A model of neurovisceral integration in emotion regulation and dysregulation

In the present paper we present the outlines of a model that integrates autonomic, attentional, and affective systems into a functional and structural network that may help to guide us in our understanding of emotion regulation and dysregulation. We will emphasize the relationship between attentional regulation and affective processes and propose a group of underlying physiological systems that serve to integrate these functions in the service of self-regulation and adaptability of the organism. We will attempt to place this network in the context of dynamical systems models which involve feedback and feedforward circuits with special attention to negative feedback mechanisms, inhibitory processes, and their role in response selection. From a systems perspective, inhibitory processes can be viewed as negative feedback circuits that allow for the interruption of ongoing behavior and the re-deployment of resources to other tasks. When these negative feedback mechanisms are compromised, positive feedback loops may develop as a result (of dis-inhibition). From this perspective, the relative sympathetic activation seen in anxiety disorders may represent dis-inhibition due to faulty inhibitory mechanisms.  2000 Elsevier Science B.V. All rights reserved.


Introduction
adapts itself with great ease to whatever may be placed before it, even something that is out of its ''The secret of success lies not so much in the accustomed line.' ' (p. 25) Swami Paramananda abundance of thought as in our ability to follow (1974)  things in order to deal effectively with others, and 0165-0327 / 00 / $ -see front matter © 2000 Elsevier Science B. V. All rights reserved. PII: S0165-0327( 00 )00338-4 is a condition which has a real opposite in the and adaptability of the organism. We will attempt to confused, dazed, scatter-brain state.'' (James, place this network in the context of dynamical 1997) systems models which involve feedback and feedforward circuits with special attention to negative The idea that inhibitory processes are important feedback mechanisms, inhibitory processes, and their for adaptability and success of both the individual role in response selection. Importantly, we will show and the species is not new. The above quotations that the arousal associated with anxiety represents a from one of the first Eastern yogic teachers to come dis-inhibition of positive feedback circuits that are to the United States and from one of the pioneers of normally under tonic inhibitory control. American psychology were written in the early part of the twentieth century. The first represents a distillation of thousands of years of yoga psychol-2. Functional and dysfunctional aspects of ogy, the second the prescient writings of one of the emotion most influential thinkers of the twentieth century. More recently, McGeer et al. (1978) note: Emotions represent a distillation of an individual's perception of personally relevant interplay with the ' 'We can think that inhibition is a sculpturing environment including not only challenges and process. The inhibition, as it were, chisels away at threats but also their ability to respond to or cope the diffuse and rather amorphous mass of excitat-with them (Frijda, 1988). As such, emotions serve as ory action and gives a more specific form to an integrative index of an individual's adjustment to neuronal performance at every stage of synaptic the constantly changing environmental demands they relay. This suppressing action of inhibition can be face. Emotions have been characterized as an orrecognized very clearly at higher levels of the ganismic response to an environmental event that brain . . . ' ' (p. 133).
allows for the rapid mobilization of multiple subsystems for action (Levenson, 1988). In this context, A comprehensive model of emotions and disorders emotions are the moment-to-moment output of a of affect must attempt to account for the complex continuous sequence of behavior. These behavioral mix of cognitive, affective, behavioral, and physio-sequences are organized around biologically imporlogical concomitants of normal and pathological tant functions and these lawful sequences of behavior affective states and dispositions. In the present paper have been termed 'behavioral systems' (Timberlake, we present the outlines of a model that integrates 1994). Put another way, emotions are self-regulatory some of these components into a functional and responses that allow the efficient coordination of the structural network that may help to guide us in our organism for goal-directed behavior. For example, understanding of emotion regulation and dysregula-when faced with a potentially dangerous situation, a tion. Functionally, this network includes attentional Defensive Behavior System might be activated. The regulation, classical conditioning, affective informa-first stage of the behavioral sequence could involve tion processing, and behavioral and physiological the experience of anxiety, a shift toward relative flexibility. Structurally, this network includes central sympathetic autonomic activity including increased nervous system structures, particularly the cingulate heart rate, and a selective search for signs of danger. cortex, and peripheral endorgans, particularly the If a source of danger is identified the next stage in cardiovascular system. We will briefly review some the behavioral sequence might involve fear and the of the work that we have done on the functional mobilization of the organism for fight or flight. As aspects of emotion regulation and dysregulation Frijda (1988) noted, specific emotions imply specific tying these functions to specific structures. We will eliciting stimuli, specific action tendencies including emphasize the relationship between attentional regu-selective attention to relevant stimuli, and specific lation and affective processes and propose a group of reinforcers. When this system works properly it underlying physiological systems that serve to inte-allows for the flexible adaptation of the organism to grate these functions in the service of self-regulation changing environmental demands. In another sense, an emotional response represents a selection of an (preferred configurations) in the state-space (beappropriate response and the inhibition of other less havioral repertoire) of the organism. One conseappropriate responses from a more or less broad quence of this idea is that the high dimensional behavioral repertoire. system may be guided through the emotional state-These behavioral systems recruit a whole host of space by a smaller number of dimensions of change organism resources in the service of goal-directed or control parameters. One of us has recently prebehavior and adaptability (cf. Damasio, 1998). From sented evidence to this effect (Johnsen et al., 1995; a dynamical systems perspective an organism is a Nyklicek et al., 1997;Thayer and Friedman, 1997). complex set of reverberating circuits or sub-systems Thus, discrete emotions may be viewed as attractors working