ArticlePDF AvailableLiterature Review

A healthy heart is not a metronome: an integrative review of the heart's anatomy and heart rate variability

Frontiers
Frontiers in Psychology
Authors:

Abstract and Figures

Heart rate variability (HRV), the change in the time intervals between adjacent heartbeats, is an emergent property of interdependent regulatory systems that operate on different time scales to adapt to challenges and achieve optimal performance. This article briefly reviews neural regulation of the heart, and its basic anatomy, the cardiac cycle, and the sinoatrial and atrioventricular pacemakers. The cardiovascular regulation center in the medulla integrates sensory information and input from higher brain centers, and afferent cardiovascular system inputs to adjust heart rate and blood pressure via sympathetic and parasympathetic efferent pathways. This article reviews sympathetic and parasympathetic influences on the heart, and examines the interpretation of HRV and the association between reduced HRV, risk of disease and mortality, and the loss of regulatory capacity. This article also discusses the intrinsic cardiac nervous system and the heart-brain connection, through which afferent information can influence activity in the subcortical and frontocortical areas, and motor cortex. It also considers new perspectives on the putative underlying physiological mechanisms and properties of the ultra-low-frequency (ULF), very-low-frequency (VLF), low-frequency (LF), and high-frequency (HF) bands. Additionally, it reviews the most common time and frequency domain measurements as well as standardized data collection protocols. In its final section, this article integrates Porges' polyvagal theory, Thayer and colleagues' neurovisceral integration model, Lehrer et al.'s resonance frequency model, and the Institute of HeartMath's coherence model. The authors conclude that a coherent heart is not a metronome because its rhythms are characterized by both complexity and stability over longer time scales. Future research should expand understanding of how the heart and its intrinsic nervous system influence the brain.
This content is subject to copyright.
REVIEW ARTICLE
published: 30 September 2014
doi: 10.3389/fpsyg.2014.01040
A healthy heart is not a metronome: an integrative review
of the heart’s anatomy and heart rate variability
Fred Shaffer1*, Rollin McCraty 2and Christopher L. Zerr1
1Center for Applied Psychophysiology, Department of Psychology, Truman State University, Kirksville, MO, USA
2HeartMath Research Center, Institute of HeartMath, Boulder Creek, CA, USA
Edited by:
J. P. Ginsberg, Dorn VA Medical
Center, USA
Reviewed by:
Andrew Kemp, Universidade de São
Paulo, Brazil
Amit Jasvant Shah, Emory
University, USA
*Correspondence:
Fred Shaffer, Center for Applied
Psychophysiology, Department of
Psychology, Truman State
University, 100 E. Normal, Kirksville
MO 63501, USA
e-mail: fredricshaffer@gmail.com
Heart rate variability (HRV), the change in the time intervals between adjacent heartbeats,
is an emergent property of interdependent regulatory systems that operate on different
time scales to adapt to challenges and achieve optimal performance. This article briefly
reviews neural regulation of the heart, and its basic anatomy, the cardiac cycle, and
the sinoatrial and atrioventricular pacemakers. The cardiovascular regulation center in the
medulla integrates sensory information and input from higher brain centers, and afferent
cardiovascular system inputs to adjust heart rate and blood pressure via sympathetic and
parasympathetic efferent pathways. This article reviews sympathetic and parasympathetic
influences on the heart, and examines the interpretation of HRV and the association
between reduced HRV, risk of disease and mortality, and the loss of regulatory capacity.
This article also discusses the intrinsic cardiac nervous system and the heart-brain
connection, through which afferent information can influence activity in the subcortical and
frontocortical areas, and motor cortex. It also considers new perspectives on the putative
underlying physiological mechanisms and properties of the ultra-low-frequency (ULF),
very-low-frequency (VLF), low-frequency (LF), and high-frequency (HF) bands. Additionally,
it reviews the most common time and frequency domain measurements as well as
standardized data collection protocols. In its final section, this article integrates Porges
polyvagal theory, Thayer and colleagues’ neurovisceral integration model, Lehrer et al.’s
resonance frequency model, and the Institute of HeartMaths coherence model. The
authors conclude that a coherent heart is not a metronome because its rhythms are
characterized by both complexity and stability over longer time scales. Future research
should expand understanding of how the heart and its intrinsic nervous system influence
the brain.
Keywords: heart rate variability, psychophysiological coherence, neurocardiology, biofeedback interventions,
emotional self-regulation
INTRODUCTION
THE HEART
The heart is about the size of a closed fist, weighs between 250 and
350 g, and beats approximately 100,000 times a day and 2.5 billion
times during an average lifetime. The muscular heart consists of
two atria and two ventricles. The atria are upper receiving cham-
bers for returning venous blood. The ventricles comprise most
of the heart’s volume, lie below the atria, and pump blood from
the heart into the lungs and arteries. Deoxygenated blood enters
the right atrium, flows into the right ventricle, and is pumped to
the lungs via the pulmonary arteries, where wastes are removed
and oxygen is replaced. Oxygenated blood is transported through
the pulmonary veins to the left atrium and enters the left ven-
tricle. When the left ventricle contracts, blood is ejected through
the aorta to the arterial system (Marieb and Hoehn, 2013; Tortora
and Derrickson, 2014).
THE CARDIAC CYCLE
The cardiac cycle consists of systole (ventricular contraction) and
diastole (ventricular relaxation). During systole, blood pressure
(BP) peaks as contraction by the left ventricle ejects blood from
the heart. Systolic BP is measured during this phase. During dias-
tole, BP is lowest when the left ventricle relaxes. Diastolic BP is
measured at this time.
PACEMAKERS
The heart contains autorhythmic cells that spontaneously gen-
erate the pacemaker potentials that initiate cardiac contractions.
These cells continue to initiate heartbeats after surgeons sever all
efferent cardiac nerves and remove a heart from the chest cavity
for transplantation. Autorhythmic cells function as pacemakers
and provide a conduction pathway for pacemaker potentials.
The sinoatrial (SA) node and atrioventricular (AV) node are
the two internal pacemakers that are primarily responsible for ini-
tiating the heartbeat. The electrocardiogram (ECG) records the
action of this electrical conduction system and contraction of the
myocardium (Figure 1).
CARDIAC CONDUCTION
In a healthy heart, the SA node initiates each cardiac cycle through
spontaneous depolarization of its autorhythmic fibers. The SA
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |1
Shaffer et al. A healthy heart is not a metronome
FIGURE 1 | The generation of the electrocardiogram. Credit: Alila Sao
Mai/Shutterstock.com.
node’s intrinsic firing rate of about 60–100 action potentials
per minute usually prevents slower parts of the conduction sys-
tem and myocardium (heart muscle) from generating competing
potentials. The AV node can replace an injured or diseased SA
node as pacemaker and spontaneously depolarizes 40–60 times
per minute. The SA node generates an electrical impulse that trav-
elsthroughtheatriatotheAVnodeinabout0.03sandcausesthe
AV node t o re (Figure 2).ThePwaveoftheECGisproducedas
muscle cells in the atria depolarize and culminates in contraction
of the atria (atrial systole).
The signal rapidly spreads through the AV bundle reaching
the top of the septum. These fibers descend down both sides
of the septum as the right and left bundle branches and con-
duct the action potential over the ventricles about 0.2 s after
the appearance of the P wave. Conduction myofibers, which
extend from the bundle branches into the myocardium, depolar-
ize contractile fibers in the ventricles (lower chambers), resulting
in the QRS complex followed by the S-T segment. Ventricular
contraction (ventricular systole) occurs after the onset of the
QRS complex and extends into the S-T segment. The repolar-
ization of ventricular myocardium generates the T wave about
0.4 s following the P wave. The ventricles relax (ventricular dias-
tole) 0.6 s after the P wave begins (Tortora and Derrickson,
2014).
REGULATION OF THE HEART
In a healthy organism, there is a dynamic relative balance between
the sympathetic nervous system (SNS) and parasympathetic ner-
vous system (PNS). PNS activity predominates at rest, resulting
in an average HR of 75 beats per minute (bpm). This is signif-
icantly slower than the SA node’s intrinsic rate, which decreases
with age from an average 107 bpm at 20 years to 90 bpm at 50
years (Opthof, 2000). The parasympathetic branch can slow the
heart to 20 or 30 bpm or briefly stop it (Tortora and Derrickson,
2014). This illustrates the response called accentuated antagonism
(Olshansky et al., 2008). Parasympathetic nerves exert their effects
more rapidly (<1 s) than sympathetic nerves (>5s;Nunan et al.,
2010).
A major cardiovascular center, located in the medulla of
the brainstem, integrates sensory information from propriocep-
tors (limb position), chemoreceptors (blood chemistry), and
mechanoreceptors (also called baroreceptors) from the heart and
information from the cerebral cortex and limbic system. The car-
diovascular center responds to sensory and higher brain center
input by adjusting heart rate via shifts in the relative balance
between sympathetic and parasympathetic outflow (Shaffer and
Venner, 2013).
In a healthy individual, the HR estimated at any given time
represents the net effect of the neural output of the parasympa-
thetic (vagus) nerves, which slow HR, and the sympathetic nerves,
which accelerate it. At rest, both sympathetic and parasympa-
thetic nerves are tonically active with the vagal effects dominant.
