An Enactive Approach to Anxiety and Anxiety Disorders
This article sketches in which way clinical practice and the science of anxiety and anxiety disorders could benefit from an enactive approach. It starts with the clinical phenomenology of anxiety and anxiety disorder. It describes why traditional theories of emotion have difficulty doing justice to aspects of anxiety that are important in psychiatric practice, most importantly the contextual embedding and the self-referential nature of manifestations of anxiety. The article shows how an enactive approach provides a possible route to do justice to these two dimensions of anxiety, both theoretically and practically. Finally, I situate the enactive approach conceptually, by interpreting it as a primarily philosophical paradigm with strong suggestions for both the theory and clinical practice of psychiatry.
Anxiety, Anxiety disorder, Enactive, Enactivism, Phenomenology, Concept of anxiety, Emotion theory
Enactive approaches to emotion are rare and to anxiety and anxiety disorder even more (see, however, Lewis, 2005; Colombetti, 2014; Colombetti & Thompson, 2008). This article aims to show how an enactive paradigm might be helpful in solving some problems in the clinical and scientific understanding of anxiety and anxiety disorder. I begin by pointing at a number of relevant clinical features of anxiety and anxiety disorder and by sketching how and why anxiety theories have difficulties with doing justice to these features. I specifically focus on two themes: a) how to conceptualize anxiety as the expression of a dynamical interaction with the environment instead of as just a reaction to the environment (or to internal cues); and b) how to make sense of the self-referential nature of anxiety as an emotion. The enactivist paradigm offers novel insights that may bring psychiatric theory further at precisely these two points. It highlights the embeddedness of anxiety in a shared biological, sensory, cognitive, and social world. And, despite using a different vocabulary, it also recognizes the self-referential dimension of anxiety.
The article closes with an attempt to locate the enactive approach conceptually. I interpret enactivism as a primarily philosophical approach, which is situated closely enough to scientific theory to provide strong enough suggestions with respect to methodology and hypothesis building. This in turn helps psychiatrists to solve some of the well-known tensions between scientific theory and clinical practice, particularly those concerning the two main issues in this article: the embeddedness of symptoms of anxiety and the self-referentiality of anxiety as an emotion. [End Page 35]
Symptoms Are Not Just Symptoms—Anxiety as Embedded Phenomenon
Anxiety is not only a very common, but also an elusive emotion. It has an enormous array of manifestations, ranging from hardly noticeable worries to storms of bodily sensations and from complete paralysis to conditions with severe bodily and mental agitation.
One important thing a clinician needs to know is whether the anxiety of the patient needs treatment. The dominant underlying model to answer this question is biomedical: treatment is needed if there is a disorder; there is disorder if there is a dysfunction; there is a dysfunction if there are sufficient symptoms that count as manifestations of the dysfunction. The criteria to decide about symptoms—their number, their nature and their intensity—are listed in classification systems, like the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2013). So, there is a close relationship between diagnosis and classification. Symptoms are the corner stone for both.
This general picture needs more nuance, of course. One of the nuances is that clinicians tend to have their own 'clinical' pictures of disorders. These pictures have a prototypical rather than a categorical structure. The clinical picture of certain illnesses differs, in other words, from the neatly distinguished categories of illness/disease in classification manuals like the Diagnostic and Statistical Manual of Mental Disorders and the International Classification of Disease in that they consider illnesses as clusters, or patterns, of overlapping symptoms. Clinical diagnosis is based on a form of pattern recognition rather than on counting symptoms. Experienced clinicians take signs and symptoms not just as bare facts, but tend to interpret them from a more holistic perspective: not every form of fatigue is a symptom of depression; there are kinds of fatigue that are typical for depression, and others, which are not.
Other nuances concern the notion of symptom itself. Are symptoms the invariant expression of an underlying dysfunction? Or, are they something else? Even within the biomedical model there are different ways in which a symptom can be a symptom. Some symptoms (for instance, redness caused by an infection) are immediate expressions of an underlying problem and others (like fever) are reactions to such problems. Some symptoms come from within and are called 'primary' or 'endogenous'; others are reactions on influences from without and are known as 'secondary' or 'reactive.' The biomedical model, furthermore, differentiates between positive and negative symptoms, which corresponds with the distinction between wrong functioning and absence of functioning. This model also recognizes that causal chains leading from dysfunctions (patterns, causal factors) to symptoms differ widely in length and complexity. Some chains are long and much more intertwined with other chains than others, which are shorter and more uniform.
One of the most important problems in current psychiatric research is to determine a preferred target level of description of symptoms/signs for causal factors to relate to. Influential researchers have lost their trust in the traditional disease oriented, categorical approach. The National Institute of Mental Health has issued a new research paradigm which focuses on research domains criteria, specifying broad clusters of dysfunctioning, which are mapped onto a wide variety of predisposing factors and their 'markers' (National Research Council, 2011). Research domains criteria puts, in other words, less emphasis on boundaries between clusters of symptoms and focuses, instead, on groups of symptoms which are loosely grouped together and associated with heterogeneous sets of possible predisposing factors (Insel, 2014).
This conceptual move might possibly not be enough, however. Despite its more realistic starting point, research domains criteria still leaves the underlying disease model largely intact. Symptoms are still caused by underlying factors within the individual. This approach precludes a more radical view which conceives symptoms as co-constructions of influences from outside and weaknesses and forces inside.
I take symptom here in the traditional medical sense as a ('subjective') sensation or feeling indicating that something is wrong; in contrast to 'signs,' which are supposed to indicate illness 'objectively,' [End Page 36] not via first-person awareness (Sims, 1988). Symptoms, however, are not only subjective, they are also intersubjective. They are not only first-person expressions of something within the individual, they are also a reflection of a learning history in which others play an important role. Bodily feelings too are socialized and have a developmental history. The way they are expressed, dealt with, and even felt, is shaped by commentaries, interpretations and valuations of others, from the earliest beginning of the individual.