together in a coordinated fashion (a set of in the state-space defined by a small set of control loosely coupled bio-oscillators). The individual ele-parameters. A vast amount of research points to the ments of the organism can be thought of as a high two dimensions of valence and arousal as being the dimensional system with a large number of degrees most important in this respect (Osgood et al., 1957; of freedom. However, in the service of goal-directed Johnsen et al., 1995;Nyklicek et al., 1997; Ohman et behavior and in the context of a behavioral system, al., 1993;Russell, 1980). These dimensions have these elements organize into coordinated assem-been viewed as representing the motivational sysblages that can be described by a smaller number of tems of approach and avoidance (valence), and control parameters. This is not unlike the factors of together with an index of the amount of vigor with factor analysis, which reveal the latent structure which these drives are pursued (arousal) have been among a set of questionnaire items thereby reducing suggested as the underpinnings of goal-directed or mapping the high dimensional item space into a behavior in a wide range of organisms (cf. Schneirla, lower dimensional factor space. For example, the 1959). coordination of breathing, the direction of blood flow Further, in the dynamical systems perspective, the to various muscles and organs, the release of appro-basin of attraction for a given emotional state defines priate hormones such as cortisol, the detailed control the set of parameter values that will lead to one of specific muscle fibers to aid in grasping or emotion as opposed to another. That is, the unique running, the control of the biochemical processes combination of internal and external conditions that support these processes, and the higher level (parameter values) that lead to the emergence of one information processing that accompanies the apprais-behavioral system or sequence of emotions (attractor) al of threat and challenge represent just some of the over another is termed the basin of attraction. Within responses at various levels of system organization, the context of the physiological constraints and the each with its own constraints and options (degrees of learning history of a given organism, the set of freedom) that must be coordinated in a fraction of a parameter values or environmental conditions that second (see Schwartz, 1986, for a detailed chart of lead to one emotion versus another can vary greatly. various functions associated with an emotional re- The map of these parameter values and related sponse). This complex system, as it moves through attractors defines the behavioral repertoire, emotional time on numerous time-scales, requires various feed-topography or state-space of each individual. back and feedforward circuits for its efficient func-We have asserted (Friedman and Thayer, 1998; tioning. Moreover, the emotional response 'emerges' Thayer and Friedman, 1997) that disorders of affect, from the interaction of the various sub-systems with including anxiety disorders, may be viewed as a kind the environmental demands and this response is not of distorted emotional state space in which an orchestrated from a central command center. Thus, it individual is unable to shift into an attractor or is a distributed system. emotion that is appropriate for a given set of These behavioral systems or preferred configura-environmental demands. As such the individual is tions and trajectories represent modes of relative 'stuck' in an attractor or behavioral pattern that is not stability in the constant flow of organism-environ-responsive to the demands placed upon it by the ment interactions. As such, from a dynamical sys-environment (see also Stormark et al., 1998). This is tems perspective emotions may represent attractors manifested in inflexibility at various levels of system organization. Put another way, the individual is as manifested in cardiac variability may be related to unable to select the appropriate response or, more both attentional regulation and affect. Cardiac vagal often the case, unable to inhibit the inappropriate tone as reflected in heart rate variability (HRV) in response. Thus, the response selection mechanism is particular has been shown to be related to attentional somehow corrupted.
control and to emotional regulation (Nyklicek et al., We have investigated the ability to respond to or 1997; Porges, 1991Porges, , 1992Thayer et al., 2000). 'read' the emotional landscape and generate an These researchers propose that measures of cardiac appropriate response in the context of self-moni-vagal tone index the efficiency of central-peripheral toring (Miller andThayer, 1988, 1989) and levels of neural feedback mechanisms. Thus, these measures emotional awareness (Lane and Schwartz, 1987). may serve to quantify the ability to self-regulate, This 'reading' of the emotional environment requires through the organization of physiological resources the selection of certain information and the disre-and appropriate response selection, in the service of garding of other information from a complex input goal-directed behavior. Specifically, high vagal tone that includes internal as well as external cues. When is associated with the ability to self-regulate and thus this ability to effectively process affective infor-to have greater behavioral flexibility and adaptability mation is compromised for whatever reason, the in a changing environment. Low vagal tone, on the individual is unable to maneuver efficiently in its other hand, is associated with poor self-regulation environment. Inefficiency in affective information and a lack of behavioral flexibility (Porges, 1992, p. processing leads to affective dysregulation. When 208). Thus, HRV appears to index vital aspects of this inefficiency becomes severe, various forms of self-regulation due to its ability to reflect neural pathology are said to exist such as alexithmia, feedback mechanisms of central nervous system depression, panic disorder, generalized anxiety disor-(CNS)-autonomic nervous system (ANS) integration der, hostile personality, hypertension, and coronary (Friedman and Thayer, 1998). heart disease to name a few.
We have presented a functional framework that A related aspect of this self-regulation involves integrates affective regulation, attentional regulation, selective attention. The ability to sustain and shift and heart rate variability. This functional system attention is an important component of organism serves to describe the behavioral processes associself-regulation and adaptability. The capacity to ated with goal-directed behavior and adaptability. We select meaningful information and disregard irrele-next describe the structural system that serves to vant information from the external and internal instantiate this functional system in the physiological environments is critical for the survival of the 'wet-ware' of an organism. organism. Information that is particularly meaningful to a given individual will attract that person's attention and resources.
