NYLOCKS, K. MARIA., Ph.D.

, August 2020 PSYCHOLOGICAL SCIENCES

FEAR-PATHOLOGY ETIOLOGY: FEAR REACTIVITY, FEAR RECOVERY, AND

REGULATORY RESOURCES (115 PP.)

Dissertation Advisor: Karin G. Coifman

Fear-based psychological disorders such as social anxiety disorder (SAD), posttraumatic stress

disorder (PTSD), generalized anxiety disorder (GAD), obsessive-compulsive disorder (OCD),

and specific phobias, are among the most prevalent and costly mental health problems today.

Despite the prevalence of these disorders, their etiology is not yet clearly defined. It remains

elusive why some individuals develop fear-pathology while others do not. A primary challenge

facing research on the etiology of fear-pathology is parsing out if increased generation of fear in

response to threat (greater fear reactivity), or poor recovery (decreased fear inhibition), is more

predictive of disease. Part of this challenge is also understanding of the association between fear

reactivity and fear inhibition. A secondary challenge is reaching a better understanding how

regulatory resources might influence fear reactivity and/or fear inhibition. The current study was

designed to explore these two challenges in order to reach a better understanding of fear-

pathology development.

Data was collected from n=101 college students during one laboratory session.

Participants were asked to complete a fear reactivity task consisting of two film clips to elicit

fear, each followed by a recovery period (without stimuli presentation), as well as two positive

emotion videos. This task was specifically designed to examine fear reactivity within the context

of threat (during the fear videos) and fear inhibition immediately following fear elicitation
(during the recovery periods). During the entire task, emotional responses were measured in real-

time on multiple response dimensions including, coded emotional facial expressions,

sympathetic arousal (autonomic activity), and self-reported emotional experience. During the

laboratory session, participants also completed questionnaires to index psychological symptoms

and threat sensitivity, as well as two computerized Stroop tasks to index executive cognitive

functioning. The current study aimed to model trajectories in fear responding (indexed by skin

conductance) by using latent growth mixture modeling (LGMM), and derive latent groups

characterized by their pattern of fear responding. We also aimed to examine latent groups for

differences in psychological symptoms, reported threat sensitivity, and regulatory resources

(Stroop task performance and positive emotion expression). Lastly, we examined the association

between fear reactivity and fear recovery (utilizing change scores) for the full sample. Results

from the current study indicate that meaningful individual differences in fear responding are

likely to emerge in response to removal of a novel threat, where fear inhibition is expected.

Further, our findings suggest a link between psychological risk (reported threat sensitivity) and

decreased habituation to a familiar threat, which is moderated by executive cognitive functioning

(indexed by Stroop reaction time in the current study). Indeed, our data suggest that increased

executive functioning might serve as a regulatory resource within the context of fear. These

findings have important clinical implications by adding to existing literature on fear-pathology.

Indeed, our findings support current models of fear-pathology, and add to the literature on how

executive cognitive resources can influence threat responding.

 FEAR-PATHOLOGY ETIOLOGY:

FEAR REACTIVITY, FEAR RECOVERY, AND REGULATORY RESOURCES

A dissertation submitted

to Kent State University in partial

Fulfillment of the requirements for the

Degree of Doctor of Philosophy

By

K. Maria Nylocks

August 2020

© Copyright

All rights reserved

Except for previously published materials

 Dissertation written by

K. Maria Nylocks

B.A., Georgia State University, 2012

M.A., Kent State University, 2016

Ph.D., Kent State University, 2020

Approved by

Karin G. Coifman, Ph.D. , Chair, Doctoral Dissertation Committee

Aaron M. Jasnow, Ph.D. , Members, Doctoral Dissertation Committee

John Gunstad, Ph.D.

Tanja Jovanovic, Ph.D.

Susan Roxburgh, Ph.D.

Accepted by

Maria Zaragoza, Ph.D. , Chair, Department of Psychological Sciences

Mandy Munro-Stasiuk, Ph.D. , Interim Dean, College of Arts and Sciences

TABLE OF CONTENTS………………………………………………………………...........................v

LIST OF FIGURES……………………………………………………………………………………...vi

LIST OF TABLES……………………………………………………………………………………...vii

ACKNOWLEDGMENTS…………………………………………………………………..................viii

INTRODUCTION………………………………………………………………………………………..1

THE CURRENT STUDY………………………………………………………………………………15

METHOD……………….………………………………………………………………………………19

RESULTS……………………………………………………………………………………………….41

DISCUSSION……………………………………………………………………………………..……53

REFERENCES……………………………………………………………………………………...…..78

APPENDICES………………………………………………………………………………………... 101

v
LIST OF FIGURES

Figure 1. Self-reported negative emotion experience across the fear reactivity task……………72

Figure 2. Fear expression across the fear reactivity task………………………………………...73

Figure 3. Mean skin conductance response across the fear reactivity task………………………74

Figure 4: Latent growth model (six time points) with covariate included……………………….74

Figure 5. First fear video segment, 1-class solution……………………………………………..75

Figure 6. First fear recovery period, 2-class solution……………………………………………75

Figure 7. Second fear video, 3-class solution……………………………………………………76

Figure 8. Second fear recovery period, 1-class solution…………………………………………76

Figure 9. Stroop task difference score (negative emotion context) moderating the relationship

between fear reactivity observed during the second fear video (indexed by skin conductance) and

reported threat sensitivity………………………………………………………………………...77

vi
LIST OF TABLES

Table 1. Sample demographic characteristics………………………………………....................64

Table 2. Stroop task interference scores……………………………………………....................65

Table 3. Missing data. ………………………………………………………………....................66

Table 4: LGMM model fit statistics, per task segment (*=best fit)………………….............67-68

Table 5. Pearson correlations between skin conductance reactivity, recovery, psychological

symptoms, Stroop task interference difference score, and positive emotion expression………...69

Table 6. Pearson correlations between reported negative emotion reactivity, recovery,

psychological symptoms, Stroop task interference difference score, and positive emotion

expression………………………………………………………………………………………..70

Table 7. Pearson correlations between fear expression reactivity, recovery, psychological

symptoms, Stroop task interference difference score, and positive emotion expression………...71

vii
 ACKNOWLEDGEMENTS

I would like to thank all the people who assisted, inspired, and conducted this research with me.

It is certainly a good feeling to finish up this document, as it is the product of many years of hard

work. First, I would like to thank my advisor Dr. Karin Coifman. Thank you for your

mentorship, guidance, feedback on countless drafts, and for showing me how to be a scientist. I

would also like to thank you for offering your continued mentorship as I leave graduate school, I

consider myself very lucky to have such a strong relationship with my academic advisor.

With my whole heart, I would also like to thank my friends. My lab mates in the ESR

Lab and my amazing graduate school cohort, I know that I would not have made it through this

journey without you. We will forever be friends. Most of all, I would like to thank my wonderful

family for showing me unconditional love, compassion, and support, when I needed it the most.

Thank you to my husband Ryan, my parents Suzanne and Lars, and brother David. I love you all.

viii
 I. INTRODUCTION

Fear-based psychological illnesses, commonly referred to as anxiety disorders (American

Psychiatric Association; [APA] 2013), have significant worldwide lifetime prevalence rates.

Across all age groups, these disorders, including social anxiety disorder (SAD) (12.1%),

posttraumatic stress disorder (PTSD) (6.8%), generalized anxiety disorder (GAD) (5.7%), panic

disorder (4.7%), agoraphobia without panic (1.4%), obsessive-compulsive disorder (OCD)

(1.6%), and specific phobias (12.5%) (Kessler et al., 2005), are prevalent in the general

population. Individuals with these disorders suffer impairment across several domains of life

including, education attainment, stability and quality of romantic relationships, employment,

parental functioning, and risk for physical illness (Kessler et al., 2014). Despite the prevalence of

these disorders, their etiology is not yet clearly defined. It remains elusive why some individuals

develop fear-pathology while others do not.

Fear

Functional theories of emotion suggest that fear has high adaptive value and functions as

a defense response within the context of a true threat (Darwin, 1872/1998; Ekman, 1992;

LeDoux, 2012; Mineka, 2013; Ohman & Mineka, 2001). Fear, like most discrete emotions (e.g.

disgust, sadness, joy, anger) consists of multiple response dimensions including behavioral (e.g.

emotional facial behaviors), experiential (e.g. self-reported emotion), and physiological (e.g.

autonomic nervous system activity). These response dimensions are loosely coordinated (Bulteel

et al., 2014; Bonanno & Keltner, 1997) in order to facilitate specific actions in response to

environmental threats or opportunities (Ekman, 1992). For example, fear can involve movement

1
away from a threat (behavioral), the experience and identification of “feeling afraid”

(experiential), and elevated heart rate and increased skin conductance (physiological). Indeed,

the individual facing threat might then be able to escape quickly to avoid the threat, a response

that is adaptive. However, sustained fear (i.e. poor fear inhibition) once the threat is no longer

present is a well-established risk factor for fear-based psychological disorders. Increased fear

reactivity (greater intensity of the fear response) has also been theorized to predict fear-

pathology, and there is evidence to suggest that increased fear reactivity in response to aversive

events might lead to sustained fear (Jovanovic et al., 2010; Norrholm et al., 2011; 2015). Indeed,

a primary challenge facing research on the etiology of fear-pathology is parsing out if increased

generation of fear in response to threat (greater fear reactivity), or poor recovery (diminished fear

inhibition) following resolution of the threat, is more predictive of disease. Part of this challenge

is also understanding the association between fear reactivity and fear inhibition.

Fear reactivity

Broadly, fear reactivity refers to increased intensity of the fear response (e.g. Gross &

Jazaieri, 2014). A hallmark of most fear-based psychological disorders is increased fear

reactivity or symptoms of hyperarousal. It has been theorized that pathological fear, as compared

to normative fear which does not cause impairment, consists of a disproportionately strong fear

response (i.e. increased fear reactivity) that is resistant to change (Foa, & Kozak, 1986; Foa,

Steketee, & Rothbaum, 1989). For example, a common symptom of PTSD is intense

physiological reaction (e.g. sweating, heavy breathing, heart racing) in response to internal and

external fear-cues, and a common symptom of GAD is muscle tension and feeling “keyed up.”

Similarly, specific phobias are characterized by a highly intense fear response to specific events

or objects, and panic disorder is characterized by unexpected fear-related physical symptoms

2
(e.g. accelerated heart rate, sweating) and fear of losing control or a fear of dying (APA, 2013).

Indeed, a meta-analysis of fear conditioning studies (i.e. experimental paradigms where a neutral

stimulus is paired with an aversive stimulus and, after repeated pairings, the neutral stimulus

likely elicits a fear response) found evidence of increased fear intensity across anxiety disorders

(Lissek et al., 2005). There is also neuroimaging evidence to support increased activation of the

limbic system across these disorders (Etkin & Wager, 2007; Shin & Liberzon, 2010). For

example, increased activity in the left amygdala in response to moderately to highly threatening

facial expressions has been found in individuals with SAD, as compared to healthy controls

(Klumpp et al., 2010). Evidence for increased amygdala activity in response to threatening cues

has also emerged in samples with panic disorder (van den Heuvel et al., 2005), specific phobia

(Dilger et al., 2003), and GAD (Krain et al., 2008; McClure et al., 2007; Monk et al., 2008).

Both theory and empirical evidence regarding fear reactivity, and associated risks, has

contributed to the development of therapeutic interventions (e.g. exposure therapy) to treat fear-

based psychological illness. Currently, exposure therapy is well-known and frequently

recommended for cases of fear-pathology such as PTSD and specific phobias (Foa et al., 1991;

Foa et al., 1999; Foa et al., 2005; Powers et al., 2010; Resick et al., 2012). Specifically, exposure

therapy consists of repeated contact with (or approach towards) feared (but not dangerous)

stimuli, and the main target of the therapy is to decrease fear reactivity in response to the feared

stimuli, or achieve habituation (i.e. ‘the decline in response to a repeatedly presented stimulus;’

Mackintosh, 1987). Indeed, exposure therapy largely assumes that successful habituation

achieved in therapy sessions will lead to decreased anxiety and hyperarousal symptoms in

everyday life. The end goal of exposure therapy is fear extinction (i.e. reduced fear reactivity in

response to stimuli that were previously experienced as dangerous). However, there is evidence

3
to suggest that habituation might not predict fear extinction that is long-lasting (Brown, LeBeau,

Chat, & Craske, 2017; Craske, Liao, Brown, & Vervliet, 2012; Phelps, Delgado, Nearing, &

LeDoux, 2004). For example, in a fear-potentiated startle paradigm, researchers found that long-

term fear extinction (assessed seven days following the initial startle paradigm) was not

predicted by habituation (measured by self-reported fear and eyeblink response) during

extinction training (Brown, LeBeau, Chat, & Craske, 2017). Thus, although current therapies that

rely on the process of habituation seem to produce symptom relief in the short-term, these

treatment protocols are often followed by relapse (Ali et al., 2017; Ginsburg et al., 2014; Loerinc

et al., 2015; Olatunji, Cisler, Deacon, 2010; Vervliet, Craske, & Hermans, 2013). Several studies

report that approximately half of individuals treated for anxiety disorders relapse following

treatment (e.g. Ali et al., 2017; Ginsburg et al., 2014). For example, Pallanti and colleagues

(2002) found that 55% of participants relapsed following exposure therapy for OCD (relapse

defined as 25% increase in symptoms from post-treatment symptom levels). These data indicate

that when fear reactivity is the main focus of treatment, benefits are likely short-term, suggesting

that fear reactivity might explain only part of the variability in fear-pathology. Further, although

these data suggest that fear reactivity plays a role in fear-pathology development, there are likely

other components of the fear response that influence etiology of these disorders as well.

Fear recovery

Sustained fear, or poor fear inhibition following the removal of threat, is another

hallmark of fear-based psychological illness. For example, individuals with SAD often

experience persistent physical symptoms of fear (e.g. sweating, elevated heart rate) in social

situations. Although it might be normative to feel increased anxiety when entering into a novel

social situation, SAD is characterized by fear that persists even after the social situation could be

4
considered “safe” or emotionally “neutral” (APA, 2013). Similarly, PTSD is characterized by

prolonged distress in response to cues that might only resemble trauma-related cues, but that do

not present an actual current threat. Further, GAD consists of persistent worry and anxiety that is

difficult to control (APA, 2013), and individuals with GAD often report fear in response to a

wide variety of abstract contexts that are most often not dangerous (e.g. but rather ambiguous

perceived danger in the community, even when most other inhabitants rate the community as

safe).

A substantial body of research suggests that sustained fear that is resistant to explicit

safety cues plays an important role in the etiology of fear-pathology (e.g. Craske et al., 2012;

Jovanovic & Ressler, 2010). For example, impaired fear inhibition and increased startle eyeblink

response to safety cues have emerged as present across anxiety disorders (Craske et al., 2008;

2009; Lissek et al., 2009; Waters, Neumann, Henry, Craske, & Ornitz, 2008). Based on this

theory and empirical research, Craske and colleagues (2012) conducted an experiment in which

adolescents (with no current psychiatric diagnoses) were presented with eight repetitions of

“safe-danger” sequences with the threat of an aversive biceps contraction during the “danger”

condition, and no threat of contraction during the “safety” condition. This study found that

increased fear response to explicit safety cues (immediately following presentation of a danger

cue) predicted increased risk of fear-pathology three to four years following the experiment. In

the same sample, findings have also emerged showing increased fear response to explicit safety

cues (immediately following presentation of a danger cue) in adolescents with current fear-

pathology (specifically, specific phobia and social anxiety disorder) (Waters et al., 2014).

Similarly, Buss (2011) found that toddlers (two years of age) with increased fear reactivity in

low threat environments were more likely to develop symptoms of anxiety (indexed by maternal

5
report of anxiety and social withdrawal) by age five years, as compared to toddlers who

demonstrate increased fear reactivity in high threat but not in low threat environments. These

data indicate that diminished fear inhibition observed during moments of safety (or low levels of

threat), might predict both the presence of current fear-pathology, as well as the onset of future

fear-pathology. Indeed, poor fear inhibition is an important component of the fear response, and

is likely to explain a significant portion of the variance in fear-pathology.

Association between Fear Reactivity and Fear Inhibition

There is neurobiological evidence to suggest that fear reactivity and fear inhibition are

linked by bottom-up/top-down attentional processes (Ochsner, 2009). Indeed, the amygdala has

been found to be highly involved in fear reactivity by relaying information from the environment

(regarding, for example, potential danger) to other brain areas (lateral hypothalamus,

parabrachial nucleus, basal forebrain) (i.e. bottom-up processing). Through this process,

signaling from the amygdala initiate engagement of coordinated response systems needed for

adaptive action (i.e. increasing fear reactivity). Other areas of the brain that are involved in the

fear response includes the hippocampus and the prefrontal cortex (PFC) (Morrison & Ressler,

2014), and connections between the amygdala and these other brain areas are largely involved in

fear inhibition (Davidson, 2002; Sotres-Bayon, Bush, & LeDoux, 2004). Specifically, there is

evidence to suggest that the ventromedial PFC might function to inhibit fear by inhibiting the

amygdala response (Pape & Pare, 2010) (i.e. top-down processing via executive resources). In

individuals with fear-pathology, attentional biases have been found, such that these individuals

might respond faster to incoming sensory information regarding threat (bottom-up processing)

and then interpret this incoming information at a slower pace (top-down processing), as

compared to healthy individuals (Bar-Haim et al., 2007). Indeed, individuals with anxiety

6
disorders have been found to be more likely, as compared to healthy controls, to direct their

attention towards threat or threat-related stimuli (Eysenck et al., 2007; Mathews & MacLeod,

1994), and this might lead to interference of top-down processing (poor fear inhibition).

Additionally, research on threat sensitivity (i.e. increased hypervigilance and threat avoidance)

(Beauchaine & Thayer, 2015), commonly measured by self-report (Behavioral Inhibition System

scale of the BIS/BAS questionnaire) (Carver & White, 1994), has shown that increased threat

sensitivity is indeed linked to biased attention towards threat and greater detection of threat in the

environment. This link might influence negative emotional states and contribute to sustained

negative emotions (Balle, Tortella-Feliu, & Bornas, 2013). Consistent with these findings are

neuroimaging studies showing both greater amygdala responding (Etkin & Wager, 2007; Shin &

Liberzon, 2010), and dampened ventromedial PFC activity across anxiety disorders, as compared

to healthy controls (Bishop, Duncan, & Lawrence, 2004; Bishop, Duncan, Brett, & Lawrence,

2004; Ohman, 2005). Taken together, this research suggests that there is indeed a link between

fear reactivity and fear inhibition, and further, that attentional processes might play an important

role in this relationship.

There is a growing body of research examining the relationship between emotional

responding and attentional processes, and current evidence suggests that this relationship might

be influenced by important individual differences in executive cognitive resources (Dennis &

Chen, 2007a; 2007b; Derryberry & Reed, 2002; Gray & Burgess, 2004; Coifman, Halachoff, &

Nylocks, 2018). For example, Dennis & Chen (2007) conducted a study where they recorded

ERPs (event-related potentials) responses (specifically, N200 a response known to be increased

when attentional control is needed) (to index attentional control) while participants completed an

emotional version (included emotional images) of the Attention Network Test (ANT; Fan et al.,

7
2002). Participants in this study were grouped by scores on the BIS scale (to index threat

sensitivity), to create a high-BIS and a low-BIS group. Results from this research show that

individuals who report high threat sensitivity in combination with increased attentional control

within the context of negative emotion, show reduced negative impact of the negative emotional

context. Notably, this study also suggests a likely “balance” between threat sensitivity and

executive resources that afford psychological benefits. Indeed, Dennis and Chen (2007a; 2007b)

report that low-BIS participants who demonstrated increased attentional control showed poor

performance on the ANT task. These findings are consistent with recent data from our lab

suggesting that greater executive resources (indexed by lower perseverative errors on the

Wisconsin Card Sorting Task) facilitates emotion-related benefits (more approach-oriented

behaviors during social threat) in individuals who report greater threat sensitivity (Coifman,

Halachoff, Nylocks, 2018). Taken together, these findings indicate that executive cognitive

resources might play an important role in moderating the relationship between fear reactivity and

fear recovery such that increased executive resources might work to facilitate top-down control

of fear reactivity, especially in highly threat sensitive individuals.

