Review of General Psychology Copyright 2006 by the American Psychological Association 2006, Vol. 10, No. 3, 229–240 1089-2680/06/$12.00 DOI: 10.1037/1089-2680.10.3.229

Heart Rate Variability as an Index of Regulated Emotional Responding

Bradley M. Appelhans and Linda J. Luecken Arizona State University

The study of individual differences in emotional responding can provide considerable insight into interpersonal dynamics and the etiology of psychopathology. rate variability (HRV) analysis is emerging as an objective measure of regulated emotional responding (generating emotional responses of appropriate timing and magnitude). This review provides a theoretical and empirical rationale for the use of HRV as an index of individual differences in regulated emotional responding. Two major theoretical frame- works that articulate the role of HRV in emotional responding are presented, and relevant empirical literature is reviewed. The case is made that HRV is an accessible research tool that can increase the understanding of emotion in social and psychopatho- logical processes.

Keywords: variability, emotion, affect, social interaction, psychopathology

Although the nature of emotions has been the The emotions that humans experience while subject of much debate, most theorists consider interacting with their environment are associ- emotions to be multifaceted processes involving ated with varying degrees of physiological coordinated changes in peripheral and central arousal (Levenson, 2003). A key system in- physiology (Thayer & Siegle, 2002), behavior volved in the generation of this physiological or behavioral tendencies, and cognitive process- arousal is the (ANS). ing. Emotions guide our decisions (Damasio, The ANS is subdivided into an excitatory sym- 2003, pp. 144–150), provide a substrate for pathetic nervous system (SNS) and an inhibi- social interaction (Keltner & Kring, 1998), and tory parasympathetic nervous system (PNS) facilitate responses to challenge (Tooby & Cos- that often interact antagonistically to produce mides, 1990). Emotions that are expressed with varying degrees of physiological arousal. Dur- sensitivity to the situational context in which ing physical or psychological , activity of they unfold, both in terms of timing/occurrence the SNS becomes dominant, producing physio- and magnitude, are more likely to facilitate logical arousal to aid in adapting to the chal- adaptive responses. Emotions are matched to lenge. An increased pulse, or heart rate, is char- their context through various regulatory strate- acteristic of this state of arousal. During periods gies implemented both during and after initial of relative safety and stability, the PNS is dom- emotional expression (Gross, 1998). Indeed, the inant and maintains a lower degree of physio- capacity to regulate emotion is vital to social logical arousal and a decreased heart rate. The functioning (Eisenberg, 2001) and maintaining ease with which an individual can transition mental health (Gross & Munoz, 1995). between high and low arousal states is depen- dent on the ability of the ANS to rapidly vary heart rate. Emotion regulation depends critically on an individual’s ability to adjust physiological Bradley M. Appelhans and Linda J. Luecken, Department of Psychology, Arizona State University. arousal on a momentary basis (Gross, 1998). A We are grateful to James Eason of Washington and Lee flexible ANS allows for rapid generation or University for his thoughtful comments on this article. modulation of physiological and emotional Correspondence concerning this article should be ad- states in accordance with situational demands. dressed to Bradley M. Appelhans, Department of Preventive Medicine, Northwestern University, 680 N. Lake Shore In contrast, autonomic rigidity results in a less- Drive, Suite 1220, Chicago, IL 60611. E-mail: ened capacity to generate or alter physiological [email protected] and emotional responses in synchrony with

229 230 APPELHANS AND LUECKEN changes in the environment. As described later sympathetic fibers has an excitatory influence in this article, (HRV) is a on the firing rate of the and, measure of the continuous interplay between among other things, results in increased heart sympathetic and parasympathetic influences on rate. In contrast, parasympathetic activation has heart rate that yields information about auto- an inhibitory influence on the pacemaking ac- nomic flexibility and thereby represents the ca- tivity of the sinoatrial node and produces de- pacity for regulated emotional responding. creased heart rate. Alternatively, it can be said This article provides theoretical and empiri- that the two autonomic branches regulate the cal support for the emergence of HRV as an lengths of time between consecutive heartbeats, important marker of emotion regulatory ability. or the interbeat intervals, with faster heart rates The first section of the article provides an in- corresponding to shorter interbeat intervals and troduction to the physiological underpinnings of vice versa. The PNS and SNS act antagonisti- HRV. The roles of autonomic and central ner- cally to influence cardiac activity. For example, vous system influences on cardiac functioning an increase in heart rate could arise from either are emphasized. The second section describes increased sympathetic activity or decreased the methods involved in measuring heart rate parasympathetic inhibition (vagal withdrawal). and deriving HRV “values” through three major Although both autonomic branches exert a con- classes of analyses. The third section describes stant influence on heart rate, parasympathetic contemporary theoretical models of the self- influence is predominant at rest and serves to regulatory function of HRV, including the poly- maintain resting heart rate well below the in- vagal theory (Porges, 1997, 2001) and the trinsic firing rate of the sinoatrial node (Bernt- model of neurovisceral integration (Thayer & son et al., 1997). Lane, 2000). A brief review of empirical re- The two branches of the ANS rely on differ- search