Symptoms are not just symptoms, not only because their meaning depends on intersubjective interpretation, but also because they themselves are products of a range of external (contextual) and internal (person-related) factors. To the internal factors belong coping skills, personality, and cognitive processes. These factors influence the way symptoms are felt and expressed. The perception of symptoms, in turn, is molded by external factors like education, socialization, and manipulation. Something similar holds for other external factors such as public awareness of illness and of risk. All these factors may interfere with one's functioning in different roles. This interference not only influences the perception of symptoms, but also what clinicians count as symptoms: the interference with daily functioning should be severe enough for a symptom to count as the manifestation of an illness.
Clinical practice, furthermore, shows that symptoms are often layered. There is an initial sensation, feeling or thought. This feeling is accompanied by a certain inclination to act and often also by certain (other) bodily manifestations. This initial complex may openly be faced, molded, suppressed, or mentally circumvented, depending on the way it is perceived and interpreted. The success of dealing with the initial symptoms depends on capacities that are based on the kind of context-dependent learning that was just mentioned. Reactions of others, including medical doctors, play an important role in coping with the symptoms.
Miss C, 26 years old, is a perfectionistic administrative worker, who has run into trouble after a reorganization at her work. Two departments have fused, there is a new manager, much less supportive than her former boss. The working climate has deteriorated, with hidden jealousy and competition between the colleagues. Old feelings of insecurity and insufficiency have recently re-emerged. On a morning, after having noticed that she has made a small mistake in her work, she feels very tense, with tingling of the fingers, shortness of breath, and the feeling of fainting. She has the inclination to go home. She does not say anything to her boss and her colleagues and, indeed, goes home. At home, the worrying does not stop. She starts weeping, cannot stop, calls a friend, who advises her to take (sedative) pills and look for positive distractions. The next day she goes to the general practitioner who advises her to stay at home for two weeks. He also prescribes sleeping pills.
This short vignette illustrates the gradual building up of layers in the experience of anxiety. The problems begin with feelings of insecurity (first layer), which, after a minor incident, are accompanied by body sensations (choking, tingling, the sensation of fainting) and a tendency to flee (second layer). Miss C has a passive and avoiding coping style, related to deficiencies in her early socialization and in the formation of her personality. This coping style contributes to the stress and anxiety. She takes her tenseness and feelings of insufficiency and insecurity as proof of not being suitable for her job, instead of as sign of incompatible demands and of inappropriate standards in her working place. This misinterpretation enhances the anxiety (third layer) and is insufficiently addressed by both her friend and the general practitioner, who both, despite their good intents, in fact reinforce her tendency to avoid (fourth layer).
In short: symptoms of anxiety are not merely expressions of underlying dysfunctions. They develop over time and are the product of interaction between factors within the patient and in the environment. They are, moreover, modulated by personality characteristics, coping style, and commentaries by important others. The biomedical model has difficulty to do justice fully to this complexity. We investigate whether the enactive paradigm is better able to do justice to the embeddedness of the symptoms of anxiety. We also briefly address the question whether, and if so how, enactivism could have an impact on treatment. [End Page 37]
Self-Referentiality—Anxiety and the Self
Let us focus now on a second theme which appears to be important in the clinical context: the self-referentiality of anxiety. A short vignette may help to give an impression of what is meant with this term.
Mr. A is a 43-year-old salesman. He is married and has two children, boys of 8 and 10 years old. He consults a psychiatrist because of panic disorder. He has initially successfully overcome the panic with a combination of medication and 12 sessions of cognitive behavioral therapy. During a follow-up session he tells his psychiatrist about an unexpected recent attack, which understandably has disturbed him. During the cognitive behavioral therapy, he has learned to recognize certain triggers for the attacks: work-related stress, arguments with his spouse, and feelings of loneliness. Some of these triggers played a role in the genesis of the most recent attack.
Mr. A is an obsessive worker with an unhappy childhood. His current feelings of loneliness resonate with episodes in his childhood, in which he felt unhappy, helpless, and alone. His parents separated when he was young, his mother was an alcoholic and frequently depressed, he was her only child and she was never involved in a new intimate relationship. The panic disorder began after conflicts at his work. There was already a longer history of marital problems, circling around his obsession with work and his lack of involvement in family life. The recurrence of the panic seems related to increased tensions in the marital relationship.
The intervention of the psychiatrist goes roughly along the following lines:
You are panicking in a situation in which you feel under pressure and in which no solution seems available. Your wife asks for more involvement with her and with your kids. But your boss is also more demanding. You feel stuck between incompatible demands and there seems to be no way out. Your anxiety is indicating this. At the same time, old feelings of unhappiness and powerlessness are recurring. This might be an indication that there is some weak spot in your personality, I am still not sure about that, however. So, what I want to discuss with you is how much the anxiety is saying about you, your personality, and how much it does say about the situation you are in.
A new aspect comes to the surface here: anxiety is not just a state or disposition that results from exposure to an external or internal danger. It also has self-referential qualities, in the sense that the anxiety (also) reveals something about the person having the anxiety. Anxiety may unveil for instance one's vulnerability for being left alone or being separated from important others (as in the case of Mr. A). Or it may indicate one's sensitivity for criticism; or one's fear for failure on an important task. A significant part of the job of the therapist consists in sorting out what (and how much) the anxiety says about the situation and what (and how much) it says about the patient.
This balancing of self- and situation-related aspects is typical for the evaluation of emotions in general, and in clinical situations particularly. Sadness is the consequence of some loss (situation), but it also indicates how important the lost one has been for the person suffering from sadness (self). Shame occurs in embarrassing situations, but also indicates what kind of things I am embarrassed about and what I try to hide for others. Emotions, in other words, are not only directed at an object or situation, but also refer to the person having the emotion. They are both object-directed and self-directed or self-referential; self-referential in the sense that they refer to my interests, the things I care about, my vulnerabilities, longings, and predilections. This self-referentiality is also present when the person having the emotion is not aware of the emotion.