Furthermore, there is clear evidence suggesting that deficits in fear inhibition are also a

function of increased fear reactivity. Indeed, evidence has emerged to suggest that individuals

with PTSD might have a higher “fear load,” as compared to individuals who do not have PTSD

(Norrholm et al., 2011; 2015). Specifically, the term “fear load,” refers to a greater magnitude of

fear responding (i.e. increased fear reactivity), and this concept has been derived from research

utilizing a fear-potentiated startle paradigm where the fear response is indexed by eyeblink

startle response. This line of research has demonstrated that individuals with PTSD, as compared

to individuals who have experienced a traumatic event but do not have PTSD, show increased

8
fear reactivity in response to both danger and safety cues during fear conditioning. Further, these

individuals (with PTSD) show increased fear reactivity during early phases of fear extinction

(Norrholm et al., 2011). This research suggests that individuals with PTSD might have a more

intense fear response to threat cues (increased fear reactivity) and that this pattern of responding

might lead to deficits in fear inhibition (i.e. continued elevation of fear after the threat has been

removed) (Norrholm et al., 2011). Related to above discussed research on executive cognitive

resources (Dennis & Chen, 2007a; 2007b; Coifman, Halachoff, & Nylocks, 2018), top-down

processing might be an important, largely unexplored, piece of this research. It is possible that

individuals with decreased executive cognitive resources show greater fear reactivity in these

paradigms, and that the lack of top-down control might lead to decreased fear inhibition

(Sarapas, Weinberg, Langenecker, & Shankman, 2017). Consistent with these findings,

individuals with more severe PTSD symptoms have been found to show greater startle responses

during early phases of fear extinction, as compared to individuals with less PTSD symptoms

(Norrholm et al., 2015). Similar findings have emerged in adolescents with a primary diagnosis

of specific phobia or SAD (Waters et al., 2014). These adolescents show greater startle response

during early “danger phases” and during “safe phases” of an explicit threat cue paradigm (Craske

et al., 2009), as compared to healthy controls. However, other evidence suggests that poor fear

inhibition might not be a function of greater fear reactivity (e.g. Craske et al., 2012). For

example, across a number of fear potentiated startle paradigms, it has been found that individuals

with an anxiety disorder, or those who are at risk for developing an anxiety disorder, do not show

elevated fear in response to threatening stimuli (Craske et al., 2009; 2012; Grillon, Dierker, &

Merikangas, 1998; Grillon et al., 2005; Reeb-Sutherland et al., 2009). Rather, these individuals

show elevated fear in response to safety cues that follow threatening stimuli, as discussed above.

9
Norrholm and colleagues (2015) also report that when controlling for level of fear acquisition

(fear load), decreased fear extinction remains a significant predictor of PTSD

symptoms. Such findings indicate that increased fear reactivity alone might not be predictive of

fear-pathology. Indeed, increased fear reactivity in response to a threat might simply be a

normative response (observed in both individuals with fear-pathology and in healthy controls).

In summary, evidence reviewed above regarding the link between fear reactivity and fear

inhibition appears mixed, and the association between these two components of the fear response

and fear-pathology remains unclear. There is indeed evidence, including neuroimaging findings,

to suggest that increased fear reactivity might lead to diminished fear inhibition, and that this

pattern might be associated with fear-pathology. However, contradictory research suggests that

fear reactivity is not associated with fear-pathology, and that poor fear inhibition might be a

unique predictor of fear pathology. Furthermore, evidence suggests that fear inhibition might be

driven by top-down executive resources and such resources are perhaps an important, but less

explored, factor in the association between fear reactivity and fear inhibition.

Methodological Challenges to Understanding the Association between Fear Reactivity and

Fear Inhibition

Mixed evidence on the associations between fear reactivity, fear inhibition, and fear-

pathology, might also stem from considerable methodological challenges. First, fear reactivity is

a complex, evolutionary based, response (Ohman & Mineka, 2001; Seligman, 1971), which can

be difficult to account for in experimental research. Indeed, Pavlovian fear conditioning

paradigms (one of the most commonly used methods for studying fear) have provided the field

with essential information regarding how a safe stimulus can come to be observed as a threat,

10
and how individuals might respond to such learning (LeDoux, 2012). However, in order to better

understand the various, complex, components of the fear response, more complex and dynamic

methodological approaches might be necessary. For example, assessment of fear reactivity might

benefit from utilizing fear-relevant stimuli that is more dynamic and closer to daily-life

experiences. However, many current fear conditioning paradigms include fear-irrelevant stimuli

as the CS (e.g. objects of different shape and color). Moreover, fear reactivity has most often

been examined in samples with current symptoms of fear-pathology. Although this body of

research has provided essential information about fear-pathology presentation, there is limited

knowledge about patterns of fear reactivity that might indicate risk for future fear-pathology

development.

Broadly, there is a need to further examine spontaneous fear responses. Much research on

fear inhibition has relied on experimental paradigms that include explicit instructions for

participants to utilize certain regulatory strategies (e.g. acceptance or reappraisal) (Campbell-

Sills, Barlow, Brown, & Hofmann, 2006; Eifert & Heffner, 2003; Low, Stanton, & Bower, 2008;

Gallo et al., 2009), or explicit safety cues (Craske et al., 2009), during and after fear elicitation.

There might be important information regarding spontaneous fear inhibition that is overlooked

by providing participants with such instructions. Indeed, participants might respond differently to

threatening stimuli (and the removal of threat) when provided with instructions, as compared to

how they would naturally respond to, and recover from, threat in real life where safe contexts are

more ambiguous. Additionally, current research on fear inhibition has largely examined pre-

determined group differences (Breuninger, et al., 2017; Davis, et al., 2014; Eifert & Heffner,

2003; Erisman & Roemer, 2010). Important variability in the sample might be overlooked when

the sample is divided into groups assumed to show common patterns of behavior (Jung &

11
Wickrama, 2008). Taken together, spontaneous fear inhibition, as well as individual differences

in fear inhibition, might not be captured in these types of paradigms.

Regulatory resources

A secondary challenge facing research on the etiology of fear-pathology is reaching a

better understanding of factors that might influence fear reactivity and fear inhibition. There is

evidence to suggest that certain regulatory resources (e.g. increased executive cognitive

resources and positive emotion expression) offer protection against development of fear-

pathology. However, it is less clear how these factors directly influence fear reactivity and fear

inhibition.

One component of executive cognitive functioning, specifically, the ability to suppress a

dominant response when necessary (Homack & Riccio, 2004; Miyake et al., 2000), has been

found to influence emotional responding (Hofmann, Schmeichel, & Baddeley, 2012). This

ability is commonly assessed by use of a Stroop task (Stroop, 1935), or an “emotional Stroop

task.” Generally, participants respond slower to words presented within the context of emotion,

as compared to a neutral context (e.g. Williams, Mathews, MacLeod, 1996). Stroop task deficits

are well-documented in samples with fear-pathology. For example, individuals diagnosed with

GAD have been found to perform more slowly and with less accuracy on Stroop tasks, as

compared to healthy controls (Chen et al., 2013; Hallion et al., 2017). Further, individuals with

affective disorders often show slower responding on Stroop tasks (or tasks measuring similar

abilities), specifically, within the context of negative emotion (Gotlib & Joormann, 2010;

Graham & Milad, 2011). For example, in an emotional Stroop task, veterans with PTSD were

found to respond slower to combat-related words (as compared to neutral or generally

threatening words), whereas veterans without PTSD were found to respond to all categories of

12
words at a similar speed (Khanna et al., 2017). Individuals with fear-pathology show greater fear

reactivity in response to emotional contexts. Such a response pattern might cause bottom-up

interference and thus, slow ability to respond accurately (Bar-Haim et al., 2007; Ohman, 2005;

Ohman & Mineka, 2001). Individuals with fear-pathology might also have decreased ability to

shift attention from emotional cues thus, impairing performance on neutral trials that follow

emotional trials (Bar-Haim et al., 2007; McKenna & Sharma, 2004). However, there is limited

research on how Stroop task responding (or responding to task that measure similar abilities)

might be related to fear responding, specifically (Schmeichel & Tang, 2015). It is indeed

possible that introducing any extraneous information (negative emotional cues, or neutral cues)

slows down performance on Stroop tasks, and that the valence of the extraneous information is

less important. As aforementioned, there is also evidence to suggest that a “balance” between

executive cognitive resources (top-down control) and increased threat sensitivity (bottom-up

interference or reactivity) might lead to optimal attentional performance on tasks that are similar

to the Stroop (i.e. Attention Network Test, which contains a flanker task) (Dennis & Chen,

2007a; 2007b).

Another factor that likely influences fear reactivity and fear inhibition, but that has not

yet been explored in the fear literature, is the expression of positive emotion within negative

emotion contexts. In general, the ability to generate positive emotions is linked to greater

psychological health (Fredrickson, 1998; Fredrickson et al., 2003; Tugade & Fredrickson, 2007).

More specifically, positive emotions might function to facilitate down-regulation of negative

emotions, as suggested by the “undoing” hypothesis of positive emotions (Fredrickson, 1998;

2001; Keltner & Bonanno, 1997), and positive emotion generation has been found to inhibit

distress (Fredrickson & Levenson, 1998; Fredrickson, Mancuso, Branigan, & Tugade, 2000;

13
Papa & Bonanno, 2008; Tugade & Fredrickson, 2004). For example, increased self-reported

positive emotion has been shown to accelerate resolve of negative emotion, following a threat

induction (speech preparation task) (Tugade & Fredrickson, 2004). It is possible that ability to

generate positive emotion could moderate the relationship between fear reactivity and fear

recovery. Indeed, according to the evidence reviewed above, positive emotion expression within

the context of threat could accelerate fear inhibition once the threat has been removed.

14
 II. The Current Study

The current study was designed to explore fear reactivity and fear inhibition in order to

further the understanding of fear-pathology development. To reach this aim, a fear reactivity task

was developed consisting of a series of film clips to elicit either fear or positive emotions. Each

film clip was approximately 90 seconds long and was followed by a 60-second recovery period.

This task was specifically designed to examine spontaneous fear reactivity (during the fear

eliciting videos) and spontaneous fear inhibition immediately following fear elicitation (during

the recovery periods). During the entire task, emotional responses were measured in real-time on

multiple dimensions including, coded emotional facial expressions, sympathetic arousal (skin

conductance), and self-reported emotional experience. Questionnaires were administered to

measure threat sensitivity and symptoms of depression and anxiety. Participants also completed

two computerized Stroop tasks.

The current study utilized latent growth mixture modeling (LGMM) to model fear

response (indexed by skin conductance) during relevant segments of the fear reactivity task (i.e.

during the fear eliciting videos and the recovery periods that follow). Rather than relying on a-

priori group differences, this approach allowed for detection of latent groups characterized by

their pattern of change in fear responding during the specified task segments (Jung & Wickrama,

2008; Muthen, 2004). In order to verify that the latent groups indeed reflect meaningful

differences in fear responding, fear facial expression and reported negative emotions across the

full fear reactivity task were examined for differences by latent group. Fear response dimensions

were examined separately, rather than collapsed together, due to limited evidence for consistent

15
emotion response coherence (Bulteel et al., 2014). Latent groups were then examined for

differences in factors that are indicative of psychological risk, specifically, psychological

symptoms and threat sensitivity.

Because LGMM is still a novel approach to these research questions, we also conducted

some conventional analyses utilizing change scores (described in detail below) (e.g. Badour &

Feldner, 2013; Codispoti, Surcinelli, Baldaro, 2008; Gomez, von Gunten, Danuser, 2016; Low,

Stanton, & Bower, 2008) capturing changes in emotional responses that occurred during the fear

reactivity task. Specifically, we examined associations between fear reactivity, fear recovery, and

factors reflective of psychiatric risk (i.e. psychological symptoms and threat sensitivity).

We then conducted two exploratory analyses to explore the influence of regulatory

resources on fear responding and psychiatric risk. First, we examined how the association

between fear reactivity and psychiatric risk might be influenced by executive cognitive

functioning by testing Stroop reaction time as a moderator in the relationship between fear

reactivity and reported threat sensitivity. Based on prior research (Dennis & Chen, 2007a; 2007b;

Coifman, Halachoff, Nylocks, 2018) suggesting that greater executive resources might mitigate

risks associated with high threat sensitivity, we conducted an exploratory analysis to test Stroop

task performance (within a negative emotion context) as a moderator between threat sensitivity

and fear reactivity. We also tested the same relationship, but using Stroop reaction time within a

neutral context as a moderator. This was to explore if there is a unique effect of negative

emotional context, or if the introduction of any extraneous stimuli might moderate the

relationship between reactivity and threat sensitivity. Lastly, to explore how the association

between fear reactivity and fear recovery might be influenced by positive emotions, we tested

16
positive emotion expression as a moderator in the relationship between fear reactivity and fear

recovery.

Hypotheses

Given that LGMM is a data driven approach, which allows for detection of groups that

best fit the data (Jung & Wickrama, 2008), we seek to carefully examine the latent groups that

emerge without specific hypotheses or predictions (Del Boca et al., 2004). Although we cannot

predict the specific fear response pattern of latent groups, we do anticipate groups to emerge in

the data that will be characterized by different patterns of fear responding. Indeed, based on prior

evidence reviewed above (e.g. Craske et al., 2009; 2012; Norrholm et al., 2015) we anticipate a

group to emerge that shows persistently elevated skin conductance (non-habituating) or that

shows a “reactive” pattern of change (increasing arousal within a segment) that is resistant to

change across the segments of the task (both during and after fear elicitation). Moreover, we

anticipate that another group could emerge from the data with consistently lower levels of skin

conductance (habituating) during the recovery periods of the task, as well as during the second

fear video of the task (when fear inhibition would be expected). We also anticipate increased

psychological symptoms and threat sensitivity to emerge in a group showing persistent arousal

(non-habituating) across the fear reactivity task. Lastly, based on prior research (e.g. Dennis &

Chen, 2007a; 2007b), we hypothesize that increased executive cognitive functioning might

moderate the relationship between higher fear reactivity and psychological risk. Lastly, we

hypothesize that increased positive emotion expression within a negative emotional context

might facilitate fear recovery (Fredrickson & Levenson, 1998). Aims of the current study are

specified below, and also detailed further in the data analytic approach section.

17
Aims

Aim 1. Model fear response (indexed by skin conductance) during the fear elicitation and

fear recovery segments of the fear reactivity task in order to detect latent subgroups within the

sample. Latent groups that are derived will be characterized by their pattern of fear responding

during these specific segments of the task.

Aim 1a. Verify that latent groups reflect meaningful differences in fear

responding. Specifically, examine latent groups for mean differences in both fear facial

expression and reported negative emotion experience across the fear reactivity task.

Aim 1b. Examine latent groups for differences in factors that might indicate

psychiatric risk, specifically, psychological symptoms and reported threat sensitivity.

Also examine latent groups for differences in regulatory resources, specifically, Stroop

reaction time and positive emotion expression.

Aim 2. Examine the association between fear reactivity and fear recovery (utilizing

change scores derived from fear responding during the fear reactivity task). Additionally,

examine associations between fear reactivity, fear recovery, and factors reflective of psychiatric

risk (i.e. psychological symptoms and reported threat sensitivity).

Exploratory aim 1. To explore how the association between fear reactivity and

psychiatric risk might be influenced by executive cognitive functioning, Stroop reaction time

will be tested as a moderator in the relationship between fear reactivity and reported threat

sensitivity.

Exploratory aim 2. To explore how the association between fear reactivity and fear

recovery might be influenced by positive emotions, positive emotion expression will be tested as

a moderator in the relationship between fear reactivity and fear recovery.

18
 Method

Procedure

Participants (n=101) were recruited over the course of two academic semesters from the

Psychology department subject pool at a large public university in the Midwest of the United

States. All participants provided informed consent upon arrival at the laboratory. Once informed

consent was obtained, participants were seated in a 4’x4’ private study room, in front of a

computer. Then, participants were asked to complete two computerized Stroop tasks. Following

the completion of the computer tasks, self-report questionnaires were administered to index

demographic information, threat sensitivity, and symptoms of depression and anxiety. Following

the completion of these questionnaires, participants were asked to engage in the fear reactivity

task. Once the laboratory session was complete, participants were compensated with research

credits. The university’s institutional review board approved the present study.

Participants

One hundred and one college-age participants (71.3% Caucasian, 76.2% female, mean

age = 20.07 years, SD = 3.43) were recruited (see Table 1 for sample characteristics). In order to

be eligible for the study, participants needed to be 18 years of age, or older, and be fluent in the

English language. No additional exclusion criteria were established.

Measures

Psychological symptoms.

Depression. The Center for Epidemiologic Studies Depression Scale (CES-D) was

utilized to index depressive symptoms. This scale contains 20 items and has been found to

demonstrate good validity and reliability in both general and clinical populations (Radloff,

1977). The mean score for this sample was 16.64 (SD=10.50), suggesting clinical levels of

19
depressive symptoms in the current sample (Lewinsohn, Seeley, Roberts, & Allen, 1997).

Indeed, the mean score in this sample is higher compared to general college-age samples (Latsko

et al., 2016; Matt, Fresco, Coifman, 2016) and 42.3% of the current sample scored above the

clinical cut-off. Notably, we did not intentionally recruit for high depression scores. Internal

consistency for this sample was good (α=.90).

Distress. A modified version of the Symptoms Checklist-90-Revised (SCL-90-R) was

utilized in the current study as an indicator of current distress. The SCL-90-R is widely used, and

the original measure has shown good internal consistency, and good test-retest reliability

(Derogatis, 1983). In the current study, the Depression, Hostility, and Anxiety scales were

administered, and an average score was calculated to represent general distress for each

participant (Almahmoud et al., 2016). The mean score for this sample was .91 (SD=.63),

indicating greater reported distress in the current sample as compared to other general college-

age samples (Todd, Deane, McKenna, 1997). Internal consistency for this sample was good

(α=.94).

Threat sensitivity. The Behavioral Inhibition (BIS) scale, a previously validated measure

of threat sensitivity (Carver & White, 1994), was used to index threat sensitivity. The BIS scale

consists of seven items, and scores were summed to create one BIS score for each participant in

the current sample, following established scoring procedures (Carver & White, 2013). The mean

BIS scale for the current sample was 21.22 (SD=4.18) which is comparable to other general

college-age samples (Coifman, Halachoff, & Nylocks, 2018). Internal consistency for this

sample was good (α=.82).

20
Tasks

Stroop tasks. Two tasks, a Color-word Stroop task and an Emotional Color-word Stroop

task (Stroop, 1935) were administered in the current study to measure the ability to suppress a

dominant or pre-potent response (Homack & Riccio, 2004). This task has previously been shown

to display good reliability (Connor, Franzen, & Sharp, 1988; Graf et al., 1995; Sacks, Clark,

Pols, & Geffen, 1991). Both Stroop tasks included in the current study were administered using

EPrime 2.0 (Psychology Software Tools, Sharpsburg, PA).

First, the Color-word Stroop task (Golden, 1978) was administered. The Color-word

Stroop task included in the current study began with a set of 16 practice trials in order to ensure

participant comprehension of the task. Following the practice, the task included three sets of 80

trials each, and each set presented 25% congruent and 75% incongruent trials.

Similar to the Color-word Stroop task, the Emotional Color-word Stroop task also

contained three sets of 80 trials each, and each set presented 25% congruent and 75%

incongruent trials. In this modified version of the Color-word Stroop task, each word was

presented superimposed over an emotionally evocative picture. All pictures included in the task

were selected from the IAPS picture set, and remained on the screen for 500ms. Each of the three

set contained 20 images that would appear four times each, and five images that would appear

five times each. Specifically, one set included 20 negatively valenced images randomized to

appear throughout the set, one set included 20 positively valenced images, and one set included

20 neutral images1. This design allows for examination of reaction time within the context of

1
The 60 pictures included in the task, listed by their IAPS identification numbers, were as follows: negative - 1111,
1205, 2095, 2352.2, 2703, 2811, 3015, 3100, 3180, 3181, 3225, 3266, 3350, 3530, 6022, 9040, 9181, 9187, 9410;
positive - 1410, 1411, 1440, 1460, 1620, 1710, 1750, 1811, 2030, 2070, 2071, 2345, 2347, 2530, 2550, 5833, 5910,
7330, 8470, 8501; neutral - 2383, 2393, 5471, 5534, 5535, 7000, 7001, 7006, 7009, 7020, 7025, 7030, 7031, 7061,
7090, 7140, 7211, 7504, 7512, 7700.

21
emotion, an ability that is often impaired in individuals with affective disorders (Gotlib &

Joorman, 2010; Graham & Milad, 2011). Lastly, between the three sets of trials, participants

were asked to watch three images of neutral landscapes (presented on the screen for 10 seconds

each)2.

