not sure how to interpret this result based on the video condition. Due to weak

experimental manipulation, we cannot draw much inference about the activation that

were related to condition difference.

49

Table 9. Change-Point Analysis of the BOLD Responses to the Video – Regression Based on Video Condition.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

Neutral Video > Positive Video

Pre-SMA; BA6; Supp Motor Area

L

8

20

52

13

832

6.42

S1; R1

S1; R1

S1

Neutral Video < Positive Video

Thalamus; thalamus; thalamus

0

-16

4

47

3008

-8.17

R1

dlPFC, vlPFC; BA46,

BA45; Frontal Inf Tri

R -48 28 16 10 640 -6.04

dACC, pre-SMA; BA32, BA8; Cingulum Mid, Frontal Sup Medial

0 20 48 19 1216 -5.88 S1; R1

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half.

Mood recovery as covariates. Low PE Recovery vs. High PE Recovery.

Participants who achieved better PE recovery had more brain activation in diverse

regions mostly during the anagram task. Although in Hypothesis D we expected that PE

recovery would associate with more vmPFC/NaC activation, we instead found a big

cluster spanning the whole mPFC and some ACC and pre-SMA regions. All these

regions are important in emotion processing and cognitive control. The findings that

more activation during the first half of the anagram task in these regions predicted better

50

PE recovery suggest more (affective) information processing during the stressor may

facilitate more resilience in the participants to achieve PE recovery (see Table 10 and

Figure 16 and Figure 17).

Table 10. Change-Point Analysis of the BOLD Responses – Regression Based on PE Recovery

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

High PE Recovery > Low PE Recovery

vmPFC, dmPFC,

pregenual ACC,

dACC, pre-SMA

R

-4

36

20

159

10176

6.19

S1

S1

S1

Thalamus; thalamus; thalamus

0 -16 4 40 2560 7.27 S; R1 S S1

Posterior cingulate

cortex, precuneus; BA31; precuneus

R -4 -56 28 37 2368 7.25 S; R1 S1

BA6; precentral R -48 4 28 11 704 5.56 S1; R1 S1 S1

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half.

51

Figure 16. Radiologist view of the medial plane of the brain. Blue means positive correlation between PE Recovery and regional activation.

Figure 17. vmPFC, dmPFC, pACC, dACC and pre-SMA activation time course at .05 ratio criterion. Black bar in the upper part of the plot represents the time points when the whole cluster shows consistent between group difference based on the .05 ratio criteria.

Those participants also had stronger thalamus activation at multiple periods in

tasks. This may mean that stronger control of executive function and more attention

during tasks may aid PE recovery. Precuneus is a key component of default mode

network (DMN) and usually becomes deactivated during a task (especially an effortful

52

one) (Jin et al., 2012). For participants who achieved better PE recovery, stronger

precuneus activation may suggest that they felt less challenged during the anagram task

and more relaxed when the recovery period started. Stronger right BA 6 area and right

precentral gyrus activation probably associated with more success for those participants

to engage in the anagram task by pressing keys and reading out anagram letters at the

same time. We were not sure about explanations behind more activation in this area at

one point during the recovery period.

Hence we found some evidence to show that PE recovery was related to vmPFC

activation, though its activation happened during the stressor and vmPFC was in a big

cluster including other medial parts of the brain.

Low NE recovery vs. high NE recovery. Participants who experienced better NE

recovery (low NE recovery residual) had weaker activation in left dlPFC during recovery

period. In healthy people, stressful tasks usually induce stronger left dlPFC activation to

enhance cognitive control (Koric et al., 2012). In our study, having persistent left dlPFC

activation during recovery may be a sign of continued stress in those

participants, as the temporal dynamic neural activation “signature” of left dlPFC post

stress is reduced activation (Sinha et al., 2016) (see Table 11).

Participants who later reported more positive video thoughts also had more

diverse regional activation during the stressor and recovery task. In Hypothesis D, we

predicted that more positive thoughts would correlate with stronger vmPFC/NaC

activation. We found that stronger vmPFC activation during the first half of the stressor

and the first half of the recovery period predicted that those participants reported more

positive thoughts about the video. This result is consistent with vmPFC’s role of tracking

53

positive valence and facilitate stress recovery (see Table 12 and see Figure 18 and Figure

19).

Table 11. Change-Point Analysis of the BOLD Responses – Regression Based on NE Recovery.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

High NE Recovery > Low NE Recovery

dlPFC; BA9, BA46, BA45

3 area's conjunction; Frontal Inf Tri

L

48

32

16

13

832

6.53

R1

R1

R1

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half.

Thought content as covariates. Low PosVT vs. High PosVT. The same group of

participants had stronger thalamus activation during the stressor and the first half of the

recovery. This suggests that those participants probably exerted more effort and were

more attentive during tasks. Stronger left vlPFC activation at some point during the first

half of the anagram task was possibly about normal stress reactivity.

Hence we found strong evidence for Hypothesis D that positive thoughts

correlated with vmPFC activation during the recovery period.

54

Table 12. Change-Point Analysis of the BOLD Responses – Regression Based on Positive Video Thoughts.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

More PosVT > Less PosVT

vmPFC; BA11; Rectus

0

40

-16

10

640

6.03

S1; R1

S1; R1

R1

Thalamus; Thalamus; Thalamus

0 -12 4 18 1152 5.84 S; R1 S; R1 S; R1

vlPFC; BA44,45; Frontal Inf Tri

L 52 24 20 15 960 5.4 S1

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half

Figure 18. Radiologist view of the medial plane of the brain. Blue means positive correlation between positive video thoughts and regional activation.

55

Figure 19. vmPFC activation time course at .05 ratio criterion. Black bar in the upper part of the plot represents the time points when the whole cluster shows consistent between group difference based on the .05 ratio criteria.

Low Decentering vs. High Decentering. Participants who had stronger thalamus

activation during the first half of the stressor predicted more decentering thoughts. Since

decentering is about being mindful and attentive towards the emotional experience,

activated thalamus may facilitate attention. Because decentering variable does not consist

of a strong positive valence component, it was reasonable that decentering, a category of

positive thoughts did not positively correlate with vmPFC/NaC activation as predicted in

Hypothesis D (see Table 13 and See Figure 20).

Personality variable as covariates. Low Rumination vs. High Rumination.

Participants with a stronger rumination personality had stronger activation in thalamus,

right precentral gyrus, and left pre-SMA mainly during the anagram task. The results

suggest that participants high in rumination tendency were also more involved in the

anagram task. We are not sure how to connect these results to rumination (see Table 14

and Figure 21).

56

Table 13. Change-Point Analysis of the BOLD Responses – Regression Based on Decentering.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

More Decentering > Less Decentering

Thalamus; Thalamus; Thalamus

L

4

-16

8

16

1024

6.27

S1

S1

S1

More Decentering < Less Decentering

dACC, pre-SMA; BA8, BA6; Cingulum Mid, Supp Motor Area

L

4

20

48

45

2880

-6.31

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half.

Figure 20. Radiologist view of the medial plane of the brain. Blue means positive correlation between decentering and regional activation. Red means negative correlation between decentering and regional activation.

57

The change-point analyses results provided evidence to support Hypothesis D that

more positive thoughts were positively correlated with more vmPFC activation.

Table 14. Change-Point Analysis of the BOLD Responses – Regression Based on Ruminative Response Scale.

Region

L/R

x

y

z

voxels

vol

MaxZ

Ratio .05

Ratio .25

Ratio .50

Higher Rumination > Lower Rumination

Thalamus; Thalamus; Thalamus

0

-12

8

17

1088

5.74

S1

S1

S1

BA6, BA44; Precentral

R -48 8 32 21 1344 6.26 S; R1 S S1

Pre-SMA; BA6, BA8,

BA32 conjunction; Supp Motor Area, Cingulum Mid

L 4 16 48 48 3072 5.99 S S1 S1

Higher Rumination < Lower Rumination

vmPFC; BA11; Rectus

0

36

-16

11

704

-7.5

Note. Regions: determined based on MNI coordinates of all the voxels in a cluster and the reference sources were from Neurosynth, MNI <-> Talairach with Brodmann Areas (1.09), Mricron with aal.nii.gz template; L/R: MNI coordinates are from radiologist perspective, here we converted L to refer to left hemisphere, R to right hemisphere; vol: based on mm3; MaxZ: the largest z value among all the voxels in a given cluster; Ratio: based on different ratio criteria, the cluster had significant change points during which periods; S: stressor; R: recovery; superscript 1: 1st half; superscript 2: 2nd half.