Self-referentiality is not a very common notion (see, however, Slaby & Stephan, 2008; Stephan, 2012; Ratcliffe, 2008; Glas 2017; 2019). It is a term which has been coined by the French philosopher Paul Ricoeur (1990/1992), who uses the term in a narrative and normative sense and applies it to (speech) acts, narratives, and societal interactions (not specifically affective life). Self-referentiality should be distinguished from the broader category of self-relatedness, which primarily has to do with stance-taking (Glas, 2006; 2019, chapter 4). In Ricoeur's work this distinction is not always clear, but for the purposes of this article [End Page 38] we need it. Self-referentiality is a form of signifying and not of relating. Emotions refer to, that is, signify, particular aspects of the persons who are having the emotion, just like speech acts and narratives refer to the persons who are speaking and narrating. This self-referring does not occur as a result of some experience or activity of the person who feels (acts, narrates), it is given with emotion itself. Self-referentiality is neither a form of self-regulation nor a form of self-observation. It is self-signification. This means that the emotion implicitly signifies certain aspects of the self. Stance-taking is different, it is adopting a stance toward what one is feeling, saying or doing. There is a certain degree of deliberateness in stance taking, which is lacking in self-referentiality. Self-relating comes close to self-regulation. Self-regulation is what one does with one's emotions, either as a preparation for their possible occurrence (e.g., by influencing the situation); or as a way of coping with emotion once it is there; or by modulating one's emotional response. Self-referentiality is more basic, it precedes self-regulation both theoretically and factually. Self-referentiality refers to the intrinsic and often implicit capacity of emotions (but also gestures, attitudes, interactions) to signify something about the person having them, whether this person is aware of this or not. Self-referentiality does not require action, nor even consciousness, it is given with the emotion, immediately and implicitly. There exists a distinction between different forms of self-referentiality: Primary self-referentiality refers to immediate referring to an aspect of the self; secondary selfreferentiality to implicit self-referring in the way the patient deals with her emotions and moods; and tertiary self-referentiality to self-referring in the way the relation to these emotions and moods has become internalized.
Can there be situations in which emotions do not have this self-referential quality? I will address this question briefly by considering the analogy with the object of emotion. Philosophers tend to deny that there are emotions without an object. Kenny (1963), for instance, states that emotions without an object are in fact sensations. Pain is not about something; it is just a feeling state. One can feel pain on a certain spot, but that does not mean that the pain is 'about' the spot. There are, of course, situations in which I do not know what my emotion is about or in which the object of emotion is vague. Vagueness, however, does not rule out object-directedness. I may fear a 'dark future,' but even then the fear is about something, a particular aspect of the future, its 'darkness.' Cases in which I simply do not know what the emotion is about, are usually cleared up later, with the help of others or by self-investigation (Kenny, 1963, 60–61).
A similar line of reasoning is defensible with respect to the self-referentiality of emotion. Emotions are by definition self-referential. If there is no self-referentiality, even after attempts at clarification, the putative emotion is just a physiological state or a sensation. This is not just a theoretical possibility. Panic disorder is a telling case: in severe panic attacks there is sometimes no intrinsic object, but just only a storm of physical sensations. The accompanying thought that one could die from a heart attack or from suffocation is a (secondary) self-interpretation, it is not self-referencing of the panic itself.
To summarize, anxiety is an emotion which not only refers to an object but also to a self (or, aspects of it), that is, the person having the emotion. Self-referentiality is intrinsic, just like the object of emotion. It is critical for adequate understanding of what happens in clinical practice. Anxiety not only indicates something about the body or the brain of the patient, but also about who the person is who suffers from anxiety. Self-referentiality also, and indirectly, helps to discern why there is an evaluative dimension in the assessment of anxiety. After all, the self-referring element in anxiety needs to be weighed against what anxiety says about the situation of the patient. Is the anxiety proportional to the situation? Does it predominantly express something about the world or about the person having the anxiety? This weighing requires a sound view on what realistically might be expected to occur in similar cases. The requirement of 'realism' adds a normative dimension to the assessment of anxiety. Whether or not anxiety is realistic depends on how others look at it and on whether I can give the right reasons for it. People constantly scan their environment and attune to the views of others. They absorb the implicit and [End Page 39] explicit interactions between people, including their gestures, attitudes, verbal expressions, and implicit suggestions. This scanning is embedded in and presupposes a shared world of interpretations, evaluations, and opinions. So, in the clinical situation the self-referential aspect of anxiety is conceptually inescapable; without it, clinicians would never be able to say something about the person with anxiety. Nor would they be able to say anything about how realistic the anxiety of this person is.
Contextuality And Self-Referentiality in Emotion Theory and Anxiety Research
Two big, clinically relevant themes have emerged, so far—the contextual embeddedness of symptoms of anxiety and the self-referential dimension of anxiety. We now briefly review the three main paradigms in anxiety research and investigate what they have to offer with respect to these two themes. We see that there is a gap between theory and practice in this respect. In the next section, we investigate whether enactivism has the resources to fill this gap. As a philosophical paradigm it offers important suggestions for scientific research and hypothesis-building, especially with respect to the contextual embeddedness of symptoms of anxiety and their self-referentiality (albeit in different words).
Before discussing this, let us first briefly review some of the main traditions in anxiety research, as a preparation for the next section. Broadly speaking, there are three of such main traditions in the scientific study of anxiety (see also Barlow, 2002; Glas, 2013):
(1). A Darwinian, ethological tradition in which anxiety is seen as a universal, biologically anchored survival response, which has become sensitive to a range of environmental cues via conditioning and other learning processes; generally speaking, most neurobiological research fits in this tradition, which has been very influential in psychopharmacology (Kandel, 1983, 2005; LeDoux, 1996).
(2). A strong empiricist, initially mainly behaviorist research tradition, which views anxiety as a theoretical term denoting observable behavior that is considered to be the expression of activation of different neural and/or behavioral systems (Gray, 1982; Lang, 1979, 1985). Within this tradition anxiety is not by definition a unite phenomenon, but composed of building blocks like physiological symptoms, behavior manifestations and verbal behavior. Most animal research is performed within this framework (Kalueff, Wheaton, & Murphy, 2007).