Stroop task data cleaning and preparation. Data from the two Stroop tasks were cleaned

in E-DataAid (Psychology Software Tools, Sharpsburg, PA) by following established procedures

(Kane & Engle, 2003). For both Stroop tasks, reaction times from incorrect trials were removed.

From the correct trials, data was removed only when the reaction time for a trial exceeded the

participant’s own mean response time for that condition by three standard deviations or more. If

the reaction time for a trial was below 200ms, it was also removed. This resulted in dropping less

than 1% of trials for the current sample. Interference scores were then derived for each

participant. Specifically, mean reaction times were calculated for incongruent and congruent

trials. Then, the congruent mean was subtracted from the incongruent mean in order to generate

interference scores for each participant (Strauss, Allen, Jorgensen, & Cramer, 2005). Interference

scores were calculated for the Color-word Stroop task, as well as from each of the emotion

conditions in the Emotional Stroop task (neutral, negative, and positive conditions thus, a total of

four Stroop task interference scores were generated for each participant (Table 3a). Larger

interference scores in this task might indicate broadly decreased executive functioning (Adleman

et al., 2002; Archibald & Kerns, 1999).

The four Stroop interference scores were examined by Pearson correlations (Table 3b).

Interference scores derived from the Color-word Stroop task were positively correlated with

interference scores derived from the neutral Stroop condition r(79)=.28, p=.014, and the negative

2
The nine neutral pictures shown during the breaks, listed by their IAPS identification numbers, were as follows:
5215, 5250, 5539, 5551, 5628, 5711, 5720, 5740, 7545.

22
emotion Stroop task condition r(79)=.33, p=003. Interference scores derived from the neutral

emotion Stroop condition were significantly correlated with interference scores derived from the

positive Stroop condition r(93)=.31, p=.003, and the negative Stroop condition r(94)=.21,

p=.040. In order to examine differences in Stroop reaction times depending on context (negative

emotional context has been shown to weaken performance on the Stroop task: Williams,

Mathews, MacLeod, 1996), a Stroop task difference score was created for the current sample by

subtracting the mean interference score derived from the Color-word Stroop task from the mean

interference score derived from the negative condition Emotional Color-word Stroop task. The

mean for the Stroop task difference score in the current sample was 19.37ms (SD=138.69). A

smaller difference score indicates a smaller difference between performance on the Color-word

Stroop and the negative Stroop task condition, and a larger score indicates a larger difference

between performance on the Color-word Stroop task and the negative Stroop task condition (i.e.

slower performance on the task when a negative emotion context is introduced). A Stroop task

difference score was also created for the current sample by subtracting the mean interference

score derived from the Color-word Stroop task from the mean interference score from the neutral

condition Emotional Color-word Stroop task. This mean in the current sample was 8.26ms

(SD=79.27). The two Stroop task difference scores will be used in the final analyses to index

Stroop responding within a negative emotional context, as well as within a neutral context (but

with extraneous content), relative to Stroop responding outside of any emotional context, or

extraneous content.

Fear reactivity task to index fear reactivity and fear inhibition.

Fear reactivity task overview. To complete the fear reactivity task, participants were

seated in front of a computer, and were then asked to engage with the content of four emotionally

23
evocative film clips. The task consisted of, specifically, two fear eliciting videos and two

positive emotion videos, and each video was followed by a recovery period (without stimulus

presentation). Each film clip was approximately 90 seconds long, and each recovery period was

60-second long. During the entire fear reactivity task, participants’ emotional responses were

recorded on multiple dimensions (reported emotional experience, emotion facial expression, and

skin conductance). The order of valence of the film clips was fixed (i.e. fear then positive), but

random in terms of specific film content. As such, the fear reactivity task was designed to

explicitly examine spontaneous fear reactivity (increased fear expression, reported negative

emotion, and autonomic arousal, during the fear eliciting videos) and then spontaneous fear

inhibition following the removal of the fear eliciting video (during the recovery periods) (see

Figure 2 for overview of the task). Video stimuli utilized in the fear reactivity task were assessed

in a pilot study, detailed in Appendix A, prior to any data collection for the current study. Results

from the pilot study showed that the four selected fear videos elicited sufficient levels of fear

thus, all evaluated videos were included in the fear reactivity task.

24
 EDA and facial behavior EDA and facial behavior
data collection begins data collection ends

120 sec 60 sec 90 sec 60 sec 90 sec 60 sec 90 sec 60 sec 90 sec 60 sec

Neutral Recovery Recovery Positive Recovery Recovery Positive Recovery

Baseline period Fear film Fear film period
period film period period film

20 sec 20 sec 20 sec 20 sec 20 sec 20 sec 20 sec 20 sec 20 sec 20 sec

= self-report of
emotional experience
provided. Approximately 20
seconds for each rating.

Figure 2. Overview of the fear reactivity task.

Fear reactivity task details. Participants were given task instructions before the fear

reactivity task started. Instructions utilized in the current study were initially adopted from Gross

(1998) and have been used in previous emotion elicitation research using video clips (Gilman et

al., 2017; Nylocks et al., 2017) (see below for all instructions given in the task). To begin the

task, and in order for participants to acclimate to study conditions, the video sequence started

with a clip of neutral content (“Alaska’s Denali”, Alaska Travel Guide), followed by a recovery

period (without stimulus presentation). Participants were then shown a fear eliciting video,

followed by a recovery period (without stimuli presentation), followed by a high arousal positive

emotion video, followed by a recovery period (without stimuli presentation), followed by a

second fear eliciting video, followed by a recovery period (without stimuli presentation),

followed by a second high arousal positive emotion video, followed by a recovery period

(without stimuli presentation). The four video clips and recovery periods were part of a larger

25
investigation which was followed by four additional video clips for a separate investigation. The

task was administered using EPrime 2.0 (Psychology Software Tools, Sharpsburg, PA).

Introduction to task (provided on-screen as well as read aloud by administrator):

“We will now be showing you a number of short video clips. It is important to us

that you watch the film clips carefully, until the end, and that you engage with the

content in each film. However, if you find any of the films too distressing, just say

‘stop’ and I will stop the film. As you watch the films, you may have some

emotional responses. After each film clip, you will complete two brief

questionnaires on how you are currently feeling. When you are ready to start,

press the spacebar.”

Instructions provided on-screen before each video: “Remember: Please watch the

film carefully and engage with it as best as you can.”

Instructions provided on-screen during each recovery period: “Please rest”

Emotion response assessment during the fear reactivity task.

Reported emotional experience. To measure emotional experience, participants were

instructed to report their emotional experience immediately following each film clip and

immediately following each recovery period by rating emotion-words on a 7-point Likert scale.

The ratings took approximately 20 seconds each time that they were completed. Emotion-words

were of both negative (fear, sadness, disgust, guilt, distress, anger) and positive (happiness,

enjoyment, amusement, affection, relief) valence. Self-reported ratings of negatively valenced

words were utilized to generate a mean negative affect score (for each participant) in response to

fear contexts, positive emotion contexts (positive emotion films), and the recovery periods.

Similarly, self-reported ratings of positively valenced words were used to generate a mean

26
positive affect score (for each participant) in response to fear contexts, positive emotion contexts

(positive emotion films), and the recovery periods. By examining broadly negative affect scores,

the current study will be sensitive to variability in affect regulation. Indeed, there are known

individual differences in emotion labelling that might make some participants able to report

experienced fear, specifically, whereas other participants might report experienced fear as, for

example, “distress” (Tomko et al, 2015; Zaki et al., 2013; Thompson et al., 2015; Demiralp et al.,

2012; Kashdan & Farmer, 2014). Emotion-words were chosen from contemporary circumplex

models of affect (e.g. Rafaeli, Rogers & Revelle, 2007; Russell, 1980) and have been used

reliably in prior studies (e.g. Coifman & Bonanno, 2010). Each affect scale demonstrated

sufficient internal consistency in this sample (negative emotion experience in response to fear

context a=.86, and positive emotion experience in response to positive emotion context; a=.91).

Emotion facial expression. To measure behavioral expression of emotion during the fear

reactivity task, participant’s emotional facial behavior was recorded with a high-resolution

camera. Emotional facial behaviors were later coded by four independent coders, naïve to any

study details3. Coders viewed participant videos individually, on a 23” computer monitor without

sound. After viewing each video, coders rated, specifically, the degree of fear facial behaviors to

each video on a 7-point Likert scale, resulting in a fear expression score for each participant.

Additionally, coders rated the degree of positive emotional facial behaviors to each video on a 7-

point Likert scale (Coifman & Bonanno, 2010), resulting in a positive emotion expression score

for each participant. Coders were reliable (fear expression: average ICC = .93, range .92-.93;

3
Evidence suggests that facial coding by relatively naïve coders may be as valid as highly trained coders (e.g.,
Dondi et al., 2007).

27
positive emotion expression: average ICC = .90, range .81-.96), and ratings were averaged across

coders to increase reliability.

Skin conductance. Electrodermal activity (EDA) was recorded in real time, in order to

calculate a value to represent Skin Conductance (SC) for each participant. This procedure was

conducted as follows. The study room was kept at a steady temperature of 74.3 degrees

Fahrenheit and participants were encouraged to drink one 8-ounce bottle of water before the

assessment began. These procedures were established in order to minimize interference with

EDA recording. To record the EDA signal, an electrolyte gel of sodium chloride was applied to

two Beckman electrodes that were attached to the palmar surface of the participant’s middle

phalanges of the first and second fingers of the non-dominant hand. The EDA signal was

acquired through a 31-channel A/D converter operating at a resolution of 12 bits and with an

input range of -2.5V to +2.5V. Amplification rates, high-pass filter (HPF), and low-pass filter

(LPF) settings were as follows: EDA (gain=0.1V/S, HPF=none/DC, LPF=10Hz, 6dB/octave,

single pole RC) (Mindware Technologies, Gahanna, OH).

EDA data were cleaned and artifacts removed using customized Bio-Lab™ Software

(MindWare Technologies, Gahanna, OH) and visual inspection of artifacts. Problematic

segments of the EDA wave were excluded following visual inspection. Mean skin conductance

response (SCR) values were calculated for each 10-second increment, for each film clip and for

each recovery period, for each participant (in uSiemens). Specifically, since each film was

approximately 90 seconds in length, these segments were divided into nine 10-second long

increments, and then a mean skin conductance value was generated for each of the nine 10-

second increments, for each participant. Similarly, each recovery period was 60 seconds in

length thus, these 60-second segments were divided into six 10-second long increments, and then

28
a mean skin conductance value was generated for each of the six 10-second increments, for each

participant. This method allows for closer inspection of fluctuations in skin conductance, as well

as for plotting of trajectories, and might be preferable to averaging across the full 60-second/90-

second segments, which might artificially alter natural fluctuations in skin conductance. The

specific length of the increments (10 seconds) was determined based on skin conductance

response latency (Dawson, Schell, Filion, 2007) and prior research on skin conductance response

to emotion elicitation (e.g. Aguado, 2016). Increase in skin conductance has previously been

suggested to indicate “alertness” or arousal, and within the context of fear (in response to a

threat) increased skin conductance might be an indicator of fear reactivity (Kreibig et al., 2007;

Kreibig et al., 2010; Jacobs et al., 1994). Mean skin conductance across the full fear reactivity

task for the current sample was M=4.78, SD=2.85, (range: .30-14.24).

Emotion response data preparation

LGMM analyses. Skin conductance response data used in LGMM analyses consisted of

raw skin conductance levels (in 10-second increments, as explained above).

Change scores. Change scores were calculated to index fear reactivity and fear recovery.

Specifically, change scores were calculated for each participant by using mean reported negative

emotion experience, mean skin conductance response, and mean fear expression, as detailed

below. Change scores were calculated separately for the two fear video segments due to

significant differences in mean fear expression and mean skin conductance response that

emerged (examined by use of paired samples t-tests) in the current sample between the first fear

video and the second fear video. Specifically, the current sample showed significantly greater

fear expression during the second fear video (M=2.27, SD=1.30) as compared to during the first

fear video (M=2.03, SD=1.15), t(86)=-1.96, p=.054. The current sample also showed

29
significantly greater skin conductance during the second fear video (M=5.03, SD=2.99) as

compared to the first fear video (M=4.32, SD=2.62), t(81)=-5.44, p<.0014.

Reactivity. To index fear reactivity during the first fear video segment of the fear

reactivity task, mean emotion response (reported negative emotion experience, skin conductance,

and fear expression) during the baseline video was subtracted from mean emotion response

during the first fear video. Larger scores indicate more reactivity, for example, if the fear

expression score for a participant is 1 (low fear expression) in response to the baseline video, and

the fear expression score of that participant is 7 (high fear expression) in response to the fear

video, the reactivity score of that participant is 6 (7-1=6). If a participant has a fear expression

score of 1 (low) at baseline, and a fear expression score of 3 (somewhat low) in response to the

fear video, the change score for that participant is 2 (3-1=2), indicating less reactivity.

Change scores were calculated in a similar way for the second set of videos. Specifically,

mean emotion response (reported negative emotion experience, skin conductance, and fear

expression) during the positive recovery period (presented prior to the second fear video) was

subtracted from mean emotion response during the second fear video (larger scores indicating

greater reactivity). This resulted in a total of six reactivity change scores for each participant

(two reactivity scores for each emotion response output).

Recovery. To index recovery during the first fear video segment, mean emotion response

(reported negative emotion experience, skin conductance, and fear expression) during the first

fear recovery period was subtracted from mean emotion response during the first fear video.

Larger scores indicate greater recovery (i.e. smaller scores indicate poor recovery).

4
This effect remains after accounting for baseline skin conductance levels (by subtracting mean baseline
skin conductance response from each fear video mean score).

30
 Similarly, change scores to indicate recovery were calculated for the second fear video

set. Specifically, mean emotion response (reported negative emotion experience, skin

conductance, and fear expression) during the second fear recovery period was subtracted from

mean emotion response during the second fear video. Larger scores again indicate greater

recovery. This resulted in a total of six recovery change scores for each participant (two recovery

scores for each emotion response output).

Positive emotion expression data. Mean positive emotion facial expression during the

two fear eliciting videos was utilized to examine positive emotion expression within the fear

context. This variable is included in the current study due to previous research showing that

positive emotion generation might facilitate the resolution of negative emotion (e.g. Fredrickson

& Levenson, 1998). Positive emotion expression in response to the first fear video (M=2.09,

SD=1.27) and positive emotion expression in response to the second fear video (M=1.92,

SD=1.09) will be examined separately in the final analyses.

Manipulation check

In order to test the validity of the fear reactivity task, a repeated measures analysis of

variance (ANOVA) was performed to examine negative emotion experience in response to the

fear reactivity task. This test followed a (film) x (time) design. A Greenhouse-Geisser correction

was used due to sphericity assumptions being violated (Gruber, Dutra, Eidelman, Johnson, &

Harvey, 2011). A significant within-person effect emerged F(8,3.55)=47.31, p<.001, partial eta2

=.35, such that the current sample reported greater negative emotion during the fear eliciting

videos, as compared to the positive videos and the recovery periods. Mean levels of self-reported

negative emotion across the task by the current sample were as follows: baseline video (M=1.07,

SD=.26), first fear video (M=2.01, SD=.88), first recovery period (M=1.50, SD=.77), first

31
positive video (M=1.17, SD=.45), second recovery period (M=1.14, SD=.36), second fear video

(M=1.89, SD=.86), third recovery period (M=1.47, SD=.63), second positive video (M=1.16,

SD=.35), fourth recovery period (M=1.15, SD=.35) (see Figure 3). To examine any potential

differences in degree of reported fear that was elicited between the first and the second fear

video, a paired samples t-test was conducted to examine self-reported emotional experience in

response to these specific task stimuli. This test showed no significant difference in levels of

self-reported negative emotion between the first fear video and the second fear video of the task

t(91)=1.12, p=.265, 95%CI[-.08-.32], dz=.12.

Similarly, in order to test the validity of the fear reactivity task, a repeated measures

ANOVA was performed to examine fear facial behavior in response to the fear reactivity task.

This test also followed a (film) x (time) design. A Greenhouse-Geisser correction was used due

to sphericity assumptions being violated (Gruber, Dutra, Eidelman, Johnson, & Harvey, 2011). A

significant within-person effect emerged F(1.97, 86)=51.07, p<.001, partial eta2 =.38, such that

the current sample showed increased fear facial behavior during the fear eliciting videos, as

compared to the positive videos and the recovery periods. Mean levels of fear facial behavior

across the task by the current sample were as follows: baseline video (M=1.07, SD=.15), first

fear video (M=2.01, SD=1.15), first recovery period (M=1.20, SD=.26), first positive video

(M=1.11, SD=.26), second recovery period (M=1.14, SD=.20), second fear video (M=2.23,

SD=1.30), third recovery period (M=1.20, SD=.35), second positive video (M=1.13, SD=.24),

fourth recovery period (M=1.10, SD=.20) (see Figure 4). To examine any potential differences in

degree of fear that was elicited between the first and the second fear video, a paired samples t-

test was conducted to examine fear expressions in response to these specific task stimuli. This

test showed a significant difference in levels of fear expression between the first and the second

32
fear eliciting video t(87)=-2.20, p=.05, 95%CI[-.48- -.01], dz= .24, such that the current sample

showed increased fear expression in response to the second fear video as compared to the first

fear video.

Lastly, mean skin conductance levels of the current sample were observed as expected. A

repeated measures ANOVA was performed to examine skin conductance in response to the fear

reactivity task. This test also followed a (film) x (time) design. A Greenhouse-Geisser correction

was used due to sphericity assumptions being violated (Gruber, Dutra, Eidelman, Johnson, &

Harvey, 2011). A significant within-person effect emerged F(2.92, 75)=49.88, p<.001, partial

eta2 =.40, such that the current sample showed increased arousal during the fear videos, followed

by decreased arousal following the fear segments. Mean levels of skin conductance response

across the task by the current sample were as follows: baseline video (M=3.25, SD=2.61), first

fear video (M=4.61, SD=2.70), first recovery period (M=5.05, SD=3.15), first positive video

(M=4.68, SD=2.89), second recovery period (M=5.06, SD=3.02), second fear video (M=5.31,

SD=3.23), third recovery period (M=5.31, SD=2.97), second positive video (M=5.10, SD=2.90),

fourth recovery period (M=5.63, SD=3.15) (see Figure 3). To examine any potential differences

in degree of fear that was elicited between the first and the second fear video, a paired samples t-

test was conducted to examine skin conductance response to these specific task stimuli. This test

showed a significant difference in mean skin conductance response between the first and the

second fear eliciting video t(80)=-5.19, p<.001, 95%CI[-1.03- -.46], dz=.58, such that the current

sample showed increased skin conductance in response to the second fear video as compared to

the first fear video.

The above reported results indicate that the fear reactivity task indeed worked as

expected by eliciting fear (observed on multiple dimensions) during the fear videos, and

33
decreased fear and arousal following the fear segments. Pearson correlations were conducted to

examine associations between the emotion response output variables included in the fear

reactivity task (see Appendix B for all statistics). Generally, fear expression and reported

negative emotion experience were significantly correlated in a positive direction during the fear

videos and fear recovery periods (range r = .32-.56, p<.01). Skin conductance was not

significantly correlated with these other emotion responses during the fear videos or fear

recovery periods. A similar pattern of response coherence emerged during the positive emotion

videos (range r = .40-.47, p<.01). These data indicate that there is some level of emotion

response coherence in the current sample, and align with research suggesting that emotion

responses are “loosely coupled” (e.g. Bulteel et al., 2014).

Missing data

Some data from n=31 participants were dropped from the current study. Four participants

were excluded from all data analyses, resulting in a sample size of n=97, due to the following

reasons. One participant did not speak English fluently and a translator device needed to be

utilized during the entire session (with difficulty), two participants did not feel well and needed

to end the session early, and one participant had previously participated in a different study

utilizing the same video stimuli as the present study, and was therefore excluded from all

analyses.

From the remaining (n=27) participants, data were dropped from some, but not all, final

analyses (see Table 4 for details regarding missing data). Thus, the degrees of freedom vary

slightly across the final analyses. Given the various sources of missing data (technological and

human error), this pairwise deletion approach to missing data was utilized in order to maximize

34
the sample while keeping variability in the sample size between analyses minimal (Roth, 1994;

Roth, Switzer, & Switzer, 1999).

In order to preserve data and maximize the sample, the following actions were taken. One

participant showed skin conductance levels that were more than three standard deviations above

the mean of the current sample. This participant’s mean skin conductance on the baseline video,

for example, was 17.23uS and the sample mean for this task segment was 3.36uS (SD=2.38).