Mediation Analyses. From previous behavioral analyses, we found that PE

recovery mediated the relationship between positive video thoughts and NE recovery.

Based on the fMRI change-point analyses results, we identified that vmPFC activation

was associated with positive video thoughts during the task and recovery. To further test

58

Figure 21. Radiologist view of the medial plane of the brain. Blue means positive correlation between ruminative response scores and regional activation. Red means negative correlation between ruminative response scores and regional activation. Hypothesis E that is PE recovery mediates the relationship between activation in vmPFC

and NE recovery, we conducted mediation analyses based on the change-point analysis

contrast results from using positive video thoughts as a covariate (see Figure 22).

Figure 22. PE recovery mediation between vmPFC activation and NE recovery model. Standardized coefficients. c is standardized coefficient of total effect of vmPFC on NE recovery. c’ is direct effect coefficient.

We first picked out time points when vmPFC cluster (MNI coordinates x = 0, y =

40, z = -16; 10 voxels) activation significantly differed between high positive video

thoughts and low positive video thoughts groups. This was based on the .05 ratio (5% of

Path b

PE Recovery

NE Recovery

Path a

Path c

Path c’ vmPFC

59

the voxels in the cluster showed above threshold activation), so in this case, there was at

least one voxel showing significant between group activation difference. The time points

were categorized into the time points during the stressor (anagram task) and those during

recovery (reflection and video task). Then we collected weights that were assigned to

each voxel within the cluster at those time points for each participant during the previous

HEWMA analyses. Averaging the weights across the voxels, we produced an averaged

vmPFC cluster weight for each participant during stressor and recovery respectively.

Using Matlab lscov() function with weights and PE recovery and NE recovery statistics,

we calculated, within the weighted least squares regression, raw regression coefficients

and estimated standard errors for path a (vmPFC activation predicting PE recovery), and

path b (vmPFC activation and PE recovery together predicting NE recovery). With these

statistics, we performed Sobel test to evaluate path c’, the mediation effect of PE

recovery on the relationship between vmPFC activation and NE recovery based on the

contrast between high positive video thoughts and low positive video thoughts groups.

We also calculated the total path statistics and carried out same calculation and analyses

based on .25 and .50 ratios (see Table 15).

We found no significant correlations (total effect) between vmPFC activation and

NE recovery at different periods and ratio criteria. When we set the ratio to .05 (at least

one voxel had significant between group activation difference), vmPFC activation was

close to positively correlate with PE recovery, p = .068. PE recovery was also close to

mediating the relationship between vmPFC activation during the stressor and NE

recovery p = .070. Given these results, we did not find strong evidence for Hypothesis E

60

such that PE recovery did not strongly mediate the relationship between activation in

vmPFC and NE recovery.

Table 15. PE Recovery Mediation of Relationship Between vmPFC Activation and NE Recovery Based on Positive Video Thoughts Contrast.

Period

Ratio

Vox

Path a

Path b

Path c

Path c’

B SE

p

B

SE

p

B

SE

p

B

SE

p

Stressor

.05

1

.315

.166

.068

-.828

.132

< .001

.020

.172

.394

-1.813

.144

.070

Stressor .25 3 .178 .132 .159 -.831 .129 < .001 -.022 .167 .393 -1.320 .112 .187

Recovery .05 1 .119 .133 .264 -.832 .127 < .001 -.040 .175 .386 -0.885 .111 .376

Recovery .25 3 .219 .146 .129 -.853 .128 < .001 -.024 .187 .393 -1.459 .128 .145

Recovery .50 5 .170 .122 .150 -.826 .130 < .001 -.015 .168 .395 -1.359 .103 .174

Note. Vox are the minimum significant voxel numbers based on the ratio. Path c’ is the indirect effect path. fMRI Data - Event-Related Analysis of Re-Exposure to Anagrams

After the anagram and recovery tasks, the participants performed a memory task.

It is a task to identify (negative) emotional responses by exposing the participants to the

anagrams again. In Hypothesis F, we expect to see that when the participants were

exposed to the negative stimuli again, the participants who experienced better PE

recovery would have weaker responses (including amygdala and dorsal ACC activation)

to the negative stimuli (compared to novel stimuli) than the participants who had poor PE

recovery.

We first assessed the accuracy for the participants to identify an anagram as seen

(they tried to solve those anagrams during the anagram task) or new. One participant

61

consistently used wrong keys to respond to the trials and if we assumed that correct keys

were used, this participant only had a 50% correct recognition rate which was about

chance. So this participant’s memory task data was excluded from analyses due to failure

to follow task instructions. Based on the average ratios, the rest of the participants

showed acceptable (better than chance) identification accuracy, suggesting they were

following instructions to recall if an anagram was old or new. See Table 16. Additionally,

our memory task design resembles some part of the subsequent memory paradigm and

according to emotion memory literature, only correctly identified trials were included for

analyses (Cabeza et al., 2003; Dolcos & Cabeza, 2002; Kensinger & Corkin, 2003). We

decided to only investigate the trials where participants correctly categorized the anagram

as old or new.

Table 1. Average Ratios of Correctly Recognized Anagrams of Different Event Types.

Old

Old Unsolved

Old Solved

New

Correct

.699

.681

.846

.716

Despite our plan to build a separate event type of recognized solved anagrams,

because of the extremely limited and varied numbers of recognized solved anagrams

among the participants (26 out of 42 participants recognized fewer than five solved

anagrams), we eventually decided to combine the recognized solved anagrams with

recognized unsolved anagrams to make sure that there were no zero trials and enough

number of trials in the type of event for each participant.

62

We then used the IL logit HTW script to conduct event-related analysis

contrasting recognized old anagrams with recognized new anagrams within the same

emotion mask that was used in the previous change-point analysis. Against our

prediction, we found no significant differences in emotion-related brain regions

activation (delay, height or width of HRF) between old and new anagrams. We then ran

the analyses with different covariates including the ratio of recognized unsolved

anagrams out of recognized old anagrams, PE recovery, NE recovery, and residual of

post memory unpleasant mood ratings controlling for pre-memory unpleasant mood

ratings. Still we found no significant cluster activation difference among the emotion-

related brain regions. Hence we found no evidence supporting Hypothesis F which was

better PE recovery would buffer participants against negative stimuli.

63

DISCUSSION

This is one of the first studies to examine the neural correlates of beneficial

effects of positive emotion on stress recovery. In other words, we try to answer the

question: how does PE impact NE recovery? We considered a top-down cognition

facilitation route and a bottom-up positive information bias route. We then examined the

behavioral data and adopted change-point analysis to investigate the fMRI data. We

found partial evidence to support PE’s both routes.

Top-down route: more decentering thoughts uniquely predicted PE recovery and

PE recovery fully mediated the relationship between decentering and NE recovery.

Bottom-up route: more positive video thoughts uniquely predicted NE recovery and there

were direct path between these thoughts and NE recovery as well as indirect path through

PE recovery; positive video thoughts were also correlated with robust vmPFC activation

during the recovery period, suggesting the role of vmPFC tracking positive valence and

facilitating stress recovery.

We confirmed Hypothesis A with behavioral data from both studies: PE recovery

correlates with NE recovery. Hypothesis B was also supported: PE recovery also

correlates with more decentering, positive anagram and video thoughts. Decentering

thoughts best predicted PE recovery whereas positive video thoughts uniquely explained

NE recovery. Then Hypothesis C was confirmed that PE recovery mediated the

relationship between positive and/or negative thoughts and NE recovery. The results from

fMRI data supported Hypothesis D PE recovery and positive thoughts were positively

correlated with vmPFC activation. We then found vmPFC activation correlated with

positive video thoughts was not associated with NE recovery in the first place. So the

64

Hypothesis E that PE recovery would mediate the relationship between activation in

vmPFC and NE recovery was not supported. With IL HRF event-related analysis, we still

failed to find evidence for Hypothesis F that is when exposed to the negative stimuli

again, participants who experienced better PE recovery would have weaker negative

affect related neural responses to the negative stimuli (compared to novel stimuli) than

participants who had poor PE recovery.

Our finding that more vmPFC activation predicted that the same group of

participants would later endorse more positive thoughts about the video is consistent with

our understanding of vmPFC as a region important for emotion coding and positive

affect. A recent study of the neural correlates of positive reappraisal concluded that

increased activation in ventral striatum and vmPFC regions tracked participants’ reports

of more positive emotional experience (Doré et al., 2017). The activation of vmPFC is

probably needed for generating and using positive input in emotion regulation due to its

connectivity with other regions associated with affective arousal, cognitive control, and

positive value (Barrett & Satpute, 2013; Doré et al., 2017). Thus vmPFC stronger

activation during recovery period in our study may be a sign of more active emotion

processing and integration with positive value such that those participants were following

the task instructions to reflect on the anagram task experience and watch the video.