Therefore, HR reflects the relative activity of the sympathetic
and parasympathetic systems; with the more important question
being, is the relative balance (HR) appropriate for the context the
person is engaged in at any given moment? In other words, is
HR higher during the daytime and when dealing with challeng-
ing tasks, and lower at night, during sleep or when not engaged in
challenging duties or activities?
The most obvious effect of vagal activity is to slow or even
stop the heart. The vagus nerves are the primary nerves for the
parasympathetic system and innervate the intrinsic cardiac ner-
vous system and project to the SA node, AV node, and atrial
cardiac muscle. Increased efferent activity in these nerves trig-
gers acetylcholine release and binding to muscarinic (mainly M2)
receptors. This decreases the rate of spontaneous depolarization
in the SA and AV nodes, slowing HR. Because there is sparse
vagal innervation of the ventricles, vagal activity minimally affects
ventricular contractility.
The response time of the sinus node is very short and the
effect of a single efferent vagal impulse depends on the phase
of the cardiac cycle at which it is received. Thus, vagal stimula-
tion results in an immediate response that typically occurs within
the cardiac cycle in which it occurs and affects only one or two
heartbeats after its onset. After cessation of vagal stimulation, HR
rapidly returns to its previous level. An increase in HR can also be
achieved by reduced vagal activity or vagal block. Thus, sudden
changesinHR(upordown)betweenonebeatandthenextare
parasympathetically mediated (Hainsworth, 1995).
An increase in sympathetic activity is the principal method
used to increase HR above the intrinsic level generated by the
SA node. Following the onset of sympathetic stimulation, there is
Frontiers in Psychology | Psychology for Clinical Settings September 2014 | Volume 5 | Article 1040 |2
Shafferetal. A healthy heart is not a metronome
FIGURE 2 | The depolarization and repolarization of the heart. Credit: Alila Sao Mai/Shutterstock.com.
a delay of up to 5 s before the stimulation induces a progressive
increase in HR, which reaches a steady level in 20–30s if the
stimulus is continuous (Hainsworth, 1995). The slowness of the
response to sympathetic stimulation is in direct contrast to vagal
stimulation, which is almost instantaneous. However, the effect
on HR is longer lasting and even a short stimulus can affect
HR for 5–10 s. Efferent sympathetic nerves target the SA node
and AV node via the intrinsic cardiac nervous system, and the
bulk of the myocardium (heart muscle). Action potentials con-
ducted by these motor neurons trigger norepinephrine (NE)
and epinephrine (E) release and binding to beta-adrenergic (β1)
receptors located on cardiac muscle fibers. This speeds up sponta-
neous depolarization in the SA and AV nodes, increases HR, and
strengthens the contractility of the atria and ventricles. In failing
hearts, the number of β1 receptors is reduced and their cardiac
muscle contraction in response to NE and E binding is weakened
(Ogletree-Hughes et al., 2001).
AFFERENT MODULATION OF CARDIAC AND BRAIN ACTIVITY
The field of neurocardiology explores the anatomy and func-
tions of the connections between the heart and brain (Davis and
Natelson, 1993; Armour, 2003) and represents the intersection
of neurology and cardiology. While efferent (descending) regula-
tion of the heart by the autonomic nervous system (ANS) is well
known, newer data have suggested a more complex modulation of
heart function by the intrinsic cardiac nervous system (Kukanova
and Mravec, 2006). These intracardiac neurons (sensory, inter-
connecting, afferent, and motor neurons) (Verkerk et al., 2012)
can operate independently and their network is sufficiently exten-
sive to be characterized as its own “little brain on the mammalian
heart (Armour, 2008, p. 165). The afferent (ascending) nerves
play a critical role in physiological regulation and affect the heart’s
rhythm. Efferent sympathetic and parasympathetic activity are
integrated with the activity occurring in the heart’s intrinsic ner-
vous system, including the afferent signals occurring from the
mechanosensory and chemosensory neurons (Figure 3).
Interestingly, the majority of fibers in the vagus nerve (approx-
imately 85–90%) are afferents, and signals are sent to the brain
via cardiovascular afferents to a greater extent than by any
other major organ (Cameron, 2002). Mechanical and hormonal
information is transduced into neurological impulses by sen-
sory neurons in the heart before being processed in the intrinsic
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |3
Shaffer et al. A healthy heart is not a metronome
FIGURE 3 | The neural communication pathways interacting between
the heart and the brain are responsible for the generation of HRV. The
intrinsic cardiac nervous system integrates information from the extrinsic
nervous system and from the sensory neurites within the heart. The extrinsic
cardiac ganglia located in the thoracic cavity have connections to the lungs
and esophagus and are indirectly connected via the spinal cord to many other
organs such as the skin and arteries. The vagus nerve (parasympathetic)
primarily consists of afferent (flowing to the brain) fibers which connect to the
medulla, after passing through the nodose ganglion. Credit: Institute of
HeartMath.
nervous system. These impulses then travel to the brain via affer-
ent pathways in the spinal column and vagus nerve (McCraty,
2011).
Short-term regulation of BP is accomplished by a complex
network of pressure-sensitive baroreceptors or mechanosensi-
tive neurons which are located throughout the heart and in the
aortic arch. Since BP regulation is a central role of the cardio-
vascular system, the factors that alter BP also affect fluctuations
in HR. Intrinsic cardiac afferent sensory neurons (Figures 4,5)
transduce and distribute mechanical and chemical information
regarding the heart (Cheng et al., 1997) to the intrinsic cardiac
nervous system (Ardell et al., 1991). The afferent impulses from
the mechanosensitive neurons travel via the glossopharyngeal and
vagal nerves to the nucleus of the solitary tract (NST), which
connects with the other regulatory centers in the medulla to mod-
ulate SNS outflow to the heart and the blood vessels. There is
also some modulation of parasympathetic outflow to the heart via
connections to the dorsal vagal complex. Thus, mechanosensitive
neurons affect HR, vasoconstriction, venoconstriction, and car-
diac contractility in order to regulate BP (Hainsworth, 1995). This
input from the heart can also modulate and impact hormonal
release (Randall et al., 1981).
The heart not only functions as an intricate information pro-
cessing and encoding center (Armour and Kember, 2004), but
is also an endocrine gland that can produce and secrete its
FIGURE 4 | Microscopic image of interconnected intrinsic cardiac
ganglia in the human heart. The thin, light blue structures are multiple
axons that connect the ganglia. Credit: Dr. Andrew Armour and the Institute
of HeartMath.
own hormones and neurotransmitters (Cantin and Genest, 1985,
1986; Mukoyama et al., 1991; Huang et al., 1996). For instance,
atrial myocytes can secrete atrial natriuretic peptide (ANP), a hor-
mone that promotes salt and water excretion, to lower BP and
Frontiers in Psychology | Psychology for Clinical Settings September 2014 | Volume 5 | Article 1040 |4
Shafferetal. A healthy heart is not a metronome
FIGURE 5 | This drawing shows the location and distribution of
intrinsic cardiac ganglia which are interconnected and form the “heart
brain. Note how they are distributed around the orifices of the major
vessels. Credit: Dr. Andrew Armour and the Institute of HeartMath.
produce vasodilation (Dietz, 2005). Additionally, intrinsic cardiac
adrenergic cells can synthesize and secrete catecholamines such as
dopamine, NE, and E (Huang et al., 1996) in addition to high
concentrations of oxytocin (Gutkowska et al., 2000).
Research insights from the field of neurocardiology have con-
firmed that the neural interactions between the heart and brain
are more complex than thought in the past. This research has
shown that complex patterns of cardiovascular afferent activity
occur across time scales from milliseconds to minutes (Armour
and Kember, 2004). This work has also shown that the intrin-
sic cardiac nervous system has both short-term and long-term
memory functions, which can influence HRV and afferent activ-
ity related to pressure, rhythm, and rate, as well as afferent activity
associated with hormonal factors (Armour, 2003; Armour and
Kember, 2004; Ardell et al., 2009).
John and Beatrice Lacey conducted heart–brain interaction
studies and were the first to suggest a causal role of the heart in
modulating cognitive functions such as sensory-motor and per-
ceptual performance (Lacey, 1967; Lacey and Lacey, 1970, 1974).
They suggested that cortical functions are modulated via affer-
ent input from pressure-sensitive neurons in the heart, carotid
arteries, and aortic arch (Lacey, 1967). Their research focused
on activity occurring within a single cardiac cycle, and they
confirmed that cardiovascular activity influences perception and
cognitive performance. Research by Velden and Wölk found that
cognitive performance fluctuates at a rhythm around 10 Hz. They
also demonstrated that the modulation of cortical function via
the heart’s influence is due to afferent inputs on the neurons in
the thalamus which globally synchronizes cortical activity (Velden
and Wölk, 1987; Wölk and Velden, 1989). An important aspect of
their work was the finding that it is the “pattern and stability” (the
rhythm) of the heart’s afferent inputs, rather than the number of
neural bursts within the cardiac cycle, that are important in mod-
ulating thalamic activity, which in turn has global effects on brain
function.
