Emotional mechanisms of social (re)production

Structuration

In the course of social life, institutions present themselves to experience as those objective and constraining entities that Durkheim had in mind when delineating the ‘social’ or ‘moral’ fact. Giddens’ problem is to relate this experience of the social world as objective and constraining facticity to a conception of the social agent as actively producing and reproducing that social world. Inspired by various sources such as interpretive sociology, hermeneutics, Garfinkel’s ethnomethodology, and Goffman’s studies of the interaction order, the view of the social agent as an active producer of their social world that emerged in the 1960s seriously took to task the conception of the agent on which sociological functionalism and structuralism, including its Marxian variants, was and is based. To put it bluntly, this is the idea that ‘all men and women … are bloody silly’ (Thompson, 1968: 40). Giddens intended to synthesize these various developments, emphasizing the active character of human agency with a more classical sociological concern with the experienced fixity of social institutions. The challenge is to articulate an account in which the constraining objectivity of institutions is achieved not in spite of, but rather thanks to, active human agency.

Even though he discards the functionalist (and structuralist) picture of the human agent as an automaton, Giddens does not entirely jettison one of the central themes of functionalist sociology, namely the preoccupation with the unintended consequences of action as a principle of institutional (or ‘system’) reproduction. However, he proposes to renew the analysis of the socially reproductive effects of unintended consequences of action by freeing it from the sway of functionalist concepts. The underlying premises are that agency is the capacity to transform the environment, and that what the agent effectively transforms through their agency may not be entirely known to the agent. That the agent is not aware of all the consequences of their action, and particularly of those consequences related to the reproduction of social institutions, does not mean that the agent is moved by external or unconscious ‘forces’. It simply means that the understanding that the agent has of their action, at least in respect of its social reproductive implications, is bounded or incomplete. One important task of social science is therefore to demonstrate the intimate connection between situated actions and the larger spatial breadth and temporal endurance of ‘objective’ social institutions, especially in those cases in which agents do not clearly see the socially reproductive consequences of their situated actions. Giddens’ additional remark, and one I wish to endorse, is that the understanding that agents hold of the consequences of their actions is not without effect for the continued reproduction of social institutions. Awareness of the social-reproductive consequences of their actions may lead agents to act differently in the future if they happen to deplore the institution whose reproduction they have unintendedly facilitated in the past. Social science may contribute to precipitate this cognitive and pragmatic shift. ²

This conception of the task of social science is incompatible with the comfortable division of labor between ‘macro-’ and ‘micro-sociology’. As Giddens points out, this way of speaking of the relationship between institutions and situated actions is misleading insofar as it carries with it the connotation that macro- and micro-sociology, as ‘sub-fields’ of the discipline, are concerned with essentially distinct phenomena. But institutions do not exist independently of, or reproduce themselves in spite of, situated actions: they are those situated actions, to the extent that the latter can be seen as ‘the same’ actions across space and time. ³ This becoming ‘stretched’ of situated activities across wide spans of time-space defines what Giddens calls structuration.

Conversely, the fleeting encounters that instantiate ‘the interaction order’, one of the central notions in what is usually called micro-sociology, cannot be rendered intelligible without reference to longer-lasting and more widespread institutions or ‘social practices’. Obvious examples are the rules of turn-taking and the face-saving techniques that confer on situated gatherings their recognizable shape as social interactions.

Stating that institutions and action are conceptually interdependent is not terribly informative, though. What I take Giddens’ originality to be in this respect is his formulation of the problem of social order as being concerned with the relationship between the integration of social interaction, on the one hand, and the integration of larger ‘systems’ or institutions, on the other (but see Mouzelis, 1989, for a criticism of this move). In order to do away with the dichotomy of the ‘macro’ and the ‘micro’, Giddens reformulated Lockwood’s (1976) classical distinction between social and system integration. In this perspective, it is in the process of skillfully ‘doing’ social interaction in a particular setting (understood as a spatial and temporal locale) that institutions or ‘systems’ spanning a range of settings are produced and reproduced. In other words, it is as a consequence of the successful ‘integration’ of social interaction that the ‘integration’ of social institutions (i.e. structuration) takes place. When the socially reproductive effects of ‘integrated’ face-to-face encounters are unknown to their participants, those effects can be considered as unintended consequences of action. Alternatively, agents may be aware of the socially reproductive effects of their actions and even attempt to ‘steer’ the process reflexively.