(3). A cognitive research tradition which focuses on anxiety as inner experience, influenced by expectancies, (mis) interpretations, (biased) attention, and cognitive schemes (Barlow, 2002; Beck, Emery, & Greenberg, 1985; Rachman, 1978).
According to contemporary leading psychologists and cognitive neuroscientists these traditions do not exclude one another (Lang, Davis & Öhman, 2000; LeDoux, 2012). The neurobiological tradition (ad 1) has begun to merge with the psychological/behaviorist paradigm (ad 2) for more than three decades now (Gray, 1982; LeDoux, 1996, 2002; Kandel, 1983, 2005). More recently, the cognitive approach (ad 3) has been re-phrased in a way that makes it more suitable for joining the other two research traditions (Barlow, 2000; Clark & Beck, 2010; Lang et al., 2000; LeDoux, 2012).
These three main traditions and their interactions should be placed against the background of fundamental problems in the philosophy of emotion and emotion theory. One of these problems concerns the perennial question whether emotions are natural kinds or the result of environmental influences. According to the 'natural kind' approach there exists a relatively small set of basic emotions, which are universal and cross-culturally similar (Izard, 1977). This approach corresponds with the Darwinian, ethological tradition, mentioned under (1). However, according to the other empiricist/behaviorist paradigm, mentioned under (2), emotions should be viewed as acquired behavioral responses that are susceptible to conditioning, modeling, and vicarious learning (Strongman, 2003, chapter 4; Saarni, 2008; for a philosophical overview, see Goldie 2010). Current conceptualizations combine these two approaches by assuming that fear, as basic emotion, is mediated [End Page 40] by environmentally sensitive neural circuits in which the amygdala plays a key role (LeDoux, 2002). These circuits are considered to be responsible for conditioning, modeling, vicarious learning, and other learning processes. From a conceptual point of view, this theoretical approach leaves the idea of basic emotions intact: fear reactions remain more or less invariant behavioral and physiological responses. However, they are now thought to be modulated in a much subtler way, by the learning (and cognitive) processes just mentioned. In short, fear remains a more or less fixed behavioral response, however, one that is susceptible for environmental modulation and fine-tuning via learning processes.
There remain questions, however, especially with respect to the supposed fixed character of the basic emotions. These questions specifically pertain one of our main topics: the contextual embeddedness of emotion. Are fear reactions, even the basic ones, so uniform? Or, could there be more behavioral and physiological variation, depending on variation in the context? Would it be possible to defend a position in which even basic emotions, or subspecies of these basic emotions, are fundamentally products of interactions? There are some indications to this effect (Scarantino & Griffiths, 2011). Current animal research suggests that even at the basic level of emotional responding there is more context-sensitivity than until recently has been assumed (Kalueff et al., 2007; Pawlak, Ho, & Schwarting, 2008). Developmental neurobiology and attachment theory show that the development of affect occurs in bio-behavioral shifts. Each shift represents a new level of organization of the maturing brain with concomitant reorganization of behavioral responses (Schore, 1994). There is, in other words, empirical support for a more varied, contextually sensitive, and developmental view on emotion, in which basic emotions are more than just invariant behavioral responses triggered by limited sets of environmental cues.
With respect to the other topic of this article, the self-referentiality of emotion, this aspect has not been overlooked in emotion theory (Slaby & Stephan, 2008; Stephan, 2012; Ratcliffe, 2010), but has been addressed in different terms and from a different angle. Frijda (1986), for instance, defends the thesis that emotions reveal our interests. They embody what is important to us, by making manifest what we cherish and what we are concerned about (see for a similar approach Goldie 2000). Applied to anxiety this means that anxiety manifests our interests in things like physical integrity, possession of physical and/or mental space, or being valued by others. Social anxiety reveals that it is important for me that others do not disapprove or humiliate me. If it would not matter what others think about me, even not in my imagination, I would never suffer from social anxiety. Similar for panic disorder, post-traumatic anxiety, and other anxieties. The panic in panic disorder is related to the importance of physical integrity and well-being. In post-traumatic stress, it embodies the interest of feeling safe, physically as well as emotionally.
Embodying an interest is conceptually not the same as self-referentiality, but it is an empirical approximation of it. Self-referentiality is caught in a theory about how emotions help us to protect our interests. Emotions reveal our concerns, not so much symbolically, as a form of signification, but psychobiologically, by providing us tools to defend and position ourselves. These tools are given with the emotion; they are parts of it. Anxiety reactions are embodiments of an alarm system, these reactions make us alert, focus our attention, and prepare us for action. Anger makes us ready for fight. Sadness embodies an inclination to withdraw, physically and/or socially, when we are confronted with a loss which makes us vulnerable.
More specific references to the self and to selfhood can be found in the work of Damasio, who, after decades of scientific neglect, provides an account of the self and of subjective aspects of emotion. Damasio (1999, 2010) makes a firm distinction between feelings and emotions. Feelings are subjective and emotions objective. Emotion terms refer to observable patterns of behavior and physiology, whereas feeling terms denote what is subjectively felt. The development of feeling is for Damasio predominantly a matter of developmental neuroscience. To have feelings a person should have an elementary awareness of self. This elementary sense of self is based [End Page 41] on interoceptive sensations, awareness of one's global organismic state and information provided by the sensory portals to the world, that is, the body parts that are involved in object-perception (visual, auditory, tactile, kinesthetic). These sensations are clustered, integrated and represented in what Damasio calls the proto-self (Damasio, 2010, 190–201). Partially apart from and partially intertwined with the developing proto-self, there is a constant transient flux of pre-reflective sensations and representations of body states under the influence of current environmental challenges and demands. These sensations provide affective quality to the stream of consciousness, which, at this stage, has no thematic content. Thematic content emerges when both objects and the protoself begin to stand out against their respective backgrounds. Objects gradually acquire salience. The perception of objects transforms into a 'feeling of knowing the object.' The proto-self, in the meantime, develops into what Damasio calls a 'protagonist,' a core-self that stands out against the primordial images and feelings that formed the basis of the proto-self. Finally, another capacity develops, the capacity of knowing that one feels, the reflective awareness of one's feeling, the feeling of feeling. This second-order feeling requires awareness of oneself. Feelings are then no longer just representations of the physiological condition of the organism. They require a connection with the self. This self is initially the core-self, which in its turn is rooted in the proto-self (Damasio, 2010, 201–209). Later on in one's life, a more sophisticated, autobiographical self develops—a self that is based on memories that are personally significant and colored by feelings that are associated with body experiences during significant life events (Damasio, 2010, p. 210ff). The memories of such bodily feelings are integrated into one's body self, that is, the landscape of images, sensations, feelings, and body-memories, that helps us to navigate through the social world and to face the challenges of human existence.
