The systematic investigation of mind-body relations forms the scientific basis for the science of psycho-physiology. Unlike the correlative view of mind-body evaluations that dominates psychology and psychiatry, psychophysiology emphasizes a continuity between neuro-physiological and psychological processing. Psycho-physiologists assume that the nervous system provides the functional units for the bidirectional transduction of psychological and physiological processes. Thus, from a psychophysiological perspective, it is possible to link, not merely in theory, but also via measurement, psychological processes with neurophysiological processes and brain structures.
The paper will focus on the neural regulation of the heart via the vagus and how this regulation evolved to facilitate specific psychological processes. The Poly-Vagal Theory, described in this paper, provides an explanation of how the vagal pathways regulate heart rate in response to novelty and to a variety of stressors. The theory proposes that via evolution, mammalian nervous systems developed two vagal systems: one a phylogenetic relic of amphibia and reptilia, and the other, an evolutionary modification unique to mammals. According to the Poly-Vagal Theory, the two vagal systems are programmed with different response strategies and may respond in a contradictory manner. Explanations for several psychophysiological phenomena and psychosomatic disturbances will be proposed. The theory is based upon an established literature in neurophysiology, neuroanatomy, and psychophysiology.
Arousal theory: Historical legacy.
Early psychophysiological research assumed that peripheral autonomic measures provided sensitive indicators of arousal or activation (Darrow, Jost, Solomon, & Mergener, 1942; Duffy, 1957; Lindsley, 1951; Malmo, 1959). This view was based upon a rudimentary understanding of the autonomic nervous system in which changes in electrodermal activity and heart rate were assumed to be accurate indicators of sympathetic activity. As activation-arousal theory developed, continuity between the peripheral autonomic response and central mechanisms was assumed. According to this assumption, any organ influenced by sympathetic efferent fibers, such as the sudomotor, vascular, or cardiac systems, was a potential indicator of limbic or cortical activity.
Although the specific pathways relating these various levels were never outlined and are still sketchy, electrodermal and heart rate measures became the primary focus of research during the early history of the Society for Psychophysiological Research. This was due to their presumed sympathetic innervation and, in part, to their measurement availability. Not by plan, but by default, this emphasis created a research environment that neglected several important factors including: 1) parasympathetic influences, 2) interactions between sympathetic and parasympathetic processes, 3) peripheral autonomic afferents, 4) central regulatory structures, 5) the adaptive and dynamic nature of the autonomic nervous system, and 6) phylogenetic and ontogenetic differences in structural organization and function.
Neglect of these concepts and an emphasis on a global construct of 'arousal' still abides within various sub- disciplines of psychology, psychiatry, and physiology. This outdated view of 'arousal' may restrict an understanding of how the autonomic nervous system interfaces with the environment and the contribution of the autonomic nervous system to psychological and behavioral processes. In contrast, more recent neurophysiological data promote a more integrative view of the autonomic nervous system.
Brain-heart communication: Historical perspective.
When we view living organisms as a collection of dynamic, adaptive, interactive, and interdependent physiological systems, it is no longer appropriate to treat the autonomic nervous system as functionally distinct from the central nervous system. We start to recognize that peripheral organs do not "float in a visceral sea." Rather, they are anchored to central structures via efferent pathways and are continuously signaling central regulatory structures via their abundant afferent pathways. Thus, the bidirectional connections between autonomic and central brain structures are becoming apparent. Accordingly, new theories and research strategies must incorporate the dynamic and interactive constructs that link central structures with peripheral organs.
Darwin (1872) provides historical insight into the potential importance of the vagus in bidirectional communication between the brain and the heart. Although Darwin focused on facial expressions in defining emotions, he acknowledged the dynamic relationship between the vagus and the central nervous system activity that accompanied the spontaneous expression of emotions. He speculated that there were identifiable neural pathways that provided the necessary communication between specific brain structures and peripheral organs to promote the unique pattern of autonomic activity associated with emotions. For example he stated:
... when the mind is strongly excited, we might expect that it would instantly affect in a direct manner the heart; and this is universally acknowledged... when the heart is affected it reacts on the brain; and the state of the brain again reacts through the pneumo-gastric [vagus] nerve on the heart; so that under any excitement there will be much mutual action and reaction between these, the two most important organs of the body (p.69).