In this respect, Giddens invites us to go beyond the usual confinement of micro-sociologists to the here-and-now situation, but also to avoid the usual blindness of macro-sociologists with regard to the inescapably situated enactment of ‘social structure’ by active agents. Pace the tenants of functional analysis, the social agent is not an unreflective automaton moved by external or unconscious ‘forces’. But pace the proponents of micro-sociology, neither are the consequences of their actions necessarily bounded by the local setting of any particular social interaction episode. Active social agents, in the skillful production of local interaction, reflectively or unintendedly bring about consequences that produce and reproduce larger social institutions.

I follow Giddens in the formulation of the general problem of ‘structuration’, but I am less inclined to think that he has provided any convincing solution. In general, he has been more preoccupied with mechanisms of social reproduction than with mechanisms of social change, which in practice reduces structuration to reproduction. The concept of ‘routinization’ is a case in point that I address in detail in the following section, given the connections it suggests between day-to-day activity, institutions, and emotions. In particular, I think that the slogan of the ‘duality of structure’ has made disproportionate noise as compared to its theoretical fecundity. In Giddens’ glossary, ‘structure’ does not refer to any actual institutions or social systems spanning space and time but to ‘rules and resources’ that have by definition only ‘virtual’ existence, in the same way as for Saussure langue is virtual as compared to the actuality of parole. Defined in this way, ‘structure’ is dual because it is at once the medium and the outcome of situated action. But as Sewell (1992) observes, Giddens’ notion of ‘structure’ as ‘rules and resources’ is inconsistent, because even though ‘rules’ (generalizable procedures) can be easily thought of as ‘outside of space and time’, no sense can be made of ‘resources’ without reference to human bodies and physical objects, which by definition occupy space and endure through time. This flaw undermines the ‘theorem’ of the duality of structure of which the notion of structure is a building block. It is important to note that the ‘duality of structure’ theorem is conceptually independent from the formulation of the general problem of structuration. The shortcomings of the former do not immediately invalidate the latter.

The problem then is accounting for the fact that the active production of interaction in spatio-temporally bounded settings has as a consequence the production and reproduction of longer-lasting and more widespread institutions, ‘practices’, or ‘social systems’. This is the problem of structuration.

Routinization

The notion of ‘routinization’ points to processes of social reproduction actively sustained by agents through their repetitive day-to-day activities. Routinization binds institutions, everyday activity, and the psychological needs of the human agent. The notion is relevant in the present context because it explicitly accords an active role to emotions in social reproduction.

Drawing on Erikson’s (1962) theory of personality development, Giddens’ thesis is that ‘ontological security’, or ‘trust’ or confidence in the continuity of the world, is a basic psychological need that agents can satisfy only through routine day-to-day activity. In a sense, routinization is a cunning of institutional reason: agents seek ontological security through habitual conduct, and in their striving they bring about social reproduction, with higher or lower awareness of the socially reproductive consequences of their habitual conduct.

Giddens states that habitual action is indirectly motivated by the need to sustain security and, concomitantly, to avoid the emotion of anxiety. This generalized motive is further assumed to be unconscious in the Freudian sense that the agent is not able to put it into words. The need for ontological security thus defines the ‘deep’ motivational component of Giddens’ ‘stratification model’ of individual personality, avowedly inspired by Freud’s classical id, ego and super-ego triad. In this model, the middle ‘stratum’ is occupied by practical consciousness, which operates in circumstances in which people pay attention to events going on around them so as to relate their conduct to those events. On top of the stratification model lies discursive consciousness, which designates the agents’ ability to provide linguistically articulated reasons for their actions. Whereas the need for ontological security defines the general unconscious motive of the agent, practical and discursive consciousness point to the bases of their ‘knowledgeable’ or ‘reflexive monitoring’ of action. The unconscious pushes blindly, and consciousness steers lucidly and skillfully.

