Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Latent physiological state dynamics underlying everyday affect, affect regulation and cognition
(USC Thesis Other)
Latent physiological state dynamics underlying everyday affect, affect regulation and cognition
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Copyright 2024 Jiani Li
Latent Physiological State Dynamics Underlying Everyday Affect, Affect Regulation and
Cognition
by
Jiani Li
A Thesis Presented to the
FACULTY OF THE USC DORNSIFE COLLEGE OF LETTERS, ARTS AND SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTERS OF SCIENCE
PSYCHOLOGY
December 2024
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
ii
TABLE OF CONTENTS
List of Tables ................................................................................................................................. iii
List of Figures................................................................................................................................ iv
Abstract............................................................................................................................................v
Chapter One: Introduction ...............................................................................................................1
Neurophysiological Mechanism of Self-Regulation............................................................2
Ambulatory Assessment ....................................................................................................12
Person-Specific Prediction of Psychopathology................................................................14
Gaps in Literature ..............................................................................................................17
Research Questions and Hypotheses .................................................................................20
Chapter Two: Method....................................................................................................................22
Participants.........................................................................................................................22
Procedure ...........................................................................................................................23
Measures............................................................................................................................24
Cardiac Indices...................................................................................................................27
Chapter Three: Data Analysis........................................................................................................31
Idiographic State Extraction ..............................................................................................31
Model Selection & Between-Person State Alignment.......................................................33
Computation of State Dynamic Statistics..........................................................................34
Multilevel Prediction of Affective and Cognitive Outcomes............................................35
Chapter Four: Results ....................................................................................................................39
Idiographic State Discovery & Theory-Based State Alignment........................................39
Multilevel Prediction of Psychological Outcomes ............................................................44
Chapter Five: Discussion ...............................................................................................................71
Negative Affect (NA) ....................................................................................................... 72
Brooding ............................................................................................................................74
Mind-Wandering................................................................................................................77
Momentary Impulsivity .....................................................................................................80
Theoretical and Clinical Implications of Ambulatory Physiological Monitoring .............81
Discretizing Ambulatory Time-Series into Multivariate States.........................................82
Unique Affect/Cognition-Physiology Coupling in rMDD ................................................84
Limitations & Future Directions........................................................................................87
Chapter Six: Conclusion ................................................................................................................89
References......................................................................................................................................90
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
iii
LIST OF TABLES
Table 1: List of Psychological Outcomes Measured via EMA......................................................26
Table 2: List of Physiological State Constituents..........................................................................30
Table 3: State-EMA Slopes Tested in the Two-Way Interactions.................................................35
Table 4: GLM Families for the EMA Outcome Variables............................................................38
Table 5: Statistically Significant Results for Stressed States.........................................................41
Table 6: Statistically Significant Results for Relaxed States.........................................................47
Table 7: Statistically Significant Results for Active States...........................................................55
Table 8: Statistically Significant Results for Average States ........................................................62
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
iv
LIST OF FIGURES
Figure 1: EMA Schedule ...............................................................................................................24
Figure 2: Stressed States Across All Participants..........................................................................41
Figure 3: Relaxed States Across All Participants..........................................................................42
Figure 4: Active States Across All Participants.............................................................................43
Figure 5: Average States Across All Participants..........................................................................44
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
v
Abstract
Human physiology reflects the body’s capacity for self-regulation that is crucial for
flexible adaptation to changing environmental demands. Leveraging wearable sensors and
machine learning, the current study aims to uncover latent physiological states in ambulatory
recordings of cardiac, respiratory and activity signals that predict momentary affective outcomes
relevant to depression. Participants with remitted major depressive disorder (rMDD) and healthy
controls (HCs) completed seven-day ecological momentary assessments of affect, affect
regulation and impulsivity while being passively monitored for heart rate, respiration and
movement. Results show that transitions into a stressed physiological state predict worse
psychological outcomes in most cases, including elevated negative affect, increased engagement
in maladaptive emotion regulation (ER), and higher momentary impulsivity. Furthermore, the
effects of stressed and active states on psychological outcomes differed by group. Compared to
individuals with rMDD, HCs showed more consistent links between stressed states and
unfavorable affective and cognitive outcomes, and more consistent relationships between active
states and favorable psychological outcomes. Findings underscore the utility of passive
physiological monitoring in tracking momentary affective processes that could otherwise be hard
to actively sample. Moreover, group differences in the relationships between physiological state
dynamics and psychological outcomes suggest potential alterations in physiology-affect coupling
related to depression history that could inform novel physiological mechanisms and targeted
treatments of depression.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
1
Chapter One: Introduction
A leading cause of disability and death, major depressive disorder (MDD) affects 322
million individuals worldwide and presents a burden of $326.2 billion to the US economy
(Greenberg et al., 2021). Central to MDD is dysfunctional emotion regulation, which is the target
of many existing psychotherapies for MDD such as the Cognitive Behavioral Therapy
(Rothbaum et al., 2000). However, the dynamic and context-dependent nature of emotion and
emotion regulation (ER) presents challenges to both researchers and clinicians. To researchers,
this means that conventional methods of studying ER, which are typically cross-sectional and
carried out in laboratory settings, may not translate to better understanding of real-world
emotional experience. To clinicians, this means that seeing their clients weekly in therapy
sessions might be insufficient to address rapid fluctuations in emotion and behavior outside of
the clinic. Thanks to technological advancements in smartphones and wearable devices,
researchers can now continuously monitor psychological processes as they occur in the real
world, which can facilitate research on the temporal dynamics of affect and affect regulation and
inform optimal timing of interventions. Furthermore, evidence suggests that atypical autonomic
nervous system (ANS) activities and movement patterns, which could be passively tracked
outside of the lab, underlie inflexible regulation and mood difficulties, raising the possibility of
finding biomarkers for mood dysregulation (Stange et al., 2017a; Yaroslavsky et al., 2016). In
this study, we leveraged ambulatory recordings of candidate physiological biomarkers for affect
and affect regulation, including cardiac signals, respiration and physical activity, to uncover
latent physiological states whose dynamics predicted momentary affective and cognitive
outcomes. In the following paragraphs, I will: 1) explain the constituents of physiological states
by surveying the literature on the neurophysiological underpinnings of self-regulation; 2)
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
2
introduce ambulatory assessment, a method for collecting data in real-time and in real-world
settings; 3) review studies that predicted clinically relevant outcomes using within-subject selfreport, behavioral and physiological data. This line of work has the potential to: 1) elucidate the
proximal physiological risk factors of emotion dysregulation; 2) identify people who are at high
risk for emotion dysregulation; and 3) enable real-time monitoring of physiological risk factors
and inform just-in-time delivery of interventions to people during moments of potential affect
dysregulation.
Neurophysiological Mechanisms of Self-Regulation
The Neurovisceral Integration Model
Physical, cognitive, affective, or social, the environmental influences placed on the body
vary in kinds and degrees over time. The capacity for self-regulation is thus necessary for the
body to maintain homeostasis in face of changing demands. According to the Neurovisceral
Integration model, the body’s ability to flexibly self-regulate depends on the extent to which it
can orchestrate complex interactions between bodily components, most notably between the
brain and the heart (Thayer & Lane, 2000). In particular, the central autonomic network (CAN)
in the brain receives information about the external environment through afferent nerves from
the rest of the body. Through complex interactions between its constituent regions, the CAN
integrates the incoming sensory information with mental representations of beliefs and goals to
produce goal-directed behavior. The CAN then outputs inhibitory control to the heart via the
sympathetic and parasympathetic nervous systems (SNS and PNS). Thus, the extent to which the
CAN can integrate information from the external world and generate context-appropriate
emotional response is reflected by the complex variability in its output to the body (Smith et al.,
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
3
2017). One important CAN output to the body is the dynamically variable heart beat rhythms
that characterize a healthy cardiovascular system.
Cardiac Output
Innervated by the SNS and the PNS via the stellate ganglia and the vagus nerves, the
heart receives inhibitory control from the CAN. In response to changing external and internal
demands, the CAN dynamically increases and withdraws output to the SNS and PNS, resulting
in complex variability in heart rate (HR). Specifically, the extent to which HR rhythms are
controlled by the PNS as opposed to the SNS, termed cardiac vagal control, is tightly associated
with self-regulation capacity (Laborde et al., 2018). Physiologically speaking, while both
branches of the ANS influence HR, the PNS exerts faster control over HR through the vagus
nerves than the SNS (Berntson et al., 1997). Therefore, beat-to-beat variability in HR is mostly
attributed to dynamic changes in vagal tone, which explains why most studies focus on vagallymediated HR complexity as an index for regulatory flexibility.
Why might HR complexity reflect self-regulation capacity? First, the physiological
system and its subcomponents exhibit properties of a complex dynamical system (Thayer and
Lane, 2000). According to dynamical systems theory, complex systems consist of constituents
that interact and mutually constrain one another through feedback loops (Bar-Yam, 2020).
Similarly, different parts of the body communicate with each other via intricate nervous and
endocrine systems. These complex internal interactions result in an intrinsic equilibrium that
enables the system to respond to an impressively wide range of inputs and produce highly
variable outputs accordingly. In other words, physiological complexity permits adaptive selfregulation. Therefore, as the primary output of the physiological system, HR displays complex
variability across time that is indicative of the underlying physiological flexibility, which
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
4
prepares an organism for various possible challenges. To support this theory, a wealth of
research suggests that inflexible autonomic regulation, manifested as diminished HR complexity,
is associated with affect dysregulation and mood disorder diagnoses (Chalmers et al., 2014;
Kemp et al., 2010, 2014; Koenig et al., 2016; Stange et al., 2020; Yaroslavsky et al., 2013a).
Therefore, reduction in HR complexity may be a viable biomarker for psychopathology.
Complexity is assessed through three domains: variability, unpredictability, and self-similarity.
Heart rate variability (HRV) is defined as variation in the length of beat-to-beat (specifically Rto-R) intervals (Shaffer & Ginsberg, 2017). Because parasympathetic activity operates in
relatively short timescales and is primarily responsible for the beat-to-beat changes in HR, most
studies on HRV and affective outcomes focus on short-term HRV as an indicator of vagal
influences. A wealth of research supports a positive relationship between HRV and physical and
mental health. Studies have associated momentary decrease in HRV with the presence of acute
stressors and negative affect (Dikecligil & Mujica-Parodi, 2010; Luo et al., 2018; Wagner et al.,
2019). Furthermore, resting-state HRV is often lower among individuals with MDD and is
inversely correlated with depressive symptom severity, suggesting that reduced vagal control
may underlie depression symptomatology (Chang et al., 2012; Kemp et al., 2010; Koenig et al.,
2016). Finally, whereas it is normal for HRV to decrease following stress and rebound
afterwards, such HRV reactivity is attenuated among individuals with MDD, indicating that
autonomic inflexibility may contribute to affect dysregulation and mood disorders (Bylsma et al.,
2014). The well-established link between HRV and regulation flexibility has led researchers to
propose HRV as a transdiagnostic biomarker for self-regulation that is heavily implicated in
psychopathology such as depression (Beauchaine & Thayer, 2015; Yaroslavsky et al., 2014).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
5
Despite the association between HRV and regulation flexibility, linear analysis of HRV
may not be sufficient for understanding a nonlinear physiological system. Mounting evidence
suggests that the physiological system is nonlinear in nature, which means that as system
components (e.g., the brain, the heart, and the peripheral organs) interact with each other, the
output (e.g., HRV) does not fluctuate in proportion to changes in the input (e.g., environmental
demands) (Young & Benton, 2015). Therefore, apart from analyzing HRV itself, it is also
important to examine patterns of HRV over time. Specifically, complex systems produce outputs
that vary in an unpredictable fashion, which is different from variability per se. An example that
dissociates variability from unpredictability would be a sine wave, which is strictly periodic yet
quite variable. Further evidence for the distinction between HR variability and unpredictability
comes from research that suggests different physiological underpinnings of these two constructs.
In contrast to short-term HRV which primarily reflects parasympathetic modulation, HR
unpredictability may index the balance between both arms of the ANS. Specifically, a shift
towards parasympathetic dominance, either through blocking the SNS via head-up tilt or through
stimulating the vagus nerve, can cause an increase in HR unpredictability. On the other hand,
pharmacological blockade of the PNS with atropine or glycopyrrolate or activation of the SNS
with propranolol reduces HR unpredictability. (Lepoluoto et al., 2005; Liu et al., 2018; Penttilä
et al., 2003; Porta et al., 2007, 2013). Thus, nonlinear analysis of the irregularity of HRV over
time might tap into dynamic sympathovagal balance, thereby complementing the traditional
linear HRV analysis that primarily reflects the vagal tone. The unpredictability of time series can
be quantified by entropy, which measures the information content (i.e., surprisal) of a time
series. Intuitively, an unpredictable sequence is more information-dense, whereas segments of a
highly regular sequence carry repetitive information and are thus redundant. Higher HR entropy
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
6
at rest has been associated with better physical and mental health outcomes (Bakhchina et al.,
2018). Conversely, diminished resting-state entropy is related to medical diseases, aging and
MDD (de la Torre-Luque et al., 2016; Pikkujämsä et al., 1999). Furthermore, momentary entropy
loss is associated with negative affect and stress states (Young & Benton, 2015). Importantly, the
experience of negative affect and stress may result from ineffective emotion regulation attempts,
as studies have explicitly linked reduced entropy to the use of maladaptive emotion regulation
strategies (Fiol-Veny et al., 2019). These findings thus suggest that autonomic inflexibility may
underlie cognitive inflexibility and ultimately negative affective outcomes.
Finally, output from a complex system is not just unpredictable at a certain time scale. To
cope with challenges that occur at various time scales (e.g., the sudden presence of a spider
versus a looming sense of worry), the body has physiological subsystems that operate at different
speeds: vagal withdrawal happens shortly after exposure to stress, followed by slower changes in
SNS activity, and the hypothalamus-pituitary-adrenal axis may then be activated if PNS and SNS
modulations are not sufficient (Erath & Pettit, 2021). The multiscale regulation mechanism that
the body is equipped with means that it is capable of producing responses that are similarly
complex across multiple time scales – a property termed self-similarity or scale invariance. The
concept of self-similarity was first applied to study fractals, which are a class of objects that have
the same geometric shape at different levels of magnification (Feder & Feder, 1988). In other
words, each subunit of a fractal resembles the macroscopic shape of the object. Similarly, the
complexity of a scale-invariant time series is preserved across time scales through long-range
correlations in the sequence. Thus, greater self-similarity indicates greater complexity, which is
then associated with positive physical and mental health outcomes (Alvares et al., 2013; Kwon et
al., 2019). Physiologically speaking, fractal dynamics of heart rate seems to be sensitive to both
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
7
branches of the ANS, as pharmacological blockade of both systems can result in changes in
cardiac self-similarity, although this is still an area of debate (Castiglioni et al., 2011).
In summary, a healthy and optimally responsive physiological system generates variable,
information-rich, and scale-invariant cardiac outputs, enabling itself to adapt to a broad range of
demands. Such HR complexity may be compromised momentarily during states of stress, but
persistent complexity loss could indicate regulation inflexibility and thus confer risks for
psychopathology.
Respiration
In conventional analysis of HR complexity, respiration is often treated as a nuisance
variable to be controlled for due to its strong influence on HR (Bernardi et al., 2001). HR
increases during inhalation and drops during exhalation, and this respiration-coupled HRV,
termed respiratory sinus arrhythmia (RSA), is understandably sensitive to changes in breathing
volume and frequency (Neff et al., 2003). Therefore, researchers have recommended controlling
for the effects of breathing on HR complexity by, for example, forcing participants to breathe at
a particular rate or statistically adjusting for them post hoc (Laborde et al., 2017). However, we
decided to include respiration as a constituent of physiological states because of two reasons.
First, evidence suggests that indices of HR complexity may be affected differentially by
breathing. Frequency-domain measures of HRV, such as the high-frequency power of HRV, are
very sensitive to respiratory rhythms. In contrast, time-domain metrics of HRV, such as the root
mean square successive differences, are relatively unaffected by breathing (Hill et al., 2009;
Penttilä et al., 2001). In addition, the impact of respiration on the unpredictability and selfsimilarity properties of HR complexity seems to be more nuanced, with some studies showing no
influence and others suggesting the opposite (Kanters et al., 1997; Weippert et al., 2015). But
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
8
more importantly, regardless of the interactions between breathing and HR complexity,
respiration should still play a nontrivial role in the current study that is concerned with finding
biomarkers of affect and affect regulation. The colloquial use of “take a deep breath” when one
is experiencing anxiety symptoms shows our intuitive appreciation of respiratory modulation of
affect. Indeed, various breathing interventions have been proposed for managing stress and
anxiety disorders (Cho et al., 2016; Meuret et al., 2008). In addition to changing mood, breathing
patterns can also be altered during moments of stress and in people with anxiety disorders
(Sartory et al., 1992; Stein et al., 1995; Tiller et al., 1987). Thus, atypical breathing may indeed
be a physiological signature of affect dysregulation at both the between-person and withinperson levels, and ambulatory monitoring of respiratory patterns might allow for automated
detection and prediction of moments of distress and psychopathology.
Perhaps more interesting is the possibility of using not just respiration per se but the
combination of respiration and cardiac and motor signals to predict affect dysregulation. Indeed,
although variation in breathing occurs at both the intraindividual (i.e., stressful vs. neutral
moments) and the interindividual levels (i.e., people with vs. without anxiety disorders), using
daily breathing patterns alone to predict diagnostic status has proven to be difficult. Pfaltz and
coworkers (2009) reported no difference between people with panic disorder and those without
in their respiratory trajectory throughout the day. Moreover, hyperventilation, measured via
blood levels of carbon dioxide, does not always co-occur with episodes of panic attacks (Garssen
et al., 1996; Hibbert & Pilsbury, 1989). We speculate two explanations for this seeming
discrepancy. First, breathing is affected by various non-affective factors, one of which is physical
activity. Elevated respiration rate and volume during exercise is normative, and can really only
be treated as hyperventilation in the absence of changes in physical activity. This underscores the
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
9
importance of assessing respiration and physical activity simultaneously, which would allow us
to define abnormal breathing patterns relative to concurrent levels of physical demand. Second,
evidence suggests that respiration affects mood through altering HR complexity (Nashiro et al.,
2023a). For example, RSA, the portion of HRV attributable to respiratory rhythms, is associated
with adaptive emotion regulation (Butler et al., 2006). In contrast, poor coupling of HRV and
breathing (i.e., low RSA) predicts worse mental health outcomes (Beauchaine, 2015). In other
words, changes in respiration can impact mood in so far as it modulates heart beat rhythms.
Therefore, a physiological state characterized by comparable levels of respiration rate and HRV,
and another that features discrepancy between the two vital signs, could be differentially related
to affect, even if absolute levels of respiration rate or HRV might be similar across the two
states. The current study, which passively measured multiple physiological channels
concurrently, is well-suited to test this hypothesis.
Physical Activity
The close connection between physical activity and emotion has long been recognized.
An active lifestyle is related to elevated positive mood, fewer symptoms of anxiety and
depression, lower relapse rates, and prevention of depression onset (Babyak et al., 2000;
Paffenbarger Jr et al., 1994; Ross & Hayes, 1988). Furthermore, studies that track people’s
behavior in real time show that higher levels of daily physical activity are momentarily
associated with increased positive affect and reduced negative affect (Kanning & Schlicht, 2010;
Kolar et al., 2020). In contrast, sedentary behavior predicts momentary negative affect withinperson (Giurgiu et al., 2019). Finally, apart from enhancing mood directly, physical activity
facilitates cognitive emotion regulation as well. Performing moderate-intensity aerobic exercise
is associated with better domain-general cognitive control, more effective use of cognitive
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
10
reappraisal, and attenuation of the relationships between regulatory difficulties and negative
affect (Bernstein & McNally, 2017; Giles et al., 2018). In contrast, maladaptive emotion
regulation (e.g., rumination) mediates the positive relationship between sedentary lifestyle and
mood disorder symptoms (Bernstein & McNally, 2018).
On the other hand, emotional disorders often feature motor disturbance. For instance, a
symptom of MDD is psychomotor retardation or agitation (American Psychiatric Association,
2013). Studies using accelerometers or actometers to track people’s daily activity suggest that
individuals with MDD exhibit reduced motor activity during daytime and elevated motor activity
at night compared to healthy controls (van Londen et al., 1998; Volkers et al., 2003). Atypical
activity patterns are also observed in individuals with other mood disorders, such as anxiety
disorders and bipolar disorder (Difrancesco et al., 2019; Jones et al., 2005). Thus, altered
physical activity may represent a transdiagnostic biomarker underlying multiple disorders of
emotion regulation.
Apart from being associated directly with mood, physical activity is also a key indicator
of cardiopulmonary health and interacts with HR complexity and respiration. Exercise enhances
cardiac health and is associated with higher baseline HR complexity (Laborde et al., 2018;
Sandercock et al., 2005). However, studies examining HR complexity, specifically HRV, during
exercise have reported inconsistent results. This is because most of these studies employed
spectral and/or time-domain analysis of HRV that assumes non-stationarity in the data, which is
violated by the increasing trends in HR during exercise (Tulppo et al., 1996). Despite
methodological limitations, there is a general consensus that during exercise, the PNS
progressively withdraws its input to the heart while the SNS activity is augmented, resulting in a
momentary decline in HR complexity and increase in HR (Aubert et al., 2003). Furthermore,
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
11
HRV decreases up to the ventilatory threshold, which is the exercise intensity at which
ventilation starts to increase at a faster rate than the volume of oxygen intake (Garcia-Tabar et
al., 2013). Beside the general inverse relationship between HR complexity and exercise,
individual differences exist in the degree of this relationship and reflect cardiorespiratory fitness.
Studies comparing individuals across levels of aerobic fitness showed that greater aerobic fitness
is associated with 1) smaller HR complexity decline during exercise; 2) faster initial HRV
response to the onset of exercise (10-15s into the exercise); 3) higher exercise intensity (i.e.,
ventilatory threshold) at which vagal modulation of HR disappears (i.e., more protracted HRV
response to exercise) (D’Agosto et al., 2014; Tulppo et al., 1998).
Because both physical activity and psychological stress can reduce HR complexity,
including physical activity into analyses would allow us to distinguish between these two
possible sources of HR complexity decline. For example, the co-occurrence of low HR
complexity and high physical activity may be normative, whereas low HR complexity
accompanied by minimal physical activity may indicate the presence of a “disordered''
physiological state that foreshadows regulation difficulties and negative affect. Researchers of
HR complexity and mental health have long recognized the influence of metabolic demands like
movement on heart beat rhythms. However, the conventional approach to the interdependency
between physical activity and HR complexity is to control for movement statistically or
experimentally (Grossman et al., 2004; Laborde et al., 2017). This is reasonable given that in
most cases, the research question is whether HR complexity might be a biomarker for affect
dysregulation, which necessitates an unconfounded measure of HR complexity. However, if the
research question is whether distinct configurations of HR complexity, respiration and physical
activity, as opposed to HR complexity per se, might be a biomarker for affect dysregulation, then
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
12
physical activity should not be merely treated as a confound. Indeed, profiles of physical activity
and cardiorespiratory variables may offer predictive power above and beyond HR complexity
itself, since strong bidirectional relationships exist between bodily movement and mood (as
mentioned above). By including as many physiological indices that are strongly related to affect
into the feature set as possible, we hope to comprehensively map out the physiological dynamics
underlying affect and affect regulation and to enhance predictions accordingly.
In short, we argue that individuals traverse through discrete multivariate physiological states
over time that underlie moments of psychological vulnerability, and that HR complexity,
respiration and physical activity are all components of these physiological profiles. To test this
hypothesis, we would need to simultaneously track movement and cardiorespiratory indices in
daily life, which is made possible via the refinement of ambulatory assessment methods.
Ambulatory Assessment
Affective and physiological processes are dynamic and context-dependent in nature. For
example, decrease in HR complexity in the presence of acute psychological stressors is
normative, while prolonged depression in HR complexity that cannot be explained by metabolic
demands may be pathological (Verkuil et al., 2016). Therefore, measuring the temporal
trajectory of affective and physiological processes in situ, in vivo, and across multiple contexts is
necessary. Ambulatory assessment (AA), which refers to methods for collecting data as the
process unfolds in real time, are well-suited for these purposes (Trull & Ebner-Priemer, 2013).
Depending on the type of data collected, AA can be divided into active and passive
measurements (Stange et al., 2019). Active AA involves intensive sampling of self-report data
over time. One example is the ecological momentary assessment (EMA), which involves
repeatedly sending out surveys on smartphones to capture the variable of interest at the moment.
On the other hand, passive AA collects data outside of participants’ conscious awareness, and
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
13
can take the form of recording via wearable devices (e.g., Fitbit wristbands that track HR). AA
stands in contrast to conventional assessments of psychological variables in three ways. First,
AA captures the psychophysiological process as it unfolds in real time, whereas previous studies
typically rely on retrospective self-report which is prone to recall biases (Ebner-Priemer & Trull,
2009). Second, repeated assessments via AA enable the study of temporal dynamics, which is an
advance from cross-sectional measurement of average values (e.g., average mood over the past
week). By charting out the exact time course of the process, we can identify proximal
antecedents of emotional difficulties, which could offer additional prognostic value on top of
distal, trait-like risk factors for affective disorders (e.g., female sex, rumination) (Hankin et al.,
1998; Stange et al., 2016). Third, idiographic data collected via AA permits the study of both
intraindividual and interindividual differences in person-specific time courses of affect and
bodily states. With the help of AA, we can address within-subject questions about when a person
is at greater risk for mood difficulties by examining associations between variables of interest
(e.g., physiological composite states and momentary affect) within each person’s time series.
