Neurobiological Effects of Context on Emotion Regulation
The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters
Citation Shermohammed, Maheen. 2019. Neurobiological Effects of Context on Emotion Regulation. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:42029802
Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA Neurobiological Effects of Context on Emotion Regulation
A dissertation presented
by
Maheen Shermohammed
to
The Department of Psychology
in partial fulfillment of the requirements
for the degree of
Doctor of Philosophy
in the subject of
Psychology
Harvard University
Cambridge, Massachusetts
April 2019
© 2019 Maheen Shermohammed
All rights reserved.
Dissertation Advisor: Dr. Leah Somerville Maheen Shermohammed
Neurobiological Effects of Context on Emotion Regulation
Abstract
The ability to regulate emotions is a critical feature of healthy psychological functioning.
It is therefore essential to understand under what conditions different emotion regulation strategies may or may not be effective. Neurobiological evidence suggests that certain contexts, including acute psychological stress and sleep deprivation, may impair emotion regulation ability. However, there is little data on the causal effects of these contexts on emotion regulation processing. In this dissertation, I present three studies that examine the neurobiological effects of stress and sleep deprivation on two emotion regulation strategies: cognitive reappraisal and affect labeling. In Paper 1, we induced acute psychosocial stress and measured its impact (relative to a control manipulation) on emotional responding during a cognitive reappraisal task while participants underwent fMRI. Findings revealed no evidence that stress modulated the effects of cognitive reappraisal on subjective or physiological measures of emotional responding. Modest effects of stress on reappraisal-related neural activation were observed in the prefrontal cortex and amygdala, but these relationships were statistically fragile. These findings were extended in
Paper 2, in which we tested the effects of one night of total sleep deprivation on the same cognitive reappraisal task. Once again, results showed no evidence that the context manipulation
(this time sleep deprivation) affected subjective, physiological, or neural responses to cognitive reappraisal. However, it was not the case that these null effects generalized to all types of
iii emotion regulation. In Paper 3, we examined the effects of sleep deprivation on affect labeling, an implicit emotion regulation strategy hypothesized to rely on the cognitive control functions of the right ventrolateral prefrontal cortex. Findings revealed up-regulated recruitment of this prefrontal region as well as increased functional connectivity with the amygdala during sleep deprivation. Increased coupling was associated with lower baseline negative affect when sleep deprived, suggesting that sleep-loss-induced increases in activation may have adaptive buffering effects on mood.
Together, findings from these papers show that two context manipulations expected to impair emotion regulation ability do not appear to impact cognitive reappraisal, despite influencing the processing of a more implicit emotion regulation strategy. This work calls for a careful examination of the way emotion regulation is presently studied and whether cognitive reappraisal, or only certain aspects of it, may in fact be robust to the effects of contextual factors like stress and sleep deprivation.
iv
Table of Contents
Background and Introduction ...... 1
Paper 1: Does psychosocial stress impact cognitive reappraisal? Behavioral and neural evidence ...... 12
Paper 2: Neurobiological effects of sleep deprivation on cognitive reappraisal ...... 44
Paper 3: Neurobiological effects of sleep deprivation on affect labeling ...... 68
General Discussion and Conclusion ...... 89
References ...... 97
Appendix ...... 112
v Acknowledgments
I would first like to extend my deepest gratitude to my advisor, Dr. Leah Somerville, for creating an environment where I could grow fearlessly as a scientist and providing nothing but support along the way. I would also like to thank the members of my dissertation committee:
Drs. Liz Phelps and Kate McLaughlin for their time and extremely helpful discussion, and Dr.
Randy Buckner, for his infectious curiosity and for showing me how discoveries are made.
I am very grateful to the members of the Affective Neuroscience and Developmental Lab, who are each brilliant scientists and even better people. I would especially like to thank Hayley
Dorfman, Katie Insel, and Alex Rodman for being my safety net; their support and friendship has meant the world to me.
Thank you to the funding sources that generously supported this research, including the
National Science Foundation, the National Institutes of Health, and the U.S. Army Natick
Soldier Research, Development, and Engineering Center.
Finally, my loving thanks to my family, who taught me that knowledge was something to be held in the highest regard, and to Jack, for being my best friend and biggest fan.
vi Background and Introduction
Emotions can play a highly adaptive role, helping us form attachments, avoid threats, communicate our thoughts and intentions, and generally respond quickly to situations in our environment (Ekman, 1992; Frijda, 1986; Keltner & Gross, 1999; Oatley & Jenkins, 2003).
However, sometimes emotional responses in the moment do not align with our higher order goals, like giggling during a serious lecture or yelling at a frustrating coworker, and in such situations we may attempt to modify or regulate them. Understanding emotion regulation is critical, as it plays a key role in healthy psychological functioning. For example, successful emotion regulation can confer beneficial academic and social outcomes in daily life. In students, a greater capacity to regulate an emotional response is associated with higher GPA, standardized achievement scores, and teacher reports of academic achievement and productivity, above and beyond the effects of IQ (Graziano, Reavis, Keane, & Calkins, 2007; Gumora & Arsenio, 2002).
Adaptive regulation is also associated with higher competence in peer relationships, improved quality of social relationships, and greater prosocial tendencies (Contreras, Kerns, Weimer,
Gentzler, & Tomich, 2000; Lopes, Salovey, Coté, & Beers, 2005).
Meanwhile, maladaptive emotion regulation tendencies are frequently associated with psychopathology. Individuals with depression tend to use less effective and often even counter- productive strategies to regulate their emotions (Berking & Wupperman, 2012; Campbell-Sills,
Barlow, Brown, & Hofmann, 2006; Ehring, Tuschen-Caffier, Schnülle, Fischer, & Gross, 2010;
Liverant, Brown, Barlow, & Roemer, 2008), and this profile of aberrant emotion regulation has been linked to deficits in inhibiting the processing of negative material (Joormann & Gotlib,
2010). Anxiety disorders such as generalized anxiety disorder and post-traumatic stress disorder
(PTSD) have similarly been linked to deficits in the use or efficacy of adaptive emotion
1 regulation strategies (Tull, Barrett, McMillan, & Roemer, 2007; Tull, Stipelman, Salters-
Pedneault, & Gratz, 2009). Longitudinal studies of individuals with depression and anxiety show that maladaptive emotion regulation attitudes and tendencies predict disease prognosis (Kassel,
Bornovalova, & Mehta, 2007; Kraaij, Pruymboom, & Garnefski, 2002), and that interventions targeting these factors can improve health outcomes (Cloitre, Koenen, Cohen, & Han, 2002;
Kumar, Feldman, & Hayes, 2008). Together, this work suggests that a better understanding of emotion regulation can shed light on healthy emotional functioning and reveal cognitive processes that may contribute to symptoms of psychopathology.
Emotion regulation strategies
To date, a great deal of research has been dedicated to investigating how emotion regulation can be implemented adaptively, examining what types of emotion regulation strategies are most effective. One emotion regulation strategy that has received immense empirical attention is cognitive reappraisal (CR), which involves re-interpreting the content of an emotional stimulus in a way that changes its meaning. For example, someone who is frustrated at getting lost could reappraise the situation as an opportunity to enjoy the scenery in a new part of town. CR is considered a highly effective method of reducing negative affect (i.e. changes in subjective feeling states). When compared to passively experiencing affect or attempting to suppress it, CR is more effective at decreasing negative and increasing positive affect (Gross,
1998) and modulating biological components of the emotional response, such as sympathetic nervous system activity and amygdala activation (Goldin, McRae, Ramel, & Gross, 2008; Gross,
1998).
2 CR has also been particularly well studied in the brain and, like many other emotion regulation strategies, is believed to rely on neural systems implicated in other forms of cognitive control. A meta-analysis of 48 neuroimaging studies has suggested that the most reliable among these are the dorsolateral prefrontal cortex (dlPFC), posterior parietal cortex, ventrolateral prefrontal cortex (vlPFC), posterior dorsomedial prefrontal cortex (dmPFC), dorsal anterior cingulate cortex (dACC), and left posterior temporal cortex (Figure 1; Buhle et al., 2014). The dlPFC and posterior parietal cortex are thought to be important for directing attention to goal- relevant features of stimuli and holding in mind generated reappraisals (Silvers, Buhle, &
Ochsner, 2013). The vlPFC is believed to be involved in choosing appropriate reappraisals, in light of its putative role in the controlled access and selection of responses from semantic memory that are congruent with one’s goals (Badre & Wagner, 2007), and to inhibit emotional information that interferes with goal-oriented behavior or processing (Hooker & Knight, 2006).
Finally, the dmPFC and dACC may be involved in their capacity as regions that monitor conflict and performance (Botvinick, Braver, Barch, Carter, & Cohen, 2001), perhaps tracking the success of current reappraisals and recruiting increased controlled processing when necessary.
Buhle & colleagues’ (2014) meta-analysis yielded only one region that is consistently down regulated by reappraisal: the amygdala. This finding falls in line with a rich literature on the amygdala’s role in facilitating attention and perception for emotional, arousing stimuli
(Phelps & LeDoux, 2005). However, the mechanisms by which cognitive control systems modulate the amygdala during CR remain under debate. Some researchers have proposed that prefrontal and parietal regions involved in CR may exert their control on affective systems indirectly by modulating lateral temporal regions supporting semantic and perceptual representations (Buhle et al., 2014; Ochsner, Silvers, & Buhle, 2012; Silvers et al., 2013).
3 Others have instead pointed to the possible mediating role of the ventromedial prefrontal cortex, which has been implicated in conditioned fear extinction and which has direct anatomical projections to the amygdala (Adhikari et al., 2015; Phelps, Delgado, Nearing, & Ledoux, 2004;
Quirk, Likhtik, Pelletier, & Pare, 2003).
Figure 1. Left-hemisphere results from a meta-analysis of 48 neuroimaging studies examining cognitive reappraisal compared to an emotional baseline condition (Buhle et al., 2014).
CR can be categorized as an emotion regulation strategy that is explicit; it is conscious, deliberate, and often effortful (Gyurak, Gross, & Etkin, 2011). More recently, there has been a growing interest in emotion regulation strategies that are implicit. Implicit strategies are those that do not involve a conscious goal or desire to change one’s emotions, and thus tend to be more efficient and require less effort (Gyurak et al., 2011). One implicit form of emotion regulation is affect labeling, which is the act of translating feelings or affective stimuli into words. This is considered a form of emotion regulation because simply assigning an emotion label or name to an affective stimulus can reduce its psychological and physiological impact (Constantinou, Van
Den Houte, Bogaerts, Van Diest, & Van den Bergh, 2014; Lieberman, Inagaki, Tabibnia, &
Crockett, 2011; Niles, Craske, Lieberman, & Hur, 2015; Tabibnia, Lieberman, & Craske, 2008).
4 Although this may not seem like a traditional form of emotion regulation, it is demonstrated to be similarly effective at reducing self-reported and physiological emotional responses as more explicit strategies like CR (Kircanski, Lieberman, & Craske, 2012; Lieberman et al., 2011).
Furthermore, similar to CR, affect labeling has been shown to reduce activation in the amygdala
(Burklund, Craske, Taylor, & Lieberman, 2015; Costafreda, Brammer, David, & Fu, 2008;
Lieberman et al., 2007; S. E. Taylor, Eisenberger, Saxbe, Lehman, & Lieberman, 2006) and increase activation in the right vlPFC, a brain region implicated in top-down inhibitory control
(Aron, Robbins, & Poldrack, 2004). This dissertation will examine both affect labeling and CR, as it is useful to consider both explicit and implicit strategies when drawing broader conclusions about emotion regulation and its neural correlates (Berkman & Lieberman, 2009).
The role of context
Despite a rich empirical history of characterizing and comparing different emotion regulation strategies, less attention has been paid to the circumstances under which these regulation strategies are more or less effective. A medical resident regulating their emotions to angering news amid an intense shift may do so in a less effective or qualitatively different manner than when responding to the same provocation while on vacation. Although there has been a recent call for the incorporation of situational factors in the study of emotion regulation
(Aldao, 2013), the factors that have been considered have been limited to variations in the components that compose the actual emotion regulation scenario. These include elements such as the controllability of the situation eliciting the emotion (Cheng, 2001), the intensity of the emotional antecedent (Sheppes, Scheibe, Suri, & Gross, 2011), the type of emotion being regulated (Katzir & Eyal, 2013), and the desired regulatory goals (Tamir, 2009).
5 Nonetheless, there exist important contextual factors that influence the nature and effectiveness of an emotion regulation response but which are not related to what is actually being regulated. One such contextual variable is the psychological state with which an agent enters an emotion regulation scenario. Before even encountering an emotion-eliciting event or stimulus, the psychological state of an individual provides a mental context that could predispose him or her towards certain emotion regulation goals or strategies, and could facilitate their success or failure. The goal of this dissertation is to examine two such contextual factors: stress and sleep levels.
Stress and emotion regulation
Acute stress is a strain on homeostasis that results in a multifaceted response at subjective psychological, endocrine, and physiological levels (McEwen, 2000). Previous work has suggested that acute stress is a psychological context in which an emotion regulation strategy like CR may be less effective. Greater perceived stress or stress-related symptoms are related to decreased tendencies to use CR (Boden, Bonn-Miller, Kashdan, Alvarez, & Gross, 2012;
Miklósi, Martos, Szabó, Kocsis-Bogár, & Forintos, 2014; Moore, Zoellner, & Mollenholt, 2008), and using CR to down-regulate negative responses to laboratory stressors can paradoxically increase levels of cortisol, the primary stress response hormone in humans (Denson, Creswell,
Terides, & Blundell, 2014a). Participants experiencing elevated stress or cortisol levels generate fewer non-negative reappraisals, are less successful at regulating conditioned fear responses, and report associating a greater number and intensity of fear-related emotions with those conditioned stimuli (Raio, Orederu, Palazzolo, Shurick, & Phelps, 2013; Tsumura, Sensaki, & Shimada,
2015).
6 Neurobiological models suggest that stress may modulate neural systems implicated in emotion regulation, such as those involved in CR. First, stress is believed to cause broad impairment in prefrontal cortex (PFC) functioning. Acute stressors result in increased release of the neurotransmitters dopamine and norepinephrine in the PFC (Finlay, Zigmond, &
Abercrombie, 1995; Roth, Tam, Ida, Yang, & Deutch, 1988). While such catecholamine projections to the PFC are sparse (Lewis, Campbell, Foote, Goldstein, & Morrison, 1987), they play a critical tuning role in prefrontal signaling. However, at the excessive levels characteristic of stress-induced catecholamine release, this tuning is actually abolished and results in impaired
PFC functioning (Arnsten, 2009). Glucocorticoids may interact with these signaling pathways to bolster such impairments: they are capable of increasing excitatory signaling of midbrain dopamine neurons (Saal, Dong, Bonci, & Malenka, 2003) and situated to block catecholamine reuptake via receptors on the catecholamine transporter (Gründemann, Schechinger, Rappold, &
Schömig, 1998). This disordered signaling results in stress-induced impairment of executive functions relying on the PFC, including deficits in working memory, attention, and cognitive flexibility (Alexander, Hillier, Smith, Tivarus, & Beversdorf, 2007; Luethi, Meier, & Sandi,
2009; Olver, Pinney, Maruff, & Norman, 2015).
As reviewed in previous sections, CR relies on these domain-general cognitive control systems to engage in regulatory control. Meanwhile, stress is believed to enhance the very affective systems that are targeted during emotion regulation. Noradrenaline and glucocorticoids bolster amygdala functioning (Arnsten, 2009; Hatfield & McGaugh, 1999; Roozendaal,
McEwen, & Chattarji, 2009), which is reflected in the enhancement of phenomena such as classical fear conditioning and memory for emotional pictures under acute stress (Buchanan &
Lovallo, 2001; Luethi et al., 2009; Payne et al., 2007). Despite this mechanistic overlap between
7 stress and CR, the effects of stress on the subjective emotional experience following CR, and the neurobiological correlates of any such interactions, have yet to be demonstrated.
Sleep and emotion regulation
Lack of sleep is a prevalent and growing condition in modern society. Almost half of
Americans report experiencing non-refreshing sleep a few nights a week or more, and an estimated 50-70 million are affected by chronic sleep loss and sleep disorders (Institute of
Medicine, 2006; Swanson et al., 2011). Meanwhile, average sleep times are declining, with a drop of 1.5-2 hours per night for American adults and adolescents over the last 50 years (Van
Cauter, Spiegel, Tasali, & Leproult, 2008).
This pervasive sleeplessness may be exacting a grave psychological toll - insomnia is a frequent symptom in most psychiatric disorders (M. T. Smith, Huang, & Manber, 2005).
Affective disorders in particular are commonly associated with sleep disturbances. Those who suffer from generalized anxiety or panic disorder have difficulty initiating or maintaining sleep
(Mellman & Uhde, 1989; M. T. Smith et al., 2005). Patients with bulimia and anorexia nervosa also report poor sleep quality and a greater number of insomnia symptoms (Latzer, Tzischinsky,
Epstein, Klein, & Peretz, 1999; Lauer & Krieg, 2004). Individuals with insomnia are more likely to develop major depression for the first time (Baglioni, Spiegelhalder, Lombardo, & Riemann,
2010; Ford & Kamerow, 1989), and greater levels of sleep disturbance are associated with subsequent recurrence of depression symptoms in patients who were in remission (Perlis, Giles,
Buysse, Tu, & Kupfer, 1997). Sleep disruption is also a central feature of PTSD. Not only are insomnia and nightmares diagnostic symptoms of PTSD, sleep disturbance exacerbates other symptoms and leads to worse clinical outcomes (Germain, Buysse, & Nofzinger, 2008).
8 The ubiquity of sleep disturbance in affect-related psychopathology motivates a deeper exploration of the causal relationship between sleep and basic affective processing. One primary theme of such research is that sleep deprivation potentiates negative affect. Sleep loss is associated with an increase in negative moods, negative emotional responses to daytime disruptive events, and depressive and anxiety symptoms in non-clinical populations (Babson,
Trainor, Feldner, & Blumenthal, 2010; Caldwell, Caldwell, Brown, & Smith, 2004; Cutler &
Cohen, 1979; Dinges et al., 1997; Kahn-Greene, Killgore, Kamimori, Balkin, & Killgore, 2007;
Prather, Bogdan, & Hariri, 2013; Sagaspe et al., 2006; Zohar, Tzischinsky, Epstein, & Lavie,
2005). These heightened affective responses are paralleled by changes in central and peripheral physiology. Sleep loss is linked to exaggerated amygdala reactivity to affective stimuli (Greer,
Goldstein, & Walker, 2013; Gujar, Yoo, Hu, & Walker, 2011; Motomura et al., 2013; Yoo,
Gujar, Hu, Jolesz, & Walker, 2007), and one night of sleep deprivation results in greater pupil dilation in response to negative images as well as broadly elevated heart rate, systolic blood pressure, and sympathetic nervous system activity (Franzen, Buysse, Dahl, Thompson, & Siegle,
2009). Further, obtaining sleep, particularly REM sleep, can reduce amygdalar and subjective emotional responses to previously viewed negative stimuli (van der Helm et al., 2011).
Some theoretical accounts suggest that the heightened sensitivity to affective stimuli associated with sleep loss may be accompanied by a decrement in specificity of affective responding, resulting in indiscriminate emotional responding to potentially non-emotional stimuli. The mechanistic explanation proposed is that a lack of sleep (particularly REM sleep) may shift the firing pattern of the locus coeruleus, which typically responds in a phasic manner to salient stimuli, to a mode in which tonic, elevated release of central noradrenaline creates a hyper-vigilant brain state that is less able to discriminate between salient and non-salient stimuli
9 (Goldstein & Walker, 2014). This account has received some empirical support. Sleep deprivation has been shown to impair the discrimination between threatening and non- threatening cues and to increase negative evaluation of neutrally valenced stimuli (Goldstein-
Piekarski, Greer, Saletin, & Walker, 2015; Tempesta et al., 2010).
In addition to its impact on basic emotional processing, sleep plays an important role in executive functioning. For example, one night of total sleep deprivation reduces the ability to both focus and sustain attention (Durmer & Dinges, 2005; Goel, Rao, Durmer, & Dinges, 2009;
Ma, Dinges, Basner, & Rao, 2015) and causes deficits in short-term and working memory (Goel et al., 2009; Krause et al., 2017). These sleep-loss-induced impairments in working memory and attention are associated with reductions in brain activation in lateral prefrontal and posterior parietal cortical regions (Chee & Choo, 2004; Lythe, Williams, Anderson, Libri, & Mehta, 2012;
Ma et al., 2015; Mu et al., 2005) . Broadly, this work suggests that sleep deprivation results in impaired higher-order controlled processing and disrupted affective signaling.
The Current Research
The findings reviewed above describe similar psychological states under both stress and sleep deprivation, characterized by heightened emotion responding and impairment of higher order executive functioning. As emotion regulation draws on control systems to modulate affective responses, such regulatory functions may be impaired under stressful or sleep deprived contexts. The three papers composing this dissertation examine this idea by inducing one of these contexts and assessing its impact on emotion regulatory functioning.
In Paper 1, I hypothesized that acute stress would impair cognitive reappraisal (CR) success, and that it would do so by impacting the adaptive responding of biological systems that
10 are engaged by both CR and stress. I tested this hypothesis by inducing an acute state of psychosocial stress in a group of participants (N = 29) and examined their ability to implement
CR compared to a non-stressed control group (N = 25) while undergoing fMRI scanning.
Paper 2 builds on Paper 1 by examining the same emotion regulation strategy, CR, under sleep deprivation. Furthermore, this examination employed a within-subjects design (N = 34) in which participants completed two experimental sessions, one under a healthy night of sleep and one after a night of total sleep deprivation. I similarly hypothesized that SD would impair CR success by reducing the engagement of cognitive control systems and heightening affective responding to negative stimuli. A supplementary hypothesis was that sleep deprivation would result in decreased specificity in emotional responding, leading to elevated affective responses to neutral stimuli.
Finally, Paper 3 extends the findings of Paper 2 by assessing the neurobiological effects of sleep deprivation on a more implicit strategy of emotion regulation: affect labeling. Affect labeling reliably recruits the right vlPFC, and the interaction between this structure and the amygdala has been of particular interest. I hypothesized that sleep loss would modulate the recruitment and dynamic interplay between these key brain regions.
11 Paper 1: Does psychosocial stress impact cognitive reappraisal? Behavioral and neural
evidence
Maheen Shermohammed, Pranjal H. Mehta, Joan Zhang, Cassandra M. Brandes, Luke J. Chang,
and Leah H. Somerville
Published in Journal of Cognitive Neuroscience (2017), Volume 29 (11), 1803-1816
ABSTRACT
Cognitive reappraisal (CR) is regarded as an effective emotion-regulation strategy. Acute stress, however, is believed to impair the functioning of prefrontal-based neural systems, which could result in lessened effectiveness of CR under stress. This study tested the behavioral and neurobiological impact of acute stress on CR. While undergoing fMRI, adult participants (n=54) passively viewed or used CR to regulate their response to negative and neutral pictures and provided ratings of their negative affect in response to each picture. Half of the participants experienced an fMRI-adapted acute psychosocial stress manipulation similar to the Trier Social
Stress Test, and a control group received parallel manipulations without the stressful components. Relative to the control group, the stress group exhibited heightened stress as indexed by self-report, heart rate, and salivary cortisol throughout the scan. Contrary to our hypothesis, we found that reappraisal success was equivalent in control and stress groups, as was electrodermal response to the pictures. Heart rate deceleration, a physiological response typically evoked by aversive pictures, was blunted in response to negative pictures and heightened in response to neutral pictures in the stress group. In the brain, we found weak evidence of stress- induced increases of reappraisal-related activity in parts of the prefrontal cortex and left amygdala, but these relationships were statistically fragile. Together, these findings suggest that
12 both the self-reported and neural effects of CR may be robust to at least moderate levels of stress, informing theoretical models of stress effects on cognition and emotion.
INTRODUCTION
The ability to flexibly direct cognitive resources to influence an emotional response, known broadly as emotion regulation, is a key feature of healthy psychological functioning. A particularly well-studied strategy to regulate emotions is cognitive reappraisal (CR), which involves re-interpreting the content of an emotional stimulus in a way that changes its meaning.
For example, a student with a messy roommate may try to down-regulate a negative affective response to a pile of dirty dishes by remembering that her roommate is a busy medical student with no intentions of being inconsiderate.
CR is considered a highly effective method of reducing negative affective responses when compared to passively experiencing affect or attempting to suppress it. It is more effective at decreasing negative and increasing positive emotions (Gross, 1998) and modulating biological components of the emotional response, such as sympathetic nervous system activity and amygdala activation (Goldin et al., 2008; Gross, 1998). In addition, using CR in everyday life predicts fewer depressive symptoms and greater psychological wellbeing (Gross & John, 2003).