Giddens sees in Goffman’s (1967) ‘face-work’ (which he renames ‘tact’) an important mechanism for the maintenance of ontological security, and interprets Garfinkel’s (1967) ‘experiments with trust’ as the demonstration that breaches of routine bring to the surface of consciousness that deep unconscious anxiety which agents actively seek to curb through day-to-day activity. He observes that another context for examining the ontological security-providing function of routine is the counterfactual study of ‘critical situations’, e.g. the personality changes that Nazi concentration-camp prisoners underwent as a result of the drastic disruption of their accustomed ways of conduct.

There can be no quarreling with the idea that the integration of personality and the perpetuation of day-to-day activity are interdependent, but I feel less persuaded by Giddens’ assimilation of the distinction between (pushing) motive and (steering) knowledgeability, on the one hand, to the distinction between the unconscious and the conscious, on the other. Why assume that motive is unconscious? Colin Campbell (1996) recalls two traditional meanings of motive, the one referring to that which ‘motivates’, ‘causes’, or ‘moves’ an action, and the other to the actor’s reasons for acting, i.e. their end-in-view. However, in the history of sociology, the final or end-in-view acception has obliterated the causal or dynamic understanding of motive. Campbell sees this state of affairs as resulting from a mistaken alignment of the dynamic meaning of motive with behaviorism, as though embracing the familiar idea that motives push towards action amounted to disavowing the project of interpretive sociology with its stress on meaning and voluntarism.

Giddens seems to build on these assumptions when he says that, while day-to-day conduct is ‘reflexively monitored’, it is ‘not directly motivated’. He does not entirely do away with the causal meaning of motive but neither does he integrate it with the ‘knowledgeable’ side of the human agent. In positing that the avoidance of ontological anxiety is an unconscious motive, he preserves all the advantages of a causal explanation of habitual action without questioning the widespread but mistaken view that the causal of motive is irrelevant, if not radically inimical, to the project of interpretive sociology.

The price of this move is high, however. For the explanation of habitual action as relying on unconscious motivation surreptitiously brings Giddens back to the picture of the human agent that he wished to discard in the first place. In lieu of the automaton of functionalist and structuralist sociology, moved by outer social forces, we find the automaton of psychoanalysis, moved by inner unconscious motives. Explaining action, a fortiori routine, as a result of unconscious motivation does no good to the construction of an account of enduring and widespread institutions in terms of situated active agency.

But the unconscious motive that Giddens has in mind is also wanting from an empirical point of view. His attempt to squeeze the results of Garfinkel’s breaching experiments into the stratification model of the agent is not particularly convincing. First, one may wonder in what sense, as examples of what is involved in loss of ontological security, Garfinkel’s isolated and after all innocuous disruptions of everyday scenes are comparable to the systematic and devastating effects of concentration-camp life. But even if one overlooks this dubious parallel, it is hard to see how the manifold emotional reactions that Garfinkel reported can be plausibly reduced to the expression of undifferentiated anxiety. For example, Garfinkel notes that ‘reports were filled with accounts of astonishment, bewilderment, shock, anxiety, embarrassment, and anger’ (Garfinkel, 1963: 232). Everybody knows from common experience that the understanding of the situation and the motivation to act characteristic of astonishment and anger, to take only these examples, are largely different. What theoretical progress do we make by diluting their specificity in an undifferentiated category of diffuse anxiety? ⁴

That unconscious motives bring us back to a passive picture of the agent, and that anxiety is inadequate to the description of breaches of routine does not mean, however, that motives cannot be invoked to explain action, and that emotions are inadequate to account for social control. It instead means that we need to assume neither that motives are unconscious, nor that anxiety is the be-all of emotion.

In fact, drawing on phenomenology and appraisal theory (Frijda, 1986), emotions can be conceived of as motivational states of which the agent is always aware, in the sense that there is something ‘it is like’ to be in an emotional state (Lambie & Marcel, 2002). In Giddens’ terms, emotional experience always presents itself in practical consciousness. The agent may have second-order awareness of the first-order experience of the emotion state, and thus the agent may be able to report on the phenomenology of (what it is like to be in) that emotion state. ⁵ In Giddens’ schema, this means that in some cases the experience of emotion may be accessible to discursive consciousness. Since the experience of emotion involves among its components the experience of a motivational state, it follows that emotional motives are always experienced in practical consciousness and sometimes available in discursive consciousness. Emotional motives need not be lodged in any inaccessible unconscious realm. They are better conceived of as states of the whole organism that carry with them the subjective experience of an urge to act and get outwardly expressed in visible bodily changes.