Common patterns of within-person associations could then be extracted to answer betweensubject questions about who is at elevated risk for affect dysregulation (Kleiman et al., 2018;
Lane et al., 2019).
Furthermore, the contrast between active and passive AA underscores the benefits of
finding physiological markers of affect regulation. Although recall biases in active AA are
somewhat attenuated by the increased sampling rate, participants are nevertheless asked to report
their experience during the past and can thus be inaccurate. Yet, researchers cannot indefinitely
increase the sampling rate either in search for accurate reporting, as it would result in participant
burden and may significantly skew the results if the construct of interest is affect. Physiological
processes, on the other hand, can be tracked passively by wearable devices. This means that the
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
14
tradeoff between measurement frequency and fidelity (i.e., that intensive assessment of a
psychological process might alter the process itself) is lessened for physiological dynamics
measured by passive AA. Therefore, if physiological signatures of affect dysregulation do exist,
which is what the current study aims to establish, then they would become a suitable proxy for
affect that we can monitor with fine time resolution and minimal disruption of one’s daily
routine. This would make them perfectly compatible for just-in-time interventions: biosensors
can detect the transient presence of physiological markers of mood dysregulation in daily life and
trigger the delivery of interventions in the exact moment of, if not preceding, an emotional crisis.
In summary, tailoring the time resolution of data collection to the speed of change of the
construct via AA has positive implications on research, as it enhances measurement accuracy,
permits the study of proximal risk factors, opens up a new dimension (i.e., time) along which
individual differences could underlie psychopathology. Furthermore, passive AA could advance
our design of just-in-time interventions by enabling fine-grained measurement while requiring
minimal user engagement.
Person-Specific Prediction of Psychopathology
Opportunities to examine within-person dynamics enabled by AA have encouraged
researchers to build person-specific prediction models of psychopathology. Here, the question of
interest is not who is at risk for mood disorders (i.e., interindividual differences), but when a
given person is in danger of affect dysregulation (i.e., intraindividual differences). The
assumption here is that the relationship between two variables at the between-subject level does
not always generalize to the within-person level, as illustrated by the example of exercise and
HR mentioned previously (i.e., HR increases with exercise within-person, but decreases with
exercise between-person). This means that group-level predictors of psychopathology are not
necessarily applicable to individuals (Fisher et al., 2018). Therefore, using intraindividual
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
15
variation in constructs of interest to predict psychopathology may lead to better prediction
accuracy and thus represent a step towards precision psychiatry.
Depending on the predictors used, approaches to building idiographic prediction models
of psychopathology can be divided into two classes: one that is concerned with the relationships
between variables, and the other that examines absolute levels of variables. A notable example of
the first approach is temporal network models of psychopathology, which assume that
psychopathology emerges from causal interactions between symptoms (Bringmann et al., 2022).
For each individual’s symptom trajectory, temporal network analysis maps out the effects of
each symptom (“node”) on all other symptoms at the next time point. Network properties, such
as the centrality of a node in sustaining the entire network, could then be derived to understand
how a disorder is being generated, exacerbated and maintained through dynamic symptom
interactions (McNally, 2021). Researchers have related network structures of a wide range of
disorders to clinical outcomes, such as MDD and anxiety disorders (Fisher et al., 2017).
Although temporal networks permit the identification of person-specific risk factors
compared to cross-sectional studies of interindividual differences, the timescale at which
network metrics are computed may still be too coarse for precise determination of intervention
timing. Because a network model is fit to the entire time series and provides a single set of
estimates at the end of model fitting, it necessarily collapses information across the time series.
To discern the evolution of network structures over time, multiple networks would need to be
computed by prolonging the measurement period, obtaining a longer time series and dividing it
into segments that still contain enough timepoints for stable network estimation. This, however,
can be especially challenging for studies that utilize AA, which already places high demand on
the participant to comply with the intensive data collection protocol (e.g., repeatedly filling out
questionnaires, wearing biosensors every day).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
16
To recover moment-by-moment changes in affect and physiology, researchers have
proposed to classify timepoints into unique profiles of variable levels that could be thought of as
“states” (Fisher & Bosley, 2020). In other words, this approach is concerned with finding relative
rank-orders of variables that commonly co-occur in the time series, and pinpointing the moments
at which these specific configurations appear. For example, a person may often exhibit a state
characterized by high positive affect (PA), low negative affect (NA), and average amount of
physical activity. An advantage of this classification approach is that because each timepoint is
labeled with a state, dynamic transitions between states can now be investigated at the finest
timescale afforded by measurement (i.e., consistent with the sampling rate of AA). However,
there have been relatively few studies that classified timepoints within-person, as classification
has traditionally been done across individuals to reveal subgroups. One of the few studies of
idiographic state discovery was conducted by Howe and Fisher (2022) who applied a clustering
algorithm on time courses of posttraumatic stress disorder (PTSD) symptoms to uncover personspecific symptom states. Moreover, because the idiographic symptom states were heterogeneous
across people, they applied a second-level clustering algorithm on idiographic symptom states to
align them across individuals. They reported that, despite substantial variation in within-subject
PTSD states across people, generalizable configurations of PTSD symptoms still emerged from
the pooling of idiographic PTSD states. A further advantage of labeling timepoints as states is
that perseverance and transitions of states can now be examined, as exemplified by Liu and
colleagues (2023)’s study on intraindividual phenotypes of depression among high-risk youths.
Specifically, they applied multilevel hidden Markov models (HMM) on repeated measures of
depressive symptoms, and found three states characterized by low-depression, high-depression
and cognitive-somatic symptoms respectively. Furthermore, they reported sex differences in state
transition patterns, such that girls were more likely to switch from low-depression states to high-
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
17
depression or cognitive-physical states than boys. Finally, whereas the two aforementioned
studies examined self-report data only, Wu and colleagues (2020) employed clustering analysis
on continuous cardiac and respiration recordings. In this study, the researchers confirmed the
existence of distinct physiological states in individuals’ recordings. Furthermore, they showed
that across individuals, probabilities of state membership at each timepoint predicted symptoms
of MDD and anxiety (assessed once), suggesting that expression of different physiological states
may reflect different severity of mood disorder symptoms and may thus be clinically relevant.
Overall, despite being a nascent field, idiographic state discovery in multivariate time series of
affect and physiology has proven to be feasible and clinically meaningful, thus justifying more
research efforts.
Gaps in Literature
While there is substantial support for the association between each of the aforementioned
physiological variables (i.e., cardiac measures, respiration and physical activity) and MDD and
ER, several gaps exist in the literature that await further study. First, no study has combined
different HR complexity indices together with respiration and movement cross-modally to
predict affective outcomes, potentially due to the difficulty of multimodal ambulatory tracking in
real life. On one hand, although cardiac vagal tone has been proposed as a summary statistic of
self-regulation capacity, exactly which aspect of HR complexity (variability, unpredictability, or
self-similarity) best reflects cardiac vagal tone is still a subject of debate (e.g., Laborde et al.,
2018; Young & Benton, 2015). This underscores the importance of examining multiple metrics
of HR complexity simultaneously. On the other hand, mounting evidence suggests that cardiac,
respiratory and motor activities both reciprocally influence each other and are independently
associated with mental health outcomes (e.g., Giurgiu et al., 2019; Kanters et al., 1997; Koenig et
al., 2016; Nashiro et al., 2023a). This supports the clinical utility of integrating HR complexity,
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
18
respiration and movement signals into the same analysis. Much like why multivariate pattern
analysis is sometimes favored over univariate analysis of single-voxel or region-based activation
in neuroimaging research, multivariate patterns of physiological indices may carry information
above and beyond what each constituent variable could offer (Weaverdyck, Lieberman, &
Parkinson, 2020). Therefore, it is worth investigating configuration of multiple physiological
channels as a potential biomarker for affect dysregulation.
Furthermore, no studies have simultaneously examined both within-person and betweenperson relationships between multimodal physiological profiles and daily affective and
behavioral regulation. Prior research has separately related HR complexity and physical activity
to mood regulation at the intraindividual and interindividual levels (e.g., Giurgiu et al., 2019;
Stange et al., 2020; Stange, Li et al., 2023). Given the benefits of multivariate assessment
discussed above, it would make sense to apply the same multilevel paradigm to multivariate
physiological time series in search for physiological markers of affect dysregulation.
In addition, few ambulatory studies have examined autonomic flexibility and affect regulation
among individuals who have recently remitted from MDD, a population that is likely to have
high rates of relapse (Möller, 2009). Researchers have hypothesized regulation inflexibility as a
vulnerability factor for mood disorders, which may not be attenuated as symptoms resolve
(Yaroslavsky et al., 2013b). For example, antidepressant medications alleviate depressive
symptoms without elevating resting HRV (Kemp et al., 2010). And, laboratory studies suggest
that people who are prone to MDD (e.g., having just remitted from MDD, having recurrent MDD
episodes, or having severe depressive symptoms) display reduced HR complexity in face of
stressors, raising the possibility that autonomic inflexibility may be a trait-like vulnerability
factor for mood disorders (Kovács et al., 2016; Stange et al., 2017b; Yaroslavsky et al., 2014).
Despite initial laboratory evidence in favor of it, the hypothesis that autonomic inflexibility is a
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
19
vulnerability factor for mood disorders has yet to be tested in ecologically valid contexts.
Through ambulatory monitoring of affect, affect regulation and physiological states of
individuals with remitted MDD, we can examine if inflexible physiological responses may be a
relatively stable phenotype that underlie negative affect and maladaptive regulation despite
current symptom resolution. If individuals with remitted MDD indeed showed inflexible
autonomic responses predictive of poor affective outcomes, then monitoring these transient
physiological signatures of affect dysregulation would help inform methods and timing of
relapse prevention among people who recovered from MDD, a population that may be especially
in need of assistance.
Finally, a source of information that has been missed in previous studies lies in dynamic
transitions between multivariate physiological profiles. The idea of uncovering recurring patterns
of variable relationships and examining transitions between them is long-established in
neuroscience research of dynamic functional connectivity (e.g., Damaraju et al., 2014; Hansen et
al., 2015; Shakil et al., 2016). These neuroimaging studies typically conceptualize a particular
configuration of region-to-region connectivity as a “brain state.” Then, researchers would
compute metrics to quantify dynamic functional reorganization of neural networks, such as the
percentage of state presence (i.e., frequency), the amount of time spent in each state (i.e., dwell
time), and frequency of transitions between each pair of states. Research supports the clinical
utility of these state metrics above and beyond brain states themselves in differentiating
individuals with cognitive deficits and/or psychiatric disorders from those without (Briley et al.,
2022; Damaraju et al., 2014; Yao et al., 2019). By discretizing continuous measurements of
physiological variables into “bodily states” in the current study, we hoped to similarly uncover
information from state transition behavior to enhance predictions of affect and affect regulation
at both the interindividual and the intraindividual levels.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
20
Research Questions and Hypotheses
The current study is data-driven in nature and is the first to relate longitudinal multimodal
physiological profiles to daily affect, affect regulation and cognition. Therefore, while having
specific research questions in mind, we remained relatively hypothesis-free and open to a wide
variety of possible findings. To better illustrate our research questions, here we proposed
exemplar hypotheses based on the limited existing empirical evidence, while acknowledging that
they do not represent a definitive or comprehensive list of possible results.
Research Question 1: What patterns of cardiac, respiratory and activity variables (i.e.,
physiological states) might people display in everyday life?
Hypothesis 1: Individuals will vary in the numbers and kinds of physiological states. However,
based on prior literature on the psychological implications of physiological indicators, we
hypothesized that four types of physiological states may be shared by individuals:
a) An “average” state, characterized by average levels of all physiological variables;
b) A “stressed” state, characterized by reduced HR complexity, average or shallow and fast
breathing, and average or low physical activity;
c) A “relaxed” state, characterized by elevated HR complexity, average or deep and slow
breathing, and average or low physical activity;
d) An “active” state, characterized by reduced HR complexity, average or deep and fast
breathing, and elevated physical activity;
Research Question 2: Within-person predictions: Do the temporal dynamics of physiological
states (i.e., frequency, dwell time and transition probability) predict momentary affect, affect
regulation and impulsivity?
Hypothesis 2: There will be within-person associations between physiological state dynamics
and momentary affect, affect regulation and impulsivity. Despite not knowing what states we
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
21
would identify, we expected positive within-person associations between the frequency, dwell
time and transitions towards less favorable physiological states (e.g., “stressed” state) and less
favorable psychological outcomes (e.g., NA, maladaptive ER).
Research Question 3: Between-person predictions: Do the average temporal dynamics of
physiological states predict average levels of affect, affect regulation and impulsivity?
Hypothesis 3: There will be between-person associations between average physiological state
dynamics and average affect, affect regulation and impulsivity over the data collection period.
We expected positive between-person associations between the average frequency, dwell time
and transitions towards less favorable physiological states and less favorable psychological
outcomes. We also expected negative between-person associations between transitions out of
less favorable physiological states and less favorable psychological outcomes.
Research Question 4: Do individuals with remitted major depressive disorder (rMDD) differ
from healthy controls (HC) in terms of:
a) The number of physiological states they exhibit in everyday life?
b) The within-person and between-person associations between physiological state
dynamics (i.e., frequency, dwell time and transition probability) and momentary affect,
affect regulation and impulsivity?
Hypothesis 4: Individuals with rMDD will differ from HCs in terms of the number of states,
state dynamics and the relationships between state dynamics and affect, affect regulation and
impulsivity. We did not have hypotheses about group differences in the relationships between
physiological state dynamics and momentary affect, affect regulation and impulsivity.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
22
Chapter Two: Method
Participants
53 adults who remitted from MDD and 46 HCs participated in the study. We excluded
four participants during the HMM fitting stage who had less than three days of recording. We
further excluded one individual who had exactly three days of recording when performing
second-level clustering, as the amount of physiological data was still insufficient to enable more
than two states to emerge at the second-level. Moreover, we excluded one individual who had
only three idiographic states during state alignment via state concordance, as this approach
indicated four states per person to be the best alignment solution. 93 participants entered our
final analyses (42 HCs, 51 rMDD). Ages ranged from 18-37 years with a mean of 25.4 years (SD
= 3.89). 60.2% of the participants identified as female, and 45.2% identified as a non-white race.
Recruitment took place in the local community through digital and print advertisements, as well
as from a pool of past research participants. In order to qualify, potential participants needed to
be fluent in English, possess normal or corrected vision, and not have any known cardiac
arrhythmias affecting HR complexity. Cash compensation was provided to participants. The
Institutional Review Boards at the University of Southern California and the University of
Illinois Chicago approved the study. The Diagnostic Interview for Genetic Studies (Nurnberger
et al., 1994) was used to evaluate participants' diagnostic status. Those with a history of
depression met DSM-5 criteria for lifetime MDD but had been free from major depressive
episodes or significant depressive symptoms for at least two months (American Psychiatric
Association, 2013). HCs did not have any current or past psychiatric disorders or first-degree
family members with known psychiatric disorders.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
23
Procedure
During the initial visit, participants gave informed consent, underwent eligibility
screening, and received training on the EMA protocol. They then participated in a seven-day
ambulatory assessment, which involved EMA and HR complexity measurements using a
biometric shirt. Throughout the week, participants were sent links to six daily surveys through
text messages (Twilio, Inc.) or email, depending on the participant's preference. We used
REDCap (Harris et al., 2009) for sending and collecting EMA survey data.
Two fixed schedules for EMA surveys were available depending on participants’ daily
routine: the "early" (8:00am-7:00pm) or "late" (10:00am-9:00pm) schedule. The specific survey
schedule was not disclosed to participants, except that they would receive surveys in the
morning, afternoon, and evening. Each day, participants received three pairs of "pre" and "post"
surveys. “Pre” surveys were sent semi-randomly during morning, afternoon, and evening time
slots, which were approximately four hours apart from each other. A “post” survey was then sent
30 minutes after completing the “pre” survey. Participants had one hour to complete each survey
and received reminders every 20 minutes until completion (up to two reminders). In the current
study, only affective items rated in the “post” surveys were included in the analyses, since
questions about regulation strategy use (more details below) were only asked in the “post’
surveys.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
24
Figure 1
EMA Schedule
Measures
Ecological Momentary Assessment (EMA)
Current Affect
Current affect was measured using a scale that assesses momentary positive (PA) and
negative affect (NA) (Hedeker et al., 2009). Participants rated their current PA and NA on ten
items (five positive, five negative), using 10-point Likert scales (1 = not at all; 10 = very much).
Summary scores of PA and NA were computed by summing the five PA items and NA items
separately.
Spontaneous Affect Regulation
Participants rated their spontaneous engagement in a range of emotion regulation
strategies and processes using questions adapted for EMA from a modified version of the
Spontaneous Affect Regulation Scale (Egloff et al., 2006; Stange et al., 2017a). Adaptive
strategies assessed included reappraisal (3 items), acceptance (2 items), and distraction (2 items).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
25
Maladaptive strategies included brooding (3 items) and mind wandering (3 items). Each item
was rated on a 10-point Likert scale (1 = not at all; 10 = very much). To operationalize strategy
use, we computed an average score for each strategy that were assessed with multiple items. We
also included an item that measured perceived regulation success: “Since the last survey, I was
successful at regulating my emotions” that was rated on the same scale. Please refer to Table 1
for a complete list of EMA variables examined in the current study.
Ambulatory Measurement of Physiological Data
We collected ambulatory electrocardiograms using Hexoskin smart shirts (Carré
Technologies, Inc.) that participants wore during the seven-day period. The shirts sample heart
beat signals at 256 Hz using three electrode sensors arranged in a triangular CC5 lead
configuration (two thoracic electrodes and one abdominal grounding electrode). Additionally,
they have a 3-axis accelerometer that samples physical activity at 64 Hz. Participants were
instructed to wear the Hexoskin shirt during their waking hours and to charge the device while
they slept.
We used the HRVanalysis software to analyze sequential R-R intervals extracted from
the Hexoskin shirts (Pichot et al., 2016). We removed visibly noisy or empty segments at the
beginning or end of the recording through visual inspection. The HRVanalysis software detected
and corrected anomalies caused by heart rhythm disturbances, such as ectopic beats, using the
algorithm developed by Kamath and Fallen (Kamath & Fallen, 1995). Erroneous beats were
identified using Cheung's algorithm, which sets a high and low threshold for relative variation in
R-R intervals (+32.5% and -24.5%, respectively) (Cheung, 1981). The software then estimates
the number of missing beats by comparing the error period's total time duration with the duration
of the beat immediately preceding it. If fewer than three recalculated beats are detected, cubic
spline interpolation is used; if four or more consecutive errors occur, missing beats are
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
26
interpolated by inserting the same number of previous R-R intervals between the first and last
valid intervals.
Subsequently, data were analyzed in batches of 5-minute segments to compute HR
complexity metrics. Any 5-minute segments with excessive noise or missing data were excluded
from further analysis based on the following criteria: (1) the proportion of artifacts in the R-R
time series exceeded 10%; (2) over five seconds of data were absent from either the beginning or
the end of the 5-minute segment (since the software demanded a minimum of five minutes of
data, this led to analysis errors); or (3) more than one-third of the data were missing from the
middle of the 5-minute segment. Using these criteria, 8.6% of the 5-minute segments were
discarded due to excessive noise or missing data; additionally, 2.9% of the segments were
removed after an earlier visual inspection of the R-R intervals revealed messy or unreliable
signals, possibly caused by improper shirt wearing (e.g., if the participant wore the shirt without
applying conductance lotion). In total, 88.5% of the available 5-minute segments were preserved
for further analysis. All participants contributed enough usable data for inclusion in the analyses.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
27
Table 1
List of Psychological Outcomes Measured via EMA
Psychological Domain Index
Affect NA
PA
Affect regulation Reappraisal
Acceptance
Distraction
Brooding
Mind-wandering
Sum of adaptive strategies
Sum of maladaptive strategies
Perceived regulation success
Impulsivity Momentary impulsivity
Cardiac Indices
Variability
Metrics of HRV differ in the timescales of variability that they capture. Short-term HRV
is commonly indexed as the root mean square successive difference (RMSSD) of R-R intervals
and the high-frequency (HF) band of HRV. As the primary time-domain measure of HRV,
RMSSD quantifies the difference between contiguous heartbeat intervals. In contrast, power
spectral analysis of heartbeat intervals reveals an HF component of HRV representing fast
oscillations in heart rate (0.15-0.4 Hz). Both measures are highly correlated with each other and
capture vagally-mediated HRV that operates in very short time scales (Shaffer & Ginsberg,
2017). Further, they both show reliable negative associations with aversive physical and mental
health outcomes, such as depressive symptoms and stress (Koch et al., 2019; Schiweck et al.,
2019).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
28
In addition, slower fluctuations in R-R intervals can be indexed by the low-frequency
(LF) band of HRV (0.04-0.15 Hz) (Shaffer & Ginsberg, 2017). The LF power is primarily an
indicator of the baroreflex activity, but its relationships with the SNS and the PNS have been
controversial. Instead of treating it as an index of the SNS activity, researchers have now started
to agree that the LF power more accurately reflects both sympathetic and parasympathetic
influences (Reyes del Paso et al., 2013). Studies have similarly revealed associations between
reduced LF HRV and depression (Dell’Acqua et al., 2020; Koch et al., 2019).
An index for longer-term HRV is the SD2 value from a Poincaré plot, which visualizes
successive differences between consecutive R-R intervals by plotting each R-R interval at time t
against its counterpart at time t+1 (Hoshi et al., 2013). By fitting an ellipse to the data points, we
can visually represent the variability of successive heartbeat intervals at short and long time
scales. In particular, the dispersion of the data points along the minor axis is termed SD1 and
reflects short-term variability just like RMSSD, while the spread of data points along the major
axis is termed SD2 and indexes long-term variability in successive heartbeat intervals. Since SD1
is statistically identical to RMSSD, we only included SD2 as an indicator of the physiological
states (Shaffer & Ginsberg, 2017). Studies suggest that SD2 is likely influenced by both the
parasympathetic and sympathetic nervous systems, and, similar to other HR complexity metrics,
is reduced during moments of stress and among individuals with depression (Brugnera et al.,
2019; Byun et al., 2019; De Vito et al., 2002; Tulppo et al., 1996).
Unpredictability
A commonly used entropy-based metric for quantifying the unpredictability of HR
signals is sample entropy (SampEn), which is defined as the negative logarithm of the
conditional probability that within a time series, segments that match for m timesteps will
continue to match for one additional timestep (Richman & Moorman, 2000). If many pairs of
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
29
similar sequences continue to be similar at the next timepoint (i.e., low SampEn), this means that
whether two sequences will match with each other or not can be reliably predicted from the first
few data points; in other words, the last timepoints of the two sequences are redundant and do
not contribute to prediction of similarity, thus indicating low information content. In
HRVAnalysis, a SampEn value is calculated over a five-minute segment of R-R intervals, which
stands in contrast to RMSSD and HF-HRV that are designed to capture rapid fluctuations of
heart beat intervals at very fine time scales. Research supports the clinical utility of SampEn as
an index of HR complexity, as low SampEn of cardiac signals has been associated with negative
affect (e.g., fear) and emotion dysregulation (Berry et al., 2019; Bornas et al., 2006; Fiskum et
al., 2017; Fiskum et al., 2018).
Self-Similarity
The Detrended Fluctuation Analysis (DFA) has been proposed to quantify scale
invariance of HR time series (Peng et al., 1995). The DFA involves removing the trend from a
time series and calculating the room-mean-square deviation within windows of increasing sizes.
The deviation is then plotted against window sizes on a log-log scale, and the slope (ɑ) of the
regression line provides a measure of the long-term correlation in the time series. In particular,
the short-term scaling exponent ɑ1 measures the degree of self-similarity over short time scales,
whereas the long-term scaling exponent ɑ2 is calculated over longer time segments. In
HRVanalysis, the time scales used for computing ɑ1 are between 4 and 11 heart beats, and those
for ɑ2 are between 12 and 64 heart beats. Values of ɑ1 and ɑ2 that are closer to 1 reflect greater
self-similarity of the time series, whereas an ɑ of 0.5 and 1.5 is characteristic of white noise and
Brownian motions respectively and indicate no self-similarity (Peng et al., 2000). In our data,
values of ɑ1 and ɑ2 range from 1 to 1.5, meaning that lower values of ɑ1 and ɑ2 reflect more self-
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
30
similarity and thus more HR complexity, which is hypothesized to support greater regulation
flexibility.
Finally, in addition to the seven HR complexity metrics outlined above, we also included
heart rate into the physiological states, as prior research has found increased heart rate to be
associated with adverse psychological outcomes such as NA (Carels et al., 2000; Simon et al.,
2021).
Respiration
We collected two respiratory measures and aggregated them over five-minute intervals.