This growing body of research indicates that afferent informa-
tion processed by this intrinsic cardiac nervous system (Armour,
1991) can influence activity in the frontocortical areas (Lane et al.,
2001; McCraty et al., 2004) and motor cortex (Svensson and
Thorén, 1979), affecting psychological factors such as attention
level, motivation (Schandry and Montoya, 1996), perceptual sen-
sitivity (Montoya et al., 1993), and emotional processing (Zhang
et al., 1986). Intrinsic cardiac afferent neurons project to nodose
and dorsal root ganglia, the brainstem (dorsal root ganglia first
project to the spinal cord), the hypothalamus, thalamus, or amyg-
dala, and then to the cerebral cortex (Kukanova and Mravec, 2006;
McCraty et al., 2009).
HEARTBEAT EVOKED POTENTIALS
Heartbeat evoked potentials (HEPs) can be used to identify the
specific pathways and influence of afferent input from the heart to
the brain. HEPs are segments of electroencephalogram (EEG) that
are synchronized to the heartbeat. The ECG R-wave is used as a
timing source for signal averaging, resulting in waveforms known
as HEPs. Changes in these evoked potentials associated with the
heart’s afferent neurological input to the brain are detectable
between 50 and 550 ms after each heartbeat. There is a replicable
and complex distribution of HEPs across the scalp. Researchers
can use the location and timing of the various components of
HEP waveforms, as well as changes in their amplitudes and
morphology, to track the flow and timing of cardiovascular affer-
ent information throughout the brain (Schandry and Montoya,
1996).
MacKinnon et al. (2013) reported that HRV influences the
amplitude of heartbeat evoked potentials (HEP N250s). In this
specific context, self-induction of either negative or positive emo-
tion conditions by recalling past events reduced HRV and N250
amplitude. In contrast, resonance frequency breathing (breath-
ing at a rate that maximizes HRV amplitude) increased HRV and
HRV coherence (auto-coherence and sinusoidal pattern) above
baseline and increased N250 amplitude. The authors speculated
that resonance frequency breathing reduces interference with
afferent signal transmission from the heart to the cerebral cortex.
WHAT IS HEART RATE VARIABILITY?
Ever since Walter Cannon introduced the concept of home-
ostasis in 1929, the study of physiology has been based on the
principle that all cells, tissues, and organs maintain a static or
constant “steady-state” condition in their internal environment.
However, with the introduction of signal processing techniques
that can acquire continuous time series data from physiologic
processes such as heart rate, BP, and nerve activity, it has become
abundantly apparent that biological processes vary in a com-
plex and nonlinear way, even during “steady-state” conditions.
These observations have led to the understanding that healthy
physiologic function is a result of continuous, dynamic interac-
tions between multiple neural, hormonal, and mechanical con-
trol systems at both local and central levels. For example, we
now know that the normal resting sinus rhythm of the heart is
highly irregular during steady-state conditions rather than being
monotonously regular, which was the widespread notion for
many years. A healthy heart is not a metronome.
With the ability to measure the ECG in 1895, and the later
development of modern signal processing which first emerged
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |5
Shaffer et al. A healthy heart is not a metronome
in the 1960s and 1970s, the investigation of the heart’s complex
rhythm rapidly exploded. The irregular behavior of the heartbeat
is readily apparent when heart rate is examined on a beat-to-beat
basis, but is overlooked when a mean value over time is calculated.
These fluctuations in heart rate result from complex, non-linear
interactions between a number of different physiological systems
(Reyes Del Paso et al., 2013).
The interactions between autonomic neural activity, BP, and
respiratory control systems produce short-term rhythms in HRV
measurements (Hirsch and Bishop, 1981, 1996; McCraty et al.,
2009)(Figure 6). The most common form for observing these
changes is the heart rate tachogram, a plot of a sequence of time
intervals between R waves. Efferent sympathetic and parasym-
pathetic activity is integrated in and with the activity occurring
in the heart’s intrinsic nervous system, including the afferent
signals occurring from the mechanosensitive and chemosensory
neurons, all of which contribute to beat-to-beat changes. HRV is
thus considered a measure of neurocardiac function that reflects
heart–brain interactions and ANS dynamics.
Circadian rhythms, core body temperature, metabolism, hor-
mones, and intrinsic rhythms generated by the heart all con-
tribute to lower frequency rhythms [e.g., very-low-frequency
(VLF) and ultra-low-frequency (ULF)] that extend below
0.04 Hz. Due to their long time periods, researchers use 24-h HRV
recordings to provide comprehensive assessment of their fluctu-
ations (Kleiger et al., 2005). In concert, these multiple influences
create a dynamic physiological control system that is never truly at
rest and is certainly never static. In healthy individuals, it remains
responsive and resilient, primed and ready to react when needed.
HOW IS HRV DETECTED?
Clinicians use ECG or photoplethysmograph (PPG) sensors to
detect the interbeat interval (IBI). While the ECG method had
been considered to be more accurate than the PPG method
because early software algorithms could more easily detect the
sharp upward spike of the R wave than the curved peak of the
blood volume pulse signal, newer algorithms have improved peak
detection from the pulse wave. The ECG method should be
used when recordings are contaminated by frequent abnormal
beats (e.g., premature ventricular contractions), since the ECG’s
morphology and timing properties allow software algorithms to
discriminate normal sinus beats from ectopic beats (Mateo et al.,
2011).
All HRV assessments are calculated from an IBI file. However,
in some cases there can be differences in the IBI files derived
from ECG and PPG data. Several studies have shown that when
the recordings are taken during a resting state (sitting quietly as
done in most resting baseline recordings), the IBI values between
ECG and PPG are highly correlated (Giardino et al., 2002; Schafer
and Vagedes, 2013). However, during ambulatory monitoring or
when a person experiences a stressor strong enough to activate
the sympathetic system, there can be significant differences due
FIGURE 6 | Display of short-term HRV activity. Credit: Institute of HeartMath.
Frontiers in Psychology | Psychology for Clinical Settings September 2014 | Volume 5 | Article 1040 |6
Shafferetal. A healthy heart is not a metronome
to changes in pulse transit time (the time it takes the BP wave
to propagate from the heart to the periphery), which result from
changes in the elasticity of the arteries. When arteries stiffen due
to sympathetic activation, the BP wave travels faster. The accuracy
of HRV measurements is primarily determined by the sampling
rate of the data acquisition system. Kuusela (2013) recommends
a sampling rate of 200 Hz unless overall variability among RR
intervals is unusually low, as in case of heart failure. In con-
trast, Berntson et al. (2007) recommend a minimum sampling
rate of 500–1000 Hz. However, for many applications, like HRV
biofeedback (HRVB), a sampling rate of 126Hz may be adequate.
There are many ECG configurations, with varying numbers of
leads, used for ambulatory and stationary monitoring. For exam-
ple, a standard three-lead ECG chest placement locates active
and reference electrodes over the right and left coracoid pro-
cesses, respectively, and a second active electrode over the xiphoid
process (Figure 7).
WHY IS HRV IMPORTANT?
An optimal level of variability within an organism’s key regulatory
systems is critical to the inherent flexibility and adaptability or
resilience that epitomizes healthy function and well-being. While
too much instability is detrimental to efficient physiological
FIGURE 7 | ECG electrode placement. Credit: Truman State University
Center for Applied Psychophysiology.
functioning and energy utilization, too little variation indicates
depletion or pathology.
HRV IS A MARKER FOR DISEASE AND ADAPTABILITY
The clinical importance of HRV was noted as far back as 1965
when it was found that fetal distress is preceded by alterations
in HRV before any changes occur in heart rate itself (Hon and
Lee, 1963). In the 1970s, HRV analysis was shown to predict auto-
nomic neuropathy in diabetic patients before the onset of symp-
toms (Ewing et al., 1976).LowHRVhassincebeenconrmedas
a strong, independent predictor of future health problems and as
a correlate of all-cause mortality (Tsuji et al., 1994; Dekker et al.,
1997). Reduced HRV is also observed in patients with autonomic
dysfunction, including anxiety, depression, asthma, and sudden
infant death (Kazuma et al., 1997; Carney et al., 2001; Agelink
et al., 2002; Giardino et al., 2004; Lehrer et al., 2004; Cohen and
Benjamin, 2006).
Based on indirect evidence, reduced HRV may correlate with
disease and mortality because it reflects reduced regulatory capac-
ity, which is the ability to adaptively respond to challenges like
exercise or stressors. For example, patients with low overall
HRV demonstrated reduced cardiac regulatory capacity and an
increased likelihood of prior myocardial infarction (MI). In this
sample, a measure of cardiac autonomic balance did not predict
previous MIs (Berntson et al., 2008).
Patient age may mediate the relationship between reduced
HRV and regulatory capacity. HRV declines with age (Umetani
et al., 1998) and aging often involves nervous system changes, like
loss of neurons in the brain and spinal cord, which may degrade
signal transmission (Jäncke et al., 2014) and reduce regulatory
capacity.
Reduced regulatory capacity may contribute to functional gas-
trointestinal disorders, inflammation, and hypertension. While
patients with functional gastrointestinal disorders often have
reduced HRV (Gevirtz, 2013), HRVB has increased vagal tone and
improved symptom ratings in these patients (Sowder et al., 2010).
The PNS may help regulate inflammatory responses via a
cholinergic anti-inflammatory system (Tracey, 2007). While the
experimental administration of lipopolysaccharide to healthy vol-
unteers decreases HRV and vagal tone (Jan et al., 2009), HRVB
training has reduced the symptoms produced by this intervention
(Lehrer et al., 2010).