Again drawing on phenomenology and appraisal theory, in lieu of the indifferentiation of context-blind anxiety, the differentiation of emotions can be thought of in terms of the characteristic understanding of the situation that they imply or the kind of action that they solicit. In Frijda’s (1986) theory, for example, that understanding always entails an evaluation of the situation or object of emotion, and that action appears as moved by an ‘action tendency’ which is also describable by the end result it aims for. ⁶ Seeing the object as an intolerable obstacle and experiencing (and bodily expressing) the urge to vigorously remove it is a way of beginning an analysis of some of the episodes that we usually consider to be instances of anger. Emotions differentiated by appraisal and action tendency need not be conflated under a general category of anxiety.

As much as phenomenology and appraisal theory provide a way to integrate differentiated emotions in practical and discursive consciousness, Garfinkel (1963) points to an alternative analysis of the need for ontological security. Giddens is probably right in positing that the integration of personality depends on experiencing the continuity of the world, which involves the active sustaining of that continuity through habitual conduct. But I do not think that we need to go very far to understand why people are motivated to produce the continuity of the world. In this respect, Garfinkel stresses the significance of analyzing ‘social affects’ in terms of ‘background expectancies’, observing that people portray themselves as concerned with how to conduct their daily affairs ‘so as to solicit enthusiasm and friendliness or avoid anxiety, guilt, shame, or boredom’ (Garfinkel, 1963: 233). Background expectancies are basic, mostly implicit anticipations as to how the world is to continue. Conforming to them preserves calmness or brings about positive emotions, while breaching them triggers negative emotions. We only need to understand that human agents prefer the ease of calmness and the euphoria of positive emotions to the pangs of negative emotions ⁷ to understand why they conform to background expectancies, thereby sustaining the continuity of the world through habitual action. There is no need to conceptualize the search for ontological security in terms of an unconscious motive to avoid diffuse anxiety.

The role of emotions in face-work

To recap, even though I endorse Giddens’ formulation of the problem of structuration as that of the stretching of ‘the same’ situated actions across space and time, for the reasons presented above I think that his concept of routinization is unconvincing. I have pointed out some ways this concept could be amended so as to make it compatible with a conception of the active human agent.

I do think, however, that Giddens has identified a set of mechanisms whereby the routine character of day-to-day situated activity is sustained, namely the interactional techniques for expressing considerateness and maintaining self-respect that Goffman (1967) places under the heading of ‘face-work’.

Keeping the above qualifications in mind, I also think that Giddens is right to complain that a significant drawback of Goffman’s work is the studied absence of an account of motivation, especially because it has fostered the naive dramaturgical reading that participants in interactions are distanced actors deliberately playing scripts for the cynical manipulation of appearances. However, as we have seen, Giddens’ postulation of an unconscious motive to avoid anxiety is not of much help in this respect. An additional shortcoming of Giddens’ solution, not mentioned in the previous discussion, is that the alignment of emotion with anxiety, and of anxiety with the unconscious, carries with it the misleading connotation that the occurrence of emotion reveals not the normal functioning but the breakdown of face-work or ‘interaction ritual’. In this section I would like to provide some empirical support for the thesis that some specific emotions are essentially constitutive of face-work. Goffman (1967) did not entirely neglect the emotional dimension of face-to-face interaction, as his chapter on embarrassment and his sporadic use of the vocabulary of affect demonstrates. But neither did he attempt to systematically integrate emotions into his analysis of interaction.

Drawing on Goffman’s sequential description of ‘corrective interchanges’, Aranguren and Tonnelat (2014) in Paris, and later on Aranguren (forthcoming) in Delhi, investigated how passengers of crowded metro trains accomplish face-work in the face of unavoidable ‘territorial’ incursions due to unwanted physical contact during rush hours. These studies found that in such circumstances face-work is enacted mostly through nonverbal signals, which the psychological literature interprets as expressions of emotions. In treating nonverbal expressions of emotions as interactional ‘moves’, this research questions the implicit equation often made between interactional move and turn at talk. On the other hand, it considers expressions of emotions, usually treated formally and in isolation, from the point of view of their function in the larger context of interaction sequences.