supporting the role of HRV in emotion ent signaling mechanisms with different tempo- regulation (in a variety of contexts) is presented ral effects. Sympathetic influence on heart rate and evaluated in the fourth section. The final is mediated by neurotransmission of norepi- section includes a brief summary and suggested nephrine and possesses a slow course of action directions for future study. The goal is to pro- on cardiac function. That is, changes in heart vide an introduction to HRV, with the hopes of rate due to sympathetic activation unfold rather stimulating future research using this promising slowly, with peak effect observed after about 4 s physiological index of emotion regulatory and return to baseline after about 20 s. In con- ability. trast, parasympathetic regulation of the heart is mediated by neurotransmission Physiological Underpinnings of HRV and has a very short latency of response, with peak effect at about 0.5 s and return to baseline HRV reflects the degree to which cardiac within 1 s (Berntson et al., 1997; Pumprla, activity can be modulated to meet changing Howorka, Groves, Chester, & Nolan, 2002). situational demands. Although HRV is influ- The ability of the PNS to rapidly modulate enced by numerous physiological and environ- cardiac activity allows for flexibility in respond- mental factors, two are particularly prominent ing to environmental demands with physiolog- and of psychophysiological importance: the in- ical and emotional arousal. Owing to the differ- fluence of the ANS on cardiac activity and ANS ence in their latencies of action, the oscillations regulation by the central autonomic network in heart rate produced by the two autonomic (CAN). The heart is innervated by the sympa- branches occur at different speeds, or frequen- thetic and parasympathetic (vagal) branches of cies. This serves as the basis for the frequency- the ANS, which exert a regulatory influence on based HRV analyses described below. heart rate by influencing the activity of the Breathing air into the temporarily gates primary pacemaker of the heart, the sinoatrial off the influence of the parasympathetic influ- node. The sinoatrial node generates action po- ence on heart rate, producing a heart rate in- tentials that course throughout the cardiac tis- crease (see Berntson, Cacioppo, & Quigley, sue, causing regions of the myocardium (heart 1993). Breathing air out of the lungs reinstates muscle) to contract in the orchestrated fashion parasympathetic influence on heart rate, result- that characterizes a heartbeat. Activation of ing in a heart rate decrease. This rhythmic os- HEART RATE VARIABILITY 231 cillation in heart rate produced by respiration is gans) through the SNS and PNS and directly called respiratory sinus (Bernardi, influences heart rate. Therefore, HRV reflects Porta, Gabutti, Spicuzza, & Sleight, 2001; the moment-to-moment output of the CAN and, Berntson et al., 1993). As only cardiac parasym- by proxy, an individual’s capacity to generate pathetic activity possesses a latency of action regulated physiological responses in the context rapid enough to covary with respiration, respi- of emotional expression (Thayer & Lane, 2000; ratory sinus arrhythmia is a phenomenon known Thayer & Siegle, 2002). to be entirely mediated by the PNS. In fact, a large majority of parasympathetically mediated Assessment and Calculation of HRV variation in heart rate is produced by respiratory sinus arrhythmia (Berntson et al., 1997), and HRV measures are derived by estimating the many researchers have reported the magnitude variation among a set of temporally ordered of respiratory sinus arrhythmia as an index of interbeat intervals. Obtaining a series of inter- parasympathetically mediated HRV. Histori- beat intervals requires a continuous measure of cally, researchers assumed that the magnitude heart rate, typically (ECG). of respiratory sinus arrhythmia could serve as a A simple lead II configuration (electrodes on the linear index of “,” or the electrochem- right arm and left leg, plus a ground) is suffi- ical activity of the (Grossman, cient for HRV analysis. Data should be sampled Stemmler, & Meinhardt, 1990; Saul & Cohen, at a rate rapid enough to provide a high-resolu- 1994). However, this view has been challenged tion signal (e.g., 500–1,000 Hz; Berntson et al., (Pyetan & Akselrod, 2003), and strictly speak- 1997). After collection, the series of interbeat ing, respiratory sinus arrhythmia is an estimate intervals must be corrected for abnormal beats of parasympathetically mediated HRV rather and artifacts (see Kamath & Fallen, 1995). than of vagal tone. Commercially available software is then used to The autonomic influences on heart rate are define the interbeat intervals within an ECG regulated remotely by the distributed network recording. Several waveforms can be identified of brain areas composing the CAN (Benarroch, from a raw ECG recording, each corresponding 1993). The CAN supports regulated emotional to different components of a heartbeat. In HRV responding by flexibly adjusting physiological analysis, interbeat intervals are almost always arousal in accordance with changing situational defined as the temporal distance between R- demands. Thus, the CAN is critically involved spikes (see Figure 1), the prominent waveform in integrating physiological responses in the that corresponds to the contraction of the heart’s services of emotional expression, responding to ventricles. Because in theory HRV analyses are environmental demands, goal-directed behav- to be applied to a set of interbeat intervals ior, and homeostatic regulation. The neuroana- defined by consecutive “normal” beats (i.e., tomical composition of the CAN includes cor- produced by sinoatrial depolarizations), the in- tical (medial prefrontal and insular cortices), terbeat intervals are often referred to as “nor- limbic (anterior cingulate cortex, hypothalamus, mal-to-normal” (NN) intervals. Although in central nucleus of the amygdala, bed nucleus of practice HRV analyses can be applied to a cor- the stria