This intriguing conception—which in some respects resembles William James's theory of emotion—gives, in fact, an explanation of the origin of the awareness of mine-ness of feelings (James, 1884). The mine-ness of feelings, again, comes close to the notion of self-referentiality, though it is not identical with it. This is, basically, because (self-) signification does not require consciousness, nor awareness, even not in the form of a (first-person) perspective. More broadly speaking, we recognize in Damasio's work an awareness of the embodied and embedded aspects of emotion and consciousness; an awareness which is so vivid and consistent that it resembles Merleau-Ponty's concept of the corps-sujet or lived body. Damasio's account, however, also offers an example of what happens when these intuitions are squeezed within more traditional conceptual frameworks. These frameworks are typically either introspectionistic/dualistic or epiphenomenalistic. Emotions remain after all objective, organismic states or dispositions. These states or dispositions are seen as only secondarily molded by environmental input. Feelings are then conceived as either epiphenomena of neuronal developments or as perceptions of body states. Damasio comes close to the notion of self-referentiality as it has been defined in this article, but strictly speaking his conceptual framework has no place for this notion. Emotions are related to the proto-self or the autobiographical self, but this relatedness is either conceptual or experiential and not a form of signification. The relation of emotions to the self is conceptual when Damasio suggests that emotional expressions by definition are integrated within body experiences and memories about a self. The relation between emotion and the self is merely experiential when he states that feelings imply a form of awareness of the self. Both interpretations fail to see that there is a (self-) signifying dimension in emotion even without awareness. Emotions do not only say something about a self when a person is aware of herself in her feelings. They also signify (aspects of) the self when there is no such self-awareness. Behind this looms, of course, a certain dualism (introspectionism, representationalism)—a dualism that comes to expression in the division of tasks between feelings and emotions and in the (implicit) assumption that referring to a self requires consciousness of the self.
The Enactive Paradigm
Let us now move on to enactivism and investigate whether this theoretical approach brings us [End Page 42] any further with respect to the conceptualization of the embeddedness of the symptoms of anxiety and of the self-referential nature of these symptoms.
I give, first, a brief review of what is meant with the term enactivism. Then, I discuss what the enactive approach to psychopathology (see also Guastello, Koopmans, & Pincus, 2009; Stewart, Gapenne, & Di Paolo, 2010) and to emotions and feelings in general could mean (Colombetti, 2014).
Enactivism is one of the more recent branches of dynamical systems theory. It can be characterized by a cluster of terms: autopoiesis, self-organization, autonomy, circular causality, emergence, structural coupling, and (participatory and other forms of) sense-making. I will explain these terms.
The concepts of self-organization and autonomy are based on ideas that first were developed by Maturana and Varela (1980; see also Varela, 1979) in their book on autopoiesis. Autopoiesis (literally: 'self-making') refers to the mutual dependence between parts and wholes, or, more technically, between the "network of dynamic transformations that produces its own components and that is essential for a boundary" and the "boundary that is essential for the operation of the network of transformations which produced it as a unity" (Maturana & Varela, 1987, p. 46). Maturana and Varela are referring here to the morphodynamic genesis of the cell, as prototype for living things. The cell is characterized by circular organization of metabolism (dynamics) and membrane (boundary). This circular organization leads to unity, that is, distinctness with respect to a background. The use of the term autopoiesis is restricted to processes in which there exists a material boundary. Autonomy is the broader concept and extends to systems with no visible (material) boundary (Varela, 1979).
Self-organization refers to the spontaneous emergence of order from interactions among the components of a dynamic system. Interactions between components are typically bidirectional (reciprocal; recursive), whereas the emergence of order occurs nonlinear. Non-linearity means that the new order is not in a linear way derivable from preceding causes. Self-organization implies that top-down and bottom-up processes go together and are complementary. 'Top' refers here to the whole and 'down' to the constituent parts. There is, in other words, not first an interaction between components at the constituent level and subsequently emergence of order at a new, supervening level. Both go together: top-down and bottom-up relations are aspects of the same process, they are co-constitutive for one another, without causal interaction between 'levels.' It is in this complementary interaction that new forms of order emerge. When the interaction between systems exerts a structuring influence on the stability and functioning of each system, this is called 'structural coupling.' In other words: the boundaries of the system are co-defined by the interaction with other systems.
From the late eighties, these ideas were applied to cognitive neuroscience and psychology, first in a book on embodied cognition (Varela, Thompson & Rosch, 1991), later in a study by Thompson (2007). Varela's notion of structural coupling returns in Thompson's account of emergence as dynamic co-emergence. Dynamic co-emergence means that what emerges in the interaction between systems, is simultaneously a 'self' or 'individual' and a correlative niche: the domain or set of interactions possible for such a system given its organization and concrete structural realization (Thompson, 2007, 419–420). Dynamic co-emergence also (more or less) coincides with the concept of circular causality.