In the Delhi and Paris metros, corrective interchanges often involve the following sequence of moves between the toucher (the offender) and the touched (the victim). First, the offender touches the victim, which in most cases involves mutual contact on the surface of the shoulders or the upper back. Second, the touched simultaneously raises and lowers the brow, thereby displaying the upper face components of the facial configuration that participants in recognition tests agree to associate with the word ‘fear’ (Ekman & Friesen, 1975). There are no good reasons, however, to think that the touched experiences fear; in fact, the content analysis that Aranguren and Tonnelat (2014) present at the outset of their study instead suggests that passengers experience annoyance at being touched by a stranger. ⁸ Third, with the ‘fear’ upper face on display, the touched turns their head and eyes toward the toucher.

In the complete repair sequence, the toucher then performs the fourth and last move by displaying some or all of the components of the facial configuration that participants in recognition studies agree to associate with the word ‘embarrassment’ (Keltner, 1995). The expression typically involves lowcast gaze, chin-raising, and lips-pressing. After this move on the part of the toucher, the touched returns their head and eyes to the initial position and the upper face components of the ‘fear’ face disappear from their forehead. These changes are taken to mark the successful termination of the repair sequence.

However, the sequence may unfold in a different manner with radically contrasting effects. The third move, namely the victim’s turning of the head and eyes toward the offender, may not be followed by the offender’s display of ‘embarrassment’. If, in addition to this, the offender subsequently repeats the annoying touch, the victim may terminate the sequence not with a direct return to neutral but with a prior expression that participants in recognition studies associate with the word ‘contempt’, involving an inward tightening of the lip corners (Ekman & Friesen, 1986; Izard & Haynes, 1988).

From a functional perspective, the expressions are not primarily outward presentations of inward feelings, but actions transforming the environment ⁹ that together form a unitary emotional transaction. The victim’s ‘fear’ upper face is a request or demand, and its function is to coerce an apology from the offender. Similarly, the offender’s ‘embarrassment’ expression functions to give ritual satisfaction to the victim’s demand. When this appeasing reply is not forthcoming, and the offender happens to repeat the touch, the victim’s ‘contempt’ expression effects symbolic exclusion or ‘degradation’ (Garfinkel, 1956) of the offender, assuming with Fischer and Roseman (2007) that the function of contempt is to exclude its target from the social world of the agent.

Thus, in the interaction sequences reported in these studies, the expressions of ‘fear’, ‘embarrassment’, and ‘contempt’, and their concomitant functions of coercion, appeasement, and exclusion, are indissociably intertwined with the workings of the repair sequence. Emotions are present not only when the repair sequence breaks down, as is the case when a repeat offense prompts the ‘contempt’ expression, but also when remedial work is effective, as is the case when the victim’s ‘fear’ face successfully coerces the appeasing ‘embarrassment’ expression of the offender.

These interactional episodes are not unrelated to the production and reproduction of specific sets of social practices or institutions. At the more proximal level, emotionally accomplished face-work contributes to the social organization of the metro ride, and therefore to the reproduction of the metro network as a complex technical and social system of mass transportation. At the most distal level, but at the same time entering into the local organization of the metro ride itself, successful face-work reproduces the institution of individualism, that ‘cult’ of the person that Durkheim deemed to be characteristic of modern society and which motivated Goffman to associate face-to-face interaction with ritual activity.

It is important to note that, even though face-work in general can be assumed to be a universal social mechanism, its form and function may present important variations across cultures and epochs, which in turn are related to the production and reproduction of specific social institutions. For example, as regards function, it is unlikely that face-work will serve the same purposes in a society that dramatizes hierarchy as compared to a society which, like ours, dramatizes equality. In other words, face-work reproduces the status of interaction partners, and the expected status to be confirmed depends on the institutional background of the larger society or specific setting in which the encounter takes place.