Hypertensive patients often present with reduced baroreflexes
and HRV (Schroeder et al., 2003). HRVB can increase barore-
flex gain, which is the amplitude of HR changes, and HRV, and
decrease BP (Lehrer, 2013). Several randomized-controlled stud-
ies have documented BP reductions in hypertensive patients who
received HRVB (Elliot et al., 2004; Reineke, 2008).
HRV is also an indicator of psychological resiliency and behav-
ioral flexibility, reflecting the individual’s capacity to adapt effec-
tively to changing social or environmental demands (Beauchaine,
2001; Berntson et al., 2008). More recently, several studies have
shown an association between higher levels of resting HRV and
performance on cognitive performance tasks requiring the use of
executive functions (Thayer et al., 2009) and that HRV, especially
HRV-coherence, can be increased in order to produce improve-
ments in cognitive function as well as a wide range of clinical
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |7
Shaffer et al. A healthy heart is not a metronome
outcomes, including reduced health care costs (Lehrer et al., 2003,
2008; McCraty et al., 2003; Bedell and Kaszkin-Bettag, 2010;
Alabdulgader, 2012).
HRV ANALYSIS METHODS
It was recognized as far back as 1979 that nomencla-
ture, analytical methods, and definitions of HRV measures
required standardization. Therefore, an International Task
Force consisting of members from the European Society of
Cardiology and the North American Society for Pacing and
Electrophysiology was established. Their report was published in
Task Force (1996).
HRV can be assessed with various analytical approaches,
although the most commonly used are frequency domain or
power spectral density (PSD) analysis and time domain analysis.
In both methods, the time intervals between each successive nor-
mal QRS complex are first determined. All abnormal beats not
generated by sinus node depolarizations are eliminated from the
record.
Analogous to the EEG, we can use power spectral analy-
sis to separate HRV into its component rhythms that operate
within different frequency ranges (Figure 8). PSD analysis pro-
vides information of how power is distributed (the variance and
amplitude of a given rhythm) as a function of frequency (the
time period of a given rhythm). The main advantages of spectral
analysis over the time domain measures are that it supplies both
frequency and amplitude information about the specific rhythms
that exist in the HRV waveform, providing a means to quantify
the various oscillations over any given period in the HRV record-
ing. The values are expressed as the PSD, which is the area under
the curve (peak) in a given segment of the spectrum. The power
or height of the peak at any given frequency indicates the ampli-
tude and stability of the rhythm. The frequency reflects the period
of time over which the rhythm occurs. For example, a 0.1Hz fre-
quency has a period of 10 s. In order to understand how power
spectral analysis distinguishes the various underlying physiolog-
ical mechanisms that are reflected in the heart’s rhythm, a brief
review of these underlying physiological mechanisms follows.
FIGURE 8 | This figure shows a typical HRV recording over a 15-min
period during resting conditions in a healthy individual. The top trace
shows the o riginal HRV waveform. Filtering techniques were used to
separate the original waveform into VLF, LF, and HF bands as shown in the
lower traces. The bottom of the figure shows the power spectra (left) and the
percentage of power (right) in each band. Credit: Institute of HeartMath.
Frontiers in Psychology | Psychology for Clinical Settings September 2014 | Volume 5 | Article 1040 |8
Shafferetal. A healthy heart is not a metronome
Figure 8 shows a typical example of an HRV recoding from
an adult human at rest. Using filtering techniques, the high-
frequency (HF), low-frequency (LF), and VLF bands have been
extracted from the original HRV signal and spectral power has
been calculated for each band.
SOURCES OF HRV
The Task Force report (1996) divided heart rhythm oscillations
into four primary frequency bands. These included the HF, LF,
VLF, and ULF bands. The Task Force report also stated that
the analysis should be done on 5-min segments, although other
recording periods are often used. When other recording lengths
are analyzed and reported, the length of the recording should be
reported since this has large effects on both HRV frequency and
time domain values.
HIGH-FREQUENCY BAND
The HF spectrum is the power in each of the 288 5-min seg-
ments (monitored during a 24-h period) in the range from 0.15
to 0.4 Hz. This band reflects parasympathetic or vagal activity
and is frequently called the respiratory band because it corre-
sponds to the HR variations related to the respiratory cycle. These
HR changes are known as respiratory sinus arrhythmia (RSA).
Heart rate accelerates during inspiration and slows during expi-
ration. During inhalation, the cardiovascular center inhibits vagal
outflow resulting in speeding the heart rate. Conversely, during
exhalation, it restores vagal outflow resulting in slowing the heart
rate via the release of acetylcholine (Eckberg and Eckberg, 1982).
The magnitude of the oscillation is variable, but can usually be
exaggerated by slow, deep breathing.
The modulation of vagal tone helps maintain the dynamic
autonomic regulation important for cardiovascular health.
Deficient vagal inhibition is implicated in increased morbidity
(Thayer et al., 2010). The mechanism linking the variability of
HR to respiration is complex and involves both central and reflex
interactions. A large number of studies have shown that total
vagal blockade essentially eliminates HF oscillations and reduces
the power in the LF range (Pomeranz et al., 1985; Malliani et al.,
1991).
Reduced parasympathetic (high frequency) activity has been
found in a number of cardiac pathologies and in patients
under stress or suffering from panic, anxiety, or worry. Lowered
parasympathetic activity may primarily account for reduced HRV
in aging (Umetani et al., 1998). In younger healthy individuals,
it is not uncommon to see an obvious increase in the HF band
at night with a decrease during the day (Lombardi et al., 1996;
Otsuka et al., 1997).
LOW-FREQUENCY BAND
The LF band ranges between 0.04 and 0.15 Hz. This region was
previously called the “baroreceptor range” or “mid-frequency
band” by many researchers, since it primarily reflects barore-
ceptoractivitywhileatrest(Malliani, 1995). The vagus nerves
are a major conduit though which afferent neurological sig-
nals from the heart and other visceral organs are relayed to
the brain, including the baroreflex signals (De Lartique, 2014).
Baroreceptors are stretch-sensitive mechanoreceptors located in
the chambers of the heart and vena cavae, carotid sinuses (which
contain the most sensitive mechanoreceptors), and the aortic arch
(Figure 9). When BP rises, the carotid and aortic tissues are dis-
tended, resulting in increased stretch and, therefore, increased
baroreceptor activation. At normal resting BPs, many barorecep-
tors actively report BP information and the baroreflex modulates
autonomic activity.
Active baroreceptors generate action potentials (“spikes”)
more frequently. The greater their stretch or detection of an
increased rate of change, the more frequently baroreceptors fire
action potentials. Baroreceptor action potentials are relayed to the
NST in the medulla, which uses baroreceptor firing frequency to
measure BP. Increased activation of the NST inhibits the vaso-
motor center and stimulates the vagal nuclei. The end-result of
baroreceptor activations tuned to pressure increases is inhibition
of the SNS and activation of the PNS. By coupling sympathetic
inhibition with parasympathetic activation, the baroreflex maxi-
mizes BP reduction when BP is detected as too high. Sympathetic
inhibition reduces peripheral resistance, while parasympathetic
activation depresses HR (reflex bradycardia) and contractility. In
a similar manner, sympathetic activation, along with inhibition of
vagal outflow, allows the baroreflex to elevate BP. Baroreflex gain
is commonly calculated as the beat-to-beat change in HR per unit
of change in BP. Decreased baroreflex gain is related to impaired
regulatory capacity and aging.
The existence of a cardiovascular system resonance frequency,
which is caused by the delay in the feedback loops in the barore-
flex system, has been long established (Vaschillo et al., 2011).
Lehrer et al. have proposed that each individual’s cardiovascular
system has a unique resonance frequency, which can be iden-
tified by measuring HRV while an individual breathes between
7.5 and 4.5 breaths per minute (Lehrer et al., 2013). When the
FIGURE 9 | Credit: Alila Sao Mai/Shutterstock.com.
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |9
Shaffer et al. A healthy heart is not a metronome
cardiovascular system oscillates at this frequency, there is a dis-
tinctive high-amplitude peak in the HRV power spectrum around
0.1 Hz. Most mathematical models show that the resonance fre-
quency of the human cardiovascular system is determined by the
feedback loops between the heart and brain (deBoer et al., 1987;
Baselli et al., 1994). In humans and many other mammals, the
resonance frequency of the system is approximately 0.1 Hz, which
is equivalent to a 10-s rhythm.
The sympathetic system does not appear to produce rhythms
much above 0.1 Hz, while the parasympathetic system can be
observed to affect heart rhythms down to 0.05 Hz (20-s rhythm).
During periods of slow respiration rates, vagal activity can eas-
ily generate oscillations in the heart rhythms that cross over into
the LF band (Ahmed et al., 1982; Tiller et al., 1996; Lehrer et al.,
2003). Therefore, respiratory-related efferent vagally-mediated
influences are particularly present in the LF band when respira-
tionratesarebelow8.5breathsperminuteor7-speriods(Brown
et al., 1993; Tiller et al., 1996)orwhenanindividualsighsortakes
a deep breath.