Reflecting about the stressor and at the same time, paying attention to the present

environmental cues may also hinder the person from being completely drawn into the

negative rumination and help the person to relate the positive contextual cues to

regulating their moment to moment emotions.

65

The regional activation that was found to be correlated with decentering was

somehow surprising. The decentering variable we used in this study resembles the

mindfulness construct. Yet we did not find much consistent brain activation associated

with this variable despite its strong correlation with PE recovery. In fact, the results we

found conflict with past mindfulness literature. Trait mindfulness (e.g., use of cognitive

reappraisal, labeling emotions and less mind-wandering) has been found related with

more PFC activation, dampened amygdala and DMN activation, and brief mindfulness

training also additionally correlates with decreased activation in thalamus (Zeidan, 2015).

This suggests that either our decentering scale is more different from mindfulness than

we expected or contextual cues influenced decentering tendencies may be different from

trait mindfulness.

A few explanations exist for the null findings with Hypothesis E. In the first

place, we did not identify a strong correlation between NE recovery and vmPFC

activation which correlated with positive video thoughts. Then only most liberal

threshold vmPFC activation during the stressor was marginally correlated with PE

recovery, so was the PE recovery mediation in that case. However, we found evidence

that the behavioral variable positive video thoughts had robust correlations with PE

recovery and NE recovery, and PE recovery actually partially mediated the relationship

between positive video thoughts and NE recovery. In addition, more positive video

thoughts correlated with increased vmPFC activation robustly during recovery. Given

these evidence, we expected to see more than one significant paths in the neural

mediation model. We believe this is due to limits in the neural mediation strategy we

66

adopted including no access to raw weighted vmPFC activation data and the use of Sobel

test instead of bootstrapping.

With Hypothesis F, despite our use of flexible IL HRF modeling to analyze the

BOLD responses to different event types with various valid behavioral covariates, we

still found no difference in brain activation based on the contrasts. However, this result

did not mean there was absolutely no brain activation difference between viewing old and

new anagrams. Because we used an emotion mask to study emotion-related processes,

other potential differences between event types such as memory for old and new items

may have been masked out. Low power in detecting the differences is another important

issue: only above-chance rate of recognized anagrams were available for analyses and

this largely limited the numbers of the trials under different event types. Despite our

original plan to perform contrasts between recognized unsolved and recognized solved

anagrams, between recognized unsolved and recognized new anagrams, because of very

limited trials for recognized new anagrams particularly, we had to mix recognized solved

and unsolved anagrams together to form an old anagram event type. This event type

inevitably increased error variance thus lowered the effect size. We also suspect that the

current memory task design actually cannot robustly induce emotional responses to

anagrams as expected. Given the null findings even when we used the mood ratings

covariates that were directly related to the anagrams and memory task, we consider that

there seemed not an emotion signature when the participants viewed the anagrams again.

Another piece of evidence is that during post interview of the participants, some of them

commented that the memory task was very surprising and even harder than anagram task.

This suggests that some participants were probably mostly invested in remembering and

67

identifying anagrams to the point that they were not thinking about their feelings or re-

experiencing the frustration with the unsolved anagrams.

One limitation we recognize is that the existing HEWMA script dichotomizes the

continuous regressors for the analyses, which decreased the sensitivity of the results. The

HEWMA script is a validated script for change-point analysis and we are currently

unable to modify it to treat the variables as continuous variables. Additionally, the

thought content questionnaires were administered at the end of the study, so the data

about the thoughts during the video may be influenced by the mood state of the

participants. It is more desirable to collect video thoughts when the participants were

watching the video. In follow-up studies, stronger stress manipulation and emotion

induction experiment paradigms are needed. In fact, the weak positive emotion induction

manipulation limited study results to pure correlational, instead of causal explanations.

When conducting vmPFC mediation analyses, we were not able to directly retrieve the

weighted thus covariance-corrected vmPFC activation data that was produced during

HEWMA analyses. So we had to acquire coefficients and standard errors for all paths in a

“roundabout” manner. In future, we would like to improve the existing HEWMA script to

make the weighted brain region activation data more easily accessible so that we can use

the raw data to do bootstrapping mediation analyses just as we did with the behavioral

data. Sobel test performs much better with very large sample to ensure distribution

assumptions (Preacher & Hayes, 2008), thus using Sobel test probably limited the power

of detecting significant mediation evidence. According to the results from memory task

analyses, we also recognize that one more possible improvement in the anagram task is to

have easy enough anagrams to ensure more solved anagrams for most participants. At the

68

same time, the memory task may be improved to become straight-forward reminders of

the participants’ errors and fake statistics, though the longer duration of the task may be

another challenge.

The results will hopefully provide information to complete the biobehavioral

models of positive emotions and stress regulation as well as the temporal interactions

between emotional context and the regulating neural circuitry. Researchers and clinicians

will also be better informed to combine positive emotions into formulating solutions for

improving mental and physical health. In this study, we put much effort in capturing the

temporal characteristics of various brain regions activation, which is an improvement

than analyzing the static brain regions states. So far we have only been examining the

regional activations separately. The next step is to investigate the changes in the

functionally connectivity among the significant clusters that we already identified. For

instance, we will look into the functional connectivity between vmPFC and dmPFC to

more directly test the bottom-up route to see if vmPFC was facilitating emotional events

interpreting regions such as dmPFC. We will also include temporal regions in analyses to

examine the communication between vmPFC and the temporal regions to find evidence

for the top-down route of PE facilitate cognition in cognitive reappraisal.

69

REFERENCES

Atique, B., Erb, M., Gharabaghi, A., Grodd, W., & Anders, S. (2011). Task-specific

activity and connectivity within the mentalizing network during emotion and

intention mentalizing. NeuroImage, 55(4), 1899–1911.

https://doi.org/10.1016/j.neuroimage.2010.12.036

Barrett, L. F., & Satpute, A. B. (2013). Large-scale brain networks in affective and social

neuroscience: Towards an integrative functional architecture of the brain. Current

Opinion in Neurobiology, 23(3), 361–372.

https://doi.org/10.1016/j.conb.2012.12.012

Bartlett, F. C. (1932). Remembering: A study in experimental and social psychology.

New York: Cambridge University Press.

Block, J., & Kremen, A. M. (1996). IQ and ego-resiliency: conceptual and empirical

connections and separateness. Journal of Personality and Social Psychology,

70(2), 349–361.

Bradshaw, J. L. (1984). A guide to norms, ratings, and lists. Memory & Cognition, 12(2),

202–206. https://doi.org/10.3758/BF03198435

Briggs, S. R., & Cheek, J. M. (1986). The role of factor analysis in the development and

evaluation of personality scales. Journal of Personality, 54(1), 106–148.

https://doi.org/10.1111/j.1467-6494.1986.tb00391.x

Buhle, J. T., Silvers, J. A., Wager, T. D., Lopez, R., Onyemekwu, C., Kober, H., …

Ochsner, K. N. (2014). Cognitive reappraisal of emotion: A meta-analysis of

human neuroimaging studies. Cerebral Cortex, 24(11), 2981–2990.

https://doi.org/10.1093/cercor/bht154

70

Carl, J. R., Soskin, D. P., Kerns, C., & Barlow, D. H. (2013). Positive emotion regulation

in emotional disorders: A theoretical review. Clinical Psychology Review, 33(3),

343–360. https://doi.org/10.1016/j.cpr.2013.01.003

Clore, G. L., & Palmer, J. (2009). Affective guidance of intelligent agents: How emotion

controls cognition. Cognitive Systems Research, 10(1), 21–30.

https://doi.org/10.1016/j.cogsys.2008.03.002

Cohen S, Janicki-Deverts D, & Miller GE. (2007). PSychological stress and disease.

JAMA, 298(14), 1685–1687. https://doi.org/10.1001/jama.298.14.1685

Costafreda, S. G., Fu, C. H. Y., Lee, L., Everitt, B., Brammer, M. J., & David, A. S.