terminalis), and brainstem (periaqua- rected series of interbeat intervals, not true nor- ductal gray matter, ventrolateral medulla, para- mal-to-normal intervals, the abbreviation NN brachial nucleus, nucleus of the solitary tract) has been incorporated into the names of many regions. The CAN receives input from visceral analyses. afferents regarding the physiological conditions The obtained set of interbeat intervals is used inside the body and input from sensory process- as input data for computing three classes of ing areas in the brain regarding the external HRV analyses. The statistical class of analyses sensory environment (Benarroch, 1993). This consists of variance-based calculations (de- input allows the CAN to dynamically adjust scribed below) performed on the set of interbeat physiological arousal, including arousal associ- intervals and yielding numerical estimates for ated with emotional expression and regulation, HRV in temporal units (e.g., milliseconds). in response to changes in internal and external Within the frequency class of analyses, a tech- conditions. The output of the CAN is transmit- nique known as power spectral analysis is used ted to the sinoatrial node (and many other or- to partition the variance of the interbeat inter- 232 APPELHANS AND LUECKEN

Figure 1. Idealized electrocardiograph segment representing two heartbeats. Waveforms are labeled with letters and correspond with specific electrophysiological events during a heart- beat. The interbeat interval is defined by the temporal distance between R-spikes, the waveforms corresponding to depolarization of the heart’s ventricles. vals into a spectrum of frequencies. The amount involving differences between interbeat inter- of variance within a given frequency range, vals, such as the mean squared difference of represented by the area under the curve, is re- successive NN intervals (MSSD), the square ferred to as power. The geometrical class of root of MSSD (RMSSD), the number of pairs of analyses derives estimates of HRV from the adjacent NN intervals differing by more than 50 geometric properties of the sample density dis- ms (NN50), and the proportion derived by di- tributions of either the interbeat intervals or the viding the NN50 by the total number of NN differences between consecutive interbeat inter- intervals (pNN50). The MSSD, RMSSD, vals (see Malik, 1995). Unlike the statistical NN50, and pNN50 are thought to represent class of analyses, geometrical analyses are less parasympathetically mediated HRV (Task affected by aberrations in the ECG input data Force, 1996). but require more extended sampling of ECG The frequency-based technique of power and provide less precise estimates of HRV spectral analysis is a relatively new and sophis- (Task Force, 1996). Therefore, they are less ticated approach to quantifying HRV. Power frequently reported and are not described here. spectral analysis parses from the ECG signal the The interested reader is referred to Malik (1995) quantity of HRV occurring at different frequen- and Task Force (1996). cies (these frequencies are expressed in hertz, Statistical analyses are frequently reported calculated as cycles per second). The technique and can be computed to represent overall HRV results in a power spectrum, a distribution of or HRV at different frequencies (see Kleiger, variance in heart rate at different frequencies Stein, Bosner, & Rottman, 1995). For example, (see Figure 2). Various approaches are used to SDNN refers to the of NN extract the oscillatory components of HRV intervals, and SDANN refers to the standard from a set of sequential interbeat intervals, the deviation of the average NN interval computed most common being the across all 5-min recording segments. SDNN and autoregressive modeling techniques. Both reflects overall HRV, whereas SDANN reflects approaches have been compared in detail HRV occurring at cycles longer than 5 min. (Cerutti, Bianchi, & Mainardi, 1995) and found Also among the statistical analyses are those to yield similar results (Task Force, 1996). HEART RATE VARIABILITY 233

Figure 2. An example of a heart rate variability power spectrum obtained using the fast Fourier transform on a 5-min recording obtained from a resting subject in supine position. The low-frequency (LF) component occurs between .04 and .15 Hz and the high-frequency (HF) component occurs between .15 and .40 Hz. Hz ϭ cycles per second.

The power spectrum contains prominent that are extraneous to respiratory sinus bands representing the major oscillatory com- arrhythmia. ponents of HRV. The high-frequency (HF) component occurs at the frequency of adult Psychophysiological Theories of HRV respiration, .15–.40 Hz and primarily reflects cardiac parasympathetic influence due to respi- Two major theories causally relate the auto- ratory sinus arrhythmia. The low-frequency nomic flexibility represented by HRV with reg- (LF) component occurs within .04–.15 Hz. ulated emotional responding. Porges’s (1997, Controversy exists over whether LF HRV is a 2001) is based within an evo- direct marker of SNS activity, is a product of lutionary framework, which understands as- both SNS and PNS activity (Malliani, Pagani, pects of human functioning in terms of ac- Lombardi, & Cerutti, 1991; Pumprla et al., quired, genetically based characteristics that are 2002; Task Force, 1996), or is mostly parasym- presumed to have aided in survival and/or re- pathetically mediated (Eckberg, 1997). Owing production throughout human phylogenetic his- to the concern that the LF component is con- tory. Specifically, the polyvagal theory posits taminated by parasympathetic influence, many that the human ANS evolved in three stages, researchers have reported the ratio of LF to HF each characterized by the acquisition of an au- power as an index of “sympathovagal balance” tonomic structure that plays a unique role in (Task Force, 1996; but also see Eckberg, 1997). social processes. First acquired was the dorsal As the SNS and PNS operate antagonistically in vagal complex, a slow-responding, unmyeli- their control