Today we will look at the work of Lynn Smith-Lovin and David Heise on what they call the affect-control theory of emotion. Smith-Lovin begins by noting that one goal of sociologists who study emotion is a powerful, general model of how people understand and react emotionally to social situations. To do this, it is necessary to develop a model that integrates the insights gained by both qualitative and quantitative research. It must recognize that emotions reflect our understanding and definitions of social events, how we see ourselves and others, and the normative aspects of emotional display. It should show how non-normative emotions lead to qualitative changes in our view of reality.

A sociological theory of emotion should link emotional response also to other aspects of social action like identity, role, ideology and culture. Ideally this will be a parsimonious model of social action where our emotional reactions are affected by social events around us, which then mobilize us to further action. It will also serve as a framework within which to fit more specialized findings of researchers with more specialized interests.

In this chapter, Smith-Lovin describes a theoretical model of emotion and social action that begins to attempt this task. Affect control theory (ACT) provides a model of social action incorporating the basic assumptions of symbolic interactionism, but also captures relationships among social institutions, emotion and social action. Today we will:

  1. Summarize the theory and discuss its view of social life;
  2. Comment on how ACT processes relate to other research traditions in the sociology of emotion;
  3. Outline questions for further research.

The Affect-Control Model:

Heise (1977-79) developed ACT from symbolic interactionist ideas and studies in the psychology of impression formation. It is based on the proposition that people perceive and create events to maintain the meanings evoked by their definition of a situation.

The argument is that when people enter a given situation, they occupy identities (e.g. doctor, guest or patient). Some may be formal and institutional, others may be less formal and institution-specific (e.g. friends encountering one another in public). Through their chosen identities, actions and physical props, actors may transform the relations between people (e.g. causing a scene at a social event by getting angry and throwing things).

People recognize social events within their definition of the situation (e.g. the Host is


ignoring me, his guest). The interpretation of behavior (e.g. its cognitive categorization or label) determines its emotional impact (e.g. its OK to discipline a child, but not to abuse them). Of course, individual variations in interpretation may occur, but these are not unlimited in such situations.

The labels we use to characterize self, others and social actions carry important meanings - fundamental assumptions about how good , how powerful and how lively such people and behaviors are. A key feature of ACT is that it conceptualizes Ameaning@ in a specific, measurable way. These fundamental sentiments of goodness, powerfulness and liveliness correspond to the 3 dimensions of emotional meaning (evaluation, potency and activity). These dimensions underlie reactions to many types of concepts in a variety of linguistic and national cultures, including our own (e.g. despite widely varying backgrounds, we largely agree, for example, that mothers are nicer than Mafiosi, that physicians have more power than patients, and that children are livelier than Acripples.@) Studies on a wide range of topics find surprising agreement on the emotional meanings associated with social identities and behaviors.

Obvious exceptions occur when subcultures develop unique meanings for identities central to the group. Yet, by adjusting the weightings to incorporate these alternate meanings, even these can be examined within this framework.

ACT focuses on the goodness, powerfulness, and activity dimensions of meaning for 4 reasons: (1) they can be used to characterize many significant elements of social situations, including identities, actions, emotions and settings; (2) they are widely shared and represent important cultural information; (3) they correspond to important social features of identities and behaviors like status, power and expressiveness; and (4) we can measure people=s reactions on these dimensions. Moreover, our measurements of affective meaning allow us to link the qualitative features of situational definitions to the quantitative processes of impression-change and control.

Social events may change impressions of people, making them seem better or worse, stronger or weaker, livelier or quieter than they were expected to be (e.g. a host who ignores his or her guest damages the reputation of each, losing goodness and power, though not liveliness). ACT uses impression-change equations, estimated from large numbers of such events, to predict this outcome on the 3 dimensions. One might think of an identity as a point in a 3 dimensional space: an event, when recognized and processed by someone, has the capacity to move the person in that identity from its original, culturally given location to a new position. In the above example, the host suffers because she has done bad to a good person, and the guest suffers from being ignored. These are reflected in changes in the evaluation, potency and activity dimensions.

When events create transient impressions that so differ from our fundamental understandings of what people and behaviors are like, we are likely to generate new events that restore these fundamental sentiments. In other words, people construct new events to confirm fundamental sentiments about self and others; they manage social life in ways that control their feelings about reality. The maintenance of meaning is accomplished by creating a new event that, when processed, will move the identities of actor, behavior, and object back to their original positions in the 3 dimensional space.