3. The central autonomic network and the What makes information meaningful to an in-anterior executive region dividual is its impact on the well being of the organism. Does the information represent something A number of researchers have identified functional aversive to be avoided or something appetitive to be units within the CNS that appear to support goalapproached? Thus, the selective attention to infor-directed behavior and adaptability. One such funcmation is often of an affective or motivational tional unit is the central autonomic network (CAN; character (Stormark et al., 1995;Stormark and Benarroch, 1993). Functionally, this network Hugdahl, 1996. In this context, one can posit that is an integrated component of an internal regulation attentional regulation and affective regulation are system through which the brain controls visfunctionally an integrated system in the service of ceromotor, neuroendocrine, and behavioral responses organism self-regulation and adaptability (see, for that are critical for goal-directed behavior and adaptexample, Heilman, 1997).
ability (Benarroch, 1993). Structurally, the CAN A number of researchers (e.g., Porges, 1991Porges, , 1992 includes the anterior cingulate, insular, and ven-Richards and Casey, 1992;Thayer et al., 2000) have tromedial prefrontal cortices, the central nucleus of suggested that autonomic nervous system regulation the amygdala, the paraventricular and related nuclei of the hypothalamus, the periaquaductal gray matter, sympathoexcitatory circuits within the CAN (Benarthe parabrachial nucleus, the nucleus of the solitary roch, 1993Masterman and Cummings, 1997;tract (NTS), the nucleus ambiguus, the ventrolateral Spyer, 1989). medulla, the ventromedial medulla, and the medul-Other functional units within the CNS subserving lary tegmental field. The primary output of the CAN executive, social, affective, attentional, and motiis mediated through the preganglionic sympathetic vated behavior in humans and animals have also and parasympathetic neurons. Importantly, these been identified (Damasio, 1998; Devinsky et al., neurons innervate the heart via the stellate ganglia 1995; Masterman and Cummings, 1997; Posner and and the vagus nerve. The interplay of these inputs to Petersen, 1990;Spyer, 1989). One such functional the sino-atrial node of the heart is the source of the unit has been termed the anterior executive region complex variability that characterizes the heart rate (AER; Devinsky et al., 1995). Functionally, the AER time series (Saul, 1990). Thus, the output of the and its projections ''assesses the motivational content CAN is directly linked to heart rate variability of internal and external stimuli and regulates context-(HRV). In addition, sensory information from the dependent behaviors.'' (Devinsky et al., 1995, p. peripheral end organs such as the heart are fed back 279). The AER and its projections has been termed to the CAN, one important example of which is the the 'rostral limbic system' and structurally includes baroreceptor reflex. As such, HRV is an index of the anterior, insular, and orbitofrontal cortices, the central-peripheral neural feedback and CNS-ANS amygdala, the periaquaductal gray, the ventral integration.
striatum, and autonomic brainstem motor nuclei. Moreover, the CAN has many features of a Another such functional unit has been identified by nonlinear dynamical system. First, the components of Damasio (1998) as the neural substrate for emotions. the CAN are reciprocally interconnected. This allows Table 1 lists the structures and functions associated for continuous positive and negative feedback inter-with the CAN, the rostral limbic system, and the actions and integration of autonomic responses.
'emotion circuit'. The structural overlap is quite Second, the CAN is comprised of a number of substantial. parallel, distributed pathways, which allows for We propose that the CAN, the AER and its multiple avenues to a given response. For a simple projections, the 'emotion circuit' (Damasio, 1998), example, a given heart rate change of 68 to 80 beats and related systems (Masterman and Cummings, per minute can be achieved by various combinations 1997;Spyer, 1989) are one and the same functional of sympathetic and parasympathetic input to the network identified by different researchers from sino-atrial node including increased sympathetic differing orientations. This network of CNS strucactivity, decreased parasympathetic activity or any tures is associated with the processes of response combination of the two, as well as via other path-organization and selection, and serves to modulate ways such as circulating hormones. Moreover, within psychophysiological resources in attention and emothe CAN direct and indirect pathways can modulate tion (Friedman and Thayer, 1998; Thayer and Friedthe output to the preganglionic sympathetic and man, 1997). parasympathetic neurons. Third, the activity of the In the following we will briefly review several CAN is state dependent and thus sensitive to initial experimental investigations that we have conducted conditions (see Glass and Mackey, 1988).