of the heart, this ratio represents nated vagus nerve that supports simple immo- relative shifts or biases toward sympathetic or bilization (e.g., freezing) in response to threat. parasympathetic dominance over cardiac func- This “vegetative vagus” slows heart rate tion (Pumprla et al., 2002). Statistical filters, through tonic inhibition of sinoatrial node ac- such as Porges’s (1986) moving polynomial tivity. The capacity for active mobilization re- algorithm and Grossman’s peak-valley statistic sponses (e.g., fight or flight) became supported (Grossman, van Beek, & Wientjes, 1990), have with the subsequent acquisition of the SNS. been developed that remove influences on HRV Most recently acquired was the ventral vagal 234 APPELHANS AND LUECKEN complex, consisting of a fast-acting, myelinated parameters that guide the organization of emo- vagus that can rapidly withdraw and reinstate its tional responses. inhibitory influence on sinoatrial node activity. More important, the model of neurovisceral A key premise of the polyvagal theory is that integration views the CAN as the neurophysio- the ventral vagal complex has afferent fibers logical command center governing cognitive, terminating in the nuclei of the facial and tri- behavioral, and physiological elements into reg- geminal nerves and includes portions of cranial ulated emotion states (Hagemann, Waldstein, & nerves that mediate facial expression, head turn- Thayer, 2003; Thayer & Lane, 2000). The CAN ing, vocalization, listening, and other socially does this by inhibiting other potential responses. relevant behaviors. The connection of the ven- This inhibition requires reciprocal communica- tral vagal complex and these pro- tion among system components (e.g., feedback vides a mechanism by which cardiac states can loops), sensitivity to the initial conditions of the be coordinated with social behaviors (Porges, system, and the existence of multiple pathways 1997, 2001). to a response (e.g., combinations of sympathetic The polyvagal theory states that the ability of and parasympathetic activity), all elements of a the ventral vagal complex to rapidly withdraw neuroviscerally integrated dynamical system its inhibitory influence allows humans to rap- (Thayer & Seigle, 2002). Such inhibition is idly engage and disengage with their environ- thought to be mediated synaptically in the brain ment without the metabolic cost of activating and vagally in the periphery (Thayer & Fried- the slower responding SNS. The dynamic na- man, 2002). From this perspective, HRV can be ture of many social processes (e.g., nonverbal considered a proxy for the CAN’s ability to communication, romantic courtship; see Porges, regulate the timing and magnitude of an emo- 1998) requires this rapid management of meta- tional response through inhibition in accordance bolic resources. Only when ventral vagal com- with contextual factors. Both theories presented above are similar in plex withdrawal is insufficient to meet demands that they (a) specify a critical role for parasym- are other autonomic subsystems enlisted. In this pathetically mediated inhibition of autonomic respect, the polyvagal theory emphasizes the arousal in emotional expression and regulation relation of respiratory sinus arrhythmia (which and (b) maintain that HRV measures are infor- purportedly indexes ventral vagal complex ac- mative about individuals’ capacity for this as- tivity; Porges, 2001) and the regulation of the pect of regulated emotional responding. How- emotional processes underlying social behavior. ever, some differences between the theories do Thayer and colleagues (Thayer & Lane, exist. For example, the model of neurovisceral 2000) have outlined the model of neurovisceral integration is premised on neuroanatomical integration, which relates emotional responding links between the ANS and brain regions asso- with HRV through a dynamical systems per- ciated with emotional processing (e.g., cortical spective. Within the dynamical systems frame- and limbic areas of the CAN), whereas the work, large-scale patterns (e.g., fluctuations in polyvagal theory largely rests on the neural bird populations) are considered to be products connections between the vagus nerve and other of a number of smaller scale interacting forces cranial nerves that control peripheral structures (e.g., food supply, presence of predators) that involved in the behavioral expression of emo- influence one another in nonlinear fashions (see tion (e.g., the muscles of the face and head). Gleick, 1987). According to the model of neu- Attending to different neuroanatomical sub- rovisceral integration, the array of behavioral, strates has led to divergent extensions of each cognitive, and physiological processes involved theory, mostly to affective dysfunction for the in emotion are considered to be subsystems of a model of neurovisceral integration and to social larger, self-organizing system. Specific emotion and developmental processes for the polyvagal states emerge from interactions among these theory. Integrating our understanding of the lower level elements along the lines of certain neuroanatomical components from both models control parameters. The model of neurovisceral should help researchers generate more intricate integration cites a growing body of literature hypotheses about the interaction of autonomic, suggesting the dimensions of valence (aversive cognitive, and behavioral aspects of emotional vs. appetitive) and arousal represent the control expression and regulation. HEART RATE VARIABILITY 235