The first measure is tidal volume, which measures the volume of air breathed in during a
respiration cycle. Second, breathing rate is defined as the number of breaths per minute. Because
we batch-analyzed the physiological time series in five-minute epochs, the respiratory variables
that were originally defined over shorter timescales (i.e., per minute and per respiration cycle)
were averaged over time to create summary statistics for five-minute segments of HR intervals
(Smith et al., 2019).
Physical Activity
The Hexoskin shirts contain 3-axis accelerometers for measuring momentary
acceleration. Activity was calculated per second as the magnitude of the displacement vector
(i.e., square root of the sum of squares of the three acceleration axes), and averaged across each
five-minute epoch (Villar et al., 2015). Please refer to Table 2 for a complete list of physiological
indicators.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
31
Table 2
List of Physiological State Constituents
Physiological Domain Physiological Sub-Domain Index
Heart Rate Heart rate
Variability RMSSD
HF power
LF power
SD2
Unpredictability Sample entropy
Self-similarity Alpha 1
Alpha 2
Respiration Volume Tidal volume
Rate Breathing rate
Physical Activity - Displacement
Chapter Three: Data Analysis
The current study was data-driven in nature due to a lack of literature on ambulatory
physiological states. To balance exploration and statistical rigor, we pre-registered three analytic
approaches that all fulfilled the same aims of state extraction, state alignment and multilevel
prediction of psychological outcomes: https://osf.io/a74wr. To narrow the scope of the current
manuscript, we focused on one of the three approaches here. Interested readers can refer to the
pre-registration for detail information about all three statistical approaches.
Idiographic State Extraction
In studies of dynamic functional connectivity that examined dynamic transitions between
brain states, hidden Markov models (HMM) were commonly used for state extraction, which we
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
32
also employed to uncover multivariate physiological states from idiographic time series (Eavani
et al., 2013; Javaheripour et al., 2023). The HMM assumes a dual stochastic process: a latent
state process governed by a first-order Markov chain, such that the probability of transitioning to
a state ss+1 at the next time point solely depends on the current state st; and an observed statedependent process that indicates the latent state process. The model also assumes conditional
independence between observations, such that an observation at time t is independent of any past
observations given the latent state at time t (McClintock et al., 2020). In the current study, the
latent state process in an HMM would correspond to the sequence of hidden physiological states
that people experienced in daily life, which was indicated by the physiological time series
modeled as an observed state-dependent process in an HMM. We fitted HMMs using the Python
package hmmlearn.
Traditionally, most studies that used HMMs for latent state extraction would fit a single HMM to
all participants’ data (e.g., Javaheripour et al., 2023). While computationally efficient and helpful
for uncovering group-level states, a group-level HMM may not be sensitive to individual
differences in the patterns or expression of states. For instance, some individuals may experience
more idiosyncratic and rarer states that could be overshadowed by more frequently displayed
states at the group level but may nevertheless be of clinical interest. Therefore, to maximally
preserve idiosyncrasies in state patterns and expression, we opted for an idiographic modelfitting procedure and trained HMMs person by person. To facilitate model selection, we fitted
HMMs with 2 to 20 states to each individual, and determined the optimal number of states
following certain criteria (see Model Selection & Between-Person State Alignment section
below).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
33
Model Selection & Between-Person State Alignment
The complications of fitting HMMs idiographically are: 1) if model selection is also done
idiographically (i.e., determining the best-fitting HMM person by person), individuals could end
up with different numbers of states; 2) state labels may carry different meanings across
participants. For example, the first state in Participant 1’s best-fitting HMM may reflect a very
different pattern of physiological features than Participant 2’s first state. Here, we described one
of the three ways we proposed to conduct model selection and align states across individuals.
Idiographic Model Selection + Theory-Based State Alignment
We used existing literature on physiological markers of affect and affect regulation to
inform state selection. As noted earlier, we hypothesized that there might be four physiological
states that participants experience in daily life:
a) An “average” state, characterized by average levels of all physiological variables;
b) A “stressed” state, characterized by reduced linear and nonlinear HRV, average or
shallow and fast breathing, and average or low physical activity;
c) A “relaxed” state, characterized by elevated linear and nonlinear HRV, average or deep
and slow breathing, and average or low physical activity;
d) An “active” state, characterized by reduced linear and nonlinear HRV, average or deep
and fast breathing, and elevated physical activity;
Hence, after performing idiographic model selection based on model fit metrics (i.e.,
AIC, BIC, cross-validated likelihood), we selected the four states that best matched our
hypothesized states within each individual’s best-fitting HMM.
Finally, the quality of alignment between each person’s idiographic states and the four
hypothesized states could differ between participants. Therefore, we computed an alignment
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
34
quality metric for each person that helped us later account for the effects of alignment quality on
the relationships between state dynamic statistics and EMA variables. For the approach of
idiographic model selection + theory-based state alignment, alignment quality was defined as the
Euclidean distance between a state and its corresponding template state (e.g., the Euclidean
distance between person A’s “stressed” state and our hypothesized “stressed” state).
Computation of State Dynamic Statistics
After model selection and state alignment, we applied the best-fitting HMM to each
individual’s physiological time series to decode state membership using the Viterbi algorithm,
which finds the most probable state sequence given the estimated emission probabilities (Forney,
1978). Then, we recoded the first-level states into second-level states determined through
alignment. After that, we partitioned each individual’s physiological time series by EMA
surveys, and calculated summary statistics of state dynamics during the segment preceding each
EMA survey. Specifically, we computed: 1) frequency, calculated by dividing the number of
occurrences of each state by the length of the segment; 2) dwell time, defined as the number of
consecutive time points spanned by a state normalized by the segment length; and 3) transition
probability, defined as the number of a transition (e.g., state 1 to state 2) divided by the segment
length. The rationale for normalizing the statistics by segment length is that the intervals between
EMA surveys were of variable lengths, which would affect the raw counts of states and
transitions. Frequency and dwell time were calculated for each state, and transition probability
was computed for each directed state-to-state transition (e.g., stressed-to-relaxed, relaxed-tostressed). Importantly, it was possible to have no physiological recording between a pair of EMA
surveys (e.g., the participant did not wear the Hexoskin shirt between two EMA surveys). In this
case, no state dynamic statistics were computed between this pair of EMA surveys.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
35
Multilevel Prediction of Affective and Cognitive Outcomes
To predict psychological outcomes both at the within-person and the between-person
levels, we fitted multilevel models with state dynamic variables (frequencies, dwell times, and
transition probabilities), depression history (HC vs. rMDD) and alignment quality of the
physiological states as predictors, and EMA variables (affect, affect regulation, and momentary
impulsivity) as outcomes. For each state dynamic variable of each state (e.g., frequency of the
stressed state), we fitted a multilevel model to predict each EMA variable. This resulted in
[n_states + n_states + n_states * (n_states - 1)] * 12 models for each approach, and hence
(4+4+4*3)*12 = 240 models for the theory-based state alignment approach. Further, we
decomposed each state dynamic variable into its within-person and between-person effects by
person-centering, so that the person-centered variable represented fluctuations around the person
mean while the person mean of the variable represented its between-person effects. Moreover,
we interacted depression history and alignment quality separately with the person-centered state
dynamic variable to examine the moderating effects of depression history and statistically control
for individual differences in state alignment quality. The rationale for not including a three-way
interaction between a state dynamics variable, depression history and alignment quality was
concerns about a lack of power, as both moderators are Level 2 variables while our Level 2
sample size was 93 participants. With two two-way interactions (State x Depression History,
State x Alignment Quality) and follow-up simple slope analyses, we could test 1) the effects of a
state dynamic variable on an EMA variable in the HC group and in the rMDD group, while
holding alignment quality at average (i.e., reference level of alignment quality); 2) the effects of
a state dynamic variable on an EMA variable at high, average and low levels of alignment
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
36
quality in the HC group (i.e., reference level of depression history). See Table 3 for a summary
of the state-EMA slopes tested at different levels of the two moderators.
Table 3
State-EMA Slopes Tested in the Two-Way Interactions
HC rMDD
AQ = Low State x AQ -
AQ = Average State x AQ
State x Depression History
State x Depression History
AQ = High State x AQ -
Note. AQ = alignment quality. State x AQ = The interaction between a state dynamic variable
and alignment quality. State x Depression History = The interaction between a state dynamic
variable and depression history.
Because all EMA variables except PA were heavily right-skewed distributed, we fitted
Linear Mixed-Effects Models (LMMs) with Gaussian distribution to PA, and Generalized Linear
Mixed-Effects Models (GLMMs) to all other EMA variables using the lme4 package (Bates et
al., 2015). Furthermore, variables pertaining to adaptive regulation strategies (reappraisal,
acceptance, distraction, and the sum of adaptive regulation strategies) and perceived regulation
success, which were zero-inflated, were predicted using hurdle models implemented in the
glmmTMB package (Brooks et al., 2017). This was because adaptive regulation strategies and
perceived regulation success were not rated on an entirely ordinal scale: an “N/A'' rating, which
would correspond to a zero, meant that the participant did not need to regulate their affect, while
a rating between 1 and 10 indicated the participant’s engagement in the adaptive ER
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
37
strategy/perceived regulation success given a regulatory need. Therefore, while scores from 1 to
10 on an adaptive ER variable/perceived regulation success reflected increasing endorsement of
the item as was the case for other EMA variables, a score of zero would be considered more
favorable than lower non-zero values (e.g., 1, which would indicate low adaptive ER
engagement despite regulatory need). A hurdle model contains two separate components: a zeroinflation model predicting the probability of zeros in the response variable, and a conditional
model predicting the non-zero values in the response variable. Thus, hurdle models allowed us to
predict the occurrence of zeros and non-zero values separately for adaptive regulation strategies
and perceived regulation success.
Among the EMA variables fitted with GLMMs, NA and momentary impulsivity, which
were positive continuous variables, were modeled with the Gamma distribution and the inverse
link function to accommodate their right-skewed distribution, and were given as follows:
While maladaptive regulation strategies (brooding, mind-wandering and the sum of
maladaptive regulation strategies) were initially modeled with Poisson distribution (Gamma
distribution could not be used due to the presence of zeros in maladaptive ER variables),
subsequent model diagnostics revealed overdispersion in the response variables, such that true
variances of the response variables vastly exceeded the model-estimated variances. Therefore,
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
38
we re-fitted GLMMs with the negative binomial distribution with quadratic parametrization and
the log link function to the maladaptive ER variables to adjust for overdispersion (Hardin &
Hilbe, 2007):
Among the EMA variables fitted with hurdle models (i.e., adaptive ER strategies and
perceived regulation success), the Poisson distribution (truncated at zero) with the log link
function was used in the conditional models to model distributions of the non-zero values, while
the zero-inflation models were binary logistic regression models predicting the probability of
zeros (Feng, 2021). Thus, values of adaptive ER strategies and perceived regulation success were
given by the hurdle Poisson models as:
Finally, after all GLMMs and hurdle models were fitted, we applied the BenjaminiHochberg procedure separately over the GLMMs and over the hurdle models to control the false
discovery rate at q = .05 (Benjamini & Hochberg, 1995). If a model contained a significant
interaction (e.g., State x Depression History), we conducted follow-up analyses of the simple
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
39
slopes to examine the marginal effect of the predictor on the outcome at each level of the
moderator. See Table 4 for a list of the GLM families used to model the EMA variables.
Table 4
GLM Families for the EMA Outcome Variables
(Truncated) Poisson Gamma Negative Binomial Gaussian
GLMM 1. NA
2. Impulsivity
1. Sum of
maladaptive ER
2. Brooding
3. Mind-wandering
1. PA
Hurdle –
Conditional
Models
1. Sum of adaptive ER
2. Reappraisal
3. Acceptance
4. Distraction
5. Perceived regulation
success
Chapter Four: Results
Idiographic State Discovery & Theory-Based State Alignment
Idiographic HMMs revealed 6-13 physiological states per person (mean = 10.1, SD =
1.71). Individuals with rMDD did not differ from HCs in the number of idiographic states (HC
vs. rMDD: t = -0.952, p = 0.344). The average deviations of the four types of states from the
templates, as measured by Euclidean distance, were as follows: stressed states (mean = 1.13, SD
= 0.40), relaxed states (mean = 1.28, SD = 0.46), active states (mean = 1.43, SD = 0.52), and
average states (mean = 1.15, SD = 0.33). Average alignment quality differed significantly across
states (f = 9.04, p < 0.0001). Post-hoc analysis revealed that stressed states and average states
were aligned equally well (t = -0.266, p = 0.791) and were both better aligned than relaxed states
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
40
(stressed vs. relaxed: t = -2.62, p = 0.0104; average vs. relaxed: t = -2.52, p = 0.0136), and that
relaxed states were better aligned than active states (relaxed vs. active: t = -2.10, p = 0.0384).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
41
Figure 2
Stressed States Across all Participants
Note. The template for stressed states was as follows: [0.5, -1, -1, 0.5, -1, 0.5, -0.5, -1, -0.5, -0.5,
-1].
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
42
Figure 3
Relaxed States Across All Participants
Note. The template for relaxed states was as follows: [-1, 2, 2, 0, 1, -1, -1, 0, 0, 0, -0.5].
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
43
Figure 4
Active States Across all Participants
Note. The template for active states was as follows: [2, -2, -2, -1, -1, 1, 1, -1, 1, 1, 2].
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
44
Figure 5
Average States Across all Participants
Note. The template for average states was as follows: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0].
Multilevel Prediction of Psychological Outcomes
Here we presented results from the theory-based state alignment approach. Due to the
large number of models, exact coefficient estimates and inferential statistics were presented in
tables and omitted from the text for simplicity (Stressed states: Table 5. Relaxed states: Table 6.
Active states: Table 7. Average states: Table 8).
Stressed States
Transitions into Stressed States. At average alignment quality and for both HC and
rMDD, transitioning from relaxed states to stressed states predicted higher levels of NA.
There was a significant interaction between depression history and inward transitions of
stressed states predicting mind-wandering, brooding and momentary impulsivity, holding
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
45
alignment quality at the mean. For individuals with rMDD, relaxed-to-stressed transitions
negatively predicted mind-wandering and active-to-stressed transitions negatively predicted
brooding, while HCs did not show such associations. For HCs, active-to-stressed transitions
positively correlated with momentary impulsivity, while these effects were non-significant
among individuals with rMDD. Finally, HCs and individuals with rMDD showed opposite
associations between inward transitions of stressed states and brooding: average-to-stressed
transitions negatively predicted brooding among HCs but positively correlated with brooding
among individuals with rMDD.
There was a significant interaction between alignment quality and inward transitions of
stressed states predicting mind-wandering and brooding, holding depression history at its
reference level (i.e., HCs). For HCs with high alignment quality, inward transitions of stressed
states were significantly associated with mind-wandering and brooding. For HCs with wellaligned active states, active-to-stressed transitions positively predicted brooding, while averageto-stressed transitions negatively correlated with brooding and mind-wandering.
Finally, inward transitions of stressed states were significantly associated with brooding
at the between-person level. In particular, individuals who showed more active-to-stressed
transitions on average showed more engagement in brooding.
Transitions out of Stressed States. At average alignment quality and for both HCs and
individuals with rMDD, stressed-to-active transitions were negatively associated with brooding.
There was a significant interaction between depression history and outward transitions of
stressed states predicting brooding, holding alignment quality at the mean. However, simple
slope analyses revealed that brooding was not significantly correlated with stressed-to-relaxed
transitions within either group.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
46
There was a significant interaction between alignment quality and outward transitions of
stressed states predicting brooding. Among HCs with well-aligned stressed states, stressed-torelaxed transitions negatively predicted brooding.
At the between-person level, outward transitions of stressed states were significantly
related to brooding. Individuals who showed more stressed-to-relaxed transitions displayed less
brooding on average, while those with more stressed-to-active transitions showed more brooding
on average.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
47
Table 5
Statistically Significant Results for Stressed States
Focal Predictor Outcome Results
Relaxed-Stressed NA Relaxed-Stressed_pc (for HC with mean align_qual): b =
8.26, SE = 2.77, t = 2.98, padj = 0.0461
Relaxed-Stressed_pc (for rMDD with mean align_qual): b =
13.0, SE = 4.85, t = 2.68, punadj = 0.00729 (sig.)
rMDD: b = -0.182, SE = 0.0465, t = -3.93, padj = 0.00539
Mind-wandering Relaxed-Stressed_pc * rMDD: b = -1.13, padj < 0.0001
Relaxed-Stressed_pc (for HC with average align_qual): b = -
1.70, SE = 0.654, z = -2.60, padj = n.s.
Relaxed-Stressed_pc (for rMDD with average align_qual): b
= -1.32, SE = 0.709, z = -18.361, punadj < 0.0001 (sig.)
rMDD: β = 4.01, padj = 0.00115
Relaxed-Stressed_pm: b = 3.43, padj < 0.0001
Relaxed-Stressed_pc * align_qual_Relaxed: b = 1.92, SE =
0.679, z = 2.82, padj = 0.0312
Relaxed-Stressed_pc (for align_qual_Relaxed = +1SD): b =
0.191, SE = 0.662, z = 0.288, punadj = n.s.
Relaxed-Stressed_pc (for align_qual_Relaxed = mean): b = -
1.17, SE = 0.295, z = -5.80, punadj < 0.0001 (sig.)
Relaxed-Stressed_pc (for align_qual_Relaxed = -1SD): b = -
3.63, SE = 0.276, z = -13.1, punadj < 0.0001 (sig.)
adaptive.ER.sum Relaxed-Stressed_pc * align_qual_Relaxed: b = 12.195, SE
= 4.30, z = 2.84, padj = 0.0421
Relaxed-Stressed_pc (align_qual_Relaxed = +1SD): b =
7.09, SE = 11.1, z = 0.636, punadj = n.s.
Relaxed-Stressed_pc (align_qual_Relaxed = mean): b = -
3.23, SE = 9.53, z = -0.340, punadj = n.s.
Relaxed-Stressed_pc (align_qual_Relaxed = -1SD): b = -
13.6, SE = 10.2, z = -1.33, punadj = n.s.
Active-Stressed Brooding Active-Stressed_pc (for HC with average align_qual): b = -
3.09, SE = 1.77, z = -1.74, padj = n.s.
Active-Stressed_pc (for rMDD with average align_qual): b =
-4.73, SE = 0.774, z = -6.11, punadj < 0.0001 (sig.)
Active-Stressed_pm: b = 5.29, SE = 1.11, z = -5.35, padj <
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
48
0.0001
Active-Stressed_pc * align_qual_Active: b = 5.29, SE =
0.638, z = 8.29, padj < 0.0001
Active-Stressed_pc (for align_qual_Active = +1SD): b =
2.21, SE = 0.578, z = 3.83, punadj = 0.000130 (sig.)
Active-Stressed_pc (for align_qual_Active = mean): b = -
3.08, SE = 9.788, z = -0.315, punadj = n.s.
Active-Stressed_pc (for align_qual_Active = -1SD): b = -
8.38, SE = 0.596, z = -14.1, punadj < 0.0001 (sig.)
Impulsivity Active-Stressed_pc (for HC with average align_qual): b =
2.00, SE = 0.655, t = 3.05, padj = 0.0159
Active-Stressed_pc (for rMDD with average align_qual): b =
0.417, SE = 0.451, t = 0.926, punadj = n.s.
Average-Stressed Brooding Average-Stressed_pc * rMDD: b = 4.42, SE = 0.967, z =
4.58, padj = 0.00014
Average-Stressed_pc (for HC with mean align_qual): b = -
3.09, SE = 0.602, z = -3.17, padj < 0.0001
Average-Stressed_pc (for rMDD with mean align_qual): b =
2.51, SE = 0.875, z = 2.87, punadj = 0.00405 (sig.)
Average-Stressed_pc * align_qual_Stressed: b = -1.16, SE =
0.563, z = -2.78, padj = 0.0350
Average-Stressed_pc (for align_qual_Stressed = +1SD): b =
-3.48, SE = 0.528, z = -6.59, punadj < 0.0001 (sig.)
Average-Stressed_pc (for align_qual_Stressed = mean): b = -
1.91, SE = 2.20, z = -0.869, punadj = n.s.
Average-Stressed_pc (for align_qual_Stressed = -1SD): b = -
0.350, SE = 0.778, z = -0.449, punadj = n.s.
Mind-wandering Average-Stressed_pc * rMDD: b = 2.49, SE = 0.410, z =
6.08, padj < 0.0001
Average-Stressed_pc (for HC with average align_qual): b =
-1.22, SE = 0.297, z = -4.12, punadj < 0.0001 (sig.)
Average-Stressed_pc (for rMDD with average align_qual):
b = 1.27, SE = 0.339, z = 3.73, punadj = 0.00019 (sig.)
rMDD: b =0.381, SE = 0.0973, z = 3.92, padj = 0.00162
Average-Stressed_pm: b = 8.05, SE = 0.512, z = 15.7, padj <
0.0001
Stressed-Relaxed PA Stressed-Relaxed_pc * align_qual_Stressed: b = 19.1, SE =
6.90, t = 2.77, padj = 0.0353
Stressed-Relaxed_pc (for align_qual_Stressed = +1SD): b =
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
49
32.1, SE = 14.4, t = 2.23, punadj = 0.0259 (n.s.)
Stressed-Relaxed_pc (for align_qual_Stressed = mean): b =
13.2, SE = 9.98, t = 1.32, punadj = n.s.
Stressed-Relaxed_pc (for align_qual_Stressed = -1SD): b = -
5.76, SE = 9.40, t = -0.613, punadj = n.s.
rMDD: b = -1.33, SE = 3.57, t = -3.71, padj = 0.00200
Brooding Stressed-Relaxed_pc * rMDD: b = 13.3, SE = 1.02, z = 13.0,
padj < 0.0001
Stressed-Relaxed_pc (for HC): b = -12.1, SE = 6.88, z = -
1.76, punadj = n.s.
Stressed-Relaxed_pc (for rMDD): b = 1.16, SE = 0.611, z =
1.90, punadj = n.s.
Stressed-Relaxed_pm: b = -15.0, SE = 0.773, z = -19.4, padj
< 0.0001
Stressed-Relaxed_pc * align_qual_Stressed: b = -9.18, padj <
0.0001
Stressed-Relaxed_pc (for align_qual_Stressed = +1SD): b =
-21.3, SE = 0.620, z = -34.3, punadj < 0.0001 (sig.)
Stressed-Relaxed_pc (for align_qual_Stressed = mean): b = -
12.1, SE = 6.88, z = -1.76, punadj = n.s.
Stressed-Relaxed_pc (for align_qual_Stressed = -1SD): b = -
2.92, SE = 0.717, z = -4.07, punadj < 0.0001 (sig.)
Stressed-Active Brooding Stressed-Active_pc (for HC with average align_qual): b = -
2.05, SE = 0.559, z = -3.67, padj = 0.00216
Stressed-Active_pc (for rMDD with average align_qual): b =
-2.83, SE = 0.618, z = -4.57, punadj < 0.0001 (sig.)
Stressed-Active_pm: b = 10.0, SE = 1.20, z = 8.35, padj <
0.0001
Stressed-Active_pc * align_qual_Active: b = 2.95, SE =
0.641, z = 4.61, padj = 0.000127
Stressed-Active_pc (for align_qual_Active = +1SD): b =
0.886, SE = 2.26, z = 0.392, punadj = n.s.
Stressed-Active_pc (for align_qual_Active = mean): b = -
2.05, SE = 1.83, z = -1.12, punadj = n.s.
Stressed-Active_pc (for align_qual_Active = -1SD): b = -
4.99, SE = 0.729, z = -6.85, punadj < 0.0001 (sig.)
Reappraisal Stressed-Active_pc:align_qual_Active: b = 6.47, padj =
0.0436
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
50
Stressed-Active (for align_qual_Active = +1SD): b = 2.42,
SE = 2.77, z = 0.874, punadj = n.s.
Stressed-Active (for align_qual_Active = mean): b = -2.49
SE = 2.32, z = -1.07, punadj = n.s.
Stressed-Active (for align_qual_Active = -1SD): b = -7.39,
SE = 3.22, z = -2.29, punadj = 0.0219 (n.s.)
Note. The “_pc” suffix represents the person-centered version of the physiological variable (i.e.,
denoting within-person effects), while the “_pm” suffix refers to the person mean of the
physiological variable (i.e., denoting between-person effects). “align_qual” = alignment quality.
“*” = interaction. padj = FDR-adjusted p-value. punadj = unadjusted p-value. adaptive.ER.sum =
sum of all adaptive ER strategies.
Relaxed States
Frequency and Dwell Time of Relaxed States. At the within-person level, there were
significant interactions between the frequency and dwell time of relaxed states and depression
history when predicting PA and engagement in distraction. Although simple slope analyses did
not support statistical significance of the effects of frequency or dwell time of relaxed states on
the outcome variables within either population, the effects nevertheless pointed in opposite
directions for HCs and individuals with rMDD. For HCs, visiting relaxed states was marginally
associated with less use of distraction while dwelling in relaxed states was marginally related to
increased PA. For individuals with rMDD, however, frequency of relaxed states marginally
predicted more use of distraction while dwell time of relaxed states marginally predicted less PA.
At the between-person level, individuals who dwelled in relaxed states more showed greater
engagement in mind-wandering on average.
Transitions into Relaxed States. At average alignment quality and for both HCs and
individuals with rMDD, average-to-relaxed transitions were negatively correlated with brooding.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
51
There was a significant interaction between depression history and inward transitions of
relaxed states predicting mind-wandering, holding alignment quality at its mean. For HCs with
average alignment quality, active-to-relaxed transitions positively predicted mind-wandering,
while individuals with rMDD showed no such associations.