Rather than dropping all data for this participant, their skin conductance data across the fear

reactivity task was truncated (replaced with the data points of the next highest participant’s

values), following standard procedures for truncated outlier filtering (Costa, 2014). Moreover, in

order to preserve skin conductance data for the current sample, skin conductance data was

imputed by following standard imputation procedures (replacing the missing value with the mean

from the data points adjacent to the missing skin conductance value) (Kreindler & Lumsden,

2006). In the current sample, skin conductance data (one data point equals 10 seconds of

recorded skin conductance responding) was imputed for some participants, and data was imputed

for a maximum of two data points within each video segment of the fear reactivity task, resulting

in less than 1% of imputed data points, out of the total number of data points in the full dataset

(including all task segments).

Participants with missing data were not found to be different from the main sample on

any key study variables, including age, sex, psychological symptoms, Stroop interference scores,

fear expression, positive emotion expression, or skin conductance.

Data analytic approach

Aim 1: Model fear response (indexed by skin conductance) during the fear elicitation and fear

recovery segments of the fear reactivity task in order to detect latent subgroups within the

35
sample. Latent groups that are derived will be characterized by their pattern of fear

responding during these specific segments of the task.

The current investigation will utilize LGMM in Mplus (7.11) (Muthen & Muthen, 2006)

to identify latent classes within the sample characterized by their pattern of fear responding

during the fear reactivity task. Based on evidence regarding fear reactivity and fear inhibition

reviewed above (e.g. Buss et al., 2011; Craske et al., 2009; 2012; Lissek et al., 2005; Norrholm

et al., 2011), fear responding in the current study will be modeled during the two fear videos and

the two fear recovery periods of the fear reactivity task (30 total segments of skin conductance

data). LGMM is well-suited for modeling behavioral patterns over time. The main goal of

LGMM is to use a data driven approach to identify, within a population, latent sub-populations

(i.e. latent classes/groups) with similar response patterns. Unlike conventional longitudinal

modeling approaches, LGMM does not assume that one single linear trajectory (central

tendencies) will adequately represent the entire sample (Berlin, Parra, & Williams, 2013; Jung &

Wickrama, 2008). Rather, LGMM is a data driven approach based on research showing that

there is meaningful heterogeneity within populations (e.g. Jackson & Sher, 2005). By identifying

latent classes that best fit the data, LGMM accounts for complexity present within a sample

(Berlin, Parra, & Williams, 2013; Jung & Wickrama, 2008).

Within the LGMM framework, a piecewise growth model will be utilized in the current

study as this particular approach allows for identification of unique slopes (patterns of change)

for different segments of the data (Flora, 2008). The piecewise model approach was selected, as

opposed to a conventional growth model approach, due to the structure of the fear reactivity task

which contains segments of data (conditions of the task) with potentially different parameters

(for example, linear or quadratic) (Galatzer-Levy et al., 2013; Li, Duncan, Duncan, Hops, 2001).

36
Thus, four separate LGMM analyses will be conducted, one for each task segment (first fear

video, first fear recovery period, second fear video, and second fear recovery period). Mean skin

conductance response during the baseline video will be entered into these analyses as a covariate,

due to known baseline differences in psychophysiology that are independent of task demands

(Dawson, Schell, & Filion, 2007). Model fit statistics are reported with the baseline covariate

included. The slope and intercept were specified as free to vary within each latent class (thus, the

model is specified as LGMM).

For each task segment, growth models will be tested with increasing number of classes.

We will first identify a single-class model solution to show the mean trajectory of the sample as

one distribution. We will then compare this single-class model to 2-class and 3-class growth

models. In order to determine model fit, the following indices will be examined: Akaike

information criteria (AIC; Akaike, 1987), Bayesian information criteria (BIC; Schwarz, 1978),

sample-size adjusted BIC, Lo-Mendell-Rubin likelihood ratio test (LMR-LRT) (Jung &

Wickrama, 2008), as well as entropy, parsimony, theoretical justification, and interpretability

(Nylund, 2007). In LGMM, these criteria indicate model fit related to the shape of the trajectory

(linear and linear/quadratic) as well as the number of latent classes that have been identified. The

best fitting model will have lower information criterion indices (AIC and BIC) indicating better

fit of the model, and a significant (p<.05) LMR-LRT value indicating a significant difference in

the model fit between different class solutions. Entropy indicates the probability of an individual

to fit into the specified class. Generally, entropy above .80 suggests that the latent classes are

likely separate enough and individual class membership can be stated with confidence.

Additionally, models with convergence issues, or models that show trajectories that seem to

explain small variations within the sample (for example, identification of classes with a low

37
number of participants such that no meaningful differences can be examined) will be ruled out

from the current study (e.g. Galatzer-Levy et al., 2013; Orcutt, Bonanno, Hannan, & Miron,

2014).

Aim 1a: Verify that latent groups reflect meaningful differences in fear responding.

Specifically, examine latent groups for mean differences in both fear facial expression and

reported negative emotion experience across the fear reactivity task.

In order to detect differences between latent groups in fear responding on multiple

emotion response dimensions across the task, a repeated measures ANOVA test will be

conducted comparing latent groups on mean levels of fear facial expression and reported

negative emotion experience. The repeated measures ANOVA will follow a 6(task segment: fear

video, first recovery, positive video, second recovery, fear video, third recovery) x 2(emotion

response output: fear expression, reported negative emotion) x 2 or 3(group) design. This

analysis will serve to verify that the latent groups derived from the LGMM analyses indeed

reflect differences in fear responding across the full task.

Aim 1b: Examine latent groups for differences in factors that might indicate

psychiatric risk, specifically, psychological symptoms and reported threat sensitivity. Also

examine latent groups for differences in regulatory resources, specifically, Stroop reaction

time and positive emotion expression.

Once latent classes have been derived from LGMM analyses, ANOVA tests will be

conducted to examine potential differences in psychological symptoms, reported threat

sensitivity (BIS), Stroop reaction time, and positive emotion expression by class membership

(Hadzi-Pavlovic, 2009).

38
Aim 2: Examine the association between fear reactivity and fear recovery (utilizing change

scores derived from fear responding during the fear reactivity task). Additionally, examine

associations between fear reactivity, fear recovery, and factors reflective of psychiatric risk

(i.e. psychological symptoms and reported threat sensitivity).

To index fear reactivity and recovery, change scores were calculated for skin

conductance, fear expression, and reported negative emotion experience (detailed above).

Pearson correlations will be utilized, in PASW Statistics 24 (SPSS Inc., Chicago, IL, USA), to

examine associations between reactivity and recovery. Pearson correlations will also be used to

examine associations between reactivity, recovery, psychological symptoms, and reported threat

sensitivity.

Exploratory aim 1: To explore how the association between fear reactivity and psychiatric

risk might be influenced by executive cognitive functioning, Stroop reaction time will be tested

as a moderator in the relationship between fear reactivity and reported threat sensitivity.

Based on prior research (Dennis & Chen, 2007a; 2007b; Coifman, Halachoff, Nylocks,

2018) suggesting that greater executive resources might mitigate risks associated with high threat

sensitivity, we will conduct an exploratory analysis to test Stroop task performance (within a

negative emotion context) as a moderator between threat sensitivity and fear reactivity (observed

during the second fear video). We will also test the same relationship, but using Stroop reaction

time within a neutral context as a moderator. This is to explore if there is a unique effect of

negative emotional context, or if the introduction of any extraneous stimuli might moderate the

relationship between reactivity and threat sensitivity. To this end, Hayes’ PROCESS Model 1

will be used. Fear reactivity will be entered into the model as the dependent variable, and

reported threat sensitivity will be entered into the model as the independent variable.

39
Exploratory aim 2: To explore how the association between fear reactivity and fear recovery

might be influenced by positive emotions, positive emotion expression will be tested as a

moderator in the relationship between fear reactivity and fear recovery.

To examine if positive emotion moderates the relationship between fear reactivity

(observed during the first fear video) and fear recovery (observed following the first fear video),

Hayes’ PROCESS Model 1 will be used. Fear recovery will be entered into the model as the

dependent variable, and fear reactivity will be entered into the model as the independent variable.

Positive emotion expression within a negative emotional context (first fear video) will be entered

into the model as the moderator.

Statistical Power

Statistical power within an LGMM framework is complex, and frequently utilized power

estimation calculators (e.g. G*Power: Faul, Erdfelder, Buchner, & Lang, 2009) do not allow for

accurate estimation of power for LGMM models. There are several approaches to statistical

power and estimation of sample size within LGMM models, and below is a discussion about

approaches that are relevant to the current study. In general, there is little information in the

current literature about sample size needed to accurately detect the number of latent classes

(Curran, Obeidat, & Losardo, 2010; Dziak, Lanza, & Tan, 2014). Samples of approximately

n=100, or more, are generally recommended (Muthen & Curran, 1997). In smaller samples, there

is potential for generating too few classes to describe the model, and LGMM is therefore more

commonly utilized to detect trajectories within large samples over time (e.g. Bonanno et al.,

2012; Greenbaum et al., 2005). However, LGMM models with “good fit” have been identified in

smaller samples (n=22) as well (Huttenlocher, Haight, Bryk, Seltzer, & Lyons, 1991), and there

are a number of studies that have successfully utilized LGMM in samples below n=100 (Bertocci

40
et al., 2014; Galatzer-Levy et al., 2013; Okado & Haskett, 2014). Statistical power in growth

models also depends on the number of observations per individual, with a general

recommendation of at least three repeated observations per participant (Curran, Obeidat, &

Losardo, 2010). In the current study, fear facial expression and reported negative emotion

experience consist of fewer data points, as compared to skin conductance data. However, all

response dimensions in the current study exceed the recommended minimum number of data

points for longitudinal data analysis.

Results

Aim 1: Model fear response (indexed by skin conductance) during the fear elicitation and fear

recovery segments of the fear reactivity task in order to detect latent subgroups within the

sample. Latent groups that are derived will be characterized by their pattern of fear

responding during these specific segments of the task.

To identify latent trajectories of skin conductance responses across the specified fear

reactivity task segments, LGMM analyses were conducted. Class membership that is derived by

this method indicates distinct subpopulations within the current sample, characterized by their

pattern of change across time. As stated above, a piecewise model was utilized in order to derive

latent trajectories for each relevant segment of the fear reactivity task (the two fear videos and

the two recovery periods following the fear videos). Model fit statistics were compared across all

models (see Table 5 for model fit statistics), and mean baseline skin conductance was included as

a covariate in all models (Figure 4). Derived trajectories for the four relevant segments of the

fear reactivity task are stated below.

First fear video segment. A 1-class solution was identified as the best fitting model for

the first fear segment of the fear reactivity task, with linear parameters, indicating that the full

41
sample mean might best represent the pattern of change during the first fear video of the fear

reactivity task. A 2-class solution was ruled-out for this segment of data despite high entropy (1),

reduced information criterion values, and a significant LMRT score. The 2-class solution was

ruled out due to the low number of participants identified as belonging to the second class

(1.2%), indicating over-fitting of the model. Similarly, the 3-class solution was ruled out due to

low numbers of participants (1.2% and 1.2%) identified as belonging to two of the classes. The

single-class model in the current sample showed a non-significant positive linear slope during

the first fear video of the fear reactivity task (ESTslope = .031, SE = .01, p = .246) and an intercept

that was significantly different from zero (ESTintercept = 4.272, SE = .08, p < .001) (Figure 5).

First fear recovery segment. The first fear recovery segment (presented approximately

20 seconds following the first fear video of the task during which time participants were asked to

rate their current emotions) indicated a 2-class solution as the best fitting model, with linear

parameters. This solution was identified by observation of high likelihood that classes were

specified correctly (entropy = .99) and reduced information criterion values, as compared to a 1-

class solution. The 2-class model also showed a significant LMRT p-value (.037), as compared

to a 3-class solution. A 3-class solution was ruled out as a good fitting model despite high

entropy (.92) and reduced information criteria, as compared to the 2-class model. The 3-class

solution was ruled-out due to the low number of participants in the third class (8.4% of the

sample), and a non-significant LMRT p-value (.401), indicating that there is not a significant

difference in the model fit between the 2-class and 3-class solutions. In the 2-class solution, the

largest class Habituating (88% of the sample) showed a significant negative slope during the first

fear recovery segment (ESTslope = -.04, SE = .12, p < .001), the Habituating group also showed an

intercept that was significantly different from zero (ESTintercept = 4.64, SE = .02, p < .001). The

42
second class Non-habituating (12% of the sample) also showed a significant negative slope

during the first fear recovery segment (ESTslope = -.37, SE = .12, p < .001) and an intercept that

was significantly different from zero (ESTintercept = 9.33, SE = .01, p < .001), indicating a higher

level of skin conductance in this smaller group within the current sample (Figure 6).

Second fear video segment. A 3-class solution was identified as the best fitting model

for the second fear video segment, with linear parameters. Although the 2-class solution showed

high likelihood that classes were specified correctly (entropy = .96) and reduced information

criterion values, as compared to a 1-class solution, this solution was ruled out due to the low

number of participants identified as belonging to the second class (7.9%) and a non-significant

LMRT p-value (.265). The 3-class solution showed high entropy (.93) and reduced information

criteria, as compared to the 2-class model, however, the 3-class solution also did not show a

significant LMRT p-value (.328). In the 3-class solution, the largest class Habituating (79.2% of

the sample) showed a marginally significant negative slope during the second fear video segment

(ESTslope = -.01, SE = .10, p = .060), the Habituating group also showed an intercept that was

significantly different from zero (ESTintercept = 4.96, SE = .02, p < .001). The second class Non-

habituating (10.3% of the sample) showed a marginally significant negative slope during the

second fear video segment (ESTslope = -.03, SE = .07, p = .066), the Non-habituating group also

showed an intercept that was significantly different from zero (ESTintercept = 8.18, SE = .02, p <

.001). The third class, Reactive (10.5% of the sample) showed a marginally significant positive

slope during the second fear video segment (ESTslope = .27, SE = .04, p = .063), the Reactive

group also showed an intercept that was significantly different from zero (ESTintercept = 3.99, SE =

.03, p < .001) (see Figure 7).

43
 Second fear recovery period. A 1-class solution was identified as the best fitting model

for the second fear recovery segment of the fear reactivity task, indicating that the full sample

mean might best represent the pattern of change during this segment of the task. A 2-class

solution was ruled out for this segment of data despite high entropy (.95) and reduced

information criterion values. The 2-class solution was ruled out due to the low number of

participants identified as belonging to the second class (6.3%) and a non-significant LMRT p-

value (.272), indicating likely over-fitting of the model. The single-class model showed a non-

significant negative slope during the second fear recovery segment (ESTslope = -.06, SE = .01, p =

.216) and an intercept that was significantly different from zero (ESTintercept = 5.45, SE = .06, p <

.001) (Figure 8).

Additional LGMM analyses. LGMM analyses were also conducted on the positive

video segments of the fear reactivity task. Due to these segments being less relevant for the

current research question, which pertains specifically to fear reactivity and fear inhibition, these

analyses are not detailed here. Briefly, 2-class solutions were identified as the best fitting models

for both positive emotion video segments. Moreover, in addition to the piecewise model, LGMM

was conducted to explore the full fear reactivity task, by modeling trajectories that consisted of

mean skin conductance values for each segment (a total of six segments of data). Mean values

were utilized to decrease the complexity of the model while including data from the full task in

one model. Similarly, a 2-class solution was identified as the best fitting model for the full task,

however, only 7.6% of the sample were identified as belonging to the second class. Lastly,

LGMM was conducted to model fear facial expression and reported negative emotion

experience, across the fear reactivity task (also by utilizing mean values for each task segment).

These analyses resulted in either a single-class solution (fear facial expression), or model non-

44
convergence (reported negative emotion). Thus, we do not report details regarding those models

here.

Class identification. Given the reported results of the LGMM piecewise models, the

current sample likely consists of two or three trajectories of skin conductance responding, a

pattern that emerges during the first fear recovery period. Specifically, one smaller group

characterized by higher levels of arousal (Non-habituating) and one larger group characterized

by lower levels of arousal (Habituating) emerged during the first fear recovery segment. During

the second fear video, reported results suggest that the current sample can be characterized by

three separate trajectories of responding, including a group that starts the segment with low

arousal and ends the segment with high arousal (Reactive), in addition to the Non-habituating

and Habituating groups already described. Follow-up analyses will examine group differences

utilizing class membership from the first fear recovery segment and the second fear video,

separately, rather than as a combined group5.

Aim 1a: Verify that latent groups reflect meaningful differences in fear responding.

Specifically, examine latent groups for mean differences in both fear facial expression and

reported negative emotion experience across the fear reactivity task.

Utilizing the two groups (Non-habituating and Habituating) that were derived from the

LGMM analyses on the first fear recovery segment, we conducted one repeated measures

ANOVA test to examine differences between these two groups in fear facial expression and

reported negative emotion experience across the fear reactivity task. The repeated measures

5
Although most participants in the current sample (n=66) remain within the same class (either remain in the Non-
habituating or in the Habituating group) during both the first fear recovery segment and the second fear video
segment, some participants (n=17) show a more inconsistent pattern of class membership during these segments of
the fear reactivity task. Specifically, n=6 shift from the Non-habituating group during the fear recovery segment to
the Habituating (n=4) or the Reactive (n=2) group during the second fear video. Further, n=11 shift from the
Habituating group during the fear recovery segment to the Non-habituating (n=5) or Reactive (n=6) group during
the second fear video.

45
ANOVA test followed a 6(task segment) x 2(emotion response output: fear expression and

reported negative emotion) x 2(group: Non-habituating versus Habituating) design. A significant

between-subject effect emerged F(1,67)=4.07, p=.048, partial eta2=.06, such that the Non-

habituating group showed greater fear expression and increased reported negative emotion

across the fear reactivity task, as compared to the Habituating group. Parameter estimates show

that the Non-habituating group had significantly greater fear expression during the first fear

video t(68)=2.16, p=.035, 95%CI[.07-1.73], and marginally greater fear expression during the

first fear recovery period t(68)=1.80, p=.08, 95%CI[-.02-.36], as compared to the Habituating

group. The Non-habituating group also reported significantly greater negative emotion

experience during the first fear video, as compared to the Habituating group t(68)=2.26, p=.027,

95%CI[.08-1.39].

Utilizing the three groups (Non-habituating, Habituating, Reactive) derived from the

LGMM analyses on the second fear video, we also conducted one repeated measures ANOVA

test to examine differences between these groups in fear facial expression and reported negative

emotion experience across the fear reactivity task. Similarly, this test followed a 6(task segment)

x 2(emotion response output) x 3(group) design. A marginally significant between-subject effect

emerged F(2, 65)=2.43, p=.096, partial eta2= .07, such that the Reactive group showed greater

fear expression and increased reported negative emotion experience across most of the fear

reactivity task. Parameter estimates show that the Reactive group had significantly greater fear

expression during the second fear video t(65)=-3.35, p=.001, 95%CI[-2.32--.59], as compared to

the other two groups. The Reactive group also reported significantly greater negative emotion

experience during the second fear video t(65)=-3.13, p=.003, 95%CI[-1.68--.37], and during the

46
following recovery period t(65)=-2.48, p=.016, 95%CI[-1.11--.12], as compared to the other two

groups.

These results indicate that latent groups derived from LGMM analyses do indeed reflect

meaningful differences in fear responding across the full task. Above reported findings also

suggest that there is some level of emotion response coherence across the full fear reactivity task

within the latent groups that were derived from the piecewise LGMM analyses. This is

particularly clear during the fear videos where increases emerged on all emotion response

dimensions for all groups, and most distinctly for the Non-habituating group.

Aim 1b: Examine latent groups for differences in factors that might indicate

psychiatric risk, specifically, psychological symptoms and reported threat sensitivity. Also

examine latent groups for differences in regulatory resources, specifically, Stroop reaction

time and positive emotion expression.

No significant differences emerged when comparing the latent groups on psychological

symptoms or reported threat sensitivity.

Examination of the three groups derived from the second fear video (Non-habituating,

Habituating, and Reactive) resulted in a marginally significant group difference on positive

emotion expression during the first fear video, F(2,74)=2.49, p=.090, partial eta2=.07.