to help elucidate the function and structure of this The CAN receives and integrates visceral, humor-network. Most of this work has been published al, and environmental information and coordinates elsewhere in detail and the interested reader is autonomic, endocrine, and behavioral responses to referred to those publications for an in-depth discusenvironmental challenges. Importantly, the CAN is sion of the experimental procedures. under tonic inhibitory control. This is achieved by g-aminobutyric acid (GABA) interneurons within the NTS. GABA is the main inhibitory neurotransmitter 4. HRV in emotion and psychopathology within the CNS. Disruption of this inhibitory pathway may lead to such things as hypertension and Autonomically mediated cardiovascular variability sinus tachycardia, and represents a dis-inhibition of is critical as an index of neurovisceral integration and organism self-regulatory ability. The interplay of pathology. Similar models of vagal inhibition have sympathetic and parasympathetic (vagal) outputs of recently been put forward to describe the relationship the CAN at the sino-atrial node produces the com-between psychological factors and physiological plex beat-to-beat variability that is characteristic of a health. Brosschot and Thayer (1998) have related healthy, adaptive organism. Vagal influences domi-vagal inhibition to hostility and the risk for carnant cardiovascular control and thus the cardiovascu-diovascular disease, and Sroka et al. (1997) have lar system is under tonic inhibitory control via the suggested vagal inhibition as the link between psyvagus nerve (Levy, 1990). There are several lines of chological factors and myocardial ischemia. The research that point to the importance of HRV in interested reader is referred to some of our other attention, emotion and affective disorders. This work on non-pathological emotion for the role of literature has recently been reviewed by Friedman vagally mediated cardiac activity in various emotionand Thayer (1998). They concluded that a relative al states (Nyklicek et al., 1997), and emotional reduction in vagally mediated HRV is consistent with regulation (Sollers et al., 1997). the cardiac symptoms of panic anxiety as well as Thayer et al. (1996) examined the autonomic with the psychological symptoms of poor attentional characteristics of GAD and its cardinal feature, control, ineffective emotional regulation, and be-worry. Autoregressive spectral analysis of heart havioral inflexibility. Recent studies on depression period variability (HPV) was used to investigate the (Thayer et al., 2000) and generalized anxiety disor-effects of a 10 min relaxation period and a 10 min der (GAD;Thayer et al., 1996) have found similar worry period in persons with GAD and non-anxious reductions in HRV. This reduction of vagally me-controls. The results provided evidence for two main diated cardiovascular control serves to dis-inhibit effects (see Fig. 1). One, persons with GAD had sympathoexcitatory influences. Due to differences in lower vagally-mediated HPV compared to non-anxithe temporal kinetics of the autonomic neuroeffec-ous controls across all experimental conditions intors, sympathetic influences on cardiac control are cluding baseline. Importantly, a second main effect relatively slow (order of magnitude seconds) com-appeared which indicated that worry in both the pared to vagal influences (order of magnitude milli-GAD and non-anxious control group was associated seconds; see Saul, 1990). Thus, when the fast vagal with a reduction in HPV. modulation of cardiac function is decreased, the This tonic reduction in HRV in GAD and the organism is less able to track the rapid changes in phasic reduction during worry represents a breakenvironmental demands and less able to organize an down of the inhibitory influences that allow for appropriate response. In the remainder of this sec-efficient self-regulation including the shifting of tion, we will use our work on generalized anxiety attentional focus. Thus an excitatory positive feeddisorder to illustrate some of the implications of this back loop is allowed to emerge. disruption of the inhibitory pathway for psycho-GAD is characterized by excessive, unrealistic apprehension. This persistent state is supported by The capacity to attend to salient events and disregard attentional mechanisms such as hypervigilance, scan-irrelevant ones greatly enhances the viability of an ning, and a pre-attentive bias for threat information organism. Porges (1992) has proposed a two com- (Mathews, 1990). As such, the normally fine-tuned ponent model of attention, each component of which ability to adjust to changing environmental factors is associated with distinct cardiac activity. Specificalbecomes a rigid, inflexible response disposition.
ly, vagally mediated changes in phasic heart rate Lyonfields et al. (1995) examined vagally me-(HR) are associated with reactive attention, whereas diated HPV in an analogue sample of GADs and a changes in HRV are associated with sustained attengroup of non-anxious controls during periods of tion. Reactive attention is thought to reflect a funcimagery and worry about an individualized topic of tional 'tuning' of the organism to novel stimuli to concern. Resting baselines were recorded both at the allow for the appropriate perception, evaluation, and beginning and at the end of the experimental session.
response selection (Sokolov, 1963). Physiologically, The anxiety subjects showed reduced HPV across all reactive attention is associated with, among other recording periods with very little change from one things, cardiac deceleration. On the other hand, period to the next (see Fig. 2). This was taken as sustained attention is associated with vigilance and evidence of a lack of behavioral flexibility and an the suppression of HRV. The appropriate short-term inability to generate an appropriate response to the suppression of HRV and the associated focusing of changing environmental demands. Whereas the non-attention are important for effective self-regulation. anxious controls did show differences among the Tonic measures of HRV may reflect general experimental conditions, the worry condition was responsiveness to changes in the internal and exterassociated with the greatest reduction in HPV. This nal environments and, as an index of neurovisceral reduction during worry was greater than the reduc-integration, represents a measure of self-regulatory tion during imagery of the same topic.
ability. Diminished tonic HRV and the associated The ability to rapidly shift and effectively sustain reduction of vagally mediated cardiovascular control attention in accord with situational demands is a has been associated with a variety of pathological critical component of self-regulation (Porges, 1992).
states and dispositions (see Friedman and Thayer, 1998;Malliani et al., 1991;Stein et al., 1994, for 'tunes' the organism to novel stimuli (Sokolov, reviews). As a measure of vagally mediated car-1963). The A1 has been associated with a number of diovascular activity, HRV is an index of a negative functions including stimulus elaboration, the signal feedback mechanism that is important for the self-value of S1 and the response requirements of S2. regulation of behavior. Vagal activity has negative The D2 component is the most robust and has been cardiac chronotropic and dromotropic effects that associated with the anticipation of S2. This comserve to produce efficient cardiovascular functioning ponent has been shown to be enhanced (greater through the restraint of cardiac rate and electrical deceleration) prior to an aversive S2 (Somsen et al., conduction speed. This restraint or inhibition is 1983). necessary for cardiac stability, responsiveness, and The stimulus value of the S2 determines the flexibility (Levy, 1990;Verrier, 1987).