In ambulatory 24-h HRV recordings, it has been suggested that
the LF band also reflects sympathetic activity and the LF/HF ratio
has been controversially reported as an assessment of the balance
between sympathetic and parasympathetic activity (Pagani et al.,
1984, 1986).Anumberofresearchers(Tiller et al., 1996; Eckberg,
1997; Porges, 2007; Rahman et al., 2011; Heathers, 2012)have
challenged this perspective and have persuasively argued that in
resting conditions, the LF band reflects baroreflex activity and not
cardiac sympathetic innervation.
The perspective that the LF band reflects sympathetic activ-
ity came from observations of 24-h ambulatory recordings where
there are frequent sympathetic activations primarily due to phys-
ical activity, but also due to emotional stress reactions, which
can create oscillations in the heart rhythms that cross over into
the lower part of the LF band. In long-term ambulatory record-
ings, the LF band fairly approximates sympathetic activity when
increased sympathetic activity occurs (Axelrod et al., 1987). This
willbediscussedinmoredetailintheVLFsection.Unfortunately,
some authors have assumed that this interpretation was also
true of short-term resting recordings and have confused slower
breathing-related increases in LF power with sympathetic activity,
when in reality it is almost entirely vagally mediated. Remember
that the baroreflex is primarily vagally mediated (Keyl et al.,
1985).
Porges (2007) suggests that under conditions when partici-
pants pace their breathing at 0.1 Hz (10-s rhythm or 6 breaths
per minute), which is a component of many HRVB training
protocols, the LF band includes the summed influence of both
efferent vagal pathways (myelinated and unmyelinated, which
reflects total cardiac vagal tone).
AUTONOMIC BALANCE AND THE LF/HF RATIO
The autonomic balance hypothesis assumes that the SNS and
PNS competitively regulate SA node firing, where increased SNS
activity is paired with decreased PNS activity. While some ortho-
static challenges can produce reciprocal changes in SNS activation
and vagal withdrawal, psychological stressors can also result in
independent changes in SNS or PNS activity. It is now generally
accepted that both branches of the ANS can be simultaneously
active (Berntson and Cacioppo, 1999). Therefore, the relation-
ship between the SNS and PNS in generating LF power appears
to be complex, non-linear, and dependent upon the experimental
manipulation employed (Berntson et al., 1997; Billman, 2013).
The ratio of LF to HF power is called the LF/HF ratio. The
interpretation of the LF/HF ratio is controversial due to the issues
regardingtheLFbanddescribedabove.However,oncethemech-
anisms are understood as well as the importance of the recording
context (i.e., ambulatory vs. resting conditions and normal vs.
paced breathing), the controversy is resolved. Recall that the
power in the LF band can be influenced by vagal, sympathetic,
and baroreflex mechanisms depending on the context, whereas
HF power is produced by the efferent vagal activity due to respi-
ratoryactivity.ItisoftenassumedthatalowLF/HFratioreects
greater parasympathetic activity relative to sympathetic activity
due to energy conservation and engaging in “tend-and-befriend”
behaviors (Taylor, 2006). However, this ratio is often shifted due
to reductions in LF power. Therefore, the LF/HR ratio should
be interpreted with caution and the mean values of HF and LF
power taken into consideration. In contrast, a high LF/HF ratio
may indicate higher sympathetic activity relative to parasympa-
thetic activity as can be observed when people engage in meeting a
challenge that requires effort and increased SNS activation. Again,
the same cautions must be taken into consideration, especially in
short-term recordings.
VERY-LOW-FREQUENCY BAND
The VLF band is the power in the HRV power spectrum range
between 0.0033 and 0.04 Hz. Although all 24-h clinical measures
of HRV reflecting low HRV are linked with increased risk of
adverse outcomes, the VLF band has stronger associations with
all-cause mortality than the LF and HF bands (Tsuji et al., 1994,
1996; Hadase et al., 2004; Schmidt et al., 2005). Low VLF power
has been shown to be associated with arrhythmic death (Bigger
et al., 1992) and PTSD (Shah et al., 2013). Additionally, low power
in this band has been associated with high inflammation in a
number of studies (Carney et al., 2007; Lampert et al., 2008)and
has been correlated with low levels of testosterone, while other
biochemical markers, such as those mediated by the HPA axis
(e.g., cortisol), did not (Theorell et al., 2007).
Historically, the physiological explanation and mechanisms
involved in the generation of the VLF component have not
been as well defined as the LF and HF components, and this
region has been largely ignored. Long-term regulation mecha-
nisms and ANS activity related to thermoregulation, the renin-
angiotensin system, and other hormonal factors may contribute
to this band (Akselrod et al., 1981; Cerutti et al., 1995; Claydon
and Krassioukov, 2008). Recent work by Dr. Andrew Armour has
shed new light on the mechanisms underlying the VLF rhythm
and suggests that we may have to reconsider both the mechanisms
and importance of this band.
Dr. Armour’s group has developed the technology to obtain
long-term single-neuron recordings from a beating heart, and
simultaneously, from extrinsic cardiac neurons (Armour, 2003).
Figure 10 shows the VLF rhythm obtained from an afferent neu-
ron located in the intrinsic cardiac nervous system in a dog heart.
Frontiers in Psychology | Psychology for Clinical Settings September 2014 | Volume 5 | Article 1040 |10
Shafferetal. A healthy heart is not a metronome
FIGURE 10 | Long-term single-neuron recordings from an afferent neuron
in the intrinsic cardiac nervous system in a beating dog heart. The top row
shows neural activity, the second row, the actual neural recording, and the third
row, the left ventricular pressure. This intrinsic rhythm has an average period of
90 s with a range between 75 and 100 s (0.013–0.01 Hz), which falls within the
VLF band. Credit: Dr. Andrew Armour and the Institute of HeartMath.
In this case, the VLF rhythm is generated from intrinsic sources
and cannot be explained by sources such as movement. The black
area in the bottom of the figure labeled “rapid ventricular pac-
ing” shows the time period where efferent spinal neurons were
stimulated. The resulting increase in efferent sympathetic activity
(bottom row) clearly elevates the amplitude of the single afferent
neuron’s intrinsic VLF rhythm (top row).
This work, combined with findings by Kember et al. (2000,
2001), implies that the VLF rhythm is generated by the stim-
ulation of afferent sensory neurons in the heart, which in turn
activate various levels of the feedback and feed-forward loops in
the heart’s intrinsic cardiac nervous system, as well as between
the heart, the extrinsic cardiac ganglia, and spinal column. Thus,
the VLF rhythm is produced by the heart itself and is an intrinsic
rhythm that appears to be fundamental to health and well-
being. Dr. Armour has observed that when the amplitude of the
VLF rhythm at the neural level is diminished, an animal sub-
ject is in danger and will expire shortly if they proceed with the
research procedures (personal communication with McCraty).
Sympathetic blockade does not affect VLF power and VLF activ-
ity is seen in tetraplegics, whose SNS innervation of the heart
and lungs is disrupted (Task Force, 1996; Berntson et al., 1997).
These findings further support a cardiac origin of the VLF
rhythm.
In healthy individuals, there is an increase in VLF power that
occurs during the night and peaks before waking (Huikuri et al.,
1994; Singh et al., 2003). This increase in autonomic activity may
correlate with the morning cortisol peak.
In summary, experimental evidence suggests that the VLF
rhythm is intrinsically generated by the heart and that the
amplitude and frequency of these oscillations are modulated by
efferent sympathetic activity. Normal VLF power appears to indi-
cate healthy function, and increases in resting VLF power may
reflect increased sympathetic activity. The modulation of the fre-
quency of this rhythm due to physical activity (Bernardi et al.,
1996), stress responses, and other factors that increase efferent
sympathetic activation can cause it to cross over into the lower
region of the LF band during ambulatory monitoring or during
short-term recordings when there is a significant stressor.
ULTRA-LOW-FREQUENCY BAND
The ULF band falls below 0.0033 Hz (333 s or 5.6 min).
Oscillations or events in the heart rhythm with a period of 5 min
or greater are reflected in this band and it can only be assessed
with 24-h and longer recordings (Kleiger et al., 2005). The cir-
cadian oscillation in heart rate is the primary source of the ULF
power, although other very slow-acting regulatory processes, such
as core body temperature regulation, metabolism, and the renin-
angiotensin system likely add to the power in this band (Bonaduce
et al., 1994; Task Force, 1996). Different psychiatric disorders
show distinct circadian patterns in 24-h heart rates, particularly
during sleep (Stampfer, 1998; Stampfer and Dimmitt, 2013).
The Task Force report (1996) stated that analysis of 24-h
recordings should divide the record into 5-min segments and that
HRV analysis should be performed on the individual segments
prior to the calculation of mean values. This effectively filters
out any oscillations with periods longer than 5 min. However, as
shown in Figure 11, when spectral analysis is applied to entire 24-
h records, several lower frequency rhythms are easily detected in
healthy individuals. At the present time, the clinical relevance of
these lower frequency rhythms is unknown, largely due to the
Task Force guidelines that eliminate their presence from most
analysis procedures.
TIME DOMAIN MEASUREMENTS OF HRV
Time domain measures are the simplest to calculate and include
the mean normal-to-normal (NN) intervals during the entire
recording and other statistical measures such as the stan-
dard deviation between NN intervals (SDNN). However, time
domain measures do not provide a means to adequately quantify
autonomic dynamics or determine the rhythmic or oscillatory
activity generated by the different physiological control systems.