Examination of parameter estimates showed that the Reactive group was marginally different

from the Non-habituating group t(75)=1.76, p=.083, 95%CI[-.14-2.33], and significantly

different from the the Habituating group, t(75)=2.20, p=.031, 95%[.10-2.02], on positive

emotion expression. Specifically, individuals in the Reactive group showed significantly lower

positive emotion expression (M=1.19, SD=.45), as compared to the Habituating group (M=2.25,

SD=.17), and as compared to individuals in the Non-habituating group (M=2.28, SD=.43). There

47
was no significant difference between the Non-habituating group and the Habituating group on

positive emotion expression during the first fear video (p=.944), and no significant group

differences emerged on positive emotion expression during the second fear video F(2,73)=.46,

p=.634, partial eta2 =.01. No other significant group differences emerged when comparing latent

groups on positive emotion expression or Stroop reaction time6.

These data suggest that although the latent groups derived from the current sample might

show distinct patterns of fear responding, this might not explain psychiatric risk, as measured in

the current study.

Aim 2: Examine the association between fear reactivity and fear recovery (utilizing change

scores derived from fear responding during the fear reactivity task). Additionally, examine the

association between fear reactivity, fear recovery, and factors reflective of psychiatric risk

(psychological symptoms and threat sensitivity).

Associations between fear reactivity and fear recovery. To index fear reactivity and

recovery, change scores were calculated for skin conductance, fear expression, and reported

negative emotion experience (detailed above), and Pearson correlations were then utilized to

examine all associations in aim 2. Broadly, fear reactivity and fear recovery were significantly

correlated in a positive direction, such that increased reactivity was associated with greater

recovery (range r = .22-.98, p<.05). This pattern was generally consistent across emotion

response dimensions (see Tables 6,7,8 for all statistics). However, skin conductance recovery

6
High SC versus Low SC (derived from fear recovery segment): No significant group differences were found on the
SCL-90 F(1,81)=.31, p=578, STAI F(1,81)=.01, p=.960, CESD F(1,81)=.47, p=.493, BIS F(1,81)=.33, p=.569,
Stroop task difference score, F(1,67)=.01, p=.956.
High SC versus Low SC versus Low-to-High SC (derived from second fear video): No significant group differences
were found on the SCL-90 F(2,80)=.99, p=.377, STAI F(2,80)=.52, p=.594, CESD F(2,80)=.74, p=.481, BIS
F(2,80)=.39, p=.679. or the Stroop task difference score, F(2,66)=1.92, p=.155.

48
following the first fear video was negatively correlated with skin conductance reactivity during

the second fear video r(81)=-.25, p=.026. This indicates that greater skin conductance recovery

following the first fear elicitation in the task is associated with less arousal (reactivity) during the

second fear elicitation in the current sample.

Fear reactivity: Fear responding during the fear videos. Greater fear reactivity

(indexed by skin conductance) during the first fear video was marginally correlated with less

psychological symptoms and reported threat sensitivity, as measured on the SCL-90 r(82)=-.24,

p=.034, CESD r(82)=-.21, p=.061, STAI r(82)=-.21, p=.061, and BIS r(82)=-.19, p=.097. Fear

reactivity during the first fear video was not significantly associated with any other factors under

investigation. These data suggest that greater arousal within the initial fear elicitation of the fear

reactivity task is associated with lower symptomatology.

During the second fear video, increased fear reactivity (indexed by fear expression) was

marginally associated with increased threat sensitivity (BIS), r(87)=.20, p=.067. This indicates

that more fear reactivity (detected in facial expression responding) during the second fear

elicitation of the task is associated with more reported threat sensitivity. This is in contrast to fear

responding detected during the first fear video, where increased reactivity was associated with

less symptomatology and less threat sensitivity. Fear reactivity during the second fear video was

not significantly associated with any other factors under investigation.

Fear recovery: Fear responding during recovery following the first fear elicitation

segment. Following the first fear video, reported negative emotion recovery was marginally

associated with BIS r(91)=-.19, p=.077, such that more reported recovery is associated with

49
lower levels of reported threat sensitivity. No other significant associations emerged between

fear recovery and factors under investigation in the current sample7.

Taken together, these findings broadly align with results that emerged from above

presented LGMM analyses. Indeed, a single-class solution emerged from LGMM analyses on the

first fear elicitation segment of the fear reactivity task, indicating that the current sample

responded in a relatively uniform manner to this first elicitation. Latent groups did, however,

emerge during the recovery segment that follows the first fear elicitation. Consistent with this,

reactivity (derived from change scores) during the first fear elicitation was not found to be

significantly associated with level of recovery during the segment that follows. Rather, more

recovery following the first fear elicitation was significantly associated with less reactivity

during the second fear elicitation of the task. Furthermore, increased reactivity during the first

fear video was associated with less symptomatology and less reported threat sensitivity,

however, reactivity during the second fear elicitation of the task was associated with more

reported threat sensitivity. Lastly, latent groups that emerged during the second fear elicitation

showed significant differences in positive emotion, such that the Reactive group showed less

positive emotion expression within a negative emotion context. Exploratory analyses (presented

below) aim to clarify some of these associations, taking into account regulatory resources that

7
In addition to the change score analyses described, we also examined slopes to indicate fear reactivity
and recovery in the current sample. Specifically, we created a lagged skin conductance variable for the
fear recovery segments and then conducted a multilevel regression model (PROC MIXED, SAS, 2008)
predicting next 10-second interval skin conductance level from current 10-second interval skin
conductance level. An autoregressive error structure was utilized, and random effects for the slope
coefficients were extracted. The random effects serve as an estimate of slope in skin conductance for the
specified task segment (e.g. Coifman, Berenson, Rafaeli, & Downey, 2012). The slope estimates were
then examined for associations with symptoms, Stroop reaction time, and positive emotion expression, by
the same procedure as the change scores described in the main text. The additional slope analyses did not
yield any unique information regarding fear reactivity or recovery thus, we do not report details from
these analyses in the current article.

50
might influence the relationship between reactivity and recovery, as well as between reactivity

and threat sensitivity. Specifically, based on prior research (Coifman, Halachoff, Nylocks, 2018)

suggesting that greater executive resources might mitigate risks associated with high threat

sensitivity, we opted to explore Stroop task performance as a moderator between reactivity and

reported threat sensitivity in the current sample. Additionally, based on models of positive

emotions suggesting that increased positive emotion might accelerate recovery from negative

emotions (Fredrickson, 1998; 2001), we also explored positive emotion expression as a

moderator in the relationship between reactivity and recovery in the current sample.

Exploratory aim 1: To explore how the association between fear reactivity and psychiatric

risk might be influenced by executive functioning, Stroop reaction time will be tested as a

moderator in the relationship between fear reactivity and threat sensitivity.

An exploratory moderation analysis was conducted to further examine the relationship

between reported threat sensitivity and fear reactivity during the second fear video. Specifically,

we sought to explore if the relationship between reactivity during the second fear video (indexed

by skin conductance) (dependent variable) and threat sensitivity (independent variable) might be

moderated by Stroop reaction time (within a negative context). Hayes’ (2013) PROCESS Model

1 was used to test this relationship. Indeed, results showed a significant moderation effect b =

.0003, 95%CI[-.00-.00], t=1.95, p=.055, such that the relationship between high reactivity and

high BIS emerged in individuals with poor Stroop task performance within a negative emotion

context (i.e. in Stroop difference scores 1 SD above the mean) t(63)=1.62, p=.055, 95%CI[-.01-

.11]. The unstandardized simple slope for participants 1 SD above the mean Stroop reaction time

(within a negative context) was .05 (p=.055), the unstandardized simple slope for participants

51
with a mean level of Stroop reaction time was .01 (p=.266), and the unstandardized simple slope

for participants 1 SD below the mean Stroop reaction time was -.03 (p=.137) (see Figure 9).

Additionally, we examined if the relationship between reactivity during the second fear

video (indexed by skin conductance) (as the dependent variable) and threat sensitivity (as the

independent variable) might be moderated by Stroop reaction time (within a neutral context).

Hayes’ (2013) PROCESS Model 1 was used to test this relationship. Similar to above reported

results, a significant moderation effect emerged b = .0005, 95%CI[-.00-.00], t=2.02, p=.047,

such that the relationship between high reactivity and high BIS emerged in individuals with poor

Stroop task performance within a neutral context (i.e. in Stroop difference scores 1 SD above the

mean) t(63)=1.68, p=.048, 95%CI[-.01-.12]. A marginal effect also emerged such that the

relationship between high reactivity and low BIS emerged in individuals with increased Stroop

task performance within a neutral context (i.e. Stroop difference scores 1 SD below the mean),

t(63)=-1.41, p=.082, 95%CI[-.09-.01]. The unstandardized simple slope for participants 1 SD

above the mean Stroop reaction time (within a neutral context) was .05 (p=.048), the

unstandardized simple slope for participants with a mean level of Stroop reaction time was .01

(p=.338), and the unstandardized simple slope for participants 1 SD below the Stroop reaction

time mean was -.04 (p=.082).

Exploratory aim 2: To explore how the association between fear reactivity and fear recovery

might be influenced by positive emotions, positive emotion expression will be tested as a

moderator in the relationship between fear reactivity and fear recovery.

An exploratory moderation analysis was conducted to further examine the relationship

between fear reactivity (indexed by skin conductance during the first fear video) and fear

recovery following the first fear video. Specifically, we sought to explore if the relationship

52
between reactivity (independent variable) and recovery (dependent variable) might be moderated

by positive emotion expression (during the first fear video). Hayes’ (2013) PROCESS Model 1

was used to test this relationship. No significant moderation effect emerged, b = -.12, 95%CI[-

.33-.09], t=-1.12, p=.267.

Discussion

The current study aimed to examine fear reactivity and fear inhibition in order to further

the understanding of fear-pathology development. To this end, college students were recruited

and asked to complete a fear reactivity task measuring spontaneous fear reactivity (during two

fear eliciting videos) and spontaneous fear inhibition immediately following fear elicitation

(during recovery periods without stimuli presentation). We aimed to examine patterns of fear

responding, as well as associations between fear reactivity, fear recovery, psychological risk

factors, and regulatory resources (Stroop reaction time and positive emotion expression).

A data driven analytical approach was employed (LGMM) to model trajectories of fear

responding (indexed by skin conductance) during the fear reactivity task, and to derive latent

groups characterized by their pattern of fear responding. Results indicated that the current sample

responded in a uniform manner (single-class solution) to the first fear elicitation of the task.

However, during the first recovery period, one smaller latent group characterized by higher

levels of arousal (Non-habituating) and one larger latent group characterized by lower levels of

arousal (Habituating) emerged in the current sample. Further, during the second fear video, a

third latent group emerged, characterized by low arousal at the beginning of the segment, and

high arousal towards the end of the segment (Reactive). Additional analyses indicated that the

Non-habituating and Reactive groups show increased fear responding on multiple dimensions

(fear facial expression and reported negative emotion) across the full task, suggesting that the

53
latent groups that emerged from the current sample indeed reflect meaningful differences in fear

responding. The groups, however, did not reflect significant differences in psychological

symptoms or regulatory resources (Stroop reaction time and positive emotion expression).

The current study also found significant associations between fear reactivity and fear

recovery (indexed by change scores). Broadly, increased reactivity (i.e. larger increase in level of

arousal) during fear elicitation was found to be associated with greater recovery (i.e. a larger

decrease in level of arousal) during the task segment that followed (recovery period). However,

notably, greater recovery (indexed by skin conductance) following the first fear elicitation was

associated with less reactivity during the second fear elicitation. Additionally, increased

reactivity (indexed by skin conductance) during the first fear elicitation was found to be

associated with less symptomatology and less reported threat sensitivity, whereas greater

reactivity during the second fear elicitation was associated with more reported threat sensitivity.

Exploratory analyses showed that Stroop reaction time (within both negative emotional

context and neutral context) moderates the relationship between fear reactivity (in response to

the second fear elicitation) and threat sensitivity. Specifically, we found a significant association

between increased reported threat sensitivity and more fear reactivity, but only in individuals

with slower performance on the Stroop task. Positive emotion expression was not found to

moderate the relationship between fear reactivity and recovery in the current sample.

Fear responding within the context of a novel threat

Results from LGMM analyses indicated that the current sample responded in a uniform

way to the first novel fear elicitation of the fear reactivity task. As expected, manipulation checks

confirm that, across dimensions of emotion (fear facial expression, reported negative emotion,

and skin conductance response), the current sample showed a significant increase in fear

54
responding during the first fear elicitation of the task, as compared to the preceding task segment

(baseline). Interestingly, increased arousal (fear reactivity) in response to the first fear elicitation

was correlated with less psychological symptoms and less reported threat sensitivity in the

current sample. These findings are in line with functional models of emotion (Darwin,

1872/1998; Ekman, 1992; LeDoux, 2012; Mineka, 2013; Ohman & Mineka, 2001) stating that in

response to a novel threat, increased fear responding is highly adaptive and functions as a

defense response. These findings also align with research showing that emotion response

dimensions are loosely coordinated (Bulteel et al., 2014; Bonanno & Keltner, 1997) and tend to

fluctuate in a generally concordant manner in order to facilitate action (Ekman, 1992). Thus, our

findings add to this body of literature by showing that increased fear generation in response to a

novel threat might afford psychological benefits.

Fear responding following the removal of a novel threat

Following the fear elicitation segments of the fear reactivity task, participants were

presented with a “recovery period” designed to measure spontaneous fear inhibition. LGMM

analyses showed that during the first recovery period, a larger latent group (Habituating) and a

smaller latent group (Non-habituating) emerged in the sample. As aforementioned, the Non-

habituating group was found to show significantly greater fear responding on multiple response

dimensions (fear facial expression and reported negative emotion) across the full task, suggesting

that the latent groups derived from the current sample indeed reflect meaningful differences in

fear responding. Consistent with results indicating a single-class solution during the first fear

elicitation segment, results indicated that fear recovery (indexed by skin conductance) during the

first recovery period was not associated with fear reactivity during the preceding fear elicitation.

These findings suggest that potentially important individual differences in fear responding

55
become detectable, not in response to threat, but following the removal of the threat when fear

inhibition is expected and largely beneficial. Indeed, a hallmark of fear-pathology is sustained

fear following the removal of a threat, a pattern of responding seen in cases of, for example,

generalized anxiety disorder, PTSD, and social anxiety disorder (APA, 2013).

Unexpectedly, we did not find significant differences in psychological symptoms or

reported threat sensitivity between the latent groups derived from the current sample. Prior

research suggests a strong association between deficits in fear inhibition and psychiatric risk (e.g.

Craske et al., 2008; 2009; 2012; Lissek et al., 2009; Waters et al., 2014). As such, we expected

the Non-habituating group (showing increased fear responding during the fear recovery period,

as compared to the Habituating group) to be characterized by increased psychological symptoms

and more reported threat sensitivity. Our findings might suggest that there are other factors that

influence the relationship between fear inhibition and symptomatology (thus, exploratory

analyses were conducted, discussed further below). Moreover, Craske and colleagues (2012)

found that poor fear inhibition during the “safety” condition of their task (immediately following

the “danger” condition) predicted increased risk of anxiety disorder development three to four

years following the experiment (in a sample of adolescents). It is possible that the Non-

habituating group of the current sample (consisting of young adults) are also at increased risk for

future development of psychological symptoms, that are not yet detectable. It will be important

for future research to further examine deficits in fear inhibition in a prospective manner.

Fear reactivity in response to a familiar threat

We found a significant association between more recovery following the first fear

elicitation of the fear reactivity task and less reactivity during the second fear elicitation,

suggesting that a larger decrease in arousal in response to removal of a novel threat correlates

56
with less arousal in response to a subsequent familiar threat. Further, results showed an

association between increased fear reactivity during the second fear elicitation and increased

reported threat sensitivity in the current sample. These data suggest that those individuals with

greater threat sensitivity show poor fear inhibition in response to a familiar threat. This aligns

with models of fear-pathology (Duits et al., 2015; Lissek et al., 2005) and literature showing a

link between fear-pathology and fear that is resistant to change (e.g. Foa & Kozak, 1986). Our

findings also add important information to the literature on threat sensitivity as core dimension in

models of fear-pathology, and emotion-related disorders more broadly (Beauchaine & Thayer,

2015). Indeed, the association between reported threat sensitivity and arousal during the second

fear elicitation suggests that individuals who report more threat sensitivity indeed show

hypervigilance in response to threat even when the threat is not novel.

Given that we did not discover increased threat sensitivity in the Non-Habituating or

Reactive latent groups derived from LGMM, but we know that this was an important place for

divergent trajectories, we opted to further examine the relationship between threat sensitivity and

fear reactivity during the second fear elicitation using a continuous indicator of reactivity. Based

on prior research showing that increased executive resources might mitigate risks associated with

increased threat sensitivity (Coifman, Halachoff, & Nylocks, 2018) we conducted an exploratory

moderation analysis to examine Stroop performance (within a negative emotion context and a

neutral context) as a moderator between reactivity and threat sensitivity. As reported, Stroop

performance indeed moderated this relationship such that increased reactivity predicted more

reported threat sensitivity, but only in individuals with slower Stroop performance. This is

consistent with our prior findings, as well as a larger body of research on the interaction between

executive resources, threat sensitivity, and emotion responding (Dennis & Chen, 2007;

57
Derryberry & Reed, 2002). Indeed, increased threat sensitivity suggests increased attention

towards threat thus, it is theoretically reasonable that individuals with this tendency would show

greater fear reactivity towards a familiar threat, as compared to individuals with lower reported

threat sensitivity. Additionally, increased speed of performance on the Stroop task would

indicate potentially greater control of attention, this ability might serve to regulate negative

emotions (and decrease arousal) when presented with a familiar threat (second fear video of the

task), also consistent with literature on top-down processing (Ochsner, 2009; Pape & Pare,

2010). It is particularly interesting that this effect emerges from examination of Stroop

performance within a negative emotional context as well as a neutral context. This suggest that

ability to control attention when extraneous content is present, not specifically when negative

emotional content is present, might be an important ability that could indicate greater ability to

also inhibit fear in response to a familiar threat.

Positive emotion generation within a negative emotion context has previously been

shown to facilitate down-regulation of negative emotions (Fredrickson, 1998). However,

findings from the current study did not align with this research. Results did indicate that the

Reactive group (characterized by increasing arousal during the second fear video) derived from

LGMM showed lower positive emotion expression during the first fear video, as compared to the

other latent groups. However, in our exploratory analysis, we did not find positive emotion

expression to moderate the relationship between reactivity and recovery. The fear eliciting

videos utilized in the current study might not have allowed participants much “room” to express

positive emotion, indeed, the videos were selected because they elicited fear fairly discretely.

58
Strengths and Limitations

Reported findings need to be considered in light of both strengths and limitations of the

current study. Notable strengths of the current study include measurement of fear on multiple

response dimensions, including behavioral (facial expressions), experiential (self-reported

emotion), and physiological (skin conductance response) (Ekman, 1992; Ohman & Mineka,

2001). As such, the current study captured more of the complexity in emotional responding than

if only one dimension had been included. Indeed, we were able to verify latent groups derived

from LGMM analyses by conducting repeated measures ANOVA tests on multiple emotional

response channels thus, increasing our confidence in the derived latent groups. Moreover, the

fear reactivity task developed for the current study provides opportunity to examine fear

responding to fear-relevant, dynamic, and long-lasting stimuli (videos that were approximately

one minute in length). Such stimuli are perhaps closer to daily life experiences, as compared to

commonly used fear elicitation paradigms were stimuli consist of static images or fear-irrelevant

stimuli (e.g. Pavlovian fear conditioning paradigms). Furthermore, the fear reactivity task

allowed for examination of spontaneous fear reactivity and spontaneous fear inhibition, rather

than relying on explicit instructions for participants to utilize certain emotion regulatory

strategies. Lastly, utilization of a data driven analytical approach (LGMM) also strengthens the

current study as this approach accounts for important variability within the sample (Jung &

Wickrama, 2008).

There are also some important limitations to the current study that need to be considered.

First, effects that emerged from the current sample were rather small, and a larger sample would

help to confirm these results. Second, LGMM analyses were only conducted on skin

conductance response and no other emotion response channels. Although skin conductance is

59
widely considered a good indicator of fear, there is likely more information to be gathered by

deriving groups from additional emotion response dimensions. Third, the fear reactivity task that

was developed and administered in the current study is novel and therefore, we can only compare

the current findings to results derived from different types of paradigms. As such, replication

studies are needed to further confirm the presented findings. Fourth, the current study did not

recruit for high depression scores, however, the current sample reported higher levels of

depression than general college-aged samples (Latsko et al., 2016; Matt, Fresco, Coifman, 2016)

and a large portion (approximately 42%) of the current sample scored above the clinical cut-off

on the CES-D. The high rate of depression in the current sample is important to consider when

interpreting the current findings as it might impact the associations between patterns in fear

responding and psychological risk. Indeed, reduced emotional reactivity (in response to emotion

eliciting laboratory paradigms) has been found in individuals with major depression (Rottenberg,

2017). Additionally, increased depressive symptoms might reduce ability to recovery from

negative emotional states (Kashdan & Rottenberg, 2010). However, there is also some research

showing no difference between depressed and healthy individuals in skin conductance reactivity

in response to negative images (Rosebrock, et al., 2017). Lastly, in the fear reactivity task, there

was a time gap between fear elicitation and fear recovery segments where participants were

asked to rate their current emotional experience. This means that for approximately 20 seconds

between segments, there might be meaningful changes in fear responding that are not captured.