However, the efficacy of CR has been primarily tested in controlled laboratory environments free of the distraction and arousal of external stressors. As such, it is less clear whether CR remains robust under more complex contexts such as stress. Stressful circumstances are precisely the time when one would benefit from intact emotion regulation, and it would therefore be especially problematic if this popular emotion regulation strategy were flimsy under such contexts. To
13 address this question, the present study evaluated the impact of acute stress on CR and its associated mechanisms.
Here we consider stress to mean a multilayered reaction to a perceived threat, which interrupts biological homeostasis and results in a multifaceted response at subjective psychological, endocrine, and physiological levels (McEwen, 2000). Previous studies suggesting a link between stress and emotion regulatory processes have used the glucocorticoid hormone cortisol as a proxy for stress. However, the exclusive use of cortisol in the measurement of stress limits understanding of these effects because cortisol is not a direct proxy for stress. Not all situations perceived as stressful result in a cortisol response (Hellhammer, Wüst, & Kudielka,
2009), and cortisol increases are not uniquely provoked by events perceived as stressful
(Anisman & Merali, 1999). As such, the present study aimed to offer a more comprehensive account of stress effects by measuring stress on multiple levels.
Initial observations suggest that negative or stressful contexts influence whether someone will choose to use CR as well as the psychological and physiological consequences of doing so.
Previous work has shown that participants were disinclined to choose to use CR while experiencing intense negative situations in the laboratory (Sheppes et al., 2011). In both healthy and trauma-exposed populations, greater perceived stress or stress-related symptoms are related to decreased tendencies to use CR (Boden et al., 2012; Miklósi et al., 2014; Moore et al., 2008).
In addition, when explicitly instructed to use CR, participants with elevated levels of cortisol are less successful at regulating conditioned fear responses (Raio et al., 2013) and generate fewer non-negative reappraisals (Tsumura et al., 2015). Further, using CR to downregulate negative responses to laboratory stressors can paradoxically increase cortisol responding (Denson,
Creswell, Terides, & Blundell, 2014b). Together, this work suggests that CR and stress interact
14 in complex ways, and that the efficacy of CR may be vulnerable while an individual is experiencing stress.
Neurobiological models indicate that CR and stress engage overlapping neural substrates, further supporting the possibility for modulatory interactions of stress on CR efficacy. CR is believed to rely on domain-general cognitive control systems largely in the prefrontal cortex
(PFC) to modulate (usually decrease) activation in limbic regions, particularly the amygdala
(Buhle et al., 2014; Ochsner & Gross, 2005). Meanwhile, acute stress is believed to inhibit the
PFC, impairing the very control systems recruited in CR, while simultaneously bolstering amygdala responsiveness (Arnsten, 2009). This is reflected in the impairment of executive functions such as working memory, attention, and cognitive flexibility under stress (Alexander et al., 2007; Luethi et al., 2009; Olver et al., 2015) and the enhancement of phenomena such as classical fear conditioning (Luethi et al., 2009) and memory for emotional pictures (Buchanan &
Lovallo, 2001; Payne et al., 2007).
The goal of the present study was to evaluate the impact of acute psychological stress on
CR and its associated neural substrates. Half of the participants experienced acute stress during fMRI scanning, induced by a variant of the Trier Social Stress Task (TSST), while attempting to use CR to regulate their emotional response to aversive images. The other participants experienced a control version of the stress task, which mimicked procedural aspects of the stress induction but did not induce stress. Analyses first validated the effectiveness of the stress manipulation, and then measured the degree to which stress modulated emotional reactivity and
CR success. We predicted that acute stress would impair CR success, and that it would do so by impacting the adaptive responding of biological systems that are engaged by both CR and stress.
15 Initial observations suggest that negative or stressful The goal of this study was to evaluate the impact of contexts influence whether someone will choose to use acute psychological stress on CR and its associated neural CR as well as the psychological and physiological conse- substrates. Half of the participants experienced acute quences of doing so. Previous work has shown that par- stress during fMRI scanning, induced by a variant of the ticipants were disinclined to choose to use CR while Trier Social Stress Task, while attempting to use CR to experiencing intense negative situations in the laboratory regulate their emotional response to aversive images. (Sheppes, Scheibe, Suri, & Gross, 2011). In both healthy The other participants experienced a control version and trauma-exposed populations, greater perceived stress of the stress task, which mimicked procedural aspects Brainor stress-related imaging symptoms data were are used related to toidentify decreased stress ten--relatedof the modulation stress induction of but activity did not in induce the neural stress. Analyses dencies to use CR (Miklósi, Martos, Szabó, Kocsis-Bogár, first validated the effectiveness of the stress manipulation circuitry& Forintos, previously 2014; Boden, implicated Bonn-Miller, in Kashdan, CR. Alvarez, and then measured the degree to which stress mod- & Gross, 2012; Moore, Zoellner, & Mollenholt, 2008). In ulated emotional reactivity and CR success. We predicted addition, when explicitly instructed to use CR, partici- that acute stress would impair CR success and that it pants with elevated levels of cortisol are less successful would do so by impacting the adaptive responding of bi- at regulating conditioned fear responses (Raio, Orederu, ological systems that are engaged by both CR and stress. METHODPalazzolo, Shurick, & Phelps, 2013) and generate fewer Brain imaging data were used to identify stress-related nonnegative reappraisals (Tsumura, Sensaki, & Shimada, modulation of activity in the neural circuitry previously 2015). Furthermore, using CR to downregulate negative implicated in CR. responses to laboratory stressors can paradoxically in- crease cortisol responding (Denson, Creswell, Terides, METHODS Participants& Blundell, 2014). Together, this work suggests that CR and stress interact in complex ways and that the efficacy Participants of CR may be vulnerable while an individual is experienc- Fifty-six young adults aged 18–23 years participated in Fiftying stress.-six young adults aged 18-23 participated in thisthis study. study. Two Two participants participants were were excluded excluded from analyses Neurobiological models indicate that CR and stress en- because of noncompliance, leaving a final sample of 54 fromgage overlappinganalyses due neural to non substrates,-compliance, further supporting leaving a final(see sample Table 1 of for 54 detailed (see Table demographic 1.1 for information). detailed All the possibility for modulatory interactions of stress on CR participants were right-handed, nonsmokers, proficient efficacy. CR is believed to rely on domain-general cognitive in English, and not currently receiving treatment for psy- demographiccontrol systems information). largely in the PFC All to participants modulate (usually were rightchological-handed, or neurological non-smokers, disorders. proficient In an effortin to re- decrease) activation in limbic regions, particularly the cruit a more representative sample, no more than 25% English,amygdala (Buhleand not et currently al., 2014; Ochsnerreceiving & Gross,treatment 2005). for psychologicalof the participants or wereneurological current or disorders. former students In an at Meanwhile, acute stress is believed to inhibit PFC, impair- Harvard University. All participants provided informed ing the very control systems recruited in CR, while simulta- written consent for their participation. Research proce- effortneously to bolstering recruit a amygdalamore representative responsiveness sample, (Arnsten, no moredures than were 25% approved of participants by the Committee were oncurrent the Use or of 2009). This is reflected in the impairment of executive func- Human Subjects at Harvard University and by the Army formertions such students as working at Harvard memory, attention,University. and Al cognitivel participantsHuman provided Research informed Protections written Office. consent for flexibility under stress (Olver, Pinney, Maruff, & Norman, 2015; Luethi, Meier, & Sandi, 2009; Alexander, Hillier, Tasks theirSmith, participation. Tivarus, & Beversdorf, Research 2007) procedures and the enhancement were approved by the Committee on the Use of Human of phenomena such as classical fear conditioning (Luethi Task Overview Subjectset al., 2009) at andHarvard memory University for emotional and pictures by the (Payne Army HumanThe goal Research of the experimental Protections session Office was to (AHRPO). elicit a stress et al., 2007; Buchanan & Lovallo, 2001). response that would persist throughout fMRI scanning to
TableTable 1.1.1Demographics Demographi ofcs Assignedof Assigned Groups Groups Control Stress Statistic
Total 25 29 Female 48% 51.72% χ2(1) = 0.074, p = .785 White (non-Hispanic) 48% 51.72% χ2(1) = 0.074, p = .785 Age 20.89 (1.71) 21.13 (1.67) t(50.53) = −0.510, p = .613 IQ 116.12 (12.83) 118.86 (12.22) t(50) = −0.800, p = .427
For age and IQ, means and standard deviations are reported, and all t tests are Welch’s t tests for unequal variances. IQ was obtained using the ForWechsler age and Abbreviated IQ, means Scale and of Intelligence standard (Seconddeviations Edition) are reported, FSIQ-2 score. and all t tests are Welch’s t tests for unequal variances. IQ was obtained using the Wechsler Abbreviated Scale of Intelligence (Second Edition) FSIQ-2 score. 1804 Journal of Cognitive Neuroscience Volume 29, Number 11
16 Tasks
Task Overview
The goal of the experimental session was to elicit a stress response that would persist throughout functional neuroimaging (fMRI) scanning in order to observe effects of stress on reappraisal processes. To do so, participants in the Stress condition completed elements of established stress induction tasks immediately before and during fMRI scanning. Participants in the Control condition completed highly similar procedures, absent the stressful components. During fMRI scanning, the primary dependent variables were behavioral and neural activity measured during the CR task. In addition, stress indices were acquired throughout fMRI scanning, including saliva samples for cortisol analysis, self-reported affect, and heart rate (see Figure 1.1A). These measures were used for manipulation checks of the stress induction and for individual difference analyses.
17
Figure 1.1 Study design. (A) Three indices of stress were acquired repeatedly throughout the experiment, denoted by checkmarks. The gray box indicates which measures occurred in the MRI. (B) Example screens from the math task for the stress and control groups. For the stress group, a blue bar growing quickly from the left to right sides of the screen indicated the time remaining to answer the math problem, and arrows falsely indicated whether participants (bottom arrow) and their peers (top arrow) were performing below average (red), average (yellow), or above average (green). The control task had similar visual features but with simpler math problems, no time pressure, and no social comparison. (C) The CR task first instructed participants to use CR (i.e., Decrease) or Look at the forthcoming picture, then displayed an image, and then acquired self-reported affect ratings on a scale of 1 (“not at all bad”) to 5 (“very bad”).
18 CR Task
Building on prior work, we employed a task frequently used to target cognitive reappraisal processes (Ochsner et al., 2004). In this task, participants viewed negative and neutral images and were asked to use cognitive reappraisal to decrease their emotional response to half of the negative images (Figure 1.1C). Before the MRI session, participants were thoroughly trained on the strategy of cognitive reappraisal. They were given example reappraisals for sample negative images (e.g. “help is on the way”, “it’s just a scene from a movie”, etc.) and generated their own reappraisals aloud to the experimenter over several practice trials until they demonstrated clear understanding of the goals of the task.
During fMRI scanning, a two second instructional cue first informed participants whether they were to passively view (“LOOK”) or reappraise (“DECREASE”) an image that followed.
The image was displayed (8 s), and then participants rated how negative they felt on a scale of 1-
5 using a button response box held in the right hand (2 s). Jittered fixation periods (2-8 s) were presented between the image display and the rating scale, and between the rating scale and the next trial. A total of 72 pictures (24 decrease-negative, 24 look-negative, 24 look-neutral) were presented over 3 scanning runs with 24 trials each. Run order was counterbalanced across participants. Images were from the International Affective Picture System (IAPS; Lang, Bradley,
& Cuthbert, 1997) database1 and all included depictions of people. Two sets of negative images were constructed with the same average normative valence and arousal and assigned to the decrease-negative or look-negative condition, counterbalanced across participants. In addition to
1 IAPS images used: 2038, 2095, 2100, 2102, 2104, 2120, 2205, 2214, 2305, 2357, 2375.1, 2383, 2385, 2393, 2441, 2455, 2480, 2487, 2490, 2493, 2512, 2575, 2579, 2590, 2595, 2661, 2683, 2691, 2700, 2702, 2703, 2710, 2749, 2750, 2799, 2811, 2840, 2870, 2900, 3160, 3180, 3220, 3280, 3300, 3350, 3500, 4605, 4621, 6211, 6212, 6250, 6311, 6312, 6313, 6530, 6555, 6561, 6821, 6840, 7493, 9007, 9070, 9331, 9341, 9404, 9423, 9424, 9429, 9584, 9903, 9905, 9927
19 self-report ratings, heart rate and electrodermal activity were acquired as measures of emotional response to each of the 3 picture conditions.
Speech Stress Induction Task
The speech stress induction task was adapted from the preparation phase of the TSST
(Kirschbaum, Pirke, & Hellhammer, 1993). Before the MRI scan, participants in the Stress group were told that immediately after the scan, they would give a speech about what makes a good friend, including the participant’s own personal strengths and weaknesses as a friend. The speech was to be delivered to a panel of judges who would code their verbal and nonverbal behavior, and video-recorded for further evaluation. Participants in the Control group were told that they would write a story immediately after the MRI scan that would not be used as data in the experiment. Both groups were given 3 minutes to prepare for their respective task. After the preparation period, participants entered the MRI, and three times over the course of the scanning session they were reminded of the forthcoming speech (Stress group) or story-writing (Control group) task (e.g. “A few more scans and then you can come out and give that speech you prepared”, “The judges for your speech just arrived, so we’ll be able to start that right after the
MRI”, “Just a little while longer and then our judges will be ready to hear your presentation”).
Upon leaving the MRI, participants were debriefed and informed that they would not have to make the speech or write the story.
Math Stress Induction Task
The math stress induction task was adapted from the Montreal Imaging Stress Task (Dedovic et al., 2005). During fMRI scanning, participants were presented with math problems, which they
20 answered by selecting a number on a rotary dial (see Figure 1.1B for a diagram). The dial began with “0” highlighted, and participants used the button box to navigate around the dial to select their chosen response.
For participants in the Stress group, these problems included up to 4 one- or two-digit integers, and operations could include any combination of addition, subtraction, multiplication, and division (e.g. 2/6*12+4). They were given a very limited amount of time to answer these questions and, to increase the sense of urgency, they saw a progress bar growing on the screen and heard a tone of escalating frequency. This time limit changed adaptively to ensure poor performance. An initial time limit was calculated from performance speed during a non-timed practice round, and then flexibly shortened or lengthened if the participant got the previous 3 problems correct or incorrect, respectively.
Participants in the Stress condition were also given negative social evaluative feedback about their task performance. First, they were falsely told before the task that they must get at least 70% of the questions correct for researchers to be able to use their data, and that most people get about 80% correct. In reality, the average performance for this group was 44%.
During the task itself, they saw two arrows pointing to a colored bar. They were told that the top arrow indicated the average performance of others in their peer group, and that the bottom arrow indicated their own performance. The colors on the bar corresponded to above average (green), average (yellow), and below average (red). The time pressure forced participants to perform in the red, while the sham performance indicator for the group stayed generally within the green zone. Finally, after every run of the task, stressed participants were given feedback from the researcher expressing surprise and disappointment at their poor performance.
21 The Control participants completed a similar version of the task that taxed mathematical processes but did not include the potentially stressful components. The control version of the task was self-paced (hence, no time pressure) and participants did not receive the social evaluative feedback. The mathematical operations were also easier, only involving adding or subtracting up to 3 one-digit numbers at a time (e.g., 5-2).
In each of 3 functional imaging runs, participants completed a two-minute block of the math task, which was preceded and succeeded by 30 seconds of resting baseline where the visual interface was displayed without a math problem. Heart rate was continuously monitored during both the math and rest blocks, serving as a measure of physiological stress in response to this task.
Study Procedure
After consenting, participants were introduced to the tasks and given the opportunity to practice them. For the math task, all participants practiced the un-timed, easier version. Next, in order to give participants time to reduce any initial stress associated with entering the laboratory environment, participants quietly watched the first 20 minutes of an episode from the Planet
Earth TV series titled “Ocean Deep,” a mildly positive, informative video about animals in the ocean (Denson, Mehta, & Ho Tan, 2013). Participants then provided their first salivary cortisol sample along with self-report mood ratings (i.e., how much they currently felt Nervous, Excited,
Stressed, Alert, and Happy on scale of 1-5). Participants then received instructions about the speech (Stress group) or story (Control group) task and completed the 3-minute preparation phase where they could outline their speech or story. At the end of the 3 minutes, participants were led to the MRI room and prepared for scanning.
22 In the MRI scanner, participants completed 3 runs each of the math and CR tasks in alternating order, with math always being first. After the first math/CR task pair (T2), participants once again provided mood ratings, which were displayed on the screen in the MRI and answered using the button response box. After the second math/CR task set (T3), participants provided their second salivary cortisol sample while remaining in the MRI and completed mood ratings for the third time. At the very end of the MRI session (T4), participants provided the final saliva sample and completed mood ratings for the fourth time.
After exiting the MRI, participants were asked if they knew what was happening next in order to give them an opportunity to volunteer any suspicions about the deceptive nature of the speech component. During the full debriefing, we asked participants whether they found the speech/story and math tasks stressful, which gave them another chance to volunteer any suspicions they had. Finally, participants completed follow-up questionnaires and assessments and provided mood ratings for the fifth time. They were then compensated for their time and thanked for their participation.
Dependent Measures
Salivary Cortisol
Salivary cortisol samples were collected by placing SalivaBio oral swabs under the tongue for 3-
4 minutes. They were then immediately stored in Swab Storage Tubes (Salimetrics, Inc.) at -
20°C. To maximize the integrity of the samples, participants were instructed to refrain from exercising, consuming caffeine, eating or drinking anything but water, and brushing their teeth for 12 hours, 2 hours, 30 minutes, and 1 hour prior to the appointment, respectively. To reduce the effects of diurnal variation, all MRI sessions were scheduled between 1230 h and 1730 h.
23 Once data collection was complete, samples were shipped frozen overnight to the Kirschbaum laboratory (Technische Universität, Dresden, Germany) for analysis with chemiluminescence immunoassay kits (IBL International, Hamburg, Germany). Samples were analyzed in singlets, and the inter-assay coefficient of variation was less than 8%. One participant whose baseline cortisol level was almost 5 standard deviations above the mean was removed from any analyses that did not correct for baseline levels.
Heart Rate
Heart rate (HR) was monitored with a fiber-optic oximetry sensor on the left ring finger using
BIOPAC® Systems MP150 and OXY-MRI (Nonin Medical Inc.) hardware, and AcqKnowledge software. Technical difficulties with the AcqKnowledge software resulted in full data loss for n=8. Any participant with at least one full run of HR data was included in these analyses, resulting in n=46 usable datasets.
These data were collected for two distinct purposes. First, HR acquired during each run of the math task was compared to the resting baseline preceding it to provide an indicator of autonomic stress responsiveness. Second, previous research has shown that viewing aversive pictures elicits a brief, parasympathetically mediated HR deceleration, and the magnitude of this deceleration varies by stimulus intensity (Bradley, Codispoti, Cuthbert, & Lang, 2001; Campbell,
Wood, & McBride, 1997). Thus, HR was acquired during the CR task as a physiological index of emotional responding to pictures. The pulse oximeter is not ideal for stimulus-locked measurements because rather than providing an electrocardiogram directly, it calculates beats- per-minute (BPM) by averaging over a window of heartbeats, causing a ~3.5 s delay between initial stimulus presentation the resulting change in HR (Nonin Medical Inc., personal
24 communication). Thus, the entire HR trace was shifted by 3.5 seconds before computing the average HR for the 4.5 s following picture onset, corrected for the average HR during the 2 s preceding.
Electrodermal Activity
Electrodermal activity (EDA) was recorded during fMRI scanning with Ag/AgCl electrodes on the left index and left middle finger using BIOPAC® Systems MP150 and EDA100C-MRI hardware and AcqKnowledge software at a sampling rate of 250 Hz. All data were first processed with a 1 Hz low-pass filter to remove high-frequency scanner noise. In accordance with the procedure in Raio et al. (2013), the amplitude of the skin conductance response (SCR) for each trial was calculated using the maximum change from base to peak in the 0.5 to 4.5 seconds following picture onset. Amplitudes below 0.02 µS were scored as 0, and the data were square root transformed to normalize distributions for standard statistical analyses. Technical difficulties with the software and the exclusion of participants who did not produce any SCRs resulted in a final sample size of n=41 usable EDA datasets.
fMRI Analysis
Acquisition and Preprocessing
Brain imaging was performed on a 3.0 Tesla Siemens Prisma scanner with a 32-channel head coil (Siemens Medical Systems, Erlangen, Germany) at the Harvard University Center for Brain
Science-Neuroimaging. A T1-weighted high-resolution anatomical image of the brain was acquired using a multiecho multiplanar rapidly acquired gradient-echo (MEMPRAGE) sequence
(176 sagittal slices, repetition time = 2200 ms, echo time = 1.67 ms, flip angle = 7°, slice
25 thickness = 1 mm, voxel size = 1x1x1 mm). Functional images were collected using an echo planar imaging T2*-weighted sequence sensitive to the BOLD response (69 axial slices per whole-brain volume, voxel size = 2.2x2.2x2.2 mm, repetition time = 2000 ms, echo time = 35 ms, flip angle = 80°, multi-band acceleration factor = 3). Functional slices were oriented to a slightly greater tilt than the anterior-posterior commissure plane to minimize signal dropout due to sinus cavities. Nonetheless, an abnormally large sinus cavity in one participant resulted in excessive frontal distortion. This participant was excluded from all imaging analyses.
Functional imaging data were preprocessed using the Functional MRI of the Brain
Software Library (FSL; version 5.0.4; Smith et al., 2004) tools implemented in Nipype (v.
0.11.0; Gorgolewski et al., 2011) using the Lyman interface (v. 0.0.7; http://www.cns.nyu.edu/~mwaskom/software/lyman/). Each functional scan was first realigned to its middle volume, spatially smoothed with a 6mm full-width at half maximum Gaussian kernel, and high-pass filtered at 128 seconds. Functional scans were co-registered to individual- subject anatomical images using bbregister (v. 5.3.1; Greve & Fischl, 2009), and all first-level or individual subject analyses were conducted in this native space. Analyses comparing participants or groups first normalized statistical maps to a Montreal Neurological Institute brain template using linear and nonlinear warping methods through the Advanced Normalization Tools software
(v. 1.9.x; Avants, Tustison, & Song, 2009).
Modeling of CR Task
Preprocessed images were entered into a standard general linear model (GLM) in FSL, which estimated neural responses to: the cue period (collapsed over trial types), the three types of picture periods (decrease-negative, look-negative, and look-neutral), and the rating period
26 (collapsed over trial types). Regressors used boxcar functions convolved with the canonical double-gamma hemodynamic response function implemented in FSL. The model also included nuisance regressors for motion parameters, temporal derivatives for each regressor of interest, and temporal filter regressors with a cut-off of 128s. To remove additional noise, functional volumes with motion greater than 1mm or whole-brain intensity values greater than 4.5 standard deviations away from the mean were censored from the model as additional regressors; no scan had greater than 10% censored volumes. All participants had at least 2 out of 3 scans of usable data.
Region-of-Interest (ROI) Analysis
Unbiased, a priori ROIs were identified using a previous meta-analysis of 48 neuroimaging studies of reappraisal (Buhle et al., 2014). Eleven 6mm spheres were centered on activation peaks from the reappraise>emotional baseline contrast reported in the meta-analysis, in addition to two 4mm spheres in the bilateral amygdalae from the emotional baseline>reappraise contrast reported in the meta-analysis (see Table 1.2 for ROI coordinates). Mean parameter estimates were extracted for the picture period from each ROI and tested for task and stress effects. P- values from these tests were corrected for 13 comparisons by controlling for the false discovery rate using the Benjamini–Hochberg method.