For Darwin, when an emotional state occurred, the beating of the heart changed instantly, the change in cardiac activity influenced brain activity, and the brain stem structures via the cranial nerves (i.e., vagus) stimulated the heart. He did not elucidate the neurophysiological mechanisms that translate the initial emotional expression to the heart. Our current knowledge of the brain stem origin and neurophysiological function of the various branches of the vagus was not available to Darwin. At that time it was not known that vagal fibers originated in several medullary nuclei and that the branches of the vagus exerted control over the periphery through different feedback systems. However, Darwin's statement is important, because it emphasizes the afferent feedback from the heart to the brain, independent of the spinal cord and the sympathetic nervous system, as well as the regulatory role of the pneumo-gastric nerve (renamed the vagus at the end of the 19th century) in the expression of emotions.
Darwin attributed the above ideas to Claude Bernard as an example of nervous system regulation of le milieu interieur. Consistent with more contemporary psychophysiology, Claude Bernard viewed the heart as a primary response system capable of responding to all forms of sensory stimulation. He explicitly emphasized the potency of central nervous system pathways to the heart (Cournand, 1979). These ideas are expressed in the following quotation:
In man the heart is not only the central organ of circulation of blood, it is a center influenced by all sensory influences. They may be transmitted from the periphery through the spinal cord, from the organs through the sympathetic nervous system, or from the central nervous system itself. In fact the sensory stimuli coming from the brain exhibit their strongest effects on the heart (Claude Bernard, 1865 quoted in Cournand, 1979).
Although seldom acknowledged as founders of modern psychophysiology, Bernard and Darwin have contributed to the theoretical basis for a neuro-psychophysiology of the autonomic nervous system. The above quotations document their view that the heart provided not only an output system from the brain, capable of indexing sensory processing, but they also recognized that the heart was a source of afferent stimulation to the brain able to change or contribute to psychological state. Consistent with this theoretical bias, psychophysiologists during the past century have investigated the functional sensitivity of heart rate measures to sensory and affective stimuli (e.g., Darrow, 1929; Graham & Clifton, 1966; Lacey, 1967), and the dynamic feedback between the brain and the heart in regulating both psychological state and the threshold for sensory stimuli (e.g., Lacey and Lacey, 1978).
Contemporary psychophysiology gained much of its current theoretical perspective from intriguing ideas regarding the interaction between autonomic and sensory processes introduced by Sokolov (1963). The Sokolovian model contained all the requisite components of an integrative theory relating autonomic function to psychological state. The model included: 1) acknow-ledgment of both afferents and efferents in both autonomic and somatic systems, 2) an autonomic feedback loop (i.e., autonomic tuning) to regulate sensory thresholds, 3) an interface between autonomic processes and psychological phenomena (i.e., orienting and defensive reflexes), and, 4) brain regulation of autonomic reactivity via habituation.
The Sokolovian model included bidirectional communication between brain and periphery. In the Sokolovian, model autonomic processes contributed to the tuning of receptor systems to engage or disengage with the external environment. Consistent with the Sokolov view, the Laceys (e.g., Lacey, 1967; Lacey and Lacey, 1978) emphasized the bidirectional communication between the afferents in the cardiovascular system and brain in the regulation of both cardiac function and sensory threshold. In contrast, to this emphasis on bidirectional communication, Obrist (1976) focused on the general concordance between metabolic demands and heart rate. Both arguments have merit. For example, afferent stimulation of the baroreceptors has immediate effects on both peripheral cardiovascular function and on central arousal state (Gellhorn, 1964), and the metabolic demands associated with exercise have deterministic influences, via vagal withdrawal, on heart rate (Obrist, 1981; Rowell, 1993).
Heart rate responses: A neurogenic emphasis
Throughout the history of the Society for Psychophysiological Research, psychophysiologists have been studying robust phenomena, such as the autonomic components of the orienting reflex, often without explanatory neurophysiological models. This paper is in response to this need. The paper will provide a theoretical model based upon the evolution of neural structures and the neural regulation of autonomic processes to explain several psychophysiological phenomena including orientation, attention, and emotion.
The orienting reflex provides an excellent point of embarkation. Based upon the convergent theoretical approaches of Sokolov (1963), Lacey (1967), and Graham and Clifton (1966), the orienting reflex is assumed to have a cardiac component. This component is characterized by a heart rate deceleration that functionally influences perceptual thresholds, facilitating the processing of information regarding the state of the external environment. However, what are the neural mechanisms mediating the cardiac orienting response? Or, as Obrist (1976) argued, is the heart rate deceleration merely an epiphenomenon associated with decreased metabolic demands accompanying the reduced motor activity that defines orienting and attending behaviors? The time course of the response, the effects of neural blockades, and studies with clinical populations support the contention that the cardiac orienting response is neurogenic. First, heart rate deceleration associated with the cardiac orienting response is rapid, occurring within a few seconds and usually returns rapidly to baseline. Second, the latency characteristics of the cardiac orienting response are similar to other neurogenic bradycardic reflexes such as opto-vagal, vaso-vagal, baroreceptor-vagal and chemoreceptor-vagal reflexes.