Thompson's account of emergence follows from what Varela says about self-organization and structural coupling. There is a network of interrelated components, which exhibits an emergent process. This process instantiates emergent properties as a consequence of the coupling of the components plus the nonlinear dynamics of their interaction. The emerging properties are not determined by the intrinsic properties of the components. It is the other way around: the constant rearrangement of the components defines both the components and the system. Thompson calls this 'relational holism.' Wholes and parts are defined by relations, rather than by intrinsic properties of the components (or the system).
It will come as no surprise that this works for systems with no or little decomposability. Non-decomposable systems [End Page 43] (Bechtel & Richardson, 1993) are systems of which the components have lost their identity and are no longer separable. The components lack intrinsically defining characteristics. Emergence usually occurs in systems that are not or minimally decomposable. From an outsider perspective it is indiscernible in such cases whether the components (for instance: local neuronal activities) emerge from the whole or the whole (dynamic patterns of large-scale neuronal integration) from the components (Thompson, 2007, p. 423). From an insider perspective, this question does not matter; there is just an immanent movement (in time) toward emergence that does not require any extra external influence to occur.
Sense-making, finally, refers to the fact that the organism is not the passive recipient of sensations that inform the organism about the environment. Organisms perceive their environment by interacting with it, in a bodily, sensory, and cognitive way. It is at this point that dynamical systems theory touches phenomenology. What Thompson and others say about sense-making comes close to Merleau-Ponty's notion of the lived body. This lived body is no longer an instrument for the fulfilment of the intentions of the mind. It is itself subject of action, it is corps-suject (Merleau-Ponty, 1945). Thompson's phenomenologically inspired enactivism entails that perception (input) and action (output) are intertwined ('coupled') and cannot be understood without the other (cf. also Gallagher, 2005; Noë, 2004, 2009). Participatory sense-making (De Jaegher & Di Paolo, 2007) refers to the joined sense-making of autonomous agents.
Thompson's account relies for its empirical support to a considerable extent on the work of Kelso and others on brain dynamics (Kelso, 1995; Le van Quyen, 2003). Kelso (1995, 257) considers the brain as "fundamentally a pattern forming self-organized system governed by potentially discoverable, nonlinear dynamical laws." This means that "behaviors such as perceiving, intending, acting, learning, and remembering arise as metastable spatiotemporal patterns of brain activity that are themselves produced by cooperative interactions among neural clusters" (Kelso, 1995, 257). With this statement, Kelso leaves primitive localiza tionism as well as Cartesian representationalism behind: the neuronal correlate of a certain activity is spread-out all over the brain; and neuronal firing has a pattern-like structure, it does not offer an image of one's role in the world.
The reference to non-linear dynamics implies that brain processes lack the predictability of linear systems and should be analyzed in terms of their behavior in phase space, which is the geometric representation of all possible states of the system. The behavior of the system appears then as the curve or trajectory in phase space. Non-linear approaches to brain processes can also be found under the heading of complexity theory and chaos theory (Kauffman, 1993). Nonlinearity, however, is not the same as randomness. Systems that behave non-linear are not unstable, but metastable or dynamically instable, indicating that they exist at the edge of instability where they can bifurcate into two (or more) different states. This dynamic instability is thought to add to the system's flexibility and adaptability (Kelso, 1995, p. 22).
Enactive theory has been applied to biology, immunology, sensorimotor coordination, kinesthetic perception, consciousness, attention and other cognitive processes (Guastello et al., 2009; Stewart et al., 2010). So far, however, relatively little has been written about emotion (see, however, Freeman, 2000; Scherer, 2000; Lewis, 2005; Thompson, 2007; Colombetti, 2010, 2014) and even less about anxiety (Glas, 2003). What follows is an attempt to show how the enactive paradigm could become relevant for our two areas of concern: the embeddedness of symptoms of anxiety and the self-referentiality of the emotion of anxiety.
Application—The Enactive Approach and the Embeddedness of Symptoms of Anxiety
Earlier, it was observed that symptoms of anxiety are not merely expressions of underlying dysfunctions but also always products of interaction between factors within the patient and between the patient and her environment. We have seen that more recent research on anxiety and emotion leaves room for a more malleable conception of [End Page 44] symptoms and of emotion, but that the underlying conceptual framework has remained traditional in the sense that this malleability is still conceived as secondary, that is, as a modulation on a more or less fixed repertoire of basic emotional responses. Can the enactive paradigm inform anxiety theory in such a way that it can do justice to clinical wisdom that says that symptoms are contextually embedded; that they are not just products of an underlying brain mechanism, but always (also) the product of interactions at the surface; interactions with contextual and person-bound factors?
I think the enactive paradigm can do this job—to a certain extent and within certain boundaries. As I see it, enactivism broadly supports clinical wisdom and it provides suggestions that are sufficiently detailed to be relevant for empirical research. However, it remains a philosophical theory; enactivism does not replace clinical knowing and, despite its relevance for empirical research, it does not offer itself an empirical theory. Having said this, it should be clear that there are still considerable distances to bridge, first between enactivism as philosophical paradigm and empirical science; and then between enactivist-informed empirical science and clinical practice. Let me give an impression of a possible route, a route that gives an impression about the conceptual steps that need to be taken to make sense of anxiety and anxiety disorder as enacted phenomena. I work my way bottom-up, again, from the clinical phenomena to theory and philosophy.
What is needed, first, is an overview (sketch, logical map, conceptual model) of the course of events that leads to the emergence of particular types or episodes of anxiety. Such a model or map can help to overcome one-sided unidirectional causal modeling of anxiety and anxiety disorder, which has been typical for much theorizing about emotion. Think of Miss C with her feelings of insecurity, which became entrenched within an amalgam of uncontrollable body sensations, worrying, misinterpretations, and reactions from the environment, reinforcing her avoidant coping style. It is possible to map all these factors and this series of events as a network of phenomena which are bi-directionally connected with one another and develop in time. Lewis (2005) sketches such a model for the relation between appraisals and the physiological and motor aspects of emotion in general. There is mutual interaction between perception and bodily arousal, between bodily arousal and attention, between perception and action tendency, between action tendency and attention, and so on, until, finally, the complete appraisal-emotion amalgam has evolved. The reconstruction of this network can also inform decisions about treatment. Relaxation exercises (to diminish bodily arousal) will not prove to be very helpful when there are many negative cognitions, as is the case with Miss C. And treatment of Miss C's cognitions should be realistic in the sense that it requires an environment that allows for such better appreciation of who she is and how she is functioning.