cardiac response subsequent to the S2. For a neutral The defensive attentional style that characterizes or appetitive stimulus, a HR deceleration is elicited. GAD is ultimately detrimental to functioning be-Importantly, for an aversive stimulus HR acceleracause it impairs the ability to flexibly respond to tion is elicited. This is thought to be associated with changing demands. It serves to reduce the range of a defensive reaction (DR) which serves to buffer the behavioral options by limiting the ability of the impact of the aversive stimulus on the organism organism to generate appropriate responses via a (Hare and Blevings, 1975). Jennings (1986) has compromised ability to inhibit inappropriate re-suggested that this HR acceleration may represent sponses. This impairment is evident even at a pre-motivated inattention and a cognitive avoidance of attentive stage of information processing. threatening information. Of particular note for our We have recently completed an experiment in model, these phasic cardiac changes found in the which we attempted to provide support for this S1-S2 paradigm have been shown to be vagally characterization of organism self-regulation through mediated (Porges, 1992;Somsen et al., 1983). attentional control, affective information processing, Porges (1992) has noted that cardiac vagal tone is and cardiac functioning. Thayer et al. (2000) ex-positively associated with the ability to self-regulate amined phasic cardiac responses in an S1-S2 via the modulation of attention. Given our previous paradigm in persons with GAD and non-anxious findings of reduced vagal activity in GAD (Lyoncontrols. The S1-S2 paradigm has been extensively fields et al. Thayer et al., 1996) and the used to examine the relationship between phasic HR reports of dysfunctional attention in GAD (Mathews, changes and attention. Briefly, this paradigm in-1990), we reasoned that non-anxious controls would volves the presentation of a series of paired stimulus have better attentional regulation relative to persons trials in which an initial cue stimulus (S1) is with GAD. This would manifest itself by individuals followed after a fixed inter-stimulus interval (ISI) by with GAD vigilantly monitoring their environment a second stimulus (S2). The second stimulus is for threat and failing to disengage from unimportant usually imperative in some way. Classical condition-events. Thus, persons with GAD would fail to ing is a type of S1-S2 procedure.
habituate to novel neutral stimuli whereas non-anxi-An extensive literature supports characteristic HR ous controls would show habituation. changes that accompany cognitive shifts over the Porges (1992) has also suggested that the magcourse of the ISI and in response to the S2. A robust nitude of cardiac responses is positively related to tri-phasic HR response has been described during the vagal tone. Therefore, relative to persons with GAD, ISI (Bohlin and Kjelberg, 1979;Somsen et al., non-anxious controls would show initially larger 1983). An initial HR deceleration following the S1 orienting responses (OR), even though they would (D1) is followed by a HR acceleration (A1) over the show rapid habituation over time. This is consistent next several cardiac beats. Finally, just prior to the not only with the reduced vagal tone found in GADs S2 a second HR deceleration occurs (D2). Each of but also with the restricted range of autonomic these phasic cardiac changes has been associated responses found in GADs (Hoehn-Saric and with specific cognitive activities. The D1 has gener- McLeod, 1988). Thus, across individuals, resting ally been interpreted as an orienting response which vagal tone should be positively associated with the magnitude of the OR. Given that the OR is important stimuli leads to a positive feedback loop that in for appropriate perception, evaluation and response essence spirals out of control. This leads to a very selection (Sokolov, 1963), a robust OR is an indica-broad basin of attraction for the 'worry' attractor. tion of a healthy and adaptive engagement of the That is, worry becomes the preferred response to an organism with its environment. ever-widening range of situations. The direction of phasic cardiac responses is also of Persons with GAD also showed HR acceleration importance. Cardiac decelerations, as noted above, to threat words and this response did not habituate. are associated with the OR and with anticipation of Persons with chronic anticipatory anxiety have been important events (Berg and Donohue, 1992). Cardiac shown to respond to perceived threat with HR accelerations are associated with the DR, stimulus acceleration that is characteristic of the defensive elaboration, or motivated inattention (Jennings, response (Graham and Clifton, 1966;Hare and 1986). Therefore, we also predicted that persons with Blevings, 1975;Sokolov, 1963). The sustained atten-GAD would show defensive reactions to threat tion that is associated with vigilance is accompanied words and that this response would not habituate. by a phasic decrease in HR variability (Porges, Non-anxious controls, on the other hand, would 1992). However, this garnering of attentional reshow an OR in early trials but would eventually sources constrains the ability to respond flexibly to show habituation. In addition, GADs would develop environmental challenges. The finding that persons a conditioned anticipatory HR deceleration to threat with GAD respond to threat-related words with HR words. This prediction is based on the findings of an accelerations fits the above characterization and attentional bias toward threat cues in GAD, an extends our previous findings of reduced HR variassociative-learning basis for the generalization of ability in GAD. This response may represent an threatening information that characterizes GAD, and attempt to shield against the impact of threat, as is the phasic HR deceleration found in anticipation of suggested for the DR elicited by electric shock in aversive stimuli found in non-anxious individuals non-anxious individuals (Somsen et al., 1983), or as (Somsen et al., 1983). The verbal-linguistic process-is observed in phobic reactions to relevant specific ing that is characteristic of worry and GAD supports stimuli (Hare and Blevings, 1975). An alternative this last supposition. Thus, verbal stimuli are per-interpretation is that the HR acceleration is indicative ceived as aversive to these individuals. of motivated inattention (Jennings, 1986), and thus The results were consistent with our predictions may be a form of cognitive avoidance of threat by and were indicative of attentional dysregulation in persons with GAD. Stormark and colleagues (Storthe GAD group as indicated by vagally mediated mark et al., 1995;Stormark and Hugdahl, 1996) have phasic HR changes. First, relative to non-anxious recently demonstrated the