Blockade studies with atropine demonstrate that short latency bradycardia associated with both orienting reflexes and classical conditioning are mediated by cholinergic pathways via the vagus (e.g., Berntson, Cacioppo, & Quigley, 1994; Obrist, 1981; Schneiderman, 1974). Studies with the aged and other clinical populations with peripheral neuropathies or autonomic regulatory problems (e.g., diabetes) document deficits in vagal function (De Meersman, 1993; Gribben, Pickering, Sleight, & Peto, 1971; Weiling, van Brederode, de Rijk, Borst, & Dunning, 1982; Weise & Heydenreich, 1991). Additionally, studies of individuals with unilateral brain damage demonstrate that heart rate responses are diminished more in individuals with right side damage (Yokoyama, Jennings, Ackles, Hood, & Boller, 1987). This latter finding is consistent with the evidence that neurophysiological regulation of heart rate is primarily via the right vagus to the sino-atrial node, and that heart rate is under control of higher ipsilateral structures in the brain (Warwick & Williams, 1975).
Vagal influences producing heart rate deceleration in response to mild stress may interact synergistically with sympathetic withdrawal (Buwalda, Koolhaas & Bohus, 1992). Moreover, in conditions of anticipation of aversive stimuli, there have been reports that heart rate deceleration is, in part, due to sympathetic withdrawal (Rau, 1991). Although there are reports of a sympathetic contribution to stimulus dependent heart rate decelerations short latency decelerations are determined primarily by the vagus. Thus, it may be argued, that since short latency heart rate reactivity is mediated by the vagus, the magnitude of the cardiac orienting response is an index of vagal regulation.
The Vagal Paradox
In attempting to structure a neurogenic model of vagal regulation to explain psychophysiological phenomena, there is an obvious inconsistency between data and theory. Physiological theory attributes both the chronotropic control of the heart (i.e., heart rate) and the amplitude of respiratory sinus arrhythmia (RSA) to direct vagal mechanisms (e.g., Jordan, Khalid, Schneiderman & Spyer, 1982; Katona & Jih, 1975). However, while there are situations in which both measures covary (e.g., during exercise and cholinergic blockade), there are other situations in which the measures appear to reflect independent sources of neural control.
Several arguments have been made to explain this discrepancy. First, it has been argued that RSA and average heart rate (during sympathetic blockade) reflect different dimensions of vagal activity. For example, average heart rate might be viewed as reflecting tonic vagal influences and RSA as reflecting phasic vagal influences (e.g., Berntson, Cacioppo, & Quigley, 1993; Jennings & McKnight, 1994; Malik & Camm, 1993). Second, it has been argued that the discrepancy is caused by variations in respiratory parameters (Grossman, Karemaker, & Wieling, 1991) with RSA being confounded by respiratory frequency and tidal volume. Third, it has been argued that variation in quantification methods may contribute to the divergence between RSA and heart rate (Porges & Bohrer, 1990; Byrne & Porges, 1993). And fourth, it has been argued that average heart rate is influenced by a complex and dynamic interaction between sympathetic and vagal systems making it difficult to extract a vagal tone dimension (Berntson, Cacioppo, and Quigley ,1991, 1993).
Often, the arguments have been linked to a definition of vagal tone determined via neural blockade. The functional effect of the neural blockade on heart rate has been used as the criterion measure of vagal tone or parasympathetic control (e.g., Katona & Jih, 1975). Researchers have argued that RSA is not an accurate index of vagal tone, because individual pre-blockade levels of RSA do not accurately map into pre-post change in heart rate (Grossman & Kollai, 1993). Contrary to this argument, Porges (1986) argued that the discrepancy was, in part, based upon the criterion measure selected. He demonstrated that RSA exhibited a more sensitive dose-dependent response curve to vagal blockade via atropine than heart rate. This suggests the possibility that RSA, monitored during periods of spontaneous breathing, may provide a better criterion variable than heart rate. Neurophysiological support may be offered for this proposal. RSA is a vagal phenomenon in contrast to heart rate, which is determined by several sources including vagal, sympathetic, and mechanical factors. Thus, the efficacy of change in heart rate following cholinergic blockade as an index of vagal tone may be challenged.