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |11
Shaffer et al. A healthy heart is not a metronome
FIGURE 11 | This figure shows the power in the various frequency
bands for 24-h HRV and 95% confidence intervals for each of
the bands. The left side of the figure reveals a number of slower
rhythms that make up the ULF band. The analysis was conducted
using the healthy sample described in Umetani et al. (1998).The
right side of the figure shows an analysis of the same data
performed on 5-min segments as is traditionally done. Credit:
Institute of HeartMath.
Since they are always calculated the same way, data collected by
different researchers are comparable, but only if the recording
lengths are exactly the same and the data are collected under the
same conditions.
Time domain indices quantify the amount of variance in the
IBI using statistical measures. For 24-h recordings, the three most
important time domain measures are the SDNN, the SDNN
index, and the RMSSD. For short-term assessments, the SDNN,
RMSSD, pNN50, and HR Max HR Min are most commonly
reported.
SDNN
The SDNN is the standard deviation of the normal (NN) sinus-
initiated IBI measured in milliseconds. This measure reflects the
ebb and flow of all the factors that contribute to heart rate vari-
ability (HRV). In 24-h recordings, the SDNN is highly correlated
with ULF and total power (Umetani et al., 1998). In short-
term resting recordings, the primary source of the variation is
parasympathetically-mediated RSA, especially with slow, paced
breathing protocols.
SDNN values are highly correlated with the lower frequency
rhythms discussed earlier (Tabl e 1 ). Low age-adjusted values
predict both morbidity and mortality. Classification within a
higher SDNN category is associated with a higher probability of
survival. For example, patients with moderate SDNN values, 50–
100 ms, have a 400% lower risk of mortality than those with low
values, 0–50 ms, in 24-h recordings (Kleiger et al., 1987).
SDANN
The SDANN is the standard deviation of the average NN intervals
(mean heart rate) for each of the 5-min segments during a 24-h
recording. Like the SDNN, it is measured and reported in mil-
liseconds. This index is correlated with the SDNN and is generally
considered redundant.
SDNN INDEX
The SDNN index is the mean of the standard deviations of all the
NN intervals for each 5-min segment of a 24-h HRV recording.
Therefore, this measurement only estimates variability due to the
factors affecting HRV within a 5-min period. It is calculated by
first dividing the 24-h record into 288 5-min segments and then
calculating the standard deviation of all NN intervals contained
within each segment. The SDNN Index is the average of these
288 values. The SDNN index is believed to primarily measure
Frontiers in Psychology | Psychology for Clinical Settings September 2014 | Volume 5 | Article 1040 |12
Shafferetal. A healthy heart is not a metronome
Table 1 | Correlations between time and frequency domain measures in 24-h recordings.
HR (ms) N-D SDNN Ln total Ln ULF SDANN SDNN Ln 5-min Ln 5-min Ln 5-min Ln 5-min Ln Ln
delta power index total power VLF LF HF RMSSD LF/HF
HR (ms) 1
N-D delta 0.29 1
SDNN 0.61 0.66 1
Ln total power 0.55 0.66 0.98 1
Ln ULF 0.47 0.67 0.95 0.99 1
SDANN 0.47 0.70 0.96 0.97 0.98 1
SDNN index 0.72 0.43 0.79 0.73 0.62 0.62 1
Ln 5-min total power 0.71 0.40 0.78 0.71 0.60 0.61 0.99 1
Ln 5-min VLF 0.74 0 .49 0.83 0.80 0.70 0.68 0.96 0.93 1
Ln 5-min LF 0.57 0.27 0.63 0.61 0.49 0.48 0.87 0.84 0.81 1
Ln 5-min HF 0.36 0.38 0.56 0.54 0.44 0.44 0.79 0.75 0.68 0.75 1
Ln RMSSD 0.54 0.41 0.68 0.64 0.54 0.54 0.90 0.86 0.80 0.82 0.95 1
Ln LF/HF 0.02 0.31 0.27 0.24 0.20 0.21 0.37 0.34 0.27 0.20 0.80 0.66 0.20
Credit: Institute Of Heartmath.
autonomic influence on HRV. This measure tends to correlate
with VLF power over a 24-h period.
RMSSD
TheRMSSDistherootmeansquareofsuccessivedifferences
between normal heartbeats. This value is obtained by first cal-
culating each successive time difference between heartbeats in
milliseconds. Then, each of the values is squared and the result
is averaged before the square root of the total is obtained. The
RMSSD reflects the beat-to-beat variance in heart rate and is
the primary time domain measure used to estimate the vagally-
mediated changes reflected in HRV. While the RMSSD is cor-
related with HF power (Kleiger et al., 2005), the influence of
respiration rate on this index is uncertain (Schipke et al., 1999;
Pentillä et al., 2001). Lower RMSSD values are correlated with
higher scores on a risk inventory of sudden unexplained death
in epilepsy (DeGiorgio et al., 2010).
pNN50
The pNN50 is the percentage of adjacent NN intervals that dif-
fer from each other by more than 50 ms. It is correlated with the
RMSSD and HF power. However, the RMSSD typically provides
a better assessment of RSA (especially in older subjects) and most
researchers prefer it to the pNN50 (Otzenberger et al., 1998).
HR MAX HR MIN
HR Max HR Min is the average difference between the high-
est and lowest HRs during each respiratory cycle. This measure is
especially used for assessment in paced breathing protocols and is
highly correlated with the SDNN and RMSSD.
POLYVAGAL THEORY
As previously discussed, increased efferent activity in the vagal
nerves (also called the 10th cranial nerve) slows the heart
rate, yet has an opposite effect in the lungs as it increases
bronchial tone. According to Porges (2011) polyvagal theory,
the ANS must be considered a “system, with the vagal nerves
containing specialized subsystems that regulate competing adap-
tive responses. His theory proposes competing roles for the
unmyelinated fibers in the vagus, which originate in the dorsal
motor complex, and newer myelinated nerves, which originate
in the nucleus ambiguus. He hypothesizes that the unmyelinated
fibers are involved in regulating the “freeze response” and respond
to threats through immobilization, feigning death, passive avoid-
ance, and shutdown (the freeze response).
In Porges’ view, the evolution of the ANS was central to
the development of emotional experience and affective processes
central to social behavior. As human beings, we are not lim-
ited to fight, flight, or freezing behavioral responses. We can
self-regulate and initiate pro-social behaviors (e.g., the tend-and-
befriend response) when we encounter stressors. Porges calls this
the social engagement system and the theory suggests that this
system depends upon the healthy functioning of the myelinated
vagus, a vagal brake, which allows for self-regulation and ability to
calm ourselves and inhibit sympathetic outflow to the heart. This
implies that standardized assessment of vagal tone could serve as
a potential marker for one’s ability to self-regulate.
The theory suggests that the evolution and healthy function
of the ANS sets the limits or boundaries for the range of one’s
emotional expression, quality of communication, and ability to
self-regulate emotions and behaviors. The theory describes the
details of the anatomical connections from higher brain struc-
tures with the centers involved in autonomic regulation and
argues that the afferent systems are an important aspect of the
ANS. The theory provides insights into the adaptive nature of
physiological states and suggests these states support different
types or classes of behavior (Porges, 2011).
The SNS, in concert with the endocrine system, responds to
threats to our safety through mobilization, fight-or-flight, and
active avoidance. The SNS responds more slowly and for a longer
period of time (i.e., more than a few seconds) than the vagus
system. According to this theory, quality communication and
pro-social behaviors can only be effectively engaged when these
defensive circuits are inhibited.
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |13
Shaffer et al. A healthy heart is not a metronome
NEUROVISCERAL INTEGRATION: THE CENTRAL AUTONOMIC
NETWORK MODEL
Thayer and Lane (2000) outline a neurovisceral integration model
that describes how a set of neural structures involved in cog-
nitive, affective, and autonomic regulation are related to HRV
and cognitive performance. In this complex systems model, the
anatomical details of a central autonomic network (CAN) are
described that link the NST in the brainstem with forebrain
structures (including the anterior cingulate, insula, ventromedial
prefrontal cortex, amygdala, and hypothalamus) through feed-
back and feed-forward loops. They propose that this network
is an integrated system for internal system regulation by which
the brain controls visceromotor, neuroendocrine, and behavioral
responses that are critical for goal-directed behavior, adaptability,
and health.
Thayer et al. (2012) contend that dynamic connections
between the amygdala and medial prefrontal cortex, which eval-
uate threat and safety, help regulate HRV through their con-
nections with the NST. They propose that vagally-mediated
HRV is linked to higher-level executive functions and that
HRV reflects the functional capacity of the brain structures
that support working memory and emotional and physiologi-
cal self-regulation. They hypothesize that vagally-mediated HRV
is positively correlated with prefrontal cortical performance
and the ability to inhibit unwanted memories and intrusive
thoughts. In their model, when the CAN decreases prefrontal
cortical activation, HR increases and HRV decreases. The pre-
frontal cortex can be taken offline” when individuals perceive
that they are threatened. Prolonged prefrontal cortical inactiv-
ity can lead to hypervigilance, defensiveness, and social isolation
(Thayer et al., 2009).
