Review
Autonomic nervous system activity in emotion: A review
Introduction
Autonomic responding in emotion has been an active research topic since, almost a century ago, Walter Cannon (1915) first studied the physiology of emotion (Brown and Fee, 2002, Dale, 1947). Still, there is no scientific consensus on whether there exists a relation between emotion and the organization of autonomic nervous system (ANS) activity and, if so, in what form. The various positions, which contemporary researchers hold on this topic, are first addressed in this article, before turning to the physical components—or the hardware—of autonomic responding in emotion. Next, a brief overview of the various theories and models that have been suggested to explain and identify mechanisms of autonomic response organization in emotion is given. The center part of this article consists of a review of the empirical basis for the postulate of emotion-specific ANS activity, considering 134 experimental studies on ANS activity in emotion. The next section summarizes and discusses how empirical emotion effects relate to models of autonomic response organization, points to the importance of choosing adequate measures of autonomic activation components, and addresses the issue of emotion terminology. A final section considers boundary conditions of the definition of emotion employed in the present article and its implications for identifying emotion-specific ANS activation.
Contemporary researchers in the field of emotion hold contrary positions on the topic of ANS activation in emotion. At one extreme, Feldman-Barrett (2006, p. 41), for example, stated that “it is not possible to confidently claim that there are kinds of emotion with unique and invariant autonomic signatures,” but rather that configurations follow general conditions of threat and challenge and positive versus negative affect. Feldman-Barrett named three points of critique regarding the evidence for autonomic differences between emotions: first, the high heterogeneity of effects in meta-analytical studies (e.g., Cacioppo et al., 2000) is interpreted to suggest the presence of moderator variables in the relation of emotion and ANS activity; second, autonomic differences that do emerge between specific emotions are viewed to be along lines of dimensional differentiation; and third, ANS activity is said to be “mobilized in response to the metabolic demands associated with actual behavior […] or expected behavior” (p. 41) and because different behaviors have been shown neither to be emotion-specific nor to be context-invariant (e.g., Lang et al., 1990), Feldman-Barrett views emotion-specific autonomic patterns as a priori improbable.
An intermediate position is suggested by meta-analyses of physiological responding in emotion (Cacioppo et al., 1997, Cacioppo et al., 2000) that report some degree of autonomic emotion specificity. Besides certain reliable differences between specific emotions, Cacioppo et al. also noted context-specific effects of ANS activity in emotion (i.e., according to different induction paradigms). Moreover, valence-specific patterning was found to be more consistent than emotion-specific patterning: negative emotions were associated with stronger autonomic responses than positive emotions (cf. Taylor, 1991). However, only one positive emotion, happiness, which subsumed joy, was used in the meta-analysis. This unequal representation of merely one positive as contrasted to a sample of five negative emotions may significantly bias the kind of distinction discerned. Due to a limited number of studies considered, a restricted range of physiological variables (only cardiovascular and electrodermal, but no respiratory measures), and the univariate nature of the meta-analytic approach, such results give only an imperfect answer to the question of autonomic patterning in emotion. Authors of review articles thus typically acknowledge that discrete emotions may still differ in autonomic patterns even if they do not differ in single variables (Larsen et al., 2008, Mauss and Robinson, 2009).
Diametrically opposed to Feldman-Barrett’s (2006) position, Stemmler (2009) argued why the ANS should not convey specific activation patterns for emotions, if those have specific functions for human adaptation. Stemmler, 2004, Stemmler, 2009 reasoned that emotions have distinct goals and therefore require differentiated autonomic activity for body protection and behavior preparation. Autonomic activity for behavior preparation is physiological activation that occurs before any behavior has been initiated that itself engages the ANS according to behavioral demands. Such autonomic activity has even been reported in experimentally paralyzed animals (Bandler et al., 2000), underlining that it is not merely overt behavior that causes this activity. This also resonates with Brener’s (1987) notion of “preparation for energy mobilization,” which contrasts to Obrist’s (1981) view of ANS activity as a component of the motoric response.
Stemmler (2004) reported on a meta-analysis on autonomic responding in fear and anger—two emotions that are believed to share similar valence and arousal characteristics—in which he found considerable specificity between the two. Taking a functional approach to autonomic responding in emotion, Stemmler, 2003, Stemmler, 2004 stressed the importance of studying autonomic regulation patterns in emotion rather than single response measures. According to the view that the central nervous system (CNS) is organized to produce integrated responses rather than single, isolated changes (Hilton, 1975), any variable which can be described or measured independently is constituent of several such patterns. Only when considering comprehensive arrays of physiological measures can such regulation patterns be discerned. Stemmler (2009) stressed that this should include variables that indicate both specific and unspecific effects of emotion. Unspecific emotion effects distinguish between control and emotion conditions, but not between emotions, whereas specific emotion effects distinguish between emotions. The pool, from which indicators of independent autonomic activation components can be drawn, is considered in the subsequent section.