Interaction ritual theory

An important achievement of Collins’ theory is to have re-infused Goffman’s analysis of face-work as ‘interaction ritual’ with the emotions that Durkheim deemed characteristic of religious practice. Collins posits that interaction ritual is the mechanism of production and reproduction of society. Interaction rituals are bounded by particular situations, but their outcomes may ‘spill over’ in time and space, thereby placing conditions upon the initial ingredients of future and distant interaction rituals. The concept that connects the outcomes of the here-and-now interaction ritual with the starting ingredients of future and distant interaction rituals is called ‘emotional energy’.

The model of the interaction ritual thus distinguishes between ingredients and outcomes. The ingredients are: physical co-presence of two or more individuals, boundaries between insiders and outsiders, a common focus of attention, and shared (or collective) emotion. Collins’ contention is that, when the common focus of attention and the shared emotion reinforce each other through ‘rhythmic entrainment’, the interaction ritual brings about solidarity (i.e. feelings of membership), changes in the level of emotional energy, group symbols or ‘sacred objects’, and feelings of morality.

In order to identify a possible emotional mechanism of structuration as institutional creation, I retain Collins’ hypothesis that one outcome of interaction ritual may be the establishment of ‘sacred objects’. However, in this respect the theory lacks the empirical detail that would make it immediately applicable. This is why discussion of sacred-object establishment, or ‘enshrinement’ for short, is deferred to a later section concerned with Herbert Blumer’s (1958) middle-range theory of the ‘sense of group position’. On the other hand, no sensible use can be made of Collins’ theory without prior acknowledgement of its limitations. These are made explicit in the rest of the present section.

The most glaring shortcoming concerns the method of theory-building itself. Collins claims to bring forward testable hypotheses, but all his materials are second-hand and produced for purposes very different from the ones he has in mind. As a result, the relationship between the ingredients and the outcomes in the interaction ritual model largely remain hypothetical constructs, as much as the notion of emotional energy.

In addition, Collins asserts that the relationship between rhythmic entrainment and solidarity is causal in nature, but it seems to me that all he demonstrates at best is that rhythmic entrainment expresses solidarity. This is particularly apparent in his discussion of conversational turn-taking rules. He contends that conversations that follow the ‘no gap no overlap’ pattern are the cause of solidarity. This argument would make sense if it were possible to separate the cause, the smooth rhythmic coordination of turns, from the effect, the feelings of solidarity between the conversationalists, and then examine if the occurrence of the cause is followed by the occurrence of the effect. What Collins does, however, is indicate that synchronization expresses or manifests solidarity, just as a facial expression expresses or manifests an emotion. Conversely, lack of synchronization expresses lack of solidarity, so that ‘in angry arguments … both participants try to talk at the same time, typically speaking louder and faster in an effort to override the other’ (Collins, 2004: 70). Not that speaking louder and faster is the cause and lack of solidarity the effect: speaking louder and faster is simply a way of expressing lack of solidarity or hostility. It would be awkward for me to be told by a sociologist that my deliberate interruption of a person’s turn at talk was not an expression but the cause of my hostility toward that person. Since the demonstrated link is expressive and not causal, the inescapable conclusion is that the vocabulary of mechanism is inadequate to the relationship between rhythmic entrainment and feelings of solidarity. More generally, as von Scheve (2011) points out, Collins offers only sparse hints as to how exactly shared or collective emotions are elicited and transmitted.

This is not to say that interaction rituals cannot in principle be shown to produce solidarity or, more precisely, to reconfigure the status of interpersonal relations (Parkinson et al., 2004). But rhythmic entrainment alone does not seem to be the process responsible for this end result. Take the example of the repair sequence discussed above. The end result is to restore the relationship between the participants that was altered or disrupted by the offensive event. A specific mechanism does this job, namely the sequence of interactional moves that Goffman (1967) calls ‘corrective interchange’. Goffman occasionally compares the ‘interchange’ to a dance. I agree with Collins that rhythm is important, but nobody would suggest that a dance is only made of rhythm. The other essential component are the steps or, in Goffmanian jargon, the ‘moves’.