In addition to differing in their neuroanatomi- and Eisenberg concluded that those with lower cal emphases, the model of neurovisceral inte- resting respiratory sinus arrhythmia experi- gration and the polyvagal theory are also based enced greater negative emotional arousal in re- in different theoretical frameworks. The evolu- sponse to stress, which interfered with their tionary perspective within which the polyvagal ability to implement adaptive coping strategies. theory is presented concerns itself with recog- Similarly, higher resting HRV was associated nizing biologically based adaptations to chal- with reduced indices of distress in grade school lenges that have recurred throughout human children watching an upsetting film (Fabes, history. On the other hand, the dynamical sys- Eisenberg, & Eisenbud, 1993) and higher social tems perspective that characterizes the model of competence in young children (Fabes, Eisen- neurovisceral integration aims to uncover berg, Karbon, Troyer, & Switzer, 1994). Al- broad-based rules that produce emerging pat- though a few studies have associated higher terns within various systems. However, it ap- resting respiratory sinus arrhythmia with lower pears that the dynamical systems and evolution- ratings of social competence (Eisenberg et al., ary frameworks are complementary rather than 1995, 1996), results have generally supported contradictory, and the evolutionary process it- the premise that higher HRV reflects a greater self can be considered a dynamical system capacity for regulated emotional responses. (Kenrick, 2001). A synthesis between these per- Other evidence has supported a relation be- spectives might provide a more complete un- tween adaptive coping strategies and HRV. Re- derstanding of the role of HRV in emotional cently bereaved individuals with higher resting responding. respiratory sinus arrhythmia scored higher on measures of active coping and acceptance and Empirical Research With HRV lower on measures of passive coping (O’Connor, Allen, & Kaszniak, 2002). Female Although interest in cardiac rhythms may graduate students classified as repressive copers have existed as early as the first half of the 18th demonstrated lower resting LF and HF HRV century, academic interest in HRV did not near the time of a major examination than emerge until the 1960s (Berntson et al., 1997). women classified as low anxious (Fuller, 1992), Since then, medical researchers have estab- and women with lower parasympathetically me- lished diminished HRV as an independent risk diated HRV (RMSSD) during experimentally factor for several negative cardiovascular out- induced fear states reported greater use of de- comes (Curtis & O’Keefe, 2002) and as a proxy fensive coping (Pauls & Stemmler, 2003). Fi- for underlying cardiovascular disease processes nally, those who exhibited submissive behavior (Stys & Stys, 1998). A growing body of psy- during an interpersonal stressor had lower HRV chological research supports an association be- (SDNN and RMSSD) at rest and during the task tween HRV (particularly parasympathetically (Sgoifo et al., 2003). mediated HRV) and various measures and out- It would be expected that attention allocation, comes relevant to regulated emotional respond- which can be considered both an emotion reg- ing. One such outcome is coping, a construct ulation strategy (e.g., attentional deployment; highly intertwined with emotion regulation. Gross, 1998) and a cognitive component of Coping refers to a set of regulatory strategies emotion (e.g., fearful vigilance), is associated that are motivated by emotions (often negative with HRV. Insofar as attention allocation is emotions) and that frequently serve an emotion involved in selecting meaningful information regulatory function but generally involve either from the sensory environment, it is linked with nonemotional actions or nonemotional goals, or the emotional responses to that information. both (Gross, 1998). Higher levels of resting Both the polyvagal theory (Porges, 1992) and respiratory sinus arrhythmia have been associ- the model of neurovisceral integration (Thayer ated with greater self-reported emotion regula- & Lane, 2000) suggest that strong vagal regu- tion and the use of constructive coping strate- lation of the heart provides the capacity to gies in university students (Fabes & Eisenberg, quickly adjust cardiac parasympathetic influ- 1997). This relation between resting respiratory ence to foster attentional engagement or disen- sinus arrhythmia and constructive coping was gagement with the sensory environment. Sup- mediated by negative emotional arousal. Fabes porting this notion is the finding that dental- 236 APPELHANS AND LUECKEN phobic persons with higher parasympathetically polar depression showed reduced overall resting mediated HRV throughout the experiment were HRV (Cohen et al., 2003). One study found better able to inhibit attentional processing of decreased resting parasympathetically mediated the linguistic content of dental-related words in HRV in depressed men, but increased HRV in a modified version of the Stroop task (Johnsen depressed women (Thayer, Smith, Rossy, et al., 2003). Sollers, & Friedman, 1998). An inability to gen- The model of neurovisceral integration con- erate well-regulated autonomic responses to siders disorders, such as generalized stress has been observed in depression, as those anxiety disorder and panic disorder, to result reporting greater depressive symptoms exhib- from a breakdown in the inhibitory processes of ited larger decreases from baseline in parasym- the CAN. CAN disinhibition is thought to ac- pathetically mediated HRV during a stressful count for the continual state of excessive worry speech task and smaller increases in parasym- and behavioral avoidance that characterizes pathetically mediated HRV during a cold pres- generalized anxiety disorder and the potential sor task (Hughes & Stoney, 2000), indicating a for seemingly uncued panic attacks found in lessened capacity to regulate cardiac activity to panic disorder (Friedman & Thayer, 1998b). meet the task demands. Resting overall and This disinhibition would also be expected to parasympathetically mediated HRV have inter- produce reduced parasympathetically mediated estingly been shown to increase with successful HRV and attentional vigilance toward threat. treatment of depression (Balogh, Fitzpatrick, Biases in attentional processing toward threat- Hendricks, & Paige, 1993), suggesting that rest- related stimuli have also been observed in panic ing HRV is related to within-person variation in