27 frontal distortion. This participant was excluded from all look-neutral), and the rating period (collapsed over trial imaging analyses. types). Regressors used boxcar functions convolved with Functional imaging data were preprocessed using the the canonical double-gamma hemodynamic response Functional MRI of the Brain Software Library (FSL, Version function implemented in FSL. The model also included 5.0.4; Smith et al., 2004) tools implemented in Nipype nuisance regressors for motion parameters, temporal de- (v. 0.11.0; Gorgolewski et al., 2011) using the Lyman inter- rivatives for each regressor of interest, and temporal filter face (v. 0.0.7; www.cns.nyu.edu/∼mwaskom/software/ regressors with a cutoff of 128 sec. To remove additional lyman/). Each functional scan was first realigned to its noise, functional volumes with motion greater than 1 mm middle volume, spatially smoothed with a 6-mm FWHM or whole-brain intensity values greater than 4.5 SDs away Gaussian kernel, and high-pass filtered at 128 sec. Func- from the mean were censored from the model as additional tional scans were coregistered to individual-participant regressors; no scan had greater than 10% censored anatomical images using bbregister (v. 5.3.1; Greve & volumes. All participants had at least two of three scans of Fischl, 2009), and all first level or individual-participant usable data. analyses were conducted in thisnativespace.Analyses comparing participants or groups first normalized statisti- ROI Analysis cal maps to a Montreal Neurological Institute (MNI) brain template using linear and nonlinear warping methods Unbiased, a priori ROIs were identified using a previous through the Advanced Normalization Tools software meta-analysis of 48 neuroimaging studies of reappraisal (v. 1.9.x; Avants, Tustison, & Song, 2009). (Buhle et al., 2014). Eleven 6-mm spheres were centered on activation peaks from the reappraise > emotional baseline contrast reported in the meta-analysis, in addi- tion to two 4-mm spheres in the bilateral amygdalae from Modeling of CR Task the emotional baseline > reappraise contrast reported in Preprocessed images were entered into a standard general the meta-analysis (see Table2forROIcoordinates). linear model in FSL, which estimated neural responses to Mean parameter estimates were extracted for the picture the cue period (collapsed over trial types), the three types period from each ROI and tested for task and stress of picture periods (decrease-negative, look-negative, and effects. p Values from these tests were corrected for
Table 1.2 ROI Analyses for the Effects of Stress on Reappraisal and Reactivity Table 2. ROI Analyses for the Effects of Stress on Reappraisal and Reactivity Reappraisal Reactivity MNI Coordinates Effect Effect Reappraisal × Group Reactivity × Group
ROI Side x y z Corrected p Corrected p (Uncorrected p) Inferior frontal gyrus R 60 24 3 <.001** .624 .643 (.346) .909 (.198) Middle frontal gyrus R 51 15 48 <.001** .010* .291 (.045*) .909 (.282) Medial frontal gyrus R 9 30 39 <.001** .219 .643 (.334) .909 (.837) Anterior cingulate gyrus L −3 24 30 <.001** .824 .936 (.870) .909 (.531) Superior frontal gyrus L −9 12 69 <.001** .824 .936 (.872) .909 (.454) Middle frontal gyrus L −33 3 54 <.001** <.001** .643 (.161) .909 (.909) Anterior insula L −36 21 −3 <.001** .032* .936 (.936) .909 (.815) Inferior frontal gyrus L −42 45 −6 <.001** .219 .643 (.288) .909 (.530) Superior temporal gyrus R 63 −51 39 <.001** .942 .804 (.586) .531 (.041*) Angular gyrus L −42 −66 42 <.001** .792 .804 (.618) .909 (.776) Middle temporal gyrus L −51 −39 3 <.001** .135 .729 (.449) .909 (.612) Amygdala R 30 −3 −15 .291 .219 .643 (.249) .909 (.341) Amygdala L −18 −3 −15 .122 .824 .202 (.016*) .909 (.884)
Each ROI was subject to a separate ANOVA testing for the effect of stress group on reappraisal or reactivity. Significance tests for each interaction are Eachreported ROI both was with subject and without to a correction separate for ANOVA the number testing of ROIs for tested. the Regioneffect labelsof stress were basedgroup on on the reappraisal Harvard-Oxford or corticalreactivity. and subcortical atlases cross-referencing (Mai, Paxinos, & Voss, 2008). Significance tests for each interaction are reported both with and without correction for the number of ROIs tested. *p < .05. Region**p < .001. labels were based on the Harvard-Oxford cortical and subcortical atlases cross-referencing (Mai, Paxinos, & Voss, 2008). *p < .05. **p < .001. 1808 Journal of Cognitive Neuroscience Volume 29, Number 11
Whole-brain fMRI Analysis
To supplement ROI analyses, a random effects analysis focused on the picture period was also performed across the whole brain. First, to corroborate established findings with this task, one- tailed t-tests for contrasts isolating the effects of reappraisal (decrease-negative > look-negative) and emotional reactivity (look-negative > look-neutral) were computed across the whole sample.
Next, two-sample t-tests comparing the Stress and Control groups were computed for these contrasts to identify activations that were significantly greater in Stress vs. Control groups.
To address individual differences in neural responses in reaction to stress reactivity measures, we conducted three separate group-level models using HR changes, cortisol changes,
28 or self-reported stress levels as continuous covariates of interest. All contrast maps were corrected at a family-wise error (FWE) threshold of p<0.05 using FSL’s cluster-based correction.
Activation maps were first thresholded (z>2.3), and then Gaussian Random Field theory was used to calculate a cluster-size threshold, below which clusters were removed.
Analysis Strategy
Primary analyses evaluated how stress influenced responses during the reappraisal task. On one hand, stress could affect a participant’s ability to use reappraisal effectively. If this was the case, stress should specifically reduce any differences between the decrease-negative and look- negative conditions, i.e., the “reappraisal contrast.” Another possibility is that stress could make participants more reactive to negative pictures in general. This would result in an exacerbation of any differences between the look-negative and look-neutral conditions, i.e., the “reactivity contrast.”
While the Stress group experienced more stress than the Control group on average, some participants in the Control group reported feeling stressed and some in the Stress group did not.
We thus decided to proceed with two parallel analysis streams. In the first, we considered the stress measure categorically based on the experimental manipulation of stress, comparing those who were stressed by the stress manipulation relative to unstressed controls. Therefore, we excluded any “non-responders” in the Stress group (i.e. those who at no point during the manipulation rated their stress level as higher than they did at baseline, n=5), and any stressed participants in the Control group (i.e. those whose average stress rating increase from baseline was greater than 0.67, which was 1.5 times the interquartile range above the upper quartile, n=2).
29 In the second analysis stream, we considered stress on a continuum, including all participants irrespective of their response levels or which experimental manipulation they received. This analysis examined the effects of how much stress someone was feeling during fMRI scanning relative to baseline. We computed three separate measures of stress levels using self-report, HR, and cortisol. To index an individual’s self-reported stress levels, we computed the average of the “Stressed” mood rating for T2, T3, and T4, corrected for the rating at baseline
(T1). As another measure of stress, we calculated the difference between the average HR for the period of the math task when participants were answering problems and the rest period preceding it (HRdiff). Finally, we assessed cortisol levels using the formula in Pruessner et al. (2003) for area under the curve with respect to the increase, which is a measure of a participant’s cortisol levels throughout the experiment correcting for the pre-task baseline level.
For both analysis streams, we conducted two-way ANOVAs to test the effects of stress and the reappraisal contrast (decrease-negative vs. look-negative) or reactivity contrast (look- negative vs. look-neutral) on the outcome variables: picture ratings, neural activation, and physiology during the CR task. All statistical analyses were performed in R 3.2.3 (R Core Team,
2015).
RESULTS
Stress Induction
Overall, the speech and math tasks were successful in inducing stress, as assessed by self-report, cortisol levels, and HR changes. Cortisol levels and self-reported stress were both expected to increase from baseline levels measured at the beginning of the study. Both of these measures were submitted to a two-factor ANOVA, using group (Stress, Control) as a between-subjects
30 factor and sample number (ordered time points) as a within-subjects factor. The distribution of cortisol values was right-skewed, and thus log10 transformed first. There was a significant group x sample number interaction for self-reported stress ratings (F(4,50)=12.89, p<0.001), which subsequent t-tests revealed was driven by no difference between groups during the initial baseline measure (p=0.90) or after participants were debriefed (p=0.58), and an increase in ratings for the Stress group for samples T2, T3, and T4 compared to the Control group (p’s <
0.001; Figure 1.2A).
There was also a significant interaction between group and sample for cortisol responding
(F(2,101)=3.880, p=0.024), with no difference at baseline (p=0.43) and an increase for the Stress group at samples T3-T4 compared to Controls (p’s < 0.005). HR levels were expected to increase during the math task relative to the fixation period preceding it. A two-factor ANOVA comparing the average HR during the math and baseline fixation periods between the two groups revealed a significant interaction between group and math task period (F(1,46)=37.52, p<0.001) driven by an increase in heart rate when answering problems in the Stress but not Control group
(Figure 1.2B). Together, these analyses provide converging evidence that the task procedures were effective at inducing a multi-faceted stress response that was selective to the Stress group.
Previous studies have shown that males exhibit a greater cortisol response to laboratory stressors than females (Childs, Dlugos, & De Wit, 2010; Kirschbaum, Kudielka, Gaab,
Schommer, & Hellhammer, 1999; Kirschbaum, Wust, & Hellhammer, 1992). We thus also examined the interaction between group and sample separately in males and females (Figure
1.2C). In line with previous work, the interaction was significant in males (F(2,50)=4.842, p=0.012) but not females (F(2,47)=0.898, p=0.414), despite both showing increases in subjective stress [males: (F(4,99)=8.493, p<0.001); females: (F(4,99)=5.738, p<0.001)].
31 During debriefing, 4 out of the 29 participants assigned to the Stress manipulation expressed suspicion about at least some component of the study (e.g. they did not believe they would have to give a speech at the end). Removing these participants’ data did not impact the results of the key behavioral analyses discussed below.
Figure 1.2 Measures of components of the stress response in the stress and control groups. (A) Participants in the stress group exhibited higher average self-reported stress ratings after the stress induction (T2–T4), but not at baseline or after debriefing. (B) Participants in the stress group showed an increase in average HR during the math task compared with the rest period preceding it, whereas the control participants did not. (C) Participants in the stress group exhibited increased cortisol from baseline after the stress induction, although this effect was greater in male participants. Cortisol values are log transformed. Error bars represent SEM.
Effects of Stress on Emotional Reactivity & CR
To evaluate the general effectiveness of the stimuli and task instructions, we first evaluated main effects of picture valence and reappraisal instruction on affect ratings. As expected, the negative pictures were rated significantly more negatively than the neutral pictures (F(1,45)=681.66, p<0.001; look-negative mean=2.95, SEM=0.075; look-neutral mean=1.27, SEM=0.03). A significant main effect of picture condition (F(1,45)=75.36, p<0.001) indicated that consistent with prior work, participants successfully reduced their negative reaction to negative pictures using reappraisal strategies (decrease-negative mean=2.27, SEM=0.075) compared to passive viewing.
32 The critical tests evaluated whether stress modulated reappraisal success and reactivity levels. Evaluating interactions with Stress group revealed no significant interaction between group and picture type on either reappraisal success (F(1,45)=1.306, p=0.259) or emotional reactivity (F(1,45)=1.266, p=0.266). Thus, assignment to the Stress or Control group did not influence these target emotional measures (Figure 1.3).
Figure 1.3 Reappraisal success and emotional reactivity in the stress and control groups. Reappraisal success is defined as the difference in self- reported affect ratings for look-negative > decrease-negative pictures. Emotional reactivity is defined as the difference in self-reported affect ratings for look-negative > look-neutral pictures. There was no effect of stress group on either measure. Error bars represent SEM.
Secondary analyses treating stress as a continuous variable assessed whether individual differences in engagement of the components of the stress response (self-report, cortisol, and
HR) influenced emotional reactivity or reappraisal. We found a significant interaction between emotional reactivity and self-reported stress (F(1,52)=4.435, p=0.04) and a trending interaction between reactivity and HRdiff (F(1,44)=2.921, p=0.094). Greater self-reported stress and a larger math-induced heart rate increase was associated with more negative affect ratings to look- negative pictures compared to look-neutral pictures. However, there was no effect of stress on decrease-negative pictures. There was no interaction between HRdiff and reappraisal success
(F(1,44)=1.731, p=0.195) and a trending interaction between self-reported stress and reappraisal
33 success (F(1,52)=3.898, p=0.054). The trending interaction was driven by an increase in response to look-negative pictures with greater stress (r(52)=0.357, p=0.008) whereas the response to decrease-negative was unaffected by stress (r(52)=0.084, p=0.547). We found no interaction between cortisol increases and emotional reactivity (F(1,51)=0.345, p=0.559) or reappraisal success (F(1,51)=0.656, p=0.422).
Effects of Stress on Neural Signatures of Emotional Reactivity & CR
ROI Analysis
Parallel to the behavioral analyses above, separate ANOVAs were conducted on each ROI testing for main effects of task condition, and interactions between task condition and group on fMRI responses. Results are summarized in Table 1.2. Consistent with meta-analytic evidence localizing the neural correlates of reappraisal, all cortical ROIs exhibited greater activation during reappraisal than while passively viewing negative images. Contrary to our predictions, no
ROI showed a significant interaction between group (Stress, Control) and reappraisal condition after correcting for multiple comparisons. To provide the most comprehensive description of the data, we also report uncorrected results. Without correction, there was a significant interaction between group and reappraisal condition in the left amygdala and the right middle frontal gyrus, with the Stress group exhibiting greater reappraisal-related activity in these regions.
Contrary to our expectations, we did not find any main effects of emotional reactivity in the amygdala ROIs. However, as discussed below, partially overlapping portions of the amygdalae were activated when examining the reactivity contrast in the whole brain. Like the reappraisal contrast, no interactions between group and reactivity condition survived correction for multiple comparisons.
34 We also examined whether individual differences in components of the stress response
(irrespective of group) predicted reappraisal success or reactivity levels. We found no significant interactions between any of the continuous subject-level stress measures and the reappraisal or reactivity contrasts. Without correction, there was a significant increase in reappraisal-related activity in the right middle frontal gyrus as HRdiff increased (F(1,43)=6.782, p=0.013), and a trending increase in emotional reactivity in the right amygdala as HRdiff increased
(F(1,43)=3.666, p=0.062).
Whole-brain Analysis
We supplemented targeted ROI analyses with exploratory whole-brain comparisons. Consistent with prior work (Buhle et al., 2014), we found that reappraisal (decrease-negative>look- negative) recruited an extensive network of lateral and medial prefrontal regions, as well as areas in the posterior temporal and parietal cortex (Figure 1.4; Table 1.3). Emotional reactivity (look- negative>look-neutral) recruited regions consistent with prior work, including the bilateral amygdala and insula (Table 1.3).
Examining the effects of group (Stress, Control) on the reactivity contrast revealed a group difference in the left superior temporal gyrus, such that a prominent deactivation in the
Control group when viewing negative pictures was not evident in the Stress group. There were no differences between groups for the reappraisal contrast. Whole-brain analyses examining different components of the stress response as continuous covariates revealed that as HRdiff increased, activation increased for the reactivity contrast in the lingual gyrus (peak at MNI coordinates: x=-12, y=-68, z=-2, 418 voxels) and decreased for the reappraisal contrast in
35 posterior cingulate cortex (x=4, y=-46, z=10, 952 voxels). No brain regions survived corrections for whole-brain analyses using self-reported stress or cortisol increase as continuous covariates.
Figure 1.4 Whole-brain fMRI analyses examining reappraisal (decrease-negative > look- negative) revealed the extensive network of prefrontal, temporal, and posterior parietal regions expected from previous work (Buhle et al., 2014). Images are p < .05, FWE corrected, and have been further thresholded at z > 5 for visualization purposes.
36 Table 1.3 Brain Regions Recruited During CR Task Table 3. Brain Regions Recruited during CR Task MNI Coordinates Region Side Extent Max Z xyz Reappraisal contrast (decrease-negative > look-negative) Middle frontal gyrus L 113,868 8.16 −44 4 50 Lateral occipital cortex L −52 −62 34 Superior frontal gyrus L −4 10 62 Caudate R 14 8 16 Inferior frontal gyrus L −46 24 −6 Temporal pole R 44 10 −34 Caudate L −16 4 18 Middle temporal gyrus L −58 −40 −6 Insula R 36 22 −6 Inferior frontal gyrus R 48 30 −8 Middle temporal gyrus R 50 −38 2 Superior temporal gyrus L −40 −52 24 Insula L −32 24 −2 Temporal pole L −50 14 −24 Superior frontal gyrus R 10 20 58 Superior temporal gyrus R 50 −54 32
Reactivity contrast (look-negative > look-neutral) Inferior temporal gyrus L 56,664 8.29 −42 −76 −8 Inferior temporal gyrus R 50 −72 2 Fusiform gyrus L −38 −44 −18 Fusiform gyrus R 42 −44 −20 Precentral gyrus L −44 2 28 Middle frontal gyrus L −28 −6 48 Medial frontal gyrus L −8 14 48 Habenula L −4 −28 −2 Thalamus R 18 −30 4 Precentral gyrus R 42 6 26 Thalamus L −14 −30 4 Habenula R 6 −28 −4 Insula L −28 20 2 Inferior frontal gyrus L −40 30 18 Superior parietal lobule L −28 −48 44 Superior parietal lobule R 30 −52 50 Insula R 30 26 −2 Amygdala L −26 −4 −20 Amygdala R 28 −2 −18
For both the reappraisal and reactivity contrasts, resulting maps contained one large cluster spanning many regions. Local maxima in these regions Forare reported both the as subpeaks.reappraisal Region and labels reactivity were based contrasts, on the Harvard-Oxford resulting corticalmaps andcontained subcortical one atlases large cross-referencing cluster spanning (Mai et al., many 2008). regions. Local maxima in these regions are reported as subpeaks. Region labels were based on the Harvard-Oxford cortical and1812 subcortical Journal ofatlases Cognitive cross Neuroscience-referencing (Mai et al., 2008). Volume 29, Number 11
37 Effects of Stress on Physiological Signals of Emotional Reactivity & CR
EDA Response to Pictures
There were no significant main effects of reappraisal or reactivity on EDA, and no interactions with group or any of the continuous stress measures (p’s > 0.1).
HR Response to Pictures
As expected, there was a prominent decrease in HR when viewing negative pictures. This effect was attenuated when participants used reappraisal, as suggested by a significant main effect of picture type for the reappraisal contrast (F(1,39)=6.003, p=0.019). There was also a significant main effect of group for the reappraisal contrast, such that participants in the Stress group showed a blunted HR deceleration to negative images, whether they were reappraised or not
(F(1,39)=4.439, p=0.042). A significant interaction between group and picture type for the reactivity contrast suggested that while HR deceleration was specific to negative images in the
Control group, participants in the Stress group continued to exhibit similar HR decelerations when viewing neutral pictures (F(1,39)=5.891, p=0.02).
Examining stress as a continuous measure largely mirrored the reactivity difference between groups, with interactions between both self-reported stress (F(1,44)=4.85, p=0.033) and
HRdiff (F(1,44)=15.956, p<0.001) and picture type in the reactivity contrast suggesting that the more stressed a participant was, the more their HR decelerated in response to neutral compared to negative pictures. In addition, both of these continuous stress measures exhibited a significant main effect on HR deceleration within the reappraisal contrast [self-report: F(1,44)=4.86, p=0.033; HRdiff: F(1,44)=6.435, p=0.015], such that as participants were more stressed, they experienced less of a HR deceleration in response to negative pictures that were just looked at as
38 well as those that were reappraised. Cortisol levels did not exhibit either of the aforementioned effects.
DISCUSSION
The present study sought to examine the effects of stress on CR and its associated neural mechanisms. The manipulation successfully induced a rich, multifaceted stress response: the
Stress group exhibited an increase in cortisol, heart rate, and self-reported stress levels relative to the Control group. Participants also responded to the emotion regulation task as expected.
Cognitive reappraisal attenuated participants’ emotional responses to negative images, and recruited the extensive network of brain regions in the temporal, parietal, and lateral and medial prefrontal cortices that has been implicated in previous work.
Nonetheless, the observed neural and behavioral effects of stress on CR were minimal.
Although we found some evidence that the more stressed participants were, the more negative affect they experienced in response to negative images, the degree to which participants successfully decreased negative affect using CR did not vary with stress. Stress similarly did not influence electrodermal response to reappraised images. However, it is difficult to interpret the electrodermal effects because we also did not observe the expected increase in response to negative images seen in prior work (Cuthbert, Bradley, & Lang, 1996; Peter J. Lang, Greenwald,
Bradley, & Hamm, 1993). These findings should therefore be interpreted with caution. In examining neural response patterns, we found that in general, stress exerted minimal modulatory influence on neural activity patterns both in a priori ROIs previously implicated in reappraisal processes and across the whole brain. We observed modest stress modulation of reappraisal-
39 related activity in the left amygdala and the right middle frontal gyrus, but these observations did not survive correction for multiple comparisons.
We also measured participants’ heart rate deceleration response while viewing the emotional and neutral images as an index of emotional responding, and tested whether it varied depending on stress. This measure is an index of a parasympathetic orienting response akin to fear bradycardia, and its magnitude increases with stimulus intensity (Bradley et al., 2001;
Campbell et al., 1997). Stress was associated with greater deceleration to neutral pictures and blunted deceleration to negative pictures, including negative pictures that were reappraised.
Previous work has suggested that stress may induce a state of hyper-vigilance characterized by a heightened sensitivity to emotional stimuli at the cost of specificity (Cousijn et al., 2010; van
Marle, Hermans, Qin, & Fernández, 2009). The present findings are consistent with this idea that stress reduces the differentiation between threat-related and non-threat-related stimuli in physiological measures such as heart rate. It should be noted, however, that other measures of emotional responding (electrodermal activity and self-reported affect) did not yield this pattern.
These results ran counter to our hypotheses. CR and stress are believed to engage overlapping neural systems, and previous behavioral work has suggested that these processes interact to influence behavior and physiology (Denson et al., 2014b; Raio et al., 2013; Tsumura et al., 2015). We therefore predicted that stress would reduce the effectiveness of CR by negatively impacting the functioning of biological systems that are jointly modulated by CR and stress. In the following paragraphs, we explore possible reasons why CR was largely robust to the effects of stress in this study and offer directions for future work that would build a more nuanced account of the relationship between stress and emotion regulation.
40 In the present study, we focused on psychological stressors. Although stress in its original and most fundamental conception is a non-specific response (Selye, 1987), it is possible that components of the response which depend on the specific nature of the stressor play a more central part in influencing emotion regulation. For instance, previous work using a physical rather than psychological stressor has found evidence of stress modulation on regulatory processing (Raio et al., 2013). The intensity of induced stress may also play an important role: as with most laboratory-based studies, the intensity of stress achieved in this study is moderate.
Thus, it remains unclear whether these findings would generalize to stressors of high intensity.
In addition, we were interested in eliciting and examining an ecologically valid, multifaceted stress response. Previous studies investigating the effects of stress on CR have focused on one component of the stress response: cortisol. It was specifically cortisol, examined either by isolating individuals who generated a robust cortisol response to a stress induction
(Tsumura et al., 2015) or by directly eliciting a cortisol response using a physiological manipulation (Raio et al., 2013), that was associated with regulatory deficits. Although we observed differences in cortisol responding between the Stress and Control groups, there was substantial variability that may have reduced our ability to isolate the impact of cortisol specifically on CR. Though male and female participants reported experiencing significant and comparable levels of subjective stress, male participants exhibited a substantially greater cortisol response compared to female participants. The sex specificity of the cortisol response, consistent with previous work (Childs et al., 2010; Kirschbaum et al., 1999, 1992), limited statistical power to examine stress effects in only cortisol responders. Future experiments could focus on the specific effects of cortisol on emotion regulation; such studies would benefit from carefully considering their sample demographics and perhaps testing only male participants, who exhibit a
41 more reliable cortisol response. That said, in the present study cortisol and subjective stress showed substantially different individual patterns- female participants reported marked subjective stress but did not exhibit a significant cortisol response. Thus, focus on cortisol- anchored mechanisms will necessarily steer away from inferences that can generalize to the multifaceted construct of “stress”.
It is useful to consider whether stress effects on CR do exist, but they are so subtle that participants could still succeed at the standard cognitive reappraisal paradigm in spite of them.
Likert scales lack the sensitivity to capture small quantitative changes in affect and are not designed to track qualitative changes in emotional experience. In addition, although the present
CR paradigm is the most widely used for neuroimaging, the cognitive processes it examines constitute only one component of what happens when people use CR in daily life. The current study instructed participants to generate one alternative appraisal that effectively reduced their emotional responses. Tsumura and colleagues (2015) instructed participants to generate multiple reappraisals to a negative stimulus under stress and found that cortisol responders generated fewer non-negative reappraisals. However, the relationship between the number of reappraisals generated and CR effectiveness is currently unknown. It is possible that the moderate level of stress evoked in laboratory studies does indeed reduce the quantity of non-negative reappraisals an individual can generate, but the participant is still able to generate at least one viable reappraisal and use it to effectively down-regulate negative affect. Future work could more thoroughly explore whether the content or range of reappraisals is reduced under acute stress, a possibility that the present study was not equipped to evaluate.
CR is believed to be a highly efficacious strategy for regulating emotions, and it is important to understand under what conditions it is effective. Despite mechanistic overlap that
42 suggested otherwise, we found that CR was effective under stress. Though we cannot rule out that there are more subtle or nuanced effects of stress on CR, we demonstrate that a successful multifaceted stress response evoked by a well-validated stress induction did not influence regulation processing during a well-validated CR task. We tentatively conclude that if stress does influence CR, this effect might be less robust or more circumscribed than expected.