This is methodologically difficult to account for, however, findings from the current study

should be interpreted with this in consideration.

60
Clinical implications

Reported findings align with current models of fear-pathology (Foa et al., 2011) and

research on anxiety-disorder development (Buss et al., 2011; Craske et al., 2012). Furthermore,

the current study adds to a growing body of literature on transdiagnostic risk factors, or core

dimensions of emotion-related disorders (Beauchaine & Thayer, 2015). Current findings suggest

that there might some benefit in focusing clinical interventions on attentional biases within

familiar negative emotion contexts. In particular, individuals reporting high threat sensitivity

might benefit from such interventions.

The current findings suggest that fear reactivity in response to a familiar threat might be a

particularly important component of the fear response to consider in intervention. Indeed, this is

in line with research suggesting that deficits in inhibitory learning is a core element of anxiety

disorders and that exposure therapy protocols can be made more effective by placing larger focus

on this component (e.g. Bouton, 1993; Craske et al., 2014; Wagner, 1981). In particular, our data

suggests, which is consistent with suggested intervention procedures laid out by Craske and

colleagues (2014), utilizing expectancy violation as a therapeutic strategy could come with great

benefits. This strategy, used during exposure therapy, places emphasis on increasing fear

inhibition by maximizing the gap between the patient’s expectations (i.e. expectations

concerning the frequency and intensity of negative/aversive outcomes from exposure to the

feared stimuli) and the actual outcome from exposure to the feared stimuli. For example, a

patient might expect to be severely injured or die if they encounter a large dog (feared stimuli),

and the actual outcome of the exposure (to a large dog) is likely no severe injuries or death.

Given the focus on incongruence between what is expected and what is experienced of this

strategy, it is also suggested that discussion with the patient (cognitive interventions) regarding

61
realistic threat of the feared stimuli, should only take place following exposure. This is different

from traditional exposure therapy protocols (Foa & Kozak, 1986) where the emphasis is on fear

habituation within the session.

Future directions

First, the current findings need to be replicated in a larger and more diverse sample. The

fear reactivity task should also be administered in conjunction with additional measures of

psychological risk and regulatory resources. Specifically, future studies should include

comprehensive measures of symptomatology, preferably by utilizing a clinical diagnostic

interview (e.g. SCID) to assess psychological symptoms. This is important in order to assess,

more thoroughly and accurately, symptoms of all fear-related disorders, without relying on self-

report measures. It will be important to both replicate current findings, as well as examine if fear

response patterns that emerge in the fear reactivity task might be predictive of future

psychopathology. Indeed, we would be interested in adding to existing literature (e.g. Buss et al.,

2011; Craske et al., 2012). To do so, a follow-up session (where symptoms could be assessed)

would need to be added to the current study protocol. These future steps are important

considering our lack of significant differences between latent groups on symptomatology.

Indeed, it seems likely that individuals with a pattern of consistent arousal during the fear

reactivity task would be at increased risk of fear-pathology, but the current study might not have

been comprehensive enough to detect such differences. Further, future studies might benefit from

including additional measures of executive cognitive functioning in order to further validate

reported associations between reactivity, threat sensitivity, and Stroop performance.

62
Conclusion

The current study aimed to examine spontaneous fear reactivity and fear inhibition in

order to further the understanding of fear-pathology development. Specifically, we aimed to

model trajectories in fear responding (indexed by skin conductance) by using LGMM, and derive

latent groups characterized by their pattern of responding. The current study then examined

derived latent groups for differences in psychological symptoms and threat sensitivity, as well as

regulatory resources (executive cognitive functioning and positive emotion expression). The

current study also tested the association between fear reactivity and fear recovery for the full

sample by utilizing more conventional data analyses (change scores). Results indicate that

contextually appropriate fear, consisting of increased fear reactivity in response to a novel threat,

is a common response that is associated with decreased psychiatric risk. Our data also suggest

that meaningful individual differences in fear responding are likely to emerge in response to

removal of a novel threat, where fear inhibition is expected. Further, our findings indicate a link

between psychological risk (reported threat sensitivity) and increased fear responding to a

familiar threat. This association is moderated by executive cognitive functioning (indexed by

Stroop reaction time in the current study), suggesting that increased executive cognitive

functioning might serve as a regulatory resource within the context of fear. The findings from the

current study align with current models of fear-pathology, and add to literature suggesting that

executive resources might mitigate psychological risks associated with threat sensitivity.

63
Tables and Figures

Table 1. Sample demographic characteristics

Demographic Variables
(n=101)

Age M=20.07, SD=3.43

Sex 77 (76.2%) Female

24 (23.8%) Male
Race
Caucasian 72 (71.3%)
African American 14 (13.9%)
Asian 3 (3%)
Other 12 (11.88%)

Ethnicity
Not Hispanic or Latino 94 (93.1%)
Hispanic or Latino 6 (5.9%)

CESD M=16.64, SD=10.50

BIS M=21.22, SD=4.18

SCL-90 M=.91, SD=.63

64
Table 2. Stroop task interference scores
a) Mean interference score per task condition

Task M (SD)

Color-word Stroop 45.53 (51.90)

Emotional Color-word Stroop

Neutral condition 61.39 (81.21)

Negative condition 57.19 (53.49)
Positive condition 65.02 (97.47)

Difference score (Negative 19.37 (138.69)

condition – Color-word Stroop)

Difference score (Neutral 8.26 (79.27)

condition – Color-word Stroop)

b) Pearson correlations between task conditions

(1) (2) (3) (4)

Color-word 1 -- -- --
Stroop
(1)

Emotional .28* 1 -- --
Stroop: Neutral
condition
(2)

Emotional .33** .21* 1 --

Stroop: Negative
emotion condition
(3)

Emotional .05 .31** .09 1

Stroop: Positive
emotion condition
(4)

* = p<.05
** = p<.01

65
Table 3. Missing data

Total number of Type of data missing Reason for missing data

participants with
missing data

1. Participant did not speak English fluently (n=1)

4 All study data 2. Participant did not feel well and need to end the
session early (n=2)
3. Participant had previously participated in a
different study utilizing the same video stimuli as
the present study (n=1)
1. BioLab froze during the task and no skin
18 EDA signal conductance data was saved (n=1)
2. Equipment malfunction resulting in lack of
detectable signal for some parts of the task (n=5).
When possible, these missing values were
preserved by imputation.
3. Administrator error (electrodes not placed
properly), resulting in lack of detectable signal for
the full task, or some parts of the task (n=7)
4. Naturally occurring irregular or low skin
conductance8 resulting in lack of detectable signal
for the full task (n=5)

1. Participant misunderstanding the instructions for

3 Self-reported emotion how to fill out the affect rating forms (n=3)
experience
1. Error in video recording (camera software
4 Emotion expression malfunction) (n=3)
2. Participant request to not be video recorded (n=1)
1. Color-word Stroop: administrator error (the task
2 Stroop tasks was not administered by mistake) (n=1)
2. Both Stroop tasks: participant was color blind
(n=1)

Initial sample size: n=101

Number of participants with some missing data: n=31

8
Notably, due to individual skin conductance variation within the general population, some
people naturally show irregular or low skin conductance (termed “nonresponders”) (Dawson et al., 2007).
When such irregularities were discovered during data collection for the present study, the research
investigator attempted to adjust the electrodes and massage fingers of the participant in order to better
collect the EDA data. Adjustments were unsuccessful for these participants, resulting in lack of a
detectable signal.

66
Table 4: LGMM model fit statistics, per task segment (*=best fit).

First fear video

Measure 1 class* 2 class 3 class

n per class 83 82/1 81/1/1

Loglikelihood -722.334 -697.465 -692.072

AIC 1476.668 1432.929 1428.145
BIC 1515.369 1478.887 1481.359
SSA-BIC 1464.901 1418.957 1411.966
Entropy n/a 1 0.996
LMR test n/a -722.334 -697.465
LMR, p-value n/a 0.0162 0.0361

First fear recovery period

Measure 1 class 2 class* 3 class
n per class 83 73/10 55/21/7

Loglikelihood -380.595 -360.228 -352.232

AIC 787.189 752.456 742.465
BIC 818.634 791.158 788.423
SSA-BIC 777.629 740.69 728.492
Entropy n/a 0.986 0.915
LMR test n/a -380.595 -360.228
LMR, p-value n/a 0.0373 0.4006

Second fear video

Measure 1 class 2 class 3 class*

n per class 82 76/6 64/8/8

Loglikelihood -675.214 -664.508 -653.172

AIC 1382.427 1367.016 1350.344
BIC 1420.935 1412.744 1403.292
SSA-BIC 1370.471 1352.819 1333.905
Entropy n/a 0.961 0.928
LMR test n/a -675.214 -664.508
LMR, p-value n/a 0.2645 0.3279

67
Second fear recovery period

Measure 1 class* 2 class 3 class

n per class 82 77/5 70/7/5

Loglikelihood -330.421 -320.557 -309.616

AIC 686.842 673.113 657.231
BIC 718.13 711.621 702.959
SSA-BIC 677.128 661.158 643.034
Entropy n/a 0.95 0.968
LMR test n/a -330.421 -320.557
LMR, p-value n/a 0.2717 0.252

68
Table 5. Pearson correlations between skin conductance reactivity, recovery, psychological
symptoms, Stroop task interference difference score, and positive emotion expression.

(1) (2) (3) (4)

First SC reactivity 1 -- -- --
(1)

First SC recovery -.03 1 -- --

(2)

Second SC reactivity .07 -.25* 1 --

(3)

Second SC recovery .08 -.32** .58** 1

(4)
SCL-90 -.24* .09 .06 .01

STAI -.21 .03 .11 .11

CESD -.21 .09 -.01 .02

First positive .01 -.02 -.04 -.04

emotion expression

Second positive .03 .03 .16 .61

emotion expression

Stroop interference .07 .01 .05 .05

difference score

*=p<.05
**=p<.01

69
Table 6. Pearson correlations between reported negative emotion reactivity, recovery,
psychological symptoms, Stroop task interference difference score, and positive emotion
expression.

(1) (2) (3) (4)

First negative 1 -- -- --
emotion experience
reactivity
(1)

First negative .54** 1 -- --

emotion experience
recovery
(2)

Second negative .22* .25* 1 --

emotion experience
reactivity
(3)

Second negative .07 .39** .79** 1

emotion experience
recovery
(4)

SCL-90 .09 -.03 -.11 -.11

STAI .04 -.05 -.16 -.11

CESD .03 .04 -.18 -.13

First positive -.31** -.01 -.01 .05

emotion expression

Second positive -.34** -.12 -.33** -.22*

emotion expression

Stroop interference -.07 -.01 -.08 -.13

difference score

*=p<.05
**=p<.01

70
Table 7. Pearson correlations between fear expression reactivity, recovery, psychological
symptoms, Stroop task interference difference score, and positive emotion expression.

(1) (2) (3) (4)

First fear expression 1 -- -- --

reactivity
(1)

First fear .98** 1 -- --

expression recovery
(2)

Second fear .56** .55** 1 --

expression reactivity
(3)

Second fear .52** .51** .96** 1

expression recovery
(4)

SCL-90 .04 .03 -.01 -.05

STAI .05 .04 -.03 -.05

CESD -.02 -.05 -.04 -.06

First positive -.24* -.21* -.12 -.09

emotion expression

Second positive -.17 -.13 -.10 -.11

emotion expression

Stroop interference .19 .21 .05 .06

difference score

*=p<.05
**=p<.01

71
Figure 1. Self-reported negative emotion experience across the fear reactivity task.

Self-reported negative emotion

2.50

2.00 M=2.01
M=1.89

1.50 M=1.50

M=1.17 M=1.15
1.00 M=1.07 M=1.14 M=1.16 M=1.15

0.50

0.00
Baseline Fear video Recovery Positive Recovery Fear video Recovery Positive Recovery
video period video period period video period

Self-reported fear

72
Figure 2. Fear expression across the fear reactivity task.

Fear Facial Expression

2.5
M=2.23
2 M=2.009

1.5
M=1.198 M=1.108 M=1.201
1 M=1.07 M=1.13 M=1.128 M=1.105

0.5

0
Baseline Fear video Recovery Positive Recovery Fear video Recovery Positive Recovery
video period video period period video period

Fear expression

73
Figure 3. Mean skin conductance response across the fear reactivity task.

Skin conductance mean across the fear reactivity task

Skin conductance mean

6
5.3121 5.4743
5.0478
5 4.86 5.3093
5.1028
4.6055 4.683
4 4.2305

3 3.2501

0
baseline baseline First fear First fear First First Second fear Second fear Second Second
video recovery video recovery positive positive video recovery positive positive
video recovery video recovery

Figure 4: Latent growth model (six time points) with covariate included.

Fear recovery segment

E1 E2 E3 E4 E5 E6

T1 T2 T3 T4 T5 T6

I S

X=
covariate C

74
Figure 5. First fear video segment, 1-class solution.

First fear video: 1-class solution

4.65
4.6
4.55
4.5
4.45
4.4
4.35
4.3
4.25
4.2
4.15
4.1
0 1 2 3 4 5 6 7 8

1-class solution

Figure 6. First fear recovery period, 2-class solution.

First fear recovery period: 2-class solution

10
9
8
7
6
5
4
3
2
1
0
0 1 2 3 4 5

High SC (12%) Low SC (88%)

75
Figure 7. Second fear video, 3-class solution.

Second fear video: 3-class solution

9
8
7
6
5
4
3
2
1
0
0 1 2 3 4 5 6 7 8

High SC (10.3%) Low SC (79.2%) Low-to-High SC (10.5%)

Figure 8. Second fear recovery period, 1-class solution.

Second fear recovery period: 1-class solution

5.5

5.4

5.3

5.2

5.1

4.9
0 1 2 3 4 5

1-class solution

76
Figure 9. Stroop task difference score (negative emotion context) moderating the relationship
between fear reactivity observed during the second fear video (indexed by skin conductance) and
reported threat sensitivity.

-. 0
3, p
= .13
7

=.266
.01, p
5
.05
p=
5,
.0

77
 References

Adleman, N. E., Menon, V., Blasey, C. M., White, C. D., Warsofsky, I. S., Glover, G. H., &

Reiss, A. L. (2002). A developmental fMRI study of the Stroop color-word

task. Neuroimage, 16(1), 61-75.

Aguado, L., Fernández-Cahill, M., Román, F. J., Blanco, I., & de Echegaray, J. (2016).

Evaluative and psychophysiological responses to short film clips of different emotional

content. Journal Of Psychophysiology, 32(1), 1-19. doi:10.1027/0269-8803/a000180

Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317–332.

doi:10.1007/BF02294359

Ali, S., Rhodes, L., Moreea, O., McMillan, D., Gilbody, S., Leach, C., Lucock, M., Lutz, W.,

& Delgadillo, J. (2017). How durable is the effect of low intensity CBT for depression

and anxiety? Remission and relapse in a longitudinal cohort study. Behaviour Rsearch

and Therapy, 94, 1-8.

Almahmoud, S. Y., Coifman, K. G., Ross, G. S., Kleinert, D., & Giardina, P. (2016). Evidence

for multidimensional resilience in adult patients with transfusion-dependent thalassemias:

Is it more common than we think? Transfusion Medicine, 26(3), 186-194.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders

(5th ed.). Washington, DC: Author.

Archibald, S. J., & Kerns, K. A. (1999). Identification and description of new tests of executive

functioning in children. Child Neuropsychology, 5(2), 115-129.

Balle, M., Tortella-Feliu, M., & Bornas, X. (2013). Distinguishing youth at risk for anxiety

disorders from self-reported BIS sensitivity and its psychophysiological concomitants.

Internals Journal of Psychology, 48, 964-977.

78
Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H.

(2007). Threat-related attentional bias in anxious and nonanxious individuals: A meta-

analytic study. Psychological Bulletin, 133(1), 1-24. doi:10.1037/0033-2909.133.1.1

Beauchaine, T. P., & Thayer, J. F. (2015). Heart rate variability as a transdiagnostic biomarker of

psychopathology. International Journal of Psychophysiology, 98(2, Part 2), 338–350.

https://doi.org/10.1016/j.ijpsycho.2015.08.004

Berlin, K. S., Parra, G. R., & Williams, N. A. (2013). An introduction to latent variable mixture

modeling (part 2): Longitudinal latent class growth analysis and growth mixture

models. Journal Of Pediatric Psychology, 39(2), 188-203. doi:10.1093/jpepsy/jst085

Bertocci, M. A., Bebko, G., Olino, T., Fournier, J., Hinze, A. K., Bonar, L., & ... Phillips, M. L.

(2014). Behavioral and emotional dysregulation trajectories marked by prefrontal–

amygdala function in symptomatic youth. Psychological Medicine, 44(12), 2603-2615.

doi:10.1017/S0033291714000087

Bishop, S. J., Duncan, J., & Lawrence, A. D. (2004). State Anxiety Modulation of the Amygdala

Response to Unattended Threat-Related Stimuli. The Journal Of Neuroscience, 24(46),

10364-10368. doi:10.1523/JNEUROSCI.2550-04.2004

Bishop, S., Duncan, J., Brett, M., & Lawrence, A. D. (2004). Prefrontal cortical function and

anxiety: Controlling attention to threat-related stimuli. Nature Neuroscience, 7(2), 184-

188. doi:10.1038/nn1173

Bonanno, G. A., & Keltner, D. (1997). Facial expressions of emotion and the course of conjugal

bereavement. Journal Of Abnormal Psychology, 106(1), 126-137. doi:10.1037/0021-

843X.106.1.126

79
Bonanno, G. A., Mancini, A. D., Horton, J. L., Powell, T. M., LeardMann, C. A., Boyko, E. J., &

... Smith, T. C. (2012). Trajectories of trauma symptoms and resilience in deployed US

military service members: Prospective cohort study. The British Journal Of

Psychiatry, 200(4), 317-323. doi:10.1192/bjp.bp.111.096552

Bouton, M. E. (1993). Context, time, and memory retrieval in the interference paradigms of

Pavlovian learning. Psychological Bulletin, 114, 80e99. http:// dx.doi.org/10.1037/0033-

2909.114.1.80.

Breuninger, C., Sláma, D. M., Krämer, M., Schmitz, J., & Tuschen-Caffier, B. (2017).

Psychophysiological reactivity, interoception and emotion regulation in patients with

agoraphobia during virtual reality anxiety induction. Cognitive Therapy And

Research, 41(2), 193-205. doi:10.1007/s10608-016-9814-9

Brown, L. A., LeBeau, R. T., Chat, K. Y., & Craske, M. G. (2017). Associative learning versus

fear habituation as predictors of long-term extinction retention. Cognition And

Emotion, 31(4), 687-698. doi:10.1080/02699931.2016.1158695

Bulteel, K., Ceulemans, E., Thompson, R. J., Waugh, C. E., Gotlib, I. H., Tuerlinckx, F., &

Kuppens, P. (2014). DeCon: A tool to detect emotional concordance in multivariate time

series data of emotional responding. Biological Psychology, 98, 29–42.

https://doi.org/10.1016/j.biopsycho.2013.10.011

Buss, K.A. (2011). Which fearful toddlers should we worry about? Context, fear, regulation, and

anxiety risk. Developmental Psychology, 47(3), 804-819.

Campbell-Sills, L., Barlow, D. H., Brown, T. A., & Hofmann, S. G. (2006). Effects of

suppression and acceptance on emotional responses of individuals with anxiety and mood

disorders. Behaviour Research And Therapy, 44(9), 1251-1263.

80
 doi:10.1016/j.brat.2005.10.001

Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective

responses to impending reward and punishment: The BIS/BAS Scales. Journal of

Personality and Social Psychology, 67(2), 319–333. https://doi.org/10.1037/0022-

3514.67.2.319

Carver, C. S., & White, T. L. (2013). behavioral avoidance/inhibition (BIS/BAS) scales.