(2006). A systematic review and quantitative appraisal of fMRI studies of verbal

fluency: Role of the left inferior frontal gyrus. Human Brain Mapping, 27(10),

799–810. https://doi.org/10.1002/hbm.20221

Cox, R. W. (1996). AFNI: Software for analysis and visualization of functional magnetic

resonance neuroimages. Computers and Biomedical Research, 29, 162-173.

Cox, R. W., Chen, G., Glen, D. R., Reynolds, R. C., & Taylor, P. A. (in press). FMRI

Clustering in AFNI: False Positive Rates Redux. Brain Connectivity.

Davidson, K. W., Mostofsky, E., & Whang, W. (2010). Don’t worry, be happy: positive

affect and reduced 10-year incident coronary heart disease: The Canadian Nova

Scotia Health Survey. European Heart Journal, 31(9), 1065–1070.

https://doi.org/10.1093/eurheartj/ehp603

71

Diekhof, E. K., & Ratnayake, M. (2016). Menstrual cycle phase modulates reward

sensitivity and performance monitoring in young women: Preliminary fMRI

evidence. Neuropsychologia, 84, 70–80.

https://doi.org/10.1016/j.neuropsychologia.2015.10.016

Dolcos, F., & Cabeza, R. (2002). Event-related potentials of emotional memory:

Encoding pleasant, unpleasant, and neutral pictures. Cognitive, Affective &

Behavioral Neuroscience, 2(3), 252–263. https://doi.org/10.3758/CABN.2.3.252

Doré, B. P., Boccagno, C., Burr, D., Hubbard, A., Long, K., Weber, J., … Ochsner, K. N.

(2017). Finding positive meaning in negative experiences engages ventral striatal

and ventromedial prefrontal regions associated with reward valuation. Journal of

Cognitive Neuroscience, 29(2), 235–244. https://doi.org/10.1162/jocn_a_01041

Engström, M., Karlsson, T., & Landtblom, A.-M. (2014). Thalamic activation in the

Kleine-Levin syndrome. Sleep: Journal of Sleep and Sleep Disorders Research,

37(2), 379–386.

Estrada, C. A., Isen, A. M., & Young, M. J. (1997). Positive affect facilitates integration

of information and decreases anchoring in reasoning among physicians.

Organizational Behavior and Human Decision Processes, 72(1), 117–135.

https://doi.org/10.1006/obhd.1997.2734

Etkin, A., Egner, T., & Kalisch, R. (2011). Emotional processing in anterior cingulate and

medial prefrontal cortex. Trends in Cognitive Sciences, 15(2), 85–93.

https://doi.org/10.1016/j.tics.2010.11.004

72

Falkenstern, M., Schiffrin, H. H., Nelson, S. K., Ford, L., & Keyser, C. (2009). Mood

over matter: Can happiness be your undoing? The Journal of Positive Psychology,

4(5), 365–371. https://doi.org/10.1080/17439760902992415

Folkman, S., & Moskowitz, J. T. (2000). Positive affect and the other side of coping.

American Psychologist, 55(6), 647–654. https://doi.org/10.1037/0003-

066X.55.6.647

Fredrickson, B. L. (2001). The role of positive emotions in positive psychology: The

broaden-and-build theory of positive emotions. American Psychologist, 56(3),

218–226. https://doi.org/10.1037/0003-066X.56.3.218

Fredrickson, B. L. (2004). The broaden-and-build theory of positive emotions.

Philosophical Transactions of the Royal Society B: Biological Sciences,

359(1449), 1367–1378. https://doi.org/10.1098/rstb.2004.1512

Fredrickson, B. L., & Branigan, C. (2005). Positive emotions broaden the scope of

attention and thought-action repertoires. Cognition and Emotion, 19(3), 313–332.

https://doi.org/10.1080/02699930441000238

Fredrickson, B. L., & Cohn, M. A. (2008). Positive emotions. In M. Lewis, J. M.

Haviland-Jones, L. F. Barrett, M. Lewis, J. M., Haviland-Jones, & L. F. Barrett

(Eds.), Handbook of emotions (3rd ed.) (pp. 777–796). New York: Guilford Press.

Fredrickson, B. L., & Levenson, R. W. (1998). Positive emotions speed recovery from

the cardiovascular sequelae of negative emotions. Cognition and Emotion, 12(2),

191–220. https://doi.org/10.1080/026999398379718

73

Fredrickson, B. L., Mancuso, R. A., Branigan, C., & Tugade, M. M. (2000). The undoing

effect of positive emotions. Motivation and Emotion, 24(4), 237–258.

https://doi.org/10.1023/A:1010796329158

Fredrickson, B. L., Tugade, M. M., Waugh, C. E., & Larkin, G. R. (2003). What good are

positive emotions in crisis? A prospective study of resilience and emotions

following the terrorist attacks on the United States on September 11th, 2001.

Journal of Personality and Social Psychology, 84(2), 365–376.

https://doi.org/10.1037/0022-3514.84.2.365

Fresco, D. M., Moore, M. T., van Dulmen, M. H. M., Segal, Z. V., Ma, S. H., Teasdale, J.

D., & Williams, J. M. G. (2007). Initial psychometric properties of the

Experiences Questionnaire: Validation of a self-report measure of decentering.

Behavior Therapy, 38(3), 234–246. https://doi.org/10.1016/j.beth.2006.08.003

Garland, E. L., Farb, N. A., R. Goldin, P., & Fredrickson, B. L. (2015). Mindfulness

broadens awareness and builds eudaimonic meaning: A process model of mindful

positive emotion regulation. Psychological Inquiry, 26(4), 293–314.

https://doi.org/10.1080/1047840X.2015.1064294

Gervais, M., & Wilson, D. S. (2005). The evolution and functions of laughter and humor:

A synthetic approach. The Quarterly Review of Biology, 80(4), 395–430.

Ghashghaei, H. T., & Barbas, H. (2002). Pathways for emotion: Interactions of prefrontal

and anterior temporal pathways in the amygdala of the rhesus monkey.

Neuroscience, 115(4), 1261–1279.

74

Ghashghaei, H. T., Hilgetag, C. C., & Barbas, H. (2007). Sequence of information

processing for emotions based on the anatomic dialogue between prefrontal cortex

and amygdala. NeuroImage, 34(3), 905–923.

https://doi.org/10.1016/j.neuroimage.2006.09.046

Gilbert, K. E. (2012). The neglected role of positive emotion in adolescent

psychopathology. Clinical Psychology Review, 32(6), 467–481.

https://doi.org/10.1016/j.cpr.2012.05.005

Gross, J. J. (2008). Emotion regulation. In M. Lewis, J. M. Haviland-Jones, L. F. Barrett,

M. Lewis, J. M. Haviland-Jones, & L. F. Barrett (Eds.), Handbook of emotions

(3rd ed.) (pp. 497–512). New York: Guilford Press.

Gross, J. J., & John, O. P. (2003). Individual differences in two emotion regulation

processes: Implications for affect, relationships, and well-being. Journal of

Personality and Social Psychology, 85(2), 348–362. https://doi.org/10.1037/0022-

3514.85.2.348

Hallam, G. P., Whitney, C., Hymers, M., Gouws, A. D., & Jefferies, E. (2016). Charting

the effects of TMS with fMRI: Modulation of cortical recruitment within the

distributed network supporting semantic control. Neuropsychologia, 93(Part A),

40–52. https://doi.org/10.1016/j.neuropsychologia.2016.09.012

Harris, L. T., McClure, S. M., van den Bos, W., Cohen, J. D., & Fiske, S. T. (2007).

Regions of the MPFC differentially tuned to social and nonsocial affective

evaluation. Cognitive, Affective & Behavioral Neuroscience, 7(4), 309–316.

https://doi.org/10.3758/CABN.7.4.309

75

Herman, J. P. (2012). Neural pathways of stress integration. Alcohol Research: Current

Reviews, 34(4), 441–447.

Hughes, A. A., & Kendall, P. C. (2008). Effect of a positive emotional state on

interpretation bias for threat in children with anxiety disorders. Emotion, 8(3),

414–418. https://doi.org/10.1037/1528-3542.8.3.414

Huntsinger, J. R., Isbell, L. M., & Clore, G. L. (2014). The affective control of thought:

Malleable, not fixed. Psychological Review, 121(4), 600–618.

https://doi.org/10.1037/a0037669

Iordan, A. D., & Dolcos, F. (2015). Brain activity and network interactions linked to

valence-related differences in the impact of emotional distraction. Cerebral

Cortex (New York, N.Y.: 1991). https://doi.org/10.1093/cercor/bhv242

Isen, A. M. (2008). Some ways in which positive affect influences decision making and

problem solving. In M. Lewis, J. M. Haviland-Jones, L. F. Barrett, M. Lewis, J.