Although physiologists at the beginning of the last century characterized the ANS as too slow and undifferentiated to quickly produce highly organized response patterns in emotion (Cannon, 1927), contemporary physiologists see considerable room for such organization (Bandler et al., 2000, Davidson et al., 2003, Folkow, 2000, Jänig and Häbler, 2000, Jänig, 2003; see also Levenson, 1988). Research over the past 50 years has invalidated the view that the sympathetic devision of the ANS functions in an ‘all-or-none’ fashion without distinction between different effector organs (Cannon, 1939). Rather, each organ and tissue is innervated by distinct sympathetic and parasympathetic pathways, with very little or no cross-talk between them (Jänig and McLachlan, 1992b, Jänig and McLachlan, 1992a, Jänig and Häbler, 2000). Pools of sympathetic neurons can be selectively engaged, such that individual systemic circuits or other effector units are independently activated (Folkow, 2000).
The originally assumed functional unity of the sympatho-adreno-medullary system is now known to consist of two separately controlled system parts—a direct-nervous and an adrenomedullary hormonal one—that under most situations have different functional roles (Folkow, 2000). Whereas the former executes precise, rapid, and often highly differentiated adjustments, the latter independently modifies important metabolic functions. In some emergency situations, where massive and generalized sympatho-adrenal system activation can occur, the two parts may also mutually support each other.
The inclusion of respiratory measures under autonomic measures also deserves some comment here. Respiratory activity evidences effects of autonomic control as well as significant independent contribution of peripheral and central chemoreceptors sensitive to CO2 (Wilhelm et al., 2005). Measures of respiratory activity may thus yield additional information on ANS functioning in emotion to that indicated by cardiovascular and electrodermal measures. There moreover exist important interactions of the respiratory system with the cardiovascular system, as, for example, attested by the phenomenon of respiratory sinus arrhythmia (Grossman and Taylor, 2007). Here, respiratory measures are important in the interpretation of effects of ANS functioning indicated by cardiovascular measures, which are modulated by respiratory effects. Finally, the cardiorespiratory control system can be viewed as one functional unit as it pursues the common aim of providing the tissues with oxygen, nutrients, protective agents, and a means of removing waste by-products (e.g., Feldman and Ellenberger, 1988, Poon and Siniaia, 2000, Taylor et al., 1999). Thus, comprehensive assessment of cardiovascular, electrodermal, and respiratory measures can provide complementary information on ANS functioning in emotion.
Central coordination of autonomic activity represents a cornerstone of current views of integrated nervous system functioning (cf. central autonomic network, CAN; Benarroch, 1993, Benarroch, 1999; see also Damasio, 1998, Thayer and Lane, 2000). Unlike the original conceptualization of the ANS as functioning independently of the rest of the nervous system (e.g., involuntary, automatic, and autonomous control), close interactions between the central and autonomic nervous systems exist in various ways. Thus, like the somatic nervous system, the ANS is integrated at all levels of nervous activity. Whereas segmental autonomic reflexes are coordinated by the spinal cord, suprasegmental integration higher in the brain stem is required for regulation of functions such as respiration, blood pressure, swallowing, and pupillary movement. More complex integrating systems in the hypothalamus influence the brain stem autonomic subsystems. Many of the activities of the hypothalamus are, in turn, governed by certain cortical areas, particularly the insular, anterior cingulate, and ventromedial prefrontal cortices as well as the central nucleus of the amygdala, that process inputs from the external environment. Thus, fundamental adjustments of the organism to its environment can only be attained by the concerted coordination and integration of somatic and autonomic activities from the highest level of neurological activity in the cortex down to the spinal cord and peripheral nervous system. This high degree of specificity in ANS organization is needed for precise neural regulation of homeostatic and protective body functioning during different adaptive challenges in a continuously changing environment. In this context, emotions may provide quick and reliable responses to recurrent life challenges. But still, the question remains how autonomic response organization in emotion might be achieved.