There was a significant interaction between alignment quality and inward transitions of
relaxed states predicting brooding, mind-wandering and PA, holding depression history at its
reference level (i.e., HCs). For HCs with high alignment quality, brooding was negatively
predicted by stressed-to-relaxed and average-to-relaxed transitions and positively predicted by
active-to-relaxed transitions. In addition, mind-wandering was positively predicted by active-torelaxed transitions. Although alignment quality also significantly moderated the relationships
between stressed-to-relaxed transitions and PA and brooding, the associations were not
significant at high levels of alignment quality.
At the between-person level, inward transitions of relaxed states were significantly
related to brooding, such that individuals who displayed more stressed-to-relaxed transitions and
fewer active-to-relaxed transitions showed less brooding on average.
Transitions out of Relaxed States. At average alignment quality and for both HCs and
individuals with rMDD, relaxed-to-stressed transitions predicted greater NA.
At average alignment quality, HCs and individuals with rMDD differed in the
relationships between outward transitions of relaxed states and mind-wandering. For individuals
with rMDD with average alignment quality, relaxed-to-stressed transitions negatively predicted
mind-wandering, while HCs showed no such associations. In addition, HCs and individuals with
rMDD displayed opposite relationships between relaxed-to-active transitions and mindwandering, such that relaxed-to-active transitions were positively correlated with mind-
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
52
wandering among HCs but negatively correlated with mind-wandering among individuals with
rMDD.
There was a significant interaction between alignment quality and outward transitions of
relaxed states predicting NA and mind-wandering, holding depression history at its reference
level (i.e., HCs). For HCs with well-aligned states, relaxed-to-active transitions were negatively
associated with NA and positively associated with mind-wandering.
At the between-person level, individuals with more relaxed-to-stressed transitions and
fewer relaxed-to-active transitions showed greater mind-wandering on average.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
53
Table 6
Statistically Significant Results for Relaxed States
Focal Predictor Outcome Results
freq.Relaxed Distraction freq.Relaxed_pc * rMDD: b = 1.41, SE = 0.492, z = 2.86,
padj = 0.0401
freq.Relaxed_pc (for HC with average align_qual): b = -
0.678, SE = 0.378, z = -1.79, punadj = n.s.
freq.Relaxed_pc (for rMDD with average align_qual): b =
0.393, SE = 0.213, z = 1.84, punadj = n.s.
dwell.Relaxed PA dwell.Relaxed_pc * rMDD: b = -2.64, SE = 0.874, t = -3.01,
padj = 0.0177
dwell.Relaxed_pc (for HC with average align_qual): b =
1.20, SE = 0.654, t = 1.84, punadj = n.s.
dwell.Relaxed_pc (for rMDD with average align_qual): b = -
1.43, SE = 0.584, t = -2.45, punadj = 0.0144 (n.s.)
rMDD: b = -1.36, SE = 0.354, t = -3.84, padj = 0.00165
Mind-wandering dwell.Relaxed_pm: b = 1.45, SE = 0.382, z = 3.80, padj =
0.00181
rMDD: b = 0.399, SE = 0.101, z = 3.96, padj = 0.00162
Stressed-Relaxed PA Stressed-Relaxed_pc * align_qual_Stressed: b = 19.1, SE =
6.90, t = 2.77, padj = 0.0353
Stressed-Relaxed_pc (for align_qual_Stressed = +1SD): b =
32.1, SE = 14.4, t = 2.23, punadj = 0.0259 (n.s.)
Stressed-Relaxed_pc (for align_qual_Stressed = mean): b =
13.2, SE = 9.98, t = 1.32, punadj = n.s.
Stressed-Relaxed_pc (for align_qual_Stressed = -1SD): b = -
5.76, SE = 9.40, t = -0.613, punadj = n.s.
rMDD: b = -1.33, SE = 3.57, t = -3.71, padj = 0.00200
Brooding Stressed-Relaxed_pc * rMDD: b = 13.3, SE = 1.02, z = 13.0,
padj < 0.0001
Stressed-Relaxed_pc (for HC): b = -12.1, SE = 6.88, z = -
1.76, punadj = n.s.
Stressed-Relaxed_pc (for rMDD): b = 1.16, SE = 0.611, z =
1.90, punadj = n.s.
Stressed-Relaxed_pm: b = -15.0, SE = 0.773, z = -19.4, padj
< 0.0001
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
54
Stressed-Relaxed_pc * align_qual_Stressed: b = -9.18, padj <
0.0001
Stressed-Relaxed_pc (for align_qual_Stressed = +1SD): b =
-21.3, SE = 0.620, z = -34.3, punadj < 0.0001 (sig.)
Stressed-Relaxed_pc (for align_qual_Stressed = mean): b = -
12.1, SE = 6.88, z = -1.76, punadj = n.s.
Stressed-Relaxed_pc (for align_qual_Stressed = -1SD): b = -
2.92, SE = 0.717, z = -4.07, punadj < 0.0001 (sig.)
Active-Relaxed Brooding Active-Relaxed_pc * align_qual_Relaxed: b = 2.59, SE =
0.761, z = 3.41, padj = 0.00485
Active-Relaxed_pc (for align_qual_Relaxed = +1SD): b =
4.50, SE = 7.91, z = 0.570, punadj = n.s.
Active-Relaxed_pc (for align_qual_Relaxed = mean): b =
1.91, SE = 6.04, z = 0.317, punadj = n.s.
Active-Relaxed_pc (for align_qual_Relaxed = -1SD): b =
0.676, SE = 0.502, z = -1.35, punadj = n.s.
Active-Relaxed_pc * align_qual_Active: b = 21.8, SE =
0.794, z = 27.4, padj < 0.0001
Active-Relaxed_pc (for align_qual_Active = +1SD): b =
23.7, SE = 5.86, z = 4.05, punadj = 0.00005 (sig.)
Active-Relaxed_pc (for align_qual_Active = mean): b =
1.91, SE = 6.04, z = 0.317, punadj = n.s.
Active-Relaxed_pc (for align_qual_Active = -1SD): b = -
19.8, SE = 1.21, z = -16.3, punadj < 0.0001 (sig.)
Active-Relaxed_pm: b = 4.13, SE = 1.22, z = 3.39, padj =
0.00504
Mind-wandering Active-Relaxed_pc (for HC with average align_qual): b =
4.39, SE = 0.503, z = 8.74, padj < 0.0001
Active-Relaxed_pc (for rMDD with average align_qual): b =
0.0167, SE = 0.381, z = 0.0437, punadj = n.s.
Active-Relaxed_pm: b = -18.2, SE = 0.466, z = -39.0, padj <
0.0001
Active-Relaxed_pc * align_qual_Relaxed: b = 1.10, SE =
0.393, z = 2.88, padj = 0.00282
Active-Relaxed_pc (for align_qual_Relaxed = +1SD): b =
5.50, SE = 0.409, z = 13.4, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Relaxed = mean): b =
4.39, SE = 0.356, z = 12.3, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Relaxed = -1SD): b =
3.29, SE = 4.86, z = 0.678, punadj = n.s.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
55
Active-Relaxed_pc * align_qual_Active: b = 4.30, SE =
0.526, z = 8.17, padj < 0.0001
Active-Relaxed_pc (for align_qual_Active = +1SD): b =
8.71, SE = 0.456, z = 19.1, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Active = mean): b =
4.39, SE = 0.356, z = 12.3, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Active = -1SD): b =
0.0381, SE = 0.322, z = 0.118, punadj = n.s.
rMDD: b = 0.401, SE = 0.0987, z = 4.06, padj < 0.0001
Average-Relaxed Brooding Average-Relaxed_pc (for HC with average align_qual): b = -
3.03, SE = 0.619, z = -4.89, padj < 0.0001
Average-Relaxed_pc (for rMDD with average align_qual): b
= -2.54, SE = 0.576, z = -4.40, punadj = 0.0001 (sig.)
Average-Relaxed_pc * align_qual_Average: b = -3.89, SE =
0.823, z = -4.73, padj < 0.0001
Average-Relaxed_pc (for align_qual_Average = +1SD): b =
-6.92, SE = 0.633, z = -10.9, punadj < 0.0001 (sig.)
Average-Relaxed_pc (for align_qual_Average = mean): b = -
3.02, SE = 0.699, z = -4.32, punadj = 0.00002 (sig.)
Average-Relaxed_pc (for align_qual_Average = -1SD): b =
0.877, SE = 0.827, z = 1.06, punadj = n.s.
Relaxed-Stressed NA Relaxed-Stressed_pc (for HC with mean align_qual): b =
8.26, SE = 2.77, t = 2.98, padj = 0.0461
Relaxed-Stressed_pc (for rMDD with mean align_qual): b =
13.0, SE = 4.85, t = 2.68, punadj = 0.00729 (sig.)
rMDD: b = -0.182, SE = 0.0465, t = -3.93, padj = 0.00539
mind-wandering Relaxed-Stressed_pc * rMDD: b = -1.13, padj < 0.0001
Relaxed-Stressed_pc (for HC with average align_qual): b = -
1.70, SE = 0.654, z = -2.60, padj = n.s.
Relaxed-Stressed_pc (for rMDD with average align_qual): b
= -1.32, SE = 0.709, z = -18.361, punadj < 0.0001 (sig.)
rMDD: β = 4.01, padj = 0.00115
Relaxed-Stressed_pm: b = 3.43, padj < 0.0001
Relaxed-Stressed_pc * align_qual_Relaxed: b = 1.92, SE =
0.679, z = 2.82, padj = 0.0312
Relaxed-Stressed_pc (for align_qual_Relaxed = +1SD): b =
0.191, SE = 0.662, z = 0.288, punadj = n.s.
Relaxed-Stressed_pc (for align_qual_Relaxed = mean): b = -
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
56
1.17, SE = 0.295, z = -5.80, punadj < 0.0001 (sig.)
Relaxed-Stressed_pc (for align_qual_Relaxed = -1SD): b = -
3.63, SE = 0.276, z = -13.1, punadj < 0.0001 (sig.)
adaptive.ER.sum Relaxed-Stressed_pc * align_qual_Relaxed: b = 12.195, SE
= 4.30, z = 2.84, padj = 0.0421
Relaxed-Stressed_pc (align_qual_Relaxed = +1SD): b =
7.09, SE = 11.1, z = 0.636, punadj = n.s.
Relaxed-Stressed_pc (align_qual_Relaxed = mean): b = -
3.23, SE = 9.53, z = -0.340, punadj = n.s.
Relaxed-Stressed_pc (align_qual_Relaxed = -1SD): b = -
13.6, SE = 10.2, z = -1.33, punadj = n.s.
Relaxed-Active NA Relaxed-Active_pc * align_qual_Relaxed: b = -2.06, SE =
0.712, t = -2.89, padj = 0.0256
Relaxed-Active_pc (for align_qual = +1SD): b = -3.40, SE =
1.05, t = -3.24, punadj = 0.00120 (sig.)
Relaxed-Active_pc (for align_qual = mean): b = -1.37, SE =
0.850, t = -1.61, punadj = n.s.
Relaxed-Active_pc (for align_qual = -1SD): b = 0.653, SE =
1.16, t = 0.561, punadj = n.s.
rMDD: b = -0.177, SE = 0.0460, t = -3.85, padj = 0.00165
Mind-wandering Relaxed-Active_pc * rMDD: b = -5.40, SE = 0.435, z = -
12.4, padj < 0.0001
Relaxed-Active_pc (for HC with average align_qual): b =
3.76, SE = 0.403, z = 9.32, padj < 0.0001
Relaxed-Active_pc (for rMDD with average align_qual): b =
-1.64, SE = 0.428, z = -3.83, punadj = 0.00013 (sig.)
Relaxed-Active_pm: b = -3.42, SE = 0.444, z = -7.70, padj <
0.0001
Relaxed-Active_pc * align_qual_Relaxed: b = 1.20, SE =
0.337, z = 3.57, padj = 0.00282
Relaxed-Active_pc (for align_qual_Relaxed = +1SD): b =
4.96, SE = 0.396, z = 12.5, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Relaxed = mean): b =
3.76, SE = 0.539, z = 6.97, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Relaxed = -1SD): b =
2.56, SE = 0.871, z = 2.93, punadj = 0.00334 (sig.)
Relaxed-Active_pc * align_qual_Active: b = 4.69, SE =
0.387, z = 12.1, padj < 0.0001
Relaxed-Active_pc (for align_qual_Active = +1SD): b =
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
57
8.46, SE = 0.335, z = 25.2, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Active = mean): b =
3.76, SE = 0.539, z = 6.97, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Active = -1SD): b = -
0.953, SE = 0.336, z = -2.84, punadj = 0.00458 (sig.)
rMDD: b = 0.396, SE = 0.0998, z = 3.97, padj = 0.00162
Acceptance Relaxed-Active_pc * align_qual_Active: b = 19.6, SE =
6.41, z = 3.06, padj = 0.0211
Relaxed-Active_pc (for align_qual_Active = +1SD): b =
13.7, SE = 7.73, z = 1.77, punadj = n.s.
Relaxed-Active_pc (for align_qual_Active = mean): b = -
2.08, SE = 5.08, z = -0.410, punadj = n.s.
Relaxed-Active_pc (for align_qual_Active = -1SD): b = -
17.8, SE = 7.22, z = -2.47, punadj = 0.0135 (n.s.)
Relaxed-Average Mind-wandering Relaxed-Average_pc * align_qual_Average: b = -1.22, SE =
0.444, z = -2.75, padj = 0.0369
Relaxed-Average_pc (for align_qual_Average = +1SD): b =
-1.46, SE = 3.35, z = -0.437, punadj = n.s.
Relaxed-Average_pc (for align_qual_Average = mean): b = -
0.247, SE = 0.575, z = -0.430, punadj = n.s
Relaxed-Average_pc (for align_qual_Average = -1SD): b =
0.983, SE = 0.271, z = 3.63, punadj = 0.00029 (sig.)
rMDD: b = 0.392, SE = 0.0964, z = 4.06, padj = 0.00124
Perceived
regulation
success
Relaxed-Average_pc * align_qual_Average: b= 5.49, SE =
1.91, z = 2.88, padj = 0.0378
Relaxed-Average_pc (for align_qual_Average = +1SD): b =
6.13, SE = 3.05, z = 2.01, punadj = 0.0444 (n.s.)
Relaxed-Average_pc (for align_qual_Average = mean): b =
1.25, SE = 2.46, z = 0.506, punadj = n.s.
Relaxed-Average_pc (for align_qual_Average = -1SD): b = -
3.65, SE = 2.94, z = -1.24, punadj = n.s.
rMDD: b = -0.291, SE = 0.0664, z = -4.38, padj = 0.000153
Note. The “_pc” suffix represents the person-centered version of the physiological variable (i.e.,
denoting within-person effects), while the “_pm” suffix refers to the person mean of the
physiological variable (i.e., denoting between-person effects). “align_qual” = alignment quality.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
58
“*” = interaction. padj = FDR-adjusted p-value. punadj = unadjusted p-value. adaptive.ER.sum =
sum of all adaptive ER strategies.
Active States
Frequency and Dwell Time of Active States. Depression history significantly
moderated the relationships between frequency of active states and NA and overall use of
adaptive ER strategies, holding alignment quality at the mean. In particular, frequency of visiting
active states negatively predicted NA among HCs but positively predicted NA among individuals
with rMDD at average levels of alignment quality. Frequency of visiting active states was also
negatively associated with overall engagement in adaptive regulation strategies among
individuals with rMDD but had no effect among HCs.
Additionally, there were significant group differences in the associations between dwell
time in active states and overall use of adaptive ER strategies at average levels of alignment
quality, such that dwelling in active states was negatively associated with overall engagement in
adaptive ER among individuals with rMDD but not among HCs.
Transitions into Active States. At average alignment quality and for both HCs and
individuals with rMDD, stressed-to-active transitions were negatively correlated with brooding.
Depression history significantly moderated the effects of inward transitions of active
states on mind-wandering and momentary impulsivity, after statistically controlling for
alignment quality. HCs and individuals with rMDD showed opposite predictions of mindwandering: 1) relaxed-to-active transitions were positively associated with mind-wandering for
HCs with average alignment quality, but were negatively correlated with mind-wandering for
individuals with rMDD; 2) average-to-active transitions negatively predicted mind-wandering
among HCs but positively predicted mind-wandering among individuals with rMDD at mean
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
59
levels of alignment quality. Furthermore, while average-to-active transitions were negatively
related to momentary impulsivity among HCs, individuals with rMDD did not show such
associations.
Alignment quality significantly moderated the relationships between inward transitions of
active states and NA and mind-wandering. For HCs with high alignment quality, relaxed-toactive transitions were negatively correlated with NA and positively correlated with mindwandering.
At the between-person level, inward transitions of active states were significantly related
to brooding and mind-wandering, such that individuals who displayed more stressed-to-active
transitions showed more brooding, and those with fewer relaxed-to-active transitions and more
average-to-active transitions showed more mind-wandering.
Transitions out of Active States. At average alignment quality and for both HCs and
individuals with rMDD, active-to-average transitions positively predicted brooding.
A significant interaction emerged between depression history and outward transitions of
active states in predicting brooding and momentary impulsivity after holding alignment quality at
its mean. For individuals with rMDD, active-to-stressed transitions were negatively correlated
with brooding, while HCs showed no such associations. In addition, active-to-stressed transitions
positively predicted momentary impulsivity, active-to-average transitions negatively predicted
momentary impulsivity, and active-to-relaxed transitions positively predicted mind-wandering
among HCs with average alignment quality, while these effects were not present among
individuals with rMDD.
Alignment quality significantly moderated the associations between outward transitions
of active states and brooding and mind-wandering when depression history was held at its
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
60
reference level (i.e., HCs). Among HCs with well-aligned active states, active-to-stressed
transitions and active-to-relaxed transitions positively predicted brooding, and active-to-relaxed
transitions also positively correlated with mind-wandering.
At the between-person level, outward transitions of active states were significantly
related to brooding and mind-wandering, such that individuals with more active-to-stressed and
active-to-relaxed transitions showed greater engagement in brooding. In addition, individuals
who displayed more active-to-relaxed transitions showed reduced mind-wandering.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
61
Table 7
Statistically Significant Results for Active States
Focal Predictor Outcome Results
freq.Active NA freq.Active_pc * rMDD: b = 0.269, SE = 0.0788, t = 3.42,
padj = 0.0124
freq.Active_pc (for HC with average align_qual): b = -0.130,
SE = 0.0585, t = -2.22, punadj = 0.0264 (n.s.)
freq.Active_pc (for rMDD with average align_qual): b =
0.139, SE = 0.0502, t = 2.78, punadj = 0.00541 (sig.)
rMDD: b = -0.179, SE = 0.0458, t = -3.90, padj = 0.00539
Brooding freq.Active_pc * rMDD: b = -0.858, SE = 0.286, t = -3.00,
padj = 0.0183
freq.Active_pc (for HC with average align_qual): b = 0.403,
SE = 0.204, z = 1.97, punadj = 0.0484 (n.s.)
freq.Active_pc (for rMDD with average align_qual): b = -
0.455, SE = 0.217, z = -2.10, punadj = 0.0361 (n.s.)
adaptive.ER.sum freq.Active_pc * rMDD: b = -0.663, SE = 0.213, z = -3.11,
padj = 0.0180
freq.Active_pc (for HC with average align_qual): b = 0.216,
SE = 0.150, z = 1.43, punadj = n.s.
freq.Active_pc (for rMDD with average align_qual): b = -
0.464, SE = 0.140, z = -3.31, punadj = 0.00095 (sig.)
dwell.Active Brooding dwell.Active_pc * rMDD: b = -1.13, SE = 0.402, z = -2.82,
padj = 0.0313
dwell.Active_pc (for HC with average align_qual): b =
0.628, SE = 0.250, z = 2.52, punadj = 0.0119 (n.s.)
dwell.Active_pc (for rMDD with average align_qual): b = -
0.504, SE = 0.283, z = -1.78, punadj = n.s.
adaptive.ER.sum dwell.Active_pc * rMDD: b = -0.999, SE = 0.313, z = -3.20,
padj = 0.0135
dwell.Active_pc (for HC with average align_qual): b =
0.327, SE = 0.211, z = 1.55, punadj = n.s.
dwell.Active_pc (for rMDD with average align_qual): b = -
0.718, SE = 0.208, z = -3.46, punadj = 0.00055 (sig.)
Distraction
(idx=67)
dwell.Active_pc * rMDD: b = -1.80, SE = 0.627, z = -2.87,
padj = 0.0388
dwell.Active_pc (for HC with average align_qual): b =
0.556, SE = 0.401, z = 1.39, punadj = n.s.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
62
dwell.Active_pc (for rMDD with average align_qual): b = -
0.941, SE = 0.384, z = -2.45, punadj = 0.0142 (n.s.)
Stressed-Active Brooding Stressed-Active_pc (for HC with average align_qual): b = -
2.05, SE = 0.559, z = -3.67, padj = 0.00216
Stressed-Active_pc (for rMDD with average align_qual): b =
-2.83, SE = 0.618, z = -4.57, punadj < 0.0001 (sig.)
Stressed-Active_pm: b = 10.0, SE = 1.20, z = 8.35, padj <
0.0001
Stressed-Active_pc * align_qual_Active: b = 2.95, SE =
0.641, z = 4.61, padj = 0.000127
Stressed-Active_pc (for align_qual_Active = +1SD): b =
0.886, SE = 2.26, z = 0.392, punadj = n.s.
Stressed-Active_pc (for align_qual_Active = mean): b = -
2.05, SE = 1.83, z = -1.12, punadj = n.s.
Stressed-Active_pc (for align_qual_Active = -1SD): b = -
4.99, SE = 0.729, z = -6.85, punadj < 0.0001 (sig.)
Reappraisal Stressed-Active_pc:align_qual_Active: b = 6.47, padj =
0.0436
Stressed-Active (for align_qual_Active = +1SD): b = 2.42,
SE = 2.77, z = 0.874, punadj = n.s.
Stressed-Active (for align_qual_Active = mean): b = -2.49
SE = 2.32, z = -1.07, punadj = n.s.
Stressed-Active (for align_qual_Active = -1SD): b = -7.39,
SE = 3.22, z = -2.29, punadj = 0.0219 (n.s.)
Relaxed-Active NA Relaxed-Active_pc * align_qual_Relaxed: b = -2.06, SE =
0.712, t = -2.89, padj = 0.0256
Relaxed-Active_pc (for align_qual = +1SD): b = -3.40, SE =
1.05, t = -3.24, punadj = 0.00120 (sig.)
Relaxed-Active_pc (for align_qual = mean): b = -1.37, SE =
0.850, t = -1.61, punadj = n.s.
Relaxed-Active_pc (for align_qual = -1SD): b = 0.653, SE =
1.16, t = 0.561, punadj = n.s.
rMDD: b = -0.177, SE = 0.0460, t = -3.85, padj = 0.00165
Mind-wandering Relaxed-Active_pc * rMDD: b = -5.40, SE = 0.435, z = -
12.4, padj < 0.0001
Relaxed-Active_pc (for HC with average align_qual): b =
3.76, SE = 0.403, z = 9.32, padj < 0.0001
Relaxed-Active_pc (for rMDD with average align_qual): b =
-1.64, SE = 0.428, z = -3.83, punadj = 0.00013 (sig.)
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
63
Relaxed-Active_pm: b = -3.42, SE = 0.444, z = -7.70, padj <
0.0001
Relaxed-Active_pc * ailgn_qual_Relaxed: b = 1.20, SE =
0.337, z = 3.57, padj = 0.00282
Relaxed-Active_pc (for align_qual_Relaxed = +1SD): b =
4.96, SE = 0.396, z = 12.5, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Relaxed = mean): b =
3.76, SE = 0.539, z = 6.97, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Relaxed = -1SD): b =
2.56, SE = 0.871, z = 2.93, punadj = 0.00334 (sig.)
Relaxed-Active_pc * ailgn_qual_Active: b = 4.69, SE =
0.387, z = 12.1, padj < 0.0001
Relaxed-Active_pc (for align_qual_Active = +1SD): b =
8.46, SE = 0.335, z = 25.2, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Active = mean): b =
3.76, SE = 0.539, z = 6.97, punadj < 0.0001 (sig.)
Relaxed-Active_pc (for align_qual_Active = -1SD): b = -
0.953, SE = 0.336, z = -2.84, punadj = 0.00458 (sig.)
rMDD: b = 0.396, SE = 0.0998, z = 3.97, padj = 0.00162
Acceptance Relaxed-Active_pc * align_qual_Active: b = 19.6, SE =
6.41, z = 3.06, padj = 0.0211
Relaxed-Active_pc (for align_qual_Active = +1SD): b =
13.7, SE = 7.73, z = 1.77, punadj = n.s.
Relaxed-Active_pc (for align_qual_Active = mean): b = -
2.08, SE = 5.08, z = -0.410, punadj = n.s.
Relaxed-Active_pc (for align_qual_Active = -1SD): b = -
17.8, SE = 7.22, z = -2.47, punadj = 0.0135 (n.s.)
