Sociologist Erving Goffman (1969) argued that people conduct themselves so as to generate impressions that maintain the identities, or ”faces,” that they have in social situations. Human action— aside from accomplishing tasks—functions expressively in reflecting actors’ social positions and in preserving social order. Affect control theory (Heise 1979; Smith-Lovin and Heise 1988; MacKinnon 1994) continues Goffman’s thesis, providing a mathematized and empirically grounded model for explaining and predicting expressive aspects of action.


Cross-cultural research among people speaking diverse languages in more than twenty-five nations around the world (Osgood, May, and Miron 1975) revealed that any person, behavior, object, setting, or property of persons evokes an affective response consisting of three components. One component consists of approval or disapproval of the entity—an evaluation based on morality (good versus bad), aesthetics (beautiful versus ugly), functionality (useful versus useless), hedonism (pleasant versus unpleasant), or some other criterion. Whatever the primary basis of evaluation, it tends to generalize to other bases, so, for example, something that is useful tends also to seem good, beautiful, and pleasant.

Another component of affective responses is a potency assessment made in terms of physical proportions (large versus small, deep versus shallow), strength (strong versus weak), influence (powerful versus powerless), or other criteria. Again, judgments on the basis of one criterion tend to generalize to other criteria, so, for example, a powerful person seems large, deep (in a metaphorical sense), and strong.

The third component of affective responses— an appraisal of activity—may depend on speed (fast versus slow), perceptual stimulation (noisy versus quiet, bright versus dim), age (young versus old), keenness (sharp versus dull), or other criteria. These criteria also generalize to some degree so, for example, a young person often seems metaphorically fast, noisy, bright, and sharp.

The evaluation, potency, and activity (EPA) structure in subjective responses is one of the best-documented facts in social science, and an elaborate technology has developed for measuring EPA responses on ”semantic differential scales” (Heise 1969). The scales consist of adjectives separated by a number of check positions. For example, a standard scale has Good-Nice at one end and Bad-Awful at the other end, and intervening positions on the scale allow respondents to record the direction and intensity of their evaluations of a stimulus. The middle rating position on such scales represents neutrality and is coded 0.0. Positions moving outward are labeled ”slightly,” ”quite,” ”extremely,” and ”infinitely,” and they are coded 1.0, 2.0, 3.0, and 4.3 respectively—positive on the good, potent, and active sides of the scales; and negative on the bad, impotent, and inactive sides (Heise 1978).

EPA responses tend to be socially shared within a population (Heise 1966, 1999), so a group’s average EPA response to an entity indexes the group sentiment about the entity. Group sentiments can be computed from as few as thirty ratings since each rater’s transient response originates from a single shared sentiment rather than a separately held position (Romney, Weller, and Batchelder 1986). Sentiments vary across cultures. For example, potent authorities such as an employer are evaluated positively in U.S. and Canadian college populations, but German students evaluate authorities negatively; small children are evaluated positively in Western nations but neutrally in Japan.


Combinations of cognitive elements bring affective meanings together and create outcome impressions through psychological processes that are complex, subtle, and yet highly predictable (Gollob 1974; Heise 1979; Anderson 1996).

One kind of impression formation amalgamates a personal attribute with a social identity resulting in a sense of how a person is different from similar others (Averett and Heise 1987). For example, among U.S. college students (the population of raters for examples henceforth), someone who is rich is evaluatively neutral, very powerful, and a little on the quiet side. Meanwhile, a professor is fairly good, fairly powerful, and a bit quiet. The notion of a ”rich professor” combines these sentiments and yields a different outcome. A rich professor is evaluated somewhat negatively, mainly because the personalized power of wealth generates an uneasiness that is not overcome by esteem for academic status. A rich professor seems very powerful because the average potency of wealth and of professors is high, and the mind adds an extra increment of potency because of the personalized power deriving from wealth. A rich professor seems even quieter than the component statuses because activity connotations do not merely average, they summate to some degree.

The processes that are involved in combining a social identity with a status characteristic like ”rich” also are involved in combining a social identity with personal traits. Thus, an authoritarian professor evokes an impression roughly similar to a rich professor since the affective association for authoritarian is similar to that for rich.

Another example involving emotion illustrates additional processes involved in combining personal characteristics with social identities (Heise and Thomas 1989). Being outraged implies that one is feeling quite bad, somewhat potent, and somewhat lively. A child is felt to be quite good, quite impotent, and very active. The combination ”outraged child,” seems fairly bad, partly because the child is flaunting personalized power deriving from an emotion and partly because the mind discounts customary esteem for a person if there is a personal basis for evaluating the person negatively: The child’s negative emotion undercuts the regard one usually has for a child. The child’s impotency is reduced because of a bad and potent emotional state. And the child’s activity is greater than usual because the activity of the emotion and the activity of the identity combine additively.