The above arguments have created a volatile environment debating the neurophysiological interpretation of RSA and the efficacy of specific methods to quantifying RSA. Common to these arguments is the assumption that there is one central source of cardiac vagal tone. The arguments attribute differences, not to central mechanisms, but to the response characteristics of heart rate and RSA. Thus, divergence has been attributed to either the transfer function of the sino-atrial node that would attenuate high frequency oscillations (Saul, Berger, Chen, & Cohen, 1989) or the statistical transfer function of the method of quantifying RSA (Byrne & Porges, 1993) and not as a function of differential neural output.
However, independent of the quantification methodology and during periods of stable respiratory parameters, data have accumulated that demonstrate that RSA and heart rate (independent of sympathetic influences) often respond differently. Although both the neurogenic bradycardia and the suppression of RSA or heart rate variability observed during attention are assumed to be vagal in origin, they often appear independent of each other or in an apparent physiological contradiction (Porges & Raskin, 1969; Porges, 1972; Richards & Casey, 1991). Similar disparities between levels of heart rate and RSA have been observed during inhalant anesthesia when RSA exhibits a massive depression, while heart rate is not altered (Donchin, Feld, & Porges, 1985). Additional examples of convergence and divergence between RSA and heart rate can be observed in both within- and between-subjects designs. For example, individual differences in heart rate and RSA monitored during resting conditions provide independent contributions to measures of cardiac vagal tone derived from vagal blockade (e.g., Grossman & Kollai, 1993). However, convergence may be observed within an individual during exercise when monotonic increases in metabolic load are reflected in both faster heart rate and lower RSA (Billman & DuJardin, 1990). Or, convergence can be observed during neural blockade, via atropine, when both cardiac indices diminish in a clear dose response manner (Cacioppo, Berntson, Binkley, Quigley, Uchino, & Fieldstone, 1994; Dellinger, Taylor, & Porges, 1987; Porges, 1986).
The relationship between RSA and heart rate may change within and between individuals. In our laboratory we have observed that the relationship between RSA and heart rate varies with behavioral state (Riniolo, Doussard-Roosevelt, & Porges, 1994). Twenty-four-hour ambulatory monitoring of adults indicates that during states of drowsiness and sleep the correlation between RSA and heart rate is significantly lower than during alert states. Thus, at times, RSA and heart rate appear to reflect the same physiological processes, while at other times they appear to reflect independent processes.
In contrast to the observable data, neuro-physiological research argues for a covariation between these two parameters, because vagal cardioinhibitory fibers to the heart have consistent functional properties characterized by bradycardia to neural stimulation and a respiratory rhythm (e.g., Jordan, Khalid, Schneiderman, & Spyer, 1982). This inconsistency, based upon an assumption of a single central vagal source is labeled the Vagal Paradox and is outlined in Table 1.
Table 1
The Vagal Paradox is critical to the interpretation of several psychophysiological and clinical conditions. For example, if the bradycardia occurring during orienting reflexes are vagal, why are bradycardia often observed during periods of reduced RSA, also an index of both attention and vagal control of the heart? If vagal tone is a positive indicator of health of a fetus or neonate when monitored with RSA, why is vagal tone a negative indicator of health when it is manifested as bradycardia? If bradycardia and RSA can both be removed by severing the vagus or by pharmacological blockade, are they both manifestations of vagal tone? If bradycardia and RSA are both indices of vagal tone, why do they respond differently? This apparent paradox provides the stimulus for the following inquiry and the development of the proposed Poly-Vagal Theory that speculates that, in mammals, there are two anatomically based vagal response systems.
Mammalian Poly-Vagal System
To understand the proposed Poly-Vagal Theory, it is necessary to provide additional information regarding the neuroanatomy and neurophysiology of the vagus in mammals. First, the vagus is not one nerve, but a family of neural pathways originating in several areas of the brain stem. Second, there are several branches of the vagus. Third, the vagus is not solely an efferent or motor pathway; rather, at least 80% of the vagal fibers are afferent (Agostoni, Chinnock, DeBurgh Daly, & Murray, 1957). Fourth, the vagus is lateralized with nerve trunks originating in the left and right sides of the brain stem. Fifth, the vagus is asymmetrical with the left and right sides performing different tasks, with the right vagus most potent in the chronotropic regulation of the heart. These points are summarized in Table 2.