finding of an attentional controls, persons with GAD showed smaller cardiac bias toward threat information with a subsequent ORs, impaired habituation to neutral words, and the attempt to avoid processing of such information. He development of a conditioned anticipatory HR de-concludes that the same emotionally relevant celeration to threat words. A natural response to stimulus can promote both initial attentional orientpotential threat is the deployment of attention to scan ing and subsequent avoidance given the necessary the environment for relevant cues and anticipate and appetitive and aversive motivational processes. Thus, evaluate the risk. Early detection of threat-related ''emotional modulation of attention involves . . . cues allows an organism to either prepare an effec-both the engagement and disengagement components tive escape or modulate the impact of the danger. In of attentional orienting.'' (Stormark, 1996, p. 23). GAD, the threat is imagined, and immediate be- The failure to habituate to harmless stimuli enhavioral responses are precluded. Rather, protracted genders a rigid response disposition based upon the scanning and hypervigilance are maintained to guard defensive reaction. Again, this cardiac acceleration is against a perpetually threatening environment. To the the opposite of the cardiac deceleration associated individual with GAD, the world is perceived as a with the detection and evaluation of novel stimuli very dangerous place. Moreover, the inability to that allows for appropriate response selection. Ininhibit the allocation of attention to innocuous deed, evidence exists that anxiety and stress can decrease the number of alternative options that are terman and Cummings (1997) identified several explored thus further reducing the chance that dis-frontal-subcortical circuits and note that these circonfirming evidence from the environment will be cuits are modulated via two opposing parallel pathdetected and processed (Keinan et al., 1987). This ways connecting the striatum to output nuclei in the positive feedback loop serves, in part, to maintain basal ganglia (see Fig. 3). They describe these disorders of affect in the face of the large amounts of pathways as follows. Excitatory, glutamateric projecdisconfirming data the individual is apt to encounter tions from the frontal cortex to specific areas within in their daily living. However, the perceived threat-the striatum form the front-end of both pathways. ening nature of the environment may lead to a These pathways then diverge via striatal output restriction of behavior such that the individual is neurons in the caudate, putamen, and ventral striatum exposed to less and less novel, disconfirming in-into a direct pathway and an indirect pathway. The formation. This again is a positive feedback mecha-direct pathway is formed by inhibitory GABA fibers nism that perpetuates the existing, dysfunctional in the striatum to neurons in the internal segment of state. The prototypical example of such restricted the globus pallidus and the substantia nigra pars behavior is seen in agoraphobia where the individual reticulata. From here, GABAergic efferents are sent can no longer leave their own home.
to discrete areas in the thalamus, which relay in-These mechanisms could operate at a precognitive formation back to the same discrete areas of the or preconscious level. This is consistent with classic frontal cortex. The indirect pathway is formed by work on perceptual defense (see Mackinnon and GABAergic outputs to the globus pallidus externa, Dukes, 1962, for a review) and the psychophysiology GABAergic fibers to the subthalamic nucleus, and of attention (e.g., Graham and Clifton, 1966; glutamateric projections to the internal segment of Sokolov, 1963), as well as more contemporary the globus pallidus and the substantia nigra pars notions of pre-attentive discrimination and selective reticulata. These areas then project to specific processing of threat in anxiety (Mathews, 1990).
thalamic targets that again relay information back to Furthermore, this rapid mobilization of resources for the circuits' origins in the frontal cortex. action is consistent with the dynamical systems Importantly, the two striatal efferent projections characterization of emotion as an emergent response have opposing influences on the output neurons from or attractor driven by motivational factors (see the internal segment of the globus pallidus and the Globus and Arpaia, 1994; similar ideas have been substantia nigra pars reticulata which in turn have expressed outside of the dynamical systems frame-opposing influences on the thalamo-cortical feedback work by Lang et al., 1992).
circuits. The direct pathway excites the thalamus Our data provide further support that these processes are supported by cardiac activity. The CAN, and related functional networks, such as the rostral limbic system, provide the mechanism by which cardiac, affective, attentional, and behavioral responses are coordinated in the service of goal-directed behavior and adaptability. Moreover, the characteristics of this network are consistent with our view that emphasizes the importance of negative feedback circuits and inhibitory processes. The CAN and related frontal-subcortical circuits (Masterman and Cummings, 1997;Spyer, 1989) are reciprocally interconnected functional units composed of parallel pathways. These functional units are neurochemical- whereas the indirect pathway inhibits the thalamus.
damage to or dysfunction of the anterior cingulate A tonic GABA-mediated inhibition of the thalamus include akinetic mutism, certain forms of epilepsy, is produced by output neurons from the basal Tourette syndrome, schizophrenia, depression, anxieganglia. The direct pathway serves to disinhibit ty, obsessive-compulsive disorder, and aggression. (excite) the thalamus via inhibitory striatal efferents That the cingulate cortex is central to emotion is to the internal segment of the globus pallidus and the not new. At least as early as 1937, Papez (1937) substantia nigra pars reticulata. The indirect pathway proposed that the cingulate cortex was an integral serves to inhibit the thalamus via disinhibition of the part of a neural network or circuit that formed the tonically inhibited subthalamic nucleus that sends neuroanatomical basis of emotional experience. excitatory output to the internal segment of the However, most recent work has focused on the role globus pallidus and the substantia nigra pars re-of the cingulate cortex in information processing ticulata. Reduced variability in attention, affect, (Cabeza and Nyberg, 1997). cardiac, and motor behavior has been linked to We propose that the cingulate cortex, specifically disruption of these feedback circuits (Masterman and the anterior cingulate, serves as a point of integration Cummings, 1997;Spyer, 1989).