The CAN model predicts reduced HRV and vagal activity
in anxiety. Friedman (2007) argues that anxiety is associated
with abnormal ANS cardiac control. HRV indices consistently
show low vagal activity in patients diagnosed with anxiety dis-
orders. This finding challenges the completeness of the sympa-
thetic overactivation explanation of anxiety. Friedman observes
that “metaphorically, investigators were searching for a ‘sticky
accelerator’ while overlooking the possibility of ‘bad brakes’”
(p. 186). From his perspective, anxiety disorders can involve vary-
ing degrees of sympathetic overactivation and parasympathetic
underactivation.
THE PSYCHOPHYSIOLOGICAL COHERENCE MODEL
McCraty and Childre (2010) at the Institute of HeartMath also
take a dynamic systems approach that focuses on increasing
individuals’ self-regulatory capacity by inducing a physiologi-
cal shift that is reflected in the heart’s rhythms. They theorize
that rhythmic activity in living systems reflects the regulation of
interconnected biological, social, and environmental networks.
The coherence model also suggests that information is encoded
in the dynamic patterns of physiological activity. For example,
information is encoded in the time interval between action poten-
tials and patterns in the pulsatile release of hormones. They
suggest that the time intervals between heartbeats (HRV) also
encode information which is communicated across multiple sys-
tems, which helps synchronize the system as whole. The afferent
pathways from the heart and cardiovascular system are given
more relevance in this model due the significant degree of afferent
cardiovascular input to the brain and the consistent generation
of dynamic patterns generated by the heart. It is their thesis
that positive emotion in general, as well as self-induced posi-
tive emotions, shift the system as a whole into a more globally
coherent and harmonious physiological mode associated with
improved system performance, ability to self-regulate, and overall
well-being.
They use the term “physiological coherence” to describe
the orderly and stable rhythms generated by living systems.
Physiological coherence is used broadly and includes all of the
specific approaches for quantifying the various types of coher-
ence measures, such as cross-coherence (frequency entrainment
between respiration, BP, and heart rhythms), or synchroniza-
tion among systems (e.g., synchronization between various EEG
rhythms and the cardiac cycle), auto-coherence (stability of a sin-
gle waveform such as respiration or HRV patterns), and system
resonance.
A coherent heart rhythm is defined as a relatively harmonic
(sine-wave-like) signal with a very narrow, high-amplitude peak
in the LF region of the HRV power spectrum with no major
peaks in the VLF or HF regions. Coherence is assessed by iden-
tifying the maximum peak in the 0.04–0.26 Hz range of the HRV
power spectrum, calculating the integral in a window 0.030 Hz
wide, centered on the highest peak in that region, and then
calculating the total power of the entire spectrum. The coher-
ence ratio is formulated as: (Peak Power/[Total Power Peak
Power])” (14).
THE HEART RHYTHM COHERENCE HYPOTHESIS
As discussed above, neurocardiology research has established
that heart-brain interactions are remarkably complex. Patterns of
baroreceptor afferent activity modulate CNS activity over time
periods that range from milliseconds to minutes; that is, not
only within a cardiac cycle (Armour and Kember, 2004). The
intrinsic cardiac ganglia demonstrate both short- and long-term
memory. This affects afferent activity rhythms produced by both
mechanical variables (e.g., pressure and HR) that occur over mil-
liseconds (single cycles) and hormonal variables that fluctuate
over periods ranging from seconds to minutes (Armour, 2003;
Armour and Kember, 2004; Ardell et al., 2009). McCraty pro-
posed the heart rhythm coherence hypothesis which states that
the pattern and stability of beat-to-beat heart rate activity encode
information over “macroscopic time scales, which can impact
cognitive performance and emotional experience. For a more
detailed discussion, see McCraty et al. (2009).
INCREASING VAGAL AFFERENT TRAFFIC
Mechanosensitive neurons (baroreceptors) typically increase
their firing rates when the rate of change in the function to which
they are tuned increases. Heart rhythm coherence, which is char-
acterized by increased beat-to-beat variability and the rate of
heart rate change, increases vagal afferent traffic from the car-
diovascular system to the brain. This perspective is supported
by the MacKinnon et al. (2013) HEP study, discussed earlier,
which showed that resonance frequency breathing increased the
Frontiers in Psychology | Psychology for Clinical Settings September 2014 | Volume 5 | Article 1040 |14
Shafferetal. A healthy heart is not a metronome
amount of HRV, HRV coherence, and N250 amplitude in the
HEPs. The authors speculated that resonance frequency breathing
may have increased vagal afferent traffic and reduced interference
with its transmission through subcortical areas to the cerebral
cortex.
There has been increasing interest in treating a wide range of
disorders with implanted pacemaker-like devices for stimulating
the vagal afferent pathways. The FDA has approved these devices
for the treatment of epilepsy and depression, and they have
been investigated in treating obesity, anxiety, and Alzheimer’s
disease (Kosel and Schlaepfer, 2003; Groves and Brown, 2005).
Neuroradiology research has established that increases in tonic
vagal afferent traffic inhibit thalamic pain pathways traveling
from the body to the brain at the level of the spinal cord. This
finding may explain why studies have shown vagal afferent stim-
ulation can reduce cluster and migraine headaches (Mauskop,
2005) and HRV coherence training reduces chronic pain (Berry
et al., 2014).
RESONANCE FREQUENCY BREATHING
Lehrer et al.s resonance frequency model proposes that the delay
in the baroreflex system’s feedback loops creates each individ-
ual’s unique cardiovascular system resonance frequency (Lehrer,
2013). While their theoretical model assumes that taller indi-
viduals and men have lower resonance frequencies than women
and shorter individuals due to the former’s larger blood vol-
umes, height only accounts for 30% of the variance in reso-
nance frequency. Breathing, rhythmic muscle tension, and emo-
tional stimulation at a person’s resonance frequency can activate
or stimulate the cardiovascular system’s resonance properties
(Lehrer et al., 2009).
They suggest that when people breathe at this rate, which
varies in adults from 4.5 to 6.5 breaths per minute, they “exercise”
the baroreflex. They have shown that during this paced period,
HR and BP oscillations are 180out of phase, and HRV amplitude
is maximized (deBoer et al., 1987; Vaschillo et al., 2002). They
also suggest that this phase relationship between HR, respiration,
and BP results in the most efficient gas exchange and oxygen sat-
uration (Bernardi et al., 2001; Vaschillo et al., 2004; Yasuma and
Hayano, 2004).
With practice, people can learn to breathe at their cardio-
vascular system’s resonance frequency. This aligns the three
oscillators (baroreflex, HR, and BP) at that frequency and moves
the peak frequency from the HF range (0.2 Hz) to the LF range
(0.1 Hz). Breathing at the resonance frequency more than dou-
bles the energy in the LF band (0.04–0.15 Hz). This corresponds
to the Institute of HeartMath’s heart rhythm coherence, which
is associated with a “narrow, high-amplitude, easily visualized
peak” from 0.09 to 0.14 Hz (McCraty et al., 2009; Ginsberg et al.,
2010, p. 54).
Resonance frequency breathing is typically used in the context
of HRVB training. Several months of steady practice can reset the
baroreflex gain so that it is sustained, even when clients are not
receiving feedback (Lehrer et al., 2003; Lehrer, 2013). Increased
baroreflex gain is analogous to a more sensitive thermostat, allow-
ing the body to regulate BP and gas exchange more effectively
(Lehrer, 2007).
AN INTEGRATIVE PERSPECTIVE
There has been a paradigm shift in the medical treatment
of diverse disorders like depression, epilepsy, and pain using
vagal nerve stimulation (Kosel and Schlaepfer, 2003; Groves and
Brown, 2005; Mauskop, 2005). Instead of exclusively targeting
sympathetic activation, physicians also attempt to increase vagal
tone. Behavioral interventions like HRVB and emotional self-
regulation strategies represent non-invasive methods of restoring
homeostasis.
HRVB exercises the baroreceptor reflex to enhance homeo-
static regulation. Both the heart rhythm coherence and resonance
frequency approaches to HRVB teach clients to produce auto-
coherent (sinusoidal) heart rhythms with a single peak in the
LF region and no significant peaks in the VLF and HF regions
(McCraty and Childre, 2010; Lehrer et al., 2013). The coherence
model and HEP research (MacKinnon et al., 2013) predict that
increased HRV will increase vagal afferent transmission to the
forebrain, activate the prefrontal cortex, and improve executive
function.
Emotional self-regulation strategies (Forman et al., 2007;
McCraty and Atkinson, 2012) can contribute to improved
client health and performance, alone, or in combination with
HRVB training. McCraty theorizes that emotional self-regulation
can increase resilience and accelerate recovery from stres-
sors. From Porges (2011) perspective, self-regulation through
social engagement and bonding can reduce SNS activation
while increasing HRV. The CAN model (Thayer et al., 2012)
predicts that perception of safety will reduce the activation
of the amygdala and increase the prefrontal cortex’s ability
to exercise top-down control of emotional responses. Finally,
from a heart rhythm coherence perspective, emotional self-
regulation reduces the SNS activation and/or vagal withdrawal
that increase short-term VLF power (Bernardi et al., 1996),
decrease shorter-term LF power, and disrupt heart rhythm
coherence.
SUMMARY
The SA node normally generates the heartbeat, which is modu-
lated by autonomic efferent neurons and circulating hormones.