Average-Active Mind-wandering Average-Active_pc (for HC with average align_qual): b = -
2.14, SE = 0.549, z = -3.90, padj = 0.00162
Average-Active_pc (for rMDD with average align_qual): b =
1.81, SE = 0.278, z = 6.52, punadj < 0.0001 (sig.)
Average-Active_pm: b = 18.9, SE = 0.591, z = 32.0, padj <
0.0001
Average-Active_pc * align_qual_Active: b = 1.83, SE =
0.539, z = 3.40, padj = 0.00485
Average-Active_pc (for align_qual_Active = +1SD): b = -
0.302. SE = 0.496, z = -0.610, punadj = n.s.
Average-Active_pc (for align_qual_Active = mean): b = -
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
64
2.14. SE = 2.63, z = -0.815, punadj = n.s.
Average-Active_pc (for align_qual_Active = -1SD): b = -
3.98. SE = 0.378, z = -10.5, punadj < 0.0001
Impulsivity Average-Active_pc (for HC with average align_qual): b = -
2.48, SE = 0.933, t = -2.66, padj = 0.0491
Average-Active_pc (for rMDD with average align_qual): b =
1.63, SE = 0.887, t =1.83, punadj = n.s.
Active-Stressed Brooding Active-Stressed_pc (for HC with average align_qual): b = -
3.09, SE = 1.77, z = -1.74, padj = n.s.
Active-Stressed_pc (for rMDD with average align_qual): b =
-4.73, SE = 0.774, z = -6.11, punadj < 0.0001 (sig.)
Active-Stressed_pm: b = 5.29, SE = 1.11, z = -5.35, padj <
0.0001
Active-Stressed_pc * align_qual_Active: b = 5.29, SE =
0.638, z = 8.29, padj < 0.0001
Active-Stressed_pc (for align_qual_Active = +1SD): b =
2.21, SE = 0.578, z = 3.83, punadj = 0.000130 (sig.)
Active-Stressed_pc (for align_qual_Active = mean): b = -
3.08, SE = 9.788, z = -0.315, punadj = n.s.
Active-Stressed_pc (for align_qual_Active = -1SD): b = -
8.38, SE = 0.596, z = -14.1, punadj < 0.0001 (sig.)
Impulsivity Active-Stressed_pc (for HC with average align_qual): b =
2.00, SE = 0.655, t = 3.05, padj = 0.0159
Active-Stressed_pc (for rMDD with average align_qual): b =
0.417, SE = 0.451, t = 0.926, punadj = n.s.
Active-Relaxed Brooding Active-Relaxed_pc * align_qual_Relaxed: b = 2.59, SE =
0.761, z = 3.41, padj = 0.00485
Active-Relaxed_pc (for align_qual_Relaxed = +1SD): b =
4.50, SE = 7.91, z = 0.570, punadj = n.s.
Active-Relaxed_pc (for align_qual_Relaxed = mean): b =
1.91, SE = 6.04, z = 0.317, punadj = n.s.
Active-Relaxed_pc (for align_qual_Relaxed = -1SD): b =
0.676, SE = 0.502, z = -1.35, punadj = n.s.
Active-Relaxed_pc * align_qual_Active: b = 21.8, SE =
0.794, z = 27.4, padj < 0.0001
Active-Relaxed_pc (for align_qual_Active = +1SD): b =
23.7, SE = 5.86, z = 4.05, punadj = 0.00005 (sig.)
Active-Relaxed_pc (for align_qual_Active = mean): b =
1.91, SE = 6.04, z = 0.317, punadj = n.s.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
65
Active-Relaxed_pc (for align_qual_Active = -1SD): b = -
19.8, SE = 1.21, z = -16.3, punadj < 0.0001 (sig.)
Active-Relaxed_pm: b = 4.13, SE = 1.22, z = 3.39, padj =
0.00504
Mind-wandering Active-Relaxed_pc (for HC with average align_qual): b =
4.39, SE = 0.503, z = 8.74, padj < 0.0001
Active-Relaxed_pc (for rMDD with average align_qual): b =
0.0167, SE = 0.381, z = 0.0437, punadj = n.s.
Active-Relaxed_pm: b = -18.2, SE = 0.466, z = -39.0, padj <
0.0001
Active-Relaxed_pc * align_qual_Relaxed: b = 1.10, SE =
0.393, z = 2.88, padj = 0.00282
Active-Relaxed_pc (for align_qual_Relaxed = +1SD): b =
5.50, SE = 0.409, z = 13.4, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Relaxed = mean): b =
4.39, SE = 0.356, z = 12.3, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Relaxed = -1SD): b =
3.29, SE = 4.86, z = 0.678, punadj = n.s.
Active-Relaxed_pc * align_qual_Active: b = 4.30, SE =
0.526, z = 8.17, padj < 0.0001
Active-Relaxed_pc (for align_qual_Active = +1SD): b =
8.71, SE = 0.456, z = 19.1, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Active = mean): b =
4.39, SE = 0.356, z = 12.3, punadj < 0.0001 (sig.)
Active-Relaxed_pc (for align_qual_Active = -1SD): b =
0.0381, SE = 0.322, z = 0.118, punadj = n.s.
rMDD: b = 0.401, SE = 0.0987, z = 4.06, padj < 0.0001
Active-Average Brooding Active-Average_pc (for HC with average align_qual): b =
2.35, SE = 0.745, z = 3.15, padj = 0.0114
Active-Average_pc (for rMDD) with average align_qual): b
= 2.05, SE = 0.568, z = 3.60, punadj = 0.00032 (sig.)
Active-Average_pm: b = 24.8, SE = 1.66, z = 14.9, padj <
0.0001
Impulsivity Active-Average_pc (for HC with average align_qual): b = -
2.91, SE = 0.748, t = -3.90, padj = 0.00162
Active-Average_pc (for rMDD with average align_qual): b =
0.0634, SE = 0.875, t =0.0724, punadj = n.s.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
66
Note. The “_pc” suffix represents the person-centered version of the physiological variable (i.e.,
denoting within-person effects), while the “_pm” suffix refers to the person mean of the
physiological variable (i.e., denoting between-person effects). “align_qual” = alignment quality.
“*” = interaction. padj = FDR-adjusted p-value. punadj = unadjusted p-value. adaptive.ER.sum =
sum of all adaptive ER strategies. freq.Active = frequency of active states. dwell.Active = dwell
time of active states.
Average States
Frequency and Dwell Time of Average States. Although alignment quality
significantly moderated the associations between frequency and dwell time of average states and
NA and PA, subsequent simple slope analyses did not reveal significant effects among
individuals with well-aligned average states.
Transitions into Average States. At average alignment quality, both HCs and
individuals with rMDD showed positive associations between active-to-average transitions and
brooding.
There was a significant interaction between depression history and active-to-average
transitions in predicting momentary impulsivity conditioned on average alignment quality. For
HCs with average alignment quality, active-to-average transitions were negatively associated
with momentary impulsivity, while individuals with rMDD did not show these associations.
Alignment quality significantly moderated the effects of relaxed-to-average transitions on
mind-wandering and perceived regulation success when depression history was held at its
reference level (i.e., HCs). However, subsequent simple slope analyses did not reveal significant
associations between relaxed-to-average transitions and either EMA variable among HCs with
high levels of alignment quality.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
67
At the between-person level, active-to-average transitions were positively associated with
brooding, such that individuals who made more active-to-average transitions over time showed
more brooding on average.
Transitions out of Average States. At average alignment quality, both HCs and
individuals with rMDD showed negative associations between average-to-relaxed transitions and
brooding.
Depression history significantly moderated the effects of outward transitions of average
states on brooding and mind-wandering when alignment quality was held at the mean. HCs and
individuals with rMDD showed opposite associations between average-to-stressed transitions
and brooding, mind-wandering and momentary impulsivity, such that brooding and mindwandering were negatively predicted by average-to-stress transitions among HCs but positively
predicted by average-to-stress transitions among individuals with rMDD. In addition, while
average-to-active transitions were negatively correlated with mind-wandering among HCs, they
were positively correlated with mind-wandering among individuals with rMDD at mean levels of
alignment quality. Moreover, average-to-active transitions negatively predicted momentary
impulsivity among HCs only and not among individuals with rMDD.
The effects of inward transitions of average states on brooding depended on alignment
quality when depression history was held at its reference level (i.e., HCs). For HCs with wellaligned states, average-to-stressed transitions negatively predicted brooding, and average-torelaxed transitions were negatively correlated with brooding.
At the between-person level, average-to-stressed transitions were positively correlated
with mind-wandering, such that individuals who experienced more average-to-stressed
transitions overall showed greater engagement in mind-wandering.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
68
Table 8
Statistically Significant Results for Average States
Focal Predictor Outcome Results
dwell.Average NA dwell.Average_pc * align_qual_Average: b = 0.217, SE =
0.0772, t = 2.81, padj = 0.0320
dwell.Average_pc (for align_qual = +1SD): b = 0.227, SE =
0.181, t = 1.25, punadj = n.s.
dwell.Average_pc (for align_qual = mean): b = 0.00997, SE
= 0.162, t = 0.0614, punadj = n.s.
dwell.Average_pc (for align_qual = -1SD): b = -0.207, SE =
0.178, t = -1.16, punadj = n.s.
rMDD: b = -0.172, SE = 0.0460, t = -3.74, padj = 0.00198
PA dwell.Average_pc * align_qual_Average: b = 1.78, SE =
0.543, t = 3.28, padj = 0.00749
dwell.Average_pc (for align_qual = +1SD): b = 0.922, SE =
0.935, t = 0.986, punadj = n.s.
dwell.Average_pc (for align_qual = mean): b = -0.857, SE =
0.807, t = -1.06, punadj = n.s.
dwell.Average_pc (for align_qual = -1SD): b = -2.63, SE =
1.01, t = -2.61, punadj = 0.00909 (n.s.)
rMDD: b = -1.35, SE = 0.351, t = -3.85, padj = 0.00165
freq.Average PA freq.Average_pc * align_qual_Average: b = 0.795, SE =
0.287, t = 2.77, padj = 0.0353
freq.Average_pc (for align_qual = +1SD): b = 0.0370, SE =
0.452, t = 0.0820, punadj = n.s.
freq.Average_pc (for align_qual = mean): b = -0.758, SE =
0.406, t = -1.87, punadj = n.s.
freq.Average_pc (for align_qual = -1SD): b = -1.55, SE =
0.539, t = -2.89, punadj = 0.00397 (sig.)
rMDD: b = -1.35, SE = 0.350, t = -3.85, padj = 0.00165
Relaxed-Average Mind-wandering Relaxed-Average_pc * align_qual_Average: b = -1.22, SE =
0.444, z = -2.75, padj = 0.0369
Relaxed-Average_pc (for align_qual_Average = +1SD): b =
-1.46, SE = 3.35, z = -0.437, punadj = n.s.
Relaxed-Average_pc (for align_qual_Average = mean): b = -
0.247, SE = 0.575, z = -0.430, punadj = n.s
Relaxed-Average_pc (for align_qual_Average = -1SD): b =
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
69
0.983, SE = 0.271, z = 3.63, punadj = 0.00029 (sig.)
rMDD: b = 0.392, SE = 0.0964, z = 4.06, padj = 0.00124
Perceived
regulation
success
Relaxed-Average_pc * align_qual_Average: b= 5.49, SE =
1.91, z = 2.88, padj = 0.0378
Relaxed-Average_pc (for align_qual_Average = +1SD): b =
6.13, SE = 3.05, z = 2.01, punadj = 0.0444 (n.s.)
Relaxed-Average_pc (for align_qual_Average = mean): b =
1.25, SE = 2.46, z = 0.506, punadj = n.s.
Relaxed-Average_pc (for align_qual_Average = -1SD): b = -
3.65, SE = 2.94, z = -1.24, punadj = n.s.
rMDD: b = -0.291, SE = 0.0664, z = -4.38, padj = 0.000153
Active-Average Brooding Active-Average_pc (for HC with average align_qual): b =
2.35, SE = 0.745, z = 3.15, padj = 0.0114
Active-Average_pc (for rMDD with average align_qual): b =
2.05, SE = 0.568, z = 3.60, punadj = 0.00032 (sig.)
Active-Average_pm: b = 24.8, SE = 1.66, z = 14.9, padj <
0.0001
Impulsivity Active-Average_pc (for HC with average align_qual): b = -
2.91, SE = 0.748, t = -3.90, padj = 0.00162
Active-Average_pc (for rMDD with average align_qual): b =
0.0634, SE = 0.875, t =0.0724, punadj = n.s.
Average-Stressed Brooding Average-Stressed_pc * rMDD: b = 4.42, SE = 0.967, z =
4.58, padj = 0.00014
Average-Stressed_pc (for HC with mean align_qual): b = -
3.09, SE = 0.602, z = -3.17, padj < 0.0001
Average-Stressed_pc (for rMDD with mean align_qual): b =
2.51, SE = 0.875, z = 2.87, punadj = 0.00405 (sig.)
Average-Stressed_pc * align_qual_Stressed: b = -1.16, SE =
0.563, z = -2.78, padj = 0.0350
Average-Stressed_pc (for align_qual_Stressed = +1SD): b =
-3.48, SE = 0.528, z = -6.59, punadj < 0.0001 (sig.)
Average-Stressed_pc (for align_qual_Stressed = mean): b = -
1.91, SE = 2.20, z = -0.869, punadj = n.s.
Average-Stressed_pc (for align_qual_Stressed = +1SD): b =
-0.350, SE = 0.778, z = -0.449, punadj = n.s.
Mind-wandering Average-Stressed_pc * rMDD: b = 2.49, SE = 0.410, z =
6.08, padj < 0.0001
Average-Stressed_pc (for HC with average align_qual): b =
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
70
-1.22, SE = 0.297, z = -4.12, punadj < 0.0001 (sig.)
Average-Stressed_pc (for rMDD with average align_qual):
b = 1.27, SE = 0.339, z = 3.73, punadj = 0.00019 (sig.)
rMDD: b =0.381, SE = 0.0973, z = 3.92, padj = 0.00162
Average-Stressed_pm: b = 8.05, SE = 0.512, z = 15.7, padj <
0.0001
Average-Relaxed Brooding Average-Relaxed_pc (for HC with average align_qual): b = -
3.03, SE = 0.619, z = -4.89, padj < 0.0001
Average-Relaxed_pc (for rMDD with average align_qual): b
= -2.54, SE = 0.576, z = -4.40, punadj = 0.0001 (sig.)
Average-Relaxed_pc * align_qual_Average: b = -3.89, SE =
0.823, z = -4.73, padj < 0.0001
Average-Relaxed_pc (for align_qual_Average = +1SD): b =
-6.92, SE = 0.633, z = -10.9, punadj < 0.0001 (sig.)
Average-Relaxed_pc (for align_qual_Average = mean): b = -
3.02, SE = 0.699, z = -4.32, punadj = 0.00002 (sig.)
Average-Relaxed_pc (for align_qual_Average = -1SD): b =
0.877, SE = 0.827, z = 1.06, punadj = n.s.
Average-Active Mind-wandering Average-Active_pc (for HC with average align_qual): b = -
2.14, SE = 0.549, z = -3.90, padj = 0.00162
Average-Active_pc (for rMDD with average align_qual): b =
1.81, SE = 0.278, z = 6.52, punadj < 0.0001 (sig.)
Average-Active_pm: b = 18.9, SE = 0.591, z = 32.0, padj <
0.0001
Average-Active_pc * align_qual_Active: b = 1.83, SE =
0.539, z = 3.40, padj = 0.00485
Average-Active_pc (for align_qual_Active = +1SD): b = -
0.302. SE = 0.496, z = -0.610, punadj = n.s.
Average-Active_pc (for align_qual_Active = mean): b = -
2.14. SE = 2.63, z = -0.815, punadj = n.s.
Average-Active_pc (for align_qual_Active = -1SD): b = -
3.98. SE = 0.378, z = -10.5, punadj < 0.0001
Impulsivity Average-Active_pc (for HC with average align_qual): b = -
2.48, SE = 0.933, t = -2.66, padj = 0.0491
Average-Active_pc (for rMDD with average align_qual): b =
1.63, SE = 0.887, t =1.83, punadj = n.s.
Note. The “_pc” suffix represents the person-centered version of the physiological variable (i.e.,
denoting within-person effects), while the “_pm” suffix refers to the person mean of the
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
71
physiological variable (i.e., denoting between-person effects). “align_qual” = alignment quality.
“*” = interaction. padj = FDR-adjusted p-value. punadj = unadjusted p-value. adaptive.ER.sum =
sum of all adaptive ER strategies. freq.Average = frequency of average states. dwell.Average =
dwell time of average states.
Chapter Five: Discussion
This was the first study to date that utilized the temporal dynamics of ambulatory
physiological states to predict momentary affect and cognition. We proposed an approach to
idiographic state discovery using the HMM, aligned states across individuals using theoryderived template matching, and validated the clinical utility of physiological state dynamics by
examining their multilevel associations with psychological outcomes. Furthermore, to uncover
potential physiological underpinnings of MDD etiology, we investigated whether currently
healthy individuals with and without a history of MDD differed in the predictive relationships
between physiological state dynamics and psychological outcomes. Overall, we found significant
associations between frequency, dwell time and transitions of physiological states and selfreported momentary NA, maladaptive affect regulation and momentary impulsivity, and some of
these associations were significantly moderated by depression history. The current study
contributed to the research on mood psychopathology in three ways. First, by examining
multivariate patterns of cardiac, respiratory and movement variables, we demonstrated the utility
of higher-dimensional physiological constructs (e.g., physiological states) as helpful predictors
of psychological outcomes. Second, we showed that naturalistic fluctuations of affect, affect
regulation and impulsivity were tracked by dynamic transitions of latent physiological states.
Given that physiology can be passively measured by wearable technology that is increasingly
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
72
accessible to the general public, our findings could move us towards timely detection of daily
psychological risk and intervention based on passive physiology monitoring. Our discretization
of continuous physiological time-series into states made early warning signs of psychological
vulnerability binary (i.e., the presence or absence of a state), which could thus intuitively inform
dichotomous decision thresholds for intervention delivery. Third, group differences in
affect/cognition-physiology coupling could advance our understanding of proximal physiological
mechanisms of depression. Given the dynamic and context-dependent nature of affect and affect
regulation, the unique temporal associations between physiology and affect and cognition that
individuals with rMDD showed in ecologically valid settings may fill an important gap in the
existing literature on proximal risk factors in depression.
Negative Affect (NA)
For HCs with average-to-high alignment quality and individuals with rMDD who had
average alignment quality, NA was positively associated with transitions from relaxed to stressed
physiological states, which is in line with the well-established relationships between reduced HR
complexity and heightened NA and related mood disorders (e.g., Beauchaine, 2015; Dell’Acqua
et al., 2020; Ellis et al., 2016). While prior studies of HR complexity and psychopathology
typically studied indices of HR variability in isolation (e.g., HR-HRV in relation to daily NA:
Simon et al., 2020), the current study examined the predictive ability of multivariate patterns of
HR complexity indices, respiration and movement, and revealed that reduction in HR complexity
in an absence of intense activity (i.e., relaxed-to-stressed transitions) was indicative of NA.
However, relaxed-to-active transitions were associated with a decrease in NA among HCs at
high alignment quality, which could be interpreted as counterintuitive given that leaving a state
of elevated HR complexity (i.e., relaxed states) should correspond to adverse mood outcomes.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
73
One explanation for this finding might be the well-established positive effect of activity on mood
and mood disorders, which could have outweighed the potentially detrimental effect of leaving a
relaxed state (Bernstein & McNally, 2018; Kanning & Schlicht, 2010). However, the beneficial
effect of physical activity on mood seemingly contradicts with the finding that NA was
positively associated with frequency of active states among individuals with rMDD at average
levels of alignment quality. This contradiction is reflected in the larger literature that has
reported rather inconsistent relationships between activity and NA, as opposed to the more
reliable positive relationships between activity and PA (Schwerdtfeger et al., 2010). Moreover,
activity was operationalized as three-axis displacement in the current study, which did not
necessarily reflect exercise and could potentially also explain the inconsistent findings. For
example, increased activity could also represent agitated bodily movement that is a symptom of
MDD (American Psychiatric Association, 2013; Walther et al., 2019). Studies of momentary
locomotor activity and mood have shown that individuals tend to exhibit intermittent bursts of
activity in response to worsened mood, and that MDD is characterized by more intermittent
locomotor activity in everyday life (Kim et al., 2015; Nakamura et al., 2007). This may explain
why frequency, as opposed to dwell time, of active states predicted elevated NA among
individuals with rMDD, who may have shown brief agitated behavior in response to worsened
mood. Another explanation for the co-occurrence of heightened NA and activity might be that
instead of experiencing greater NA after engaging in activity, individuals with rMDD engaged in
more activity in response to heightened NA, while HCs did not engage in this regulatory
mechanism. Truly disentangling the directionality of the relationships between activity and NA
could be challenging, as it might require greater EMA sampling frequency that may either be
infeasible or perturb naturalistic behavior in unintended ways. Thus, it remains possible that
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
74
activity was used as a regulatory strategy among individuals with rMDD. This explanation is in
line with Schewerdtfeger and colleagues’ (2015) finding that momentary NA predicted higher
levels of bodily movement and less time spent in sedentary behaviors.
Brooding
For HCs with average-to-high alignment quality, all outward transitions of active states
coincided with a momentary increase in brooding. This finding echoes the aforementioned
finding that relaxed-to-active transitions, a type of inward transitions of active states, predicted
reduced NA among HCs, which provides further evidence for the importance of activity for daily
mood and affect regulation among healthy participants. The anti-correlation between activity and
brooding is supported by Clancy et al. (2020)’s study, which revealed a between-person negative
association between daily brooding and daily walking among healthy adults. Moreover, stressed
states appeared to be conducive to brooding among HCs: transitioning into stressed states
(active-to-stressed) predicted greater brooding, while transitioning out of them (stressed-torelaxed and stressed-to-active) predicted less brooding. This is paralleled by the apparently
protective effects of relaxed states: inward transitions of relaxed states predicted less brooding.
The opposite effects of stressed and relaxed states on brooding are consistent with the existing
literature on an anti-correlation between HRV and perseverative cognition (Ottaviani, 2018;
Ottaviani et al., 2015; Schumann et al., 2022; Verkuil et al., 2009). However, most studies on
HRV and perseverative cognition examined these constructs either cross-sectionally (e.g., trait
rumination: Verkuil et al., 2009) or through experimental manipulation (e.g., experimental
induction of rumination: Ottaviani et al., 2015), and only indices that capture linear variability of
heart rate (e.g., RMSSD) were assessed in relation to perseverative cognition. The current study
added to this literature by 1) revealing negative within-person associations between brooding and
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
75
HR complexity; 2) examining spontaneous brooding in everyday life, and 3) defining HR
complexity multidimensionally (i.e., variability, unpredictability, and self-similarity).
Interestingly, average states seemed to also facilitate brooding among HCs: transitioning into
average states (active-to-average) was associated with increased brooding, while leaving them
(average-to-stressed and average-to-relaxed) corresponded to less brooding. That active-toaverage transitions correlated with greater brooding may not be surprising given that a still
bodily state may be more conducive to sustained cognition such as brooding. However, it is less
clear why leaving average states, which were characterized by person-mean levels of HR
complexity, respiration and activity, would accompany less brooding. One hypothesis is that
average states might have been the “interchange stations” that bridged transitions between the
other three states. For instance, as one showed less brooding, one might have needed to go
through an average state in order to transition from a stressed state to a relaxed state, and simply
measuring first-order transitions could yield the finding that moving from an average state to a
relaxed state predicted less brooding. Thus, the effects of a first-order outward transition of
average states might depend on the unmeasured upstream state. Future research that examines
higher-order physiological state transitions would better shed light on this possibility.
For individuals with rMDD who had average alignment quality, an increase in brooding
was observed often when they made transitions that involved average states (i.e., average-tostressed, and active-to-average), except when the target state was a relaxed state (i.e., average-torelaxed) which corresponded to less brooding. This coincides with the aforementioned finding
that average states appeared to be conducive to brooding among HCs, and may similarly be
explained by the omission of higher-order transitions in the current analyses. Nevertheless,
relaxed states may have counteracted the effects of average states on perseverative cognition, as
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
76
average-to-relaxed transitions uniquely predicted less brooding, which was found among HCs as
well. Since previous literature has associated diminished HR complexity with perseverative
cognition, it seems reasonable that an increase in HR complexity, which could reflect cardiac
recovery, would correspond to less brooding. While HCs consistently showed increased
brooding when leaving active states, the effects of leaving active states were variable among
individuals with rMDD. This echoes the aforementioned hypothesis that individuals with rMDD
may have performed a mix of activity (e.g., psychomotor agitation, instrumental daily activity)
that all manifested as an increase in displacement, leading to inconsistent relationships between
active states and psychological outcomes. In addition, individuals with rMDD showed reduced
brooding when transitioning between active and stressed states, which is striking given that
stressed states did not seem to increase brooding in this case. This may highlight either the role
of activity in reducing perseverative cognition, or the relative lack of predictive utility of
physiological stress for perseverative cognition among individuals with rMDD. This second
hypothesis may be supported by previous studies that have implicated a lack of contextuallyappropriate vagal withdrawal in depression vulnerability: individuals with current and remitted
MDD show less decline in the variability (e.g., RSA) and unpredictability (e.g., entropy) of
cardiac signals in response to regulation cues such as increased NA and cognitive tasks (Stange
et al., 2023; Yaroslavsky et al., 2014). Similarly, Stange and colleagues (2020) showed that
individuals with less contextually-appropriate vagal withdrawal showed greater rumination
regardless of context, suggesting that physiological inflexibility may underlie cognitive
inflexibility (e.g., rumination) that is tightly linked to depression. Although people do in general
show vagal withdrawal in response to perseverative cognition (e.g., Ottaviani et al., 2016), no
studies have examined whether individuals with and without (remitted) depression differ in the
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
77
extent of vagal withdrawal following perseverative cognition. Thus, it could be argued that
autonomic inflexibility, a biomarker of depression vulnerability supported by previous research,
may have manifested as reduced synchrony between stressed physiological states and brooding
that uniquely characterized the rMDD sample in the current study.