M. Haviland-Jones, & L. F. Barrett (Eds.), Handbook of emotions (3rd ed.) (pp.

548–573). New York: Guilford Press.

Jacob, G. A., Ower, N., & Buchholz, A. (2013). The role of experiential avoidance,

psychopathology, and borderline personality features in experiencing positive

emotions: A path analysis. Journal of Behavior Therapy and Experimental

Psychiatry, 44(1), 61–68. https://doi.org/10.1016/j.jbtep.2012.07.006

Jin, G., Li, K., Qin, Y., Zhong, N., Zhou, H., Wang, Z., … Zeng, Q. (2012). fMRI study

in posterior cingulate and adjacent precuneus cortex in healthy elderly adults

using problem solving task. Journal of the Neurological Sciences, 318(1-2), 135–

139. https://doi.org/10.1016/j.jns.2012.02.032

76

Johnson, K. J., Waugh, C. E., & Fredrickson, B. L. (2010). Smile to see the forest:

Facially expressed positive emotions broaden cognition. Cognition and Emotion,

24(2), 299–321. https://doi.org/10.1080/02699930903384667

Kadosh, R. C., Bien, N., & Sack, A. (2012). Automatic and intentional number

processing both rely on intact right parietal cortex: A combined fMRI and

neuronavigated TMS study. Frontiers in Human Neuroscience, 6, 2.

Koepp, M. J., Hammers, A., Lawrence, A. D., Asselin, M. C., Grasby, P. M., & Bench,

C. J. (2009). Evidence for endogenous opioid release in the amygdala during

positive emotion. NeuroImage, 44(1), 252–256.

https://doi.org/10.1016/j.neuroimage.2008.08.032

Kolesar, T. A., Kornelsen, J., & Smith, S. D. (2017). Separating neural activity associated

with emotion and implied motion: An fMRI study. Emotion, 17(1), 131–140.

https://doi.org/10.1037/emo0000209

Koric, L., Volle, E., Seassau, M., Bernard, F. A., Mancini, J., Dubois, B., … Levy, R.

(2012). How cognitive performance-induced stress can influence right VLPFC

activation: An fMRI study in healthy subjects and in patients with social phobia.

Human Brain Mapping, 33(8), 1973–1986. https://doi.org/10.1002/hbm.21340

Lang, P. J., Davis, M., & Öhman, A. (2000). Fear and anxiety: Animal models and

human cognitive psychophysiology. Journal of Affective Disorders, 61(3), 137–

159. https://doi.org/10.1016/S0165-0327(00)00343-8

Larsen, R. J., & Prizmic, Z. (2004). Affect regulation. In R. F. Baumeister, K. D. Vohs,

R. F. Baumeister, & K. D. Vohs (Eds.), Handbook of self-regulation: Research,

theory, and applications (pp. 40–61). New York: Guilford Press.

77

Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer

Pub. Co.

Lieberman, M. D., & Eisenberger, N. I. (2015). The dorsal anterior cingulate cortex is

selective for pain: Results from large-scale reverse inference. PNAS Proceedings

of the National Academy of Sciences of the United States of America, 112(49),

15250–15255. https://doi.org/10.1073/pnas.1515083112

Lindquist, K. A., Wager, T. D., Kober, H., Bliss-Moreau, E., & Barrett, L. F. (2012). The

brain basis of emotion: A meta-analytic review. The Behavioral and Brain

Sciences, 35(3), 121–143. https://doi.org/10.1017/S0140525X11000446

Lindquist, M. A., Waugh, C., & Wager, T. D. (2007). Modeling state-related fMRI

activity using change-point theory. NeuroImage, 35(3), 1125–1141.

https://doi.org/10.1016/j.neuroimage.2007.01.004

Lu, S., Liang, P., Yang, Y., & Li, K. (2010). Recruitment of the pre-motor area in human

inductive reasoning: An fMRI study. Cognitive Systems Research, 11(1), 74–80.

https://doi.org/10.1016/j.cogsys.2008.08.009

Maldjian, J. A., Baer, A. H., Kraft, R. A., Laurienti, P. J., & Burdette, J. H. (2009). Fully

automated processing of fMRI data in SPM: From MRI scanner to PACS.

Neuroinformatics, 7(1), 57–72. https://doi.org/10.1007/s12021-008-9040-z

Maple, H., Chilcot, J., Lee, V., Simmonds, S., Weinman, J., & Mamode, N. (2015).

Stress predicts the trajectory of wound healing in living kidney donors as

measured by high-resolution ultrasound. Brain, Behavior, and Immunity, 43, 19–

26. https://doi.org/10.1016/j.bbi.2014.06.012

78

Mathews, A. (1990). Why worry? The cognitive function of anxiety. Behaviour Research

and Therapy, 28(6), 455–468.

McCrae, R. R., & John, O. P. (1992). An introduction to the five-factor model and its

applications. Journal of Personality, 60(2), 175–215.

https://doi.org/10.1111/j.1467-6494.1992.tb00970.x

McEwen, B. S. (2000). Allostasis and allostatic load: Implications for

neuropsychopharmacology. Neuropsychopharmacology, 22(2), 108–24.

https://doi.org/http://dx.doi.org.go.libproxy.wakehealth.edu/10.1016/S0893-

133X(99)00129-3

Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal cortex function.

Annual Review of Neuroscience, 24, 167–202.

https://doi.org/10.1146/annurev.neuro.24.1.167

Miller, G., Chen, E., & Cole, S. W. (2009). Health psychology: Developing biologically

plausible models linking the social world and physical health. Annual Review of

Psychology, 60, 501–524.

https://doi.org/10.1146/annurev.psych.60.110707.163551

Morris, J. A., Leclerc, C. M., & Kensinger, E. A. (2014). Effects of valence and divided

attention on cognitive reappraisal processes. Social Cognitive and Affective

Neuroscience, 9(12), 1952–1961. https://doi.org/10.1093/scan/nsu004

Neta, M., & Tong, T. T. (2016). Don’t like what you see? Give it time: Longer reaction

times associated with increased positive affect. Emotion, 16(5), 730–739.

https://doi.org/10.1037/emo0000181

79

Nugent, K. L., Chiappelli, J., Rowland, L. M., & Hong, L. E. (2015). Cumulative stress

pathophysiology in schizophrenia as indexed by allostatic load.

Psychoneuroendocrinology, 60, 120–129.

https://doi.org/10.1016/j.psyneuen.2015.06.009

Oatley, K., Keltner, D., & Jenkins, J. M. (2006). Understanding emotions, 2nd ed.

Malden: Blackwell Publishing.

Ochsner, K. N., Ray, R. R., Hughes, B., McRae, K., Cooper, J. C., Weber, J., … Gross, J.

J. (2009). Bottom-up and top-down processes in emotion generation: Common

and distinct neural mechanisms. Psychological Science, 20(11), 1322–1331.

https://doi.org/10.1111/j.1467-9280.2009.02459.x

Ochsner, K. N., Silvers, J. A., & Buhle, J. T. (2012). Functional imaging studies of

emotion regulation: A synthetic review and evolving model of the cognitive

control of emotion. Annals of the New York Academy of Sciences, 1251, E1–24.

https://doi.org/10.1111/j.1749-6632.2012.06751.x

Ong, A. D., & Allaire, J. C. (2005). Cardiovascular intraindividual variability in later life:

The influence of social connectedness and positive emotions. Psychology and

Aging, 20(3), 476–485. https://doi.org/10.1037/0882-7974.20.3.476

Ong, A. D., Bergeman, C. S., Bisconti, T. L., & Wallace, K. A. (2006). Psychological

resilience, positive emotions, and successful adaptation to stress in later life.

Journal of Personality and Social Psychology, 91(4), 730–749.

https://doi.org/10.1037/0022-3514.91.4.730

80

Pishnamazi, M., Nojaba, Y., Ganjgahi, H., Amousoltani, A., & Oghabian, M. A. (2016).