Acknowledgements
We thank Kate McLaughlin, Kateri McRae, Jens Pruessner, and Jennifer Silvers for helpful discussion and for sharing study materials, and Alexandra Rodman, Maggie Schell, Efthalia
Kaynor, and Megan Garrad for their help conducting this study. Research reported in this publication was supported through a contract with the U.S. Army Natick Soldier Research,
Development, and Engineering Center (Natick, MA) under award number W911QY-14-C-0009 to L.H.S., by the National Science Foundation (DGE1144152 to M.S.), and by the National
Institutes of Health Shared Instrumentation Grant (S10OD020039).
43 Paper 2: Neurobiological effects of sleep deprivation on cognitive reappraisal
Maheen Shermohammed, Laurel Kordyban, Leah H. Somerville
(In Preparation)
INTRODUCTION
Cognitive reappraisal (CR) is a form of emotion regulation that involves reinterpreting the content of an emotional stimulus in a way that changes its meaning. CR is regarded as an especially potent emotion regulation strategy, more effectively modulating the experiential, physiological, and neurobiological components of an emotional response than attempts to suppress or just passively experiencing the affect (Goldin et al., 2008; Gross, 1998).
Furthermore, the tendency to use CR in daily life predicts fewer depressive symptoms, better interpersonal functioning, and greater psychological well-being (Gross & John, 2003).
Because of these perceived assets, CR has been the subject of great empirical attention.
However, the effectiveness of this regulation strategy has been primarily tested in laboratory environments, free from the challenges that may encumber emotion regulation abilities in the real world. One common such challenge is lack of sleep, a condition faced by almost half of
Americans on a regular basis (Swanson et al., 2011). Sleep loss may be especially relevant to emotion regulation processing. Psychopathologies that critically implicate dysregulated emotion are consistently associated with disrupted sleep. Sleep disturbance can promote the development and worsening of depression and PTSD symptoms (Baglioni et al., 2010; Ford & Kamerow,
1989; Germain et al., 2008; Perlis et al., 1997), and frequently coincides with the occurance of
44 anxiety disorders such as generalized anxiety, panic disorders, and eating disorders (Latzer et al.,
1999; Lauer & Krieg, 2004; Mellman & Uhde, 1989; M. T. Smith et al., 2005). The current study sought to evaluate the neurobiological mechanism behind this putative relationship between sleep and affect dysregulation.
CR has been particularly well studied in the brain. It is believed to draw on domain- general cognitive control systems in the prefrontal and temporo-parietal cortices to modulate activation in limbic regions, like the amygdala (Buhle et al., 2014; Ochsner & Gross, 2005).
Sleep deprivation, meanwhile, has been shown to perturb these same cognitive and affective circuits. First, it impairs executive functions, which rely on similar cognitive control systems.
Sleep deprivation causes deficits in short-term and working memory, and decreases the ability to focus or sustain attention (Durmer & Dinges, 2005; Goel et al., 2009; Krause et al., 2017; Ma et al., 2015). These functional impairments are associated with reductions in brain activation in lateral prefrontal and posterior parietal cortical regions (Chee & Choo, 2004; Lythe et al., 2012;
Ma et al., 2015; Mu et al., 2005).
Second, lack of sleep can heighten affective responding. It is associated with increases in negative emotions and mood as well as rises in depressive and anxiety symptoms in otherwise healthy populations (Babson et al., 2010; Caldwell et al., 2004; Cutler & Cohen, 1979; Dinges et al., 1997; Kahn-Greene et al., 2007; Prather et al., 2013; Sagaspe et al., 2006; Zohar et al., 2005).
These affective changes accompanying sleep deprivation are paralleled by exaggerated amygdala reactivity to emotional stimuli (Greer et al., 2013; Gujar et al., 2011; Motomura et al., 2013; Yoo et al., 2007). Additionally, obtaining sleep can be restorative, reducing subjective affect and amygdalar responding to previously viewed negative stimuli (van der Helm et al., 2011).
45 To our knowledge, two previous studies have directly tested the effects of sleep on CR in a community sample of young adults. Mauss and colleagues (2013) tested how individual differences in self-reported sleep were associated with CR-induced changes in affect in response to sad films, and found that poorer sleep quality was correlated with decreases in CR ability.
However, a separate study testing reappraisal of negative images found that self-reported sleep duration or quality were not related to CR ability or associated neural activation (Minkel et al.,
2012). The conflicting results of these studies remain unresolved, motivating further investigation. Furthermore, although findings from Mauss and colleagues (2013) may suggest that sleep impairs CR, it is just as likely that a person’s inability to effectively reappraise their emotions impairs their sleep (Kahn, Sheppes, & Sadeh, 2013). Indeed, there is evidence that individuals who do not tend to use CR as an emotion regulation strategy are more vulnerable to the effects of lack of sleep on neural markers of sustained attention to negative stimuli (Cote,
Jancsar, & Hunt, 2015). Without experimentally manipulating sleep, these studies necessarily fall short of making critical causal inferences.
In the present study, we experimentally manipulated sleep by inducing 24 hours of sleep deprivation (SD). We employed a within-subjects crossover design to examine participants both under SD and rested-wakefulness (RW). Under both conditions, participants attempted to use CR to regulate their emotional response to aversive images while undergoing fMRI scanning.
Analyses first validated the effectiveness of the sleep manipulation and then measured the degree to which deprivation modulated emotional reactivity and CR success. Our primary hypotheses were that lack of sleep would impair CR success, and that it would do so by impairing the recruitment of prefrontal and parietal brain regions involved in CR while heightening amygdala activity.
46 METHOD
Participants
Thirty-six young adults participated in this study to completion. Data from one participant was excluded due to MRI scanner malfunction and from another due to non-compliance, leaving a final sample of N = 34 (aged 18-30, 17 female). Recruited participants were screened for the following exclusion criteria: history of sleep or neurological disorders, current use of antidepressant or hypnotic medication, current diagnosis of Axis I psychiatric conditions, engagement in shift work within the 3 months before participation, travel to time zones > 3 hours away in within the 3 months before participation, daily consumption of more than ~140 mg of caffeine (approximately 1 cup of brewed coffee), and any contraindications for MRI. In addition, participants were screened for irregular sleep habits, which include typical bedtime before 10pm or after 2am, sleep duration of less than 6.5 hrs or greater than 9.5 hrs, or highly variable bedtimes. Finally, participants were right-handed, nonsmokers, and proficient in English. All participants provided informed written consent for their participation. Research procedures were approved by the Committee on the Use of Human Subjects at Harvard University.
Cognitive reappraisal (CR) task.
Building on prior work, we employed a task frequently used to target cognitive reappraisal processes (Ochsner et al., 2004). In this task, participants viewed negative and neutral images and were instructed to use cognitive reappraisal to decrease their emotional response to half of the negative images (Figure 2.1). Before the MRI session, participants were thoroughly trained on the strategy of cognitive reappraisal. They were given example reappraisals for sample
47 negative images (e.g. “help is on the way”, “it’s just a scene from a movie”, etc.) and generated their own reappraisals aloud to the experimenter over several practice trials until they demonstrated clear understanding of the goals of the task.
During fMRI scanning, a two second instructional cue first informed participants whether they were to passively view (“LOOK”) or reappraise (“DECREASE”) an image that followed.
The image was displayed (8 s), and then participants rated how negative they felt on a scale of 1
(“Not at all bad”) to 5 (“Very bad”) using a button response box held in the right hand (2 s).
Jittered fixation periods (2-8 s) were presented between the image display and the rating scale, and between the rating scale and the next trial. A total of 72 pictures (24 decrease-negative, 24 look-negative, 24 look-neutral) were presented over 3 scanning runs with 24 trials each. Run order was counterbalanced across participants. Images were from the International Affective
Picture System (IAPS; Lang, Bradley, & Cuthbert, 1997) database2 and all included depictions of people. Two sets of negative images were constructed with the same average normative valence and arousal and assigned to the decrease-negative or look-negative condition, counterbalanced across participants. In addition to self-report ratings, heart rate and electrodermal activity were acquired as measures of emotional response to each of the 3 picture conditions.
2 IAPS images used: 2038, 2095, 2100, 2102, 2104, 2120, 2205, 2214, 2305, 2357, 2375.1, 2383, 2385, 2393, 2441, 2455, 2480, 2487, 2490, 2493, 2512, 2575, 2579, 2590, 2595, 2661, 2683, 2691, 2700, 2702, 2703, 2710, 2749, 2750, 2799, 2811, 2840, 2870, 2900, 3160, 3180, 3220, 3280, 3300, 3350, 3500, 4605, 4621, 6211, 6212, 6250, 6311, 6312, 6313, 6530, 6555, 6561, 6821, 6840, 7493, 9007, 9070, 9331, 9341, 9404, 9423, 9424, 9429, 9584, 9903, 9905, 9927
48 How bad do DECREASE you feel? or Not at all Very LOOK bad bad
Figure 2.1 Schematic for trials in the cognitive reappraisal task.
Study Procedure
Participation in this study consisted of three visits. The first visit was an orientation, during which participants consented to study procedures and were given instructions as well as the opportunity to practice the CR Task. Following the orientation, participants were scanned on two separate visits, once under rested wakefulness (RW) and once after approximately 24 hrs of sleep deprivation (SD). The order of the RW and SD visits were counterbalanced across participants
(RW first = 16; SD first = 18). The orientation occurred at least 3 days before any subsequent visit, and the RW and SD sessions occurred at least 1 week apart to prevent any residual effects of SD. For the 3 days preceding each MRI session, participants were instructed to obtain full nights of sleep at their regular schedules and to refrain from taking naps or consuming alcohol or caffeine.
For the RW session, participants were instructed to arrive at the research facility in the morning, within one hour of the time they previously reported typically waking up. Participants were reminded of the tasks they would complete in the scanner and given a chance to practice them. Next, they completed mood and sleepiness assessments and were prepared for scanning.
Mood assessments consisted of the Positive and Negative Affective Scale (PANAS), a 20-item
49 questionnaire that yields positive and negative sub-scores (Watson, Clark, & Tellegen, 1988) and was used to assess changes in affect across visits; the state form of the State-Trait Anxiety
Inventory (Spielberger, Gorsuch, & Lushene, 1970), a 20-item questionnaire that was used to assess changes in anxiety levels across visits; and a question asking participants to rate to what extent they felt stressed “right now” (using the same scale as the PANAS). Sleepiness was assessed using the single-item Stanford Sleepiness Scale (Hoddes, Zarcone, Smythe, Phillips, &
Dement, 1973) as well as the psychomotor vigilance task, a 10 minute task in which participants press a button every time they see a stimulus. The psychomotor vigilance task is highly sensitive to the increases in attentional lapses and slowing of reaction times that accompany sleep deprivation (Dinges & Powell, 1985; Mueller & Piper, 2014). After this task, participants were led to the MRI room and prepared for scanning. During scanning, participants’ eyes were monitored through a live video feed to facilitate wakefulness.
For the SD session, participants arrived at the research facility in the evening, approximately 1 hr before their self-reported bedtime. The facility was a basement lounge with no windows. During the overnight period of this visit, participants were permitted to engage in non-strenuous activities such as reading, watching movies, taking walks, working, and conversing. Participants were provided non-caffeinated snacks (such as chips, cereal, granola bars, juice, etc.) ad-libitum throughout the overnight period. Between 2-4 participants engaged in the overnight session together, during which they were permitted to interact freely and were monitored by a researcher at all times. In the morning, participants were led through the same procedures as in the RW session. MRI scanning for the two sessions were conducted within approximately one hour of one another (mean difference = 18 mins, median = 22, sd = 29.74).
50 Physiological Measures
Heart Rate
Previous research has shown that viewing aversive pictures elicits a brief, parasympathetically mediated deceleration in heart rate (HR), and the magnitude of this deceleration varies by stimulus intensity (Bradley et al., 2001; Campbell et al., 1997). Furthermore, in prior work we have found that this measure was sensitive to the effects of a context manipulation on emotional reactivity (Shermohammed et al., 2017). We therefore measured heart rate during the CR task as a physiological index of emotional responding to pictures.
This measure was acquired from the left ring finger using a wireless pulse oximetry sensor provided with the Siemens Physiological Monitoring Unit. For every participant, the resulting 400 Hz pulsatile signal was converted into a continuous beats-per-minute HR trace using AcqKnowledge software and was visually inspected for high quality. Technical difficulties with the data collection hardware resulted in data loss for several scans. Any participant with at least 3 runs of HR data in both RW and SD sessions was included in these analyses, resulting in n=23 usable datasets. Consistent with our previous protocol, the HR trace was first shifted 3.5 seconds due to the specific shortcomings of using a pulse oximeter rather than an electrocardiography for stimulus-locked measurements (Shermohammed et al., 2017). It should nonetheless be noted that the specific estimation of this delay at 3.5 s may be less accurate because of differences in hardware and processing compared to this previous report. Using the shifted trace, the average HR for the 4.5 s following picture onset was computed, corrected for the average HR during the 2 s preceding.
Eye Recordings
51 In addition to monitoring participants’ eyes during fMRI scanning in real-time, a video of the right eye was recoded for further evaluation. Each second of these videos was manually coded by an experimenter assessing whether the eye was open or closed. Classifications were subsequently used to exclude trials in which participants may have been too sleepy to engage in the CR task.
These were defined as trials for which the participant’s eye was fully closed for a second or more during the cue or picture period, or in which no rating response was made. Such trials were excluded from behavioral analyses and were modeled separately in fMRI analyses. Technical difficulties with eye recoding software resulted in data loss for 3 RW and 2 SD scanning sessions.
Analysis Strategy
Primary analyses evaluated how SD influenced responses during the reappraisal task. On one hand, SD could affect a participant’s ability to use reappraisal effectively. If this was the case,
SD should specifically reduce any differences between the decrease-negative and look-negative conditions, i.e., the “reappraisal contrast.” Another possibility is that SD could make participants more reactive to negative pictures in general. This would result in an exacerbation of any differences between the look-negative and look-neutral conditions, i.e., the “reactivity contrast.”
We therefore conducted two-way ANOVAs to test the effects of sleep session (RW or SD) and the reappraisal contrast (decrease-negative vs. look-negative) or reactivity contrast (look- negative vs. look-neutral) on the outcome variables: picture ratings, neural activation, and heart- rate deceleration during the CR task. Because these ANOVAs only contain 2 of the 3 total conditions, we also report results from a paired t-test with data merged across task conditions to assess the general effects of SD on outcome variables.
52 In addition, descriptive statistics and within-subjects tests were employed to examine participant compliance, validate the effectiveness of the sleep manipulation, and assess changes in mood across visits. All statistical analyses were performed in R 3.2.3 (R Core Team, 2015).
fMRI Analysis
Acquisition & preprocessing
Brain imaging was performed on a 3.0 Tesla Siemens Prisma scanner with a 32-channel head coil (Siemens Medical Systems, Erlangen, Germany) at the Harvard University Center for Brain
Science-Neuroimaging. A T1-weighted high-resolution anatomical image of the brain was acquired using a multiecho multiplanar rapidly acquired gradient-echo (MEMPRAGE) sequence
(176 sagittal slices; repetition time = 2200 ms; multi-echo times = 1.69, 3.55, 5.41, & 7.27 ms; flip angle = 7°; slice thickness = 1 mm; voxel size = 1x1x1 mm). Functional images were collected using an echo planar imaging T2*-weighted sequence sensitive to the BOLD response
(87 axial slices per whole-brain volume, voxel size = 1.7x1.7x1.7 mm, repetition time = 2000 ms, echo time = 28 ms, flip angle = 80°, multi-band acceleration factor = 3). Functional slices were oriented to a slightly greater tilt than the anterior-posterior commissure plane to minimize signal dropout due to sinus cavities.
Functional imaging data were preprocessed using the Functional MRI of the Brain
Software Library (FSL; version 5.0.4; Smith et al., 2004) tools implemented in Nipype (v.
0.11.0; Gorgolewski et al., 2011) using the Lyman interface (v. 0.0.7; http://www.cns.nyu.edu/~mwaskom/software/lyman/). Each functional scan was first realigned to its middle volume, spatially smoothed with a 6mm full-width at half maximum Gaussian kernel, and high-pass filtered at 128 seconds. Functional scans were co-registered to individual-
53 subject anatomical images using bbregister (Freesurfer v. 5.3.1; Greve & Fischl, 2009).
Subsequently, for analyses comparing across participants, statistical maps were first normalized to a Montreal Neurological Institute (MNI) brain template using linear and nonlinear warping methods through the Advanced Normalization Tools software (v. 1.9.x; Avants, Tustison, &
Song, 2009).
First-level modeling of CR task
Preprocessed images were entered into a standard general linear model (GLM) in FSL, which estimated neural responses to: the cue period (collapsed over trial types), the three types of picture periods (decrease-negative, look-negative, and look-neutral), and the rating period
(collapsed over trial types). Regressors used boxcar functions convolved with the canonical double-gamma hemodynamic response function implemented in FSL. The model also included nuisance regressors for motion parameters, temporal derivatives for each regressor of interest, and temporal filter regressors with a cut-off of 128s. To remove additional noise, functional volumes with motion greater than 1mm or whole-brain intensity values greater than 4.5 standard deviations away from the mean were censored from the model as additional regressors; one RW scan run had greater than 10% censored volumes and was excluded from analysis. No scan volume had greater than 5 mm of motion.
In addition, there were some trials for which there was indication that the participant may have been too sleepy to engage in the task (see Eye Recordings section above). Cue, picture, and rating periods for these trials were modeled separately as regressors of no interest. If more than 2 trials of any task condition were flagged in a single scan run, that run was excluded (7 out of 136 scan runs excluded). Parameter estimates from all runs of a given session were entered into a
54 fixed-effects analysis in FSL. Resultant maps were normalized to an MNI template to compare across participants and sessions in subsequent analyses.
Region-of-interest (ROI) analysis
Unbiased, a priori ROIs were identified using a previous meta-analysis of 48 neuroimaging studies of reappraisal (Buhle et al., 2014). Eleven 6mm spheres were centered on activation peaks from distinct regions reported for the reappraise>emotional baseline contrast in the meta- analysis, in addition to two 4mm spheres in the bilateral amygdalae from the emotional baseline>reappraise contrast reported in the meta-analysis (see Table 2.1 for ROI coordinates).
Mean parameter estimates were extracted for the picture period from each ROI and tested for task and stress effects. P-values from these tests were corrected for 13 comparisons by controlling for the false discovery rate using the Benjamini–Hochberg method.
Table 2.1 ROI Analyses for the Effects of Sleep Deprivation on Reappraisal & Reactivity Main Effect Interaction with Sleep ROI Side MNI Coordinates Reappraisal Reactivity Sleep Reappraisal Reactivity x y z Corrected p Corrected p (Uncorrected p) Inferior frontal gyrus R 60 24 3 <.001** .012* .343 .909 (.793) .865 (.266) Middle frontal gyrus R 51 15 48 <.001** .115 .042* .909 (.619) .890 (.434) Medial frontal gyrus R 9 30 39 .001** .115 .338 .909 (.781) .890 (.507) Anterior cingulate gyrus L -3 24 30 <.001** .115 .460 .909 (.824) .890 (.606) Superior frontal gyrus L -9 12 69 <.001** .016* .183 .909 (.840) .972 (.921) Middle frontal gyrus L -33 3 54 <.001** .001** .119 .950 (.950) .258 (.035*) Anterior insula L -36 21 -3 <.001** .001** .392 .909 (.747) .258 (.040*) Inferior frontal gyrus L -42 45 -6 <.001** .034* .574 .909 (.209) .972 (.970) Superior temporal gyrus R 63 -51 39 .011* .921 .784 .909 (.575) .972 (.972) Angular gyrus L -42 -66 42 <.001** .627 .343 .909 (.603) .890 (.510) Middle temporal gyrus L -51 -39 3 <.001** .115 .460 .909 (.679) .865 (.240) Amygdala R 30 -3 -15 .704 .390 .001** .909 (.635) .890 (.616) Amygdala L -18 -3 -15 .114 .994 .055† .909 (.772) .972 (.892) Each ROI was subject to a separate ANOVA testing for the effect of sleep condition on reappraisal or reactivity. Both FDR-corrected (correcting for the number of ROIs) and uncorrected p-values are reported for each interaction. Main effects of sleep are reported from a paired t-test collapsed across picture type. Region labels were based on the Harvard-Oxford cortical and subcortical atlases, cross-referencing Mai, Paxinos, & Voss (2008). p<0.1†, p<0.05*, p<0.005**.
55
Whole-brain fMRI analysis
To supplement ROI analyses, we also performed a random effects analysis focused on the picture period across the whole brain. We first wanted to corroborate established findings with this task irrespective of sleep condition. To do this, contrast maps from one-tailed t-tests examining reappraisal (decrease-negative > look-negative) and emotional reactivity (look- negative > look-neutral) from each session were first entered into separate fixed-effects analyses for each subject. The resulting map for each participant was then entered into a group level mixed-effects analysis implemented in FEAT (Woolrich, Behrens, Beckmann, Jenkinson, &
Smith, 2004), producing maps of brain activity during the task irrespective of sleep status. Next, to test the primary question of the effects of sleep condition on reappraisal and emotional reactivity, paired t-tests comparing the SD and RW sessions were computed for these contrasts.
All contrast maps were corrected at a family-wise error (FWE) threshold of p<0.05 using FSL’s cluster-based correction and an initial threshold of z>3.1.
RESULTS
Compliance & Sleep Manipulation Check
Participants complied with instructions to obtain a full night of sleep before the RW session, as evidenced by self-reported sleep duration estimates (7.58 +/- 0.75 hr), although these estimates were higher than those estimated by actigraphy (6.62 +/- 1.15 hr). Participants also complied with instructions on the night before the SD scan, getting 0 hr of sleep as confirmed by experimenter monitoring.
56 As expected, the SD manipulation successfully induced sleepiness and impairment on the psychomotor vigilance task. One participant did not complete the self-reported sleepiness ratings during the RW visit, leaving a final sample of N=33 for this measure. As expected, compared to
RW, during SD participants rated themselves as sleepier (t(32) = -11.33, p < 0.001; Figure 2.2A).
Psychomotor vigilance task reaction times were first log transformed to mitigate skew.
Participants exhibited significantly slower reaction times on the psychomotor vigilance task under SD compared to RW (t(33) = -5.54, p < 0.001; Figure 2.2B).
A B Sleepiness Psychomotor Vigilance Task
6 6.10
●
● 5 6.05
4 6.00 Rating 3 5.95 Mean Reaction Time (log) 2 ● 5.90
●
1 5.85
Rested Sleep Rested Sleep Wakefulness Deprivation Wakefulness Deprivation Figure 2.2 SD successfully induced sleepiness, as assessed by self-report on the Stanford Sleepiness Scale (A), and slowed reaction time on the psychomotor vigilance task (B). Error bars represent within-subject SE.
Changes in Baseline Affective Measures in Response to SD
One participant did not complete the self-reported stress rating during the RW visit, leaving a final sample of N=33 for this measure. As expected, sleep deprivation resulted in an increase in self-reported stress (RW: mean = 1.30, SEM = 0.11; SD: mean = 2.09, SEM = 0.17; t(32) = 3.88, p < 0.001), anxiety (RW: mean = 32.18, SEM = 1.35; SD: mean = 41.94, SEM = 1.71; t(33) =
57 6.82, p < 0.001), and negative affect (RW: mean = 12.12, SEM = 0.44; SD: mean = 14.59, SEM
= 0.59; t(33) = 4.43, p < 0.001) and a decrease in positive affect (RW: mean = 2441, SEM =
1.28; SD: mean = 19.73, SEM = 1.25; t(33) = -3.44, p = 0.002).
Effects of SD on Emotional Reactivity & CR
To evaluate the general effectiveness of the task manipulation, we first evaluated main effects of picture valence and reappraisal instruction on affect ratings. As expected, the negative pictures were rated significantly more negatively than the neutral pictures (F(1,99) = 744.85, p < 0.001).
A significant main effect of picture condition (F(1,99) = 135.84, p < 0.001) indicated that consistent with prior work, participants successfully reduced their negative reaction to negative pictures using reappraisal strategies compared to passive viewing.
The critical tests evaluated whether SD modulated reappraisal success and reactivity levels (Figure 2.3). Evaluating interactions with sleep condition revealed a trending interaction between sleep condition and picture type on emotional reactivity (F(1,99) = 2.86, p = 0.094), such that participants exhibited marginally attenuated emotional reactivity under SD. There was no significant interaction between sleep condition and picture type on reappraisal success
(F(1,99) = 1.93, p = 0.167). Finally, a paired t-test collapsed across picture type showed no effect of sleep condition on affect ratings more generally (t(33) = 0.89, p = 0.378).
58 Figure 2.3 Reappraisal success and emotional reactivity in the stress and control groups. Reappraisal success is defined as the difference in self- reported affect ratings for look-negative > decrease-negative pictures. Emotional reactivity is defined as the difference in self-reported affect ratings for look- negative > look-neutral pictures. Error bars represent SEM.