William James is often credited for originating the idea of peripheral physiological response specificity in emotion (e.g., Ellsworth, 1994, Fehr and Stern, 1970; see also Friedman, this issue, for a historical overview). James’s (1884) proposal that the feeling component of emotion derives from bodily sensations, i.e., the perceived pattern of somatovisceral activation, reversed the causality of emotion and bodily responding. Acknowledging a high degree of idiosyncrasy in emotion, James stated “that the symptoms of the same emotion vary from one man to another, and yet […] the emotion has them for its cause” (1894, p. 520). Even more so, James believed that the physiological responses were “almost infinitely numerous and subtle” (1884, p. 250), reflecting the infinitely nuanced nature of emotional life. Still, James recognized limits to bodily variations in emotion: “the symptoms of the angers and of the fears of different men still preserve enough functional resemblance, to say the very least, in the midst of their diversity to lead us to call them by identical names” (1894, p. 520, emphasis in original). James thus strongly argued for “a deductive or generative principle” (James, 1890, p. 448) that may explain physiological response specificity in emotion.
James’ claims associated with his peripheral perception theory of emotion were met with differentiated reactions—they instigated critique (most prominently the five-point rebuttal by Cannon, 1927), support (e.g., Angell, 1916), as well as various propositions for general organizing principles of autonomic responding in emotion. Although a number of different models have been proposed since then, these co-exist in a rather disjunct fashion, without clear empirical rejection of one or the other. As detailed in Kreibig (in press), the various models of autonomic responding in emotion can be organized by recognizing that these models address different conceptual levels, on which an organizing principle of autonomic responding in emotion may operate. Table 1 shows how the various theories map onto different conceptual levels that span from the physiological over the behavioral to the psychological level. A first class of models is identified, which draw on a basic physiological systems level; these are models that see the organizing principle of autonomic responding in emotion in the structure and functioning of the ANS or in the functioning of transmitter substances. A second class of models is based on brain–behavior interactions and views the organizing principle of autonomic responding in emotion in the functioning of brain–behavioral systems and refined behavioral modes. A third class of models centers on psychological processes of meaning assessment and memory retrieval; these models place particular emphasis on the functioning of psychological appraisal modules and associative networks as a general organizing principle of autonomic responding in emotion. A detailed discussion of the various models on each level can be found in Kreibig (in press). It is of note that from a component-view of emotion (Scherer, 2009), models on the same conceptual level rival each other. In contrast, models on different levels have complementary value, as they address different levels of response organization (cf. Mausfeld, 2003). It will be seen in the discussion section how these models fit with the empirical findings that are presented next.
Section snippets
Empirical findings of ANS activity in emotion
To what extent are postulated differences between emotion reflected in empirical data on ANS functioning? To address this question, a qualitative review of research findings was carried out, focusing on effects of experimentally manipulated emotions on ANS responding in healthy individuals. To cover both the psychological and medical literature, an exhaustive literature search using the databases PsycINFO, PsycARTICLES, and PubMed was conducted with the following search terms:
[emotion] and
Discussion
ANS activity is viewed as a major component of the emotion response in many recent theories of emotion (see Table 1). Different levels, on which an organizing principle of autonomic responding in emotion might be located, were identified in the introduction and the complementary nature of these approaches was pointed out. The empirical review compiled a large database that can be drawn on to evaluate such statements. What is the empirical evidence for positions of various degrees of ANS
Acknowledgements
Sylvia D. Kreibig is now at the Department of Psychology, Stanford University, Stanford, CA 94305. I thank Guido Gendolla, Klaus Scherer, and Tom Cochrane as well as two anonymous reviewers and special issue editor Bruce H. Friedman for helpful comments on earlier versions of this manuscript. Thanks also to research assistant Nora Meier for assisting in the literature search. This research was supported by the Swiss National Science Foundation (PBGEP1-125914) as well as by the National Center
References (303)
- et al.
Physiological responses to sexual and unpleasant film stimuli
Journal of Psychosomatic Research
(1972) - et al.
Basic emotions evoked by odorants: comparison between autonomic responses and self-evaluation
Physiology and Behavior
(1997) - et al.
First evidence for differential and sequential efferent effects of stimulus relevance and goal conduciveness appraisal
Biological Psychology
(2007) - et al.
Central circuits mediating patterned autonomic activity during active vs. passive emotional coping
Brain Research Bulletin
(2000) - et al.
Columnar organization in the midbrain periaqueductal gray: modules for emotional expression?
Trends in Neurosciences
(1994) The central autonomic network: functional organization, dysfunction, and perspective
Mayo Clinic Proceedings
(1993)The effects of emotional behaviour on components of the respiratory cycle
Biological Psychology
(1998)- et al.
Fear of pain and defensive activation
Pain
(2008) - et al.
Autonomic specificity of discrete emotion and dimensions of affective space: a multivariate approach
International Journal of Psychophysiology
(2004) - et al.
Watching emotional movies: affective reactions and gender differences
International Journal of Psychophysiology
(2008)