disorder and generalized anxiety disorder regulated emotional responding over time. Fi- (Bradley, Mogg, White, Groom, & de Bono, nally, there is accumulating evidence that sug- 1999; Mathews, Ridgeway, & Williamson, gests reduced parasympathetic control of the 1996). As predicted by the model, patients with heart (as indexed by HRV) may partially medi- generalized anxiety disorder have shown lower ate the well-documented link between depres- parasympathetically mediated HRV relative to sion and heart disease (Grippo & Johnson, controls during rest and during intense worry 2002; Sheffield et al., 1998). (Thayer, Friedman, & Borkovec, 1996). Lower overall and parasympathetically mediated HRV Summary and Future Directions (aggregated across several tasks) have been ob- served in nonclinical panickers and blood pho- HRV is emerging as an objective measure of bics relative to controls (Friedman & Thayer, individual differences in regulated emotional 1998a). Other manifestations of anxiety, such as responding, particularly as it relates to social trait anxiety (Fuller, 1992), social anxiety (Mez- processes and mental health. HRV can be de- zacappa et al., 1997), and self-perceived stress- rived through a number of statistical, geometri- induced anxiety (Sgoifo et al., 2003) have been cal, and frequency-based analyses that provide associated with reduced resting parasympatheti- information about sympathetic, parasympa- cally mediated or overall HRV, suggesting the thetic, and/or overall autonomic regulation of possibility that diminished autonomic flexibility the heart. Two theoretical models hypothesize may be an underlying causal factor. that the ability of the nervous system to track As with anxiety, it would be expected that changes in the environment and respond with diminished HRV would accompany depressive physiological arousal that is both proportionate states given that a core feature of depression is to the context in which it occurs and well- the inability to generate appropriate positive integrated with behavior and cognition is criti- and negative emotions (Rottenberg, Kasch, cal to emotional expression and regulation. The Gross, & Gotlib, 2002). Consistent with this polyvagal theory posits that evolutionary forces view, bereaved individuals (O’Connor, Allen, led to the acquisition of a rapidly responding & Kaszniak, 2002) and patients being treated vagus nerve that supports emotional expression for melancholic major depression with amitrip- and regulation through its neural connections tyline (Rechlin, Weis, Spitzer, & Kaschka, with peripheral structures involved with social 1994) exhibited diminished resting parasympa- functioning. The model of neurovisceral inte- thetically mediated HRV, and patients with bi- gration equates the CAN to a dynamical system HEART RATE VARIABILITY 237 that organizes changes in the physiological, sponding in adulthood (Luecken, Rodriguez, & cognitive, and behavioral subsystems from Appelhans, 2005). Potential clinical applica- which emotion states emerge along the dimen- tions of HRV also exist, as increasing an indi- sions of valence and arousal. vidual’s capacity for inhibitory emotion regula- These theories have generated studies linking tion through HRV (Lehrer et al., greater HRV to the use of adaptive emotion 2003) may have therapeutic implications for regulation and coping strategies and reduced mood, anxiety, and impulse control disorders. HRV with various outcomes indicative of emo- tional dysregulation, such as anxiety, depres- Conclusions sion, and rigid attentional processing of threat. Findings generally support the predictions of Research and theory support the utility of the polyvagal theory and the model of neuro- HRV as a noninvasive, objective index of the visceral integration, but future research is brain’s ability to organize regulated emotional needed to establish that the relation of HRV to responses through the ANS and as a marker of regulated emotional responding is independent individual differences in emotion regulatory ca- of other factors, such as intelligence or person- pacity. Unlike other psychophysiological vari- ality. Further investigation should attempt to ables, HRV provides information regarding elucidate those contexts in which the autonomic both PNS and SNS activity, thereby permitting flexibility represented by greater HRV is partic- inferences about both inhibitory and excitatory ularly adaptive, as well as situations in which processes in emotion regulation. As regulated greater HRV may be maladaptive. Although emotional responding plays a vital role in social theorists and researchers have emphasized the processes and mental health, researchers have importance of parasympathetically mediated much to gain by incorporating HRV into their HRV in regulated emotional responding, the research designs. HRV analysis also has great relative contribution of sympathetic regulation potential to illuminate the role of individual of the heart has not yet been clarified. It would differences in emotion regulation in other as- be worthwhile to identify those contexts in pects of human functioning. It is hoped and which the two branches of the ANS differen- anticipated that only a fraction of the utility of tially contribute to emotional responding. Fi- HRV for understanding emotional responding nally, research assessing the role of HRV in the has thus far been realized, which is a truly an regulation of positive emotions is generally exciting prospect for social scientists. lacking and would be beneficial in defining the relation of HRV to emotion regulation. HRV analysis may be applicable to numerous References areas of investigation. For example, a new view of pain as an emotion that motivates homeo- Balogh, S., Fitzpatrick, D. F., Hendricks, S. E., & static behaviors (Craig, 2003) suggests that in- Paige, S. R. (1993). Increases in heart rate vari- dividual differences in emotional responding ability with successful treatment in patients with major depressive disorder. Psychopharmacologi- (as assessed via HRV) may account for