Measurement Instrument Database for the Social Science. Retrieved from www.midss.ie

Chen, J., Wang, Z., Wu, Y., Cai, Y., Shen, Y., Wang, L., & Shi, S. (2013). Differential

attentional bias in generalized anxiety disorder and panic disorder. Neuropsychiatric

Disease And Treatment, 9.

Coifman, K.G., & Bonanno, G.A. (2010). When distress does not become depression: Emotion

context sensitivity and adjustment to bereavement. Journal of Abnormal Psychology,

119(3), 479-490.

Coifman, K.G., Halachoff, D.J., & Nylocks, K.M. (2018). Mitigating risk? Set-shifting ability in

high threat sensitive individuals predicts approach behavior during simulated peer-

rejection. Journal of Social and Clinical Psychology, 37(7), 481-513.

Connor, A., Franzen, M. D., & Sharp, B. (1988). Effects of practice and differential instructions

on Stroop performance. International Journal of Clinical Neuropsychology, 10(1), 1-4.

Craske, M. G., Waters, A. M., Nazarian, M., Mineka, S., Zinbarg, R. E., Griffith, J. W., & ...

Ornitz, E. M. (2009). Does neuroticism in adolescents moderate contextual and explicit

threat cue modulation of the startle reflex?. Biological Psychiatry, 65(3), 220-226.

doi:10.1016/j.biopsych.2008.07.020

81
Craske, M. G., Waters, A. M., Bergman, R. L., Naliboff, B., Lipp, O. V., Negoro, H., & Ornitz,

E. M. (2008). Is aversive learning a marker of risk for anxiety disorders in

children?. Behaviour Research And Therapy, 46(8), 954-967.

doi:10.1016/j.brat.2008.04.011

Craske, M. G., Wolitzky-Taylor, K. B., Mineka, S., Zinbarg, R., Waters, A. M., Vrshek-

Schallhorn, S., & ... Ornitz, E. (2012). Elevated responding to safe conditions as a

specific risk factor for anxiety versus depressive disorders: Evidence from a longitudinal

investigation. Journal Of Abnormal Psychology, 121(2), 315-324. doi:10.1037/a0025738

Craske, M. G., Liao, B., Brown, L., & Vervliet, B. (2012). Role of inhibition in exposure

therapy. Journal Of Experimental Psychopathology, 3(3), 322-345.

doi:10.5127/jep.026511

Craske, M. G., Treanor, M., Conway, C. C., Zbozinek, T., & Vervliet, B. (2014). Maximizing

exposure therapy: An inhibitory learning approach. Behaviour Research and Therapy, 58,

10–23. https://doi.org/10.1016/j.brat.2014.04.006

Curran, P. J., Obeidat, K., & Losardo, D. (2010). Twelve frequently asked questions about

growth curve modeling. Journal Of Cognition And Development, 11(2), 121-136.

doi:10.1080/15248371003699969

Darwin, C. (1872/1998). The Expression of the Emotions in Man and Animals. (pp. xxi-xxxvi;

33-54). Oxford University Press: New York

Davidson, R. J. (2002). Anxiety and affective style: Role of prefrontal cortex and

amygdala. Biological Psychiatry, 51(1), 68-80. doi:10.1016/S0006-3223(01)01328-2

82
Davis, T. S., Mauss, I. B., Lumian, D., Troy, A. S., Shallcross, A. J., Zarolia, P., & ... McRae, K.

(2014). Emotional reactivity and emotion regulation among adults with a history of self-

harm: Laboratory self-report and functional MRI evidence. Journal Of Abnormal

Psychology, 123(3), 499-509. doi:10.1037/a0036962

Dawson, M. E., Schell, A. M., & Filion, D. L. (2007). The electrodermal system. In J. T.

Cacioppo, L. G. Tassinary, & G. G. Berntson (Eds.), in Handbook of psychophysiology

(3rd ed., pp. 159–181). New York, NY: Cambridge University Press.

Del Boca, F. K., Darkes, J., Greenbaum, P. E., & Goldman, M. S. (2004). Up close and

personal: Temporal variability in the drinking of individual college students during their

first year. Journal Of Consulting And Clinical Psychology, 72(2), 155-164.

doi:10.1037/0022-006X.72.2.155

Demiralp, E., Thompson, R. J., Mata, J., Jaeggi, S. M., Buschkuehl, M., Barrett, L. F., & ...

Jonides, J. (2012). Feeling blue or turquoise? Emotional differentiation in major

depressive disorder. Psychological Science, 23(11), 1410-1416.

doi:10.1177/0956797612444903

Dennis, T. A., & Chen, C.-C. (2007a). Neurophysiological mechanisms in the emotional

modulation of attention: The interplay between threat sensitivity and attentional

control. Biological Psychology, 76(1–2), 1–10.

https://doi.org/10.1016/j.biopsycho.2007.05.001

Dennis, T. A., & Chen, C.-C. (2007b). Emotional face processing and attention performance in

three domains: Neurophysiological mechanisms and moderating effects of trait

anxiety. International Journal of Psychophysiology, 65(1), 10–19.

https://doi.org/10.1016/j.ijpsycho.2007.02.006

83
Derryberry, D., & Reed, M. A. (2002). Anxiety-related attentional biases and their regulation by

attentional control. Journal of Abnormal Psychology, 111(2), 225–236.

https://doi.org/10.1037/0021-843X.111.2.225

Derogatis, L. R., & Melisaratos, N. (1983). The brief symptom inventory: an introductory

report. Psychological medicine, 13(3), 595-605.

Dilger, S., Straube, T., Mentzel, H.J., Fitzek, C., Reichenbach, J.R., Hecht, H., Krieschel, S.,

&…Miltner, W.H. (2003). Brain activation to phobia-related pictures in spider phobic

humans: an event- related functional magnetic resonance imaging study. Neuroscience

Letters, 348(1), 29–32.

Duits, P., Cath, D. C., Lissek, S., Hox, J. J., Hamm, A. O., Engelhard, I. M., … Baas, J. M. P.

(2015). Updated meta-analysis of classical fear conditioning in the anxiety

disorders. Depression and Anxiety, 32(4), 239–253. https://doi.org/10.1002/da.22353

Dziak, J. J., Lanza, S. T., & Tan, X. (2014). Effect size, statistical power, and sample size

requirements for the bootstrap likelihood ratio test in latent class analysis. Structural

Equation Modeling, 21(4), 534-552. doi:10.1080/10705511.2014.919819

Eifert, G. H., & Heffner, M. (2003). The effects of acceptance versus control contexts on

avoidance of panic-related symptoms. Journal Of Behavior Therapy And Experimental

Psychiatry, 34(3-4), 293-312. doi:10.1016/j.jbtep.2003.11.001

Ekman, P. (1992). An argument for basic emotions. Cognition & Emotion, 6(3-4), p. 169-200.

Erisman, S. M., & Roemer, L. (2010). A preliminary investigation of the effects of

experimentally induced mindfulness on emotional responding to film

clips. Emotion, 10(1), 72-82. doi:10.1037/a0017162

Etkin, A., & Wager, T. D. (2007). Functional neuroimaging of anxiety: A meta-analysis of

84
 emotional processing in PTSD, social anxiety disorder, and specific phobia. The

American Journal Of Psychiatry, 164(10), 1476-1488.

doi:10.1176/appi.ajp.2007.07030504

Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007). Anxiety and cognitive

performance: Attentional control theory. Emotion, 7(2), 336-353. doi:10.1037/1528-

3542.7.2.336

Fan J., McCandliss B.D., Sommer T., Raz A., Posner M.I. (2002). Testing the efficiency and

independence of attentional networks. Journal of Cognitive Neuroscience, 14, 340–347.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using

G*Power 3.1: Tests for correlation and regression analyses. Behavior Research

Methods, 41, 1149-1160.

Flora, D. B. (2008). Specifying piecewise latent trajectory models for longitudinal

data. Structural Equation Modeling, 15(3), 513-533. doi:10.1080/10705510802154349

Foa, E.B., Dancu, C.V., Hembree, E.A., Jaycox, L.H., Meadows, E.A., & Street, G.P. (1999). A

comparison of exposure therapy, stress inoculation training, and their combination for

reducing posttraumatic stress disorder in female assault victims. J. Consult. Clin.

Psychol. 67(2), 194–200

Foa, E. B., Hembree, E. A., Cahill, S. P., Rauch, S. M., Riggs, D. S., Feeny, N. C., & Yadin, E.

(2005). Randomized trial of prolonged exposure for posttraumatic stress disorder with

and without cognitive restructuring: Outcome at academic and community

clinics. Journal Of Consulting And Clinical Psychology, 73(5), 953-964.

doi:10.1037/0022-006X.73.5.953

85
Foa, E.B., & Kozak, M.J. (1986). Emotional processing of fear: Exposure to corrective

information. Psychological Bulletin, 99(1), 20-35.

Foa, E.B., Rothbaum, B.O., Riggs, D.S., & Murdock, T.B. (1991). Treatment of posttraumatic

stress disorder in rape victims: a comparison between cognitive-behavioral procedures

and counseling. J. Consult. Clin. Psychol. 59(5), 715–23

Foa, E. B., Steketee, G., & Rothbaum, B. O. (1989). Behavioral/cognitive conceptualizations of

post-traumatic stress disorder. Behavior Therapy, 20(2), 155-176. doi:10.1016/S0005-

7894(89)80067-X

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.

doi:10.1037/0003-066X.56.3.218

Fredrickson, B.L., Tugade, M.M., Waugh, C.E., & Larkin, G.R. (2003). What good are positive

emotions in crises? 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.

Fredrickson, B. L. (1998). What good are positive emotions?. Review Of General

Psychology, 2(3), 300-319. doi:10.1037/1089-2680.2.3.300

Fredrickson, B.L., & Levenson, R.W. (1998). Positive emotions speed recovery from the

cardiovascular sequelae of negative emotions. Cognition and Emotion, 12, 191-220.

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.

doi:10.1023/A:1010796329158

86
Galatzer-Levy, I. R., Bonanno, G. A., Bush, D. A., & LeDoux, J. E. (2013). Heterogeneity in

threat extinction learning: Substantive and methodological considerations for identifying

individual difference in response to stress. Frontiers In Behavioral Neuroscience, 7

Gallo, I. S., Keil, A., McCulloch, K. C., Rockstroh, B., & Gollwitzer, P. M. (2009). Strategic

automation of emotion regulation. Journal Of Personality And Social Psychology, 96(1),

11-31. doi:10.1037/a0013460

Gilman, T. L., Shaheen, R., Nylocks, K. M., Halachoff, D., Chapman, J., Flynn, J. J., & ...

Coifman, K. G. (2017). A film set for the elicitation of emotion in research: A

comprehensive catalog derived from four decades of investigation. Behavior Research

Methods, 49(6), 2061-2082. doi:10.3758/s13428-016-0842-x

Ginsburg, G. S., Becker, E. M., Keeton, C. P., Sakolsky, D., Piacentini, J., Albano, A. M., & ...

Kendall, P. C. (2014). Naturalistic follow-up of youths treated for pediatric anxiety

disorders. JAMA Psychiatry, 71(3), 310-318. doi:10.1001/jamapsychiatry.2013.4186

Golden, C. J., & Freshwater, S. M. (1978). Stroop color and word test. Chicago, IL: Stoetting.

Gotlib, I.H., & Joormann, J. (2010). Cognition and depression: Current status and future

directions. Annual review of Psychology, 6, 285-312.

Graf, P., Uttl, B., & Tuokko, H. (1995). Color-and picture-word Stroop tests: performance

changes in old age. Journal of Clinical and Experimental Neuropsychology, 17(3), 390-

415.

Graham, B. M., & Milad, M. R. (2011). The study of fear extinction: Implications for anxiety

disorders. The American Journal Of Psychiatry, 168(12), 1255-1265.

doi:10.1176/appi.ajp.2011.11040557

87
Gray J.R., & Burgess G.C. (2004). Personality differences in cognitive control? BAS, processing

efficiency, and the prefrontal cortex. Journal of Research in Personality, 38, 35–36.

Greenbaum, P. E., Del Boca, F. K., Darkes, J., Wang, C., & Goldman, M. S. (2005). Variation in

the Drinking Trajectories of Freshmen College Students. Journal Of Consulting And

Clinical Psychology, 73(2), 229-238. doi:10.1037/0022-006X.73.2.229

Grillon, C., Warner, V., Hille, J., Merikangas, K. R., Bruder, G. E., Tenke, C. E., & ...

Weissman, M. M. (2005). Families at High and Low Risk for Depression: A Three-

Generation Startle Study. Biological Psychiatry, 57(9), 953-960.

doi:10.1016/j.biopsych.2005.01.045

Grillon, C., Dierker, L., & Merikangas, K. R. (1998). Fear-potentiated startle in adolescent

offspring of parents with anxiety disorders. Biological Psychiatry, 44(10), 990-997.

doi:10.1016/S0006-3223(98)00188-7

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

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

Social Psychology, 74, 224–237. doi:10.1037//0022-3514.74.1.224

Gross, J. J., & Jazaieri, H. (2014). Emotion, emotion regulation, and psychopathology: An

affective science perspective. Clinical Psychological Science, 2(4), 387-401.

doi:10.1177/2167702614536164

Gross, J.J., & Levenson, R.W. (1995). Emotion elicitation using films. Cognition and Emotion,

9(1), 87-108. doi: 10.1080/02699939508408966

Gruber, J., Dutra, S., Eidelman, P., Johnson, S. L., & Harvey, A. G. (2011). Emotional and

physiological responses to normative and idiographic positive stimuli in bipolar

disorder. Journal Of Affective Disorders, 133(3), 437-442. doi:10.1016/j.jad.2011.04.045

88
Hadzi-Pavlovic, D. (2009). Finding patterns and groupings: I Introduction to latent class

analysis. Acta Neuropsychiatrica, 21(6), 312-313. doi:10.1111/j.1601-5215.2009.00429.x

Hallion, L. S., Tolin, D. F., Assaf, M., Goethe, J., & Diefenbach, G. J. (2017). Cognitive control

in generalized anxiety disorder: Relation of inhibition impairments to worry and anxiety

severity. Cognitive Therapy And Research, 41(4), 610-618. doi:10.1007/s10608-017-

9832-2

Hofmann, W., Schmeichel, B. J., & Baddeley, A. D. (2012). Executive functions and self-

regulation. Trends in cognitive sciences, 16(3), 174-180.

Homack, S., & Riccio, C. A. (2004). A meta-analysis of the sensitivity and specificity of the

Stroop Color and Word Test with children. Archives of Clinical Neuropsychology, 19(6),

725-743.

Huttenlocher, J., Haight, W., Bryk, A., Seltzer, M., & Lyons, T. (1991). Early vocabulary

growth: Relation to language input and gender. Developmental Psychology, 27(2), 236-

248. doi:10.1037/0012-1649.27.2.236

Jackson, K. M., & Sher, K. J. (2005). Similarities and Differences of Longitudinal Phenotypes

Across Alternate Indices of Alcohol Involvement: A Methodologic Comparison of

Trajectory Approaches. Psychology Of Addictive Behaviors, 19(4), 339-351.

doi:10.1037/0893-164X.19.4.339

Jacobs, S.C., Friedman, R., Parker, J.D., Tofler, G.H., Jimenez, A.H., Muller, J.E., Benson, H.,

& Stone, P.H. (1994). Use of skin conductance changes during mental stress testing as an

index of autonomic arousal in cardiovascular research. American Heart Journal, 128(6),

1170-1177.

Jovanovic, T., Norrholm, S. D., Blanding, N. Q., Davis, M., Duncan, E., Bradley, B., & Ressler,

89
 K. J. (2010). Impaired fear inhibition is a biomarker of PTSD but not

depression. Depression And Anxiety, 27(3), 244-251. doi:10.1002/da.20663

Jovanovic, T., & Ressler, K. J. (2010). How the neurocircuitry and genetics of fear inhibition

may inform our understanding of PTSD. The American Journal of Psychiatry, 167(6),

648–662. https://doi.org/10.1176/appi.ajp.2009.09071074

Jung, T., & Wickrama, K. S. (2008). An introduction to latent class growth analysis and growth

mixture modeling. Social And Personality Psychology Compass, 2(1), 302-317.

doi:10.1111/j.1751-9004.2007.00054.x

Kane, M. J., & Engle, R. W. (2003). Working-memory capacity and the control of attention: The

contributions of goal neglect, response competition, and task set to Stroop

interference. Journal of Experimental Psychology: General, 132(1), 47–70.

https://doi.org/10.1037/0096-3445.132.1.47

Kashdan, T. B., & Farmer, A. S. (2014). Differentiating emotions across contexts: Comparing

adults with and without social anxiety disorder using random, social interaction, and daily

experience sampling. Emotion, 14(3), 629-638. doi:10.1037/a0035796

Kashdan, T. B., & Rottenberg, J. (2010). Psychological flexibility as a fundamental aspect of

health. Clinical Psychology Review, 30(7), 865–878.

https://doi.org/10.1016/j.cpr.2010.03.001

Keltner, D., & Bonanno, G. A. (1997). A study of laughter and dissociation: Distinct correlates

of laughter and smiling during bereavement. Journal Of Personality And Social

Psychology, 73(4), 687-702. doi:10.1037/0022-3514.73.4.687

Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005). Prevalence, Severity, and

Comorbidity of 12-Month DSM-IV Disorders in the National Comorbidity Survey

90
 Replication. Archives Of General Psychiatry, 62(6), 617-627.

doi:10.1001/archpsyc.62.6.617

Kessler, R. C., Alonso, J., Chatterji, S., & He, Y. (2014). Disability and costs. In P. Emmelkamp,

T. Ehring, P. Emmelkamp, T. Ehring (Eds.), The Wiley handbook of anxiety disorders,

Volume I: Theory and research; Volume II: Clinical assessment and treatment (pp. 47-

57). Wiley-Blackwell.

Khanna, M. M., Badura-Brack, A. S., McDermott, T. J., Embury, C. M., Wiesman, A. I.,

Shepherd, A., & ... Wilson, T. W. (2017). Veterans with post-traumatic stress disorder

exhibit altered emotional processing and attentional control during an emotional Stroop

task. Psychological Medicine, 47(11), 2017-2027. doi:10.1017/S0033291717000460

Klumpp, H., Angstadt, M., Nathan, P. J., & Phan, K. L. (2010). Amygdala reactivity to faces at

varying intensities of threat in generalized social phobia: An event-related functional

MRI study. Psychiatry Research: Neuroimaging, 183(2), 167-169.

doi:10.1016/j.pscychresns.2010.05.001

Krain, A. L., Gotimer, K., Hefton, S., Ernst, M., Castellanos, F. X., Pine, D. S., & Milham, M. P.

(2008). A functional magnetic resonance imaging investigation of uncertainty in

adolescents with anxiety disorders. Biological Psychiatry, 63(6), 563-568.

doi:10.1016/j.biopsych.2007.06.011

Kreibig, S. D., Wilhelm, F. H., Roth, W. T., & Gross, J. J. (2007). Cardiovascular,

electrodermal, and respiratory response patterns to fear- and sadness-inducing

films. Psychophysiology, 44(5), 787-806. doi:10.1111/j.1469-8986.2007.00550.x

Kreibig, S. D. (2010). Autonomic nervous system activity in emotion: A review. Biological

Psychology, 84(3), 394–421. doi: 10.1016/j.biopsycho.2010.03.010

91
Kriendler, D.M., & Lumsden, C.J. (2006). Effects of the irregular sample and mission data in

time series analysis. Nonlinear Dynamics Psychology and Life Sciences, 10(2), 187-214.

Latsko, M. S., Gilman, T. L., Matt, L. M., Nylocks, K. M., Coifman, K. G., & Jasnow, A. M.

(2016). A novel interaction between tryptophan hydroxylase 2 (TPH2) gene

polymorphism (rs4570625) and BDNF Val⁶⁶Met predicts a high-risk emotional

phenotype in healthy subjects. Plos ONE, 11(10),

LeDoux, J. (2012). A neuroscientist’s perspective on debates about the nature of emotion.

Emotion Review, 4(4), 375-379.

Lewinsohn, P.M., Seeley, J.R., Roberts, R.E., & Allen, N.B. (1997). Center for Epidemiologic

Studies Depression Scale (CES-D) as a screening instrument for depression among

community-residing older adults. Psychology and Aging, 12, 277-287. doi:10.1037/0882-

7974.12.2.277

Lissek, S., Powers, A. S., McClure, E. B., Phelps, E. A., Woldehawariat, G., Grillon, C., & Pine,

D. S. (2005). Classical fear conditioning in the anxiety disorders: A meta-

analysis. Behaviour Research And Therapy, 43(11), 1391-1424.

doi:10.1016/j.brat.2004.10.007

Lissek, S., Rabin, S. J., McDowell, D. J., Dvir, S., Bradford, D. E., Geraci, M., & ... Grillon, C.