Effects of SD on Neural Signatures of Emotional Reactivity & CR
ROI Analysis
Parallel to the behavioral analyses above, separate ANOVAs were conducted on each ROI testing for main effects of task condition, and interactions between task condition and sleep condition on fMRI responses. Results are summarized in Table 2.1. Consistent with meta- analytic evidence localizing the neural correlates of reappraisal, all cortical ROIs exhibited greater activation during reappraisal than while passively viewing negative images. Contrary to our predictions, no ROI showed a significant interaction between sleep condition (RW, SD) and reappraisal condition, even before correcting for multiple comparisons.
With respect to emotional reactivity, which is measured using a comparison of look- negative versus look-neutral picture types, there was a main effect of picture type in the left
59 anterior insula, left inferior frontal gyrus, left middle frontal gyrus, right inferior frontal gyrus, and left superior frontal gyrus; each of these regions exhibited greater activation when participants were viewing negative compared to neutral pictures. Like the reappraisal contrast, no interactions between sleep and reactivity condition survived correction for multiple comparisons.
However, to provide the most comprehensive description of the data, we also report uncorrected results for this analysis. Without correction, there was a significant interaction between sleep and reactivity condition in the left anterior insula and left middle frontal gyrus. In both of these ROIs, differences in activation due to emotional reactivity during RW were blunted during SD.
Finally, we examined the main effect of sleep irrespective of task condition to assess the general effects of SD on neural activation in each ROI. There was a significant effect of sleep condition on activation in the right amygdala and the right middle frontal gyrus as well as a trending effect in the left amygdala. For each of these ROIs, SD resulted in reduced overall activation compared to RW. Statistics from this analysis in each ROI are reported in Table 2.1.
Whole-brain Analysis
We supplemented targeted ROI analyses with exploratory whole-brain comparisons at whole- brain p<0.05 FWE thresholding. Consistent with prior work (Buhle et al., 2014), group analyses
(where participants’ RW and SD sessions were collapsed) revealed that reappraisal (decrease- negative>look-negative) recruited an extensive network of lateral and medial prefrontal regions, as well as areas in the posterior temporal and parietal cortex (Figure 2.4; Supplementary Table
S1). Emotional reactivity (look-negative>look-neutral) also recruited regions consistent with our previous findings (Shermohammed et al., 2017), including the bilateral anterior insula
(Supplementary Table S1).
60 Examining the effects of sleep condition on emotional reactivity revealed activation differences bilaterally in the insula and middle temporal gyrus, as well as regions in occipital and parietal cortices (Table 2.2). In previous work, these regions have coded for negative affect, exhibiting greater activation for negative compared to neutral picture viewing (Shermohammed et al., 2017). Participants in the present study exhibited the same pattern under RW. However SD disrupted this pattern; it either extinguished any reactivity differences (in the case of the precuneus and the right middle temporal gyrus), or resulted in greater activation to neutral compared to negative pictures (in the case of the left middle temporal gyrus). There were no effects of sleep condition for the reappraisal contrast.
3.1 6 Z
Figure 2.4 Whole-brain fMRI analyses examining reappraisal (decrease-negative > look- negative) revealed the extensive network of prefrontal, temporal, and posterior parietal regions expected from previous work (Buhle et al., 2014). Images are p < .05, FWE corrected.
TableTable 2 .2.2 BrainBrain Regions Regions Recruited Recruited During During Emotional Emotional Reactivity Reactivity for RW for> SD RW > SD MNI coordinates Region Side Voxel Extent Max Z x y z Middle Temporal Gyrus R 1155 6.16 60 -46 14 Middle Temporal Gyrus L 1123 6.08 -58 -48 8 Precuneus L 1028 5.66 -8 -60 42 Parietal Lobule L 522 4.83 -54 -34 50 Insula R 355 5.26 34 26 0 Insula L 351 4.99 -40 18 -4 Lateral Occipital Cortex L 270 4.65 -30 -72 20
Brain regions that were more activated during emotional reactivity (look-negative > Brain regions that were more activated during emotional reactivity (look-negative > look-neutral) during RW than look-neutral) during RW than SD. Reported are peak coordinates. SD. Reported are peak coordinates.
61
Effects of SD on Heart Rate Changes in Response to Pictures
Baseline-corrected HR provided a physiological index of emotional responding. A paired t-test collapsed across picture type showed that, in general, this HR measure was lower under SD compared to RW (t(22) = 2.62, p = 0.016). There was no effect of reappraisal on HR (F(1,66) =
0.004, p = 0.948) and no interaction between reappraisal and sleep condition (F(1,66) = 0.151, p
= 0.699).
There was, however, a significant interaction between emotional reactivity and sleep condition on baseline-corrected HR (F(1,66) = 4.28, p = 0.042). Compared to neutral pictures, participants exhibited decreased HR in response to negative pictures under RW. However there was no such valence difference under SD; instead, participants exhibited decreased HR in response to both negative and neutral pictures relative to the preceding baseline cue period. Thus
SD appears to have extinguished the valence-specificity of the HR deceleration response that is typically observed for negative pictures.
DISCUSSION
This study tested the effects of sleep deprivation on CR and its underlying neural processes.
Sleep deprivation resulted in expected increases in sleepiness, impairments in cognitive functioning, and elevations in baseline negative affect, state anxiety, and stress. Findings from the CR task were also highly consistent with previous work. Participants self-reported greater negative affect in response to negative pictures compared to neutral pictures, and these affect ratings were attenuated when participants were instructed to use CR. Furthermore, CR recruited
62 a network of prefrontal, temporal, and posterior parietal brain regions that was strongly convergent with meta-analytic evidence from prior work (Buhle et al., 2014).
Despite the success of these manipulation checks, the effects of sleep deprivation on the
CR task were limited. Negative affect ratings in response to negative compared to neutral pictures, an index of emotional reactivity, showed a small but non-significant decrease under deprivation. However, the crucial test of the effects of sleep on CR revealed similar reductions in negative affect when participants were reappraising their emotions under rested wakefulness and under sleep deprivation. In the brain, we targeted our analyses on regions of interest consistently shown to be involved in reappraisal. We found overall reductions of neural activation in the right middle frontal gyrus and bilateral amygdalae in response to sleep deprivation. However, these effects were not specific to any condition. We found no evidence for neurobiological effects of the sleep manipulation on emotional reactivity or CR in a priori regions known to be involved in reappraisal. With CR in particular, we did not even find evidence for modest effects under more liberal statistical criteria.
These findings ran contrary to our predictions. First, with respect to basic emotional reactivity, we expected to observe an increase in negative affect ratings in response to negative pictures. We expected this because sleep loss has been associated with increased negative emotions and heightened amygdala responding to affective stimuli (Motomura et al., 2013;
Prather et al., 2013; Yoo et al., 2007; Zohar et al., 2005). However, the present findings are at least partially consistent with another theoretical account that suggests that the heightened sensitivity to affective stimuli associated with sleep loss may be accompanied by a decrement in specificity, which would in turn result in indiscriminate emotional responding to potentially non- emotional stimuli (Goldstein & Walker, 2014). This model is supported by previous work
63 showing that sleep deprivation impairs the discrimination between threatening and non- threatening cues and increases negative evaluation of neutrally valenced stimuli (Goldstein-
Piekarski et al., 2015; Tempesta et al., 2010). One previous correlational study found a similar weak trend between poor self-reported sleep quality and decreased subjective emotional reactivity in the present task (Minkel et al., 2012).
A specificity-focused account is also supported by additional evidence in the present work. In the brain, we observed marked effects of sleep deprivation on emotional reactivity in the precuneus and the right middle temporal gyrus, as well as weak effects in the left middle frontal gyrus and the left anterior insula. Similar to the pattern observed with self-report, clear differences in activation in these regions due to emotional reactivity under rested wakefulness were blunted under deprivation. Finally, we examined a physiological index of emotional responding to images, heart rate deceleration. This measure reflects a parasympathetic orienting response akin to fear bradycardia, and its magnitude increases with stimulus intensity (Bradley et al., 2001; Campbell et al., 1997). Under rested wakefulness, we observed the expected pattern of a prominent deceleration of heart rate in response to negative but not neutral pictures. However, sleep deprivation abolished such valence differences, producing a deceleration in response to neutral pictures as well. Together, these findings provide preliminary evidence that sleep loss may lessen the degree to which someone is able to discriminate between negative or salient stimuli and neutral or non-salient stimuli.
The lack of sleep deprivation effects on CR presents another puzzle. CR is believed to draw on general cognitive control systems to regulate affective responding (Buhle et al., 2014;
Ochsner & Gross, 2005). Sleep deprivation has reliably been shown to reduce neural activity in these systems and to impair executive functions that rely on them, like working memory and
64 attention (Chee & Choo, 2004; Durmer & Dinges, 2005; Goel et al., 2009; Krause et al., 2017;
Lythe et al., 2012; Ma et al., 2015; Mu et al., 2005). We therefore expected that sleep loss would hinder the effective mobilization of these regions to down-regulate negative affect. However, we saw no evidence of such sleep-related impairments in self-reported, physiological, or neural indices of CR. One interpretation of these findings is that the ability to implement CR could be robust to the effects of sleep deprivation.
We cannot rule out the possibility that participants were simply not sleep deprived enough. After all, staying awake for an entire night is likely not common behavior among healthy adults, and it is possible that more ecologically reflective mild sleep restriction that accumulates over time may actually be more potent. However, Dongen and colleagues (2003) systematically characterized the level of cognitive impairment under different amounts of sleep deprivation and restriction, and found that one night of total sleep deprivation caused similar levels of impairment as 2 weeks of sleeping 4-6 hours a night. Furthermore, the reliably observed cognitive deficits that motivated our hypotheses were largely founded on studies that also employed a single night of total sleep deprivation, as this is the most common method of experimentally manipulating sleep levels (Goel et al., 2009; Ma et al., 2015).
It is also important to remember that although the task we employed to measure CR is widely used, it only assess a limited part of what implementing reappraisal in daily life entails.
This task was designed to assess a participant’s ability to generate and effectively implement at least one reappraisal of an affective image. In daily life, a person must choose to use CR in the first place, and subsequently may generate a number of reappraisals that vary in content. It is possible that lack of sleep may modulate these other components of CR rather than the aspect of
CR ability that this task is constructed to measure. Indeed, previous work has suggested that
65 people may be less inclined to choose CR as a strategy when a negative situation is of high intensity, which sleep deprivation may promote (Sheppes et al., 2011). However, this explanation does not resolve the inconsistency between the results of the present study and that of Mauss and colleagues (2013), who found a negative relationship (albeit correlational) between sleep quality and regulation success on a different task also designed to assess CR ability.
One key difference between the present study and Mauss et al. (2013) is the use of IAPS images compared to sad movie clips to induce an emotional response. Previous work has found that sleep deprivation produces increases in baseline levels of negative affect and anxiety that are reliable and robust (Dinges et al., 1997; Goldstein & Walker, 2014; Minkel et al., 2012; Reddy,
Palmer, Jackson, Farris, & Alfano, 2017), and the few studies that have assessed emotional responding to stimuli in daily life have found the same pattern (Gordon & Chen, 2014; Zohar et al., 2005). However, these emotional changes are not reflected in participants’ response to IAPS pictures; rather, as in the present study, evidence of changes in emotional reactivity has been weak or mixed (Minkel et al., 2012; Palmer & Alfano, 2017; Reddy et al., 2017). Such findings beg the question of what, if any, real-world phenomena these emotional picture ratings generalize to. It may be the case that the fuller, richer emotional experience provided by a sad movie clip is needed to accurately assess the effects of sleep loss on emotions that we experience in daily life.
The mere observations that lack of sleep impairs executing functioning and heightens affective responding have often been cited as evidence that emotion regulation is impaired as well (Goldstein & Walker, 2014; Gruber & Cassoff, 2014). The present study served as a critical test of this assumption, experimentally manipulating sleep to examine behavioral and neurobiological responses to a form of emotion regulation that has been hailed as particularly
66 effective, using a widely employed laboratory task. We found no evidence of impairment of CR in response to sleep deprivation. Perhaps just as likely as providing evidence that CR could be robust to sleep deprivation, these results may betray the limitations of this task in reflecting real- world CR ability.
Acknowledgements
We thank Kate McLaughlin, Randy Buckner, and Chuck Czeisler for helpful discussion and
Gina Falcone, Hyun Young Cho, Asi Graham, Katya Kabotyanski, Amma Ababio, and Biniam
Andargie for their help conducting this study. Research reported in this publication was supported by the National Science Foundation (DGE1144152 to M. S.) and the National
Institutes of Health Shared Instrumentation Grant (S10OD020039).
67 Paper 3: Neurobiological effects of sleep deprivation on affect labeling
Maheen Shermohammed and Leah H. Somerville
(In Preparation)
INTRODUCTION
Effective emotion regulation, which involves influencing an emotional response to conform to one’s goals, can be highly dependent on the regulatory circumstances. One common such circumstance is insufficient sleep, a condition almost half of Americans suffer from regularly
(Swanson et al., 2011). Poor sleep quality has been associated with impaired ability to implement otherwise effective emotion regulation strategies (Baum et al., 2014; Mauss et al., 2013).
Furthermore, disorders implicating emotion regulation are frequently accompanied by sleep disturbance. Sleep disruption can precipitate the development and exacerbation of depression and
PTSD symptoms (Baglioni et al., 2010; Ford & Kamerow, 1989; Germain et al., 2008; Perlis et al., 1997), and individuals with anxiety disorders such as generalized anxiety, panic disorders, and eating disorders routinely exhibit sleep dysregulation (Latzer et al., 1999; Lauer & Krieg,
2004; Mellman & Uhde, 1989; M. T. Smith et al., 2005). Therefore, it is important to understand the mechanism by which sleep deprivation may interact with emotion regulation processing.
Sleep deprivation is believed to impact the cognitive and affective circuits underlying emotion regulation. It impairs executive functioning, including cognitive control systems that are often essential for the initiation and implementation of emotion regulation. Sleep deprivation reduces the ability to both focus and sustain attention (Durmer & Dinges, 2005; Goel et al.,
68 2009; Ma et al., 2015) and causes deficits in short-term and working memory (Goel et al., 2009;
Krause et al., 2017). These sleep-loss-induced impairments in working memory and attention are associated with reductions in brain activation in cortical regions including, among others, regions in the lateral prefrontal cortex (Lythe et al., 2012; Ma et al., 2015; Mu et al., 2005).
In addition, sleep loss can amplify the very affective signals that are targeted during emotion regulation, and is associated with an increase in negative mood, negative emotional responses to daytime disruptive events, and depressive and anxiety symptoms in non-clinical populations (Babson et al., 2010; Caldwell et al., 2004; Cutler & Cohen, 1979; Dinges et al.,
1997; Kahn-Greene et al., 2007; Prather et al., 2013; Sagaspe et al., 2006; Zohar et al., 2005). In the brain, loss of sleep is linked to exaggerated amygdala reactivity to affective stimuli (Greer et al., 2013; Gujar et al., 2011; Motomura et al., 2013; Yoo et al., 2007), and obtaining sleep can reduce amygdalar and subjective emotional responses to previously viewed negative stimuli (van der Helm et al., 2011).
Despite these findings and evidence linking sleep disruption and emotion regulation impairment, the neural systems underlying interactions between sleep and emotion regulation have not been directly tested. We examine these neural systems in the present study using affect labeling, an implicit emotion regulation strategy. Many well-studied emotion regulation strategies, such as cognitive reappraisal, distraction, distancing, and response suppression, are explicit; they are conscious, deliberate, and often effortful (Gyurak et al., 2011). Implicit strategies, on the other hand, tend to be more efficient and require less effort, occurring outside of conscious awareness (Gyurak et al., 2011). It is useful to consider both explicit and implicit strategies when drawing broader conclusions about emotion regulation and its neural correlates
(Berkman & Lieberman, 2009). We chose to focus on an implicit strategy in the present report;
69 however, see Paper 2 (this dissertation) for an examination of this question using a widely studied explicit strategy.
Affect labeling is the act of putting feelings into words. The beneficial effects of verbalizing one’s emotions have long been demonstrated by the success of interventions like talk therapy and expressive writing (Greenberg, 2004; Pennebaker, 1997). These benefits of translating affect to language can extend to both labeling one’s own feelings or the emotional aspects of an external affective stimulus (McRae, Taitano, & Lane, 2010; Torre & Lieberman,
2018). More recent laboratory work has found that even the simple act of selecting an affect label for an emotionally evocative stimulus reduces its psychological and physiological impact
(Constantinou et al., 2014; Lieberman et al., 2011; Niles et al., 2015; Tabibnia et al., 2008).
Although this may not seem like a traditional form of emotion regulation, it is demonstrated to be similarly effective at reducing self-reported and physiological emotional responses as more explicit strategies like reappraisal and distraction (Kircanski et al., 2012; Lieberman et al., 2011).
Interestingly, participants report these changes despite predicting that labeling will not have regulatory benefits, consistent with its categorization as an incidental or implicit form of emotion regulation (Lieberman et al., 2011).
In the brain, labeling affect-laden stimuli such as emotional faces has been demonstrated to reduce activation in the amygdala (Burklund et al., 2015; Costafreda et al., 2008; Lieberman et al., 2007; S. E. Taylor et al., 2006) and increase activation in the right ventrolateral prefrontal cortex (vlPFC), a region implicated in top-down emotion regulation and control (Mitchell, 2010).
The interplay between these structures may especially be of interest. Previous work has found that vlPFC-amygdala coupling facilitates the ameliorative effects of emotion regulatory manipulations like mindfulness-based intervention (Hölzel et al., 2013). It should be noted,
70 however, that findings regarding the direction of this coupling remain mixed, as some work with affect labeling has instead found inverse vlPFC-amygdala connectivity during regulatory processing (Foland et al., 2008; Hariri, Bookheimer, & Mazziotta, 2000; Payer, Baicy,
Lieberman, & London, 2012). Meanwhile, in addition to its involvement in affect labeling, modulation of the right vlPFC has been associated with cognitive impairment resulting from sleep deprivation (Lythe et al., 2012; Mu et al., 2005). Furthermore, REM sleep deprivation in particular results in an increase in emotional reactivity and a maintenance rather than blunting of vlPFC activation in response to previously observed emotional stimuli (Rosales-Lagarde et al.,
2012).
These findings suggest that sleep deprivation may interfere with or tax the functioning of regulatory systems implicated in emotion regulation through affect labeling. In the present study, we sought to directly test the effects of sleep deprivation on implicit regulatory processes by employing a within-subjects crossover design. Using fMRI scanning, we examined neural responses to affect labeling compared to a non-emotional control labeling task, both after participants had experienced 24 hr of total sleep deprivation (SD) and during rested wakefulness
(RW). Analyses first validated the effectiveness of the SD manipulation and then examined changes in affect labeling processing in the amygdala and vlPFC compared to RW. This work aims to expand the understanding of sleep-associated deficits in emotion regulation by examining the interaction between the neurobiological processes underlying SD and emotion regulation via affect labeling.
71 METHOD
Participants
Thirty-six young adults participated in this study to completion. Data from one participant was excluded due to MRI scanner malfunction and from another due to non-compliance, leaving a final sample of N = 34 (aged 18-30, 17 female). Recruited participants were screened for the following exclusion criteria: history of sleep or neurological disorders, current use of antidepressant or hypnotic medication, current diagnosis of Axis I psychiatric conditions, engagement in shift work within the 3 months before participation, travel to time zones > 3 hours away in within the 3 months before participation, daily consumption of more than ~140 mg of caffeine (approximately 1 cup of brewed coffee), and any contraindications for MRI. In addition, participants were screened for irregular sleep habits, which include typical bedtime before 10pm or after 2am, sleep duration of less than 6.5 hrs or greater than 9.5 hrs, or highly variable bedtimes. Finally, participants were right-handed, nonsmokers, and proficient in English. All participants provided informed written consent for their participation. Research procedures were approved by the Committee on the Use of Human Subjects at Harvard University.
Affective Labeling Task
Implicit emotion regulation via affective labeling was indexed using a variant of an established fMRI paradigm (Lieberman et al., 2007). Three task conditions (affect label, gender label, and shape match) were administered as blocks over four fMRI scanning runs, with one of each block type in each run.
72 In the affect label condition, participants viewed pictures of faces displaying emotional expressions and were instructed to indicate what emotion was displayed on the face by selecting one of two emotion label options. In the gender label condition, participants viewed similar face stimuli but instead selected which of two name options was consistent with the gender of the person pictured. Finally, the shape match condition served as a low-level motor control task, in which participants chose which of a pair of shapes at the bottom of the screen was identical to a target shape displayed at the top of the screen.
Each block began with an instruction screen (3s) followed by 5 task trials (5s each), and was succeeded by an inter-block interval of 8s (see Figure 3.1). Face stimuli were drawn from a standardized set of images (Tottenham et al., 2009). Each block consisted of at least one and no more than two happy, fearful, and angry faces. The assignment of blocks of images to task conditions (affect and gender label) and sessions (RW & SD) was counterbalanced across participants, as was the order of conditions. No image was displayed to given participant more than once.
A B C Affect Label Gender Label Shape Match
Happy Angry Helen Aaron
Figure 3.1 Schematic for the 3 trial types in the affective labeling task.
73 Study Procedure
Participation in this study consisted of three visits. The first visit was an orientation, during which participants consented to study procedures and were given instructions as well as the opportunity to practice the Affective Labeling Task. Following the orientation, participants were scanned on two separate visits, once under rested wakefulness (RW) and once after approximately 24 hrs of sleep deprivation (SD). The order of the RW and SD visits were counterbalanced across participants (RW first = 16; SD first = 18). The orientation occurred at least 3 days before any subsequent visit, and the RW and SD sessions occurred at least 1 week apart to prevent any residual effects of SD. For the 3 days preceding each MRI session, participants were instructed to obtain full nights of sleep at their regular schedules and to refrain from taking naps or consuming alcohol or caffeine.
For the RW session, participants were instructed to arrive at the research facility in the morning, within one hour of the time they previously reported typically waking up. Participants were reminded of the tasks they would complete in the scanner and given a chance to practice them. Next, they completed mood and sleepiness assessments and were prepared for scanning.
Mood assessments consisted of the Positive and Negative Affective Scale (PANAS), a 20-item questionnaire that yields positive and negative sub-scores (Watson et al., 1988) and was used to assess changes in affect across visits; the state form of the State-Trait Anxiety Inventory
(Spielberger et al., 1970), a 20-item questionnaire that was used to assess changes in anxiety levels across visits; and a question asking participants to rate to what extent they felt stressed
“right now” (using the same scale as the PANAS). Sleepiness was assessed using the single-item
Stanford Sleepiness Scale (Hoddes et al., 1973) as well as the psychomotor vigilance task, a 10 minute task in which participants press a button every time they see a stimulus. The psychomotor
74 vigilance task is highly sensitive to the increases in attentional lapses and slowing of reaction times that accompany sleep deprivation (Dinges & Powell, 1985; Mueller & Piper, 2014). After this task, participants were led to the MRI room and prepared for scanning. During scanning, participants’ eyes were monitored through a live video feed to facilitate wakefulness.
For the SD session, participants arrived at the research facility in the evening, approximately 1 hr before their self-reported bedtime. The facility was a basement lounge with no windows. During the overnight period of this visit, participants were permitted to engage in non-strenuous activities such as reading, watching movies, taking walks, working, and conversing. Participants were provided non-caffeinated snacks (such as chips, cereal, granola bars, juice, etc.) ad-libitum throughout the overnight period. Between 2-4 participants engaged in the overnight session together, during which they were permitted to interact freely and were monitored by a researcher at all times. In the morning, participants were led through the same procedures as in the RW session. MRI scanning for the two sessions were conducted within approximate one hour of one another (mean diff = 18 mins, median = 22, sd = 29.74).
Sleep Measures for RW
Sleep behavior for the night before the RW scan was assessed using ambulatory actigraphy wrist monitoring (Actiwatch, Philips Respironics, Bend, OR, USA). The Actiwatch uses a motion sensor to predict sleep and wake times. These data were manually quality checked and combined with real-time sleep and wake information provided by participants (by pressing a button on the wrist monitor) to generate estimates of sleep duration. Technical difficulties with wrist monitors resulted in loss of actigraphy data for 5 participants. To supplement actigraphy estimates, on the morning of the RW visit participants self-reported their bed and wake times for the night before.
75
Behavioral Analyses
The primary outcome measures of interest from the Affect Labeling Task are results from fMRI analyses. We nonetheless calculated overall performance accuracy, the percent of responses in which the correct label was selected, and reaction time (RT). In general, participants perform the task with high accuracy and were similarly expected to do so here. The gender labeling condition serves as a control for the affect labeling, and ideally these judgments would take a similar amount of time to make. However, some previous work has shown that RT may differ between these conditions (Lieberman et al., 2007). We therefore tested for differences in RT between task conditions, and whether these RT differences interacted with sleep condition (RW vs. SD) using a two-factor within-subjects ANOVA. RT values were log transformed to mitigate a skewed distribution, and a single value was removed that was too low to be biologically plausible (<1 ms).