varia- cal Bulletin, 29, 201–206. tion in pain sensitivity and the ability to regulate Benarroch, E. E. (1993). The central autonomic net- pain through various strategies (e.g., attention work: Functional organization, dysfunction, and allocation). HRV can also be assessed through perspective. Mayo Clinic Proceedings, 68, 988– the use of ambulatory ECG monitors, allowing 1001. researchers to assess changes in both sympa- Bernardi, L., Porta, C., Gabutti, A., Spicuzza, L., & thetic and parasympathetic activity throughout Sleight, P. (2001). Modulatory effects of respira- the course of emotionally charged episodes tion. Autonomic Neuroscience: Basic and Clini- (e.g., interpersonal conflict) in naturalistic set- cal, 90, 47–56. tings. Finally, it is unknown whether HRV is Berntson, G. G., Bigger, J. T., Eckberg, D. L., Gross- man, P., Kaufmann, P. G., Malik, M., et al. (1997). influenced by early environmental influences Heart rate variability: Origins, methods and inter- during periods of greater neural plasticity. HRV pretive caveats. , 34, 623–648. analysis might help elucidate a neurophysiolog- Berntson, G. G., Cacioppo, J. T., & Quigley, K. S. ical pathway linking early family experiences (1993). Respiratory sinus arrhythmia: Autonomic with documented alterations in autonomic re- origins, physiological mechanisms, and psycho- 238 APPELHANS AND LUECKEN

physiological implications. Psychophysiology, 30, Fuller, B. F. (1992). The effects of stress-anxiety and 183–196. coping styles on heart rate variability. Interna- Bradley, B. P., Mogg, K., White, J., Groom, C., & de tional Journal of Psychophysiology, 12, 81–86. Bono, J. (1999). Attentional bias for emotional Gleick, J. (1987). Chaos: Making a new science. faces in generalized anxiety disorder. British Jour- New York: Viking Penguin. nal of Clinical Psychology, 38(3), 267–278. Grippo, A. J., & Johnson, A. K. (2002). Biological Cerutti, S., Bianchi, A. M., & Mainardi, L. T. (1995). mechanisms in the relationship between depres- Spectral analysis of the heart rate variability signal. sion and heart disease. Neuroscience and Biobe- In M. Malik & A. J. Camm (Eds.), Heart rate havioral Reviews, 26, 941–962. variability (pp. 63–74). Armonk, NY: Futura. Gross, J. J. (1998). The emerging field of emotion Cohen, H., Kaplan, Z., Kotler, M., Mittelman, I., regulation: An integrative review. Review of Gen- Osher, Y., & Bersudsky, Y. (2003). Impaired heart eral Psychology, 2, 271–299. rate variability in euthymic bipolar patients. Bipo- Gross, J. J., & Munoz, R. F. (1995). Emotion regu- lar Disorders, 5, 138–143. lation and mental health. Clinical Psychology: Sci- Craig, A. D. (2003). A new view of pain as a ho- ence and Practice, 2, 151–164. meostatic emotion. Trends in Neurosciences, 26, Grossman, P., Stemmler, G., & Meinhardt, E. (1990). 303–307. Paced respiratory sinus arrhythmia as an index of Curtis, B. M., & O’Keefe, Jr., J. H. (2002). Auto- cardiac parasympathetic tone during varying be- nomic tone as a cardiovascular risk factor: The havioral tasks. Psychophysiology, 27, 404–416. dangers of chronic flight or fight. Mayo Clinic Grossman, P., van Beek, J., & Wientjes, C. (1990). A Proceedings, 77, 45–54. comparison of three quantification methods for the Damasio, A. (2003). Looking for Spinoza: Joy, sor- estimation of respiratory sinus arrhythmia. Psy- row, and the feeling brain. Orlando, FL: Harcourt. chophysiology, 27, 702–714. Eckberg, D. (1997). Sympathovagal balance: A crit- Hagemann, D., Waldstein, S. R., & Thayer, J. F. ical appraisal. Circulation, 96, 3224–3232. (2003). Central and autonomic nervous system in- Eisenberg, N. (2001). The core and correlates of tegration in emotion. Brain and Cognition, 52, affective social competence. Social Develop- 79–87. ment, 10, 120–124. Hughes, J. W., & Stoney, C. M. (2000). Depressed Eisenberg, N., Fabes, R. A., Karbon, M., Murphy, mood is related to high-frequency heart rate vari- B. C., Wosinski, M., Polazzi, L., et al. (1996). The ability during stressors. Psychosomatic Medi- relations of children’s dispositional prosocial be- cine, 62, 796–803. havior to emotionality, regulation, and social func- Johnsen, B. H., Thayer, J. F., Laberg, J. C., tioning. Child Development, 67, 974–992. Wormnes, B., Raadal, M., Skaret, E., et al. (2003). Eisenberg, N., Fabes, R. A., Murphy, B., Maszk, P., Attentional and physiological characteristics of pa- Smith, M., & Karbon, M. (1995). The role of tients with dental anxiety. Anxiety Disorders, 17, emotionality and regulation in children’s social 75–87. functioning: A longitudinal study. Child Develop- Kamath, M. V., & Fallen, E. L. (1995). Correction of ment, 66, 1360–1384. the heart rate variability signal for ectopics and Fabes, R. A., & Eisenberg, N. (1997). Regulatory missing beats. In M. Malik & A. J. Camm (Eds.), control and adults’ stress-related responses to daily Heart rate variability (pp. 75–85). Armonk, NY: life events. Journal of Personality and Social Psy- Futura. chology, 73, 1107–1117. Keltner, D., & Kring, A. M. (1998). Emotion, social Fabes, R. A., Eisenberg, N., & Eisenbud, L. (1993). function, and psychopathology. Review of General Behavioral and physiological correlates of chil- Psychology, 2, 320–342. dren’s reactions to others in distress. Developmen- Kenrick, D. T. (2001). Evolutionary psychology, tal Psychology, 29, 655–663. cognitive science, and dynamical systems: Build- Fabes, R. A., Eisenberg, N., Karbon, M., Troyer, D., ing an integrative paradigm. Current Directions in & Switzer, G. (1994). The relations of children’s Psychological Science, 10, 13–17. emotion regulation to their vicarious emotional Kleiger, R. E., Stein, P. K., Bosner, M. S., & Rott- responses and comforting behaviors. Child Devel- man, J. N. (1995). Time domain measurements of opment, 65, 1678–1693. heart rate variability. In M. Malik & A. J. Camm Friedman, B. H., & Thayer, J. F. (1998a). Anxiety (Eds.), Heart rate variability (pp. 33–45). Ar- and autonomic flexibility: A cardiovascular ap- monk, NY: Futura. proach. Biological Psychology, 49, 303–323. Lehrer, P. M., Vaschillo, E., Vaschillo, B., Lu, S.-E., Friedman, B. H., & Thayer, J. F. (1998b). Autonomic Eckberg, D. L., Edelberg, R., et al. (2003). Heart balance revisited: Panic anxiety and heart rate vari- rate variability biofeedback increases baroreflex ability. Journal of Psychosomatic Research, 44, gain and peak expiratory flow. Psychosomatic 133–151. Medicine, 65, 796–805. HEART RATE VARIABILITY 239