Now, looking at the moves of the repair sequence as instantiated in expressions of emotions leads me to question a further component of Collins’ model, namely the assumption that, for interaction rituals to produce their outcomes, the emotion generated must be common to the co-present individuals. The corrective interchange, when sufficient attention is accorded to its emotionally expressive components, is made of different emotions and not of just the one. What is more, these different emotions are also complementary, as is apparent in the relationship between the victim’s annoyance, which functions to coerce an apology, and the offender’s embarrassment, which acts to appease the victim. Of course, acknowledging that the emotions involved in interaction rituals may be complementary and not just common does not fundamentally invalidate Collins’ and Durkheim’s thesis that interaction rituals produce emotions. In Durkheimian terminology, it only adds an ‘organic’ variety of ritual emotion to the ‘mechanic’ variety of the original model. ¹⁰

The last critical comment I would like to make on Collins’ model prepares the ground for the next section and concerns the process hypothesized to establish new ‘sacred objects’. Collins thus summarizes the mechanism of sacred-object establishment:

There is the raw experience… Next comes the transformation of these experiences into symbols… Yet more temporally remote, and more remote, too, in the kinds of social networks involved, is a second order of circulation of newly created symbols among persons who are far away from the initial experiences. (Collins, 2004: 95)

In other words, first come the situated emotions that arise in face-to-face interactions, second the symbolization of these situated emotions, and third a process of circulation, remote in time and space, of the disembodied symbols that stand for these situated emotions.

Interaction ritual theory contends that emotions emerge from the coordinated behaviors of co-present individuals and that these emotions (or the solidarity that they are supposed to bring about) may become symbolized or ‘stored’ in the objects that were the common focus of attention at the time of intense emotion. In Berger and Luckmann’s (1967) terms, the contention is that ‘objectified’ emotions become ‘detached’ from the local circumstances of their emergence. However, interaction ritual theory adds little to the understanding of how emotionally charged objects become disembodied symbols that circulate in space–time sectors remote from the original situation of their emergence, with the eventual effect that such objects acquire the power to command the awe (e.g. the UNICEF logo) or the abhorrence (e.g. the Nazi Swastika) of large audiences. Not much is said about the latter effect either, and this is all the more unfortunate because the power of symbols to compel respect or rejection, rooted in situated emotions but stretching its influence across time and space, reconciles interaction ritual theory with more traditional ‘structural’ approaches in the sociology of emotions. Indeed symbols can be thought of as carriers of emotion rules, as I propose to rephrase Hochschild’s (1979) ‘feeling rules’. Otherwise put, as a result of interaction rituals and the subsequent process of circulation, symbols become compelling reminders of emotional obligations.

Emotion rules

Hochschild (1979) coined the phrase ‘feeling rules’ to refer to sanctioned pairings of types of emotions and types of situations. For example, the participants in a wedding must be happy; the participants in a funeral must be sad. ‘Feeling rules’ are also hypothesized to govern the intensity and duration of situated emotions: a slighted person can be ‘too’ angry or angry ‘for too long’. ‘Feeling rules’ determine situated emotions through ‘emotion work’, i.e. the activity of aligning actual with expected emotion, especially in those cases in which actual emotions are utterly inappropriate (e.g. being amused at a funeral).

Hochschild’s ‘feeling rules’ that compel ‘emotion work’ provide a useful account of how social structure determines or constrains situated emotions. But the terminological asymmetry between rules that regulate ‘feelings’ and a work that concerns ‘emotion’ is not unimportant. It actually reflects Hochschild’s internalist conception of emotion work, and of emotion processes more generally (see Aranguren, 2014 for a review). The activity of conforming emotion to social norms operates on inner feelings (‘deep acting’), and not on the outer behaviors that were at the center of Mauss’ (1968) classical analysis of the obligatory expression of emotions. However, contemporary theories of emotions no longer separate ‘feeling’, i.e. the content of emotional experience, from other components of emotion such as expressive behavior ¹¹ (Lambie & Marcel, 2002). I therefore retain Hochschild’s emphasis on the normative constraints that regulate emotions, but I suggest dropping the restrictive focus on inner feeling. This is why I prefer ‘emotion rules’ to the original ‘feeling rules’.