Descriptive statistics and within-subjects tests were employed to examine participant compliance, validate the effectiveness of the sleep manipulation, and assess changes in mood across visits.
fMRI Analysis
Acquisition & preprocessing
Brain imaging was performed on a 3.0 Tesla Siemens Prisma scanner with a 32-channel head coil (Siemens Medical Systems, Erlangen, Germany) at the Harvard University Center for Brain
Science-Neuroimaging. A T1-weighted high-resolution anatomical image of the brain was acquired using a multiecho multiplanar rapidly acquired gradient-echo (MEMPRAGE) sequence
76 (176 sagittal slices; repetition time = 2200 ms; multi-echo times = 1.69, 3.55, 5.41, & 7.27 ms; flip angle = 7°; slice thickness = 1 mm; voxel size = 1x1x1 mm). Functional images were collected using an echo planar imaging T2*-weighted sequence sensitive to the BOLD response
(87 axial slices per whole-brain volume, voxel size = 1.7x1.7x1.7 mm, repetition time = 2000 ms, echo time = 28 ms, flip angle = 80°, multi-band acceleration factor = 3). Functional slices were oriented to a slightly greater tilt than the anterior-posterior commissure plane to minimize signal dropout due to sinus cavities.
Functional imaging data were preprocessed using the Functional MRI of the Brain
Software Library (FSL; version 5.0.4; Smith et al., 2004) tools implemented in Nipype (v.
0.11.0; Gorgolewski et al., 2011) using the Lyman interface (v. 0.0.7; http://www.cns.nyu.edu/~mwaskom/software/lyman/). Each functional scan was first realigned to its middle volume, spatially smoothed with a 6mm full-width at half maximum Gaussian kernel, and high-pass filtered at 128 seconds. Functional scans were co-registered to individual- subject anatomical images using bbregister (Freesurfer v. 5.3.1; Greve & Fischl, 2009).
Subsequently, for analyses comparing across participants, statistical maps were first normalized to a Montreal Neurological Institute (MNI) brain template using linear and nonlinear warping methods through the Advanced Normalization Tools software (v. 1.9.x; Avants, Tustison, &
Song, 2009).
First-level modeling of affective labeling task
Preprocessed images were entered into a standard general linear model in FSL. Neural responses were modeled in a block design, with a separate regressor for each of the 3 task condition blocks
(affect label, gender label, and shape match). Instruction screens were modeled with the task
77 trials that succeeded them. Regressors used boxcar functions convolved with the canonical double-gamma hemodynamic response function implemented in FSL. The model also included nuisance regressors for motion parameters, temporal derivatives for each regressor of interest, and temporal filter regressors with a cutoff of 128 sec. To remove additional noise, functional volumes with motion greater than 1 mm or whole-brain intensity values greater than 4.5 SDs away from the mean were censored from the model as additional regressors; two scans had greater than 10% censored volumes and were excluded from analysis. No scan volume had greater than 5 mm of motion. Finally, to guard against including data in which participants were too sleepy to engage in the task, scans in which participants responded to fewer than 3 out of 5 trials in any block were excluded from analysis (two scan runs excluded). All participants had at least three of four runs of usable data in each visit. Parameter estimates from all runs of a given session were entered into a fixed-effects analysis in FSL. Resultant maps were normalized to an
MNI template to compare across participants and sessions in subsequent group-level analyses.
Higher-level modeling
We first wanted to establish where in the brain was distinctly sensitive to affect compared to gender labeling (irrespective of sleep condition). To examine this, affect>gender labeling contrast maps from each session for a given subject were first entered into a fixed-effects analysis. The resulting map for each participant was then entered into a group level mixed- effects analysis implemented in FEAT with FLAME1+2 (Woolrich et al., 2004). Because we had a priori hypotheses about the role of the right vlPFC and amygdala in affect labeling, this analysis was constrained to an anatomical mask containing only these regions (delineated by combining the Harvard-Oxford atlas parcellations for the right inferior frontal gyrus and bilateral
78 amygdalae) using a small-volume correction for multiple comparisons at a FWE threshold of p <
.05 with FSL’s cluster-based correction. Activation maps were first thresholded (z > 3.1), and then Gaussian Random Field theory was used to calculate a cluster size threshold, below which clusters were removed.
This analysis resulted in a single cluster of functional activation in the right vlPFC. Mean parameter estimates were extracted from this cluster for the affect>gender labeling contrast for the SD and RW sessions of each subject. These values were submitted to a paired-samples t-test to examine the effects of SD on vlPFC recruitment during affective labeling. In addition, these extracted parameter estimates were used for additional control analyses.
Finally, previous work using a similar paradigm implicated right vlPFC connectivity with the amygdala during affect labeling (Foland et al., 2008; Hariri et al., 2000; Payer et al., 2012).
We therefore performed a psychophysiological interaction (PPI) analysis using the vlPFC cluster identified in the previous analysis as a seed. Session-level maps were submitted to a paired t-test model in FSL to compare between sleep conditions, and results were small-volume corrected within a mask of the bilateral amygdalae using the same parameters as above. As an exploratory follow-up, we sought to examine the relationship between functional connectivity with the amygdala and self-reported negative affect (from the PANAS), and how this relationship differed by sleep deprivation. To do this, we conducted a linear mixed effects model predicting negative affect score using sleep condition (RW or SD), parameter estimates from the PPI analysis, and their interaction as predictors. To control for the within- subject nature of the data, we included a random effect of participant.
RESULTS
79 Compliance & Sleep Manipulation Check
Participants complied with instructions to obtain a full night of sleep before the RW session, as evidenced by self-reported sleep duration estimates (7.58 +/- 0.75 hr), although these estimates were higher than those estimated by actigraphy (6.62 +/- 1.15 hr). Participants also complied with instructions on the night before the SD scan, getting 0 hr of sleep as confirmed by experimenter monitoring.
As expected, the SD manipulation successfully induced sleepiness and impairment on the psychomotor vigilance task. As with reaction times in the affect labeling task, psychomotor vigilance task reaction times were first log transformed to mitigate skew. One participant did not complete the self-reported sleepiness ratings during the RW visit, leaving a final sample of N=33 for this measure. As expected, compared to RW, during SD participants rated themselves as sleepier (t(32) = -11.33, p < 0.001; Figure 3.2A) and exhibited slower reaction times on the psychomotor vigilance task (t(33) = -5.54, p < 0.001; Figure 3.2B).
A B Sleepiness Psychomotor Vigilance Task
6 6.10
●
● 5 6.05
4 6.00 Rating 3 5.95 Mean Reaction Time (log) 2 ● 5.90
●
1 5.85
Rested Sleep Rested Sleep Wakefulness Deprivation Wakefulness Deprivation
Figure 3.2 SD successfully induced sleepiness, as assessed by self-report on the Stanford Sleepiness Scale (A), and slowed reaction time on the psychomotor vigilance task (B). Error bars represent within-subject SE.
80 Changes in Baseline Affective Measures in Response to SD
One participant did not complete the self-reported stress rating during the RW visit, leaving a final sample of N=33 for this measure. As expected, sleep deprivation resulted in an increase in self-reported stress (t(32) = 3.88, p < 0.001), anxiety (t(33) = 6.82, p < 0.001), and negative affect (t(33) = 4.43, p < 0.001) and a decrease in positive affect (t(33) = -3.44, p = 0.002).
Behavioral Response to Labeling Task
Consistent with prior work (Torrisi, Lieberman, Bookheimer, & Altshuler, 2013), performance accuracy was very high across both experimental sessions (RW: mean=97.5% ± 4.8, median=98.3%; SD: mean=96.6% ± 4.4, median=98.3%). Comparing log-transformed RT across sleep conditions and block types yielded a main effect of sleep condition such that RT was greater in the SD session (F(1, 99) = 5.141, p = .026), and a main effect of block type such that
RT was greater in the affect condition (F(1, 99) = 12.786, p < .001), but no interaction between session and task condition (F(1, 99) = 0.397, p = .530). Because RT was faster for gender than affect labeling, it is possible that task difficulty contributed to neural differences observed between these conditions. We therefore submitted results from the affect>gender labeling imaging contrast to control analyses in order to evaluate whether differences in RT had an influence on the resulting activations. However, RT differences between conditions cannot explain any effects of SD on these activations, as there was not a significant interaction between sleep and task condition on RT.
Effects of SD on Neural Signatures of Affective Labeling
81 Consistent with prior work, group level analyses (where participants’ RW and SD sessions were collapsed) revealed that affect labeling yielded higher right vlPFC activation than gender labeling (peak at MNI coordinates: x = 58, y = 30, z = 6; 116 voxels; Figure 3.3A). A paired t- test on the resultant vlPFC cluster showed that activation in this contrast was greater for the SD compared to RW session (t(33) = 3.32, p = 0.002; Figure 3.3B). Control analyses testing for a correlation between vlPFC activation and RT differences between affect and gender labeling during RW revealed that baseline activation of vlPFC was not explained by these RT differences
(r(32) = 0.17, p = 0.334). Contrary to expectations, we did not observe any difference between affect and gender labeling on activity in the amygdala.
A B Affect>Gender Labeling
40
●
30
20 Right vlPFC Activation 10
●
0 Rested Sleep 3.1 7.5 Wakefulness Deprivation Z Figure 3.3 (A) Affect labeling was associated with increased activation in the right vlPFC compared to gender labeling. Images are p < .05, FWE corrected. (B) Activation in the resulting cluster was greater under SD than RW (for affect > gender labeling contrast). Error bars represent within-subject SE.
Effects of SD on vlPFC - Amygdala Connectivity
PPI analysis of the right vlPFC identified a region in the left amygdala (peak at MNI coordinates: x = -16, y = -4, z = -22; 5 voxels) that exhibited greater functional connectivity with the vlPFC during affect labeling (affect>gender contrast) in SD compared to RW (Figure 3.4A).
82 Previous work has highlighted to role of right vlPFC-amygdala connectivity in implicit emotion regulation (Foland et al., 2008; Hariri et al., 2000; Payer et al., 2012). We reasoned that the same implicit emotion regulatory changes exhibited in the scanner may be occurring outside the scanner and could in turn influence negative affect. We found that right vlPFC-amygdala connectivity interacted with deprivation condition to predict negative affect, such that participants exhibited a negative relationship between negative affect and vlPFC-amygdala connectivity under SD but not RW (F(1, 31) = 6.67, p = 0.015; Figure 3.4B). In other words, the greater vlPFC-amygdala connectivity an individual exhibited, the less negative affect they reported, but only under SD.
A B
25 Rested Wakefulness Sleep Deprivation
20
15 Negative Affect Negative
10
3.1 7.5 −0.50 −0.25 0.00 0.25 0.50 Z vlPFC − Amygdala Connectivity Figure 3.4 (A) PPI analysis of affect labeling (affect>gender contrast) using the right vlPFC as a seed region. Results were small-volume corrected within the bilateral amygdala (p < .05, FWE correction), and revealed a region in the left amygdala exhibiting greater connectivity with the vlPFC during SD compared to RW. (B) Functional connectivity during affect labeling correlated with baseline negative affect during SD, but not during RW. Shaded areas around fit lines represent 95% confidence intervals.
83 DISCUSSION
The present study examined the effects of sleep deprivation on the neural systems underlying implicit emotion regulation. We used the affect labeling task to probe these systems, and consistent with previous work we found that the right vlPFC is specifically up-regulated during affect labeling. Using a within-subjects repeated-measures design, we found that this vlPFC recruitment was heightened under sleep deprivation. This heightened recruitment was accompanied by greater vlPFC-amygdala connectivity during affect labeling under sleep deprivation. Finally, greater vlPFC-amygdala connectivity was associated with less negative affect outside the scanner, but only under sleep deprivation. Together these findings suggest the brain engages in heightened implicit regulatory processing in response to the pressures of sleep deprivation.
Right vlPFC activation has reliably been observed during affective labeling in previous work, and is believed to be the mechanism by which regulatory control is exerted in this task. In some of these studies, this activation was accompanied by reductions in amygdala activity
(Burklund et al., 2015; Lieberman et al., 2007; S. E. Taylor et al., 2006), although such reductions were not observed in the present work. A regulatory control function of the vlPFC is consistent with brain imaging studies demonstrating its recruitment during other forms of emotion regulation including cognitive reappraisal, distraction, and suppression (Buhle et al.,
2014; Lévesque et al., 2003; McRae et al., 2010; Ochsner, Bunge, Gross, & Gabrieli, 2002; for reviews, see Berkman & Lieberman, 2009; Hooker & Knight, 2006). More broadly, this aligns with the proposed role of the vlPFC in the inhibition of emotional information that interferes
84 with goal-oriented behavior or processing (Hooker & Knight, 2006) and inhibitory control generally, especially in the right hemisphere (Aron et al., 2004).
The regulatory control exerted by the vlPFC in this and other emotion regulation tasks are thought of as domain-general, offering just one example case of how the vlPFC exerts control on subcortical and posterior-cortical systems. Interactions with sleep deprivation observed in this study can thus be contextualized by the well-documented prefrontal deficits and associated impairments in executive functioning that accompany sleep deprivation. Prefrontal impairment resulting in the increased vlPFC activation observed in the present study suggests two potential interpretations. First, it could be the case that greater vlPFC activation is reflective of this impairment: that SD is exerting its deleterious effects by up-regulating the vlPFC. This view is inconsistent with findings that cognitive impairments due to sleep loss are typically associated with reductions in right lateral prefrontal activation (Lythe et al., 2012; Ma et al., 2015; Mu et al., 2005).
An alternative interpretation is that this activation pattern reflects a compensatory mechanism. One way to evaluate this view is to consider other conditions marked by aberrant emotion regulation processing and to examine whether vlPFC activation is associated with better functioning. Clinical work with generalized anxiety disorder (GAD) has provided such a test case. GAD patients also show elevated vlPFC activation in response to emotional stimuli, and those that show the greatest activation exhibit the least severe symptoms, suggesting that the increased vlPFC activity may be related to some kind of compensation or resilience (Monk et al.,
2006). This compensatory mechanism could extend to communication with the amygdala, as
GAD symptom improvements in response to an emotion-regulation focused intervention are associated with increased vlPFC-amygdala connectivity (Hölzel et al., 2013). The compensatory
85 view is also consistent with follow-up analyses in the present work, which demonstrated that increases in vlPFC-amygdala connectivity under sleep deprivation were associated with decreased negative affect. Such a finding is consistent with the possibility that greater coupling within this system facilitates naturally occurring incidental emotion regulatory processes, which in turn counteract negative mood effects of SD.
The affect labeling task is meant in the current study to serve as a probe of the implicit modulation of affective systems due to selecting emotion labels. Although some previous studies
(Constantinou et al., 2014; Lieberman et al., 2011) have amended this task using emotionally evocative scenes and subsequently collected affect ratings (which were decreased by labeling), the present task was not meant to provoke a conscious emotional response and thus does not yield any such behavioral index of regulation. Future research should extend this work to further assess the experiential consequences of lack of sleep on emotion regulation via affect labeling.
Reaction time differences between the affect and gender labeling tasks may invite an effort-based interpretation to account for the present findings, in light of the recruitment of the vlPFC across non-specific cognitive demands (Duncan & Owen, 2000). However, reaction time differences between the affect and gender labeling conditions, which would presumably index such differences in effort, were not related to the vlPFC activations observed. This provides evidence that the vlPFC activation in response to affect labeling observed here are unlikely to be explained by differences in effort.
Nonetheless, the psychological mechanism by which affect labeling exerts regulatory control remains under debate. For example, it is not clear to what extent language is critical to this mechanism. It may be the case that linguistic processing facilitates the conversion of affective stimuli into symbolic representations, and this abstraction is what reduces affective
86 responding (Torre & Lieberman, 2018). This idea is supported by previous work demonstrating that the left vlPFC, at least, is critically implicated in the processing of abstract words (Hoffman,
Jefferies, & Ralph, 2010), and that recalled negative experiences that are construed more abstractly result in less negative affect (Kross, Ayduk, & Mischel, 2005). Another mechanism that has been proposed is the reduction of uncertainty (Lindquist, Gendron, Satpute, & Lindquist,
2016; Lindquist, Satpute, & Gendron, 2015). According to the psychological constructionist view of emotion, language helps make meaning of inherently ambiguous affective signals. Thus, accessibility to emotion words may facilitate the reduction of such ambiguity or uncertainty, which would in turn reduce amygdala responding (Cunningham & Brosch, 2012). See Torre &
Lieberman (2018) for a more detailed discussion of the possible mechanisms of emotion regulation through affect labeling.
Lack of sleep is a ubiquitous condition in modern society. With average sleep times continuing to decline (Van Cauter et al., 2008), it is more pressing than ever to understand how this depletion interacts with processes that are pivotal for psychological well-being like emotion regulation. Although a great deal of scientific work has explored the effects of sleep deprivation on the component processes of emotion regulation, to our knowledge this and a parallel report examining an explicit regulation strategy (Paper 2) are the first studies to directly test the neurobiological effects of sleep deprivation on emotion regulation (Palmer & Alfano, 2017). We found that deprivation increased vlPFC recruitment during affect labeling, evincing less efficient processing. This increase in vlPFC translated to greater coupling with the amygdala, which was in turn associated with less negative affect, suggesting an adaptive compensatory role of this system. We conclude that sleep deprivation may burden regulatory systems underlying affect labeling. Future work is needed to extend these findings to other forms of implicit emotion
87 regulation and to characterize the conditions and temporal dynamics of effective recovery from sleep-deprivation-induced impairment.
Acknowledgements
We thank Kate McLaughlin, Randy Buckner, Chuck Czeisler for helpful discussion, Jared Torre for sharing study materials, and Gina Falcone, Laurel Kordyban, Hyun Young Cho, Asi Graham,
Katya Kabotyanski, Amma Ababio, and Biniam Andargie for their help conducting this study.
Research reported in this publication was supported by the National Science Foundation
(DGE1144152 to M. S.) and the National Institutes of Health Shared Instrumentation Grant
(S10OD020039).
88 General Discussion and Conclusion
The goal of this dissertation was to examine the effects of contextual factors on emotion regulation processing. In Paper 1, we tested the effects of acute psychosocial stress on the ability to cognitively reappraise emotional pictures. Participants engaged in a canonical cognitive reappraisal (CR) task during fMRI imaging after having experienced a stress or control manipulation. Despite the induction of a rich, multifaceted stress response, stress did not impact
CR success. This was mirrored by findings in the brain, which showed that neural areas implicated in CR were recruited to a similar degree in participants who experienced a stressful or non-stressful context. Although stress did not affect CR, it was associated with greater reported negative affect in response to negative images and resulted in non-specific physiological responding to neutral pictures as if they were negative.
Paper 2 examined the effects of sleep deprivation on the same CR paradigm. Participants completed the task twice, once after a full night of sleep and once after a night of total sleep deprivation. As expected, sleep deprivation increased sleepiness levels, cognitive impairment, and baseline negative affect. However, similar to stress in Paper 1, lack of sleep did not impact the successful implementation of CR to reduce negative affect and resulted in no discernible differences in the brain. With respect to basic emotional responding, differences in subjective, physiological, and neural responding to negative versus neutral pictures that were observed under rested wakefulness were blunted under sleep deprivation.
Paper 3 examined the effects of sleep deprivation on the neurobiological processes underlying affect labeling, an implicit form of emotion regulation. Analyses focused on the amygdalae and right vlPFC, regions implicated in affect labeling. We observed activation of the right vlPFC during regulatory processing, and this recruitment was heightened under sleep
89 deprivation. Furthermore, sleep deprivation resulted in greater vlPFC-amygdala coupling, which was in turn associated with lower levels of negative mood. These findings suggest that the brain may engage in heightened implicit regulatory processing in response to the pressures of sleep deprivation.
Stress and sleep deprivation were selected as contexts in this investigation because both have repeatedly been shown to impair cognitive control functions, heighten emotional responding, and similarly impact the neural systems underlying these processes. Thus, one way to interpret Paper 2 is as a conceptual replication and extension of Paper 1, as both explore how
CR is managed under a system that is similarly dysregulated. Examining these studies together, two themes of findings emerge.
The first is with respect to basic emotional reactivity. Our hypotheses predicted that contextual factors would potentiate negative affect and result in increased emotional sensitivity.
We did see some evidence of this, with increased subjective reports of emotional reactivity as participants experienced more stress, and greater baseline negative affect after sleep deprivation.
However, we also observed another pattern across the studies. Heart rate deceleration, a physiological orienting response that is typically specific to negative or threat-related stimuli, was evoked nonspecifically to negative and neutral pictures under both stress and sleep deprivation. This lack of valence specificity in emotional responding was mirrored by subjective report and neural activation under sleep deprivation. Ratings of negative and neutral pictures shifted closer together, and several areas of the brain that differentiated the two picture types under rested wakefulness showed little or no difference when sleep deprived.
These findings add to an existing but small body of evidence that both stress and sleep deprivation induce a state of hyper-vigilance in which salient and non-salient stimuli are less
90 discriminable (Cousijn et al., 2010; Goldstein-Piekarski et al., 2015; Tempesta et al., 2010; van
Marle et al., 2009). It may be over-simplistic to just say that someone under stress or sleep deprivation is more emotionally reactive. Rather, in addition to responding more negatively to negative situations, a person may perceive more situations as negative in the first place. It is unclear from the present work whether this loss of specificity is the result of an increase in the salience of stimuli in general, or a shift to mood-congruent responding (which, in the case of these typically unpleasant contexts, would be negative). Previous research showing that reactivity to positive stimuli is amplified under sleep deprivation points to the former interpretation (Gujar et al., 2011); however, further work testing emotional responses to positive stimuli under stress is still needed.
The second major theme is the lack of contextual effects on CR. Evidence from real- world behavior and laboratory studies has suggested that these contexts impair some forms of emotion regulation (Baum et al., 2014; Mauss et al., 2013; Raio et al., 2013). However, across papers 1 and 2, we found no robust evidence that stress or sleep deprivation affected the subjective, physiological, or neural processing of CR. In the following paragraphs, I discuss the various factors that could be considered to help reconcile these divergent findings.
In Paper 1, there were some experimental factors related to the stress manipulation procedure that could have contributed to the lack of an observed effect. First, although we argue that the repeated nature of the stress manipulation created a protracted state of elevated stress, the external stressors did occur before rather than during CR. The stress response is multi-faceted, and certain components like cortisol, which slowly increased over the course of the experimental session, may have had more carry over to the CR task than other components, like elevated heart rate and certain feelings of arousal, that could have dissipated more quickly. Second, all
91 measures indicated that this study elicited a clear stress response in the experimental group on average. Nonetheless, there was appreciable variability in exactly how stressful an individual found the procedure, and thus variation in how strongly the experimental context was induced.
Finally, Paper 1 was a between-subjects design and may not have been sensitive to subtler differences in an individual’s CR ability under stressed and non-stressed conditions. It is with respect to these concerns that the findings in Paper 2 can offer some clarity. Sleep deprivation is a state that is less acute, for which we have more experimental control in assuring that the context manipulation is delivered, and that is more amenable to a within-subjects design. We nonetheless observed no effect of sleep deprivation on CR. The lack of effect across both studies, despite the relative advantages and disadvantages of each, makes it less likely that these findings are due to experimental concerns about delivering the context manipulation.
Although we successfully induced the stress and sleep manipulations, we cannot rule out the possibility that a moderate level of stress or one night of deprivation was simply not a strong enough manipulation to disrupt cognitive reappraisal. These findings are still unexpected, as the previous studies that motivated this investigation, demonstrating heightened affective responding and cognitive impairment as a result of these contexts, employed levels of psychological stress and sleep deprivation similar to those implemented in the present studies (Goel et al., 2009;
Luethi et al., 2009; Ma et al., 2015; Olver et al., 2015). Nevertheless, it remains unclear whether these results would generalize to more severe stressors and more extreme sleep loss. However, the goal is to assess how normative levels of these potentially disruptive contexts impact regulation ability. If it is the case that greater sleep loss or a stressor of higher intensity is needed to evoke CR deficits, it is worth reflecting on both how robust this interaction is and to what real- world situations it may be applicable.
92 In addition to the contexts, we must consider the generalizability of the phenomenon we are studying with the CR task used. As discussed within the papers, this task only assesses CR ability, which is one of many components involved when using CR in daily life. If CR ability is not disrupted by challenging circumstances that are hypothesized to affect CR in daily life, perhaps we should broaden our focus to examine other aspects of real-world CR, such as making the choice to use an adaptive strategy like CR in the first place, generating several reappraisal narratives to select from, and the ability to flexibly navigate among different regulation strategies or specific reappraisals.