A second step consists in the application of the concept of self-organization, as a basically philosophical concept, to this process. It is an attempt to find an empirical translation and define an empirical version of this concept. Anxiety could, for instance, be conceived as an emergent phenomenon in a temporally evolving network which includes all kinds of environmental and person-bound factors; a phenomenon which is amenable for analysis in terms of self-organization. I see at least two possible directions for the application of the concept of self-organization: a diachronic and a synchronic approach.
The diachronic perspective understands symptom formation as evolving over time and describes it in terms of a learning history. This learning history begins already in the preverbal period. It consists of an interaction between parental personality characteristics, educational practices, and temperament of the child. It leads to patterns of sensorimotor coupling and of interaction (Colombetti, 2014; Schore, 1994). Think of Miss C, whose deficiencies in coping style could be traced back to early socialization processes; to parents who were overly protective; to explorative behavior which was discouraged, nonverbally and verbally; and who not had learned to cope with unexpected events and negative comments by peers.
The synchronic perspective describes the occurrent, short-term dynamics between internal and external factors and between bottom-up and [End Page 45] top-down processes. It is this synchronic perspective that Lewis (2005) has in mind in his thorough analysis of the unfolding of emotional episodes in time. This short-term unfolding can be analyzed in terms of dynamic systems theory. Lewis (2005) describes such episodes as characterized by triggering events, self-amplification, self-stabilization, and learning processes. The triggering event is a perturbation that initiates a phase transition, that is, a temporary change of the order of the system, which in its turn activates certain 'control parameters,' that is, sensitivities of the system for particular events. Attentional bias in anxiety disorder is a well-known example of such a change in sensitivity (Mathews, May, Mogg, & Eysenck, 1990). Self-amplification refers to positive feedback loops (and feed-forward processes) within the organism. The activation of one element leads to the activation of others and forms the beginning of a recursive cycle, which sustains itself above a certain threshold. Self-stabilization occurs when negative feedback cycles balance the self-amplificatory tendencies. Self-organization typically has to do with the balance between positive and negative feedback. Long lasting emotional episodes may, finally, initiate a learning processes in which typical reactions to, for instance, threatening events lead to the genesis of behavioral patterns and/or personality types, exhibiting, for instance, a tendency to inhibition and/or avoidance.
Such insights may also have an impact on treatment. Psychotherapeutic interventions may themselves become triggering events which initiate phase transitions. Such interventions have already a name in the psychotherapeutic literature: they are known as 'corrective emotional experiences.' Interventions targeting at such a transforming experience require a certain level of emotional arousal in the client and consist, usually, of a remark or interpretation which is sufficiently empathic, but also sufficiently disturbing. Knowing how to time and to phrase such intervention is very difficult to explain and to plan beforehand. It is part of the tacit knowledge and competence of the experienced professional.
The third step could consist of making distinctions with respect to the time scale (Lewis, 2000; Thompson, 2007). Emotional episodes, like the ones just discussed, typically last from seconds to hours. Mood states usually take days to months. Personality traits exist almost lifelong. Each time scale brings its own dynamic with it, especially a particular balance between activity and receptivity. Emotions usually diminish in intensity when the subject acts according to the action tendency that characterizes the emotion. Moods and personality traits, on the contrary, do not resolve as a result of action. Moods can be seen as enduring intentional orientations, whereas in personality traits this intentional orientation acquires a connection with the sense of self. The subject no longer has a depressive mood, she is depressed. These temporal distinctions have also therapeutic significance. People sometimes talk about their personality as if they have this personality ('my depressive me') as if it were a temporal thing or entity they have to deal with. Such quasi-reification may help people to get grasp on their functioning, for instance in early stages of the treatment. But it may also obscure how ingrained the depressive reaction pattern has become and that it has become part of oneself. In such cases the ingrained reaction pattern influences the way the client deals with her depression. Not seldom, it is more important to deal with these ingrained depressive reaction types than with the depression 'as such.' How people feel, is in such cases more determined by the ingrained depressive reaction than by the depression 'as such.' Clients should be learned to recognize this. They, in other words, should be learned how they can come closer to their 'selves,' by focusing on the implicit, emotionally colored self-referentiality of their moods. Immersion in the self-referentiality may, then, gradually replace the quasi-objectifying of the 'depressive self.'
One final step could be to translate this rough framework to the field of neuroscience and provide a cognitive neuroscience account of anxiety and anxiety disorder on the basis of the enactivist paradigm. A rare and ingenious attempt to such an account for emotions in general can again be found in Lewis (2005). [End Page 46]
Application—The Enactive Approach and The Self-Referentiality of Anxiety
What about the notion of self-referentiality? Does the enactive paradigm offer a richer conceptual framework to address this aspect of anxiety than more traditional theories? Self-referentiality means, as we saw, that anxiety refers to (or: signifies) an aspect of the self. Although caught in different terms and concepts, enactivism indeed offers a conceptual framework that enables clinicians and researchers to acknowledge and do justice to the self-referential quality of emotions. The enactive approach even extends this (self-referential) perspective by introducing concepts like valence, salience, value and sense-making and by applying these concepts to a whole range of behaviors, including the behavior of animals and even bacteria. Part of these valences and values are connected with the organism that owes the emotion (instead of with the object of emotion).