There is a dynamic balance between sympathetic and parasym-
pathetic nervous outflows in a healthy, resilient, and responsive
nervous system. HRV is generated by multiple regulatory mech-
anisms that operate on different time scales. Recent findings
demonstrate the importance of the intrinsic cardiac nervous
system and cardiac afferents in generating the heart rhythm
and modulating the time interval between heartbeats. Vagally-
mediated HRV appears to represent an index of self-regulatory
control, such that individuals with greater resting HRV perform
better on tests of executive functions.
Since the LF band primarily reflects the vagally-mediated
transmission between the heart and the central nervous system
in the context of short-term BP regulation, resting measurements
should not be used as markers of SNS activity. Based on 24-h
monitoring, ULF and VLF rhythms are more strongly associated
with overall health status than HF rhythms. When age-adjusted
values are low, they are also more strongly associated with future
health risk and all-cause mortality.
www.frontiersin.org September 2014 | Volume 5 | Article 1040 |15
Shaffer et al. A healthy heart is not a metronome
HRVB exercises the baroreceptor reflex to enhance homeo-
static regulation and restore regulatory capacity. Both the heart
rhythm coherence and resonance frequency approaches to HRVB
train clients to produce auto-coherent heart rhythms with a single
peak in the LF region (typically around 0.1 Hz) and no signifi-
cant peaks in the VLF and HF regions. Emotional self-regulation
strategies can contribute to improved client health and perfor-
mance, alone, or in combination with HRVB training. A coherent
heart is not a metronome since its rhythms are characterized by
dynamic complexity with stability over longer time scales.
ACKNOWLEDGMENTS
The authors want to express their profound thanks to Mike
Atkinson, Richard Gevirtz, Paul Lehrer, Donald Moss, and John
Venner for their generous contributions to this article.
REFERENCES
Agelink, M. W., Boz, C., Ullrich, H., and Andrich, J. (2002). Relationship
between major depression and heart rate variability. Clinical consequences and
implications for antidepressive treatment. Psychiatry Res. 113, 139–149. doi:
10.1016/S0165-1781(02)00225-1
Ahmed, A. K., Harness, J. B., and Mearns, A. J. (1982). Respiratory control of heart
rate. Eur. J. Appl. Physiol. 50, 95–104. doi: 10.1007/BF00952248
Akselrod, S., Gordon, D., Ubel, F. A., Shannon, D. C., Barger, A. C., and Cohen,
R. J. (1981). Power spectrum analysis of heart rate fluctuation: a quantita-
tive probe of beat-to-beat cardiovascular control. Science 213, 220–222. doi:
10.1126/science.6166045
Alabdulgader, A. A. (2012). Coherence: a novel nonpharmacological modality for
lowering blood pressure in hypertensive patients. Glob. Adv. Health Med. 1,
56–64. doi: 10.7453/gahmj.2012.1.2.011
Ardell, J. L., Butler, C. K., Smith, F. M., Hopkins, D. A., and Armour, J. A. (1991).
Activ ity of in vivo atrial and ventricular neurons in chronically decentralized
canine hearts. Am. J. P hysiol. 260, H713–H721.
Ardell, J. L., Cardinal, R., Vermeulen, M., and Armour, J. A. (2009). Dorsal spinal
cord stimulation obtunds the capacity of intrathoracic extracardiac neurons to
transduce myocardial ischemia. Am.J.Physiol.Regul.Integr.Comp.Physiol.297,
R470–R477. doi: 10.1152/ajpregu.90821.2008
Armour, J. A. (1991). Intrinsic cardiac neurons. J. Cardiovasc. Electrophysiol. 2,
331–341. doi: 10.1111/j.1540-8167.1991.tb01330.x
Armour, J. A. (2003). Neurocardiology: Anatomical and Functional Principles.
Boulder Creek, CA: Institute of HeartMath.
Armour, J. A. (2008). Potential clinical relevance of the “little brain” on the mam-
malian heart. Exp. Physiol. 93, 165–176. doi: 10.1113/expphysiol.2007.041178
Armour, J. A., and Kember, G. C. (2004). “Cardiac sensory neurons, in Basic and
Clinical Neurocardiology,edsJ.A.ArmourandJ.L.Ardell(NewYork,NY:
Oxford University Press), 79–117.
Axelrod, S., Lishner, M., Oz, O., Bernheim, J., and Ravid, M. (1987). Spectral anal-
ysis of fluctuations in heart rate: an objective evaluation of autonomic nervous
control in chronic renal failure. Nephron 45, 202–206. doi: 10.1159/000184117
Baselli, G., Cerutti, S., Badilini, F., Biancardi, L., Porta, A., Pagani, M., et al. (1994).
Model for the assessment of heart period and arterial pressure variability inter-
actions and of respiration influences. Med.Biol.Eng.Comput. 32, 143–152. doi:
10.1007/BF02518911
Beauchaine, T. (2001). Vagal tone, development, and Gray’s motivational the-
ory: toward an integrated model of autonomic nervous system functioning
in psychopathology. Dev. Psychopathol. 13, 183–214. doi: 10.1017/S095457940
1002012
Bedell, W., and Kaszkin-Bettag, M. (2010). Coherence and health care cost—RCA
actuarial study: a cost-effectiveness cohort study. Altern. Ther. Health Med. 16,
26–31.
Bernardi, L., Gabutti, A., Porta, C., and Spicuzza, L. (2001). Slow breathing reduces
chemoreflex response to hypoxia and hypercapnia, and increases baroreflex sen-
sitivity. J. Hypertens. 19, 2221–2229. doi: 10.1097/00004872-200112000-00016
Bernardi, L., Valle, F., Coco, M., Calciati, A., and Sleight, P. (1996). Physical
activity influences heart rate variability and very-low-frequency components
in Holter electrocardiograms. Cardiovasc. Res. 32, 234–237. doi: 10.1016/0008-
6363(96)00081-8
Berntson, G. G., Bigger, J. T. Jr., Eckberg, D. L., Grossman, P., Kaufmann, P. G.,
Malik, M., et al. (1997). Heart rate variability: origins, methods, and interpretive
caveats. Psychophysiology 34, 623–648. doi: 10.1111/j.1469-8986.1997.tb02140.x
Berntson, G. G., and Cacioppo, J. T. (1999). Heart rate variability: a neuroscien-
tific perspective for further studies. Card. Electrophysiol. Rev. 3, 279–282. doi:
10.1023/A:1009920002142
Berntson, G. G., Norman, G. J., Hawley, L. C., and Cacioppo, J. T. (2008). Cardiac
autonomic balance versus regulatory capacity. Psychophysiology 45, 643–652.
doi: 10.1111/j.1469-8986.2008.00652
Berntson, G. G., Quigley, K. S., and Lozano, D. (2007). “Cardiovascular psy-
chophysiology, in Handbook of Psychophysiology, eds J. T. Cacioppo, L. G.
Tassinary, and G. G. Berntson (New York, NY: Cambridge University Press),
182–210.
Berry, M. E., Chapple, I. T., Ginsberg, J. P., Gleichauf, K. J.,Meyer, J. A., and Nagpal,
M. L. (2014). Non-pharmacological intervention for chronic pain in veterans: a
pilot study of heart rate variability biofeedback. Glob. Adv. HealthMed. 3, 28–33.
doi: 10.7453/gahmj.2013.075
Bigger,J.T.Jr.,Fleiss,J.L.,Steinman,R.C.,Rolnitzky,L.M.,Kleiger,R.E.,
and Rottman, J. N. (1992). Frequency domain measures of heart period vari-
ability and mortality after myocardial infarction. Circulation 85, 164–171. doi:
10.1161/01.CIR.85.1.164
Billman, G. E. (2013). The LF/HF ratio does not accurately measure cardiac
sympatho-vagal balance. Front. Physiol. 4:26. doi: 10.3389/fphys.2013.00026
Bonaduce, D., Petretta, M., Morgano, G., Villari, B., Binachi, V., Conforti, G., et al.
(1994). Left ventricular remodelling in the year after myocardial infarction:
an echocardiographic, haemodynamic, and radionuclide angiographic study.
Coron. Artery Dis. 5, 155–162. doi: 10.1097/00019501-199402000-00009
Brown, T. E., Beightol, L. A., Koh, J., and Eckberg, D. L. (1993). Important influence
of respiration on human R-R interval power spectra is largely ignored. J. Appl.
Physiol. (1985) 75, 2310–2317.
Cameron, O. G. (2002). Visceral Sensory Neuroscience: Interoception.NewYork,NY:
Oxford University Press.
Cantin, M., and Genest, J. (1985). The heart, an endocrine gland. Ann. Endocrinol.
46, 219–228.
Cantin, M., and Genest, J. (1986). The heart as an endocrine gland. Clin. Invest.
Med. 9, 319–327.
Carney, R. M., Blumenthal, J. A., Stein, P. K., Watkins, L., Catellier, D., Berkman,
L. F., et al. (2001). Depression, heart rate variability, and acute myocardial
infarction. Circulation 104, 2024–2028. doi: 10.1161/hc4201.097834
Carney, R. M., Freedland, K. E., Stein, P. K., Miller, G. E., Steinmeyer, B., Rich,
M. W., et al. (2007). Heart rate variability and markers