for visceral, attentional, and affective information that is critical for self-regulation and adaptability. This area appears to be associated with the conscious 5. Neural imaging of emotion and attention allocation of attention, intrinsic to which is greater inhibition of irrelevant information. Moreover, we Advances in our ability to observe the brain in feel that the evidence supports a central role for the action have shed much light on the neural substrates anterior cingulate in response selection due to its of behavior. In a recent comprehensive review of function as a source of negative feedback, via vagal both the animal and human literature, Devinsky et al. pathways, to the cardiovascular system. Recent (1995) propose a system of interconnected areas, neuroimaging and neuroanatomical studies have termed the anterior executive region, that ''assesses suggested that the anterior cingulate is comprised of the motivational content of internal and external three interactive regions. The rostral and ventral stimuli and regulates context-dependent behaviours'' regions are associated with affective, motivated, and (p. 279). The anterior cingulate has been singled out autonomic behavior, whereas the dorsal region is of this larger system of structures that includes the associated with response selection as well as pain. amygdala, the periaquaductal gray, ventral striatum, Mayberg (1997) has proposed that the rostral cinguorbitofrontal and anterior insular cortices. They late is particularly important for the integration of conclude that the ''anterior cingulate cortex appears visceral, attentional, and affective information. As to play a crucial role in initiation, motivation, and such, this area may play a critical role in the goal-directed behaviours.'' (p. 279). These conclu-coordination of the complex mix of cognitive, affecsions are based upon the role of the anterior cingu-tive, behavioral, and physiological concomitants of late in 'executive behavior' and its projections into normal affective states and dispositions. In the motor areas associated with visceral control. Affect, following, we will briefly overview the literature on cognition, pain, social interactions, response selec-the role of the anterior executive region, particularly tion, and autonomic activity are some of the func-the anterior cingulate cortex, in emotion and attentions associated with the anterior cingulate cortex.
tion. Then we will review some of our work on the Together, they help to guide the organism through its emotional, attentional, and autonomic correlates of environment in relative safety. Numerous studies activity of the CAN and the anterior executive region have indicated that damage to the anterior cingulate as revealed by positron emission tomography (PET) is associated with a wide variety of dysfunctions studies. including altered attention, flattened affect, auto- Reiman (1997) reviewed six studies in which PET nomic dysfunctions such as tachycardia, inappro-was used to investigate normal and pathological priate social behavior, and poor associative learning. emotions. A number of the areas within the anterior Clinical syndromes or disorders associated with executive region appeared to be associated with affective behavior. Of particular note for our studies The regulation of attention has also been linked to of anxiety disorders, the anterior cingulate was the anterior cingulate. The anterior executive region, involved with both normal and pathological forms of particularly the anterior cingulate, has been implianxiety. Specifically, Reiman states that the anterior cated in an anterior attentional system associated cingulate ''may participate in the conscious ex-with attentional shifting and response selection perience of emotion, the attentional or behavioral (Posner and Petersen, 1990). Cabeza and Nyberg response to an anxiety-provoking situation, the inhi- (1997) recently reviewed the PET studies on cognibition of excessive emotion, or the process of tion in normal subjects. They report that several monitoring the individual's emotional state in order studies investigating attention have found activation to make a personally relevant decision'' (italics in the anterior cingulate during tasks that involve added, p. 9). In addition, it has been reported that the competition between processing alternatives and anterior cingulate is involved in the processing of inhibition of competing responses. Thus, the anterior both positive (Dolan et al., 1996) and negative cingulate seems to be involved in response selection (Morris et al., 1998) facial expressions of emotion. and self-regulation via modulation of competing Viewing emotional pictures has become a very system inputs and outputs. This response selection fruitful technique in the experimental investigation of and self-regulation may extend to social interactions. emotion. Lane et al. (1998a) used this technique to Devinsky et al. (1995) note in their review that examine cortical activation during selective attending damage to the anterior cingulate has been associated to subjective emotional responses. Lane et al. with a host of social affective impairments including (1998a) had 10 healthy male participants focus their blunted affect, impulsivity, disinhibition, and poor attention on the emotional valence of a subset of social judgement (p. 291). pictures from the International Affective Picture Lane and Schwartz (1987) put forward a novel System (Lang et al., 1995) or focus on the setting idea concerning a cognitive-developmental model of (indoors, outdoors, or neither) of a subset of the social affective behavior. Their model of 'levels of same pictures. The former condition was thought to emotional awareness' treats the ability to make be associated with an internal focus of attention, socially appropriate judgements as an individual whereas the latter condition was thought to be difference in the capacity to experience emotion in a associated with an external focus of attention.
complex and differentiated manner. This ability Twelve PET derived measures of cerebral blood flow involves the selection of relevant social information were obtained from each participant. Six were re-and the inhibition of socially inappropriate cues and corded during the internal focus condition and six responses. Lane et al. (1998b) recently investigated were recorded during the external focus condition.
the neural correlates of levels of emotional aware-The internal and external focus conditions were ness using PET. They correlated scores on the Levels counterbalanced across subjects. Results indicated of Emotional Awareness Scale (LEAS; Lane et al., that during the external focus condition blood flow 1990) with cerebral blood flow during film-and increased bilaterally in the parieto-occipital cortex recall-induced emotions in a group of 12 healthy (Brodmann's area (BA) 19 / 37 and BA 39). Im-women. The results suggested that the anterior portantly, during the internal focus condition blood cingulate (BA 24) was associated with the accurate flow significantly increased maximally in the anterior detection of interoceptive and exteroceptive emotioncingulate (BA 32) with secondary activations in the al signals. These findings implicate the anterior medial prefrontal cortex (BA 9), right temporal pole cingulate in the response selection processes neces-(BA 38) extending into the frontal operculum and sary for the experiential processing and response insula, and of lesser magnitude in the ventral cingu-generation related to social affective information. late. These results support the early work of Papez Whereas the anterior cingulate has been highthat identified the anterior cingulate as central to the lighted here, it is clear that the anterior cingulate is 'experience' of emotion. Moreover, they are con-just one area in the larger CAN and rostral limbic sistent with the review of Reiman (1997) implicating system responsible for the central underpinnings of the anterior cingulate in the conscious experience of goal-directed behavior and adaptability. Vagally meemotion.