Finally, between-person associations were found between state transitions and brooding.
Individuals who made more stressed-to-relaxed transitions and less active-to-average transitions
showed less brooding, which were consistent with their within-person relationships. However,
individuals who made more transitions from stressed to active states showed more brooding on
average, which is in contrast to the negative within-person relationships between the two
variables. This raises the interesting possibility of activity as an affect regulation mechanism:
when participants engaged more in brooding, they may have responded to this cognitive pattern
by increasing movement, which would manifest physiologically as stressed-to-active transitions;
however, individuals who showed more stressed-to-active transitions on average may habitually
engage in brooding and thus have a greater need to use physical activity to curtail brooding,
which would explain the positive between-person associations between stressed-to-active
transitions and brooding. Future research that tests the prospective associations between physical
activity and perseverative cognition and employs more refined measurement of physical activity
could better examine this hypothesis.
Mind-Wandering
For HCs with average alignment quality, unlike brooding which was positively associated
with inward transitions of stressed states, transitioning into stressed states actually predicted less
mind-wandering. This was paralleled by the finding that both inward and outward transitions of
relaxed states corresponded to increased mind-wandering, except when the destination state was
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
78
a stressed state. Moreover, there was a between-person association between dwelling in relaxed
states and increased mind-wandering. These findings are unexpected given the supposedly
maladaptive nature of mind-wandering given regulatory need. Indeed, prior studies have found
associations between mind-wandering and worsened mood and psychopathology (Killingsworth
& Gilbert, 2010; Smallwood et al., 2007). Similarly, most studies have documented a positive
effect of mindfulness on momentary mood and mood disorders (Hofmann & Gómez, 2017).
However, recent studies have suggested that the adaptiveness of mind-wandering may depend on
where the mind wanders to. For instance, Welz and colleagues (2017) found that momentary
mind-wandering actually led to an increase in PA and a decrease in NA, and that NA was the
lowest when participants had pleasant thoughts during mind-wandering episodes. Similarly,
Franklin et al. (2013) found that mind-wandering was associated with enhanced mood when it
engendered thoughts that participants deemed interesting. Furthermore, temporal orientation of
the thought also affects the consequence of mind-wandering: future-oriented thoughts tend to
lead to better mood than past-oriented thoughts (Spronken et al., 2016). Thus, it could be
possible that healthy participants’ mind-wandering episodes entailed primarily positive and
future-oriented thoughts, which was reflected physiologically as leaving states of diminished HR
complexity and experiencing states of heightened HR complexity, although it remains a
speculation without direct measures of the thought content. Another finding was that leaving
average states was associated with decreased mind-wandering among HCs, which could suggest
that average states were conducive to mind-wandering among HCs. Although no studies have
reported that being at one’s characteristic level of HR complexity is associated with mindwandering, this finding is in line with the previous speculation that mind-wandering may not
necessarily have engendered negative thoughts among HCs. Moreover, this phenomenon may
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
79
also be explained by the ubiquity of mind-wandering: laboratory and ambulatory studies have
shown that people spend 25-50% of their daily time on mind-wandering (Schooler et al., 2014).
Therefore, mind-wandering may have been a “default” cognitive state for the healthy
participants, which was physiologically indicated by person-mean levels of HR complexity,
respiration and activity.
The effects of active states were again variable in the rMDD group, showing different
relationships with mind-wandering. This is consistent with the aforementioned hypothesis that
individuals with rMDD may have engaged in more heterogeneous daily activities. Similar to
HCs, individuals with rMDD also showed a negative relationship between inward transitions of
stressed states and mind-wandering, suggesting that mind-wandering was not a behavioral
correlate of stressed physiology among individuals with rMDD. However, unlike HCs for whom
inward and outward transitions of relaxed states predicted more mind-wandering, individuals
with rMDD showed less mind-wandering when they had more outward transitions of relaxed
states, which may imply that relaxed states were conducive to mind-wandering. These findings
may also reflect the previous speculation that mind-wandering might have produced positive
thoughts and might thus be less maladaptive than originally hypothesized. Finally, unlike HCs
for whom departing from average states consistently correlated with less mind-wandering,
average states seemed to have less consistent effects on mind-wandering among individuals with
rMDD. Coupled with the discovery that average states seemed to facilitate brooding among
individuals with rMDD, the finding might suggest that brooding was a “default” cognitive state
among individuals with rMDD instead of mind-wandering, which would remain speculative
without more refined assessment of mind-wandering and brooding (e.g., through experimental
manipulations).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
80
At the between-person level, individuals who made more transitions into stressed states
showed greater mind-wandering over the week, which is in contrast to their negative withinperson relationships discussed previously yet consistent with the broader literature that supports
their positive between-person associations. For instance, Stange et al. (2023) showed that higher
HRV was associated with less mind-wandering on average but failed to show a within-person
association between the two. Also, most studies that reported the detrimental effects of mindwandering on psychological well-being employed cross-sectional measures, and were thus
unable to speak to the temporal associations between the two (e.g., Killingsworth & Gilbert,
2010). Therefore, it is possible that mind-wandering and stress physiology indeed exhibit
opposite within- and between-person dynamics, where the momentary effects of mind-wandering
on HR complexity might depend on the thought content yet a general tendency for spontaneous
thoughts may be contextually-inflexible and thus maladaptive. Furthermore, individuals who
made more transitions between active and relaxed states showed less mind-wandering, which is
in line with the existing literature on the cardioprotective effects of mindfulness yet contradictory
to the positive within-person associations between the two variables found in the current study.
This might again highlight the nuanced differences between within- and between-person
associations: mind-wandering might have primarily entailed positive thoughts and thus
momentarily led to more favorable physiological profiles, yet individuals who in general are
more active and showed elevated HR complexity might engage less in mind-wandering.
Momentary Impulsivity
Momentary impulsivity was only significantly associated with physiological state
dynamics among HCs and not among individuals with rMDD. Specifically, transitioning from
active to stressed states predicted increased momentary impulsivity. To date, few studies have
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
81
examined the associations between momentary impulsivity and ambulatory physiology. Stange
et al. (2023) reported that at the between-person level, higher momentary impulsivity was
associated with lower unpredictability (i.e., sample entropy) of heart rate signals over time,
which partially supports the current finding about a positive within-person association between
momentary impulsivity and stressed physiology. Moreover, studies that examined the predictive
utility of HR complexity for impulsive behaviors in everyday life, such as binge eating episodes,
have shown that reduction in HR complexity often precedes momentary impulsive behaviors
(Ranzenhofer et al., 2016). Our study advanced this literature by showing that a stressed
physiological state that features diminished HR complexity and low-to-average respiration and
activity may underlie episodes of elevated momentary impulsivity among healthy adults.
Furthermore, bidirectional transitions between average and active states were both correlated
with decreased momentary impulsivity. The effects of activity on momentary impulsivity have
not been documented in prior studies. However, intervention studies have supported the role of
physical activity in reducing trait-level impulsivity and factors that contribute to impulsive
choices such as delay discounting (Ghahramani et al., 2016; Sofis et al., 2017). Thus, it may be
possible that activity also helps alleviate impulsivity at the within-person level, although this
possibility remains to be explored.
Theoretical and Clinical Implications of Ambulatory Physiological Monitoring
Measuring human physiology as it naturally unfolds in everyday life could advance our
understanding of mood psychopathology and inform just-in-time adaptive interventions (JITAIs)
for mood disorders. Most studies of physiological correlates of psychological constructs were
limited to either examining cross-sectional relationships (e.g., resting HRV and depression) or
measuring HR complexity in the lab (e.g., HRV reactivity to mood induction), which are
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
82
undoubtedly crucial to understanding the role of vagal control in emotion psychopathology while
leaving their temporal associations in daily life an open question (Bylsma, 2021; Stange et al.,
2023). Studying the naturalistic trajectories of affect, cognition and physiology is important
given that these constructs are dynamic in nature and susceptible to environmental influences,
which means that cross-sectional associations may mask the rich temporal relationships between
these variables, and that laboratory settings may be limited in the kinds of dynamics they can
elicit in affect, cognition and physiology (Stange et al., 2019). Leveraging wearable technology
and EMA, we showed that daily fluctuations in NA, maladaptive ER strategies (brooding and
mind-wandering), and impulsivity were tracked by transitions of latent physiological states, and
that individuals with and without rMDD differed in some of the affect/cognition-physiology
associations. These findings could inform our understanding of proximal physiological
mechanisms of mood psychopathology that could not be examined in prior studies with single
time-point measurements. For instance, relaxed-to-stressed transitions, which predicted moments
of elevated NA in both HCs and individuals with rMDD, might highlight momentary loss of HR
complexity without changes in activity and respiration as a transdiagnostic physiological marker
of adverse mood states. In addition, the reduced synchrony between stressed physiology and
brooding that individuals with rMDD uniquely showed may be a relatively stable biomarker for
depression vulnerability, given that both HCs and individuals with rMDD do not have current
depression. Proximal mechanisms of mood psychopathology could complement the already
prolific literature on distal depression vulnerability factors (e.g., female sex) to further elucidate
mechanisms that predispose individuals to emotional psychopathology.
Furthermore, by validating the psychological relevance of physiological state dynamics,
the current study could inform the input (i.e., what variables to track), timing (i.e., when) and
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
83
modality (i.e., how) of JITAI for mood disorders. Our findings revealed that momentary
fluctuations in affect and cognition can be predicted by jointly monitoring HR complexity,
respiration and activity, suggesting that continuous streams of cardiorespiratory and motor
signals may serve as helpful input for JITAIs. The specific kinds of physiological dynamics that
predicted unfavorable psychological outcomes, such as relaxed-to-stressed state transitions, may
then represent physiological signatures of moments of risk that inform when JITAIs are most
needed. Finally, that reduced HR complexity was in most cases associated with unfavorable
affective and cognitive outcomes also highlights HR complexity as a potential intervention
target, suggesting that interventions for mood disorders may benefit from altering HR
complexity. Indeed, research on biofeedback breathing interventions, which involve elevating
HR complexity through paced breathing, has shown promising effects of the interventions on
emotion regulation and cognitive performance (e.g., inhibitory control) that are central to
emotion psychopathology (Mather & Thayer, 2018; Nashiro et al., 2023b). The current study not
only adds further evidence to the potential utility of biofeedback breathing interventions for
affect dysregulation but may also suggest alternative treatment targets (e.g., physical activity) for
JITAIs. In summary, tracking physiological data may then allow us to promptly detect early
warning signs of affect dysregulation non-intrusively and to deliver just-in-time adaptive
interventions maximally close to the moment of risk.
Discretizing Ambulatory Time-Series into Multivariate States
A unique methodological advance made in the present study was the discretization of
continuous time-series into multivariate states, which facilitates the examination of physiological
dynamics and could inform clinical decision-making in JITAIs. Whereas existing studies
typically examined singular physiological indices in relation to psychological outcomes (e.g.,
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
84
HRV and NA), the current study showed that concurrent levels of HR complexity, respiration
and activity were also indicative of momentary affect and cognition. Conceptualizing human
physiology multidimensionally is more in line with the reality that any physiological
measurement we make in fact results from an intricate symphony of multiple bodily components.
Moreover, imbuing time-points with categorical definitions (e.g., stressed, relaxed) based on
multivariate physiological patterns allows straightforward computation of summary statistics
(e.g., frequency, dwell time, transitions) that quantify their temporal trajectories, which may be
related to ambulatory self-report measures to validate their within-person psychological
relevance. Without such discretization, constructs such as the perseverance and switching of
physiological states would be challenging to study but may nevertheless be indicative of
psychological vulnerability.
A further advantage of classifying time-points is that the resulting discrete bodily states
may become intuitive decision criteria for JITAIs, as the timing of JITAIs would likely depend
on categorical thresholds (e.g., is the individual in a high-risk state now?). Indeed, dimensional
associations between HR complexity and affective/cognitive outcomes could indicate which
variables to monitor and intervene at in everyday life but do not directly speak to when
interventions may be warranted, which often necessitates dichotomous decision thresholds to be
set in a post-hoc fashion (e.g., criteria for what counts as “too much” NA). Therefore,
conceptualizing time-points as discrete states would provide more intuitive indication of when an
individual is in a state of vulnerability and would maximally benefit from intervention.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
85
Unique Affect/Cognition-Physiology Coupling in rMDD
Individuals with and without past MDD showed different associations between
physiological state dynamics and affect, affect regulation and impulsivity, which could suggest
proximal physiological mechanisms of MDD that are independent of active depression. First,
activity seemed to have different psychological implications for HCs and individuals with
rMDD. For HCs, activity appeared to be consistently psychologically favorable: in the vast
majority of cases, transitioning into active states predicted decreased NA, brooding, mindwandering and momentary impulsivity, while leaving active states predicted greater brooding,
mind-wandering and momentary impulsivity. For individuals with rMDD, however, the effects
of activity on psychological outcomes appeared to be mixed: visiting active states more
frequently corresponded to elevated NA, while transitions that involved active states had varying
effects on psychological outcomes depending on the other states involved. Because we
operationalized activity as three-axis displacement, activity was not synonymous with exercise.
Therefore, we speculate that these discrepancies might be explained by the different forms of
activity that HCs and individuals with rMDD engaged in: HCs could have engaged in more
exercise than individuals with rMDD, whereas individuals with rMDD might have performed a
mix of exercise, agitated behavior in response to negative mood, and instrumental daily activities
such as running errands. Even if the intensity of activity was the same across groups, HCs and
individuals with rMDD might have also differed in the temporal associations between activity
and stress: did they experience elevated stress first and responded to it with increased activity, or
did they engage in more activity in the absence of negative mood and felt even better afterwards?
Findings from the current study suggest that individuals with rMDD may have used activity as a
regulatory mechanism to counteract negative mood or cognitive states while HCs might have
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
86
simply enjoyed the psychological benefits of activity without experiencing the initial regulatory
need.
Moreover, stressed states showed varying associations with psychological outcomes
among HCs and individuals with rMDD, raising the possibility that inflexible autonomic
regulation may characterize the rMDD group. For HCs, transitioning into stressed states typically
predicted adverse psychological outcomes such as elevated NA, increased brooding, and
heightened momentary impulsivity, while moving out of stressed states typically predicted
favorable psychological outcomes including reduced brooding and mind-wandering. In contrast,
individuals with rMDD did not show many of these associations. These findings echo the
literature that has implicated the sensitivity of context-dependent vagal withdrawal in depression
etiology. For example, studies have found that compared to healthy individuals, individuals with
MDD and rMDD show blunted RSA reactivity to stressors (Bylsma et al., 2014; Yaroslavsky et
al., 2014). Studies utilizing ambulatory measures also confirmed the role of physiological
insensitivity to regulation cues (e.g., less synchrony between entropy of HR signals and NA) in
mood psychopathology in everyday life (Stange et al., 2023). Although the current study was not
optimized to test group differences in vagal reactivity, the relatively inconsistent relationships
between stress physiology and psychological outcomes that uniquely featured the rMDD group
nevertheless offered further evidence to the theory of contextually-appropriate vagal control that
supports adaptive self-regulation.
In addition to suggesting potential physiological mechanisms of MDD, group differences
in affect/cognition-physiology coupling may also present unique challenges and opportunities to
design interventions that utilize widely recognized affect/cognition-physiology relationships as
benchmarks for decision-making. First, the finding that diminished HR complexity at average
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
87
activity and respiration may be less indicative of unfavorable psychological states among
individuals with rMDD calls into question the universality of physiological indicators of stress –
how generalizable is the loss of physiological regulatory capacity a marker of psychological
stress? Without further examining individual differences in stress physiology, researchers could
easily design JITAIs that rely on assumed signs of physiological stress as triggers and stopping
criteria, which may consequently be less effective for individuals who show idiosyncratic affectphysiology associations. Thus, the current result highlights the importance of further
investigating the dynamic relationships between affect, cognition and physiology in diverse
populations. Alternatively, the current result could also underscore the need to broaden our
definitions of physiological stress. For example, if we incorporate more data sources, such as
cortisol levels, into the physiological states, could we create a more comprehensive and
multimodal indicator of physiological stress that shows more robust associations with affect and
cognition? Future research that utilizes diverse streams of data may be apt to answer this
question.
On the other hand, the finding that activity may carry different psychological
implications among HCs and individuals with rMDD might open an alternative avenue for
intervention. If individuals with rMDD are more likely to show agitated behavior in response to
elevated NA, could physical relaxation be an especially helpful just-in-time technique for this
population? Although not currently used as an JITAI, progressive muscular relaxation has been
shown to ameliorate depressive and anxiety symptoms in diverse populations (Essa et al., 2017;
Jacob & Sharma, 2018; Li et al., 2020). Thus, the current study might highlight the potential
utility of relaxation-based JITAIs for at-risk individuals. In addition, if individuals with rMDD
are more likely to use activity as a regulatory mechanism than HCs, could physical activity
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
88
JITAIs be more readily received by them? Physical activity JITAIs are one of the most tested
JITAIs with relatively well-established protocols and abundant empirical support for their
effectiveness (Hardeman et al., 2019; King et al., 2013). Given that physical activity JITAIs have
not been widely used for mood management, the current study may suggest the importance of
future research on physical activity as an intervention target for individuals with mood disorders.
Granted, the current study left many unresolved questions about the associations between
physiological stress, activity and affect. Nevertheless, it points out novel issues and opportunities
in intervention design that warrant further examination before clinical translation.
Limitations & Future Directions
One limitation of the current study is the overlapping and inconsistent timescales of the
physiological state dynamic variables and the EMA variables. While frequency, dwell time and
transitions of physiological states were calculated between each pair of post EMA surveys sent
approximately four hours apart, the EMA variables were measured at different timescales: affect
was measured instantaneously (i.e., just before the survey), affect regulation strategies and
perceived regulation success were rated based on the 30-to-90-minute window before the survey,
and momentary impulsivity was rated based on the past four hours. Therefore, only the time
window of momentary impulsivity was identical to that of the physiological state dynamics
variables, while the other EMA variables were measured in narrower windows. We chose to
have overlapping windows and tested the concurrent prediction of affective and cognitive
variables using physiological state dynamics because we aimed to uncover biomarkers of affect
and cognition that could inform early warning signs for just-in-time adaptive interventions.
However, there could also be utility to examining the associations between affect and cognition
and subsequent physiological state dynamics, which could reveal the physiological aftereffects of
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
89
affective and cognitive patterns (e.g., brooding). Moreover, overlapping time windows did not
permit the examination of the directionality of associations. For instance, we could not resolve
the temporal order of active states and brooding due to both processes being measured in
parallel, which rendered interpreting the different associations between active states and NA
among HCs and individuals with rMDD challenging. There is evidence that affect exerts lagged
effects on physical activity that peak at 7-9 hour time intervals (Ruissen et al., 2022). Thus,
future research could gain a clearer understanding of the temporal relationships between
ambulatory physiology and daily affect and cognition by testing lagged associations.
Another limitation of the current study is that most of the physiological state dynamic
variables and EMA variables were zero-inflated or heavily right-skewed. The zero-inflation of
physiological state dynamic variables may have been a byproduct of the theory-based state
alignment approach: each person was limited to having four states regardless of how many
idiographic states they expressed in total, which increased the chance of not observing any of the
four theory-driven states during an interval. On the other hand, the right-skewed distributions of
most EMA variables in the current study were mirrored by the larger literature that frequently
reports positively-skewed-distributed EMA variables (e.g., Shao et al., 2023). One explanation
for this phenomenon may be the low base-rates of some of the variables (e.g., impulsivity) in
naturalistic settings in non-clinical populations (Ruf et al., 2023). Alternatively, participants may
have shown reporting bias or careless responding (Jaso et al., 2022). To address the right-skewed
distributions of the EMA variables, future research could utilize in-lab paradigms to
experimentally induce different affective and cognitive states while continuously tracking
physiological signals to obtain more robust associations between physiological state dynamics
and affect and cognition.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
90
Finally, given the interesting discrepancies between HCs and individuals with rMDD in
terms of the psychological effects of activity, future research could benefit from more targeted
measurement of physical activity. For example, researchers could classify activity by intensity,
and specifically examine the effects of moderate-to-vigorous activity on affective states (e.g.,
Liao et al., 2017). Given that exercise has been utilized widely as an intervention for mood
disorders, it would be crucial in future research to establish the within-person positive effects of
physical activity on mood, and investigate any potential group moderators (e.g., past depression)
for this association.
Chapter Six: Conclusion
In summary, the present study employed a novel data-driven approach to examine the
temporal dynamics of ambulatory physiological states in relation to naturalistic fluctuations in
affect, affect regulation and cognition. We showed that transitions of physiological states tracked
changes in NA, engagement in maladaptive ER strategies, and momentary impulsivity, and that
individuals with and without prior depression history differed in affect-physiology coupling.
Both the methodology and results of the current study could have implications on understanding
physiological mechanisms of depression and informing JITAIs, while raising unanswered
questions that may inspire future research directions.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
91
References
Alvares, G. A., Quintana, D. S., Kemp, A. H., Van Zwieten, A., Balleine, B. W., Hickie, I. B., &
Guastella, A. J. (2013). Reduced heart rate variability in social anxiety disorder:
associations with gender and symptom severity. PloS one, 8(7), e70468.
https://doi.org/10.1371/journal.pone.0070468
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders
(5th ed.). https://doi.org/10.1176/appi.books.9780890425596
Aubert, A. E., Seps, B., & Beckers, F. (2003). Heart rate variability in athletes. Sports Medicine,
33(12), 889–919. https://doi.org/10.2165/00007256-200333120-00003
Bakhchina, A. V., Arutyunova, K. R., Sozinov, A. A., Demidovsky, A. V., & Alexandrov, Y. I.
(2018). Sample entropy of the heart rate reflects properties of the system organization of
behaviour. Entropy, 20(6), Article 6. https://doi.org/10.3390/e20060449
Bar-Yam, Y. (2020). Dynamics of Complex Systems. CRC Press.
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting Linear Mixed-Effects Models
Using lme4. Journal of Statistical Software, 67(1). https://doi.org/10.18637/jss.v067.i01
Beauchaine, T. P. (2015). Respiratory sinus arrhythmia: A transdiagnostic biomarker of emotion
dysregulation and psychopathology. Current Opinion in Psychology, 3, 43–47.
https://doi.org/10.1016/j.copsyc.2015.01.017
Beauchaine, T. P., & Thayer, J. F. (2015). Heart rate variability as a transdiagnostic biomarker of
psychopathology. International Journal of Psychophysiology, 98(2, Part 2), 338–350.
https://doi.org/10.1016/j.ijpsycho.2015.08.004
Babyak, M., Blumenthal, J. A., Herman, S., Khatri, P., Doraiswamy, M., Moore, K., Edward
Craighead, W., Baldewicz, T. T., & Ranga Krishnan, K. (2000). Exercise treatment for
major depression: Maintenance of therapeutic benefit at 10 months. Psychosomatic
Medicine, 62(5), 633.
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and
powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B
(Methodological), 57(1), 289–300.
Beauchaine, T. P., & Thayer, J. F. (2015). Heart rate variability as a transdiagnostic biomarker of
psychopathology. International Journal of Psychophysiology, 98(2, Part 2), 338–350.
https://doi.org/10.1016/j.ijpsycho.2015.08.004
Bernardi, L., Porta, C., Gabutti, A., Spicuzza, L., & Sleight, P. (2001). Modulatory effects of
respiration. Autonomic Neuroscience, 90(1–2), 47–56.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
92
Bernstein, E. E., & McNally, R. J. (2017). Acute aerobic exercise helps overcome emotion
regulation deficits. Cognition and Emotion, 31(4), 834–843.
Bernstein, E. E., & McNally, R. J. (2018). Exercise as a buffer against difficulties with emotion
regulation: A pathway to emotional wellbeing. Behaviour Research and Therapy, 109,
29–36. https://doi.org/10.1016/j.brat.2018.07.010
Berntson, G. G., Thomas Bigger Jr., J., Eckberg, D. L., Grossman, P., Kaufmann, P. G., Malik,
M., Nagaraja, H. N., Porges, S. W., Saul, J. P., Stone, P. H., & Van Der Molen, M. W.
(1997). Heart rate variability: Origins, methods, and interpretive caveats.
Psychophysiology, 34(6), 623–648. https://doi.org/10.1111/j.1469-8986.1997.tb02140.x
Berry, D., Palmer, A. R., Distefano, R., & Masten, A. S. (2019). Autonomic complexity and
emotion (dys-)regulation in early childhood across high- and low-risk contexts.