So, in the above example, ACT suggests that being ignored is unlikely since it doesn=t support and maintain the identities of the interactants. There may be other, perhaps accidental reasons for this (e.g. the host is busy, preoccupied with another emotional episode, or has a mistaken impression of the actor). If the guest points out the ignoring behavior, and the host accepts this account, the host is likely to produce a restorative act - a behavior that, when processed, will move impressions back in line with fundamental meanings (e.g. soothe, apologize and or appreciate).

ACT produces such predictions about impression change and behavioral reactions with a computer program called INTERACT. This contains the formal structure of the theory and empirically derived estimates of its parameters. Specifically, it consists of equations that show how impressions change on the three dimensions of evaluation, potency and activity for actors, object-persons, and behavior. It uses a mathematical transformation to represent the proposition that people maintain fundamental meanings. It solves for the 3-number EPA profile of possible behaviors that would return the identities of the actor and object closer to their positions before the event was processed. This enables us to link verbal descriptions of identity and behavioral labels to their numerical values in EPA terms.

INTERACT also can model situations where no new event or action could possibly restore the original sentiments of participants (e.g. shocking crimes by one actor, divorce). Such events may lead to a redefinition of the situation. In these cases, there is a search for new identities that Amake sense@ of the event. The program solves for a new identity profile that fits (e.g. Acriminal@ instead of Afather@; Abitch@ or Apsychotic@ instead of loving wife).

ACT is mathematical in form, but both qualitative and quantitative in its predictions. Although it is based on empirical equations, mathematical transformations, and computer simulations, it also works with natural language descriptions of the social situation and what is happening in it. It also produces natural language outputs showing how people might interpret the situation and how they might respond.

The Affect-Control Model of Emotion:

Early versions of ACT modeled only behavioral responses and labeling, and did not deal explicitly with emotional reaction. This was rectified as later work developed the empirical base necessary to describe emotions within an ACT framework.

ACT assumes that emotions provide signals about how well events are maintaining social meanings. Events may produce transient impressions that vary from our fundamental notions of how good, how powerful, and how lively we, or others, are, or ought to be. Emotions are the Acode@ for representing the degree and kind of confirmation or disconfirmation of identities that is occurring (e.g. if someone insults you in a positive identity such as student, one is put down, deflected away from a fundamentally good, pleasant identity by a negative act).

In the ACT model, emotion words describe feelings that characterize the combination of a fundamental EPA profile (i.e. the person=s original identity in a situation) and the transient profile produced by an event. The insulted person above is predicted to feel flustered, embarrassed and shook up, while the person insulting him might feel spiteful or bitchy).

The emotional reactions and the behaviors predicted by INTERACT are, in effect, hypotheses derived formally from the theory. If people actually define the situation as posed, and view the action as expressed, then the behavioral predictions and emotions should be the ones that are actually experienced.

Identity and Emotion:

Emotions combine with identities to form impressions of emotion - identity composites (e.g. sad friend). The model uses these equations to describe how identities and impressions from events imply emotion. The equations reveal that emotion is a function of: (1) the transient impression created by an event, and (2) the difference between the transient impression and the fundamental meaning. Not surprisingly, nice events lead to positive transient impressions and positive emotions; nasty behaviors lead to negative emotions. But a transient impression that is nicer than one=s identity fosters an especially positive emotion. However, if one occupies very high status, then one simply expects the nice things that are directed one=s way.

If people interpret events and act to maintain their transient states near fundamental meanings, then ACT points to an interesting conclusion: that the character of one=s emotions is sharply determined by his or her identity. Maintenance of positive identities in the kinds of roles one normally occupies creates positive emotions. Maintenance of a negative identity (e.g. a deviant role) normally fosters negative emotions. For example, people in powerful identities may be expected to feel deep, high potency emotions like pride and fury, while people in powerless positions should feel fear, anxiety and depression. People inhabiting less lively identities are likely to be prone to quieter emotions (e.g. contentment, remorse), while those in more lively identities would be inclined toward emotions like euphoria, fury and passion. ACT computer simulations suggest a variety of emotions that would most likely be associated with people in various roles in normal circumstances (e.g. boyfriend would feel happy; grandfather kind, etc.)

Of course, occupying stigmatized identities requires people to feel negative emotions if they remain part of the dominant culture (e.g. mentally ill people and juvenile offenders often incorporate society=s negative view and have lowered self-esteem). Many powerless people share society=s Anormal@ view of their worthlessness.

Although negative emotions can be the mechanism by which deviants are reclaimed to normal society, they may also undermine one=s commitment to the dominant ideology. When no avenues for identity change are available, they may create or search for a subculture where they are offered positive feelings through a competing ideology. They may develop that makes their emotional life tolerable, even fulfilling (e.g. a gay church).

Situations Leading to Disconfirmation and Emotional Response:

ACT also alerts us to situations where events fail to confirm identities. Since disconfirmation (i.e. the deflection of a transient identity from fundamental meaning) is experienced as an emotional response, ACT indicates the types of situations that are likely to lead to this kind of emotional arousal.

Disconfirmation of identity is likely when people differ in their definition of a situation (e.g. student mistaking professor=s identity as secretary). In such cases actors will intend and expect events that confirm differing views of the interaction: actions that are confirming for 1 will be disconfirming for the other. Disconfirmation of identity also occurs when people come from differing subcultures and have different meanings for identities and behaviors: 2 people may agree on the definition of the situation, but different cultural backgrounds may entail different EPA profiles (e.g. professors may expect more lively, egalitarian exchange with students, but those from some cultural backgrounds are taught to be quiet and deferential). This may make confirming events from one party=s standpoint disturbing for the other (e.g. differences in expected power/liveliness between husband and wife in various cultures illustrates different predicted behaviors and emotions should members of these groups interact on these role-identities). While various subcultural pairings - and the deflections of meaning they imply - still produce many positive emotions in this model, the variations in power and expressivity lead to substantial differences in normative emotional climates (e.g. the lighthearted affection predominating in the view of North Americans, and the calm contentment expected by more traditional Japanese women).

Such situations produce identity disconfirmation because interaction partners come into association with differing views (either because they disagree about the relevant identities or because they hold different meanings for the same identities). Yet, full scale disagreement isn=t necessary for disconfirmation and emotion. Events can push impressions away from fundamental meanings simply because few actions can perfectly confirm all identities and behaviors in a complex situation. Many situations require us to operate simultaneously in multiple identities of varying importance (e.g. work and home): acts that confirm one may disconfirm another, and some settings may have identity conflict built into them (e.g. remarriage).

Disconfirmation and Emotion:

When disconfirming events do occur, emotions signal the character of the deflection. When especially good fortune strikes, ACT predicts positive emotions for the parties (e.g a child even a gift).Negative events cause negative emotions for both the actor and object (e.g. the punished child and his/her punishers).

Emotion and Labeling:

ACT reveals interesting dynamics for the object of a negative act. While victims may be blameless in principle, they suffer negative consequences beyond material damage since the tendency to derogate victims is strong. In ACT equations, victimization may require that the person feel and express negative emotions to confirm the vileness and inappropriateness of the act. A victim who leaks a display of positive emotion invites being stigmatized and labeled - seen as someone who got what s/he deserved. On the other hand, if one feels and displays the proper distress, one is then free to engage in constructive acts of recovery. Yet, recovery implies a certain amount of power over circumstances - and unfortunately victimization is most likely for those with least power. The negative consequences of victimization - unresolved negative feelings or permanent stigmatization by self and others - are greatest in those groups without power and prestige.

Displays of appropriate emotion cues are used by actors to avoid labeling during self­disclosure of negative information. Negative information typically leads to firm impressions and to dislike, but, if we become close to others and disclose much, how is it that others sometimes like us anyway? Experimental research showed that sadness on the speaker=s part moderates listener=s reactions to one=s negative self-disclosure content (e.g. of having harmed another). ACT has a ready explanation for this mitigating effect of emotion. Through our sadness, we express our recognition of the negative character of our behavior, showing that our self-concept or identity is fundamentally positive. Our emotion signals that the negative behavior that we have described has caused negative deflection. Since only bad people are expected to do bad things and feel neutral or good about them, we display negative emotions when revealing our shortcomings to others. Such emotions are often played to and for an audience to authenticate identities and mitigate the effects of negative information revealed through circumstance or self­disclosure.

Others= emotional displays help us infer the character of actions or actors in ambiguous circumstances (e.g. if visiting another culture and acting in a fashion that surprisingly produces horror, one quickly realizes the behavior is immoral or dangerous in that culture). Similarly, one can infer one=s expected character from other=s emotional reactions to the same behavior (e.g. when a man shortchanges a child, will others express disgust, cynical amusement, or relief). The man=s identity can be inferred from the event and the observers= reactions. In ACT terms, you have solved for an affective profile that makes sense of the peoples= reactions, then searcher for a language label that corresponds to the EPA profile of the identity. Afterwards, one is able to generate further expectations for the man in realms other than interaction with the child.