diated cardiac activity has been linked to attentional regulation and affective behavior. We (Lane and Thus, the perseverative mode of thinking that is Thayer, unpublished data) have recently investigated characteristic of worry that can lead to increased the neural correlates of heart rate variability. An anxiety is also associated with an increased risk for index of vagally mediated heart rate variability was cardiovascular disease. We have also demonstrated correlated with cerebral blood flow. The single that individuals that worry a significant portion of the highest correlation between heart rate variability and day show less physiological flexibility and adaptcerebral blood flow occurred in the insular cortex. A ability to changing task demands. secondary association was found between heart rate The poor orienting to novel stimuli, poor habituavariability and a subcortical region that included the tion to innocuous words, conditioned anticipatory thalamus. The association of these two areas with HR deceleration to threat words, and the HR acceleheart rate variability is consistent with the neural ration to the presentation of threat words represents a feedback circuits that comprise the CAN and the dysfunction of attentional and affective information associated rostral limbic system. The primary output processing that does not allow for the appropriate of the CAN is modulated via the preganglionic response to environmental demands. This inability to sympathetic and parasympathetic neurons. These disengage the threat detection or Defensive Behavior same neurons via input to the sino-atrial node are system serves to perpetuate arousal and anxiety even responsible for the complex variability that char-when no real threat exists. Thus, individuals with acterizes a healthy and adaptive cardiac time series.
low HRV are less able to detect and experience Thus, neural imaging data supports a direct associa-'safety' when it is in fact present. We (Sollers et al., tion between activity of the CAN and heart rate 1997) have recently shown that individuals with low variability.
HRV experience a blunting of subjective emotional reactions. Thus, low HRV is associated with poor affective information processing.

Summary
Importantly, we have shown elsewhere that following intervention with cognitive-behavioral We have presented evidence that autonomic con-therapy, a small sample of GAD patients made trol of the heart as measured by HRV is related to significant changes in the direction of increasing attentional regulation, affective information process-vagal cardiac control, with concomitant elevations in ing, physiological flexibility, and cerebral blood HR variability and reductions in anxiety, flow. A relative deficit in vagally mediated HRV was symptamotology, and HR (Friedman et al., 1993). found in individuals diagnosed with GAD. Impor-We are continuing to follow these individuals, as tantly, HRV was also decreased in both non-anxious well as additional patients, and now have 2-year controls and patients during worry. Anxiety disorders follow-up data on a larger sample with similar have been linked to excess cardiovascular morbidity findings. and mortality (Hayward, 1995). Traditionally, anxie-We have also presented evidence that links HRV ty has been associated with increased arousal and to the CAN and related neural circuits. Whereas sympathetic nervous system activity. However, we these preliminary results are promising and conhave suggested that the relative increase in arousal sistent with our model, additional studies are needed observed in various anxiety disorders is due to a to further specify the nature of this relationship. deficit in the inhibitory activity of the parasympa-Additional psychophysiological and neuroimaging thetic nervous system (Friedman and Thayer, 1998). studies using other attentional and cognitive tasks, Thus, the relative increase in arousal is a dis-inhibi-different aspects of affective information processing tion of sympathoexcitatory mechanisms of cardiac and expression, and further investigations of gender control.
differences in HRV, cognition, and affect are current-In addition, the decrease in vagally mediated HRV ly underway. during worry highlights the importance of verbal-We have shown that the neural circuits identified linguistic information processing. Kubzansky et al.
by other researchers to underlie autonomic regulation (1997) have recently shown that chronic worry is (Benarroch, 1997), attentional regulation (Devinsky related to an increased risk of coronary heart disease. et al., 1995), and affective regulation (Damasio, 1998) show substantial structural overlap. These cardiac vagal tone represents a psychophysiological various circuits most likely reflect a single structural resource that the organism can bring to bear, as it unit that subserves numerous functions associated were, on an environmental challenge (Friedman and with goal-directed behavior and adaptability. As . Third, framing the diverse self-regsuch, it forms the basis for an integrative self-ulatory functions and dysfunctions observed in terms regulation system (Benarroch, 1997).
of vagal processes as opposed to sympathetic processes may be a more parsimonious representation of the data (Friedman and Thayer, 1998; Thayer and 7. Conclusions Friedman, 1997). From this perspective, the relative sympathetic activation seen in anxiety disorders may Autonomic regulation, attentional regulation, and represent dis-inhibition due to faulty inhibitory affective regulation allow an organism to meet the mechanisms. challenges of an ever-changing environment. How-In sum, we have presented a model of neuroever, the importance of inhibitory processes in this visceral integration in emotion regulation and self-regulatory behavior has not yet made its way dysregulation that stresses inhibitory processes. We into the dominant thinking in this area. From a have placed this model in a systems framework systems perspective, inhibitory processes can be involving the interplay among positive and negative viewed as negative feedback circuits that allow for feedback circuits that allows for the parsimonious the interruption of ongoing behavior and the re-explication of affective behavior. These systems are deployment of resources to other tasks. When these active in the service of goal-directed and motivated negative feedback mechanisms are compromised, behavior. As such, they might provide the underpinpositive feedback loops may develop as a result of nings of a comprehensive model of the cognitive, dis-inhibition. These positive feedback loops pro-affective, behavioral, and physiological concomitants mote perseveration and continued activation of sys-of normal and pathological emotional states and tems thereby limiting their availability for other dispositions. processes. These positive feedback loops can have disastrous consequences. For example, at the level of the synapse, McGeer et al. (1978, p. 134) note that