The combining of attributions and identities involves somewhat different processes in different domains. For example, North American males and females process attributions in the same way, though both make attributions about females that are more governed by morality considerations than attributions about males (Heise 1999). Meanwhile, Japanese males are socialized to be more concerned than Japanese females with evaluative consistency in attributions and with matching goodness and weakness, so Japanese males process attributions about everything with more concern for morality than do Japanese females (Smith, Matsuno, and Ike 1999).

Events are another basis for impression formation. A social event—an actor behaving on an object or person within some setting—amalgamates EPA impressions of the elements comprising the event and generates a new impression of each element. For example, when an athlete strangles a coach, the athlete seems bad, and the coach seems cowardly because the coach loses both goodness and potency as a result of the event. The complex equations for predicting outcome impressions from input impressions have been found to be similar in different cultures (Heise 1979; Smith-Lovin 1987a; Smith-Lovin 1987b; Smith, Matsuno, and Umino 1994).

To a degree, the character of a behavior diffuses to the actor who performs the behavior. For example, an admired person who engages in a violent act seems less good, more potent, and more active than usual. Impressions of the actor also are influenced by complex interplays between the nature of the behavior and the nature of the object. For example, violence toward an enemy does not stigmatize an actor nearly so much as violence toward a child. That is because bad, forceful behaviors toward bad, potent objects seem justified while such behaviors toward good, weak objects seem ruthless. Moreover, the degree of justification or of ruthlessness depends on how good the actor was in the first place; for example, a person who acts violently toward a child loses more respect if initially esteemed than if already stigmatized.

Similarly, diffusions of feeling from one event element to another and complex interplays between event elements generate impressions of behaviors, objects, and settings. The general principle is that initial affective meanings of event elements combine and thereby produce new impressions that reflect the meaning of the event. Those impressions are transient because they, in turn, are the meanings that are transformed by later events.


Normal events produce transient impressions that match sentiments, whereas events that generate impressions deviating widely from sentiments seem abnormal (Heise and MacKinnon 1987). For example, ”a parent assisting a child” creates impressions of parent, child, and assisting that are close to sentiments provided by our culture, and the event seems normal. On the other hand, ”a parent harming a child” seems abnormal because the event produces negative impressions of parent and child that are far different than the culturally-given notions that parents and children are good.

According to affect control theory, people manage events so as to match transient impressions with sentiments and thereby maintain normality in their experience. Expressive shaping of events occurs within orderly rational action, and ordinarily the expressive and the rational components of action complement each other because cultural sentiments incite the very events that are required by the logic of social institutions like the family, law, religion, etc.

Having adopted an appropriate identity at a scene and having cast others in complementary identities, a person intuits behaviors that will create normal impressions. For example, if a person in the role of judge is to act on someone who is a proven crook, then she must do something that confirms a judge’s power and that confirms the badness of a crook, and behaviors like ”convict” and ”sentence” produce the right impressions. Fitting behaviors may change in the wake of prior events. For example, a father who is fulfilling his role in a mediocre manner because his child has disobeyed him strives to regain goodness and power by controlling the child or by dramatizing forgiveness. Other people’s identities may serve as resources for restoring a compromised identity (Wiggins and Heise 1987); for example, a father shaken by a child’s disobedience might recover his poise by supporting and defending mother.

Behaviors that confirm sentiments are the intrinsically motivated behaviors in a situation. Actors sometimes comply to the demands of others and thereby forego intrinsically motivated behaviors. Yet compliance also reflects the basic principle since compliant behavior is normal in one relationship even though it may be abnormal in another relationship. For example, a child acts normally when calling on a playmate though also abnormally if his mother has ordered him not to do so and he disobeys her. The actor maintains the relationship that is most salient.

Sometimes other people produce events that do not confirm sentiments evoked by one’s own definition of a situation. Affect control theory suggests several routes for restoring consistency between impressions and sentiments in such cases. First, people may try to reinterpret other’s actions so as to optimize expressive coherence. For example, an actor’s movement away from another person can be viewed as departing, leaving, escaping, fleeing, deserting—and one chooses the interpretation that seems most normal, given participants’ identities and prior events (Heise 1979). Of course, interpretations of a behavior are bound by deter-minable facts about the behavior and its consequences, so some behaviors cannot be interpreted in a way that completely normalizes an event.

Another response to disturbing events is construction of new events that transform abnormal impressions back to normality. Restorative events with the self as actor might be feasible and enacted, as in the example of a father controlling a disobedient child. Restorative events that require others to act might be elicited by suggesting what the other should do. For example, after a child has disobeyed his mother, a father might tell the child to apologize.

Intractable disturbances in interaction that cannot be handled by reinterpreting others’ actions or by instigating new events lead to changes in how people are viewed, such as attributing character traits to people in order to form complex identities that account for participation in certain kinds of anomalous events. For example, a father who has ignored his child might be viewed as an inconsiderate person.