Neural correlates of audiotactile phonetic processing in early-blind readers: An

fMRI study. Experimental Brain Research, 234(5), 1263–1277.

https://doi.org/10.1007/s00221-015-4515-2

Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for

assessing and comparing indirect effects in multiple mediator models. Behavior

Research Methods, 40(3), 879–891. https://doi.org/10.3758/BRM.40.3.879

Proctor, R. W., & Vu, K.-P. L. (1999). Index of norms and ratings published in the

Psychonomic Society journals. Behavior Research Methods, Instruments &

Computers, 31(4), 659–667. https://doi.org/10.3758/BF03200742

Pruessner, J. C., Dedovic, K., Khalili-Mahani, N., Engert, V., Pruessner, M., Buss, C., …

Lupien, S. (2008). Deactivation of the limbic system during acute psychosocial

stress: Evidence from positron emission tomography and functional magnetic

resonance imaging studies. Biological Psychiatry, 63(2), 234–240.

https://doi.org/10.1016/j.biopsych.2007.04.041

Puliafico, A. C., & Kendall, P. C. (2006). Threat-related attentional bias in anxious

youth: A review. Clinical Child and Family Psychology Review, 9(3-4), 162–180.

https://doi.org/10.1007/s10567-006-0009-x

Ray, R. D., & Zald, D. H. (2012). Anatomical insights into the interaction of emotion and

cognition in the prefrontal cortex. Neuroscience and Biobehavioral Reviews,

36(1), 479–501. https://doi.org/10.1016/j.neubiorev.2011.08.005

81

Roy, M., Shohamy, D., & Wager, T. D. (2012). Ventromedial prefrontal-subcortical

systems and the generation of affective meaning. Trends in Cognitive Sciences,

16(3), 147–156. https://doi.org/10.1016/j.tics.2012.01.005

Schiller, D., Levy, I., Niv, Y., LeDoux, J. E., & Phelps, E. A. (2008). From fear to safety

and back: Reversal of fear in the human brain. The Journal of Neuroscience,

28(45), 11517–11525. https://doi.org/10.1523/JNEUROSCI.2265-08.2008

Schmeichel, B. J., & Demaree, H. A. (2010). Working memory capacity and spontaneous

emotion regulation: High capacity predicts self-enhancement in response to

negative feedback. Emotion, 10(5), 739–744. https://doi.org/10.1037/a0019355

Seeman, T. E., McEwen, B. S., Rowe, J. W., & Singer, B. H. (2001). Allostatic load as a

marker of cumulative biological risk: MacArthur studies of successful aging.

Proceedings of the National Academy of Sciences of the United States of America,

98(8), 4770–4775.

Shiota, M. N., & Levenson, R. W. (2012). Turn down the volume or change the channel?

Emotional effects of detached versus positive reappraisal. Journal of Personality

and Social Psychology, 103(3), 416–429. https://doi.org/10.1037/a0029208

Sinha, R., Lacadie, C. M., Constable, R. T., & Seo, D. (2016). Dynamic neural activity

during stress signals resilient coping. PNAS Proceedings of the National Academy

of Sciences of the United States of America, 113(31), 8837–8842.

Spielberg, J. M., Miller, G. A., Warren, S. L., Engels, A. S., Crocker, L. D., Banich, M.

T., … Heller, W. (2012). A brain network instantiating approach and avoidance

motivation. Psychophysiology, 49(9), 1200–1214. https://doi.org/10.1111/j.1469-

8986.2012.01443.x

82

Spielberg, J. M., Miller, G. A., Warren, S. L., Sutton, B. P., Banich, M., & Heller, W.

(2014). Transdiagnostic dimensions of anxiety and depression moderate

motivation-related brain networks during goal maintenance. Depression and

Anxiety, 31(10), 805–813. https://doi.org/10.1002/da.22271

Spielberg, J. M., Stewart, J. L., Levin, R. L., Miller, G. A., & Heller, W. (2008).

Prefrontal cortex, emotion, and approach/withdrawal motivation. Social and

Personality Psychology Compass, 2(1), 135–153. https://doi.org/10.1111/j.1751-

9004.2007.00064.x

Tedlie Moskowitz, J. (2003). Positive affect predicts lower risk of AIDS mortality.

Psychosomatic Medicine, 65(4), 620–626.

https://doi.org/10.1097/01.PSY.0000073873.74829.23

Treynor, W., Gonzalez, R., & Nolen-Hoeksema, S. (2003). Rumination reconsidered: A

psychometric analysis. Cognitive Therapy and Research, 27(3), 247–259.

https://doi.org/10.1023/A:1023910315561

Trompetter, H. R., Kleine, E., & Bohlmeijer, E. T. (2016). Why does positive mental

health buffer against psychopathology? An exploratory study on self-compassion

as a resilience mechanism and adaptive emotion regulation strategy. Cognitive

Therapy and Research. https://doi.org/10.1007/s10608-016-9774-0

Trueba, A. F., Simon, E., Auchus, R. J., & Ritz, T. (2016). Cortisol response to acute

stress in asthma: Moderation by depressive mood. Physiology & Behavior, 159,

20–26. https://doi.org/10.1016/j.physbeh.2016.03.005

83

Tugade, M. M., Devlin, H. C., & Fredrickson, B. L. (2014). Infusing positive emotions

into life: The broaden-and-build theory and a dual-process model of resilience. In

M. M. Tugade, M. N. Shiota, L. D. Kirby, M. M. Tugade, M. N. Shiota, & L. D.

Kirby (Eds.), Handbook of positive emotions (pp. 28–43). New York: Guilford

Press.

Tugade, M. M., & Fredrickson, B. L. (2004). Resilient individuals use positive emotions

to bounce back from negative emotional experiences. Journal of Personality and

Social Psychology, 86(2), 320–333. https://doi.org/10.1037/0022-3514.86.2.320

Tugade, M. M., Fredrickson, B. L., & Barrett, L. F. (2004). Psychological resilience and

positive emotional granularity: Examining the benefits of positive emotions on

coping and health. Journal of Personality, 72(6), 1161–1190.

https://doi.org/10.1111/j.1467-6494.2004.00294.x

Turowski, T. K., Man, V. Y., & Cunningham, W. A. (2014). Positive emotion and the

brain: The neuroscience of happiness. In J. Gruber, J. T. Moskowitz, J. Gruber, &

J. T. Moskowitz (Eds.), Positive emotion: Integrating the light sides and dark

sides. (pp. 95–115). New York: Oxford University Press.

Urry, H. L., van Reekum, C. M., Johnstone, T., Kalin, N. H., Thurow, M. E., Schaefer, H.

S., … Davidson, R. J. (2006). Amygdala and ventromedial prefrontal cortex are

inversely coupled during regulation of negative affect and predict the diurnal

pattern of cortisol secretion among older adults. The Journal of Neuroscience,

26(16), 4415–4425. https://doi.org/10.1523/JNEUROSCI.3215-05.2006

84

Vanlessen, N., De Raedt, R., Koster, E. H. W., & Pourtois, G. (2016). Happy heart,

smiling eyes: A systematic review of positive mood effects on broadening of

visuospatial attention. Neuroscience and Biobehavioral Reviews, 68, 816–837.

https://doi.org/10.1016/j.neubiorev.2016.07.001

Waugh, C. E., & Fredrickson, B. L. (2006). Nice to know you: Positive emotions, self-

other overlap, and complex understanding in the formation of a new relationship.

The Journal of Positive Psychology, 1(2), 93–106.

https://doi.org/10.1080/17439760500510569

Waugh, C. E., Hamilton, J. P., Chen, M. C., Joormann, J., & Gotlib, I. H. (2012). Neural

temporal dynamics of stress in comorbid major depressive disorder and social

anxiety disorder. Biology of Mood & Anxiety Disorders, 2, 11.

https://doi.org/10.1186/2045-5380-2-11

Waugh, C. E., Hamilton, J. P., & Gotlib, I. H. (2010). The neural temporal dynamics of

the intensity of emotional experience. NeuroImage, 49(2), 1699–1707.

https://doi.org/10.1016/j.neuroimage.2009.10.006

Waugh, C. E., Lemus, M. G., & Gotlib, I. H. (2014). The role of the medial frontal cortex

in the maintenance of emotional states. Social Cognitive and Affective

Neuroscience, 9(12), 2001–2009. https://doi.org/10.1093/scan/nsu011

Welborn, B. L., Papademetris, X., Reis, D. L., Rajeevan, N., Bloise, S. M., & Gray, J. R.

(2009). Variation in orbitofrontal cortex volume: Relation to sex, emotion

regulation and affect. Social Cognitive and Affective Neuroscience, 4(4), 328–

339. https://doi.org/10.1093/scan/nsp028

85

Winecoff, A., Clithero, J. A., Carter, R. M., Bergman, S. R., Wang, L., & Huettel, S. A.

(2013). Ventromedial prefrontal cortex encodes emotional value. The Journal of

Neuroscience, 33(27), 11032–11039. https://doi.org/10.1523/JNEUROSCI.4317-

12.2013

Xue, S., Cui, J., Wang, K., Zhang, S., Qiu, J., & Luo, Y. (2013). Positive emotion

modulates cognitive control: An event-related potentials study. Scandinavian

Journal of Psychology, 54(2), 82–88. https://doi.org/10.1111/sjop.12031

Yang, Y., Zhong, N., Friston, K., Imamura, K., Lu, S., Li, M., … Hu, B. (2017). The

functional architectures of addition and subtraction: Network discovery using fmri

and dcm. Human Brain Mapping. https://doi.org/10.1002/hbm.23585

Zeidan, F. (2015). The neurobiology of mindfulness meditation. In K. W. Brown, J. D.

Creswell, R. M. Ryan, K. W. Brown, J. D. Creswell, & R. M. Ryan (Eds.),

Handbook of mindfulness: Theory, research, and practice. (pp. 171–189). New

York: Guilford Press.

86

APPENDIX A

Modified Experiences Questionnaire

Introduction:

The Modified Experiences Questionnaire (EQ) is a 17-item self-report inventory

designed to measure a decentering or dis-identification with content of negative thinking,

which is hypothesized to be a process of change in Mindfulness-Based Cognitive

Therapy. The EQ was originally designed to possess two rationally-derived subscales:

rumination (6 items) and wider perspective (11 items). Rumination items (e.g., “I think

over and over again about what others have said to me”) were included in the EQ to help

rule out the possibility that any increases in wider perspective might be explained by an

acquiescent response bias. Items are rated on a 5-point likert scale with a rating of 1

corresponding to the anchor “Not at All” and 5 corresponding to the anchor “Very

much.”

Content:

Please read the following statements and indicate how much time you engaged in such

thoughts or how much you endorse such thoughts while you were watching the video just

now.

1. EQ01 I thought about what would happen next.

2. EQ03 I was able to accept myself as I am.

3. EQ04 I noticed all sorts of little things and details in the world around me.

4. EQ06 I slowed my thinking at the time of stress.

5. EQ07 I wondered what kind of person I really am.

6. EQ09 I noticed that I didn’t take difficulties so personally.

87

7. EQ10 I separated myself from my thoughts and feelings.

8. EQ11 I analyzed why things turned out the way they did.

9. EQ12 I took the time to respond to difficulties.

10. EQ13 I thought over and over again about what others have said to me.

11. EQ14 I treated myself kindly.

12. EQ15 I observed unpleasant feelings without being drawn into them.

13. EQ16 I had the sense that I was fully aware of what was going on around me and

inside me.

14. EQ17 I actually saw that I am not my thoughts.

15. EQ18 I was consciously aware of a sense of my body as a whole.

16. EQ19 I thought about the ways in which I am different from other people.

17. EQ20 I viewed things from a wider perspective.

More information on factors:

Factor 1 (Decentering)

EQ03 I was able to accept myself as I am.

EQ15 I observed unpleasant feelings without being drawn into them.

EQ09 I noticed that I didn’t take difficulties so personally.

EQ14 I treated myself kindly.

EQ10 I separated myself from my thoughts and feelings.

EQ16 I had the sense that I was fully aware of what was going on around me and inside

me.

EQ06 I slowed my thinking at the time of stress.

EQ17 I actually saw that I am not my thoughts.

88

EQ18 I was consciously aware of a sense of my body as a whole.

EQ12 I took the time to respond to difficulties.

EQ20 I viewed things from a wider perspective.

Factor 2 (Rumination)

EQ13 I thought over and over again about what others have said to me.

EQ11 I analyzed why things turned out the way they did.

EQ19 I thought about the ways in which I am different from other people.

EQ04 I noticed all sorts of little things and details in the world around me.

EQ01 I thought about what would happen in the future.

EQ07 I wondered what kind of person I really am.

89

APPENDIX B

Thought Content Questionnaire

Introduction:

The Thought Content Questionnaire is a 14-item self-report inventory designed to

record the thoughts participants have during the 3-minutes video watching period. It

possesses five past thought data derived subscales: negative thought and emotion about

the anagram test, positive thought and emotion about the anagram test, negative thought

and emotion about the video, positive thought and emotion about the video, other

thoughts and emotions unrelated to the test or the video. Items are rated on a 5-point

likert scale with a rating of 1 corresponding to the anchor “Not at All” and 5

corresponding to the anchor “Very much.”

Content:

We are interested in what thoughts you might have had while you were watching the

video just now. Please read the following statements and indicate for how much time you

engaged in these thoughts while you were watching the video. Please try as much as

possible to indicate

1. I had pleasant thoughts about the content of the video.

2. I had unpleasant thoughts about the content of the video.

3. I thought that the video was calming.

4. I thought that the video was boring.

5. I thought that the video was exciting.

6. I thought that the video was annoying.

7. I had mostly positive thoughts about the anagram test.

90

8. I had mostly negative thoughts about the anagram test.

9. I thought about how frustrated I was with the anagram test.

10. I thought about how sad I was about the anagram test.

11. I thought about how fine I was with the anagram test.

12. I thought about how content I was with the anagram test.

13. I thought about things unrelated to the anagram test or the video.

14. I thought about other things besides this experiment.

91

APPENDIX C

Emotion Regulation Questionnaire

ERQ

We would like to ask you some questions about your emotional life, in particular,

how you control (that is, regulate and manage) your emotions. We are interested in two

aspects of your emotional life. One is your emotional experience, or what you feel like

inside. The other is your emotional expression, or how you show your emotions in the

way you talk, gesture, or behave. Although some of the following questions may seem

similar to one another, they differ in important ways. For each item, please answer using

the following scale:

1 2 3 4 5 6 7

strongly neutral strongly disagree agree

------------------------------------------------------------------------------------------------------------ 1. ____ When I want to feel more positive emotion (such as joy or amusement), I change

what I’m thinking about.

2. ____ I keep my emotions to myself.

3. ____ When I want to feel less negative emotion (such as sadness or anger), I change

what I’m thinking about.

4. ____ When I am feeling positive emotions, I am careful not to express them.

5. ____ When I’m faced with a stressful situation, I make myself think about it in a way

that helps me stay calm.

6. ____ I control my emotions by not expressing them.

92

7. ____ When I want to feel more positive emotion, I change the way I’m thinking about

the situation.

8. ____ I control my emotions by changing the way I think about the situation I’m in.

9. ____ When I am feeling negative emotions, I make sure not to express them.

10. ____ When I want to feel less negative emotion, I change the way I’m thinking about

the situation.

93

APPENDIX D

Ego-Resiliency Scale

ER89

Please read each item below and circle the number that best corresponds to the

following scale:

1-----------------2---------------3-----------------4

Does not apply at all applies slightly applies somewhat applies very strongly

1. I am generous with my friends. 1 2 3 4

2. I quickly get over and recover from being startled. 1 2 3 4

3. I enjoy dealing with new and unusual situations. 1 2 3 4

4. I usually succeed in making a favorable impression 1 2 3 4

on people.

5. I enjoy trying new foods I have never tasted before. 1 2 3 4

6. I am regarded as a very energetic person. 1 2 3 4

7. I like to take different paths to familiar places. 1 2 3 4

8. I am more curious than most people. 1 2 3 4

9. Most of the people I meet are likeable. 1 2 3 4

10.I usually think carefully about something before acting. 1 2 3 4

11.I like to do new and different things. 1 2 3 4

12.My daily life is full of things that keep me interested. 1 2 3 4

13.I would be willing to describe myself as a pretty 1 2 3 4

"strong" personality.

94

14.I get over my anger at someone reasonably quickly. 1 2 3 4

95

APPENDIX E

Neuroticism and Extraversion Subscales from Neuroticism-Extraversion-Openness

Personality Inventory (NEO)

Instructions: Read each statement carefully. For each statement, circle the number that

best represents your opinion, i.e., the extent to which you agree or disagree that the

statement applies to you. Use a number from the following scale:

Strongly Disagree Disagree Neutral Agree Strongly Agree

1 2 3 4 5

1. I’m not a worrier.

2. I like to have a lot people around me.

3. I often feel inferior to others.

4. I laugh easily.

5. When I’m under a great deal of stress, sometimes I feel like I’m going to pieces.

6. I don’t consider myself especially “light-hearted”.

7. I rarely feel lonely or blue.

8. I really enjoy talking to people.

9. I often feel tense and jittery.

10. I like to be where the action is.

11. Sometimes I feel completely worthless.

12. I usually prefer to do things alone.

13. I rarely feel fearful or anxious.

14. I often feel as if I’m bursting with energy.

96

15. I often get angry at the way people treat me.

16. I am a cheerful, high-spirited person.

17. Too often, when things go wrong, I get discouraged and feel like giving up.

18. I am not a cheerful optimist.

19. I am seldom sad or depressed.

20. My life is fast-paced.

21. I often feel helpless and want someone else to solve my problems.

22. I am a very active person.

23. At times, I have been so ashamed I just want to hide.

24. I would rather go my own way than be a leader of others.

97

APPENDIX F

Ruminative Response Scale

People think and do many different things when they feel depressed. Please read

each of the items below and indicate whether you almost never, sometimes, often, or

almost always think or do each one when you feel down, sad, or depressed. Please

indicate what you generally do, not what you think you should do.