Affect Control Theory and Other Research Traditions:

While I won=t go into this part in much detail, it is clear that ACT=s Apredictions@ correspond to important insights by other researchers. Smith-Lovin hopes that this model can add to the sociology of emotion by creating a formal apparatus for predicting which emotion occurs in which circumstances. It may provide a framework, she feels, for integrating and interpreting findings in a wide range of research traditions.

Smith-Lovin, for example, argues that ACT has some compatibility with:

- Denzin= s focus on situational definitions;

- Kemper= s focus on power, status, and predictive outcomes

- Goffman= s focus on embarrassment and shame as reflective of identity in interaction

- Clark= s work on sympathy, including its tactical use as a micropolitical strategy

ACT is also seen as compatible with normative, ideological approaches to emotion:

- Collins= ritual sequences can create powerful, predictable emotional outputs

- Hochschild= s emphasis on ideological norms, emotional management, its successes and failures, and the consequent outcomes

- Thoits= research on emotional deviance (e.g. when feelings don= t match expectations).

The Research Agenda: Testing the Model:

Smith-Lovin argues that one of the first things that needs to be done is to test the predictions of the INTERACT simulations. While they seem roughly correct and agree with previous research, more systematic work needs to be done to see whether the emotions the model produces are actually what people expect and feel in real situations. There are 3 possibilities:

  1. Making use of past, naturally occurring experiences and trying to match them to predictions generated by using affect control dynamics (this can be difficult as it is often an interpretive process);
  2. Present a hypothetical scenario and ask people what they expect as an emotional response;
  3. Compare theoretical predictions to people=s behavior in real social situations (ethical problems come in here if done in an experimental fashion, since disconfirming identities may be harmful). Examples could be experiments involving manipulation of self-esteem or emotional cues.

Smith-Lovin says she doesn=t want to limit testing to traditional survey or experimental work, but to engage creatively in evaluating this model (e.g. through the use of videotape, archival materials such as court transcripts).

Already, a number of studies derived from ACT principles lend some empirical support to some of the fundamental notions. Robinson, Smith-Lovin and Tsoudis (1994) used simulations based on the theory to predict the effect of emotion displays on identity attributions in a setting involving sentencing recommendations after criminals had confessed to certain crimes. A display of remorse after confession affected the severity of the sentence throught its impact on the assessment of the identity of the confessor, as predicted by the theory. Robinson and Smith-Lovin (1992) also applied ACT derivations to propose a counterintuitive set of hypotheses about the effect of self-esteem (a mood indicator) on the tendency to solicit identity­supporting sentiments and behaviors from others. They found that experimental subjects reacted homeostatically to maintain their identity as either high or low in self-esteem. Both high and low self-esteem subjects felt good when praised and bad when criticized, but low self-esteem subjects felt the criticism was accurate and liked the critics more than did the high self-esteem subjects. When given a chance to choose interaction partners, high and low self-esteem subjects tended to select those who would support their view of themselves.

New Substantive Directions:

All of these suggestions focus on testing the theory by matching the productions of the simulation program to other data. Yet, if ACT is found to be a useful way of viewing emotional phenomena, it may provide insights into other issues of substantive concern. For example, using ACT it may be possible to compare cross-national data samples in terms of cultural practices, emotion norms, and labeling behaviors. It may also be possible to compare subcultures within a given society with the broader society in terms of (1) emotion norms; (2) socialization; (3) conflict between ideologies; (4) rituals. Indeed, studies within our own culture can concentrate on areas likely to cause strong emotions (e.g. role strain, differences in the definition of the situation and strong emotions, such as in sexual harassment).

Theoretical Questions:

Computer simulations have also suggested some interesting theoretical questions that need further work:

- the motivational features of emotions (i.e.) Do they realign behavior with identities?

- do emotions produce and explain out of role behavior?

- the emotional responses of bystanders or observers not directly involved (vicarious emotional reactions)

- whether ACT dynamics may be linked to models of rational action

Based on interpretive principles, ACT contains the impact of both cultural values (through the fundamental sentiments) and individual thought (through the definition of the situation and its maintenance). It is specific enough to be falsifiable (i.e. tested). That is why some of the research discussed is so important: it will tell us whether ACT is fundamentally correct in its view of emotions as signals of identity confirmation and disconfirmation. If it is, it will also illustrate how this model needs to be elaborated to reflect the emotional complexities of social life.