(2009). Impaired discriminative fear-conditioning resulting from elevated fear responding

to learned safety cues among individuals with panic disorder. Behaviour Research And

Therapy, 47(2), 111-118. doi:10.1016/j.brat.2008.10.017

Loerinc, A. G., Meuret, A. E., Twohig, M. P., Rosenfield, D., Bluett, E. J., & Craske, M. G.

(2015). Response rates for CBT for anxiety disorders: Need for standardized

criteria. Clinical Psychology Review, 4272-82. doi:10.1016/j.cpr.2015.08.004

92
Low, C. A., Stanton, A. L., & Bower, J. E. (2008). Effects of acceptance-oriented versus

evaluative emotional processing on heart rate recovery and habituation. Emotion, 8(3),

419-424. doi:10.1037/1528-3542.8.3.419

Mackintosh, N. J. (1987). Neurobiology, psychology and habituation. Behaviour Research And

Therapy, 25(2), 81-97. doi:10.1016/0005-7967(87)90079-9

Mathews, A., & MacLeod, C. (1994). Cognitive approaches to emotion and emotional

disorders. Annual Review Of Psychology, 4525-50.

doi:10.1146/annurev.ps.45.020194.000325

Matt, L. M., Fresco, D. M., & Coifman, K. G. (2016). Trait anxiety and attenuated negative

affect differentiation: a vulnerability factor to consider? Anxiety, Stress, & Coping, 1-14.

McClure, E. B., Monk, C. S., Nelson, E. E., Parrish, J. M., Adler, A., Blair, R. R., & ... Pine, D.

S. (2007). Abnormal attention modulation of fear circuit function in pediatric generalized

anxiety disorder. Archives Of General Psychiatry, 64(1), 97-106.

doi:10.1001/archpsyc.64.1.97

McKenna, F. P., & Sharma, D. (2004). Reversing the Emotional Stroop Effect Reveals That It Is

Not What It Seems: The Role of Fast and Slow Components. Journal Of Experimental

Psychology: Learning, Memory, And Cognition, 30(2), 382-392. doi:10.1037/0278-

7393.30.2.382

Mineka, S. (2013). Individual differences in the acquisition of fears. In D. Hermans, B. Rimé, B.

Mesquita, D. Hermans, B. Rimé, B. Mesquita (Eds.), Changing emotions (pp. 47-52).

New York, NY, US: Psychology Press.

Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., & Howerter, A. (2000). The unity

and diversity of executive functions and their contributions to complex 'frontal lobe'

93
 tasks: A latent variable analysis. Cognitive Psychology, 41(1), 49-100.

doi:10.1006/cogp.1999.0734

Monk, C. S., Telzer, E. H., Mogg, K., Bradley, B. P., Mai, X., Louro, H. C., & ... Pine, D. S.

(2008). Amygdala and ventrolateral prefrontal cortex activation to masked angry faces in

children and adolescents with generalized anxiety disorder. Archives Of General

Psychiatry, 65(5), 568-576. doi:10.1001/archpsyc.65.5.568

Morrison, F. G., & Ressler, K. J. (2014). From the neurobiology of extinction to improved

clinical treatments. Depression And Anxiety, 31(4), 279-290. doi:10.1002/da.22214

Muthén, B.O., & Curran, P.J. (1997). General longitudinal modeling of individual differences

in experimental designs: A latent variable framework for analysis and power

estimation. Psychological Methods, 2(4), 371-402. doi:10.1037/1082-989X.2.4.371

Muthén, L. K., & Muthén, B. (2006). Mplus user’s guide (Version 4). Los Angeles, CA: Muthén

& Muthén.

Muthén, B. (2004). Latent variable analysis: Growth mixture modeling and related techniques

for longitudinal data. In D. Kaplan (Ed.), Handbook of quantitative methodology for the

social sciences (pp. 345–368). Newbury Park, CA: Sage Publications.

Norrholm, S. D., Glover, E. M., Stevens, J. S., Fani, N., Galatzer-Levy, I. R., Bradley, B., & ...

Jovanovic, T. (2015). Fear load: The psychophysiological over-expression of fear as an

intermediate phenotype associated with trauma reactions. International Journal Of

Psychophysiology, 98(2, Part 2), 270-275. doi:10.1016/j.ijpsycho.2014.11.005

Norrholm, S. D., Jovanovic, T., Olin, I. W., Sands, L. A., Karapanou, I., Bradley, B., & Ressler,

K. J. (2011). Fear extinction in traumatized civilians with posttraumatic stress disorder:

Relation to symptom severity. Biological Psychiatry, 69(6), 556-563.

94
 doi:10.1016/j.biopsych.2010.09.013

Nylocks, K. M., Gilman, T. L., Latsko, M. S., Jasnow, A. M., & Coifman, K. G. (2018).

Increased parasympathetic activity and ability to generate positive emotion: The

influence of the bdnf val66met polymorphism on emotion flexibility. Motivation And

Emotion, doi:10.1007/s11031-018-9679-1

Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in

latent class analysis and growth mixture modeling: A Monte Carlo simulation

study. Structural Equation Modeling, 14(4), 535-569. doi:10.1080/10705510701575396

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. doi:10.1111/j.1467-

9280.2009.02459.x

Öhman, A., & Mineka, S. (2001). Fears, phobias, and preparedness: Toward an evolved module

of fear and fear learning. Psychological Review, 108(3), 483-522. doi:10.1037/0033-

295X.108.3.483

Öhman, A. (2005). The role of the amygdala in human fear: Automatic detection of

threat. Psychoneuroendocrinology, 30(10), 953-958. doi:10.1016/j.psyneuen.2005.03.019

Okado, Y., & Haskett, M. E. (2015). Three-year trajectories of parenting behaviors among

physically abusive parents and their link to child adjustment. Child & Youth Care

Forum, 44(5), 613-633. doi:10.1007/s10566-014-9295-5

Olatunji, B. O., Cisler, J. M., & Deacon, B. J. (2010). Efficacy of cognitive behavioral therapy

for anxiety disorders: A review of meta-analytic findings. Psychiatric Clinics Of North

America, 33(3), 557-577. doi:10.1016/j.psc.2010.04.002

95
Orcutt, H. K., Bonanno, G. A., Hannan, S. M., & Miron, L. R. (2014). Prospective trajectories of

posttraumatic stress in college women following a campus mass shooting. Journal of

Traumatic Stress, 27(3), 249–256. https://doi.org/10.1002/jts.21914

Pallanti, S., Hollander, E., Bienstock, C., Koran, L., Leckman, J., Marazziti, D., & ... Zohar, J.

(2002). Treatment non-response in OCD: Methodological issues and operational

definitions. International Journal Of Neuropsychopharmacology, 5(2), 181-191.

doi:10.1017/S1461145702002900

Papa, A., & Bonanno, G. (2008). Smiling in the face of adversity: the interpersonal and

intrapersonal functions of smiling. Emotion (15283542), 8(1), 1–12. doi: 10.1037/1528-

3542.8.1.1

Pape, H. C., & Pare, D. (2010). Plastic synaptic networks of the amygdala for the acquisition,

expression, and extinction of conditioned fear. Physiological Reviews, 90(2), 419-463.

Phelps, E.A., Delgado, M.R., Nearing, K.I., & LeDoux, J.E. (2004). Extinction learning in

humans: role of the amygdala and vmPFC. Neuron, 43(6), 897-905.

Powers, M.B., Halpern, J.M., Ferenschak, M.P., Gillihan, S.J., & Foa, E.B. (2010). A meta-

analytic review of prolonged exposure for posttraumatic stress disorder. Clin. Psychol.

Rev. 30(6), 635–41

Radloff, L.S. (1977). The CES-D scale a self-report depression scale for research in the general

population. Applied Psychological Measurement, 3(1), 385-401.

Rafaeli, E., Rogers, G. M., & Revelle, W. (2007). Affective synchrony: Individual differences in

mixed emotions. Personality and Social Psychology Bulletin, 33(7), 915-932.

Reeb-Sutherland, B. C., Vanderwert, R. E., Degnan, K. A., Marshall, P. J., Pérez-Edgar, K.,

Chronis-Tuscano, A., & ... Fox, N. A. (2009). Attention to novelty in behaviorally

96
 inhibited adolescents moderates risk for anxiety. Journal Of Child Psychology And

Psychiatry, 50(11), 1365-1372. doi:10.1111/j.1469-7610.2009.02170.x

Resick, P. A., Bovin, M. J., Calloway, A. L., Dick, A. M., King, M. W., Mitchell, K. S., & ...

Wolf, E. J. (2012). A critical evaluation of the complex PTSD literature: Implications for

DSM-5. Journal of Traumatic Stress, 25, 239–249. doi:10.1002/jts.21699

Rottenberg, J. (2017). Emotions in depression: What do we really know? Annual Review of

Clinical Psychology, 13, 241–263. https://doi.org/10.1146/annurev-clinpsy-032816-

045252

Roth, P.L. (1994). Missing data: A conceptual review for applied psychologists. Personnel

Psychology, 47, 537-560.

Roth, P. L., Switzer, F. I., & Switzer, D. M. (1999). Missing data in multiple item scales: A

Monte Carlo analysis of missing data techniques. Organizational Research

Methods, 2(3), 211-232. doi:10.1177/109442819923001

Rosebrock, L. E., Hoxha, D., Norris, C., Cacioppo, J. T., & Gollan, J. K. (2017). Skin

conductance and subjective arousal in anxiety, depression, and comorbidity: Implications

for affective reactivity. Journal of Psychophysiology, 31(4), 145–157.

https://doi.org/10.1027/0269-8803/a000176

Russell, J. A. (1980). A circumplex model of affect. Journal of personality and social

psychology, 39(6), 1161.

Sacks, T. L., Clark, C. R., Pols, R. G., & Geffen, L. B. (1991). Comparability and stability of

performance of six alternate forms of the dodrill-stroop colour-word test. The Clinical

Neuropsychologist, 5(3), 220-225.

Sarapas, C., Weinberg, A., Langenecker, S. A., & Shankman, S. A. (2017). Relationships among

97
 attention networks and physiological responding to threat. Brain and Cognition, 111, 63–

72. https://doi.org/10.1016/j.bandc.2016.09.012

Schmeichel, B. J., & Tang, D. (2015). Individual differences in executive functioning and their

relationship to emotional processes and responses. Current Directions In Psychological

Science, 24(2), 93-98. doi:10.1177/0963721414555178

Schwarz, G. (1978). Estimating the dimension of a model. The annals of statistics, 6(2), 461-

464. doi:10.1214/aos/1176344136

Seligman, M. E. (1971). Phobias and preparedness. Behavior Therapy, 2(3), 307-320.

doi:10.1016/S0005-7894(71)80064-3

Shin, L. M., & Liberzon, I. (2010). The neurocircuitry of fear, stress, and anxiety

disorders. Neuropsychopharmacology, 35(1), 169-191. doi:10.1038/npp.2009.83

Sotres-Bayon, F., Bush, D. A., & LeDoux, J. E. (2004). Emotional perseveration: An update on

prefrontal-amygdala interactions in fear extinction. Learning & Memory, 11(5), 525-535.

doi:10.1101/lm.79504

Strauss, G. P., Allen, D. N., Jorgensen, M. L., & Cramer, S. L. (2005). Test-Retest Reliability of

Standard and Emotional Stroop Tasks: An Investigation of Color-Word and Picture-Word

Versions. Assessment, 12(3), 330-337. doi:10.1177/1073191105276375

Stroop, J.R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental

Psychology, 18(6), 643-662

Thompson, R. J., Kuppens, P., Mata, J., Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Gotlib, I.

H. (2015). Emotional Clarity as a Function of Neuroticism and Major Depressive

Disorder. Emotion, 15(5), 615-624.

Todd, D. M., Deane, F. P., & McKenna, P. A. (1997). Appropriateness of SCL-90-R adolescent

98
 and adult norms for outpatient and nonpatient college students. Journal Of Counseling

Psychology, 44(3), 294-301. doi:10.1037/0022-0167.44.3.294

Tomko, R. L., Lane, S. P., Pronove, L. M., Treloar, H. R., Brown, W. C., Solhan, M. B., ... &

Trull, T. J. (2015). Undifferentiated negative affect and impulsivity in borderline

personality and depressive disorders: A momentary perspective. Journal of abnormal

psychology, 124(3), 740-753.

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), 1161–90.

Tugade, M. M., & Fredrickson, B.J. (2007). Regulation of positive emotions: Emotion regulation

strategies that promote resilience. Journal of Happiness Studies, 8, 311-333.

van den Heuvel, O. A., Veltman, D. J., Groenewegen, H. J., Witter, M. P., Merkelbach, J., Cath,

D. C., & ... van Dyck, R. (2005). Disorder-Specific Neuroanatomical Correlates of

Attentional Bias in Obsessive-compulsive Disorder, Panic Disorder, and

Hypochondriasis. Archives Of General Psychiatry, 62(8), 922-933.

doi:10.1001/archpsyc.62.8.922

Vervliet, B., Craske, M. G., & Hermans, D. (2013). Fear extinction and relapse: State of the art.

Annual Review of Clinical Psychology, 9, 215–248. https://doi.org/10.1146/annurev-

clinpsy-050212-185542

Wagner, A. R. (1981). SOP: a model of automatic memory processing in animal behavior. In N.

E. Spear, & R. R. Miller (Eds.), Information processing in animals: Memory

mechanisms. Hillsdale, NJ: Erlbaum.

Waters, A. M., Nazarian, M., Mineka, S., Zinbarg, R. E., Griffith, J. W., Naliboff, B., & ...

99
 Craske, M. G. (2014). Context and explicit threat cue modulation of the startle reflex:

Preliminary evidence of distinctions between adolescents with principal fear disorders

versus distress disorders. Psychiatry Research, 217(1-2), 93-99.

doi:10.1016/j.psychres.2014.01.047

Waters, A. M., Neumann, D. L., Henry, J., Craske, M. G., & Ornitz, E. M. (2008). Baseline and

affective startle modulation by angry and neutral faces in 4--8-year-old anxious and non-

anxious children. Biological Psychology, 78(1), 10-19.

doi:10.1016/j.biopsycho.2007.12.005

Williams, J. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and

psychopathology. Psychological Bulletin, 120(1), 3-24. doi:10.1037/0033-2909.120.1.3

Zaki, L. F., Coifman, K. G., Rafaeli, E., Berenson, K. R., & Downey, G. (2013). Emotion

differentiation as a protective factor against nonsuicidal self-injury in borderline

personality disorder. Behavior Therapy, 44(3), 529–40.

100
 Appendix A.

Fear reactivity task development and stimulus pilot testing

In order to develop the fear reactivity task for the current investigation, a pilot study was

first conducted to assess the efficacy of the video stimuli. For the pilot study, four fear eliciting

video clips were selected from publically available clips. In the four selected videos, the

following content was depicted: 1) a doctor is examining an infected wound on the back of a

young woman. At the end of the clip, spiders crawl out of the wound and the young woman

screams (Fear Clinic, Anchor Bay Entertainment, 2014), 2) a woman wakes up to find a demon-

like character next to her in bed (Bedfellows, www.youtube.com), 3) in a dark house, arms and

hands extend through the walls and ceiling forming a monster or demon-like creature (Real

Demons, www.youtube.com), 4) a large spider suddenly jumps down from the ceiling on a man

standing on a ladder (Spider Attacks Daddy, www.youtube.com). Four positive emotion video

clips, and one neutral video clip, were also selected from previous emotion elicitation research

(Gilman et al., 2017) and from publically available clips. In the selected positive emotion videos,

the following content was depicted: 1) a fire drill exercise goes awry in the office of a small

Pennsylvania company (The Office, ABC), 2) comedian Zach Galifanakis interviews actress

Charlize Theron on an internet “talk show” (Between Two Ferns, www.comedyordie.com), 3)

funny end credits and bloopers from a comedy movie starring Will Ferrell and John Reilly

(Talladega Nights Bloopers, Sony Pictures, 2006), 4) famous celebrities read mean tweets about

themselves out loud on the Jimmy Kimmel show (Celebrities Read Mean Tweets #5, ABC). The

neutral video clip consisted of nature scenes from Alaska, shown along with audio information

about tourism in Alaska and Denali (Alaska Travel Guide, www.videosource.com).

101
 All selected video clips were evaluated as part of an online pilot study, consistent with

well-established methodology for choosing video stimuli (Gross & Levenson, 1995). A sample

of n=318 college students was recruited from the Psychology department subject pool for the

pilot study. Data was dropped from a total of n=47 participants due to non-completion of the

study (n=3) and poor engagement with the video content (n=44) (detailed below) thus, the final

pilot sample consisted of n=271 participants (78.6% female, 85.2% Caucasian, age: M=19.58,

SD=1.73 years). After giving informed consent, participants were asked to provide demographic

information, and then they viewed the above detailed video clips (presented in random order).

Following each video clip, participants were asked to report on their emotional experience by

rating emotion-words on a 7-point Likert scale. Emotion-words were of both negative (fear,

sadness, disgust, guilt, distress, anger) and positive (happiness, enjoyment, amusement, affection,

relief) valence. These emotion-words demonstrated sufficient internal consistency in this sample

(negative emotion experience in response to fear context a=.90, and positive emotion experience

in response to positive emotion context; a=.92). Following each video clip, participants were

also asked to answer one question regarding the content of the video clip, in order to ensure

engagement with the video. For example, following the “Bedfellows” video clip, participants

were asked “In what room did most of the action in this clip take place?” (correct answer: “the

bedroom”). Participants who did not answer the questions correctly were dropped from the study

(n=44). Participants received course credit upon completion of the study.

Results. In order to test the efficacy of the selected video clips in eliciting the target

emotion (fear), responses were analyzed following the example of Gross and Levenson (1995)

(c.f. Gilman et al., 2017). A “success index” was established for each video clip by utilizing the

standardized mean intensity of the clip for the target emotion (mean rating of fear for each fear

102
video) and a discreteness value (percentage of participants who rated the target emotion, fear, at

least one point above all non-target emotions). Mean intensity scores and mean discreteness

scores were then standardized as z-scores and the sum of these two z-scores represents the

success index of each video clip, relative to other video clips assessed for the same target

emotion (see Table 1). Video clips with the highest success rate have been suggested for use

(Gross & Levenson, 1995).

103
Table 1. Pilot study success indices for task stimuli

Video Emotion Mean intensity Discreteness Success

Index

Bedfellows Fear 4.91(1.93) 22.89% 2.42

Big Spider Fear 4.36(2.12) 9.23% -.35

Fear Clinic Fear 3.94(1.99) 1.85% -2.14

Real Demons Fear 4.14(2.02) 18.08% .07

Between Two Ferns Amusement 4.16()1.94 15.50% .05

The Office Amusement 4.65(1.94) 11.44% -.56

Talladega Nights Amusement 4.32(1.95) 15.13% .09

Celebrity Tweets Amusement 5.48(1.62) 12.55% .42

104
 Appendix B.

Emotion response coherence

a.
Fear Reported Skin
First fear video facial negative conductance
expression emotion
experience

Fear facial expression 1 -- --

Reported negative .43** 1 --

emotion experience

Skin conductance .08 -.04 1

**=p<.01

b.
Fear Reported Skin
Second fear video facial negative conductance
expression emotion
experience

Fear facial expression 1 -- --

Reported negative .43** 1 --

emotion experience

Skin conductance .00 .08 1

**=p<.01

105
c.
First fear recovery Fear facial Reported Skin
period expression negative conductance
emotion
experience

Fear facial expression 1 -- --

Reported negative .56** 1 --

emotion experience

Skin conductance .04 .09 1

d.

Fear facial Reported Skin

Second fear recovery expression negative conductance
period emotion
experience
Fear facial expression 1 -- --

Reported negative .32** 1 --

emotion experience

Skin conductance -.10 .21 1

*=p<.05
**=p<.01

106
e.
First positive emotion video Positive Reported Skin
emotion facial positive emotion conductance
expression experience

Positive emotion facial 1 -- --

expression

Reported positive emotion .40** 1 --

experience

Skin conductance -.02 .17 1

**=p<.01

f.
Positive Reported Skin
Second positive emotion video emotion facial positive emotion conductance
expression experience

Positive emotion facial 1 -- --

expression

Reported positive emotion .47** 1 --

experience

Skin conductance -.04 .06 1

**=p<.01

107