Development and Psychopathology, 31(3), 1173–1190.
https://doi.org/10.1017/S0954579419000683
Bornas, X., Llabrés, J., Noguera, M., López, A. M., Gelabert, J. M., & Vila, I. (2006). Fear
induced complexity loss in the electrocardiogram of flight phobics: A multiscale entropy
analysis. Biological Psychology, 73(3), 272–279.
https://doi.org/10.1016/j.biopsycho.2006.05.004
Briley, P. M., Webster, L., Boutry, C., Cottam, W. J., Auer, D. P., Liddle, P. F., & Morriss, R.
(2022). Resting-state functional connectivity correlates of anxiety co-morbidity in major
depressive disorder. Neuroscience & Biobehavioral Reviews, 138, 104701.
https://doi.org/10.1016/j.neubiorev.2022.104701
Bringmann, L. F., Albers, C., Bockting, C., Borsboom, D., Ceulemans, E., Cramer, A., Epskamp,
S., Eronen, M. I., Hamaker, E., Kuppens, P., Lutz, W., McNally, R. J., Molenaar, P., Tio,
P., Voelkle, M. C., & Wichers, M. (2022). Psychopathological networks: Theory,
methods and practice. Behaviour Research and Therapy, 149, 104011.
https://doi.org/10.1016/j.brat.2021.104011
Brooks, M. E., Kristensen, K., Benthem, K. J. van, Magnusson, A., Berg, C. W., Nielsen, A.,
Skaug, H. J., Maechler, M., & Bolker, B. M. (2017). glmmTMB Balances Speed and
Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. The
R Journal, 9(2), 378–400. https://doi.org/10.32614/RJ-2017-066
Brugnera, A., Zarbo, C., Tarvainen, M. P., Carlucci, S., Tasca, G. A., Adorni, R., Auteri, A., &
Compare, A. (2019). Higher levels of Depressive Symptoms are Associated with
Increased Resting-State Heart Rate Variability and Blunted Reactivity to a Laboratory
Stress Task among Healthy Adults. Applied Psychophysiology and Biofeedback, 44(3),
221–234. https://doi.org/10.1007/s10484-019-09437-z
Butler, E. A., Wilhelm, F. H., & Gross, J. J. (2006). Respiratory sinus arrhythmia, emotion, and
emotion regulation during social interaction. Psychophysiology, 43(6), 612–622.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
93
https://doi.org/10.1111/j.1469-8986.2006.00467.x
Bylsma, L. M. (2021). Emotion context insensitivity in depression: Toward an integrated and
contextualized approach. Psychophysiology, 58(2), e13715.
Bylsma, L. M., Salomon, K., Taylor-Clift, A., Morris, B. H., & Rottenberg, J. (2014). RSA
reactivity in current and remitted major depressive disorder. Psychosomatic Medicine,
76(1), 66–73. https://doi.org/10.1097/PSY.0000000000000019
Byun, S., Kim, A. Y., Jang, E. H., Kim, S., Choi, K. W., Yu, H. Y., & Jeon, H. J. (2019).
Detection of major depressive disorder from linear and nonlinear heart rate variability
features during mental task protocol. Computers in Biology and Medicine, 112, 103381.
https://doi.org/10.1016/j.compbiomed.2019.103381
Carels, R. A., Blumenthal, J. A., & Sherwood, A. (2000). Emotional responsivity during daily
life: Relationship to psychosocial functioning and ambulatory blood pressure.
International Journal of Psychophysiology, 36(1), 25–33.
Castiglioni, P., Parati, G., Lombardi, C., Quintin, L., & Di Rienzo, M. (2011). Assessing the
fractal structure of heart rate by the temporal spectrum of scale exponents: a new
approach for detrended fluctuation analysis of heart rate variability.
Chalmers, J. A., Quintana, D. S., Abbott, M. J. A., & Kemp, A. H. (2014). Anxiety disorders are
associated with reduced heart rate variability: a meta-analysis. Frontiers in Psychiatry, 5,
80. https://www.frontiersin.org/articles/10.3389/fpsyt.2014.00080
Chang, J. S., Yoo, C. S., Yi, S. H., Her, J. Y., Choi, H. M., Ha, T. H., Park, T., & Ha, K. (2012).
An integrative assessment of the psychophysiologic alterations in young women with
recurrent major depressive disorder. Psychosomatic Medicine, 74(5), 495.
https://doi.org/10.1097/PSY.0b013e31824d0da0
Cheung, M. N. (1981). Detection of and recovery from errors in cardiac interbeat intervals.
Psychophysiology, 18(3), 341–346. https://doi.org/10.1111/j.1469-8986.1981.tb03045.x
Cho, H., Ryu, S., Noh, J., & Lee, J. (2016). The effectiveness of daily mindful breathing
practices on test anxiety of students. PloS One, 11(10), e0164822.
https://doi.org/10.1371/journal.pone.0164822
Clancy, F., O’Connor, D. B., & Prestwich, A. (2020). Do Worry and Brooding Predict Health
Behaviors? A Daily Diary Investigation. International Journal of Behavioral Medicine,
27(5), 591–601. https://doi.org/10.1007/s12529-020-09898-1
Curtiss, J. E., Pinaire, M., Fulford, D., McNally, R. J., & Hofmann, S. G. (2022). Temporal and
contemporaneous network structures of affect and physical activity in emotional
disorders. Journal of Affective Disorders, 315, 139–147.
https://doi.org/10.1016/j.jad.2022.07.061
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
94
D’Agosto, T., Peçanha, T., Bartels, R., Moreira, D. N., Silva, L. P., Nóbrega, A. C. L., & Lima,
J. R. P. (2014). Cardiac autonomic responses at onset of exercise: Effects of aerobic
fitness. International Journal of Sports Medicine, 35(10), 879–885.
https://doi.org/10.1055/s-0034-1370911
Damaraju, E., Allen, E. A., Belger, A., Ford, J. M., McEwen, S., Mathalon, D. H., ... & Calhoun,
V. D. (2014). Dynamic functional connectivity analysis reveals transient states of
dysconnectivity in schizophrenia. NeuroImage: Clinical, 5, 298-308.
https://doi.org/10.1016/j.nicl.2014.07.003
de la Torre-Luque, A., Bornas, X., Balle, M., & Fiol-Veny, A. (2016). Complexity and nonlinear
biomarkers in emotional disorders: A meta-analytic study. Neuroscience & Biobehavioral
Reviews, 68, 410–422. https://doi.org/10.1016/j.neubiorev.2016.05.023
De Vito, G., Galloway, S. D. R., Nimmo, M. A., Maas, P., & McMurray, J. J. (2002). Effects of
central sympathetic inhibition on heart rate variability during steady‐state exercise in
healthy humans. Clinical Physiology and Functional Imaging, 22(1), 32–38.
Dell’Acqua, C., Dal Bò, E., Messerotti Benvenuti, S., & Palomba, D. (2020). Reduced heart rate
variability is associated with vulnerability to depression. Journal of Affective Disorders
Reports, 1, 100006. https://doi.org/10.1016/j.jadr.2020.100006
Difrancesco, S., Lamers, F., Riese, H., Merikangas, K. R., Beekman, A. T., van Hemert, A. M.,
Schoevers, R. A., & Penninx, B. W. (2019). Sleep, circadian rhythm, and physical
activity patterns in depressive and anxiety disorders: A 2‐week ambulatory assessment
study. Depression and Anxiety, 36(10), 975–986. https://doi.org/10.1002/da.22949
Dikecligil, G. N., & Mujica-Parodi, L. R. (2010). Ambulatory and challenge-associated heart rate
variability measures predict cardiac responses to real-world acute emotional stress.
Biological Psychiatry, 67(12), 1185–1190.
https://doi.org/10.1016/j.biopsych.2010.02.001
Du, N., Bai, S., Oguri, K., Kato, Y., Matsumoto, I., Kawase, H., & Matsuoka, T. (2005). Heart
rate recovery after exercise and neural regulation of heart rate variability in 30-40 year
old female marathon runners. Journal of Sports Science & Medicine, 4(1), 9–17.
Eavani, H., Satterthwaite, T. D., Gur, R. E., Gur, R. C., & Davatzikos, C. (2013). Unsupervised
learning of functional network dynamics in resting state fMRI. Information Processing in
Medical Imaging: 23rd International Conference, IPMI 2013, Asilomar, CA, USA, June
28–July 3, 2013. Proceedings 23, 426–437.
Ebner-Priemer, U. W., & Trull, T. J. (2009). Ecological momentary assessment of mood
disorders and mood dysregulation. Psychological Assessment, 21(4), 463.
https://doi.org/10.1037/a0017075
Egloff, B., Schmukle, S. C., Burns, L. R., & Schwerdtfeger, A. (2006). Spontaneous emotion
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
95
regulation during evaluated speaking tasks: Associations with negative affect, anxiety
expression, memory, and physiological responding. Emotion, 6(3), 356–366.
https://doi.org/10.1037/1528-3542.6.3.356
Ellis, A. J., Shumake, J., & Beevers, C. G. (2016). The effects of respiratory sinus arrhythmia on
anger reactivity and persistence in major depression. Psychophysiology, 53(10), 1587–
1599. https://doi.org/10.1111/psyp.12722
Erath, S. A., & Pettit, G. S. (2021). Coping with relationship stress in adolescence: A decade in
review. Journal of Research on Adolescence, 31(4), 1047–1067.
https://doi.org/10.1111/jora.12603
Essa, R. M., Ismail, N. I. A. A., & Hassan, N. I. (2017). Effect of progressive muscle relaxation
technique on stress, anxiety, and depression after hysterectomy. Journal of Nursing
Education and Practice, 7(7), 77.
Feng, C. X. (2021). A comparison of zero-inflated and hurdle models for modeling zero-inflated
count data. Journal of Statistical Distributions and Applications, 8(1), 8.
https://doi.org/10.1186/s40488-021-00121-4
Franklin, M. S., Mrazek, M. D., Anderson, C. L., Smallwood, J., Kingstone, A., & Schooler, J.
(2013). The silver lining of a mind in the clouds: Interesting musings are associated with
positive mood while mind-wandering. Frontiers in Psychology, 4.
https://doi.org/10.3389/fpsyg.2013.00583
Feder, J., & Feder, J. (1988). The fractal dimension. Fractals, 6–30.
Fiol-Veny, A., Balle, M., De la Torre-Luque, A., & Bornas, X. (2019). Negative cognitive
emotion regulation as a predictor of adolescent heart rate variability and entropy under
social stress. Anxiety, Stress, & Coping, 32(6), 641–653.
https://doi.org/10.1080/10615806.2019.1641199
Fisher, A. J., & Bosley, H. G. (2020). Identifying the presence and timing of discrete mood states
prior to therapy. Behaviour Research and Therapy, 128, 103596.
https://doi.org/10.1016/j.brat.2020.103596
Fisher, A. J., Medaglia, J. D., & Jeronimus, B. F. (2018). Lack of group-to-individual
generalizability is a threat to human subjects research. Proceedings of the National
Academy of Sciences, 115(27), E6106–E6115. https://doi.org/10.1073/pnas.1711978115
Fisher, A. J., Reeves, J. W., Lawyer, G., Medaglia, J. D., & Rubel, J. A. (2017). Exploring the
idiographic dynamics of mood and anxiety via network analysis. Journal of Abnormal
Psychology, 126(8), 1044. https://doi.org/10.1037/abn0000311
Garcia-Tabar, I., Sánchez-Medina, L., Aramendi, J. F., Ruesta, M., Ibañez, J., & Gorostiaga, E.
M. (2013). Heart rate variability thresholds predict lactate thresholds in professional
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
96
world-class road cyclists. J Exerc Physiol Online, 16, 38–50.
Garssen, B., Buikhuisen, M., & van Dyck, R. (1996). Hyperventilation and panic attacks. The
American Journal of Psychiatry. https://doi.org/10.1176/ajp.153.4.513
Ghahramani, M. H., Sohrabi, M., & Besharat, M. A. (2016). The effects of physical activity on
impulse control, attention, decision-making and motor functions in students with high
and low impulsivity. Biosciences Biotechnology Research Asia, 13(3), 1689-1696.
Giles, G. E., Eddy, M. D., Brunyé, T. T., Urry, H. L., Graber, H. L., Barbour, R. L., Mahoney, C.
R., Taylor, H. A., & Kanarek, R. B. (2018). Endurance exercise enhances emotional
valence and emotion regulation. Frontiers in Human Neuroscience, 12.
https://www.frontiersin.org/articles/10.3389/fnhum.2018.00398
Giurgiu, M., Koch, E. D., Ottenbacher, J., Plotnikoff, R. C., Ebner‐Priemer, U. W., & Reichert,
M. (2019). Sedentary behavior in everyday life relates negatively to mood: An
ambulatory assessment study. Scandinavian Journal of Medicine & Science in Sports,
29(9), 1340-1351. https://doi.org/10.1111/sms.13448
Gootzeit, J., & Markon, K. (2011). Factors of PTSD: Differential specificity and external
correlates. Clinical Psychology Review, 31(6), 993-1003.
https://doi.org/10.1016/j.cpr.2011.06.005
Greenberg, P. E., Fournier, A.-A., Sisitsky, T., Simes, M., Berman, R., Koenigsberg, S. H., &
Kessler, R. C. (2021). The economic burden of adults with major depressive disorder in
the United States (2010 and 2018). Pharmacoeconomics, 39(6), 653–665.
https://doi.org/10.1007/s40273-021-01019-4
Grossman, P., Wilhelm, F. H., & Spoerle, M. (2004). Respiratory sinus arrhythmia, cardiac vagal
control, and daily activity. American Journal of Physiology-Heart and Circulatory
Physiology, 287(2), H728-H734. https://doi.org/10.1152/ajpheart.00825.2003
Hankin, B. L., Abramson, L. Y., Moffitt, T. E., Silva, P. A., McGee, R., & Angell, K. E. (1998).
Development of depression from preadolescence to young adulthood: Emerging gender
differences in a 10-year longitudinal study. Journal of Abnormal Psychology, 107(1),
128. https://doi.org/10.1037/0021-843x.107.1.128
Hansen, E. C., Battaglia, D., Spiegler, A., Deco, G., & Jirsa, V. K. (2015). Functional
connectivity dynamics: modeling the switching behavior of the resting state. Neuroimage,
105, 525-535. https://doi.org/10.1016/j.neuroimage.2014.11.001
Hardeman, W., Houghton, J., Lane, K., Jones, A., & Naughton, F. (2019). A systematic review
of just-in-time adaptive interventions (JITAIs) to promote physical activity. International
Journal of Behavioral Nutrition and Physical Activity, 16, 1-21.
Hardin, J. W., & Hilbe, J. M. (2007). Generalized linear models and extensions. Stata press.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
97
Harris, P. A., Taylor, R., Thielke, R., Payne, J., Gonzalez, N., & Conde, J. G. (2009). Research
electronic data capture (REDCap)—A metadata-driven methodology and workflow
process for providing translational research informatics support. Journal of Biomedical
Informatics, 42(2), 377–381. https://doi.org/10.1016/j.jbi.2008.08.010
Haynos, A. F., Utzinger, L. M., Lavender, J. M., Crosby, R. D., Cao, L., Peterson, C. B., Crow,
S. J., Wonderlich, S. A., Engel, S. G., & Mitchell, J. E. (2018). Subtypes of adaptive and
maladaptive perfectionism in anorexia nervosa: Associations with eating disorder and
affective symptoms. Journal of Psychopathology and Behavioral Assessment, 40, 691–
700. https://doi.org/10.1007/s10862-018-9672-8
Hedeker, D., Mermelstein, R. J., Berbaum, M. L., & Campbell, R. T. (2009). Modeling mood
variation associated with smoking: An application of a heterogeneous mixed‐effects
model for analysis of ecological momentary assessment (EMA) data. Addiction, 104(2),
297–307.
Hibbert, G., & Pilsbury, D. (1989). Hyperventilation: Is it a cause of panic attacks? The British
Journal of Psychiatry, 155(6), 805–809. https://doi.org/10.1192/bjp.155.6.805
Hill, L. K., Siebenbrock, A., Sollers, J. J., & Thayer, J. F. (2009). Are all measures created
equal? Heart rate variability and respiration. Biomed. Sci. Instrum, 45, 71–76.
Hofmann, S. G., & Gómez, A. F. (2017). Mindfulness-based interventions for anxiety and
depression. Psychiatric Clinics, 40(4), 739–749.
Hoshi, R. A., Pastre, C. M., Vanderlei, L. C. M., & Godoy, M. F. (2013). Poincaré plot indexes
of heart rate variability: Relationships with other nonlinear variables. Autonomic
Neuroscience, 177(2), 271–274. https://doi.org/10.1016/j.autneu.2013.05.004
Howe, E. S., & Fisher, A. J. (2022). Identifying and predicting posttraumatic stress symptom
states in adults with posttraumatic stress disorder. Journal of Traumatic Stress, 35(5),
1508-1520. 10.1002/jts.22857
Jacob, S., & Sharma, S. (2018). Efficacy of progressive muscular relaxation on coping strategies
and management of stress, anxiety and depression. Int J Indian Psychol, 6(1), 106-124.
Jaso, B. A., Kraus, N. I., & Heller, A. S. (2022). Identification of careless responding in
ecological momentary assessment research: From posthoc analyses to real-time data
monitoring. Psychological Methods, 27(6), 958.
Javaheripour, N., Colic, L., Opel, N., Li, M., Maleki Balajoo, S., Chand, T., Van der Meer, J.,
Krylova, M., Izyurov, I., Meller, T., Goltermann, J., Winter, N. R., Meinert, S.,
Grotegerd, D., Jansen, A., Alexander, N., Usemann, P., Thomas-Odenthal, F., Evermann,
U., … Walter, M. (2023). Altered brain dynamic in major depressive disorder: State and
trait features. Translational Psychiatry, 13(1), Article 1. https://doi.org/10.1038/s41398-
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
98
023-02540-0
Jones, S. H., Hare, D. J., & Evershed, K. (2005). Actigraphic assessment of circadian activity
and sleep patterns in bipolar disorder. Bipolar Disorders, 7(2), 176–186.
https://doi.org/10.1111/j.1399-5618.2005.00187.x
Kamath, M., & Fallen, E. (1995). Correction of the heart rate variability signal for ectopics and
missing beats. In Heart Rate Variability (pp. 75–85). Futura Publishing Company.
Kanning, M., & Schlicht, W. (2010). Be active and become happy: An ecological momentary
assessment of physical activity and mood. Journal of Sport and Exercise Psychology,
32(2), 253–261. https://doi.org/10.1123/jsep.32.2.253
Kanters, J. K., Hojgaard, M. V., Agner, E., & Holstein-Rathlou, N. H. (1997). Influence of
forced respiration on nonlinear dynamics in heart rate variability. American Journal of
Physiology-Regulatory, Integrative and Comparative Physiology, 272(4), R1149–R1154.
https://doi.org/10.1152/ajpregu.1997.272.4.r1149
Kemp, A. H., Quintana, D. S., Gray, M. A., Felmingham, K. L., Brown, K., & Gatt, J. M. (2010).
Impact of depression and antidepressant treatment on heart rate variability: A review and
meta-analysis. Biological Psychiatry, 67(11), 1067–1074.
https://doi.org/10.1016/j.biopsych.2009.12.012
Killingsworth, M. A., & Gilbert, D. T. (2010). A Wandering Mind Is an Unhappy Mind. Science,
330(6006), 932–932. https://doi.org/10.1126/science.1192439
Kim, J., Nakamura, T., Kikuchi, H., Yoshiuchi, K., Sasaki, T., & Yamamoto, Y. (2015).
Covariation of Depressive Mood and Spontaneous Physical Activity in Major Depressive
Disorder: Toward Continuous Monitoring of Depressive Mood. IEEE Journal of
Biomedical and Health Informatics, 19(4), 1347–1355.
https://doi.org/10.1109/JBHI.2015.2440764
King, A. C., Hekler, E. B., Grieco, L. A., Winter, S. J., Sheats, J. L., Buman, M. P., ... &
Cirimele, J. (2013). Harnessing different motivational frames via mobile phones to
promote daily physical activity and reduce sedentary behavior in aging adults. PloS one,
8(4), e62613.
Kleiman, E. M., Turner, B. J., Fedor, S., Beale, E. E., Picard, R. W., Huffman, J. C., & Nock, M.
K. (2018). Digital phenotyping of suicidal thoughts. Depression and Anxiety, 35(7), 601-
608. https://doi.org/10.1002/da.22730
Koch, C., Wilhelm, M., Salzmann, S., Rief, W., & Euteneuer, F. (2019). A meta-analysis of heart
rate variability in major depression. Psychological Medicine, 49(12), 1948–1957.
Koenig, J., Kemp, A. H., Beauchaine, T. P., Thayer, J. F., & Kaess, M. (2016). Depression and
resting state heart rate variability in children and adolescents—A systematic review and
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
99
meta-analysis. Clinical Psychology Review, 46, 136–150.
https://doi.org/10.1016/j.cpr.2016.04.013
Kolar, D. R., Neumayr, C., Roth, M., Voderholzer, U., Perthes, K., & Schlegl, S. (2020). Testing
an emotion regulation model of physical activity in adolescents with anorexia nervosa: A
pilot ecological momentary assessment. European Eating Disorders Review, 28(2), 170–
183. https://doi.org/10.1002/erv.2706
Kovács, M., Yaroslavsky, I., Rottenberg, J., George, C. J., Kiss, E., Halas, K., ... & Kapornai, K.
(2016). Maladaptive mood repair, atypical respiratory sinus arrhythmia, and risk of a
recurrent major depressive episode among adolescents with prior major depression.
Psychological Medicine, 46(10), 2109-2119. https://doi.org/10.1017/s003329171600057x
Kwon, H. B., Yoon, H., Choi, S. H., Choi, J. W., Lee, Y. J., & Park, K. S. (2019). Heart rate
variability changes in major depressive disorder during sleep: Fractal index correlates
with BDI score during REM sleep. Psychiatry Research, 271, 291-298.
https://doi.org/10.1016/j.psychres.2018.11.021
Laborde, S., Mosley, E., & Mertgen, A. (2018). Vagal tank theory: The three Rs of cardiac vagal
control functioning – resting, reactivity, and recovery. Frontiers in Neuroscience, 12,
458. https://doi.org/10.3389/fnins.2018.00458
Lane, S. T., Gates, K. M., Pike, H. K., Beltz, A. M., & Wright, A. G. C. (2019). Uncovering
general, shared, and unique temporal patterns in ambulatory assessment data.
Psychological Methods, 24(1), 54–69. https://doi.org/10.1037/met0000192
Liao, Y., Chou, C. P., Huh, J., Leventhal, A., & Dunton, G. (2017). Associations of Affective
Responses During Free-Living Physical Activity and Future Physical Activity Levels: an
Ecological Momentary Assessment Study. International journal of behavioral medicine,
24(4), 513–519. https://doi.org/10.1007/s12529-016-9626-z
Lepoluoto, A., Nino, J., Tahvanainen, K., Ylitalo, R., Kuusela, T., Kähönen, M., & Kaila, T.
(2005). Propranolol increases the complexity of heart rate fluctuations: A mode of
antiarrhythmic action? Int. Journal of Clinical Pharmacology and Therapeutics, 43(02),
101–108. https://doi.org/10.5414/CPP43101
Li, M., Wang, L., Jiang, M., Wu, D., Tian, T., & Huang, W. (2020). Relaxation techniques for
depressive disorders in adults: a systematic review and meta-analysis of randomised
controlled trials. International Journal of Psychiatry in Clinical Practice, 24(3), 219-226.
Liu, H., Yang, Z., Meng, F., Huang, L., Qu, W., Hao, H., Zhang, J., & Li, L. (2018). Chronic
vagus nerve stimulation reverses heart rhythm complexity in patients with drug-resistant
epilepsy: An assessment with multiscale entropy analysis. Epilepsy & Behavior, 83, 168–
174. https://doi.org/10.1016/j.yebeh.2018.03.035
Liu, Q., Cole, D., Tran, T., Quinn, M., McCauley, E., Diamond, G., & Garber, J. (2023).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
100
Intraindividual phenotyping of depression in high-risk youth: An application of a
multilevel hidden Markov model. Development and Psychopathology, 1–10.
https://doi.org/10.1017/S0954579423000500
Luo, X., Qiao, L., & Che, X. (2018). Self-compassion modulates heart rate variability and
negative affect to experimentally induced stress. Mindfulness, 9(5), 1522–1528.
https://doi.org/10.1007/s12671-018-0900-9
Mather, M., & Thayer, J. F. (2018). How heart rate variability affects emotion regulation brain
networks. Current Opinion in Behavioral Sciences, 19, 98-104.
McNally, R. J. (2021). Network analysis of psychopathology: Controversies and challenges.
Annual Review of Clinical Psychology, 17(1), 31–53. https://doi.org/10.1146/annurevclinpsy-081219-092850
Meuret, A. E., Wilhelm, F. H., Ritz, T., & Roth, W. T. (2008). Feedback of end-tidal pCO2 as a
therapeutic approach for panic disorder. Journal of Psychiatric Research, 42(7), 560–
568. https://doi.org/10.1016/j.jpsychires.2007.06.005
Michel, C. M., & Koenig, T. (2018). EEG microstates as a tool for studying the temporal
dynamics of whole-brain neuronal networks: a review. Neuroimage, 180, 577-593.
https://doi.org/10.1016/j.neuroimage.2017.11.062
Möller, H. J. (2008). Outcomes in major depressive disorder: the evolving concept of remission
and its implications for treatment. The World Journal of Biological Psychiatry, 9(2), 102-
114.
Nakamura, T., Kiyono, K., Yoshiuchi, K., Nakahara, R., Struzik, Z. R., & Yamamoto, Y. (2007).
Universal scaling law in human behavioral organization. Physical Review Letters, 99(13),
138103.
Nashiro, K., Min, J., Yoo, H. J., Cho, C., Bachman, S. L., Dutt, S., Thayer, J. F., Lehrer, P. M.,
Feng, T., Mercer, N., Nasseri, P., Wang, D., Chang, C., Marmarelis, V. Z., Narayanan, S.,
Nation, D. A., & Mather, M. (2023a). Increasing coordination and responsivity of
emotion-related brain regions with a heart rate variability biofeedback randomized trial.
Cognitive, Affective, & Behavioral Neuroscience, 23(1), 66–83.
https://doi.org/10.3758/s13415-022-01032-w
Nashiro, K., Yoo, H. J., Cho, C., Min, J., Feng, T., Nasseri, P., ... & Mather, M. (2023b). Effects
of a randomised trial of 5-week heart rate variability biofeedback intervention on
cognitive function: possible benefits for inhibitory control. Applied Psychophysiology
and Biofeedback, 48(1), 35-48.
Neff, R. A., Wang, J., Baxi, S., Evans, C., & Mendelowitz, D. (2003). Respiratory sinus
arrhythmia. Circulation Research, 93(6), 565–572.
https://doi.org/10.1161/01.RES.0000090361.45027.5B
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
101
Nurnberger, J. I., Blehar, M. C., Kaufmann, C. A., Yorkcooler, C., Simpson, S. G.,
Harkavyfriedman, J., Severe, J. B., Malaspina, D., Reich, T., Miller, M., Bowman, E. S.,
Depaulo, J. R., Cloninger, C. R., Robinson, G., Modlin, S., Gershon, E. S., Maxwell, E.,
Guroff, J. J., Kirch, D., … Ritz, A. L. (1994). Diagnostic Interview for Genetic Studies:
Rationale, unique features, and training. Archives of General Psychiatry, 51, 849–859.
https://doi.org/10.1001/archpsyc.1994.03950110009002
Ottaviani, C. (2018). Brain-heart interaction in perseverative cognition. Psychophysiology, 55(7),
e13082. https://doi.org/10.1111/psyp.13082
Ottaviani, C., Shahabi, L., Tarvainen, M., Cook, I., Abrams, M., & Shapiro, D. (2015).
Cognitive, behavioral, and autonomic correlates of mind wandering and perseverative
cognition in major depression. Frontiers in Neuroscience, 8.
https://doi.org/10.3389/fnins.2014.00433
Ottaviani, C., Thayer, J. F., Verkuil, B., Lonigro, A., Medea, B., Couyoumdjian, A., &
Brosschot, J. F. (2016). Physiological concomitants of perseverative cognition: A
systematic review and meta-analysis. Psychological Bulletin, 142(3), 231.
Paffenbarger Jr, R. S., Lee, I.-M., & Leung, R. (1994). Physical activity and personal
characteristics associated with depression and suicide in American college men. Acta
Psychiatrica Scandinavica, 89(s377), 16–22. https://doi.org/10.1111/j.1600-
0447.1994.tb05796.x
Peng, C. K., Hausdorff, J. M., & Goldberger, A. L. (2000). Fractal mechanisms in neuronal
control: human heartbeat and gait dynamics in health and disease. Self-Organized
Biological Dynamics and Nonlinear Control, 66-96.
Peng, C. ‐K., Havlin, S., Stanley, H. E., & Goldberger, A. L. (1995). Quantification of scaling
exponents and crossover phenomena in nonstationary heartbeat time series. Chaos: An
Interdisciplinary Journal of Nonlinear Science, 5(1), 82–87.
https://doi.org/10.1063/1.166141
Penttilä, J., Helminen, A., Jartti, T., Kuusela, T., Huikuri, H. V., Tulppo, M. P., Coffeng, R., &
Scheinin, H. (2001). Time domain, geometrical and frequency domain analysis of cardiac
vagal outflow: Effects of various respiratory patterns. Clinical Physiology, 21(3), 365–
376. https://doi.org/10.1046/j.1365-2281.2001.00337.x
Penttilä, J., Helminen, A., Jartti, T., Kuusela, T., Huikuri, H. V., Tulppo, M. P., & Scheinin, H.
(2003). Effect of cardiac vagal outflow on complexity and fractal correlation properties
of heart rate dynamics. Autonomic and Autacoid Pharmacology, 23(3), 173–179.
https://doi.org/10.1046/j.1474-8673.2003.00293.x
Pfaltz, M. C., Michael, T., Grossman, P., Blechert, J., & Wilhelm, F. H. (2009). Respiratory
pathophysiology of panic disorder: An ambulatory monitoring study. Psychosomatic
Medicine, 71(8), 869. https://doi.org/10.1097/PSY.0b013e3181b492ff
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
102
Pichot, V., Roche, F., Celle, S., Barthélémy, J.-C., & Chouchou, F. (2016). HRVanalysis: A free
software for analyzing cardiac autonomic activity. Frontiers in Physiology, 7.
https://doi.org/10.3389/fphys.2016.00557
Pikkujämsä, S. M., Mäkikallio, T. H., Sourander, L. B., Räihä, I. J., Puukka, P., Skyttä, J., Peng,
C.-K., Goldberger, A. L., & Huikuri, H. V. (1999). Cardiac interbeat interval dynamics
from childhood to senescence. Circulation, 100(4), 393–399.
https://doi.org/10.1161/01.CIR.100.4.393
Porta, A., Castiglioni, P., Bari, V., Bassani, T., Marchi, A., Cividjian, A., Quintin, L., & Di
Rienzo, M. (2013). K -nearest-neighbor conditional entropy approach for the assessment
of the short-term complexity of cardiovascular control. Physiological Measurement,
34(1), 17–33. https://doi.org/10.1088/0967-3334/34/1/17
Porta, A., Gnecchi-Ruscone, T., Tobaldini, E., Guzzetti, S., Furlan, R., & Montano, N. (2007).
Progressive decrease of heart period variability entropy-based complexity during graded
head-up tilt. Journal of Applied Physiology, 103(4), 1143–1149.
https://doi.org/10.1152/japplphysiol.00293.2007
Ranzenhofer, L. M., Engel, S. G., Crosby, R. D., Haigney, M., Anderson, M., McCaffery, J. M.,
& Tanofsky-Kraff, M. (2016). Real-time assessment of heart rate variability and loss of
control eating in adolescent girls: A pilot study. International Journal of Eating
Disorders, 49(2), 197–201. https://doi.org/10.1002/eat.22464
Reyes del Paso, G. A., Langewitz, W., Mulder, L. J., Van Roon, A., & Duschek, S. (2013). The
utility of low frequency heart rate variability as an index of sympathetic cardiac tone: A
review with emphasis on a reanalysis of previous studies. Psychophysiology, 50(5), 477–
487.
Richman, J. S., & Moorman, J. R. (2000). Physiological time-series analysis using approximate
entropy and sample entropy. American Journal of Physiology-Heart and Circulatory
Physiology, 278(6), H2039–H2049. https://doi.org/10.1152/ajpheart.2000.278.6.H2039
Ross, C. E., & Hayes, D. (1988). Exercise and psychologic well-being in the community.
American Journal of Epidemiology, 127(4), 762–771.
https://doi.org/10.1093/oxfordjournals.aje.a114857
Rottenberg, J., Gross, J. J., & Gotlib, I. H. (2005). Emotion context insensitivity in major
depressive disorder. Journal of Abnormal Psychology, 114(4), 627–639.
https://doi.org/10.1037/0021-843X.114.4.627
Rothbaum, B. O., Meadows, E. A., Resick, P., & Foy, D. W. (2000). Cognitive-behavioral
therapy. In Effective treatments for PTSD: Practice guidelines from the International
Society for Traumatic Stress Studies. (pp. 320–325). The Guilford Press.
Ruf, A., Neubauer, A. B., Koch, E. D., Ebner-Priemer, U., Reif, A., & Matura, S. (2023).
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
103
Microtemporal Dynamics of Dietary Intake, Physical Activity, and Impulsivity in Adult
Attention-Deficit/Hyperactivity Disorder: Ecological Momentary Assessment Study
Within Nutritional Psychiatry. JMIR Mental Health, 10(1), e46550.
https://doi.org/10.2196/46550
Ruissen, G. R., Beauchamp, M. R., Puterman, E., Zumbo, B. D., Rhodes, R. E., Hives, B. A.,
Sharpe, B. M., Vega, J., Low, C. A., & Wright, A. G. C. (2022). Continuous-Time
Modeling of the Bidirectional Relationship Between Incidental Affect and Physical
Activity. Annals of Behavioral Medicine: A Publication of the Society of Behavioral
Medicine, 56(12), 1284–1299. https://doi.org/10.1093/abm/kaac024
Sandercock, G. R. H., Bromley, P. D., & Brodie, D. A. (2005). Effects of exercise on heart rate
variability: Inferences from meta-analysis. Medicine & Science in Sports & Exercise,
37(3), 433. https://doi.org/10.1249/01.MSS.0000155388.39002.9D
Sartory, G., Roth, W. T., & Kopell, M. L. (1992). Psychophysiological assessment of driving
phobia. Journal of Psychophysiology.
Schiweck, C., Piette, D., Berckmans, D., Claes, S., & Vrieze, E. (2019). Heart rate and high
frequency heart rate variability during stress as biomarker for clinical depression. A
systematic review. Psychological Medicine, 49(2), 200–211.
Schooler, J. W., Mrazek, M. D., Franklin, M. S., Baird, B., Mooneyham, B. W., Zedelius, C., &
Broadway, J. M. (2014). Chapter One - The Middle Way: Finding the Balance between
Mindfulness and Mind-Wandering. In B. H. Ross (Ed.), Psychology of Learning and
Motivation (Vol. 60, pp. 1–33). Academic Press. https://doi.org/10.1016/B978-0-12-
800090-8.00001-9
Schumann, A., Helbing, N., Rieger, K., Suttkus, S., & Bär, K.-J. (2022). Depressive rumination
and heart rate variability: A pilot study on the effect of biofeedback on rumination and its
physiological concomitants. Frontiers in Psychiatry, 13.
https://doi.org/10.3389/fpsyt.2022.961294
Schwerdtfeger, A., Eberhardt, R., Chmitorz, A., & Schaller, E. (2010). Momentary affect
predicts bodily movement in daily life: An ambulatory monitoring study. Journal of
Sport & Exercise Psychology, 32(5), 674–693. https://doi.org/10.1123/jsep.32.5.674
Shaffer, F., & Ginsberg, J. P. (2017). An overview of heart rate variability metrics and norms.
Frontiers in Public Health, 5, 258. https://doi.org/10.3389/fpubh.2017.00258
Shakil, S., Lee, C. H., & Keilholz, S. D. (2016). Evaluation of sliding window correlation
performance for characterizing dynamic functional connectivity and brain states.
Neuroimage, 133, 111-128. https://doi.org/10.1016/j.neuroimage.2016.02.074
Shao, S., Xu, Z., Liu, Q., McClure, K., Jacobucci, R., Maxwell, S. E., & Zhang, Z. (2023,
September 24). Zero inflation in intensive longitudinal data: why is it important and how
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
104
should we deal with it?. https://doi.org/10.31234/osf.io/8fscd
Simon, S. G., Sloan, R. P., Thayer, J. F., & Jamner, L. D. (2021). Taking context to heart:
Momentary emotions, menstrual cycle phase, and cardiac autonomic regulation.
Psychophysiology, 58(4), e13765. https://doi.org/10.1111/psyp.13765
Smallwood, J., O’Connor, R. C., Sudbery, M. V., & Obonsawin, M. (2007). Mind-wandering
and dysphoria. Cognition and Emotion, 21(4), 816–842.
https://doi.org/10.1080/02699930600911531
Smith, C. M., Chillrud, S. N., Jack, D. W., Kinney, P., Yang, Q., & Layton, A. M. (2019).
Laboratory validation of hexoskin biometric shirt at rest, submaximal exercise, and
maximal exercise while riding a stationary bicycle. Journal of Occupational and
Environmental Medicine, 61(4), e104. https://doi.org/10.1097/JOM.0000000000001537
Smith, R., Thayer, J. F., Khalsa, S. S., & Lane, R. D. (2017). The hierarchical basis of
neurovisceral integration. Neuroscience & Biobehavioral Reviews, 75, 274–296.
https://doi.org/10.1016/j.neubiorev.2017.02.003
Sofis, M. J., Carrillo, A., & Jarmolowicz, D. P. (2017). Maintained physical activity induced
changes in delay discounting. Behavior Modification, 41(4), 499-528.
Spronken, M., Holland, R. W., Figner, B., & Dijksterhuis, A. (2016). Temporal focus, temporal
distance, and mind-wandering valence: Results from an experience sampling and an
experimental study. Consciousness and Cognition, 41, 104–118.
https://doi.org/10.1016/j.concog.2016.02.004
Stange, J. P., Connolly, S. L., Burke, T. A., Hamilton, J. L., Hamlat, E. J., Abramson, L. Y., &
Alloy, L. B. (2016). Inflexible cognition predicts first onset of major depressive episodes
in adolescence. Depression and Anxiety, 33(11), 1005–1012.
https://doi.org/10.1002/da.22513
Stange, J. P., Hamilton, J. L., Fresco, D. M., & Alloy, L. B. (2017a). Flexible parasympathetic
responses to sadness facilitate spontaneous affect regulation. Psychophysiology, 54(7),
1054-1069. https://doi.org/10.1111/psyp.12856
Stange, J. P., Hamilton, J. L., Olino, T. M., Fresco, D. M., & Alloy, L. B. (2017b). Autonomic
reactivity and vulnerability to depression: A multi-wave study. Emotion, 17(4), 602.
https://doi.org/10.1037/emo0000254.supp
Stange, J. P., Hamilton, J. L., Shepard, R., Wu, J., Fresco, D. M., & Alloy, L. B. (2020).
Inflexible autonomic responses to sadness predict habitual and real-world rumination: A
multi-level, multi-wave study. Biological Psychology, 153, 107886.
https://doi.org/10.1016/j.biopsycho.2020.107886
Stange, J. P., Kleiman, E. M., Mermelstein, R. J., & Trull, T. J. (2019). Using ambulatory
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
105
assessment to measure dynamic risk processes in affective disorders. Journal of Affective
Disorders, 259, 325–336. https://doi.org/10.1016/j.jad.2019.08.060
Stange, J. P., Li, J., Xu, E. P., Ye, Z., Zapetis, S. L., Phanord, C. S., Wu, J., Sellery, P., Keefe,
K., Forbes, E., Mermelstein, R. J., Trull, T. J., & Langenecker, S. A. (2023). Autonomic
complexity dynamically indexes affect regulation in everyday life. Journal of
Psychopathology and Clinical Science, 132(7), 847–866.
https://doi.org/10.1037/abn0000849
Stanley, J., Peake, J. M., & Buchheit, M. (2013). Cardiac parasympathetic reactivation following
exercise: Implications for training prescription. Sports Medicine, 43(12), 1259–1277.
https://doi.org/10.1007/s40279-013-0083-4
Stein, M. B., Millar, T. W., Larsen, D. K., & Kryger, M. H. (1995). Irregular breathing during
sleep in patients with panic disorder. American Journal of Psychiatry, 152(8), 1168–
1173. https://doi.org/10.1176/ajp.152.8.1168
Thayer, J. F., & Lane, R. D. (2000). A model of neurovisceral integration in emotion regulation
and dysregulation. Journal of Affective Disorders, 61(3), 201–216.
https://doi.org/10.1016/S0165-0327(00)00338-4
Tiller, J., Pain, M., & Biddle, N. (1987). Anxiety disorder and perception of inspiratory resistive
loads. Chest, 91(4), 547–551. https://doi.org/10.1378/chest.91.4.547
Trull, T. J., & Ebner-Priemer, U. (2013). Ambulatory Assessment. Annual Review of Clinical
Psychology, 9(1), 151–176. https://doi.org/10.1146/annurev-clinpsy-050212-185510
Tulppo, M. P., Makikallio, T. H., Takala, T. E., Seppanen, T., & Huikuri, H. V. (1996).
Quantitative beat-to-beat analysis of heart rate dynamics during exercise. American
Journal of Physiology-Heart and Circulatory Physiology, 271(1), H244–H252.
https://doi.org/10.1152/ajpheart.1996.271.1.H244
van Genugten, C. R., Schuurmans, J., Hoogendoorn, A. W., Araya, R., Andersson, G., Baños, R.
M., Berger, T., Botella, C., Cerga Pashoja, A., & Cieslak, R. (2022). A data-driven
clustering method for discovering profiles in the dynamics of major depressive disorder
using a smartphone-based ecological momentary assessment of mood. Frontiers in
Psychiatry, 13, 321. https://doi.org/10.3389/fpsyt.2022.755809
van Londen, L., Kerkhof, G. A., van den Berg, F., Goekoop, J. G., Zwinderman, K. H.,
Frankhuijzen-Sierevogel, A. C., Wiegant, V. M., & de Wied, D. (1998). Plasma arginine
vasopressin and motor activity in major depression. Biological Psychiatry, 43(3), 196–
204. https://doi.org/10.1016/0924-977x(96)83066-5
Verkuil, B., Brosschot, J. F., de Beurs, D. P., & Thayer, J. F. (2009). Effects of explicit and
implicit perseverative cognition on cardiac recovery after cognitive stress. International
Journal of Psychophysiology, 74(3), 220–228.
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
106
Villar, R., Beltrame, T., & Hughson, R. L. (2015). Validation of the Hexoskin wearable vest
during lying, sitting, standing, and walking activities. Applied Physiology, Nutrition, and
Metabolism, 40(10), 1019-1024. https://doi.org/10.1139/apnm-2015-0140
Volkers, A. C., Tulen, J. H., van den Broek, W. W., Bruijn, J. A., Passchier, J., &
Pepplinkhuizen, L. (2003). Motor activity and autonomic cardiac functioning in major
depressive disorder. Journal of Affective Disorders, 76(1–3), 23–30.
https://doi.org/10.1016/s0165-0327(02)00066-6
Walther, S., Bernard, J. A., Mittal, V. A., & Shankman, S. A. (2019). The utility of an RDoC
motor domain to understand psychomotor symptoms in depression. Psychological
medicine, 49(2), 212-216.
Wagner, J. A., Feinn, R., Lampert, R., Bermúdez-Millán, A., & Pérez-Escamilla, R. (2019).
Changes in negative affect and changes in heart rate variability among low-income
latinos with type 2 diabetes in a randomized, controlled stress management trial. Journal
of Psychosomatic Research, 124, 109774.
https://doi.org/10.1016/j.jpsychores.2019.109774
Weaverdyck, M. E., Lieberman, M. D., & Parkinson, C. (2020). Tools of the Trade Multivoxel
pattern analysis in fMRI: a practical introduction for social and affective neuroscientists.
Social Cognitive and Affective Neuroscience, 15(4), 487-509.
https://doi.org/10.1093/scan/nsaa057
Weippert, M., Behrens, K., Rieger, A., Kumar, M., & Behrens, M. (2015). Effects of breathing
patterns and light exercise on linear and nonlinear heart rate variability. Applied
Physiology, Nutrition, and Metabolism, 40(8), 762–768. https://doi.org/10.1139/apnm2014-0493
Welz, A., Reinhard, I., Alpers, G. W., & Kuehner, C. (2018). Happy Thoughts: Mind Wandering
Affects Mood in Daily Life. Mindfulness, 9(1), 332–343. https://doi.org/10.1007/s12671-
017-0778-y
Wu, C., Fisher, A., & Schnyer, D. (2022). Gaussian Latent Dirichlet Allocation for discrete
human state discovery. arXiv. http://arxiv.org/abs/2206.14233
Yaroslavsky, I., Bylsma, L. M., Rottenberg, J., & Kovacs, M. (2013a). Combinations of resting
RSA and RSA reactivity impact maladaptive mood repair and depression symptoms.
Biological Psychology, 94(2), 272–281. https://doi.org/10.1016/j.biopsycho.2013.06.008
Yaroslavsky, I., Rottenberg, J., & Kovacs, M. (2014). Atypical patterns of respiratory sinus
arrhythmia index an endophenotype for depression. Development and Psychopathology,
26(4pt2), 1337–1352.
Yaroslavsky, I., Rottenberg, J., Bylsma, L. M., Jennings, J. R., George, C., Baji, I., ... & Kovacs,
M. (2016). Parasympathetic nervous system activity predicts mood repair use and its
PHYSIOLOGICAL STATE DYNAMICS AND AFFECT
107
effectiveness among adolescents with and without histories of major depression. Journal
of Abnormal Psychology, 125(3), 323.
Young, H., & Benton, D. (2015). We should be using nonlinear indices when relating heart-rate
dynamics to cognition and mood. Scientific Reports, 5(1), 16619.
https://doi.org/10.1038/srep1661
Abstract (if available)
Abstract
Human physiology reflects the body’s capacity for self-regulation that is crucial for flexible adaptation to changing environmental demands. Leveraging wearable sensors and machine learning, we aimed to uncover latent physiological states from ambulatory recordings of cardiac, respiratory and activity signals to predict momentary affective outcomes, with implications for informing just-in-time adaptive interventions. 51 participants with remitted major depressive disorder (rMDD) and 42 healthy controls (HCs) completed seven-day ecological momentary assessments of affect, affect regulation and impulsivity while their heart rate variability, respiration, and movement were passively monitored. Using Hidden Markov Models for state decoding, we found that frequency, dwell time, and transitions of physiological states predicted momentary affect, affect regulation, and impulsivity, with depression history moderating some of the associations. Findings underscore the utility of passive physiological phenotyping for tracking momentary affective processes that would otherwise be difficult to actively sample via self-report, but that may be crucial to informing optimal moments for intervention. Moreover, temporal associations between state dynamics and psychological variables may elucidate proximal mechanisms of affect dysregulation and suggest novel physiological treatment targets.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Impulsivity interferes with emotion regulation strategy prioritization in everyday life
PDF
Autonomic inertia as a proximal risk marker for moments of perseverative cognition in everyday life in remitted depression
PDF
Young adult dating couple interactions in daily life: links to family aggression and physiological processes
PDF
Aging and emotion regulation in the judgment of facial emotion
PDF
Arterial stiffness and resting-state functional connectivity in older adults
PDF
Visual and audio priming of emotional stimuli and their relationship to intertemporal preference shifts
PDF
Structure and function of the locus coeruleus across the lifespan
PDF
Understanding the dynamic relationships between physical activity and affective states using real-time data capture techniques
PDF
Age-related differences in amygdala subregion functional connectivity with Ventromedial Prefrontal Cortex and Dorsolateral Prefrontal Cortex
PDF
Prenatal and brain factors shape appetite regulation and weight from childhood onwards
PDF
The acute and longitudinal associations between sedentary behaviors, affective states, and emotional disorder symptoms among youth
PDF
Contributions of dynamic cerebrovascular function to cognitive decline and dementia: development and validation of a novel neuroimaging approach
PDF
Social attributes of adolescents’ problem-talk partners predict changes in depression
PDF
Associations between longitudinal loneliness, epigenetic age, and dementia risk
PDF
Family accommodation and youth anxiety: the roles of familial and child factors
PDF
Understanding emotional regulation and mood of young adults in the context of homelessness using geographic ecological momentary assessment
PDF
The role of blood pressure variability in cognitive decline, cerebrovascular disease and Alzheimer’s disease
PDF
The old ball and linkage: couples’ prenatal conflict behavior, cortisol linkage, and postpartum depression risk
PDF
The neuroscience of ambivalent and ambiguous feelings
PDF
Self-reported and physiological responses to hate speech and criticisms of systemic social inequality: an investigation of response patterns and their mediation…
Asset Metadata
Creator
Li, Jiani
(author)
Core Title
Latent physiological state dynamics underlying everyday affect, affect regulation and cognition
School
College of Letters, Arts and Sciences
Degree
Master of Science
Degree Program
Psychology
Degree Conferral Date
2024-12
Publication Date
01/12/2025
Defense Date
04/23/2024
Publisher
Los Angeles, California
(original),
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
affect,affect regulation,ambulatory physiological phenotyping,depression,ecological momentary assessment,OAI-PMH Harvest
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Stange, Jonathan (
committee chair
), Beam, Christopher (
committee member
), Mather, Mara (
committee member
)
Creator Email
jiani@usc.edu,jiani-li@outlook.com
Unique identifier
UC11399F8N7
Identifier
etd-LiJiani-13739.pdf (filename)
Legacy Identifier
etd-LiJiani-13739
Document Type
Thesis
Format
theses (aat)
Rights
Li, Jiani
Internet Media Type
application/pdf
Type
texts
Source
20250112-usctheses-batch-1233
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
affect
affect regulation
ambulatory physiological phenotyping
depression
ecological momentary assessment