We must also consider whether the CR task used is truly measuring what it is purported to measure. It is possible that the observed and highly replicable decreases in self-reported negative affect in response to the CR condition do not, in part or in full, reflect emotion regulation processes. Although we instructed participants to respond honestly about how they felt and assured them that it was not a problem if the CR strategy did not work well, the task explicitly called upon them to use CR to decrease their negative affect. This knowledge that the
CR condition was expected to reduce negative affect could very well have induced experimental demand effects that influenced their affect ratings, whether consciously or subconsciously. In implicit tasks, on the other hand, decreasing negative affect is not an overt goal. Indeed Paper 3, which employed an implicit task and thus obviated this concern, did find an effect of sleep deprivation on regulation processing. As a matter of fact, the right vlPFC region that was modulated during regulation processing under sleep deprivation in Paper 3 was immediately adjacent and partially overlapping with one of the ROIs in Paper 2 that exhibited absolutely no effects of the same context. These findings suggest that there is some specificity to the null effects observed in the CR task, at least in the brain.
93 Limitations and outstanding questions
There are several important limitations that constrain the interpretations of the findings presented here and offer directions for future study. First, although the sample size in this study was comparable to or larger than related fMRI studies of the effects of stress or sleep deprivation on cognition (Lythe et al., 2012; Ma et al., 2015; Strang, Pruessner, & Pollak, 2011; van Ast et al., 2014), we were only powered to observe medium to large effects. Perhaps it really is the case that the stress and sleep deprivation contexts examined do not impact CR ability, or perhaps such interactions do exist and we simply did not observe them here. There are many sources of statistical noise in experimental data, some of which have been discussed and likely others we did not consider. As the old adage goes, absence of evidence is not evidence of absence; from a strict frequentist statistical standpoint, null effects do not constitute evidence, and we must be cautious in over-interpreting the findings from papers 1 and 2. Nonetheless, it is not and should not be the case that an argument for the lack of an effect cannot be made. Further replication and the use of non-frequentist methods that do not carry the same statistical limitations, such as
Bayes factors, can help bolster the tentative claims made presently.
One benefit of Paper 3 is that it provides specificity to the preceding results. It is not the case that this specific sleep deprivation protocol simply does not cause any observable changes in the brain, or even that it does not affect neural regions implicated in regulation. Nonetheless, the task employed in Paper 3 only targets the neural processes underlying emotion regulation.
Although sleep deprivation induced changes in activation correlated with mood measures outside the scanner, we cannot truly make the reverse inference that changes in vlPFC activation in response to deprivation reflect changes in emotion regulation. To do this, an implicit regulation task with a reliable behavioral indicator of emotional responding and regulation is needed.
94 Another limitation of the current work is that in addition to the variation in how strongly the contextual manipulations were induced among participants, there is also variation in how individuals respond to the same level of stress or sleep deprivation. There are likely individual differences in the level of intensity required to produce the hypothesized deficits. Furthermore, there are qualitative differences in the way individuals may respond to these contexts. For example, in line with previous work, male and female participants displayed distinct response profiles to the stressors in Paper 1. Female participants reported marked subjective stress but did not exhibit a significant cortisol response, while male participants exhibited both. The present studies were optimized to examine group-level differences, and were not situated to make strong claims about the effects of such individual difference characteristics. However, understanding differences across individuals in profiles of responding to these contexts and how they interact with emotion regulation functioning could present an avenue for future research.
The sex differences in cortisol responding in Paper 1 alluded to above warrant deeper consideration. The two primary arms of the stress response are activation of the HPA axis, which results in the release of cortisol, and the sympathetic nervous system, which results in the release of norepinephrine (NE). In a study testing the effects of stress on working memory, Elzinga &
Roelofs (2005) argued that cortisol elevation, in concert with sympathetic activation, was required to induce cognitive deficits in humans, as only cortisol responders (N = 11, 9 male) exhibited working memory impairment. It is possible that the limited effects of the stress manipulation on cortisol in Paper 1, restricted largely to males, may not have been sufficient to induce impairment of the cognitive control systems implicated in reappraisal. However, supplementary analyses examining only male participants (N = 24, 11 stressed; Supplementary
Figure S1) also exhibited no effects of stress on affect ratings in response to CR (F(1,22)=1.182,
95 p=0.289). Furthermore, previous work in animal models contradicts the idea that sympathetic activation alone is insufficient to induce prefrontal or cognitive impairment. Administration of an agonist of α1 NE receptors (which are selectively activated at the elevated NE levels exhibited during stress) impairs working memory and disrupts prefrontal physiology, while blockade of these receptors prevents stress-induced cognitive deficits (Arnsten, 2009, 2015). These findings are consistent with human work showing that the α1-receptor antagonist prazosin is useful in treating PTSD and that stress-induced cognitive deficits can be rescued by propanolol, a beta- adrenergic receptor blocker (Alexander et al., 2007; Raskind et al., 2002; F. Taylor & Raskind,
2002). Nonetheless, we cannot be sure that the neuro-hormonal outputs of the stress induction in
Paper 1 were sufficient to cause the prefrontal deficits on which our hypotheses are predicated.
These limitations of the stress induction were in part what motivated the examination of a different context in Paper 2, which could serve as a conceptual replication.
Finally, although I have discussed in detail the potential limitations of the CR task, some of the generalizability concerns of this paradigm may be limitations of the fMRI scanner context.
Even though viewing emotional pictures and a simple CR manipulation that can be turned on and off every few seconds may not be the most reflective of real-world CR, such repetition is often needed for sufficient power to conduct neuroimaging analyses. More generally, it is difficult to induce any kind of high intensity emotion, like ecstasy, rage, or anguish, into the laboratory setting, both for experimental and ethical considerations.
Nonetheless, we can likely do better. One direction of improvement could be to employ emotional stimuli that elicit responses closer to the rich and complex emotional responses exhibited in daily life. For example, studies using stimuli like emotional movies may allow for a more ecologically reflective, dynamic emotional response to unfold over time (Chang et al.,
96 2018; Troy, Wilhelm, Shallcross, & Mauss, 2010). Another is to reflect on how CR is situated within the practice of Cognitive Behavioral Therapy (CBT), often cited as one of the reasons CR is important to study. Within CBT, patients identify problematic thoughts, assess the accuracy of those thoughts, use CR to gain a more balanced perspective, and then reexamine their emotional experience (Barlow, 2014). Considered in these circumstances, CR is likely used for thoughts that are highly self-relevant. Perhaps paradigms using participant-specific, autobiographical scenarios would better incorporate this self-relevant element (Doré et al., 2018). Furthermore,
CBT interventions typically take weeks, and it may be that we need more than a short instructional session to teach the use of CR skills in a meaningful way.
Conclusion
This dissertation sought to examine the mechanism by which challenging contexts like acute stress and sleep deprivation impaired emotion regulation processing. To our surprise, we found over the course of two studies that neither stress nor sleep deprivation impacted the psychological, physiological, or neurobiological outputs of a canonical task testing a regulation strategy of wide interest, cognitive reappraisal. These null results were observed despite evidence from a third study demonstrating that sleep deprivation modulated the neural correlates of an implicit emotion regulation task. Together, these findings call for a careful examination of the way emotion regulation is presently studied and whether cognitive reappraisal, or only certain aspects of it, may in fact be robust to the effects of stress and sleep deprivation.
97 References
Adhikari, A., Lerner, T. N., Finkelstein, J., Pak, S., Jennings, J. H., Davidson, T. J., … Deisseroth, K. (2015). Basomedial amygdala mediates top-down control of anxiety and fear. Nature, 527(7577), 179–185.
Aldao, A. (2013). The Future of Emotion Regulation Research: Capturing Context. Perspectives on Psychological Science, 8(2), 155–172.
Alexander, J. K., Hillier, A., Smith, R. M., Tivarus, M. E., & Beversdorf, D. Q. (2007). Beta- adrenergic modulation of cognitive flexibility during stress. Journal of Cognitive Neuroscience, 19(3), 468–478.
Anisman, H., & Merali, Z. (1999). Understanding stress: characteristics and caveats. Alcohol Research & Health : The Journal of the National Institute on Alcohol Abuse and Alcoholism, 23(4), 241–249.
Arnsten, A. F. T. (2009). Stress signalling pathways that impair prefrontal cortex structure and function. Nature Reviews. Neuroscience, 10(6), 410–422.
Arnsten, A. F. T. (2015). Stress weakens prefrontal networks: molecular insults to higher cognition. Nature Neuroscience, 18(10), 1376–1385.
Aron, A. R., Robbins, T. W., & Poldrack, R. A. (2004). Inhibition and the right inferior frontal cortex. Trends in Cognitive Sciences.
Avants, B. B., Tustison, N., & Song, G. (2009). Advanced Normalization Tools (ANTS). Insight Journal, 2, 1–35.
Babson, K. A., Trainor, C. D., Feldner, M. T., & Blumenthal, H. (2010). A test of the effects of acute sleep deprivation on general and specific self-reported anxiety and depressive symptoms: An experimental extension. Journal of Behavior Therapy and Experimental Psychiatry, 41(3), 297–303.
Badre, D., & Wagner, A. D. (2007). Left ventrolateral prefrontal cortex and the cognitive control of memory. Neuropsychologia, 45(13), 2883–2901.
Baglioni, C., Spiegelhalder, K., Lombardo, C., & Riemann, D. (2010, August 1). Sleep and emotions: A focus on insomnia. Sleep Medicine Reviews. W.B. Saunders.
Barlow, D. H. (Ed.). (2014). Clinical handbook of psychological disorders: A step-by-step treatment manual. Guilford publications.
Baum, K. T., Desai, A., Field, J., Miller, L. E., Rausch, J., & Beebe, D. W. (2014). Sleep restriction worsens mood and emotion regulation in adolescents. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 55(2), 180–190.
98
Berking, M., & Wupperman, P. (2012). Emotion regulation and mental health: recent findings, current challenges, and future directions. Current Opinion in Psychiatry, 25(2), 128–134.
Berkman, E. T., & Lieberman, M. D. (2009). Using Neuroscience to Broaden Emotion Regulation: Theoretical and Methodological Considerations. Social and Personality Psychology Compass, 3(4), 475–493.
Boden, M. T., Bonn-Miller, M. O., Kashdan, T. B., Alvarez, J., & Gross, J. J. (2012). The interactive effects of emotional clarity and cognitive reappraisal in Posttraumatic Stress Disorder. Journal of Anxiety Disorders, 26(1), 233–238.
Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict monitoring and cognitive control. Psychological Review, 108(3), 624–652.
Bradley, M. M., Codispoti, M., Cuthbert, B. N., & Lang, P. J. (2001). Emotion and motivation I: defensive and appetitive reactions in picture processing. Emotion (Washington, D.C.), 1(3), 276–298.
Buchanan, T. W., & Lovallo, W. R. (2001). Enhanced memory for emotional material following stress-level cortisol treatment in humans. Psychoneuroendocrinology, 26(3), 307–317.
Buhle, J. T., Silvers, J. A., Wage, T. D., Lopez, R., Onyemekwu, C., Kober, H., … Ochsner, K. N. (2014). Cognitive reappraisal of emotion: A meta-analysis of human neuroimaging studies. Cerebral Cortex, 24(11), 2981–2990.
Burklund, L. J., Craske, M. G., Taylor, S. E., & Lieberman, M. D. (2015). Altered emotion regulation capacity in social phobia as a function of comorbidity. Social Cognitive and Affective Neuroscience, 10(2), 199–208.
Caldwell, J. A., Caldwell, J. L., Brown, D. L., & Smith, J. K. (2004). The Effects of 37 Hours of Continuous Wakefulness On the Physiological Arousal, Cognitive Performance, Self- Reported Mood, and Simulator Flight Performance of F-117A Pilots. Military Psychology, 16(3), 163–181.
Campbell-Sills, L., Barlow, D. H., Brown, T. A., & Hofmann, S. G. (2006). Acceptability and suppression of negative emotion in anxiety and mood disorders. Emotion (Washington, D.C.), 6(4), 587–595.
Campbell, B. A., Wood, G., & McBride, T. (1997). Origins of orienting and defensive responses: An evolutionary perspective. In Attention and Orienting: sensory and motivational processes. (pp. 41–67).
Chang, L. J., Jolly, E., Cheong, J. H., Rapuano, K., Greenstein, N., Chen, P.-H. A., & Manning, J. R. (2018). Endogenous variation in ventromedial prefrontal cortex state dynamics during naturalistic viewing reflects affective experience. Preprint bioRxiv.
99
Chee, M. W. L., & Choo, W. C. (2004). Functional Imaging of Working Memory after 24 Hr of Total Sleep Deprivation. Journal of Neuroscience, 24(19), 4560–4567.
Cheng, C. (2001). Assessing Coping Flexibility in Real-Life and Laboratory Settings: A Multimethod Approach. Journal of Personality and Social Psychology, 80(5), 814–833.
Childs, E., Dlugos, A., & De Wit, H. (2010). Cardiovascular, hormonal, and emotional responses to the TSST in relation to sex and menstrual cycle phase. Psychophysiology, 47(3), 550– 559.
Cloitre, M., Koenen, K. C., Cohen, L. R., & Han, H. (2002). Skills training in affective and interpersonal regulation followed by exposure: a phase-based treatment for PTSD related to childhood abuse. Journal of Consulting and Clinical Psychology, 70(5), 1067–1074.
Constantinou, E., Van Den Houte, M., Bogaerts, K., Van Diest, I., & Van den Bergh, O. (2014). Can words heal? Using affect labeling to reduce the effects of unpleasant cues on symptom reporting. Frontiers in Psychology, 5, 807.
Contreras, J. M., Kerns, K. A., Weimer, B. L., Gentzler, A. L., & Tomich, P. L. (2000). Emotion regulation as a mediator of associations between mother-child attachment and peer relationships in middle childhood. Journal of Family Psychology : JFP : Journal of the Division of Family Psychology of the American Psychological Association (Division 43), 14(1), 111–124.
Costafreda, S. G., Brammer, M. J., David, A. S., & Fu, C. H. Y. (2008). Predictors of amygdala activation during the processing of emotional stimuli: A meta-analysis of 385 PET and fMRI studies. Brain Research Reviews, 58(1), 57–70.
Cote, K., Jancsar, C., & Hunt, B. (2015). Event-related neural response to emotional picture stimuli following sleep deprivation. Psychology & Neuroscience, 8(1), 102–113.
Cousijn, H., Rijpkema, M., Qin, S., van Marle, H. J. F., Franke, B., Hermans, E. J., … Fernández, G. (2010). Acute stress modulates genotype effects on amygdala processing in humans. Proceedings of the National Academy of Sciences of the United States of America, 107(21), 9867–9872.
Cunningham, W. A., & Brosch, T. (2012). Motivational salience: Amygdala tuning from traits, needs, values, and goals. Current Directions in Psychological Science, 21(1), 54–59.
Cuthbert, B. N., Bradley, M. M., & Lang, P. J. (1996). Probing picture perception: Activation and emotion. Psychophysiology, 33(2), 103–111.
Cutler, N. R., & Cohen, H. B. (1979). The effect of one night’s sleep loss on mood and memory in normal subjects. Comprehensive Psychiatry, 20(1), 61–66.
100 Dedovic, K., Renwick, R., Mahani, N. K., Engert, V., Lupien, S. J., & Pruessner, J. C. (2005). The Montreal Imaging Stress Task: Using functional imaging to investigate the effects of perceiving and processing psychosocial stress in the human brain. In Journal of Psychiatry and Neuroscience (Vol. 30, pp. 319–325).
Denson, T. F., Creswell, J. D., Terides, M. D., & Blundell, K. (2014a). Cognitive reappraisal increases neuroendocrine reactivity to acute social stress and physical pain. Psychoneuroendocrinology, 49C, 69–78.
Denson, T. F., Creswell, J. D., Terides, M. D., & Blundell, K. (2014b). Cognitive reappraisal increases neuroendocrine reactivity to acute social stress and physical pain. Psychoneuroendocrinology, 49, 69–78.
Denson, T. F., Mehta, P. H., & Ho Tan, D. (2013). Endogenous testosterone and cortisol jointly influence reactive aggression in women. Psychoneuroendocrinology, 38(3), 416–424.
Dinges, D. F., Pack, F., Williams, K., Gillen, K. A., Powell, J. W., Ott, G. E., … Pack, A. I. (1997). Cumulative Sleepiness, Mood Disturbance, and Psychomotor Vigilance Performance Decrements During a Week of Sleep Restricted to 4–5 Hours per Night. Sleep, 20(4), 267–277.
Dinges, D. F., & Powell, J. W. (1985). Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behavior Research Methods, Instruments, & Computers, 17(6), 652–655.
Doré, B. P., Rodrik, O., Boccagno, C., Hubbard, A., Weber, J., Stanley, B., … Ochsner, K. N. (2018). Negative Autobiographical Memory in Depression Reflects Elevated Amygdala- Hippocampal Reactivity and Hippocampally Associated Emotion Regulation. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 3(4), 358–366.
Duncan, J., & Owen, A. M. (2000). Common regions of the human frontal lobe recruited by diverse cognitive demands. Trends Neurosci (Vol. 23). Retrieved from http://psych.colorado.edu/~munakata/teaching/prosem07/DuncanOwen00.pdf
Durmer, J. S., & Dinges, D. F. (2005). Neurocognitive Consequences of Sleep Deprivation. Seminars in Neurology, 25(1), 117–129.
Ehring, T., Tuschen-Caffier, B., Schnülle, J., Fischer, S., & Gross, J. J. (2010). Emotion regulation and vulnerability to depression: spontaneous versus instructed use of emotion suppression and reappraisal. Emotion (Washington, D.C.), 10(4), 563–572.
Ekman, P. (1992). Facial Expressions of Emotion: New Findings, New Questions. Psychological Science, 3(1), 34–38.
Elzinga, B. M., & Roelofs, K. (2005). Cortisol-induced impairments of working memory require acute sympathetic activation. Behavioral Neuroscience, 119(1), 98–103.
101
Finlay, J. M., Zigmond, M. J., & Abercrombie, E. D. (1995). Increased dopamine and norepinephrine release in medial prefrontal cortex induced by acute and chronic stress: effects of diazepam. Neuroscience, 64(3), 619–628.
Foland, L. C., Altshuler, L. L., Bookheimer, S. Y., Eisenberger, N., Townsend, J., & Thompson, P. M. (2008). Evidence for deficient modulation of amygdala response by prefrontal cortex in bipolar mania. Psychiatry Research, 162(1), 27–37.
Ford, D. E., & Kamerow, D. B. (1989). Epidemiologic study of sleep disturbances and psychiatric disorders. An opportunity for prevention? JAMA, 262(11), 1479–1484.
Franzen, P. L., Buysse, D. J., Dahl, R. E., Thompson, W., & Siegle, G. J. (2009). Sleep deprivation alters pupillary reactivity to emotional stimuli in healthy young adults. Biological Psychology, 80(3), 300–305.
Frijda, N. (1986). The emotions: studies in emotion and social interaction. Cambridge University Press. Cambridge.
Germain, A., Buysse, D. J., & Nofzinger, E. (2008, June 1). Sleep-specific mechanisms underlying posttraumatic stress disorder: Integrative review and neurobiological hypotheses. Sleep Medicine Reviews. W.B. Saunders.
Goel, N., Rao, H., Durmer, J. S., & Dinges, D. F. (2009). Neurocognitive consequences of sleep deprivation. Seminars in Neurology, 29(4), 320–339.
Goldin, P. R., McRae, K., Ramel, W., & Gross, J. J. (2008). The Neural Bases of Emotion Regulation: Reappraisal and Suppression of Negative Emotion. Biological Psychiatry, 63(6), 577–586.
Goldstein-Piekarski, A. N., Greer, S. M., Saletin, J. M., & Walker, M. P. (2015). Sleep Deprivation Impairs the Human Central and Peripheral Nervous System Discrimination of Social Threat. The Journal of Neuroscience : The Official Journal of the Society for Neuroscience, 35(28), 10135–10145.
Goldstein, A. N., & Walker, M. P. (2014). The Role of Sleep in Emotional Brain Function. Annual Review of Clinical Psychology, 10(1), 679–708.
Gordon, A. M., & Chen, S. (2014). The Role of Sleep in Interpersonal Conflict: Do Sleepless Nights Mean Worse Fights? Social Psychological and Personality Science, 5(2), 168–175.
Gorgolewski, K., Burns, C. D., Madison, C., Clark, D., Halchenko, Y. O., Waskom, M. L., & Ghosh, S. S. (2011). Nipype: a flexible, lightweight and extensible neuroimaging data processing framework in python. Frontiers in Neuroinformatics, 5, 13.
Graziano, P. A., Reavis, R. D., Keane, S. P., & Calkins, S. D. (2007). The Role of Emotion
102 Regulation and Children’s Early Academic Success. Journal of School Psychology, 45(1), 3–19.
Greenberg, L. S. (2004). Emotion-focused therapy. Clinical Psychology and Psychotherapy, 11(1), 3–16.
Greer, S. M., Goldstein, A. N., & Walker, M. P. (2013). The impact of sleep deprivation on food desire in the human brain. Nature Communications, 4.
Greve, D. N., & Fischl, B. (2009). Accurate and robust brain image alignment using boundary- based registration. NeuroImage, 48(1), 63–72.
Gross, J. J. (1998). Antecedent- and response-focused emotion regulation: divergent consequences for experience, expression, and physiology. Journal of Personality and Social Psychology, 74(1), 224–237.
Gross, J. J., & John, O. P. (2003). Individual differences in two emotion regulation processes: implications for affect, relationships, and well-being. Journal of Personality and Social Psychology, 85(2), 348–362.
Gruber, R., & Cassoff, J. (2014). The Interplay Between Sleep and Emotion Regulation: Conceptual Framework Empirical Evidence and Future Directions. Current Psychiatry Reports, 16(11), 500.
Gründemann, D., Schechinger, B., Rappold, G. A., & Schömig, E. (1998). Molecular identification of the corticosterone-sensitive extraneuronal catecholamine transporter. Nature Neuroscience, 1(5), 349–351.
Gujar, N., Yoo, S.-S., Hu, P., & Walker, M. P. (2011). Sleep Deprivation Amplifies Reactivity of Brain Reward Networks, Biasing the Appraisal of Positive Emotional Experiences. Journal of Neuroscience, 31(12), 4466–4474.
Gumora, G., & Arsenio, W. F. (2002). Emotionality, Emotion Regulation, and School Performance in Middle School Children. Journal of School Psychology, 40(5), 395–413.
Gyurak, A., Gross, J. J., & Etkin, A. (2011). Explicit and implicit emotion regulation: a dual- process framework. Cognition & Emotion, 25(3), 400–412.
Hariri, A. R., Bookheimer, S. Y., & Mazziotta, J. C. (2000). Modulating emotional responses: Effects of a neocortical network on the limbic system. NeuroReport, 11(1), 43–48.
Hatfield, T., & McGaugh, J. L. (1999). Norepinephrine infused into the basolateral amygdala posttraining enhances retention in a spatial water maze task. Neurobiology of Learning and Memory, 71(2), 232–239.
Hellhammer, D. H., Wüst, S., & Kudielka, B. M. (2009). Salivary cortisol as a biomarker in
103 stress research. Psychoneuroendocrinology, 34(2), 163–171.
Hoddes, E., Zarcone, V., Smythe, H., Phillips, R., & Dement, W. C. (1973). Quantification of Sleepiness: A New Approach. Psychophysiology, 10(4), 431–436.
Hoffman, P., Jefferies, E., & Ralph, M. A. L. (2010). Ventrolateral prefrontal cortex plays an executive regulation role in comprehension of abstract words: convergent neuropsychological and repetitive TMS evidence.
Hölzel, B. K., Hoge, E. A., Greve, D. N., Gard, T., Creswell, J. D., Brown, K. W., … Lazar, S. W. (2013). Neural mechanisms of symptom improvements in generalized anxiety disorder following mindfulness training. NeuroImage: Clinical, 2, 448–458.
Hooker, C. I., & Knight, R. T. (2006). The role of lateral orbitofrontal cortex in the inhibitory control of emotion (pp. 1–18).
Institute of Medicine. (2006). Sleep Disorders and Sleep Deprivation. Washington, D.C.: National Academies Press.
Joormann, J., & Gotlib, I. H. (2010). Emotion regulation in depression: relation to cognitive inhibition. Cognition & Emotion, 24(2), 281–298.
Kahn-Greene, E. T., Killgore, D. B., Kamimori, G. H., Balkin, T. J., & Killgore, W. D. S. (2007). The effects of sleep deprivation on symptoms of psychopathology in healthy adults. Sleep Medicine, 8(3), 215–221.