In sum, the process of sacred-object creation begins with situated emotions and ends with the establishment of emotion rules carried by definite symbols. In between, situated emotions are symbolized, and the resulting symbols enter a process of circulation. The challenge is to spell out the connections between the processes of situated emotion, emotion symbolization, symbol circulation, and symbol enshrinement, a sequence that I take to embody an emotional mechanism of structuration. Herbert Blumer’s (1958) concept of ‘sense of group position’ provides an illuminating avenue in this respect.

The sense of group position: emotion rules in interethnic relations

In the domain of interethnic relations, an apt concept to stand for emotion rules is Blumer’s (1958) ‘sense of group position’. With this notion Blumer intends to sociologize the psychological concept of race prejudice as it applied to ‘race relations’ between Blacks and Whites in the United States. Without denying the emotional character of race prejudice, Blumer contends that the feelings of antipathy, hostility, hatred, intolerance, and aggressiveness that enter into the traditional definition of prejudice are not the outcome of individual inclination but of a social process.

For Blumer, ‘[a] basic understanding of race prejudice must be sought in the process by which racial groups form images of themselves and of others’ (Blumer, 1958: 3) In this collective process of characterization ‘definitions are presented and feelings are expressed’ (1958: 5). The outcome of the process is a socially sanctioned ‘sense of group position’, which governs the emotions that members of one racial group ought to have with regard to members of the other racial group. For example, in post-slavery race relations, the sense of group position implies the following components from the standpoint of Whites: (1) a feeling of White superiority, (2) a feeling that Blacks are intrinsically different and alien, (3) a feeling of White legitimate claim to certain areas of privilege and advantage, and (4) a fear and suspicion that Blacks threaten these privileges and advantages.

The sense of group position can be seen as a complex of socially enforced emotion rules. Conversely, these emotion rules come into being through a collective process of characterization in which situated emotions play an important role. ‘The happening that seems momentous, that touches deep sentiments, that seems to raise fundamental questions about relations, and that awakens strong feelings of identification with one’s racial group is the kind of event that is central in the formation of the racial image’ (1958: 6). The definition process starts with the narrativization of situated emotions and involves the circulation of narratives through various media such as stories, gossip, anecdotes, or pronouncements. These sediment into a stable characterization of one’s and the others’ ethnic group when the interaction becomes circular and reinforcing. An additional aspect that Blumer highlights is the abstract and categorical quality of the characterization thus formed. Indeed, it is the very nature of the resulting stereotypes to make individuals into symbols of their social category, who must therefore be treated not as singular persons but as undifferentiated embodiments of that category. The members of venerated or stigmatized social categories thus become ‘sacred objects’ that command respect or rejection, awe or abhorrence, compassion or disgust.

Bringing together Collins’ interaction ritual theory, Hochschild’s emphasis on the structural determination of emotion, and Blumer’s concept of the sense of group position, the following emotional mechanism of structuration appears to be relevant to the study of interethnic relations. Encounters between two or more individuals assigning one another to different ethnic groups give rise to situated emotions. Within or after the encounter, such emotions are represented or symbolized in various media. This involves among others the evaluative characterization of the relevant ethnic groups, and the corresponding emotional attitudes that one is expected to hold toward members of the groups so characterized. For example, a group characterized as offensive is the appropriate target of anger; a group held to be superior, of awe; a group depicted as polluting, of disgust; a group seen as a victim, of compassion. These evaluative characterizations, including the emotional rules they imply, then become disembodied symbols that enter a process of interaction with other comparable disembodied characterizations. All these symbols are candidates for sacred objects, but only those that reinforce each other to the extent of marginalizing competing depictions become enshrined, that is pervasive, recurrent, self-evident, or taken for granted. Since the enshrined characterizations involve emotional obligations toward the members of such and such an ethnic group, at the end of the process they arise as the embodiment of definite emotion rules that constrain, by anticipation, future and distant encounters with the members of those groups. For example, if the characterization of the Roma that becomes self-evident depicts this group as dangerous (as opposed to vulnerable), fear (as opposed to compassion) becomes the obligatory emotion that the non-Roma are expected and expect to experience in encounters with the Roma. In this example, the individuals assigned to the Roma category become symbols of mandatory emotions. This is why, in interethnic contacts, their presence commands fear, that is the tendency to avoid.