1 almost never 2 sometimes 3 often 4 almost always

1. think about how alone you feel

2. think “I won’t be able to do my job if I don’t snap out of this”

3. think about your feelings of fatigue and achiness

4. think about how hard it is to concentrate

5. think “What am I doing to deserve this?”

6. think about how passive and unmotivated you feel.

7. analyze recent events to try to understand why you are depressed

8. think about how you don’t seem to feel anything anymore

9. think “Why can’t I get going?”

10. think “Why do I always react this way?”

11. go away by yourself and think about why you feel this way

12. write down what you are thinking about and analyze it

13. think about a recent situation, wishing it had gone better

14. think “I won’t be able to concentrate if I keep feeling this way.”

98

15. think “Why do I have problems other people don’t have?”

16. think “Why can’t I handle things better?”

17. think about how sad you feel.

18. think about all your shortcomings, failings, faults, mistakes

19. think about how you don’t feel up to doing anything

20. analyze your personality to try to understand why you are depressed

21.go someplace alone to think about your feelings

22. think about how angry you are with yourself

99

CURRICULUM VITAE

Xi Yang

EDUCATION

08/2015 – present Master of Psychology GPA: 3.71

Wake Forest University, Graduate School of Arts and Science, Winston Salem, NC

First Year Project: Positive Emotions Facilitate Stress Recovery Master Thesis: Neural Correlates of Positive Emotions in

Reflection

Facilitating Stress Recovery (forthcoming)

09/2013 – 05/2015 Master of Psychology GPA: 3.89

New York University, Graduate School of Arts and Science, New York, NY

Master Thesis: Self-Regulation of Disgust and Fear Through

Cognitive Reappraisal

09/2010 – 06/2013 Bachelor of Arts in English GPA: 3.61 Rank: 1/118 China Pharmaceutical University, School of Foreign Languages, Nanjing, JS, China Thesis: Lexical Analysis of Depressed Novelists’ Works

RESEARCH

07/2015 – present Research Assistant, Project Coordinator

Dr. Christian Waugh’s Emotion, Adaptation, and Psychophysiology Lab at Wake Forest University • Reviewed literature on positive emotions and stress. • Conducted behavioral and fMRI studies testing how positive

emotions facilitate stress recovery, including designing experiment, recruiting community participants, performing experiments, analyzing data and writing up results.

• Lead data coding and performed modified contrast analysis to identify the resilience thought pattern.

• Trained on comprehensive psychophysiological tools, trained undergraduate RAs and assisted with IRB communications.

• SANS 2017 poster accepted and paper preparation under way.

100

04/2016 – present Research Assistant Dr. Eric Stone, independent project at Wake Forest University

• Reviewed literature on statistical power and small sample studies.

• Calculated and designed a statistical model in R to show highest result validity is achieved when Bayesian odds ratio matches the power.

• Now designing questionnaire items to collect information with misconceptions about power and results validity and prepare paper for submission in summer.

01/2014 – 05/2015 Research Assistant

Dr. Hanah Chapman’s Emotion and Morality Lab in Brooklyn College (City University of New York) • Reviewed literature on basic emotions with a deep focus on

disgust. • Conducted behavioral experiments on disgust regulation,

including designing studies, performing experiments, coding data and data analysis.

• Collected data with psychophysiological tools including GSR, EMG and EGG.

• Now performing further data analysis to synthesize results from three studies for paper preparation.

INTERNSHIPS

04/2014 – 05/2015 Rehabilitative Counselor Intern Asian American Recovery Service at Hamilton Madison House

• Participated in individual, group supervisions, weekly clinical meetings, and in-service training seminars.

• Co-lead group therapy weekly and responsible for the social and vocational skills training for one client.

• Wrote chart notes including utilization review and progress notes under supervision.

• Performed screening and assessment for the intake and presented on club drugs to educate patients’ significant others.

• Tracked patients’ medication records and monitored on metabolic syndrome for patients taking atypical antipsychotics.

• Provided case management and program administrative service, and involved in outreach activities.

05/2013 – 07/2013 Counselor Assistant, Zhimian Psychotherapy Institute

• Attended weekly volunteer counselors group supervision. • Engaged in organizing seminars in Existential, Humanistic

and Gestalt Psychology, and performed administrative work.

101

08/2012 Therapist Assistant Children Mental Health Research Center, Nanjing Brain

Hospital • Participated in recovery training of children with Sensory

Integration Disorder and/or Autism Spectrum Disorder.

AWARDS/HONORS

2017 Nominated for Wake Forest University Gordon A. Melson Outstanding Student Award (pending)

Summer 2016 National Institute of Health Grant ($2,040) Project: Investigating the neural systems that support the beneficial effects of positive emotion on stress regulation

Role: Project Coordinator, Research Assistant Summer 2015 National Institute of Health Grant ($1,012.5)

Project: Investigating the neural systems that support the beneficial effects of positive emotion on stress regulation

Role: Research Assistant 04/2014 New York University Wasserman Career Center Internship Grant

($1,000) 05/2013 First Place and Best Debater in 17th “FLTRP DANGDANG.COM

CUP” National Universities English Debating Competition (campus wide)

05/2013 Individual Debater Rank 6th, Semi-final Team and Silver Medal in NANLIN CUP Spring Tournament by China Debate Education (Eastern China Area)

05/2013 First student in department to be granted early graduation from the four-year bachelor’s education due to academic excellence

2011 – 2012 China Pharmaceutical University Chancellor's Scholarship ($1,300) 2011 – 2012 China National Scholarship for Excellent College Students ($850) 11/2012 Second Prize in 9th China-Australia English Lecture Competition

(Eastern China Area) 2010 – 2011 China National Scholarship for Excellent College Students ($850) 2010 – 2011 First Prize School Scholarship for Excellent College Students

($150) PRESENTATIONS

Talks Yang, X., Reigada, L., & Chapman, H. (September, 2015). Self-regulation of disgust and fear

through cognitive reappraisal. Talk presented at Wake Forest University Psychology Department in Seminar in Self-Regulation, Winston-Salem, NC.

Yang, X., Reigada, L., & Chapman, H. (November, 2014). Self-regulation of disgust and fear

through cognitive reappraisal. Talk presented at Brooklyn College in-house conference.

Posters Yang, X., Garcia, K., & Waugh, C. (March, 2017). Change-point analyses of positive emotions

supporting stress recovery. Poster presented at the 10th annual Social and Affective Neuroscience Society meeting, Los Angeles, CA.

Yang, X., & Zhang, Y. X. (July, 2015). When empathy questionnaires meet a narcissist. Poster presented at the International Meeting of the Psychometric Society, Beijing, China.

102

Yang, X., Reigada, L., & Chapman, H. (May, 2015). Self-regulation of disgust and fear

through cognitive reappraisal. Poster presented at the 27th Association for Psychological Science Annual Convention, New York, NY.

Yang, X., & Zhang, Y. X. (April, 2015). The validity of empathy measures assessing narcissism. Poster presented at the 19th Annual New York University MA Psychology Research Conference, New York, NY.

SKILLS

E-Prime, Biopac system (EDA, facial EMG, Noninvasive BP, ECG, ICG (Impedance Cardiography), EGG (Electrogastrogram)) WordSmith, MatLab (SPM12), SPSS, R

LANGUAGE PROFICIENCY 2012 – 2016 GRE Verbal 160 85th percentile

Quantitative 169 97th percentile Assay 4.0 59th percentile

2012 TOEFL Total 108 Reading 29 Listening 27 Speaking 24 Writing 28

PROFESSIONAL MEMBERSHIP

2017-2018 Association for Psychological Science 2015-2016 Association for Psychological Science