Kahn, M., Sheppes, G., & Sadeh, A. (2013). Sleep and emotions: Bidirectional links and underlying mechanisms.
Kassel, J. D., Bornovalova, M., & Mehta, N. (2007). Generalized expectancies for negative mood regulation predict change in anxiety and depression among college students. Behaviour Research and Therapy, 45(5), 939–950.
Katzir, M., & Eyal, T. (2013). When stepping outside the self is not enough: A self-distanced perspective reduces the experience of basic but not of self-conscious emotions. Journal of Experimental Social Psychology, 49(6), 1089–1092.
Keltner, D., & Gross, J. J. (1999). Functional accounts of emotions. Cognition and Emotion, 13(5), 467–480.
Kircanski, K., Lieberman, M. D., & Craske, M. G. (2012). Feelings Into Words: Contributions of Language to Exposure Therapy. Psychological Science, 23(10), 1086–1091.
Kirschbaum, C., Kudielka, B. M., Gaab, J., Schommer, N. C., & Hellhammer, D. H. (1999). Impact of gender, menstrual cycle phase, and oral contraceptives on the activity of the hypothalamus-pituitary-adrenal axis. Psychosomatic Medicine, 61(2), 154–162.
104
Kirschbaum, C., Pirke, K. M., & Hellhammer, D. H. (1993). The “Trier Social Stress Test”--a tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology, 28(1–2), 76–81.
Kirschbaum, C., Wust, S., & Hellhammer, D. (1992). Consistent sex differences in cortisol responses to psychological stress. Psychosomatic Medicine, 54(6), 648–657.
Kraaij, V., Pruymboom, E., & Garnefski, N. (2002). Cognitive coping and depressive symptoms in the elderly: a longitudinal study. Aging & Mental Health, 6(3), 275–281.
Krause, A. J., Simon, E. Ben, Mander, B. A., Greer, S. M., Saletin, J. M., Goldstein-Piekarski, A. N., & Walker, M. P. (2017). The sleep-deprived human brain. Nature Reviews. Neuroscience, 18(7), 404–418.
Kross, E., Ayduk, O., & Mischel, W. (2005). When asking “why” does not hurt: Distinguishing rumination from reflective processing of negative emotions. Psychological Science, 16(9), 709–715.
Kumar, S., Feldman, G., & Hayes, A. (2008). Changes in Mindfulness and Emotion Regulation in an Exposure-Based Cognitive Therapy for Depression. Cognitive Therapy and Research, 32(6), 734–744.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1997). International Affective Picture System (IAPS): Technical Manual and Affective Ratings. NIMH Center for the Study of Emotion and Attention, 39–58.
Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O. (1993). Looking at pictures: Affective, facial, visceral, and behavioral reactions. Psychophysiology, 30(3), 261–273.
Latzer, Y., Tzischinsky, O., Epstein, R., Klein, E., & Peretz, L. (1999). Naturalistic Sleep Monitoring in Women Suffering from Bulimia Nervosa. Int J Eat Disord, 26, 315–321.
Lauer, C. J., & Krieg, J.-C. (2004). Sleep in eating disorders. Sleep Medicine Reviews, 8(2), 109– 118.
Lévesque, J., Eugène, F., Joanette, Y., Paquette, V., Mensour, B., Beaudoin, G., … Beauregard, M. (2003). Neural circuitry underlying voluntary suppression of sadness. Biological Psychiatry, 53(6), 502–510.
Lewis, D. A., Campbell, M. J., Foote, S. L., Goldstein, M., & Morrison, J. H. (1987). The distribution of tyrosine hydroxylase-immunoreactive fibers in primate neocortex is widespread but regionally specific. The Journal of Neuroscience : The Official Journal of the Society for Neuroscience, 7(1), 279–290.
Lieberman, M. D., Eisenberger, N. I., Crockett, M. J., Tom, S. M., Pfeifer, J. H., & Way, B. M.
105 (2007). Putting Feelings Into Words. Psychological Science, 18(5), 421–428. Lieberman, M. D., Inagaki, T. K., Tabibnia, G., & Crockett, M. J. (2011). Subjective Responses to Emotional Stimuli During Labeling, Reappraisal, and Distraction. Emotion, 11(3), 468– 480.
Lindquist, K. A., Gendron, M., Satpute, A. B., & Lindquist, K. A. (2016). Language and emotion: Putting words into feelings and feelings into words. In Handbook of Emotions (Vol. 4). The Guilford Press, New York, NY.
Lindquist, K. A., Satpute, A. B., & Gendron, M. (2015). Does Language Do More Than Communicate Emotion? Current Directions in Psychological Science, 24(2), 99–108.
Liverant, G. I., Brown, T. A., Barlow, D. H., & Roemer, L. (2008). Emotion regulation in unipolar depression: the effects of acceptance and suppression of subjective emotional experience on the intensity and duration of sadness and negative affect. Behaviour Research and Therapy, 46(11), 1201–1209.
Lopes, P. N., Salovey, P., Coté, S., & Beers, M. (2005). Emotion regulation abilities and the quality of social interaction. Emotion (Washington, D.C.), 5(1), 113–118.
Luethi, M., Meier, B., & Sandi, C. (2009). Stress effects on working memory, explicit memory, and implicit memory for neutral and emotional stimuli in healthy men. Frontiers in Behavioral Neuroscience, 2, 5.
Lythe, K. E., Williams, S. C. R., Anderson, C., Libri, V., & Mehta, M. A. (2012). Frontal and parietal activity after sleep deprivation is dependent on task difficulty and can be predicted by the fMRI response after normal sleep. Behavioural Brain Research, 233(1), 62–70.
Ma, N., Dinges, D. F., Basner, M., & Rao, H. (2015). How acute total sleep loss affects the attending brain: a meta-analysis of neuroimaging studies. Sleep, 38(2), 233–240.
Mai, J. K., Paxinos, G., & Voss, T. (2008). Atlas of the Human Brain (3rd ed.). Academic Press Inc.
Mauss, I. B., Troy, A. S., & LeBourgeois, M. K. (2013). Poorer sleep quality is associated with lower emotion-regulation ability in a laboratory paradigm. Cognition and Emotion, 27(3), 567–576.
McEwen, B. S. (2000). The neurobiology of stress: from serendipity to clinical relevance. Brain Research, 886(1–2), 172–189.
McRae, K., Hughes, B., Chopra, S., Gabrieli, J. D. E., Gross, J. J., & Ochsner, K. N. (2010). The Neural Bases of Distraction and Reappraisal. Journal of Cognitive Neuroscience, 22(2), 248–262.
McRae, K., Taitano, E. K., & Lane, R. D. (2010). The effects of verbal labelling on
106 psychophysiology: Objective but not subjective emotion labelling reduces skin-conductance responses to briefly presented pictures. Cognition & Emotion, 24(5), 829–839.
Mellman, T. A., & Uhde, T. W. (1989). Sleep panic attacks: new clinical findings and theoretical implications. American Journal of Psychiatry, 146(9), 1204–1207.
Miklósi, M., Martos, T., Szabó, M., Kocsis-Bogár, K., & Forintos, D. P. (2014). Cognitive emotion regulation and stress: A multiple mediation approach. Translational Neuroscience, 5(1), 64–71.
Minkel, J. D., McNealy, K., Gianaros, P. J., Drabant, E. M., Gross, J. J., Manuck, S. B., & Hariri, A. R. (2012). Sleep quality and neural circuit function supporting emotion regulation. Biology of Mood & Anxiety Disorders, 2(1), 22.
Mitchell, D. G. V. (2010). The nexus between decision making and emotion regulation: A review of convergent neurocognitive substrates. Behavioural Brain Research, 217, 215– 231.
Monk, C. S., Nelson, E. E., McClure, E. B., Mogg, K., Bradley, B. P., Leibenluft, E., … Pine, D. S. (2006). Ventrolateral Prefrontal Cortex Activation and Attentional Bias in Response to Angry Faces in Adolescents With Generalized Anxiety Disorder. American Journal of Psychiatry, 163(6), 1091–1097.
Moore, S. A., Zoellner, L. A., & Mollenholt, N. (2008). Are expressive suppression and cognitive reappraisal associated with stress-related symptoms? Behaviour Research and Therapy, 46(9), 993–1000.
Motomura, Y., Kitamura, S., Oba, K., Terasawa, Y., Enomoto, M., Katayose, Y., … Mishima, K. (2013). Sleep Debt Elicits Negative Emotional Reaction through Diminished Amygdala- Anterior Cingulate Functional Connectivity. PLoS ONE, 8(2), e56578.
Mu, Q., Nahas, Z., Johnson, K. a, Yamanaka, K., Mishory, A., Koola, J., … George, M. S. (2005). Decreased cortical response to verbal working memory following sleep deprivation. Sleep, 28(1), 55–67.
Mueller, S. T., & Piper, B. J. (2014). The Psychology Experiment Building Language (PEBL) and PEBL Test Battery. Journal of Neuroscience Methods, 222, 250–259.
Niles, A. N., Craske, M. G., Lieberman, M. D., & Hur, C. (2015). Affect labeling enhances exposure effectiveness for public speaking anxiety. Behaviour Research and Therapy, 68, 27–36.
Oatley, K., & Jenkins, J. M. (2003). Human Emotions: Function and Dysfunction. Annual Review of Psychology, 43(1), 55–85.
Ochsner, K. N., Bunge, S. A., Gross, J. J., & Gabrieli, J. D. E. (2002). Rethinking feelings: an
107 FMRI study of the cognitive regulation of emotion. Journal of Cognitive Neuroscience, 14(8), 1215–1229.
Ochsner, K. N., & Gross, J. J. (2005). The cognitive control of emotion. Trends in Cognitive Sciences, 9(5), 242–249.
Ochsner, K. N., Ray, R. D., Cooper, J. C., Robertson, E. R., Chopra, S., Gabrieli, J. D. E., & Gross, J. J. (2004). For better or for worse: Neural systems supporting the cognitive down- and up-regulation of negative emotion. NeuroImage, 23(2), 483–499.
Ochsner, K. N., Silvers, J. A., & Buhle, J. T. (2012). Functional imaging studies of emotion regulation: a synthetic review and evolving model of the cognitive control of emotion. Annals of the New York Academy of Sciences, 1251, E1-24.
Olver, J. S., Pinney, M., Maruff, P., & Norman, T. R. (2015). Impairments of Spatial Working Memory and Attention Following Acute Psychosocial Stress. Stress and Health, 31(2), 115–123.
Palmer, C. A., & Alfano, C. A. (2017). Sleep and emotion regulation: An organizing, integrative review.
Payer, D. E., Baicy, K., Lieberman, M. D., & London, E. D. (2012). Overlapping neural substrates between intentional and incidental down-regulation of negative emotions. Emotion (Washington, D.C.), 12(2), 229–235.
Payne, J. D., Jackson, E. D., Hoscheidt, S., Ryan, L., Jacobs, W. J., & Nadel, L. (2007). Stress administered prior to encoding impairs neutral but enhances emotional long-term episodic memories. Learning & Memory, 14(12), 861–868.
Pennebaker, J. W. (1997). Writing About Emotional Experiences as a Therapeutic Process. Psychological Science, 8(3), 162–166.
Perlis, M. L., Giles, D. E., Buysse, D. J., Tu, X., & Kupfer, D. J. (1997). Self-reported sleep disturbance as a prodromal symptom in recurrent depression. Journal of Affective Disorders, 42(2–3), 209–212.
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.
Phelps, E. A., & LeDoux, J. E. (2005). Contributions of the amygdala to emotion processing: from animal models to human behavior. Neuron, 48(2), 175–187.
Prather, A. A., Bogdan, R., & Hariri, A. R. (2013). Impact of Sleep Quality on Amygdala Reactivity, Negative Affect, and Perceived Stress. Psychosomatic Medicine, 75(4), 350– 358.
108 Pruessner, J. C., Kirschbaum, C., Meinlschmid, G., & Hellhammer, D. H. (2003). Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent change. Psychoneuroendocrinology, 28(28), 916–931.
Quirk, G. J., Likhtik, E., Pelletier, J. G., & Pare, D. (2003). Stimulation of Medial Prefrontal Cortex Decreases the Responsiveness of Central Amygdala Output Neurons. J. Neurosci., 23(25), 8800–8807.
R Core Team. (2015). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing.
Raio, C. M., Orederu, T. A., Palazzolo, L., Shurick, A. A., & Phelps, E. A. (2013). Cognitive emotion regulation fails the stress test. Proceedings of the National Academy of Sciences of the United States of America, 110(37), 15139–15144.
Raskind, M. A., Thompson, C., Petrie, E. C., Dobie, D. J., Rein, R. J., Hoff, D. J., … Peskind, E. R. (2002). Prazosin reduces nightmares in combat veterans with posttraumatic stress disorder. The Journal of Clinical Psychiatry, 63(7), 565–568.
Reddy, R., Palmer, C. A., Jackson, C., Farris, S. G., & Alfano, C. A. (2017). Impact of sleep restriction versus idealized sleep on emotional experience, reactivity and regulation in healthy adolescents. Journal of Sleep Research, 26(4), 516–525.
Roozendaal, B., McEwen, B. S., & Chattarji, S. (2009). Stress, memory and the amygdala. Nature Reviews. Neuroscience, 10(6), 423–433.
Rosales-Lagarde, A., Armony, J. L., del Río-Portilla, Y., Trejo-Martínez, D., Conde, R., & Corsi-Cabrera, M. (2012). Enhanced emotional reactivity after selective REM sleep deprivation in humans: an fMRI study. Frontiers in Behavioral Neuroscience, 6, 25.
Roth, R. H., Tam, S. Y., Ida, Y., Yang, J. X., & Deutch, A. Y. (1988). Stress and the mesocorticolimbic dopamine systems. Annals of the New York Academy of Sciences, 537, 138–147.
Saal, D., Dong, Y., Bonci, A., & Malenka, R. C. (2003). Drugs of Abuse and Stress Trigger a Common Synaptic Adaptation in Dopamine Neurons. Neuron, 37(4), 577–582.
Sagaspe, P., Sanchez-Ortuno, M., Charles, A., Taillard, J., Valtat, C., Bioulac, B., & Philip, P. (2006). Effects of sleep deprivation on Color-Word, Emotional, and Specific Stroop interference and on self-reported anxiety. Brain and Cognition, 60(1), 76–87.
Selye, H. (1987). Stress without Distress. London: Transworld.
Sheppes, G., Scheibe, S., Suri, G., & Gross, J. J. (2011). Emotion-Regulation Choice. Psychological Science, 22(11), 1391–1396.
109 Shermohammed, M., Mehta, P. H., Zhang, J., Brandes, C. M., Chang, L. J., & Somerville, L. H. (2017). Does Psychosocial Stress Impact Cognitive Reappraisal? Behavioral and Neural Evidence. Journal of Cognitive Neuroscience, 29(11), 1803–1816.
Silvers, J., Buhle, J., & Ochsner, K. N. (2013). The neuroscience of emotion regulation: basic mechanisms and their role in development, aging, and psychopathology. The Oxford Handbook of Cognitive Neuroscience, 2, 52–78.
Smith, M. T., Huang, M. I., & Manber, R. (2005, July). Cognitive behavior therapy for chronic insomnia occurring within the context of medical and psychiatric disorders. Clinical Psychology Review. Pergamon.
Smith, S. M., Jenkinson, M., Woolrich, M. W., Beckmann, C. F., Behrens, T. E. J., Johansen- Berg, H., … Matthews, P. M. (2004). Advances in Functional and Structural MR Image Analysis and Implementation as FSL Technical Report TR04SS2. NeuroImage, 23, S208– S219.
Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety Inventory.
Strang, N. M., Pruessner, J., & Pollak, S. D. (2011). Developmental changes in adolescents’ neural response to challenge. Developmental Cognitive Neuroscience, 1(4), 560–569.
Swanson, L. M., Arnedt, J. T., Rosekind, M. R., Belenky, G., Balkin, T. J., & Drake, C. (2011). Sleep disorders and work performance: Findings from the 2008 National Sleep Foundation Sleep in America poll. Journal of Sleep Research, 20(3), 487–494.
Tabibnia, G., Lieberman, M. D., & Craske, M. G. (2008). The Lasting Effect of Words on Feelings: Words May Facilitate Exposure Effects to Threatening Images. Emotion, 8(3), 307–317.
Tamir, M. (2009). What Do People Want to Feel and Why? Current Directions in Psychological Science, 18(2), 101–105.
Taylor, F., & Raskind, M. A. (2002). The alpha1-adrenergic antagonist prazosin improves sleep and nightmares in civilian trauma posttraumatic stress disorder. Journal of Clinical Psychopharmacology, 22(1), 82–85.
Taylor, S. E., Eisenberger, N. I., Saxbe, D., Lehman, B. J., & Lieberman, M. D. (2006). Neural Responses to Emotional Stimuli Are Associated with Childhood Family Stress.
Tempesta, D., Couyoumdjian, A., Curcio, G., Moroni, F., Marzano, C., De Gennaro, L., & Ferrara, M. (2010). Lack of sleep affects the evaluation of emotional stimuli. Brain Research Bulletin, 82(1–2), 104–108.
Torre, J. B., & Lieberman, M. D. (2018). Putting Feelings Into Words: Affect Labeling as
110 Implicit Emotion Regulation. Emotion Review, 10(2), 116–124.
Torrisi, S. J., Lieberman, M. D., Bookheimer, S. Y., & Altshuler, L. L. (2013). Advancing understanding of affect labeling with dynamic causal modeling. NeuroImage, 82, 481–488.
Tottenham, N., Tanaka, J. W., Leon, A. C., McCarry, T., Nurse, M., Hare, T. A., … Nelson, C. (2009). The NimStim set of facial expressions: judgments from untrained research participants. Psychiatry Research, 168(3), 242–249.
Troy, A. S., Wilhelm, F. H., Shallcross, A. J., & Mauss, I. B. (2010). Seeing the silver lining: Cognitive reappraisal ability moderates the relationship between stress and depressive symptoms. Emotion, 10(6), 783–795.
Tsumura, H., Sensaki, J., & Shimada, H. (2015). Stress-induced cortisol is associated with generation of non-negative interpretations during cognitive reappraisal. BioPsychoSocial Medicine, 9(1), 23.
Tull, M. T., Barrett, H. M., McMillan, E. S., & Roemer, L. (2007). A preliminary investigation of the relationship between emotion regulation difficulties and posttraumatic stress symptoms. Behavior Therapy, 38(3), 303–313.
Tull, M. T., Stipelman, B. A., Salters-Pedneault, K., & Gratz, K. L. (2009). An examination of recent non-clinical panic attacks, panic disorder, anxiety sensitivity, and emotion regulation difficulties in the prediction of generalized anxiety disorder in an analogue sample. Journal of Anxiety Disorders, 23(2), 275–282. van Ast, V. A., Spicer, J., Smith, E. E., Schmer-Galunder, S., Liberzon, I., Abelson, J. L., & Wager, T. D. (2014). Brain Mechanisms of Social Threat Effects on Working Memory. Cerebral Cortex (New York, N.Y. : 1991).
Van Cauter, E., Spiegel, K., Tasali, E., & Leproult, R. (2008). Metabolic consequences of sleep and sleep loss. Sleep Medicine, 9 Suppl 1(0 1), S23-8. van der Helm, E., Yao, J., Dutt, S., Rao, V., Saletin, J. M., & Walker, M. P. (2011). REM sleep depotentiates amygdala activity to previous emotional experiences. Current Biology, 21(23), 2029–2032.
Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F. (2003). The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep, 26(2), 117–126. van Marle, H. J. F., Hermans, E. J., Qin, S., & Fernández, G. (2009). From Specificity to Sensitivity: How Acute Stress Affects Amygdala Processing of Biologically Salient Stimuli. Biological Psychiatry, 66(7), 649–655.
111 Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: the PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070.
Woolrich, M. W., Behrens, T. E. J., Beckmann, C. F., Jenkinson, M., & Smith, S. M. (2004). Multilevel linear modelling for FMRI group analysis using Bayesian inference. NeuroImage, 21(4), 1732–1747.
Yoo, S. S., Gujar, N., Hu, P., Jolesz, F. A., & Walker, M. P. (2007, October 23). The human emotional brain without sleep - a prefrontal amygdala disconnect. Current Biology. Cell Press.
Zohar, D., Tzischinsky, O., Epstein, R., & Lavie, P. (2005). The Effects of Sleep Loss on Medical Residents’ Emotional Reactions to Work Events: a Cognitive-Energy Model. Sleep, 28(1), 47–54.
112 Appendix
SupplementarySupplementary Table Table 1. Brain S1 .Regions Brain RegionsRecruited RecruitedDuring CR TaskDuring CR Task (Paper 2) MNI coordinates Region Side Extent Max Z x y z Reappraisal contrast (decrease-negative > look-negative) Inferior Frontal Gyrus L 46299 7.18 -56 24 6 Inferior Temporal Gyrus L 5.96 -60 -24 -22 Middle Temporal Gyrus L 5.54 -60 -38 0 Temporal Pole L 5.51 -52 12 -28 Inferior Frontal Gyrus L 5.05 -52 26 22 Angular Gyrus L 6.03 -50 -60 30 Frontal Pole L 6.25 -48 40 -12 Orbital Frontal Cortex L 6.65 -44 22 -12 Middle Frontal Gyrus L 6.42 -42 14 50 Anterior Insula L 6.97 -36 20 -6 Cerebellum L 5.58 -36 -68 -26 Lateral Occipital Cortex L 5.36 -36 -76 46 Putamen L 6.02 -16 8 0 Left Caudate L 5.81 -16 -2 20 Globus Pallidus L 5.4 -12 -4 -2 Paracingulate Gyrus L 6.05 -8 20 44 Left Thalamus L 5.1 -8 -14 4 Substantia Nigra L 5.12 -6 -16 -12 Superior Frontal Gyrus L 6.64 -4 12 62 Posterior Cingulate L 5.39 -2 -52 8 Cingulate Gyrus R 5.96 8 30 22 Caudate/Putamen R 5.66 12 14 2 Caudate R 6.1 16 4 16 Superior Frontal Gyrus R 6.3 18 34 48 Cerebellum R 6.18 30 -60 -32 Anterior Insula R 5.64 34 20 -4 Orbital Frontal Cortex R 5.74 42 24 -14 Frontal Pole R 5.45 50 42 -10 Inferior Temporal Gyrus R 5.17 52 -4 -34 Temporal Pole R 5.13 52 14 -26 Inferior Frontal Gyrus R 5.93 58 24 6 Angular Gyrus R 616 4.62 52 -54 30 Cerebellum R 521 4.64 8 -48 -48
Reactivity contrast (look-negative > look-neutral) Fusiform Gyrus R 8027 6.49 42 -60 -18 Lateral Occipital Cortex, superior division R 6.46 42 -62 18 Lateral Occipital Cortex, inferior division R 6.35 50 -74 -6 Lateral Occipital Cortex, inferior division L 6852 6.57 -48 -82 0 Lateral Occipital Cortex, superior division L 5.95 -48 -68 14 Fusiform Gyrus L 5.68 -40 -44 -20 Anterior Insula L 2294 5.35 -30 26 0 Middle Frontal Gyrus L 5.29 -36 0 50 Precentral Gyrus L 4.79 -48 4 32 Frontal Operculum L 4.49 -44 26 0 Inferior Frontal Gyrus L 4.48 -44 24 20 Inferior Frontal Gyrus, pars triangularis R 2052 5.49 58 30 2 Anterior Insula R 5.04 34 30 -2 Inferior Frontal Gyrus, pars opercularis R 4.48 50 16 30 Precentral Gyrus R 4.32 34 2 34 Periaqueductal Gray R 1912 6.05 4 -30 -2 Ventral Tegmental Area L 5.02 -6 -8 -14 Caudate L 4.75 -8 6 8 Thalamus L 4.7 -10 -14 10 Posterior Thalamus R 4.59 20 -30 0 Posterior Thalamus L 4.24 -24 -24 -8 Thalamus R 4.12 10 -12 10 Superior Frontal Gyrus R 1766 6.33 6 12 58 Paracingulate Gyrus L 5.31 -6 18 50 Superior Frontal Gyrus L 4.62 -14 18 62 Superior Parietal Lobule L 526 5.61 -26 -50 40 Cerebellum R 402 5.58 2 -50 -34
113 A 2.0 Males B 2.0 Females Control Control Stress Stress
1.5 1.5
1.0 1.0 Rating Difference Rating Difference 0.5 0.5
0.0 0.0 Reappraisal Success Emotional Reactivity Reappraisal Success Emotional Reactivity
Supplementary Figure S1. Reappraisal success and emotional reactivity examined separately in male (A) and female (B) participants. Selective cortisol responding in males motivated us to examine the effects of stress on CR in this sub-sample. Similar to the full sample, males showed no effect of stress on reappraisal success (F(1,22)=1.182, p=0.289).
114