Hans Jonas (1968), who inspired Maturana and Varela, describes the urge to individuation as the most fundamental expression of self-concern which is inherent to all biological life. Both Jonas and Frijda refer, in this context, to Spinoza's conatus essendi, the fundamental drive to exist and to preserve oneself. Emotions, in other words, are driven by self-concern. We already discussed Frijda's conception of emotion as (a.o.) embodying one's interests. From here, it is not difficult to imagine that there are many types of valences, values, and opportunities for sense-making; and that many of them are related to aspects of the self. These values and valences can be put in a scheme or list with an increasing complexity. Sense-making may be related to personal survival, but also to maintaining a certain image of oneself. Sense-making may result from emotions that evoke certain memories—memories which explain why I have become the person I am. But emotions may also indicate my aesthetic preferences and what these say about me.
The self-referential aspect of emotion could, furthermore, be conceptualized by taking into account theories about the development of 'the' self. Earlier, we noticed how, according to Damasio (2010), feelings result from couplings between representations of organismic change and representations of elementary forms of self-awareness (the core-self). We saw how later in life a much more differentiated, autobiographical self develops that is based on significant memories and feelings of body states during important life events. These ideas can be reconceptualized in enactive terms by interpreting the development of feeling as a form of meta-synchronization between two independent self-synchronizing systems, one monitoring the ongoing stream of object-related sensations and the other integrating sensory experiences of oneself (kinesthetic, tactile, auditory, visual) together with memories and images. Later on in life, with the emergence of the autobiographical self, feelings become subtler and more differentiated and help to mold and refine these primordial feelings and emotions.
One of the merits of enactivism is philosophical, finally. It consists of the explicit rejection of the idea that for an experience to be my experience, the person needs to have a (separate) consciousness of him- or herself. The mine-ness of experience is phenomenologically given with the experience itself. It is not the result of an introspective act directed at an inner representation of (an aspect of) myself, nor the product of a tacit form of self-reflection. It is a perspective that is given with the experience itself and that is anchored in our bodily constitution. We look with our eyes, we touch with our skin, and hear with our ears and it is because of this embodiment that we have appropriated feelings and have learned to experience them as ours.
To summarize, the self-referential aspect of emotion can be approached by paying attention to how emotions embody values that belong to the self and by making distinctions between different types of values. Another route is more theoretical. It attempts to construe self-referentiality as a resonance of self-related experiences within the overall experience of emotion. The relative autonomy of these self-related experiences could be conceived as the result of co-synchronizing of two different systems by an overarching form of synchronizing. Feeling one's own feeling would then mean that there is an overarching feeling state (the feeling of feeling) that co-synchronizes both [End Page 47] the neural networks that mediate one's sense of self (the feeling of 'mineness') and the networks that mediate the ongoing stream of situationally bound sensations and feelings.
Discussion and Conclusion
The ambitions of enactivism are both bold and far-reaching. This is not the place to discuss these ambitions in general and the objections that have been raised against them. I instead focus on the merits and weaknesses of enactivism in the present context: for the study of anxiety.
As the previous two sections have shown, a lot can be won by applying the enactive framework to the scientific study of anxiety, especially with respect to the two main subjects of this article: the contextual embeddedness of symptoms of anxiety and the self-referential qualities of these symptoms. This is one of the great merits of the enactive paradigm. Another merit is that it supports and philosophically legitimizes clinical intuitions about anxiety and anxiety disorder as embedded and self-referential phenomena; and that it helps clinicians to withstand one-sided biomedical models according to which anxiety disorder is just a brain disease. The conceptual innovation of the enactive approach is considerable and important. Consequent rethinking of anxiety from a contextually embedded and self-referential perspective helps to escape from both objectivistic reductionism (which views anxiety as just an organismic state) and subjectivist introspectionism (which sees anxiety as just an inner state, unconnected with the body and the world). Traces of these older frameworks appeared to be present even in the work of Damasio, who in his intuition, as we saw, came close to the ideas that lie at the basis of enactivism (and existential phenomenology). With enactivism we envision a theory of emotion and of anxiety in which the first-person perspective is not added to an already existing, objective framework, but wired in, as it were. Enactivism suggests that this wiring-in can be reconstructed by combining first, second, and third person methodologies.
I pointed briefly to the most important caveat, that is, the risk of blurring of the distinction between philosophy and empirical theory. As I see it, enactive theory has empirical implications, but it is not itself an empirical theory. It offers an intriguing and intricate conceptual framework that helps to understand how emotional episodes unfold in time and how self- and object-directed aspects of emotion co-evolve. It has given strong impulses to conceptual renewal in scientific emotion theory. It has also been instrumental in the reviving of scientific interest in the experiential (subjective) aspects of emotion. But as a body of theory it remains philosophical. Its fundamental concepts are not only highly abstract, but lend itself to translation into a wide variety of fields, from mathematics, physics, and molecular biology to neuroscience, psychology and the humanities. This rich potential for translation should not seduce us to endorse the idea that enactivism should move in the direction of a science of everything. Enactivism's key concepts do open-up scientific imagination and have fruitfully been applied in a number of scientific fields. But we should remain aware of the fact that this applicability is based on the analogical power and fruitfulness of enactivism's basic concepts. Analogical relations differ from identity relations. Self-organization in biology may behave similar to, and therefore show analogies with, self-organization in the social world, but it cannot be held to be identical with it. There exists a certain irreducibility in the key concepts and explanatory frameworks of the various sciences. Enactivist theory has generated a number of concepts that overarch more than one field of science and of which the analogical use is powerful enough to actually influence hypothesis building in different fields of science. This is no small result. But this should not bring us so far as to embrace the idea that the concepts in the special sciences are just extensions of key concepts within enactivism.
Gerrit Glas is professor of Philosophy of Neuroscience at Amsterdam UMC and holds an endowed chair in Christian philosophy at the Vrije Universiteit in Amsterdam, the Netherlands. He is also practicing psychiatrist and director of residency training in Dimence Groep, hospital for mental health care in the province of Overijssel (Netherlands). He publishes on topics at the interface between psychiatry, philosophy, neuroscience, ethics, and religion. His most recent book is Person-centered Care in Psychiatry. Self-relational, Contextual, and Normative Perspectives (Routledge 2019).
The research for this article was co-funded by a grant from the Templeton World Charity Foundation.