Levenson, R. W. (2003). Blood, sweat, and fears: ternational Journal of Psychophysiology, 42, 123– The autonomic architecture of emotion. Annals of 146. the New York Academy of Sciences, 1000, 348– Pumprla, J., Howorka, K., Groves, D., Chester, M., & 366. Nolan, J. (2002). Functional assessment of heart Luecken, L. J., Rodriguez, A. P., & Appelhans, B. M. rate variability: Physiological basis and practical (2005). Cardiovascular stress responses in young applications. International Journal of Cardiol- adulthood associated with family-of-origin rela- ogy, 84, 1–14. tionship experiences. Psychosomatic Medicine, 67, Pyetan, E., & Akselrod, S. (2003). Do the high- 514–521. frequency indexes of HRV provide a faithful as- Malik, M. (1995). Geometrical methods for heart rate sessment of cardiac vagal tone? A critical theoret- variability assessment. In M. Malik & A. J. Camm ical evaluation. IEEE Transactions on Biomedical (Eds.), Heart rate variability (pp. 47–61). Ar- Engineering, 50, 777–783. monk, NY: Futura. Rechlin, T., Weis, M., Spitzer, A., & Kaschka, W. P. Malliani, A., Pagani, M., Lombardi, F., & Cerutti, S. (1994). Are affective disorders associated with al- (1991). Cardiovascular neural regulation explored terations of heart rate variability? Journal of Af- in the frequency domain. Circulation, 84 (2), 482– fective Disorders, 32, 271–275. 492. Rottenberg, J., Kasch, K. L., Gross, J. J., & Gotlib, Mathews, A., Ridgeway, V., & Williamson, D. A. I. H. (2002). Sadness and amusement reactivity (1996). Evidence for attention to threatening stim- differentially predict concurrent and prospective uli in depression. Behaviour Research & Ther- functioning in major depressive disorder. Emo- apy, 34, 695–705. tion, 2, 135–146. Mezzacappa, E., Tremblay, R. E., Kindlon, D., Saul, Saul, J. P., & Cohen, R. J. (1994). Respiratory sinus J. P., Arseneault, L., Seguin, J., et al. (1997). arrhythmia. In M. N. Levy & P. J. Schwartz (Eds.), Anxiety, antisocial behavior, and heart rate regu- Vagal control of the heart: Experimental basis and lation in adolescent males. Journal of Child Psy- clinical implications (pp. 511–536). Armonk, NY: Futura. chology and Psychiatry, 38, 457–469. Sgoifo, A., Braglia, F., Costoli, T., Musso, E., O’Connor, M.-F., Allen, J. J. B., & Kaszniak, A. W. Meerlo, P., Ceresini, G., et al. (2003). Cardiac (2002). Autonomic and emotion regulation in be- autonomic reactivity and salivary cortisol in men reavement and depression. Journal of Psychoso- and women exposed to social stressors: Relation- matic Research, 52, 183–185. ship with individual ethological profile. Neuro- Pauls, C. A., & Stemmler, G. (2003). Repressive and science and Biobehavioral Reviews, 27, 179–188. defensive coping during fear and anger. Emo- Sheffield, D., Krittayaphong, R., Cascio, W. E., tion, 3, 284–302. Light, K. C., Golden, R. N., Finkel, J. B., et al. Porges, S. W. (1986). Respiratory sinus arrhythmia: (1998). Heart rate variability at rest and during Physiological basis, quantitative methods, and mental stress in patients with coronary artery dis- clinical implications. In P. Grossman, K. H. L. ease: Differences in patients with high and low Janssen, & D. Vaitl (Eds.), Cardiorespiratory and depression scores. International Journal of Behav- cardiosomatic psychophysiology (pp. 101–116). ioral Medicine, 5(1), 31–47. New York: Plenum Press. Stys, A., & Stys, T. (1998). Current clinical applica- Porges, S. W. (1992). Autonomic regulation and at- tions of heart rate variability. Clinical Cardiol- tention. In B. A. Campbell, H. Hayne, & R. Rich- ogy, 21, 719–724. ardson (Eds.), Attention and information process- Task Force of the European Society of Cardiology ing in infants and adults: Perspectives from human and the North American Society of Pacing and and animal research (pp. 201–223). Hillsdale, NJ: Electrophysiology. (1996). Heart rate variability: Erlbaum. Standards of measurement, physiological interpre- Porges, S. W. (1997). Emotion: An evolutionary by- tation, and clinical use. European Heart Jour- product of the neural regulation of the autonomic nal, 17, 354–381. nervous system. In C. S. Carter, B. Kirkpatrick, & Thayer, J. F., & Friedman, B. H. (2002). Stop that! I. I. Lederhendler (Eds.), Annals of the New York Inhibition, sensitization, and their neurovisceral Academy of Sciences: Vol. 807. The integrative concomitants. Scandinavian Journal of Psychol- neurobiology of affiliation (pp. 62–77). New York: ogy, 43, 121–130. New York Academy of Sciences. Thayer, J. F., Friedman, B. H., & Borkovec, T. D. Porges, S. W. (1998). Love: An emergent property of (1996). Autonomic characteristics of generalized the mammalian autonomic nervous system. Psy- anxiety disorder and worry. Biological Psychia- choneuroendocrinology, 23, 837–861. try, 39, 255–266. Porges, S. W. (2001). The polyvagal theory: Phylo- Thayer, J. F., & Lane, R. D. (2000). A model of genetic substrates of a social nervous system. In- neurovisceral integration in emotion regulation 240 APPELHANS AND LUECKEN

and dysregulation. Journal of Affective Disor- Tooby, J., & Cosmides, L. (1990). The past explains ders, 61, 201–216. the present: Emotional adaptations and the struc- Thayer, J. F., & Siegle, G. J. (2002). Neurovisceral ture of ancestral environments. Ethology and So- integration in cardiac and emotional regulation. ciobiology, 11, 375–424. IEEE Engineering in Medicine and Biology, 21, 24–29. Thayer, J. F., Smith, M., Rossy, L. A., Sollers, J. J., & Friedman, B. H. (1998). Heart period variability Received May 8, 2005 and depressive symptoms: Gender differences. Bi- Revision received June 20, 2005 ological Psychiatry, 44, 304–306. Accepted September 15, 2005 Ⅲ