Table 2
Mammals are poly-vagal. The different vagi have different roles in the regulation of visceral function and originate in different brain stem nuclei with their respective viscerotropic organization. The different vagi may have oppositional outputs to the same target organ. For example, it is possible that during orienting there is an increase in vagal outflow from one branch to produce bradycardia and a withdrawal of vagal outflow from the other branch to produce a suppression of RSA (e.g., Richards and Casey, 1991). Thus, the concept of vagal tone may not be generalized to all vagal efferent pathways or even to the same target organ (e.g., heart), as has been assumed (e.g., Grossman & Kollai, 1993), but may need to be limited to a specific branch or subsystem of the vagus being evaluated. And, the intriguing concept of autonomic space proposed by Berntson, Cacioppo, and Quigley (1991, 1993) to deal with dynamic sympathetic-parasympathetic interactions may require an additional dimension to deal with potential vago-vagal interactions.
The Poly-Vagal Theory proposes that neurogenic bradycardia and RSA are mediated by separate branches of the vagus. Thus, the two commonly used, but not interchangeable measures of cardiac vagal tone may represent different dimensions of vagal tone.
In mammals, the primary motor fibers of the vagus originate from two separate and definable nuclei in the medulla: the dorsal motor nucleus of the vagus (DMNX) and the nucleus ambiguus (NA). DMNX is in the dorsomedial medulla. NA is ventral to DMNX in the ventrolateral reticular formation (Warwick & Williams, 1975). The name ambiguus emphasizes the initial difficulties associated with determining its borders and connections within the reticular formation (Mitchell & Warwick, 1955). A third medullary nucleus, located near DMNX, the nucleus tractus solitarius (NTS), is the terminus of many of the afferent pathways travelling through the vagus from peripheral organs. This trinity of neural structures in the medulla, forms the primary central regulatory component of the vagal system. The relative locations of these medullary nuclei are illustrated in Figure 1.
Most cells originating in DMNX project to subdiaphragmatic structures (e.g., stomach, intestines, etc). In contrast, only the rostral portion of NA provides vagal innervation of subdiaphragmatic structures (Kalia & Mesulam, 1980), while most cells in NA project to supradiaphragmatic structures (larynx, pharynx, soft palate, esophagus, bronchi, and heart).
Neurotracing and electrophysiological techniques with mammals provide additional evidence that the two vagal nuclei may function independently and have different central connections. These studies have demonstrated that there are no apparent connections between the two nuclei, although both nuclei have input from NTS, central nucleus of the amygdala, and hypothalamus, (Hopkins, 1987; Leslie, Reynold, & Lawes, 1992). It is well accepted that in mammals the primary cardioinhibitory motoneurons are located in NA. However, motor fibers from DMNX join the cardiac vagus (Bennett, Ford, Kidd, & McWilliam, 1984).
Cardioinhibitory and bronchoconstrictor neurons located in NA have myelinated vagal axons that conduct in the fast B fiber range (McAllen & Spyer, 1976, 1978). In contrast, neurons located in DMNX have axons projecting to the cardiac vagal branches that are non-myelinated and conduct in the slower C fiber range. Although there are reports of cardioinhibitory vagal neurons with efferent axons conducting in the B fiber range being located in both DMNX and NA, neurons with axons conducting in the C fiber range are restricted to DMNX (Jordan, Khalid, Schneiderman & Spyer, 1982). The role of these non-myelinated vagal fibers on the heart is not well understood. In research with cats (Ford, Bennett, Kidd, & McWilliam, 1990) and dogs (Donald, Samueloff & Ferguson, 1967) stimulation of these fibers did not affect heart rate. However, although unsubstantiated at this time, the function of these fibers may be dependent upon the outflow of the myelinated NA fibers and may change during conditions such as hypoxia. For example, the influence of the unmyelinated fibers on the heart may be potentiated when the outflow from the mylenated NA fibers are blocked. In contrast, in the rabbit, stimulation of the non-myelinated vagal fibers results in heart rate slowing (Woolley, McWilliam, Ford, & Clarke, 1987).
The cytoarchitecture of NA illustrates that the dorsal portion contains source nuclei for special visceral efferents (i.e., voluntary motor fibers) and the ventral portion contains source nuclei for general visceral efferents (i.e., involuntary motor fibers). Motor projections from the dorsal portion go to target organs including the larynx, pharynx, soft palate and esophagus. Motor projections from the ventral portion go to several target organs including the heart, and the bronchi. In fact, these projections account for the primary cardiac and bronchomotor pathways and far outnumber the pathways originating in DMNX.
There is an obvious distinction between the viscerotropic organization of the two vagal nuclei. DMNX provides the primary vagal efferents to subdiaphragmatic organs that regulate digestive and alimentary processes. In contrast, NA provides the primary vagal efferents to the supra-diaphragmatic target organs including soft palate, pharynx, larynx, esophagus, bronchi, and heart.