Changing base identities also can produce the kind of person who would participate in certain events. For example, an employee cheating an employer would be expressively coherent were the employee known to be a lawbreaker, and a cheating incident may instigate legal actions that apply a lawbreaker label and that withdraw the employee identity. The criminal justice system changes actors in deviant events into the kinds of people who routinely engage in deviant actions, thereby allowing the rest of us to feel that we understand why bad things happen.

Trait attributions and labels that normalize particular incidents are added to conceptions of people, and thereafter the special identities may be invoked in order to set expectations for a person’s behavior in other scenes or to understand other incidents. Everyone who interacts with a person builds up knowledge about the person’s capacities in this way, and a person builds up knowledge about the self in this way as well.


Affect control theory is a central framework in the sociology of emotions (Thoits 1989; Smith-Lovin 1994), and its predictions about emotions in various situations match well with the predictions of real people who imagine themselves in those situations (Heise and Weir 1999). According to affect control theory, spontaneous emotion reflects the state a person has reached as a result of events and also how that state compares to the ideal experience of a person with a particular social identity. For example, if events make a person seem neutral on goodness, potency, and activity, then the tendency is to feel emotionally neutral, but someone in the sweetheart role ends up feeling blue because he or she is experiencing so much less than one expects in a romantic relationship.

Because emotions reflect the impressions that events have generated, they are a way of directly sensing the consequences of social interaction. Because emotions simultaneously reflect what kinds of identities people are taking, emotions also are a way of sensing the operative social structure in a situation. Moreover, because displays of emotion broadcast a person’s subjective appraisals to others, emotions contribute to intersubjective sharing of views about social matters.

People sometimes mask their emotions or display emotions other than those that they feel spontaneously in order to hide their appraisal of events from others or to conceal personal definitions of a situation. For example, an actor caught in misconduct might display guilt and remorse beyond what is felt in order to convince others that he believes his behavior is wrong and that he is not the type who engages in such activity. Such a display of negative emotion after a deviant act makes the actor less vulnerable to a deviant label— an hypothesis derived from the mathematics of affect control theory (Heise 1989).


Affect control theory provides a comprehensive social-psychological framework relating to roles, impression formation, behavior, emotion, attribution, labeling, and other issues (Stryker and Statham 1985). Consequently it is applicable to a variety of social-psychological problems. For example:

• Smith-Lovin and Douglass (1992) showed that sentiments about relevant identities and behaviors are positive in a deviant subculture, and therefore subcultural interactions are happier than outsiders believe.

• MacKinnon and Langford (1994) found that moral evaluations determine the prestige of occupations with low and middle but not high levels of education and income; and they found that income affects occupational prestige partly by adjusting feelings about the potency of the occupation.

• Robinson and Smith-Lovin (1992) found that people with low self-esteem prefer to associate with their critics rather than their flatterers. Robinson (1996) showed that networks can emerge from self-identities, with cliques reflecting differing levels of self-esteem, and dominance structures reflecting differing levels of self-potency.

• Francis (1997a, 1997b) showed that therapists often promote emotional healing by embedding clients in a social structure where key identities have particular EPA profiles; the identities are associated with different functions in different therapeutic ideologies.

• Studies of courtroom scenarios (Robinson, Smith-Lovin, and Tsoudis 1994; Tsoudis and Smith-Lovin 1998) showed that people (such as jurors) deal more leniently with convicted criminals who show remorse and guilt over their crimes, as predicted by affect control theory. In a related study, Scher and Heise (1993) suggested that perceptions of injustice follow justice-related emotions of anger or guilt, so social interactional structures that keep people happy can prevent mobilization regarding unjust reward structures.

• Heise (1998) suggested that solidarity comes easier when a group identity is good, potent, and lively so that group members engage in helpful actions with each other and experience emotions in parallel. Britt and Heise (forthcoming) showed that successful social movements instigate a sequence of member emotions, culminating in pride, which reflect a good, potent, and lively group identity.

Affect control theory’s mathematical model is implemented in a computer program that simulates social interactions and predicts the emotions and interpretations of interactants during expected or unexpected interpersonal events. Simulations can be conducted with EPA measurements obtained in a variety of nations—Northern Ireland, Canada, Germany, Japan, and the United States. The computer program, the datasets, and other materials are available at the affect control theory web site at:


Goffman (1967, p. 9) called attention to the expressive order in social relations:

By entering a situation in which he is given a face to maintain, a person takes on the responsibility of standing guard over the flow of events as they pass before him. He must ensure that a particular expressive order is sustained-an order that regulates the flow of events, large or small, so that anything that appears to be expressed by them will be consistent with his face.

Affect control theory’s empirically based mathematical model offers a rich and productive foundation for studying the expressive order.

Next post:

Previous post: